text
stringlengths 7
107k
|
|---|
1C, D). The relation between LB and µ (Fig. E) contained a number of intriguing
features. First, as ppGpp decreased, both LB and µ increased initially (Fig. 1E), in
agreement with the well-known finding that faster growing cells are larger. However,
decreasing ppGpp further led to an inverted trend, in which slower growing cells are larger
(Fig. 1E). This deviation began at near-endogenous ppGpp levels (Fig. C-E). A counter-
intuitive consequence of this inversion is that excursions above and below this
endogenous ppGpp level lead to cells that differ in size but grow equally fast (Fig. 1E, two
black dots on vertical line). The same trends were observed for the ppGpp0 strain with
RelA* and Mesh1 (Supplementary Fig 2B-C). The data are thus inconsistent with models in which ppGpp is a regulator of growth,
and in turn, growth sets cell size, as described by the general growth law [1,17–19]. In
such hierarchical models, LB would follow the increase-optimum-decrease trend observed
for µ (Fig. 1C). Instead, the monotonic increase of LB with increasing ppGpp (Fig. 1D)
suggested that ppGpp affects LB in a way that is not mediated by µ. If ppGpp indeed
modulates the size of cells in a µ-independent manner, we surmised it should play a role
in the adder mechanism, which maintains the cell size constant against stochastic
variations in birth size. ppGpp dynamically controls added cell size
In order to investigate the effect of ppGpp on the added size, we quantified the
added length (DL) each cell cycle. First, we found that the adder principle was obeyed at
all ppGpp concentrations: for the different levels of Dox and IPTG induction, DL was birth-
size independent (Fig. 2A). In line with previous adder principle observations, we find that
Tcyc rather than µ is modulated to achieve a constant DL, as larger-born cells divide sooner
(Fig. 2B). Indeed, DL increased monotonically with decreasing ppGpp (Supplementary
Fig. 2D), and thus did not follow the trend observed for µ (Fig. 1C). These data indicated
4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. that the added size correlated with population-average ppGpp concentration rather than
the rate of growth. Next, we considered how shifts in ppGpp concentration affect cell size and its
control dynamically. Within the microfluidic flow-cell, we followed individual cells as they
were exposed to a shift from basal ppGpp concentrations (-Dox) to different levels of
RelA* induction (+Dox), in various growth media. First, the growth response underscored
the important differences between the moderate induction that we focus on here (1 and
2 ng/ml) and the strong induction that mimics the stringent response (10 ng/ml Dox), with
strong induction yielding a rapid growth arrest while moderate induction led to an
approximately two-fold slower decrease in growth rate and allowed exponential growth to
continue (Fig. |
2C). Notably however, the added size did respond rapidly even to low RelA* induction
levels. Added size DL decreased halfway at about 25 min., and subsequently reached its
final value at about 55 min (Fig. 2D, red trace). In contrast, µ responded substantially
slower, and required about 100 min. to decrease halfway and over 300 min. to stabilize
(Fig. 2C, blue trace). A similar pattern of rapid DL and slow µ responses were observed
for different media and Dox induction levels (Fig. 2E), as well as for Mesh1-CFP induction
(Fig. 2F). Consistently, DL increased more rapidly than µ decreased upon Mesh1-CFP
induction (Fig. 2F). The data show a temporal order in which DL responds to ppGpp deviations prior
to µ. Indeed, DL typically has already stabilized to its post-shift level when µ has
decreased only half-way from pre-induction to post-induction level (Fig. 2D-F). These data
support the idea that µ and DL are decoupled, and further extend the notion that ppGpp
affects cell size independently from its role in growth rate control. We hypothesize that a
change in ppGpp concentration sets a new added size by affecting the division rate, which
in turn results in a new steady-state birth size. Division accelerates transiently to achieve constant added size
We used mathematical modeling to understand the interplay between growth,
division, and size, and compare different possible scenarios. Current size homeostasis
models consider constant growth conditions, as well as a strict coupling between μ and
5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ΔL [2,3,20,21]. We first tested a hierarchical model (Fig. 3A), which thus preserves the μ-
ΔL coupling as growth conditions change. This model takes the observed initial and final
values of μ and ΔL as input, lets μ decrease exponentially with the observed rate, and
averages over multiple resulting simulated stochastic trajectories in ΔL and division rate
1/Tcyc. The results of the hierarchical model (Fig. 3A) appeared inconsistent with the
experimental observations (Fig. 3B). In particular, DL and µ decrease at similar rates in
this model, while LB decreases slower than µ (Fig. 3A), while the experiments show that
both DL and LB decrease faster than µ (Fig. 3B). Notably, the experimental data also
indicate a transient increase in 1/Tcyc before it decreases (Fig. 3B), unlike the hierarchical
control model (Fig. 3A). Next, we considered a model of direct ppGpp control (Fig. 3C), which allows μ and
ΔL to respond to ppGpp changes via independent routes and timescales. In this model,
as also observed in the data (Fig. 2D), the mean ΔL responds directly by decreasing
linearly to its post-shift value in about two pre-shift cell-cycles (note that division events
in the averaged lineages are not synchronized), while μ decreases in the same way as in
the previous model (Supp. |
Fig. 3A). The direct control model reproduces many features
of the experimental data. Specifically, LB responds slower than ΔL but faster than µ, and
surprisingly, 1/Tcyc showed a transient increase (Fig. 3B and C). This increase in 1/Tcyc may appear paradoxical, as it must decrease ultimately to
match the lower μ. However, as illustrated by the model, it is a logical consequence of the
μ and ΔL speed differences: ΔL reaches the lower post-shift value rapidly, while μ is still
close to its high pre-shift value. With a comparatively high μ, the post-shift ΔL is achieved
early, and hence divisions are early as well. Yet, it is indeed notable that the division rate
appears readily modulated upwards and downwards. Together with the observation that
ΔL is already stable at its new value while µ and 1/Tcyc are still varying towards theirs (Fig. 3B), these findings support a picture in which the cells act to realize their target ΔL (set in
part by the concentration of ppGpp), and consequently LB, regardless of the rate of
growth, and thus yielding a corresponding division rate. Further observations are consistent with this picture. First, 1/Tcyc changes on a
timescale (within 25 minutes of the shift) that is shorter than the cell cycle duration, or
even the C+D period (~65-75 minutes) [22]. Scenarios in which division occurs a fixed
6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. time after the moment of replication initiation [23–25], with ppGpp affecting the added size
by modulating the initiation rate, are inconsistent with these observations. These
scenarios require completion of ongoing cell cycles before registering ppGpp effects. The
1/Tcyc response would then exceed the C+D period and decrease exclusively, to match
the new μ. Second, the growth rate independent model predicts the 1/Tcyc increase is too
small to detect in minimal media (Fig. 3D and E), owing to the lower µ, which is indeed
observed in the experiments (Fig. 3F). Third, upon Mesh1-CFP induction, Tcyc increases
within 20 minutes and then stays constant (Fig. S3B). These results indicate that ppGpp
can exert control over division, and hence over cell size, in a way that is not mediated
hierarchically through its effects on initiation and the rate of growth. Discussion
Elucidating the coordination between cell growth and cell cycle progression is a
foundational challenge of microbial physiology. The adder mechanism has emerged as a
key principle in recent years: it is observed across diverse domains of life and
experimental conditions, and elegantly explains how size remains constant despite
(division) stochasticity and exponential volume growth [2–4,15,23–27]. By adding a
constant size every cycle, stochastic size variations are averaged out without needing a
specific response to size deviations. |
How the adder mechanism relates to ppGpp is
central, given the inherently intertwined nature of size and growth, yet remains poorly
understood. Indeed, ppGpp is implicated in diverse cellular metabolism and growth
processes, including regulating ribosome production [28–30], modulating membrane
synthesis [31,32] DNA replication initiation [5,33–35], and triggering the stringent
response [36–38], a stress reaction that arrests growth until conditions improve. Here we found that ppGpp is a cell division regulator, and hence serves as a link
between growth and size control mechanisms. More specifically, we showed that E. coli
cells do not follow a hierarchical model, in which cell size adjusts to the growth rate (as in
the general growth law [1,17–19], and ppGpp controls growth (by tuning ribosome
production depending on amino acid availability, for instance). Rather, the (added) size
correlates with the level of ppGpp (instead of the growth rate), and adjusts rapidly to
7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ppGpp deviations, prior to the growth rate response, indicating that ppGpp exerts
independent control over cell division and growth. The findings lead to a number of speculations and implications. ppGpp is known
to reflect diverse signals, including nutritional conditions, stress, biosynthetic activity, and
metabolite availability [39,40]. The proposed mechanism thus allows division control to
integrate a wide spectrum of relevant growth factors. Indeed, these factors may influence
division through ppGpp rather than in direct manner, given that division was here found
to respond to ppGpp changes alone, under constant nutritional conditions, and before
growth itself adjusted. In addition, one may speculate that ppGpp-mediated cell cycle
control helps to accommodate different physiological limitations. For instance, a ribosome
excess may incur a growth cost while also favoring larger cells, rather than the smaller
cell size that would be imposed by the strictly positive size-growth correlation in
hierarchical models. Consistently, overexpression of a non-functional protein was shown
to yield larger cells [41]. The growth adjustments caused by the small ppGpp changes
that are the central focus here were also notably slower than those that characterize the
stringent response and large ppGpp changes. The former could be caused by ppGpp
inhibition of ribosome production, and the resulting dilution of ribosomes by volume
growth. In recent years, diverse mechanisms have been proposed to explain cell size
homeostasis and its dependence on growth [2–4,15,23–27]. Our findings suggest these
mechanisms are under the control of ppGpp. For instance, the constant added size is
proposed to result from the accumulation of a signaling molecule throughout the cell
cycle, which triggers division when a threshold is exceeded [24,42]. |
One may speculate
that ppGpp alters the production of this molecule and its threshold. Owing to the central
role of ppGpp in metabolism and its many regulatory targets (hundreds of genes [43] and
dozens of proteins [44]), ppGpp could control size in many possible ways. It was found
that OpgH can suppress FtsZ ring formation depending on the growth rate [45]. One may
consider whether OpgH mediates the ppGpp division effects, though it is not a known
ppGpp interactor [46]. Nucleoid occlusion mechanisms have also been proposed to
explain variations in added size [2]. Nucleoid volume could be modulated by ppGpp-
8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. induced decreases in the overall DNA replication initiation [47–49] or transcription rates
[50,51]. In conclusion, our findings establish a link between ppGpp, the central signaling
molecule in bacterial growth, and cell size homeostasis, which has implications for the
diverse cellular components and processes that are involved in these two pivotal cellular
control mechanisms. Acknowledgements
We thank Rebecca McKinzie, Martijn Wehrens, Nicole Imholz and Marek Noga for
experimental assistance and helpful discussions throughout the project and Daan J. Kiviet
for sharing his mold for the microfluidic device. Author contributions
G.B., S.J.T., and F. B. conceived the experiments, F.B. performed the experiments, F. B.,
M.C.L. and J.G. performed the modelling, all authors contributed to the writing of the
manuscript. Declaration of interest
The authors declare no competing interests. 9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figures & Legends
Figure 1. ppGpp exerts cell size control. (A) Scheme to control the cellular ppGpp
concentration. A RelA truncate from E. coli (relA*), which synthesizes ppGpp, is fused to
YFP and induced by Dox. The ppGpp hydrolysis enzyme Mesh1 is fused to CFP and
induced by IPTG. See Fig. S1 and S2 for characterization of these constructs. (B)
Measured length of single cells grown in a microfluidic device. For each cell cycle we
quantify: size (length) at birth (LB), cell cycle duration (Tcyc), added size (DL), and the
growth rate (μ) by exponential fitting. (C) Growth rate for decreasing ppGpp levels. ppGpp
increases above basal levels when inducing relA* with Dox and decreases below when
inducing mesh1-CFP by IPTG, in WT (relA+, spot+) cells. Left to right: N = 42, 316, 257,
403, 241 cell cycles. μ peaks at basal (endogenous) ppGpp levels, and then decreases. |
(D) Birth size for decreasing ppGpp levels. Conditions as in panel C. LB increases
continuously while μ decreases for below-basal ppGpp levels. The LB thus follows the
ppGpp trend rather than that of μ. (E) Birth length against growth rate. Closed circles:
10
1
2
3
4
5
6
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. single cell cycles in minimal media, colors and conditions as panels C and D. Drawn lines
are guides to the eye. For below-basal ppGpp, slower growing cells are larger, owing to
an inversion of the growth law. Cells of different size can thus have the same growth rate
(black dots). Open circles: single cell cycles in rich media, with 2 ng/ml, 1 ng/ml, or 0
ng/ml Dox, and N = 66, 35, 336 cell cycles (dark to light gray). 11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 2. ppGpp dynamically controls added cell size. (A) Cell length added (ΔL) per
cell cycle against birth length (LB) of that cycle, for different constant ppGpp levels. Dots
are single cell cycles, squares are means for LB bin, bars are s.e. Left to right: clouds for
decreasing ppGpp levels of ppGpp, starting with Dox in ng/ml: 2 (red), 1 (orange), 0
(yellow), and then 100 uM IPTG (blue). N = 42, 316, 257, 241 cell cycles. For each cloud,
ΔL is constant for different LB, consistent with “adder” principle. (B) Cell cycle duration
(Tcyc) against birth length (LB) of that cycle. When LB is smaller, Tcyc is larger, indicating it
is modulated as cells compensate for stochastic variations in LB, which is consistent with
the adder principle. Colors and conditions as in panel A. (C – F) Cells during ppGpp shift,
from basal to above and below basal levels. (C) Growth rate (μ) during ppGpp increases,
by RelA* induction with Dox, in rich and minimal media. μ response timescale for these
various conditions is assessed by normalizing rate to initial (pre-shift) and final (post-shift)
value. Top bar indicates Dox induction. In minimal media, shift from 0 ng/ml Dox to: 2
(pink), and 1 (orange). In rich media, shift from 0 ng/ml to: 1 (dark gray), 2 (intermediate
12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. gray), and 10 (light gray). Bars are s.e.m., averaged over multiple cell cycles. Curves
show similar adaptation time for moderate shifts, in contrast to the large 10 ng/ml shift,
which is significantly faster, and serves as a control: growth indeed arrests at high ppGpp
levels, as studied before, in line with the stringent response. |
(D) Growth (μ) and added
size (DL) during ppGpp increase. Circles are single cell cycles, lines are moving averages,
for shift from 0 to 2 ng/ml Dox, in rich medium. DL responds faster than μ and stabilizes
to post-shift value while μ still varies. (E) Stabilization time for μ (blue) and DL(red). See
methods for quantification approach. Each black dot shows the results of the analysis
from 1/3rd of the available data points. DL responds faster than μ under all moderate shifts. (F) Growth (μ) and added size (DL) during ppGpp decrease. Circles are single cell cycles,
lines are moving averages, for shift from 0 to 100 μM IPTG Dox, in minimal medium. 13
1
2
3
4
5
6
7
8
9
10
11
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 3 Division accelerates transiently to achieve constant added size. (A)
Predictions of hierarchical model, in which ppGpp affects the growth rate (μ), and size in
turn adjusts to the growth rate. To compare response speeds, indicated quantities are
normalized to initial (pre-shift) and final (post-shift) values. Top bar indicates ppGpp
increase. (B) Experimental data. Conditions and display as in panel A. (C) Predictions of
direct model, in which ppGpp exerts control over division and hence (added) size
independently of μ. Conditions and display as in panel A. Experiments agree with direct
model in terms of the temporal order of the responses and the transient acceleration of
divisions. (D) Quantification of transient effects in the division frequency (1/Tcyc). Data is
averaged in purple and green zones to obtain data in panels E and F. Circles are single
14
1
2
3
4
5
6
7
8
9
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. cell cycles for shift from 0 to 2 ng/ml Dox, in minimal medium. Drawn line is moving
average. (E) Pre- and post-shift division frequency (1/Tcyc), as defined in panel D. In rich
media, a ~15% increase is observed between the blue and green zones. Star: p < 0.01. (N = 336, 186, 132, 64, 257, 191, 355, 214 left to right) (F) Direct model reproduces
transient acceleration in rich media but not in minimal media, as seen in the experiments
(N = 300, randomly taken from the simulations before induction, and N = 150 after
induction). 15
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. STAR Methods
Strains and plasmids
E. coli K-12 strains NCM3722 (CGSC# 12355) and ppGpp0 (CF10237) which were
transformed with a combination of pRelA*-YFP, pMesh1-CFP or pCFP induction plasmids
were used in all the experiments as described in the figures. |
Plasmid pRelA*-YFP was
constructed by replacing mCherry on a pBbS2k-RFP plasmid (kanamycin resistance, Tet
promoter, sc101** origin) with a DNA sequence encoding the first 455 amino acids of the
native RelA gene. YFP fluorophore mVenus was fused to RelA* via a glycine-serine linker
using restriction cloning. Similarly, a codon optimized sequence of Mesh1 or CFP
replaced mCherry in the plasmid pBbA5a-RFP (ampicillin resistance, lacUV5 promoter,
p15A origin) leading to pMesh1 and pCFP. pMesh1-CFP was built similarly to pRelA*-
YFP using restriction cloning to fuse CFP with Mesh1 via a glycine-serine linker. Chemocompotent NCM3722 cells were transformed with the pRelA*-YFP plasmid and
spread on LB Agar plates with 25 ug/ml kanamycin. pMesh1-CFP and pCFP plasmids
were transformed and plated on LB Agar plates with 50 ug/ml ampicillin. Plates older than
3 weeks were discarded and fresh transformations were prepared to prevent possible
mutants. Chemocompotent ppGpp0 cells were co-transformed with pRelA*-YFP and pMesh1-CFP
and plated on LB Agar plates with 50 μg/ml ampicillin, 25 μg/ml kanamycin and 100 μM
IPTG to allow growth. Without 100 μM IPTG, leakage from pRelA*-YFP inhibits growth
enough to prevent visible colonies next day morning (data not shown). Plates with
colonies were only used on the day where the colonies first appear (next morning after
transformation) as older plates loose viability rapidly. Culture conditions
16
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Cells for bulk and microscopy experiments were grown using defined MOPS medium
containing 0.2% (with volume) carbon source [52] with 100 μM MnCl2 (minimal media). Rich medium is the same as minimal media except for the supplementation of 0.2%
Casamino acids, 400 μg/ml serine and 40 μg/ml tryptophan. 50 μg/ml ampicillin and 25
μg/ml kanamycin were added to the media along with appropriate plasmid bearing strains. For the microscopy experiments, cells were initially inoculated in 10 mL tubes with 5 ml
MOPS rich medium from a single colony in the morning. The tube was placed in a 37°C
room on an orbital shaker until the growth becomes visible (OD ≈ 0.1). Cells were then
spun down at 4000 G for 5 minutes and re-inoculated in 10 µL top media. 2 µL of the
concentrated cells were injected into the microfluidic chip by hand via a p2.5 pipette and
appropriate pipette tip. ppGpp0 strain requires different handling due to its inability to respond to stress. For the
96 well plate experiments, a single colony was inoculated in 5ml Glucose rich media with
100 μM IPTG and placed on a shaker for up to 4 hours until OD600 reaches 0.4. After that
40 ng/ml Dox is added to prime the culture with high ppGpp production which allows it to
handle stress and initiate growth in minimal media. |
This culture is then diluted in fresh
media (MOPS minimal with different carbon sources) without any inducers. Immediately
after the dilution, 98 μL of the culture is pipetted into wells of 96 well plates together with
1 μL Dox and 1 μL IPTG (both at 100x final concentration), reaching a final volume of 100
μL. 96 well plates which were prepared as above were placed in a BioTek Synergy HTX
plate reader @37°C with constant orbital shaking. OD was measured every 10 minutes. A similar method is applied for the microscopy experiments with the ppGpp0 strain. A
single colony is grown in rich media with proper antibiotics from the morning and when
OD reached 0.1, 40 ng/ml Dox is added to induce ppGpp production, which promotes cell
survival during chip loading. The chip is then placed under the objective in a warm chamber set to 37°C for all the
experiments. After cells populate the growth chamber input and output tubes are
connected to the chip and appropriate media is pushed through at 500 μl/hr. This
17
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. corresponds to 250-500x dilution per hour since the chip’s inner volume is between 1-2
μL. Microfluidic flow cell
The microfluidic chip’s Epoxy mold which was kindly sent by Daan J. Kiviet from
Ackermann Lab is a variant of the mothermachine from Jun lab. Each flow line consists
of an input which splits up into 2 arms, in each arm there are a number of extruding growth
chambers varying in depth and width (80, 60, 40, 20, 10, 5 µm width, 60, 30, 50, 40 µm
depth) and a single output after the 2 arms reconnect into a single line. Chips were built by first preparing the PDMS mix using the protocol from the reference
[15]. Polymer and curing agent (Sylgard 184 elastomer, Dow Corning) were prepared by
mixing 7.7 g of polymer with 1mL of curing agent. The slight deviation from the suggested
10 g per 1 mL was implemented to create a more rigid chip allowing low height growth
chambers to remain intact. Mixture was then thoroughly mixed using a vortexer and a
plastic mixer. Then the mix was poured into the Epoxy mold (provided by Ackermann lab)
and placed in a desiccator for 30 minutes to remove air bubbles formed during mixing. Then the mold is baked at 80°C for 1hr. After the baking period, the PDMS chip was
removed from the mold using scalpels and rough edges were cut to allow for better
binding to glass. Inlet and outlet holes were punched using a hole puncher. The PDMS
chip was then covalently attached to a glass slide by using a hand-help corona treatment
device (model BD-20ACV, Electro-Technic Products). Application was done by passing
the corona treatment device 6-7 times, each pass lasting ~5 seconds, 5-10 mm away
from the surface of both the PDMS chip and glass cover slip. |
After the corona application,
PDMS chips were placed on the treated glass surface and tapped by a gloved finger to
assure full bonding. Prepared chips could be used couple weeks after preparation
however after more than a month, chambers start to collapse, therefore chips older than
1 month were not often used. Imaging and Image Analysis
18
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Cells growing in the microfluidic chambers were imaged using an inverted Nikon TE2000. Using 100x and a 1.5x zoom lenses in tandem a pixel size of 0.041 µm was achieved. Imaging was done using a CMOS camera (Hamamatsu Orca Flash 4.0) via illumination
from an LED light source (ImSpec, HPX-L5) with a liquid light guide. The microscope
stance was equipped with a computer-controlled stage (Marzhauser, SCAN IM 120 3 100)
allowing the stage to move between several chambers for imaging. A phase contrast
image was taken every minute and a fluorescence image every 5 minutes using Chroma
filter set 49003 and a computer-controlled shutter (Sutter, Lambda 10-3 with
SmartShutter). Control of the automated microscopy systems was achieved through
MetaMorph software. Each experiment lasts between 24-36 hours. Images were initially visually checked for issues such as cells washing away from the
wells or halting of growth due to clogs. After the initial checks, a MatLab based software
customized by the Tans Lab was used to quantify growth rates and cell sizes. Individual
cells were identified and tracked from phase contrast images. Cell’s lengths and volumes
were estimated assuming the shape of cylinders with semicircular caps and fitting a
polynomial to skeletons of binary cell masks. Estimated length data through time was
used to calculate instantaneous growth rate and average growth rate using exponential
fits. Fluorescence values were calculated from a strip inside the cell area to decrease
errors caused by fluorescence falloff that occurs at the edges of the cell. Added length
was calculated by subtracting length at division (end of cycle) from length at birth
(beginning of cycle). Duration of the cycles were calculated as the time between birth and
division. Any cell that did not divide within the growth chamber was ignored along with
cells that approached the exits of the wells due to tracking issues caused by increased
cell speeds near the exit. Calculation of stabilization time
Data points within a sliding window of width ~Tcyc were compared against the data points
where growth rate became stable after induction. As the sliding window moves through
19
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. |
The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. time, both growth rate and added length decrease and approach the final stable value. When the t-test between the data points in the sliding window and in the final stable
growth regime yielded a p-value that exceeded 0.05 in three consecutive sliding time
steps, the center of the window defined the “stabilization time”. For each experiment the
data was randomly split into 3 parts and each part was analyzed using the same method
(black dots, Figure 2D). Division frequency analysis
Box plots (Figure 3D-F) were determined for measured values of 60/Tcyc, in a period
between 150 min. before induction and the moment of induction (green), and in a period
between the moment of induction until two times the average pre-shift cell cycle time
(blue). Models
We considered two alternative models describing size control during the transient. In both
models a cell divides with a hazard rate [21] that depends on the added size and a target
added size ΔL (which in stationary conditions corresponds to the average added size). Both the target added size ΔL(t) and the growth rate µ(t) are functions of time during the
transient. In both the models the growth rate µ(t) is exponentially relaxing to the stationary
value observed in each of the experiments. The two models differ for the relation between growth rate and size scale during transient. In the hierarchical model, the typical size is a deterministic function of the growth rate
(ΔL(t) = D*exp(µ(t)*T). Parameters were as follows: T = log(ΔLFinal/ΔLInitial)/(µFinal - µInitial)
and D = ΔLInitial*exp(-µInitial*T). In the direct model, ΔL(t) is relaxing to the stationary value
linearly in two pre-shift cycle duration time. 20
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Supplemental Information
Supplementary Figure 1: Expression, growth, and size characterization. (A-D)
Histograms of cellular fluorescence for different conditions. For all indicated induction and
growth conditions, fluorescence reporter distributions show that the cellular populations
are induced homogeneously. Cells were grown in either rich (with amino acids) or minimal
(without amino acids) MOPS-Glucose media. Fluorescence values of individual cells were
measured before or after induction of RelA*-YFP with 1 or 2 ng/ml Dox. (E-F) Cellular
growth rates and birth sizes for different strains and conditions. The strains carry one or
two plasmids that encode: RelA*-YFP, inducible by Dox (a-b), RelA*-YFP, inducible by
Dox, and CFP, inducible by IPTG (c) or Mesh1-CFP, inducible by IPTG (d). The cells
were grown in MOPS minimal media in a microfluidic device. |
Cells were characterized
before and after RelA*-YFP or Mesh1-CFP induction, in terms of their growth rates (panel
E) and birth lengths (panel F). For strain c, CFP was induced always with 200 μM IPTG;
21
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. both before and after RelA*-YFP induction. The results show minimal growth rate
changes caused by CFP expression only (a,b vs c, left). After the induction of RelA*-YFP
or Mesh1-CFP, all the strains grow slower, as discussed in the main text. Induction of
Mesh1-CFP with 100 μM IPTG leads to an increase of cell size, which is not observed for
CFP induction even at 200 μM IPTG (d, right vs c, left, respectively), indicating the size
effect is caused by Mesh1-CFP rather than CFP. 2 ng Dox induction decreases the growth
rate to ~0.25 db/hr (a, right). Consistently, 1 ng Dox induction leads to a higher final growth
rate that is similar for b and c. Red dotted horizontal line shows the average of 4
experiments before induction of RelA*-YFP or Mesh1-CFP. Number of cell cycles
measured: N=257, 264, 355, 403, 150, 316, 389, 241 left to right (panels E and F). 22
1
2
3
4
5
6
7
8
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Supplementary Figure 2: Dependence of cell growth and size on RelA* and Mesh1. (A) Optical density (OD) measurements (range 0-0.5 AU) in time (range 0-900 min) for a
ppGpp0 (DRelA, DSpoT) strain that is co-transformed with both the pRelA*-YFP (ppGpp
synthesis) and pMesh1-CFP (ppGpp hydrolysis) plasmids, grown in minimal media with
different carbon sources (black-glucose, green-glycerol, red-malate). Without Mesh1-
CFP induction by IPTG (bottom row) no growth is detected. This is consistent with
expression from the pRelA*-YFP plasmid being lethal, due to the resulting accumulation
23
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. of active ppGpp. Similarly, if Mesh1-CFP is maximally induced with IPTG (maximal
hydrolysis) while RelA*-YFP is not induced with Dox (minimal synthesis), growth is
minimal (top right plots), as the resulting lack of ppGpp is detrimental to growth in minimal
media. Growth effects caused by induction of one of the enzymes can be compensated
by induction of the other enzyme, as shown by the diagonal boxes (top-left to bottom-
right) displaying the fastest growth. |
This result is consistent with RelA*-YFP synthesizing
ppGpp, and Mesh1-CFP hydrolyzing it. (B) Growth rates of single cells against ppGpp
level, as quantified by YFP/CFP, for a ppGpp0 strain with pRelA*-YFP and pMesh1-CFP
plasmids. Variation along YFP/CFP axis indicates stochasticity of RelA*-YFP and Mesh1-
CFP expression. Growth shows peak for increasing YFP/CFP, while the (birth) size
decreases systematically (panel C). These findings support similar findings in main text
Fig. 1C-E. (C) Birth size of single cells against ppGpp level, as quantified by YFP/CFP. At low ppGpp (low YFP/CFP), cells grow slower yet are larger. (D) DL increases
systematically with decreasing ppGpp levels (in similar vein as the birth size in Fig 1D),
rather than with the growth rate, which peaks with ppGpp (Fig. 1C). Experimental details:
see Figure 1 C-D.
24
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Supplemental Figure 3. Cell cycle parameters during ppGpp shifts of different size
in different media. (A) Normalized cellular growth rate and added size during ppGpp
shifts. A strain carrying RelA*-YFP was grown in rich (with amino acids) or minimal
(without amino acids) MOPS-Glucose media. Induction of RelA*-YFP with 2 or 1 ng/ml
Dox occurs at T = 0 mins (vertical dashed line). Indicated are moving averages of the
growth rate (blue curve) and added length (red curve) from experiments, and computed
added lengths from two models (solid black curve: direct model, dashed black curve:
hierarchical model). The objective here is to assess the speed of adjustment from initial
to final values. Hence, the data sets were normalized to all start at the same position on
the Y-axis, and to also end at the same position (Initial and Final, respectively). Horizontal
dashed lines are half-way between initial and final values. Hierarchical adder model does
not follow the experimentally observed ΔL trajectory (red curve), while the direct adder
model does show a close match. (B) Cell cycle duration during ppGpp down-shift. Induction of Mesh1-CFP (T=0 mins, vertical dashed line) leads to rapid increase in Tcyc
suggesting inhibition of division by decreased ppGpp levels. 25
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. References
1. Schaechter, M., MaalOe, O., and Kjeldgaard, N.O. (1958). Dependency on Medium and
Temperature of Cell Size and Chemical Composition during Balanced Growth of
Salmonella typhimurium. J. Gen. Microbiol. 19, 592–606. |
2. Campos, M., Surovtsev, I. V., Kato, S., Paintdakhi, A., Beltran, B., Ebmeier, S.E., and
Jacobs-Wagner, C. (2014). A constant size extension drives bacterial cell size
homeostasis. Cell 159, 1433–1446. Available at:
http://dx.doi.org/10.1016/j.cell.2014.11.022. 3. Taheri-Araghi, S., Bradde, S., Sauls, J.T., Hill, N.S., Levin, P.A., Paulsson, J.,
Vergassola, M., and Jun, S. (2015). Cell-size control and homeostasis in bacteria. Curr. Biol. 25, 385–391. Available at: http://dx.doi.org/10.1016/j.cub.2014.12.009. 4. Sauls, J.T., Li, D., and Jun, S. (2016). Adder and a coarse-grained approach to cell size
homeostasis in bacteria. Curr. Opin. Cell Biol. 38, 38–44. Available at:
http://dx.doi.org/10.1016/j.ceb.2016.02.004. 5. Schreiber, G., Ron, E.Z., and Glaser, G. (1995). ppGpp-Mediated Regulation of DNA
Replication and Cell Division in. Curr. Microbiol. 30, 27–32. 6. Vadia, S., Tse, J.L., Lucena, R., Yang, Z., Kellogg, D.R., Wang, J.D., and Levin, P.A. (2017). Fatty Acid Availability Sets Cell Envelope Capacity and Dictates Microbial Cell
Size. Curr. Biol. 27, 1757–1767. 7. Magnusson, L.U., Gummesson, B., Joksimović, P., Farewell, A., and Nyström, T. (2007). Identical, independent, and opposing roles of ppGpp and DksA in Escherichia coli. J.
Bacteriol. 189, 5193–5202. 8. Xiao, H., Kalman, M., Ikehara, K., Zemel, S., Glaser, G., and Cashel, M. (1991). Residual
guanosine 3’ ,5’-bispyrophosphate synthetic activity of relA null mutants can be
eliminated by spoT null mutations. J. Biol. Chem. 266, 5980–5990. 9. Gropp, M., Strausz, Y., Gross, M., and Glaser, G. (2001). Regulation of Escherichia coli
RelA requires oligomerization of the C-terminal domain. J. Bacteriol. 183, 570–579. 10. Schreiber, G., Metzger, S., Aizenman, E., Roza, S., Cashel, M., and Glaser, G. (1991). Overexpression of the relA gene in Escherichia coli. J. Biol. Chem. 266, 3760–3767. 11. Sun, D., Lee, G., Lee, J.H., Kim, H.-Y., Rhee, H.-W., Park, S.-Y., Kim, K.-J., Kim, Y., Kim,
B.Y., Hong, J.-I., et al. (2010). A metazoan ortholog of SpoT hydrolyzes ppGpp and
functions in starvation responses. Nat. Struct. Mol. Biol. 17, 1188–94. Available at:
http://dx.doi.org/10.1038/nsmb.1906. 12. Zhu, M., and Dai, X. (2019). Growth suppression by altered (p)ppGpp levels results from
26
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. non-optimal resource allocation in Escherichia coli. Nucleic Acids Res. 47, 4684–4693. 13. Paul, B.J., Barker, M.M., Ross, W., Schneider, D.A., Webb, C., Foster, J.W., and Gourse,
R.L. (2004). DksA: A Critical Component of the Transcription Initiation Machinery that
Potentiates the Regulation of rRNA Promoters by ppGpp and the Initiating NTP. Cell 118,
311–322. 14. Paul, B.J., Berkmen, M.B., and Gourse, R.L. |
(2005). DksA potentiates direct activation of
amino acid promoters by ppGpp. Proc. Natl. Acad. Sci. U. S. A. 102, 7823–7828. 15. Wehrens, M., Ershov, D., Rozendaal, R., Walker, N., Schultz, D., Kishony, R., Levin,
P.A., and Tans, S.J. (2018). Size Laws and Division Ring Dynamics in Filamentous
Escherichia coli cells. Curr. Biol. 28, 972-979.e5. Available at:
https://doi.org/10.1016/j.cub.2018.02.006. 16. Kiviet, D.J., Nghe, P., Walker, N., Boulineau, S., Sunderlikova, V., and Tans, S.J. (2014). Stochasticity of metabolism and growth at the single-cell level. Nature 514, 376–9. Available at: http://dx.doi.org/10.1038/nature13582. 17. Pierucci, O. (1978). Dimensions of Escherichia coli at various growth rates: model for
envelope growth. J. Bacteriol. 135, 559–574. 18. Si, F., Li, D., Cox, S.E., Jun, Y., Si, F., Li, D., Cox, S.E., Sauls, J.T., Azizi, O., Sou, C., et
al. (2017). Invariance of Initiation Mass and Predictability of Cell Size in Escherichia coli. Curr Biol. 27, 1278–1287. 19. Vadia, S., and Levin, P.A. (2015). Growth rate and cell size: A re-examination of the
growth law. Curr. Opin. Microbiol. 24, 96–103. Available at:
http://dx.doi.org/10.1016/j.mib.2015.01.011. 20. Amir, A. (2014). Cell size regulation in bacteria. Phys. Rev. Lett. 112, 1–5. 21. Osella, M., Nugent, E., and Lagomarsino, M.C. (2014). Concerted control of Escherichia
coli cell division. Proc. Natl. Acad. Sci. U. S. A. 111, 3431–3435. 22. Bremer, H., and Dennis, P.P. (2008). Modulation of Chemical Composition and Other
Parameters of the Cell at Different Exponential Growth Rates. EcoSal Plus 3. 23. Witz, G., van Nimwegen, E., and Julou, T. (2019). Initiation of chromosome replication
controls both division and replication cycles in E. coli through a double-adder mechanism. bioRxiv, 593590. 24. Si, F., Le Treut, G., Sauls, J.T., Vadia, S., Levin, P.A., and Jun, S. (2019). Mechanistic
Origin of Cell-Size Control and Homeostasis in Bacteria. Curr. Biol. 29, 1760-1770.e7. 25. Wallden, M., Fange, D., Lundius, E.G., Baltekin, O., and Elf, J. (2016). The
Synchronization of Replication and Division Cycles in Individual E. coli Cells. Cell 166,
27
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 729–739. 26. Barber, F., Ho, P.Y., Murray, A.W., and Amir, A. (2017). Details matter: Noise and model
structure set the relationship between cell size and cell cycle timing. Front. Cell Dev. Biol. 5, 1–16. 27. Cadart, C., Monnier, S., Grilli, J., Sáez, P.J., Srivastava, N., Attia, R., Terriac, E., Baum,
B., Cosentino-Lagomarsino, M., and Piel, M. (2018). Size control in mammalian cells
involves modulation of both growth rate and cell cycle duration. Nat. Commun. 9. Available at: http://dx.doi.org/10.1038/s41467-018-05393-0. |
28. Molodtsov, V., Sineva, E., Zhang, L., Huang, X., Cashel, M., Ades, S.E., and Murakami,
K.S. (2018). Allosteric Effector ppGpp Potentiates the Inhibition of Transcript Initiation by
DksA. Mol. Cell 69, 828-839.e5. 29. Giordano, N., Mairet, F., Gouzé, J.-L., Geiselmann, J., and de Jong, H. (2016). Dynamical Allocation of Cellular Resources as an Optimal Control Problem: Novel
Insights into Microbial Growth Strategies. PLoS Comput. Biol. 12, e1004802. Available at:
http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1004802. 30. Burgos, H.L., O’Connor, K., Sanchez-Vazquez, P., and Gourse, R.L. (2017). Roles of
transcriptional and translational control mechanisms in regulation of ribosomal protein
synthesis in Escherichia coli. J. Bacteriol. 199, 1–14. 31. Noga, M.J., Büke, F., van den Broek, N.J., Imholz, N., Scherer, N., Bokinsky, G., Broek,
N.J. van den, Imholz, N., Scherer, N., and Bokinsky, G. (2019). Post-translational control
is sufficient to coordinate membrane synthesis with growth in Escherichia coli. bioRxiv,
728451. Available at:
https://www.biorxiv.org/content/10.1101/728451v2.abstract?%3Fcollection=. 32. Merlie, J.P., and Pizer, L.I. (1973). Regulation of phospholipid synthesis in Escherichia
coli by guanosine tetraphosphate. J. Bacteriol. 116, 355–366. 33. McAuley, S., Vadia, S., Jani, C., Huynh, A., Yang, Z., Levin, P.A., and Nodwell, J.R.
(2019). A Chemical Inhibitor of Cell Growth Reduces Cell Size in Bacillus subtilis. ACS
Chem. Biol. 14, 688–695. 34. Traxler, M.F., Summers, S.M., Nguyen, H.T., Zacharia, V.M., Hightower, G.A., Smith,
J.T., and Conway, T. (2008). The global, ppGpp-mediated stringent response to amino
acid starvation in Escherichia coli. Mol. Microbiol. 68, 1128–1148. 35. Liang, S.-T., Bipatnath, M., Xu, Y., Chen, S., Dennis, P., Ehrenberg, M., and Bremer, H.
(1999). Activities of constitutive promoters in Escherichia coli. J. Mol. Biol. 292, 19–37. 36. Jishage, M., Kvint, K., Shingler, V., and Nyström, T. (2002). Regulation of σ factor
28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. competition by the alarmone ppGpp. Genes Dev. 16, 1260–1270. 37. Ferullo, D.J., and Lovett, S.T. (2008). The stringent response and cell cycle arrest in
Escherichia coli. PLoS Genet. 4, 21–23. 38. Traxler, M.F., Summers, S.M., Nguyen, H.T., Zacharia, V.M., Hightower, G.A., Smith,
J.T., and Conway, T. (2008). The global, ppGpp-mediated stringent response to amino
acid starvation in Escherichia coli. Mol. Microbiol. 68, 1128–1148. 39. Ronneau, S., and Hallez, R. (2019). Make and break the alarmone: Regulation of
(p)ppGpp synthetase/hydrolase enzymes in bacteria. FEMS Microbiol. Rev. 43, 389–400. 40. Hauryliuk, V., Atkinson, G.C., Murakami, K.S., Tenson, T., and Gerdes, K. (2015). |
Recent
functional insights into the role of (p)ppGpp in bacterial physiology. Nat. Rev. Microbiol. 13, 298–309. Available at:
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=4659695&tool=pmcentrez&ren
dertype=abstract. 41. Basan, M., Zhu, M., Dai, X., Warren, M., Sévin, D., Wang, Y., and Hwa, T. (2015). Inflating bacterial cells by increased protein synthesis. Mol. Syst. Biol. 11, 836. 42. Harris, L.K., and Theriot, J.A. (2016). Relative rates of surface and volume synthesis set
bacterial cell size. Cell 165, 1479–1492. Available at:
http://dx.doi.org/10.1016/j.cell.2016.05.045. 43. Liu, K., Bittner, A.N., and Wang, J.D. (2015). Diversity in (p)ppGpp metabolism and
effectors. Curr. Opin. Microbiol. 24, 72–79. Available at:
http://dx.doi.org/10.1016/j.mib.2015.01.012. 44. Zhang, Y., Zborníková, E., Rejman, D., and Gerdes, K. (2018). Novel (p)ppGpp binding
and metabolizing proteins of Escherichia coli. MBio 9, 1–20. 45. Hill, N.S., Buske, P.J., Shi, Y., and Levin, P.A. (2013). A Moonlighting Enzyme Links
Escherichia coli Cell Size with Central Metabolism. PLoS Genet. 9. 46. Kanjee, U., Ogata, K., and Houry, W.A. (2012). Direct binding targets of the stringent
response alarmone (p)ppGpp. Mol. Microbiol. 85, 1029–1043. 47. Cooper, S., and Helmstetter, C.E. (1968). Chromosome replication and the division cycle
of Escherichia coli B r. J. Mol. Biol. 48. Kraemer, J.A., Sanderlin, A.G., and Laub, M.T. (2019). The Stringent Response Inhibits
DNA Replication Initiation in E. coli by Modulating Supercoiling of oriC . MBio 10, 1–18. 49. Fernández-Coll, L., Maciag-Dorszynska, M., Tailor, K., Vadia, S., Levin, P.A., Szalewska-
Palasz, A., and Cashel, M. (2020). The Absence of (p)ppGpp Renders Initiation of
Escherichia coli Chromosomal DNA Synthesis Independent of Growth Rates. MBio 11,
29
1
2
3
4
5
6
7
8
9
10
bioRxiv preprint
doi:
https://doi.org/10.1101/2020.06.16.154187
;
this version posted June 17, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1–14. Available at: http://mbio.asm.org/lookup/doi/10.1128/mBio.03223-19. 50. Nastaran H. Yazdi, Calin C. Guet, Reid C. Johnson, and J.F.M. (2013). Variation of the
folding and dynamics of the Escherichia coli chromosome with growth conditions. Mol
Microbiol. 86, 1318–1333. 51. Jin, D.J., and Cabrera, J.E. (2006). Coupling the distribution of RNA polymerase to global
gene regulation and the dynamic structure of the bacterial nucleoid in Escherichia coli. J.
Struct. Biol. 156, 284–291. 52. Neidhardt, F.C., Bloch, P.L., and Smith, D.F. (1974). Culture Medium for Enterobacteria. J. Bacteriol. 119, 736–747. 30 |
1
2
3
4 5 6 7 8 9 10 11
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Quantitative Mining of Compositional Heterogeneity in Cryo- EM Datasets of Ribosome Assembly Intermediates
Jessica N. Rabuck-Gibbons1, Dmitry Lyumkis1,2, James R. Williamson1
1. Department of Integrative Structural and Computational Biology, Department of
Chemistry, and The Skaggs Institute for Chemical Biology, The Scripps Research Institute, La Jolla, CA 92037, USA
2. Laboratory of Genetics and Helmsley Center for Genomic Medicine, The Salk
Institute for Biological Studies, La Jolla, CA 92037, USA
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Summary
Macromolecular complexes are dynamic entities whose function is often intertwined with their many structural configurations. Single particle cryo-electron microscopy (cryo-EM) offers a unique opportunity to characterize macromolecular structural heterogeneity by virtue of its ability to place distinct populations into different groups through computational classification. However, current workflows are limited, and there is a dearth of tools for surveying the heterogeneity landscape, quantitatively analyzing heterogeneous particle populations after classification, deciding how many unique classes are represented by the data, and accurately cross-comparing reconstructions. Here, we develop a workflow that contains discovery and analysis modules to quantitatively mine cryo-EM data for a set of structures with maximal diversity. This workflow was applied to a dataset of E. coli 50S ribosome assembly intermediates, which is characterized by significant structural heterogeneity. We identified new branch points in the assembly process and characterized the interactions of an assembly factor with immature intermediates. While the tools described here were developed for ribosome assembly, they should be broadly applicable to the analysis of other heterogeneous cryo-EM datasets. Keywords
Cryo-electron microscopy; Single particle analysis; Ribosome biogenesis; Heterogeneity analysis
1
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Introduction
Cryo-electron microscopy (cryo-EM) is a rapidly evolving, powerful technology for solving the structures of a wide variety of biological assemblies. |
The “resolution revolution” in cryo-EM (Kühlbrandt, 2014), caused in part by advances in direct electron detectors and improved data acquisition and analysis workflows, has led to high- resolution structural insights into a wide variety of biological processes performed by macromolecular assemblies (Fernandez-Leiro and Scheres, 2016). There have been steady, but consistent improvements to achievable resolution, and the collective tools are now enabling structure determination at true atomic resolution (Bartesaghi et al., 2015; Tan et al., 2018; Nakane et al., 2020; Yip et al., 2020; Zhang et al., 2020). There have also been numerous advances for analyzing structurally heterogeneous particle populations, and data processing software now routinely include strategies for handling distributions of structures that arise from compositional or conformational changes in the macromolecular species of interest (Elmlund and Elmlund, 2012; Gao et al., 2004; Klaholz, 2015; Liao, Hashem and Frank, 2015; Nakane et al., 2018; Scheres, 2016; Spahn and Penczek, 2009; Wang et al., 2013; White et al., 2017; Zhong et al., 2021; Grant, Rohou and Grigorieff, 2018; Lyumkis et al., 2013; Punjani and Fleet, 2021b; Punjani and Fleet, 2021a). However, most current cryo-EM workflows still focus on achieving the maximum possible resolution, which requires selecting and averaging potentially heterogeneous subsets of the data in the interest of increasing the particle count for the homogeneous regions of a map. This strategy comes at the expense of either eliminating particle populations that do not conform to the predominant species or neglecting dynamic and labile regions of reconstructed maps, which are often of biological interest. in workflows
Another challenge in cryo-EM heterogeneity analysis is that there is no way to define the number of distinct structures in a given dataset a priori. It is up to the researcher to employ a classification strategy and to heuristically determine the number of distinct classes. Furthermore, there is no set procedure to determine the threshold for examining map features and differences between maps. Thresholds are often set in a subjective manner in order to best display the features of interest in the maps, although an approach was recently described where a voxel-based false discovery rate could be determined to establish a noise threshold for contouring (Beckers, Jakobi and Sachse, 2019). Thus, determining the final number of classes in a dataset and quantitatively comparing a set of maps in order to tell a concise biological story with statistical significance remains a challenge. The process of ribosome assembly provides a useful case study for mining and quantitatively assessing structural heterogeneity in cryo-EM data. The bacterial 70S
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. |
The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ribosome is a complex macromolecular machine composed of three ribosomal RNAs (rRNAs) and ~50 ribosomal proteins (r-proteins) that form a large 50S subunit and a small 30S subunit. Ribosome assembly occurs within several minutes in vivo, and the process includes transcription and translation of the rRNAs and r-proteins, folding of the rRNA and r-proteins, and docking of the r-proteins on the rRNA scaffold. rRNA folding events and proper r-protein binding are facilitated by ~100 ribosome assembly factors. Given the efficiency and speed of the assembly process, structural intermediates are difficult to isolate and purify. However, perturbations in ribosome assembly lead to the accumulation of numerous structural intermediates, which collectively inform molecular mechanisms of ribosome assembly (Shajani, Sykes and Williamson, 2011; Stokes et al., 2014; Sashital et al., 2014; Sykes et al., 2010; Jomaa et al., 2014; Li et al., 2013; Ni et al., 2016; Davis et al., 2016; Rabuck-Gibbons et al., 2020). The major parts of the ribosome that are often present or missing in assembly intermediates are the central protuberance (CP), the L7/12 and L1 stalks, and the base (Figure 1A). We previously developed a genetic approach by which the amount of a given r-protein, in our case bL17, could be titrated by the addition of the small molecule homoserine lactone (HSL) (Davis et al., 2016). Limiting the amount of bL17 induced a roadblock in ribosome assembly, causing intermediates to accumulate. In the first work using the bL17-lim strain (Davis et al., 2016), we identified thirteen distinct structures that fell into four main structural classes (Figure 1B). Here, we will continue to use the nomenclature for the main classes used by Davis and Tan, et al. These categories, ordered least to most mature, are the B class which is missing the base, CP, and both stalks, the C class in which the base is formed, but the central protuberance (CP) is either misdocked or altogether missing, the D class in which the base of the 50S ribosome is missing, and the E class, which contains both the base and the CP, but has variability in the presence or absence of the stalks. Some of the “missing” regions (primarily rRNA, but they may also include r-proteins) described above are not present in the reconstructed maps but are in fact present in the sample and within individual particle images, meaning that they contribute to “biological noise”. This becomes relevant for some of the decisions that need to be made in the data analysis workflow, as will be discussed below. In previous work, the four main initial classes belonging to the 50S assembly intermediates (B, C, D, E) were each further subdivided by an additional round of subclassification, resulting in thirteen distinct structures. While several different subclassification schemes were attempted at that time using heuristics to determine the number of subclasses, no attempt was made to establish quantitative criteria by which the subclassification or coverage of relevant classes would be complete. |
While classes were identified belonging to the 30S and 70S (F class and A class in Davis et al., 2016), they are not explicitly described in our previous work or in the work described here. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. As our goal is to define broad trends in ribosome assembly through various perturbations, it is important to quantitatively assess differences between intermediates that accumulate under various specific conditions and to organize them into a ribosome assembly landscape (Davis et al., 2016; Bernstein et al., 2004; Harnpicharnchai et al., 2001; Jomaa et al., 2011; Loerke, Giesebrecht and Spahn, 2010; Nikolay et al., 2018; Razi, Guarné and Ortega, 2017; Uicker, Schaefer and Britton, 2006). To this end, we developed a data processing framework to analyze cryo-EM datasets methodically and quantitatively in order to assess the number of distinct structures, the significant differences among them, and to place these structures into a biological context. When we apply our complete workflow to a dataset of ribosome assembly intermediates from bL17-lim, we discover a total of forty-one different structures that are identifiable based on a defined set of cutoff parameters. These structures include several novel intermediates, such as the most immature assembly intermediate observed to date, and an independent pathway contingent on the binding of a ribosome assembly factor, as well as late-stage assembly intermediates. Together, these are organized into a revised assembly landscape for the 50S ribosomal subunit under bL17-lim conditions. Results and Discussion
An overview of the heterogeneity processing workflow There are two main phases in the framework for systematic analysis of heterogeneous ensembles of macromolecular conformations (Figure 2). The first phase is a discovery phase, which begins with iterative rounds of hierarchical classification and sub- classification using a defined set of thresholding parameters. The goal of this first phase to uncover the broad spectrum of distinct classes in a cryo-EM dataset, starting with traditional pre-processing and data cleaning steps (e.g. motion correction, particle picking, CTF estimation, and initial 2D and 3D classification). The initial data cleaning steps defined here are intended to be very lenient, such that the only particles removed from the dataset are clear artifacts or molecular species that are not of interest. For example, in the case of 50S ribosome assembly intermediate analysis, we remove particles that are obvious 30S or 70S ribosomes and proteasomes from the stack, but we do not remove any classes that could possibly be 50S assembly intermediates. |
After the cleaning steps, an iterative subclassification strategy is used to parse out molecular heterogeneity. After an initial round of classification, each class (class X) is subjected to a n=2 subclassification, resulting in two potential subclasses, X1 and X2. Both subclasses are processed and binarized, and then difference maps X1-X2 and X2- X1 are calculated, to determine if there is more heterogeneity that can be mined from each class X. If the difference volumes don’t reach a chosen molecular weight or resolution threshold, then the subclassification is rejected, and further subclassification
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. is terminated. If neither of these two criteria are reached, the binary subclassification process is iteratively repeated until one of the convergence criteria are met. The second phase is an analysis phase, which is intended to quantitatively define and distinguish structural features between maps, and further, to establish the number of structural states using a given set of quantitative cutoffs. During this hierarchical difference analysis, the full matrix of difference maps is calculated, and the molecular weights of the difference maps are used as a metric to cluster the classes, which can be visualized as a particle dendrogram. A line can be drawn through the dendrogram at a chosen molecular weight threshold, which identifies similar maps that can be combined. Next, to qualitatively differentiate between classes, the resulting set of maps are compared to a catalog of coarse-grained structural features that are calculated from a reference structure, in this case the bacterial 50S ribosome. It is convenient to use features such as rRNA helices and r-proteins, that may be present or absent in various classes. The presence of these coarse-grained reference features is quantitatively analyzed using hierarchical clustering to organize and visualize the patterns of variation among the final set of particle classes. For our dataset of bacterial ribosome assembly intermediates, these features are used to place the observed classes into a putative assembly pathway, based on a principle of parsimonious folding and unfolding. A divisive resolution-limited subclassification approach facilitates identifying novel species A major challenge in the analysis of heterogeneous datasets is the accurate identification of a broad diversity of structural states. To address this, we developed a classification strategy to mine an experimental cryo-EM dataset for distinct particle populations. Classification and refinement of particle classes can be undertaken using a variety of software packages, and we have adopted the latest version of FrealignX, whose code base is also implemented within cisTEM (Grant, Rohou and Grigorieff, 2018; Lyumkis et al., 2013). |
We note that most processing packages that are capable of classifying single-particle cryo-EM data can be employed for this purpose(Scheres, 2016; Nakane et al., 2018; Zhong et al., 2021; Punjani and Fleet, 2021b; Punjani and Fleet, 2021a). In typical cryo-EM workflows, 3D classification is performed several times, with different choices for the total number of classes (n). If n is too small, the resulting classes may have averaged properties leading to loss of structural diversity but potentially higher resolution in the homogeneous regions. If n is too large, the data is subdivided into nearly identical classes, but each class is characterized by lower resolution, because the particle count contributing to the class decreases. For the characterization of intrinsically heterogeneous datasets such as those encountered during ribosome
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. assembly, the goal of 3D classification is to capture the full range of structural diversity, as opposed to a select few well-resolved classes. Therefore, we developed an iterative subclassification strategy to systematically mine the data and identify distinct structural intermediates, including species that are rare and underpopulated. With the knowledge that our test dataset harbored at least thirteen intermediates (Davis et al., 2016), we started with n=10 in order to evaluate parameters for subclassification. The ten initial classes are shown in Figure 3A. While we expected that we would find the previous B, C, D and E classes in the dataset, the B-class was not present, and rather, multiple classes that are subtle variations of the E-class were present. This exemplifies one of the pitfalls of classification that we term “hiding”, where subclasses can be mixed, only to emerge at subsequent stages of subclassification. A survey of various classification parameters within FrealignX revealed the res_high_class parameter, which is the resolution of the data to be used for classification, ameliorated class hiding and had a strong effect on the classes that emerged. This parameter is typically set to just below the estimated resolution limit of the data. However, by setting res_high_class to 20Å, the gross class heterogeneity increased, and the expected B-class emerged (Figure 3B). The resolution threshold for classification is frequently defaulted and determined automatically during classification, but it may also be explicitly set by the user or limited to the resolution of the first Thon ring (Scheres, 2012; Scheres, 2016; Scheres et al., 2008). With the well-defined ribosome assembly case study, we show that a lower resolution threshold during classification helps to identify particle subsets that are substantially distinct from the predominant species. |
that
lowering
We also examined different iterative subclassification strategies, with various numbers of classes used for each stage of subclassification. In order to test the success of these strategies, we selected a final n of ~30, which was chosen because it provided a convenient number to evaluate a variety of subclassification schemes, and because it was close to twice the final number of classes found in the original bL17-lim dataset (Davis et al., 2016). The five classification schemes (Figure 3C) tested were: (1) a simple 1-round classification with n=30, (2) a 2-round hierarchical classification of 6 initial classes, each subdivided into 5 (n1=6,n2=5; total n=30,), (3) a 3-round hierarchical subclassification of 2 initial classes each subdivided into 3, with a second subdivision into 5 (n1=2,n2=3,n3=5; total n=30), (4) a 3-round hierarchical classification of 5 initial classes subdivided into 3, then subdivided into 2 n1=5,n2=3,n3=2; total n=30,), and (5) a 5-round hierarchical binary subclassification strategy, where 2 initial classes were subdivided into 2 until n=32 was reached (n1=2,n2=2,n3=2,n4=2,n5=2; total n=32,). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. With the sole exception of the simple single-round classification with n=30, all of these divisive schemes yielded new classes not previously identified (Supplemental Figure 1, indicated by *). Furthermore, the iterative divisive approaches produced the greatest range of structural diversity and avoided grouping together dissimilar classes. This observation is perhaps not unexpected, as it is well known that a divisive classification approach avoids local minima within the search space and is more robust than attempting to produce a final number of classes directly (Gray, 1984; Sorzano et al., 2010). Qualitatively, a first round of classification where n1 is on the order of the number of major classes works well, followed by smaller subdivisions. As an example, a three round subclassification scheme with [n1 = 5, n2 = 3, n3 = 2], for a total of 30 final classes, identified the greatest number of new structures, as shown in Supplemental Figure 1. For this reason, we proceeded with the n1 = 5, n2 = 3, n3 = 2 approach for our work, although we note that the optimal classification scheme will likely vary with the distinct heterogeneity spectrum for each unique dataset. Given that the observed classes are relatively independent of the details of the subclassification, we turned our attention to the criteria for termination of subclassification. Defining an endpoint for subclassification The determination of when subclassification is complete is a key question in cryo-EM analysis. |
Many times, classification is considered finished if a specific region of interest can be resolved to a satisfactory resolution, depending on what question(s) the user wishes to address. However, this subjective approach may be insufficient for the purpose of uncovering hidden features and discovering new structural states, especially if there are multiple datasets to be compared. To guide the analysis of our bL17-lim dataset, and to establish a protocol that can be used to analyze other data with statistical significance, our goal was to establish metrics by which we could confidently terminate the subclassification. We adopted a simple metric to determine the endpoint of subclassification. For any given class at any stage of subclassification, a test subclassification is performed with n=2. If the two resulting subclasses differ by less than a chosen noise threshold, or by less than a chosen molecular weight threshold, then subclassification is complete, and the subdivision is rejected. Conversely, if the thresholds are exceeded, the subclassification is retained, and the two resulting classes are iteratively subjected to additional subclassification until the termination thresholds are met (Figure 2). There are at least two types of noise that need to be considered in the difference analysis that are used to conclude subclassification. First, there is the intrinsic noise floor in the map that arises from averaging noisy image data during the reconstruction process. Second, there is biological noise, which can be broadly attributed to conformational and compositional heterogeneity, resulting in density above the intrinsic
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. noise floor that cannot be interpreted in terms of a structure or slight shifts of well- defined elements that may or may not be significant (Supplemental Figure 2). For example, in the case of ribosome assembly, there are portions of rRNA that are present in the sample, but do not resolve to a reasonable structure (Davis et al., 2016). To characterize a diverse set of classes, the goal is to identify significant differences that exceed chosen thresholds for these noise components. A three-step process was developed to remedy the above challenges, based on the estimation of the real space noise in a given map. First, a low-pass filter is used to reduce high-frequency information in the map (low-pass filter threshold, Table 1) Clearly, this is inadvisable if high resolution is the goal for the experiment, but for heterogeneity analysis, resolution is secondary to differentiating between broader conformational and compositional differences. Second, it is important that the soft spherical mask typically applied during classification is removed, and the standard deviation of the unmasked map (σmap) is calculated using standard cryo-EM analysis programs. |
While the signal from the macromolecular object is included in this calculation, that contribution to the standard deviation is negligible if the box size is sufficiently large, so that voxels containing true signal represents 1-2% of the total map volume. Effectively, σmap provides a crude estimate of the intrinsic map noise. There are several other ways to calculate a noise threshold, most recently the program developed by Beckers et al. (Beckers, Jakobi and Sachse, 2019) which uses a false discovery rate (FDR) to determine the threshold used for visualization and analysis, or one can use the noise sampled from the periphery of the map. The values of 3σmap are highly correlated to the contour levels based on FDR as shown in Figure S3 but the 3σmap threshold generally exceeds the FDR threshold, and is thus more conservative. Due to the prevalence of unresolved features in the ribosome data, we have used 3σmap as a convenient threshold to eliminate noise. Third, each map is then binarized using a 3σmap threshold such that intensities greater than 3σmap were set to 1, and intensities less than 3σmap were set to 0 (binarization threshold, Table 1). Other thresholds could be devised and implemented, as long as they are applied consistently across classes. These thresholded, binarized maps are used for the remainder of the analysis. Using these maps is advantageous because the “noise” from flexible regions is removed from the map, and there is a clear boundary of which parts of a structure are analyzed. Further, binarization facilitates coarse-grained analysis and eliminates the need for scaling. To define the endpoint to classification, the above filtering and binarization steps are applied after a test n=2 subclassification of a given class X into class X1 and X2. If either class X1 or class X2 do not have a resolvable map, as defined by the resolution limit (r-limit, Table 1), then the classification process is terminated. If the differences
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. between class X1 and class X2 are less than the volume limit (v-limit, Table 1), the subclassification is terminated. However, if the differences between class X1 and X2 are greater than the v-limit, then the subclassification is retained, and classes X1 and X2 are in turn further subdivided into 2 classes. This process then repeats on all classes until either the r-limit or the v-limit are reached. This set of limits provides a consistent and quantitative basis for iterative subclassification. Segmented difference analysis between map identifies the exact number of structural states Having discovered the structural variants in the data, we then asked how the different maps compare to one another and where/what are the major differences. |
To address this question, we developed a strategy to quantitatively assess similarities between the classes. While the classification approach in the discovery phase is designed to terminate once the structural features were no longer distinguishable using the r-limit or v-limit, this procedure does not guarantee that individual structures within the collective set of reconstructions are all distinct from one another. More specifically, a situation can arise where two similar classes emerge (from hiding) in different branches of the subclassification tree. In the first step, difference maps are calculated between all of the binarized, thresholded maps. Such difference maps are useful to identify regions of density that are distinct between classes, and in our case, provide both qualitative and quantitative insight into structural relationships between distinct assembly intermediates. Two specific examples for distinct “D-classes” are shown in Figure 4A-B. The first two columns display two distinct maps arising from some point during classification. The raw difference maps are shown in the third column (map1-map2, red; map2-map1, blue). The approximate molecular weight of these differences is also indicated. These difference maps are then segmented to remove “dust” that may arise from minor conformational or compositional variations between maps. This dust cannot be interpreted in biological terms at the target resolution but may add up to a significant molecular weight (Table 1 segmentation threshold, Figure 4). Such difference maps can be computed for all pairwise combinations of reconstructions arising from the classification procedure. The pairwise difference maps are useful for both qualitative and quantitative downstream analyses. To parse through structural differences, define an accurate final number of unique structural variants in the data, and combine particles contributing to similar maps, we employed a simple hierarchical clustering approach based on the positive/negative molecular weight differences between structures. Based on the clustering, it is possible to pare down the maps and combine particles from similar reconstructions, even if they arise from different starting points in the classification
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. (Figure 5A). At this stage, two classes can be combined if the molecular weight differences between the two classes are less than a given threshold. Since the branchpoints of the dendrogram provide a measure of molecular weight differences between maps, they can serve as a guide for analyzing the similarity between classes overall based on the nodes of the dendrogram (Figure 5B). |
In the example in Figure 5B, the dendrogram reveals that the leftmost structure is distinct from the other two and needs to be treated independently, whereas the latter two can be combined into a single class. Thus, although there are 42 distinct structures in Figure 5A, after hierarchical clustering analysis and the subsequent merging of similar maps, there are 41 distinct structures that will go forward in the analysis pathway. Collectively, these procedures enable us to identify the exact number of structural states within the data, given the limitations associated with identifying novel classes in the discovery phase and according to the established criteria in the analysis phase, defined above. Defining relationships between distinct structures An important step in analyzing differences between classes discovered within the above procedures for heterogenous cryo-EM data analysis is to define where differences between two maps are located. If a model (e.g. an atomic model or a cryo-EM structure) exists as a reference, and if the reconstructed maps differ primarily by compositional variation, then it is straightforward to use the model for interpreting the collective set of maps (Davis et al., 2016) in an “occupancy analysis.” In the case of bacterial ribosome assembly, we have a well-defined reference model (Figure 6A). This reference structure is broken into its individual r-RNA and r-protein parts, yielding theoretical cryo-EM densities for each component (Figure 6B). Such individual densities can then be directly compared to densities arising from experimental cryo-EM classification. It is important that the reference densities are generated at (approximately) the same resolution as the experimental densities arising from hierarchical clustering and difference analyses. The theoretical maps are then binarized, which enables comparing the theoretical maps to the binarized experimental maps arising from subclassification. Each binarized class (Figure 6C) is then compared to each theoretical feature map by counting overlapping voxels and normalizing to the theoretical volume, to define the fractional occupancy of the selected feature in the map that can be completely present (Figure 6D), partially present due to partial flexibility or a misdocked figure (Figure 6E), or completely missing (Figure 6F). The complete set of fractional occupancies are given as an n by m matrix of values between 0 and 1, where n describes the set of classes and m defines the number of features. The resulting fractional occupancies can be visualized as a heat map and subjected to hierarchical clustering to organize the classes and features (Figure 6G). Clustering along the feature (x-axis) groups elements (in this case, r-proteins and rRNAs), and
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. |
All rights reserved. No reuse allowed without permission. clustering along the map (y-axis) groups the maps according to their occupancy. As expected, the B, C, D, and E, maps cluster well together. The occupancy matrix facilitates the visualization of large blocks of structural features that co-vary across the particle classes, providing cooperative folding blocks (Figure 6H) (Davis et al., 2016). This procedure enables a quantitative comparison of distinct sets of maps that differ by compositional variants. We note that this procedure is not currently compatible with conformational variability or density that is not represented in the reference. However, if there are multiple reference models that differ by discrete conformational changes, the current protocol can be extended to competitively compare occupancies against different reference models. Ordering structures in a ribosome assembly pathway In the final step that is relevant to defining an assembly process, we developed a to place ribosome assembly module intermediates into a pathway. In this analysis, a “folding” matrix is calculated from the molecular weight difference that would need to be added to a given map to create a second map, and the “unfolding” matrix is calculated from the molecular weight that would need to be subtracted from one map to create a second. Each element of the folding/unfolding matrix can be considered as the driving force/barrier for a structural transition between two classes. By postulating that folding proceeds by incremental assembly, with minimal unfolding, a parsimonious transition graph can be constructed with allowed passages between classes based on simple criteria – there is a molecular weight cutoff unfolding transitions, and there is a limit set to the number of transitions emanating from each class. Large unfolding events are unlikely, given the large number of states that are close in molecular weight, but small unfolding events must be permitted to allow for structural rearrangements required to transition between classes. Finally, it is likely that structural transitions proceed from a finite manifold of close intermediates. The folding and unfolding matrices can be used to construct a directed graph of allowed transitions using these criteria, as shown in Figure 7.
that uses molecular weight differences
Analysis of bL17-lim data using the quantitative heterogeneity mining protocol Our quantitative mining protocol was developed using data collected from newly purified assembly intermediates from the previously characterized bL17-limitation strain (Davis et al., 2016). We collected new cryo-EM data (Supplementary Table 1 and subjected it to our workflow. In the discovery phase, we employed an updated high-resolution limit for refinement (res_high_class=20Å) and an initial n5>n3>n2 hierarchical classification scheme, followed by additional rounds of binary subdivision. All maps were binarized according to the 3 map threshold determined individually for each map. |
To determine if subclassification was complete, we selected a v-limit of 1.5 kDa. The rationale for this choice is that 1.5 kDa represents the size of the smallest RNA helix present in the
σ
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bacterial ribosome and therefore corresponds to the smallest feature that we would like to capture in the data. For our purposes, smaller features can be assumed to be either biological and/or experimental noise. After iterative subclassification, the total number of classes is 42. The similarity between all of the maps was then analyzed by the hierarchical clustering analysis as described above, and at this stage, pairs of classes were combined with a 10 kDa difference threshold (Figure 5A, dotted red line). This cutoff was chosen because it is close to the average molecular weight of all proteins and rRNA features, and we wished to reduce the complexity of our data. We found one pair of structures that were similar to one another according to our established criteria for biological significance, and the particles belonging to these classes were accordingly combined (Figure 5A). Thus, using this protocol, a total of forty-one ribosome assembly intermediates were identified using quantitative metrics and similarity analysis, with minimal heuristic intervention. The classes were then subjected to an occupancy analysis to view the sets of cooperative folding blocks across the different classes. For the bL17-lim dataset here, the maps are compared to the reference crystal structure (PDB 4ybb) (Figure 6A). The reference 4ybb structure is filtered to 10Å and segmented into volumes corresponding to individual r-proteins and rRNA helices, resulting in 139 theoretical map segments (Figure 6B). The fractional analysis revealed five major structural blocks (Figure 6G,H) The largest, block I (red) is composed of structural elements that are largely present in all of the classes. These elements are found on the back of the ribosome and represent the structural core that can form without bL17. Block II (green) represents the central protuberance, which is fully formed in the D and E classes but is either missing or misdocked in the B and C classes. Block III (yellow) maps to the base of the ribosome and the L1 stalk. These features are mostly present in the C and E classes but are missing in the B and D classes. These two blocks represent parallel pathways in assembly (Davis et al., 2016), as it is unlikely that the base of the ribosome would be unfolded or disordered in order to form the central protuberance, and vice versa. Block IV (blue) represents density that is specific to the base of the L7/12 stalk and is mostly present in the D and some of the E classes. |
Finally, block V (purple) represents density that is mostly missing in all maps, and is composed of h68, bL9, the L7/12 stalk, and the top of the L1 stalk. These represent features that are among the last of the ribosome to fold (like h68) or are flexible elements (the stalks). bL9 is a special case, as the conformation in the crystal structure is an artifact due to crystallization; in cryo-EM structures, bL9 wraps around to the interface between the 30S and 50S subunits and is often flexible. These central blocks are very similar to the ones that we discovered previously (Davis et al., 2016), but this updated occupancy matrix will allow us to compare the blocks that arise from other depletion or deletion strains in order to explore the cooperative block-like behavior or ribosome assembly in future work. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The ordering module was used to calculate an initial pathway in the absence of bL17- lim, which was modified by hand, as elements like the misdocked central protuberance and non-native structural elements can have large effects on molecular weight differences but may arise earlier in the order of assembly. We found the same initial super classes as previously reported (B, C, D and E classes). While the classes we found were similar to the initial bL17 data (Supplemental Figure 4), the new classes enabled refinement of our bL17-lim ribosome assembly pathway. First, we found a YjgA-dependent pathway through the assembly process (Figure 7, classes denoted by *). YjgA was only bound if the central protuberance and the L1 stalks were present. We also discovered three potential parallel processes in the C class where the earliest event could either be the completion of the L1 stalk, the partial docking of the central protuberance, or the formation of the base of the L7/12 stalk. We did not previously observe the formation of the base of the L7/12 structure in the assembly pathway for any class. We also found an immature B class (Figure 7, structure 1) and an immature D class where the base was missing, but the L7/12 stalk was absent or present (Figure 7, structures 2 and 3), which were not present in the original set of 13 structures (Davis et al., 2016). In particular, the immature B class represents the least mature pre-50S intermediate identified to date. We also identified several structures that seem to be transition points between the two classes (Figure 7, structures 6 and 7), and we observe formation of density at the base of the structures, which is lacking in other D classes and is present in other E classes. These new discoveries inform a better understanding of ribosome assembly in the context of bL17 limitation, and the data analysis process will allow us to quantitatively assess cryo-EM data from other limitation strains and ribosome assembly defects. |
Conclusions
Heterogeneity analysis in cryo-EM provides exciting opportunities to discover new biology, but current workflows suffer from numerous challenges. The work here addresses three challenges that researchers face in the analysis of cryo-EM data, as exemplified using a case study of ribosome assembly intermediates: establishing a divisive approach to classification with well-defined endpoints to discover novel stats, a comprehensive difference analysis between distinct structures, and the application of well-defined criteria (thresholds) for limiting classification. The application of specific thresholds and limits (Table 1) has been critical to the success of analyzing ribosome assembly intermediate data. The implementation of this workflow has allowed us to identify an additional 28 ribosome assembly intermediates (counting the 41 assembly intermediates after merging similar classes), which include an independent pathway for the assembly factor YjgA and the earliest intermediate discovered to date in the
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ribosome assembly process. The discovery and analysis modules of this workflow provide a powerful analysis for quantitatively interrogating heterogeneous cryo-EM data for complex biological processes. Acknowledgements
Molecular graphics and analyses were performed with the USCF Chimera package (supported by NIH P41 GM103311). This work was supported by grants from the NIH DP5-OD021396 and U54 AI150472 (to D.L.) R35-GM136412 (to JRW). Author Contributions Jessica N. Rabuck-Gibbons: Conceptualization, Investigation, Methodology, Software, Formal Analysis, Data Curation, Writing – Original Draft, Writing – Review & Editing, Visualization. Dmitry Lyumkis: Conceptualization, Resources. Investigation, Writing – Review & Editing,
James R. Williamson: Conceptualization, Methodology, Software, Data Curation, Writing – Review & Editing, Visualization, Supervision, Project Administration, Funding Acquisition. Declaration of Interests The authors declare no competing interests. References
Baldwin, P. R. and Lyumkis, D. (2020) 'Non-uniformity of projection distributions attenuates resolution in Cryo-EM', Progress in biophysics and molecular biology, 150, pp. 160-183. Baldwin, P. R. and Lyumkis, D. (2021) 'Tools for visualizing and analyzing Fourier space sampling in Cryo-EM', Progress in Biophysics and Molecular Biology, 160, pp. 53-65. Bartesaghi, A., Merk, A., Banerjee, S., Matthies, D., Wu, X., Milne, J. L. and Subramaniam, S. (2015) '2.2 Å resolution cryo-EM structure of -galactosidase in complex with a cell-permeant inhibitor', Science, 348(6239), pp. 1147-1151. Beckers, M., Jakobi, A. J. and Sachse, C. (2019) 'Thresholding of cryo-EM density maps by false discovery rate control', IUCrJ, 6(1), pp. |
18-33. Bernstein, K. A., Gallagher, J. E., Mitchell, B. M., Granneman, S. and Baserga, S. J. (2004) 'The small-subunit processome is a ribosome assembly intermediate', Eukaryotic cell, 3(6), pp. 1619-1626.
β
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Davis, J. H., Tan, Y. Z., Carragher, B., Potter, C. S., Lyumkis, D. and Williamson, J. R. (2016) 'Modular assembly of the bacterial large ribosomal subunit', Cell, 167(6), pp. 1610-1622. e15. Elmlund, D. and Elmlund, H. (2012) 'SIMPLE: Software for ab initio reconstruction of heterogeneous single-particles', Journal of structural biology, 180(3), pp. 420-427. Fernandez-Leiro, R. and Scheres, S. H. (2016) 'Unravelling biological macromolecules with cryo-electron microscopy', Nature, 537(7620), pp. 339-346. Gao, H., Valle, M., Ehrenberg, M. and Frank, J. (2004) 'Dynamics of EF-G interaction with the ribosome explored by classification of a heterogeneous cryo-EM dataset', Journal of structural biology, 147(3), pp. 283-290. Grant, T., Rohou, A. and Grigorieff, N. (2018) 'cisTEM, user-friendly software for single- particle image processing', elife, 7, pp. e35383. Gray, R. (1984) 'Vector quantization', IEEE Assp Magazine, 1(2), pp. 4-29. Harnpicharnchai, P., Jakovljevic, J., Horsey, E., Miles, T., Roman, J., Rout, M., Meagher, D., Imai, B., Guo, Y. and Brame, C. J. (2001) 'Composition and functional characterization of yeast 66S ribosome assembly intermediates', Molecular cell, 8(3), pp. 505-515. Jomaa, A., Jain, N., Davis, J. H., Williamson, J. R., Britton, R. A. and Ortega, J. (2014) 'Functional domains of the 50S subunit mature late in the assembly process', Nucleic Acids Research, 42(5), pp. 3419-3435. Jomaa, A., Stewart, G., Martín-Benito, J., Zielke, R., Campbell, T. L., Maddock, J. R., Brown, E. D. and Ortega, J. (2011) 'Understanding ribosome assembly: the structure of in vivo assembled immature 30S subunits revealed by cryo-electron microscopy', Rna, 17(4), pp. 697-709. Klaholz, B. P. (2015) in heterogeneous cryo-EM samples by 3D multivariate statistical analysis', Open Journal of Statistics, 5(07), pp. 820. Kühlbrandt, W. (2014) 'The resolution revolution', Science, 343(6178), pp. 1443-1444. Li, N., Chen, Y., Guo, Q., Zhang, Y., Yuan, Y., Ma, C., Deng, H., Lei, J. and Gao, N. (2013) 'Cryo-EM structures of the late-stage assembly intermediates of the bacterial 50S ribosomal subunit', Nucleic acids research, 41(14), pp. 7073-7083. Liao, H. Y., Hashem, Y. and Frank, J. (2015) 'Efficient estimation of three-dimensional covariance and its application in the analysis of heterogeneous samples in cryo-electron microscopy', Structure, 23(6), pp. 1129-1137. Loerke, J., Giesebrecht, J. and Spahn, C. M. (2010) 'Multiparticle cryo-EM of ribosomes', Methods in enzymology: Elsevier, pp. |
161-177. Ludtke, S. J. (2016) 'Single-particle refinement and variability analysis in EMAN2. 1', Methods in enzymology: Elsevier, pp. 159-189. Lyumkis, D., Brilot, A. F., Theobald, D. L. and Grigorieff, N. (2013) 'Likelihood-based classification of cryo-EM images using FREALIGN', Journal of structural biology, 183(3), pp. 377-388. Nakane, T., Kimanius, D., Lindahl, E. and Scheres, S. H. (2018) 'Characterisation of molecular motions in cryo-EM single-particle data by multi-body refinement in RELION', Elife, 7, pp. e36861. 'Structure sorting of multiple macromolecular states
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Nakane, T., Kotecha, A., Sente, A., McMullan, G., Masiulis, S., Brown, P. M., Grigoras, I. T., Malinauskaite, L., Malinauskas, T. and Miehling, J. (2020) 'Single-particle cryo-EM at atomic resolution', Nature, 587(7832), pp. 152-156. Ni, X., Davis, J. H., Jain, N., Razi, A., Benlekbir, S., McArthur, A. G., Rubinstein, J. L., Britton, R. A., Williamson, J. R. and Ortega, J. (2016) 'YphC and YsxC GTPases assist the maturation of the central protuberance, GTPase associated region and functional core of the 50S ribosomal subunit', Nucleic acids research, 44(17), pp. 8442-8455. Nikolay, R., Hilal, T., Qin, B., Mielke, T., Bürger, J., Loerke, J., Textoris-Taube, K., Nierhaus, K. H. and Spahn, C. M. (2018) 'Structural visualization of the formation and activation of the 50S ribosomal subunit during in vitro reconstitution', Molecular cell, 70(5), pp. 881-893. e3. Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C. and Ferrin, T. E. (2004) 'UCSF Chimera—a visualization system for exploratory research and analysis', Journal of computational chemistry, 25(13), pp. 1605-1612. Punjani, A. and Fleet, D. J. (2021a) '3D Flexible Refinement: Structure and Motion of Flexible Proteins from Cryo-EM', bioRxiv. Punjani, A. and Fleet, D. J. (2021b) '3D Variability Analysis: Resolving continuous flexibility and discrete heterogeneity from single particle cryo-EM', Journal of Structural Biology, 213(2), pp. 107702. Rabuck-Gibbons, J. N., Popova, A. M., Greene, E. M., Cervantes, C. F., Lyumkis, D. and Williamson, J. R. (2020) 'SrmB rescues trapped ribosome assembly intermediates', Journal of molecular biology, 432(4), pp. 978-990. Razi, A., Guarné, A. and Ortega, J. (2017) 'The cryo-EM structure of YjeQ bound to the 30S subunit suggests a fidelity checkpoint function for this protein in ribosome assembly', Proceedings of the National Academy of Sciences, 114(17), pp. E3396- E3403. Russo, C. J. and Passmore, L. A. (2014) 'Ultrastable gold substrates for electron cryomicroscopy', Science, 346(6215), pp. |
1377-1380. Sashital, D. G., Greeman, C. A., Lyumkis, D., Potter, C. S., Carragher, B. and Williamson, J. R. (2014) 'A combined quantitative mass spectrometry and electron microscopy analysis of ribosomal 30S subunit assembly in E. coli', Elife, 3, pp. e04491. Scheres, S. H. (2012) 'RELION: implementation of a Bayesian approach to cryo-EM structure determination', Journal of structural biology, 180(3), pp. 519-530. Scheres, S. H. (2016) 'Processing of structurally heterogeneous cryo-EM data in RELION', Methods in enzymology: Elsevier, pp. 125-157. Scheres, S. H., Núñez-Ramírez, R., Sorzano, C. O., Carazo, J. M. and Marabini, R. (2008) 'Image processing for electron microscopy single-particle analysis using XMIPP', Nature protocols, 3(6), pp. 977-990. Shajani, Z., Sykes, M. T. and Williamson, J. R. (2011) 'Assembly of bacterial ribosomes', Annual review of biochemistry, 80, pp. 501-526. Sorzano, C., Bilbao-Castro, J., Shkolnisky, Y., Alcorlo, M., Melero, R., Caffarena- Fernández, G., Li, M., Xu, G., Marabini, R. and Carazo, J. (2010) 'A clustering approach to multireference alignment of single-particle projections in electron microscopy', Journal of structural biology, 171(2), pp. 197-206. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
40 41 42 43 44 45
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Spahn, C. M. and Penczek, P. A. (2009) 'Exploring conformational modes of macromolecular assemblies by multiparticle cryo-EM', Current opinion in structural biology, 19(5), pp. 623-631. Stokes, J. M., Davis, J. H., Mangat, C. S., Williamson, J. R. and Brown, E. D. (2014) 'Discovery of a small molecule that inhibits bacterial ribosome biogenesis', Elife, 3, pp. e03574. Sykes, M. T., Shajani, Z., Sperling, E., Beck, A. H. and Williamson, J. R. (2010) 'Quantitative proteomic analysis of ribosome assembly and turnover in vivo', Journal of molecular biology, 403(3), pp. 331-345. Tan, Y. Z., Aiyer, S., Mietzsch, M., Hull, J. A., McKenna, R., Grieger, J., Samulski, R. J., Baker, T. S., Agbandje-McKenna, M. and Lyumkis, D. (2018) 'Sub-2 Å Ewald curvature corrected structure of an AAV2 capsid variant', Nature communications, 9(1), pp. 1-11. Tan, Y. Z., Baldwin, P. R., Davis, J. H., Williamson, J. R., Potter, C. S., Carragher, B. and Lyumkis, D. (2017a) 'Addressing preferred specimen orientation in single-particle cryo-EM through tilting', Nature methods, 14(8), pp. 793-796. Tan, Y. Z., Baldwin, P. R., Davis, J. H., Williamson, J. R., Potter, C. S., Carragher, B. and Lyumkis, D. (2017b) 'Addressing preferred specimen orientation in single-particle cryo-EM through tilting', Nature methods, 14(8), pp. 793. Uicker, W. C., Schaefer, L. and Britton, R. A. (2006) 'The essential GTPase RbgA (YlqF) in Bacillus subtilis', Molecular microbiology, 59(2), pp. |
528-540. Wang, Q., Matsui, T., Domitrovic, T., Zheng, Y., Doerschuk, P. C. and Johnson, J. E. (2013) 'Dynamics in cryo EM reconstructions visualized with maximum-likelihood derived variance maps', Journal of structural biology, 181(3), pp. 195-206. White, H., Ignatiou, A., Clare, D. and Orlova, E. (2017) 'Structural study of heterogeneous biological samples by cryoelectron microscopy and image processing', BioMed research international, 2017. Wolfram Research, I. (2020) Mathematica. Version 12.2 edn. Champaign, Illinois: Wolfram Research, Inc. Yip, K. M., Fischer, N., Paknia, E., Chari, A. and Stark, H. (2020) 'Atomic-resolution protein structure determination by cryo-EM', Nature, 587(7832), pp. 157-161. Zhang, K., Pintilie, G. D., Li, S., Schmid, M. F. and Chiu, W. (2020) 'Resolving individual atoms of protein complex by cryo-electron microscopy', Cell research, 30(12), pp. 1136- 1139. Zhong, E. D., Bepler, T., Berger, B. and Davis, J. H. (2021) 'CryoDRGN: reconstruction of heterogeneous cryo-EM structures using neural networks', Nature Methods, 18(2), pp. 176-185.
is required
for 50S ribosome assembly
Materials and Methods
Cell Growth and Isolation of Ribosomal Particles Cells were grown and ribosomal particles were isolated as in (Davis et al., 2016). Briefly, strain JD321 was grown in M9 media (48mM Na2HPO4, 22mM KH2PO4, 8.5mM NaCl, 10mM MgCl2, 10mM MgSO4, 5.6mM glucose, 50mM Na3*EDTA, 25mM CaCl2, 50mM FeCl3, 0.5mM ZnSO4, 0.5mM CuSO4, 0.5mM MnSO4, 0.5mM CoCl2,
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 0.04mM d-biotin, 0.02mM folic acid, 0.08mM vitamin B1, 0.11mM calcium pantothenate, 0.4nM vitamin B12, 0.2mM nicotinamide, 0.07mM riboflavin, and 7.6mM (14NH4)2SO4]) with tetracycline (10 mg/mL), chloramphenicol (35 mg/mL),and limiting conditions HSL (0.1 nM) and harvested at OD=0.5. Cells were lysed in Buffer A (20mM Tris-HCl, 100mM NH4Cl, 10mM MgCl2, 0.5mM EDTA, 6mM b-mercaptoethanol; pH 7.5) by a mini bead beater, and the clarified lysate was fractionated on a 10-40% w/v sucrose gradient (50mM Tris-HCl, 100mM NH4Cl, 10mM MgCl2, 0.5mM EDTA, 6mM b- mercaptoethanol; pH 7.5). Electron Microscopy Data Collection Fractions containing the ribosomal intermediates were spin-concentrated with a 100 kDa MW filter (Amicon) and buffer exchanged into Buffer A. 3 l of this sample was added to a plasma cleaned (Gatan, Solarus) 1.2mm hole, 1.3mm spacing holey gold grids (Russo and Passmore, 2014). Grids were manually frozen in liquid ethane, and single particle data was collected using Leginon on a Titan Krios microscope (FEI) with a K2 summit direct detector (Gatan) in super-resolution mode (pixel size of 0.66Å at 22,500 magnification). |
A dose rate of ~5.8e-/pix/sec was collected across 50 frames with a fluence of 33-35e-/Å2 at a tilt of -20° to compensate for preferred orientation (Tan et al., 2017b). μ
FrealignX Classifications After conversion from Relion to FrealignX parameters, global refinements were performed in FrealignX, and all occupancies were randomized across the parameter files. A final value of 20Å was selected for res_high_class, and after every 10 cycles of classification/refinement, all classes were aligned to a C class scaffold using custom scripts for a 3D alignment with Chimera (Pettersen et al., 2004) while running FrealignX. For each classification step, 50 refinement/classification cycles were performed. After initial classification, each class was selected in a parameter file for subsequent rounds of classification using the merge_classes.exe in cisTEM (Grant, Rohou and Grigorieff, 2018) and custom scripts. The occupancies were randomized across the parameter files, and the same cycle of 50 cycles of refinement/classification interspersed with 3D alignment with Chimera every 10 cycles. FSC curves and Euler plots were generated by FrealignX and cisTEM (Grant, Rohou and Grigorieff, 2018), and 3DFSC plots were calculated by the 3DFSC server (Tan et al., 2017a). The SCF was calculated according to the process in (Baldwin and Lyumkis, 2021; Baldwin and Lyumkis, 2020). The 3DFSCs and all maps shown were visualized in Chimera (Pettersen et al., 2004), and the details for each map are indicated in Table S1. σ
Calculation of
values
1 2 3 4 5 6 7 8 9 10 11 12 13
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. σ
which was used as a For analysis, each map was first filtered to 10Å. To calculate measure of noise, each map was unmasked by expanding the outer_radius in FrealignX so that the spherical particle mask would be larger than the box size. The Fourier folding of signal along the edges of the box was negligible. Relion 2.1 was used to calculate the value using the relion_image_handler command. Relion 2.1 was then used to create binarized maps using the relion_image_handler command, and the binarization threshold was set to 3
σ
σ
. Hierarchical clustering analysis Thresholded, binarized maps were given as input to a custom Mathematica script (Wolfram Research, 2020). The Mathematica script calculated the segmented difference maps between all maps and calculated the molecular weights of the differences maps (in kilodaltons) using Equation 1(Ludtke, 2016):
(cid:0)(cid:2) (cid:3) (cid:4)
(cid:0)(cid:2)(cid:3)(cid:4)(cid:5)(cid:6)
(cid:5) (cid:6)(cid:7)(cid:8)(cid:9)(cid:10)(cid:11)(cid:7)(cid:12)(cid:9)(cid:7) (cid:5) (cid:13)/1000
Density ρ is 0.81 daltons/Å3. The MW difference matrix was clustered using the Euclidean distance metric and Ward’s linkage and displayed in a dendrogram. |
Similar maps were averaged together after hierarchical clustering analysis using EMAN2 ((Ludtke, 2016). Occupancy Analysis The thresholded and binarized maps were given as input, and the reference map from the E. coli 50S subunit crystal structure (PDB ID 4YBB) was segmented into 139 elements comprised of individual ribosomal proteins and rRNA helices according to the 23S secondary structure. Theoretical densities for each r-protein and rRNA helix were calculated for each element at 10Å using the pdb2mrc command from EMAN. Prior to binarization, voxels that had overlapping theoretical density from two structural elements, were assigned to the smaller of the two theoretical volumes so that each pair of volumes is nonoverlapping. Each voxel density was binarized to either 0 or 1 using a threshold of 0.016, which is the threshold that gave the approximately correct molecular weight for individual r-proteins and rRNAs helices. The relative volumes in the binarized experimental and reference maps were calculated, which gave a fractional occupancy between 0 and 1 for each element. The occupancy values were clustered across the rows (classes) and columns (rRNA/protein elements) using an unsupervised hierarchical clustering using the Euclidean distance metric and Ward’s linkage method, as implemented in Mathematica. Parsimonious folding/unfolding matrices. A pathway diagram was constructed by using the n x n molecular weight difference matrices, Mf and Mu, from a set of n structures. Each difference map (Mi-Mj) has negative elements corresponding to folding that occurs in the transition from class i to class j, and positive elements that correspond to
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. unfolding that occurs in the transition from class i to class j. The volume changes for T . The folding and unfolding form the elements of Mf or Mu , noting that Mu = Mf matrices Mf and Mu are used to construct a directed graph G, comprised of the set of vertices vi, and a set of directed edges, eij, representing the allowed transitions between classes, The set of edges is initialized as the set of eij where Mu,i,j > Mf,i,j, such that only net folding transitions are allowed. The set of edges is pruned using two global parameters: θunf as a maximum threshold for unfolding, and nbranch, as a limit on the number of transitions emanating from a single class. The unfolding threshold limits unreasonable structural rearrangements, while the branching threshold limits transitions to a small set of the closest transitions. Edges are eliminated if the unfolding exceeds the threshold such that Mu,i,j > θunf, unless elimination of the edge results in a disconnected graph G. Next, for each vertex vi, the set of remaining edges eik emanating from vi, are sorted into the order based on the Mf,i,k, retaining at most the nbranch edges, again, unless deleteing the edge would result in a disconnected graph G. The resulting transition graph G should have one or more source vertices (classes) that are the earliest classes in the assembly pathway, and one or more sink vertices that are the most mature classes in the pathway. |
Tuning of the parameters θunf and nbranch, adjusts the connectivity and degree of branching of the resulting graph. The graph vertices are annotated with thumbnails of the map, followed by manual layout of the graph into a sensible order in Adobe Illustrator. The values of θunf and nbranch used to generate the graph in Figure 7 were 390 kDa and 3, respectively. Data Deposition and Software Availability Mathematica scripts and example parameter files, where needed, will be available upon request. All maps are deposited at EMPIAR as noted in the Key Resources Table. Figure Titles and Legends
Figure 1. Description of the bacterial large ribosomal subunit and prior assembly intermediates identified by cryo-EM. (A) PDB ID 4YBB labeled with prominent features identifiable on the large ribosomal subunit, including the central protuberance (CP), base, L1 stalk, and L7/12 stalks. These terms are used throughout the paper. (B) Primary classes identified within the original bL17-lim dataset (Davis et al., 2016). From left to right: B class (red), C class (yellow), D class (green), and the E class (blue). Figure 2. Workflow for cryo-EM heterogeneity analysis. Figure 3. A divisive resolution-limited subclassification approach facilitates identifying rare structural variants. (A) FrealignX classification with res_high_class parameter set to Nyquist (5.24Å). (B) FrealignX classification with the res_high_class parameter set to
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 20Å. Using a lower resolution cutoff leads to the identification of a broader range of classes. (C) Results of the five different classification schemes. The colors (A,B) correspond to the classes found in (C). Figure 4. Segmented difference analysis helps to define molecular weight differences between map pairs. (A) Example where two maps would have been considered different before segmentation, but are not different after segmentation. (B) Example where two maps are different both before and after segmentation. Numbers indicate positive (Map1-Map2) and negative (Map2-Map1) molecular weight differences. Figure 5. Hierarchical clustering is used to combine similar maps under a given threshold. (A) Hierarchical clustering analysis of the maps that result after the terminal subclassification (total n=42). The red dashed line indicates the 10kDa MWCO used to combine similar maps at this step, and the red stars indicate maps that are combined after this analysis. After combining similar maps, the final number of classes is thus 41. (B) Close-up example of two combined maps in (A). The leftmost structure is distinct from the other two by ~135 kDa and needs to be treated independently, whereas the two rightmost structures can be combined into a single class. |
Figure 6. Results of occupancy analysis on the full dataset mapped onto the ribosomal scaffold (A) Reference crystal structure 4ybb. (B) Binarized maps of the individual proteins and rRNA helices created by segmenting the crystal structure into 139 individual helices and proteins, and calculating theoretical 10A maps in Chimera. (C) An example of a binarized experimental map arising from sub-classification. The pixels from the binarized experimental map that are located in the theoretical binarized map are counted and normalized to an occupancy value of 0-1. (D) Example of an E class (blue) where the occupancy of an rRNA helix (h82, salmon) is fully occupied, with the corresponding occupancy block underneath. (E) Partial occupancy example of h82 (salmon) with a C class (yellow). (F) Example where rRNA (h82, salmon) is missing in the experimental data (B class, red). G. Occupancy analysis plot, where the individual proteins and helices are shown on the x-axis, the experimental maps are on the y-axis, and the normalized occupancy values are shown from white (0) to dark blue (1). Hierarchical clustering of both structure elements and experimental maps was performed on the occupancy matrix using a squared Euclidean distance metric and Ward’s linkage. (H) Occupancy analysis blocks mapped back to the reference structure 4YBB, and the numbering system is the same as in (G). Figure 7. Revised ribosome assembly map from bL17-lim. (Assembly pathway drawn by analyzing the folding and unfolding molecular weight matrices and revised by hand). 1
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Supplemental Information Titles and Legends Supplemental Figure 1. Results of the five tested classification schemes grouped by class. The structures are colored by classification scheme. Unique classes are shown by an asterisk (*), and classes that are similar are underscored by red brackets. Clustering of (A) the B classes, (B) the C classes, (C) the D classes, and (D) the E classes resulting from the tested classification schemes. Any 70S or “junk” classes that result from the subclassifications are omitted for clarity. Supplemental Figure 2. Example of ambiguous density and features for B class particles. From left to right: (B class filtered to 5Å and shown at 3 1.5 to the main particle that is likely due to disordered rRNA (black arrows). σ
σ
map, 2
map, and
σ
σ
map. In particular, at the 2
map threshold, noise above background is visible proximal
σ
map threshold that is used in our Supplemental Figure 3. Comparison of the the 3 current analysis versus the confidence map FDR threshold (Beckers, 2019). The black line represents y=x, and the red and black dots represesent thresholds at 1% and map threshold 0.01% FDR, respectively. |
The measures are highly correlated, and the 3 is generally more conservative than either FDR threshold. σ
Supplemental Figure 4. Hierarchical clustering analysis of the original bL17-lim data (orange) together with the structures solved by the new data processing workflow (blue). The red dotted line indicates the 10.0 kDa cutoff applied to determine similarity between classes. The original classes typically have counterparts within the new data (red underlined structures), but the new workflow is able to identify many more structural intermediates. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.23.449614
;
this version posted June 24, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. limits
Thresholds
Threshold/limit r-limit
v-limit
low pass filter threshold
binarization threshold
Description The minimum resolution necessary for a map volume limit: molecular weight difference limit for terminal subdivision used to normalize resolution between maps and to focus on lower- resolution differences between maps threshold at which maps are binarized; pixel values below this limit are set to 0, values above this limit are set to 1
Value used in this paper 10Å
1.5 kDa
10Å
3𝛔map
segmentation threshold defines the volume of dust to be removed from difference maps defines the lower limit for acceptable differences between maps
difference threshold
1.5 kDa 10 kDa |
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Dissecting the molecular basis of human interneuron migration in forebrain assembloids from Timothy syndrome
Fikri Birey1,2, Min-Yin Li1,2, Aaron Gordon3, Mayuri V. Thete1,2, Alfredo M. Valencia1,2, Omer Revah1,2, Anca M. Pașca4, Daniel H. Geschwind3, 4, 5, 6, and Sergiu P. Pașca1,2*#
1Department of Psychiatry and Behavioral Sciences, Stanford University, Stanford, CA, USA 2Stanford Brain Organogenesis, Wu Tsai Neurosciences Institute, Stanford University, Stanford, CA, USA 3Program in Neurogenetics, Department of Neurology, University of California Los Angeles, Los Angeles, CA, USA 4Department of Human Genetics, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA, USA 5Center for Autism Research and Treatment, Semel Institute, University of California Los Angeles, Los Angeles, CA, USA 6Institute of Precision Health,
University of California Los Angeles, Los Angeles, CA, USA
4Department of Pediatrics, Division of Neonatology, Stanford University, Stanford, CA, USA
Corresponding Author, #Lead contact: [email protected]
SUMMARY
Defects in interneuron migration during forebrain development can disrupt the assembly of cortical circuits and have been associated with neuropsychiatric disease. The molecular and cellular bases of such deficits have been particularly difficult to study in humans due to limited access to functional forebrain tissue from patients. We previously developed a human forebrain assembloid model of Timothy Syndrome (TS), caused by a gain-of-function mutation in CACNA1C which encodes the L-type calcium channel (LTCC) Cav1.2. By functionally integrating human induced pluripotent stem cell (hiPSC)-derived organoids the dorsal and ventral forebrain from patients and control individuals, we uncovered that migration is disrupted in TS cortical interneurons. Here, we dissect the molecular underpinnings of this acute phenotype pharmacological modulation of Cav1.2 can rescue the saltation length but not the saltation frequency of TS migrating interneurons. Furthermore, we find that the defect in saltation length in TS interneurons is associated with aberrant actomyosin function and is rescued by pharmacological MLC saltation phosphorylation, whereas
resembling
and
report
that
modulation
of
frequency phenotype in TS interneurons is driven by enhanced GABA sensitivity and can be restored by GABA receptor antagonism. Overall, these findings uncover multi-faceted roles of LTCC function in human cortical interneuron migration in the context of disease and suggest new strategies to restore interneuron migration deficits. INTRODUCTION
The formation of the cerebral cortex involves the assembly of glutamatergic neurons and GABAergic interneurons into circuits during fetal development (Silva et al., 2019). |
Glutamatergic neurons are generated in the dorsal forebrain, whereas GABAergic interneurons are generated in the ventral forebrain, and undergo extensive tangential migration to reach the cerebral cortex (Anderson et al., 1997; Marín, 2013; Wamsley and Fishell, 2017). Migration of cortical interneurons has been investigated in rodents, and it involves motion of a leading branch that subsequently gets stabilized in the direction of forward movement; triggers various cytoskeletal rearrangements that create a pulling force by the leading branch and a pushing force by the cell rear. The process is referred to as
this
the
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. nucleokinesis, and underlies the saltatory mode of migration of cortical interneurons (Schaar and McConnell, 2005; Bellion et al., 2005; Martini and Valdeolmillos, 2010). The migration
cortical interneurons spans a significantly fraction of in utero development and, unlike in rodents (Inamura et al., 2012), continues after birth in humans (Silbereis et al., 2016). Genetic perturbations that affect this process are thought to lead to miswiring of cortical circuits and contribute to disease, including schizophrenia, autism spectrum disorder (ASD) and epilepsy (Meechan et al., 2012; Powell, 2013; Muraki and Tanigaki, 2015; Buchsbaum and Cappello, 2019; Maset et al., 2021). However, how these genetic, disease-related events impact neuronal in migration and cortical circuit assembly humans remains elusive, mainly due to the lack of brain tissue from patients for functional studies. To start addressing this limitation, we forebrain have assembloids (Birey et al., 2017) – a 3D in vitro platform forebrain organoids or spheroids derived from human induced pluripotent stem cells (hiPSCs) are physically and functionally integrated. We used this system to model the migration of human cortical interneurons into glutamatergic networks and demonstrated that their migration kinetics was reminiscent of GABAergic neurons in ex vivo human forebrain specimens. of
human
developed
previously
in which
region-specific
Cortical interneuron migration has been previously shown to be dependent on LTCC function (Bortone and Polleux, 2009; Kamijo et al., 2018), and genetic variants that alter LTCC function have been in various neuropsychiatric disorders (Bhat et al., 2012; Purcell et al., 2014; Ripke et al., 2014). For instance, gain-of-function mutations the CACNA1C gene, which encode the pore-forming subunit of Cav1.2, cause Timothy Syndrome (TS) – a highly penetrant but rare neurodevelopmental disorder associated with ASD and epilepsy (Splawski et al., 2004). Using forebrain assembloids derived from patients with TS, we previously uncovered a migration phenotype in TS interneurons in which the saltation length was reduced, but the saltation frequency was increased; overall leading to a defect in migration (Birey et al., 2017). |
However, the molecular mechanisms underlying the compound nature of this migration defect is unknown. Understanding the molecular basis of
implicated
in
cellular with neurodevelopmental disorders is a key step towards developing novel therapeutic strategies. phenotypes
associated
Here, we investigate the molecular basis of the TS migration defect using calcium imaging, long-term live imaging, pharmacology, whole- cell patch clamping and RNA-sequencing in forebrain assembloids and ex vivo primary human cortical slice cultures. We discover that defects in TS interneuron saltation length and frequency are driven by distinct molecular pathways, elucidating a unique mode of LTCC- mediated regulation of interneuron migration in the context of disease. RESULTS
Uncoupling of cell rear-front coordination in migrating cortical interneurons from TS patients
To investigate the molecular basis of the human cortical interneuron migration deficits in TS, we assembled human cortical spheroids (hCS) and human subpallial spheroids (hSS) to generate forebrain assembloids from hiPSCs derived from 8 control individuals and 3 patients with TS carrying the G406R mutation in the alternatively spliced exon 8a of CACNA1C (Splawski et al., 2004) (Figure 1A, 1B; Supplementary Table 1 summarizes the use of hiPSCs lines in various experiments). The point mutation to delayed inactivation of the Cav1.2 channel and increased intracellular calcium entry (Figure 1B, 1C). To visualize interneurons in forebrain assembloids, we used a lentiviral reporter for the interneuron lineage (Dlxi1/2b-eGFP or Dlxi1/2- mScarlet). Cav1.2 was abundantly expressed in Dlxi1/2b+ human interneurons (Figure S1A). Furthermore, the G406R-containing exon 8a isoform is the dominant isoform present during development, which is replaced in relative abundance by exon 8 isoform in the postnatal cerebral cortex (Panagiotakos et al., 2019). Due to aberrant splicing caused by G406R, the developmental switch from exon 8a isoform to exon 8 isoform is hindered in TS hSS, prolonging the enrichment of the TS mutation-containing 8a isoform at later timepoints (Figure S1B), and this is consistent with previous studies (Paşca et al., 2011; Panagiotakos et al., 2019). leads
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 1. Uncoupling of soma rear-front coordination during nucleokinesis in TS interneurons. (A) Schematic illustrating the generation of forebrain assembloids by integrating human Cortical Spheroids (hCS) and human Subpallial Spheroids (hSS) from hiPSCs derived from 8 control (Ctrl) individuals and 3 patients with Timothy Syndrome (TS). (B) Schematic illustrating the impact of TS mutation G406R on the Cav1.2 function. (C) Calcium currents recorded from monolayer hSS neurons (day 144) labeled with Dlxi1/2b-eGFP. |
(D) Distance versus time plots demonstrating inefficient migration of TS interneurons. (Ctrl, n = 9 cells from 2 hiPSC lines, 1 assembloid per line; TS, n = 11 cells from 2 hiPSC lines, 1 assembloid per line. Two-way ANOVA for subjects, F18,540 = 48.59, ****P< 0.0001). (E) Saltation length and frequency phenotypes in TS interneurons and rescue of saltation length phenotype by nimodipine (Nim) (Ctrl, n = 8 cells from 3 hiPSC lines, 1 assembloid per line; TS, n = 7 cells from 3 hiPSC lines, 1 assembloid per line. Paired t-test, ***P< 0.001). (F) Schematic illustrating DeepLabCut analysis pipeline. (G) Representative live migration imaging snapshots of a migrating Ctrl and TS interneuron where soma rear and front were tracked using DeepLabCut. (H) Distance versus time plots of individual saltation events demonstrating soma rear-front uncoupling in TS interneurons. (I) Quantification of correlation coefficient (Pearson’s r) between soma rear and soma front per saltation event (left; pooled, right; separated by line. Ctrl, n = 16 cells from 3 hiPSC lines, 1–3 assembloid per line; TS, n = 21 cells from 3 hiPSC lines, 1–3 assembloid per line. Paired t-test, **P< 0.01). Bar charts: mean ±s.e.m, violin plots: center line, median. Scale bars: 40 μm (1F); 10 μm (1G). n.s.= not significant. For (1E), (1F), different shades of gray represent individual Ctrl hiPSC lines. Different shades of cyan represent individual TS hiPSC lines. that TS interneurons migrate inefficiently (Figure 1D, P< 0.0001, two-way ANOVA for Ctrl vs. TS) due to
We have previously
found
a reduction in the distance of individual saltations (“saltation length”) concomitant with an increase
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. in (“saltation frequency”) (Figure 1E, (Birey et al., 2017)). Cav1.2 activity is modulated by several classes of drugs. Interestingly, we found that the 1,4- dihydropyridine LTCC blocker nimodipine (5 μM) rescued the saltation length, but not the saltation frequency phenotype suggesting that the two phenotypes might be driven by distinct molecular pathways (Figure 1E, Ctrl saltation length: P< 0.001; TS saltation length, P< 0.001). the number of saltations
To elucidate the nature of this differential effect, we first performed fast, high-resolution imaging of migrating cortical interneurons in quantitative forebrain understanding interneuron migration dynamics is limited by the lack of tools to precisely track multiple, simultaneously moving cellular compartments over long periods of time. To overcome this issue, we applied a transfer learning-based pose estimation algorithm DeepLabCut (Mathis et al., 2018) to extract features associated with TS subcellular interneuron migration. Used for marker-less pose estimation of user-defined body parts in animal behavioral studies (Mathis and Mathis, 2020), DeepLabCut uses a small number of training frames (19) with which the user manually defines (ROIs). |
We reasoned that this method could be used to train a network to track discrete subcellular ROIs during interneuron migration without the need for localized fluorescent reporters (Figure 1F, Supplementary Video 1). Indeed, we verified that the pipeline could detect a reduction in the saltation length in TS (Figure S1C, soma front: P< 0.0001; soma rear: P< 0.0001). Previous work has extensively the the migration functional polarization of machinery interneurons, with microtubules in the leading branch and an actomyosin network in the soma rear, which mediate pulling and pushing forces during nucleokinesis respectively (Bellion et al., 2005). To estimate the contribution of pulling versus pushing in TS forces during nucleokinesis interneurons, we focused on monitoring soma cell front and rear. This analysis revealed an uncoupling of soma rear-front coordination in TS interneurons: while soma rear and front moved in synchrony during Ctrl interneuron saltations, the rear appeared uncoupled from the front of the soma in TS interneuron saltations (Figure 1G–1H, Supplementary Video 2). More specifically, the soma rear lagged behind the
assembloids. of
A
regions-of-interests
characterized
in cortical
soma front in the course of a saltation which was the correlation quantified by calculating coefficient between soma front versus soma rear movements (Figure 1I, P< 0.01). Overall, these results indicated that cell rear contractility may in migrating TS be selectively disrupted interneurons. Acute modulation of LTCCs and intracellular length, but not calcium affect saltation saltation forebrain assembloids and ex vivo forebrain tissue
frequency
in TS
Contractility in cells and tissues, such as skeletal muscle cells (Kuo and Ehrlich, 2015), is a calcium-dependent process (Bers, 2002). Given the cell rear contractility deficits in TS, where the G406R mutation leads to increased intracellular calcium ([Ca2+]i), we asked which aspects of the phenotype are calcium-related (Figure 2A), as opposed to calcium-independent mechanisms underlying dendrite retraction in TS cortical glutamatergic neurons (Krey et al., 2013). We performed live imaging in forebrain assembloids in a medium with modified extracellular calcium concentrations ([Ca2+]e) (Figure 2B, S2A), and found that compared to control levels (2 mM [Ca2+]e), a reduction to 0.5 mM [Ca2+]e severely impacted interneuron migration bringing most cells to a stop in both genotypes (Figure S2B). Migration in 1 mM [Ca2+]e was not as severely affected, with 63% and 27% of migrating interneurons coming to a stop in Ctrl and TS interneurons, respectively (Figure 2C, P< 0.01). Interestingly, out of the cells that performed at least one saltation in 1 mM [Ca2+]e (“mobile cells”), we found that the saltation length in TS interneurons was partially restored, while the saltation length in Ctrl interneurons was reduced. Saltation frequency was significantly reduced in both groups likely due to intolerance to longer exposure to less than 2mM [Ca2+]e. (Figure 2D, Ctrl saltation length: P< 0.05; TS saltation length: P< 0.01; Ctrl saltation frequency: P< 0.01; TS saltation frequency: P< 0.0001). |
the acute modulation of LTCCs and [Ca2+]i levels on the saltation in an ex vivo preparation, we prepared slices of human fetal forebrain at 18 postconceptional weeks (PCW18). We labeled interneurons with the same lentiviral reporter used for assembloid migration imaging (Dlxi1/2-eGFP). We then performed migration imaging in slices in the
To validate
the effect of
length phenotype
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 2. Acute modulation of intracellular calcium levels influences saltation length but not saltation frequency in TS interneurons. (A) Schematic illustrating the presumptive direct effect of intracellular calcium influx on saltation length and/or frequency. (B) Schematic illustrating the migration imaging experiment in 1 mM extracellular calcium [Ca2+]e. (C) Proportion of Ctrl and TS interneurons that are either mobile (>1 saltation) or non-mobile in 1 mM [Ca2+]e. (Chi-square test, χ2 = 7.0, **P< 0.01). (D) Quantification of saltation length and frequency at 1 mM [Ca2+]e. (Ctrl, n = 9 cells from 4 hiPSC lines, 1-3 assembloid per line; TS, n = 22 cells from 3 hiPSC lines, 1-3 assembloid per line. Paired t-test, *P< 0.05, **P< 0.01, ****P< 0.0001). (E) Schematic illustrating ex vivo primary fetal cortical slice culture protocol to perform interneuron migration imaging and FPL 64176 pharmacology. (F) Representative live migration imaging snapshots showing migrating Dlxi1/2b-eGFP-labeled interneurons in a primary cortical slice. (G) Quantification of saltation length and frequency following FPL 64176 administration. (n = 40 cells from 4 fields across 1 cortical section. Two-tailed t-test, ****P< 0.0001). Scale bars: 100 μm (1F). n.s.= not significant. For (2D), different shades of gray represent individual Ctrl hiPSC lines. Different shades of cyan represent individual TS hiPSC lines. presence of a potent non-dihydropyridine LTCC activator, FPL 64176 (5 μM, Figure 2E-2F). FPL 64176 application reduced the saltation length of migrating primary cortical interneurons, while it had no effect on their saltation frequency, supporting the view that LTCC-mediated calcium influx acutely modulates saltation length, but not the saltation frequency in migrating cortical interneuron (Figure 2G, saltation length: P< 0.0001). Excess phosphorylation of MLC influences saltation length, but not saltation frequency of TS cortical interneurons
We next asked whether downstream changes in cell-rear actomyosin signaling mediates the saltation length difference described above in TS interneurons. calmodulin activates the Myosin Light Chain Kinase (MLCK), which in turn phosphorylates Myosin Light Chain (MLC) and induces MLC interaction
Calcium-bound
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. |
The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. with F-actin and contraction at the soma rear (Figure 3A, 3B). We reasoned that increased ([Ca2+]i in TS interneurons may lead to excess phosphorylation of MLC, aberrant actomyosin function and consequently reduced saltation length. To test this, we first quantified pMLC2s19 levels, which has been previously shown correlate with myosin ATPase activity and contractility (Tan et al., 1992). Western blotting of protein lysates extracted from whole hSS (2– 3 hSS pooled per sample) revealed increased pMLC2s19 levels in TS hSS (Figure 3C–3D, P< 0.05, Figure S3A). Immunocytochemistry pMLC2s19 in MAP2+ neuronal soma of plated hSS also showed an increase in pMLCs19 levels in TS neurons (Figure 3E, 3G, Base Ctrl vs Base TS: P< 0.0001). Moreover, a 15-minute application of FPL 64176 (5 μM) on plated hSS neurons induced an increase in pMLC2s19 levels in Ctrl interneurons, demonstrating a link between MLC phosphorylation and LTCC activation (Figure 3G, Ctrl Base vs FPL: P< 0.05). Of note, we found no detectable increase of pMLC2s19 levels in TS interneurons treated with FPL 64176, which is likely due to saturation of pMLC2s19 in TS neurons (P= 0.23). Finally, we asked if inhibiting MLC phosphorylation would be sufficient to restore saltation length and/or that application of ML-7 (5 μM) – a selective MLCK inhibitor, rescued saltation length, but not saltation frequency of TS interneurons (Fig 3I, Ctrl saltation length: P< 0.0001; TS saltation length: P< 0.05), further emphasizing that two components of the TS migration phenotype are likely mediated by different molecular pathways downstream of the TS Cav1.2 channel. frequency
in TS. We
found
Transcriptional profiling suggests changes in GABA receptor signaling and membrane potential in TS cortical interneurons
Given that the TS saltation frequency phenotype remained unchanged upon acute modulation of LTCCs function, changes in [Ca2+]i levels and actomyosin function, we asked whether there are other, independent pathways that could explain the increased saltation frequency in TS interneurons. To investigate this, we performed RNA-sequencing in hCS and hSS (day 40–75) derived from 3 patients with TS and 4 control individuals (Figure 4A, Supplementary Table 3 includes information on samples), and first performed differential expression analysis in
hSS found upregulation of 3 genes encoding GABA receptor subunits (GABRA4: logFC = 3.3, FDR = 0.011; GABRG1: logFC = 4.8, FDR = 0.001; GABRG2: logFC = 1.1, FDR = 0.093; Figure 4B, 4C and Supplementary Table 4). Gene set enrichment analysis (GSEA) analysis further indicated a robust enrichment of genes associated with GABA transmission (“Synaptic transmission GABAergic”, “Chloride channel complex”) as well as resting membrane potential (“Regulation of membrane potential”, “Voltage- gated ion channel activity”) in hSS (Figure 4D, Supplementary Table 5), but not in hCS where we found primarily upregulation of glutamate receptor signaling changes (“Glutamate receptor signaling S4A, Supplementary Table 3). |
samples. Intriguingly, we
pathway”)
(Figure
Previous work has shown that gene co- expression relationships of gene expression can be leveraged to better understand disease biology (Zhang and Horvath, 2005; Voineagu et al., 2011; Parikshak et al., 2015). In light of this, co- we weighted expression network analysis to identify modules of highly correlated genes in the hSS dataset. This analysis identified 7 modules that were significantly associated with TS (FDR < 0.05; Figure 4E, Figure S4B, Supplementary Table 6). Two modules were down-regulated in TS hSS (M6 and M7; Figure S4B), one of which, module M7, was enriched in terms related to cell migration (“positive regulation of cell migration”, Supplementary Table 6). Modules M2, M3 and M4 (adjusted R2 M3 = 0.71, adjusted R2 M4 = 0.66; Figure S4B), which were upregulated in TS hSS, were enriched for SNP the basis of genome-wide heritability on association studies for several psychiatric disorders (BD: Bipolar Disorder, MDD: Disorder; M2BDEnrichment = 2.8, FDR = 6.2 × 10-3, M3BDEnrichment = 2.5, FDR = 4 × 10-5, M4ADHDEnrichment = 1.9, FDR = 1.3 × 10-2, M4BDEnrichment = 2.1, FDR = 5 × 10-6, M4MDD Enrichment = 2.2, FDR = 7.2 × 10-7, M4SCZEnrichment = 2.1, FDR = 3.9 × 10-6; Figure 4F). Importantly, M3 and M4 were also enriched for genes associated with ASD (ORM3 = 2.3, FDR = 1.1 × 10-3, ORM3 = 2.8, FDR = 1.1 × 10-10, Figure 4F). In agreement with the GSEA analysis, modules M2 and M4 were enriched for (Figure 4G, GABAergic neurotransmission
performed
gene (WGCNA)
M2 = 0.74, adjusted R2
(GWAS)
Major
Depressive
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure S4C, Supplementary Table 6) and ion channel activity (Figure 4H, Supplementary Table 6). To functionally validate changes in gene expression in hSS, we performed calcium imaging in interneurons in intact hSS labeled
with a genetically encoded calcium indicator (GCaMP) (Figure S4D) and found an increased rate of spontaneous calcium transients in TS (Figure S4E–S4F, P< 0.01). We then performed whole-cell patch clamping to investigate the intrinsic electrophysiological properties of
Figure 3. MLC hyper-phosphorylation at the cell rear is related to the saltation length but not saltation frequency in TS interneurons. (A) Schematic illustrating the effect of calcium on actomyosin contractility pathway. (B) Representative immunocytochemistry images of pMLC2s19 localized to Dlx5/6-eGFP+ interneurons from plated Ctrl hSS (d101). (C) Representative images of western blot scans showing total MLC2, pMLC2s19 and β-actin protein levels from Ctrl and TS hSS. (D) Western blot quantification of pMLC2s19 / total MLC2 protein level ratio in hSS (day 41-79). Values normalized to Ctrl levels per run (Ctrl, n = 18 samples from 3 hiPSC lines, 2-3 hSS per sample; TS, n = 18 samples from 3 hiPSC, 2-3 hSS pooled per sample. |
Two- tailed t-test, *P< 0.05). (E) Schematic illustrating the FPL 64176 stimulation experiment. (F) Representative ICC images of pMLC2s19 and MAP2 on plated hSS neurons (day 90). (G) Quantification of pMLC2s19 levels in MAP2+ somas in at baseline or FPL 64176-stimulated plated hSS neurons. (left: pooled, right: separated by line. Ctrl, n = 237 cells from 3 hiPSC lines; Ctrl+FPL, n = 269 cells from 3 hiPSC lines; TS n = 198 cells from 3 hiPSC lines; TS+FPL, n = 275 cells from 3 hiPSC lines, 2– 3 hSS pooled per line per condition. Kruskal-Wallis test with Dunn’s multiple comparison test. *P< 0.05, ****P< 0.0001) (H) Schematic illustrating the interneuron migration imaging and ML-7 pharmacology. (I) Quantification of saltation length and frequency following ML-7 administration (Ctrl, n = 10 cells from 3 hiPSC lines, 1–2 assembloid per line; TS, n = 13 cells from 3 hiPSC lines, 1–2 assembloid per line. Paired t-test, *P< 0.05, ****P< 0.0001). Bar charts: mean ± s.e.m. Boxplots: center, median; lower hinge, 25% quantile; upper hinge, 75% quantile; whiskers; minimum and maximum values. Scale bars: 10 μm (1B); 20 μm (1F). n.s.= not significant. For (3D), (3G), (3I), different shades of gray represent individual Ctrl hiPSC lines. Different shades of cyan represent individual TS hiPSC lines. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Dlxi1/2-eGFP+ migrating interneurons in intact forebrain assembloids. We placed forebrain assembloids in cell culture inserts for 7–14 days to achieve a flattened geometry that was more amenable patch-clamping migrating interneurons in intact assembloids (Figure S4G- S4H). Most interneurons fired single action potentials (Figure S4I), although interneurons with repetitive firing were also observed in both (Figure S4J). Ctrl and TS assembloids Importantly, transcriptional in signatures suggestive of changes in membrane potential and ion channel activity, we observed a difference in the resting membrane potential (RMP) in TS without changes in the capacitance and input resistance (Figure S4K, RMP: P< 0.0001). to
line with
the
Taken together, these results indicate that the TS Cav1.2 channel could alter specific transcriptional programs in hSS. Activity- dependent gene expression programs are tightly regulated by Cav1.2 and associated calcium signaling pathways (Dolmetsch, 2003; Yap and Greenberg, 2018) and have been previously shown to be altered in TS (Paşca et al., 2011; Servili et al., 2020). For example, cAMP response element-binding protein (CREB) is known to modulate gene expression via calcium influx-mediated phosphorylation (Wheeler et al., 2012). To probe potential changes in the calcium signaling-mediated gene expression pathways in TS hSS, we first quantified the phospho- CREB-ser133 (pCREBs133) in Dlxi1/2-eGFP+ levels of found interneurons and pCREBs133 were significantly higher in TS interneurons (P< 0.0001; Figure S5A–S5C). |
Next, we used a chronic depolarization paradigm where intact hSS (2–3 hSS per sample) were exposed to a Tyrode’s solution containing 67 mM KCl for 24 hours. Depolarization induces calcium influx through LTCC and non-LTCC pathways, receptors and calcium- such as NMDA permeable AMPA and Greenberg, 2018). To resolve the contribution of LTCC, we also separately applied nimodipine to depolarization (Figure S5D). This prior experiment revealed enhanced induction of activity-regulated genes such as FOS, NPAS4 in TS, which was prevented by nimodipine application (Figure S5E, FOS: P< 0.05 NPAS4: P= 0.053). Overall, these results suggested upregulation of transcriptional programs related to GABA neurotransmission, ion channel activity and membrane potential in TS interneurons
that
the
receptors
(Yap
potentially through altered recruitment of LTCC- transcriptional related, programs. activity-dependent
Modulation of GABAA receptor function rescues saltation frequency phenotype in TS cortical interneurons
Given that ambient GABA is known to influence interneuron migration (Manent et al., 2005; Cuzon et al., 2006; Heck et al., 2007; Bortone and Polleux, 2009) and because our results suggested gene expression changes in GABA receptors in TS hSS, we next asked whether changes in GABA neurotransmission could underlie the saltation frequency phenotype in TS. GABA has been shown to be depolarizing in developing interneurons (Bortone and Polleux, 2009). Indeed, application of 1 mM GABA to intact hSS labeled with an interneuron-specific GCaMP (Dlxi1/2b-GCaMP6s-P2A-GCaMP6s) revealed that GABA is depolarizing in hSS interneurons at this stage of development (Supplementary Video 3). We verified the effect of GABA application on calcium activity using ratiometric imaging (Fura-2) in plated hSS neurons (Figure 5A), and found that 100 μM GABA could induce a rise in calcium influx in a subset of neurons (Figure 5B). Importantly, the amplitude of the GABA-induced calcium rise was higher in TS neurons (Figure 5B, 5C, P< 0.01), suggesting enhanced GABA-induced calcium influx in TS interneurons. To further verify the enhanced effect of GABA on TS interneurons, we used whole-cell patch clamping and puff- applied 100 μM GABA (Figure 5D). We found larger GABA-induced currents in TS across various puff duration (Figure 5E–G, 5 ms: P< 0.05, 10 ms: P< 0.01, 20ms: P< 0.01, 50 ms: P< 0.01, 100 ms: P< 0.05, 200 ms: P< 0.05, 500 ms: P< 0.05). These results suggested enhanced GABA sensitivity in TS hSS neurons. Finally, we wondered if modulating GABA receptor activity could restore the saltation frequency phenotype of TS interneurons. We performed migration imaging in the presence of GABA-A receptor blocker picrotoxin (25 μM; Fig 5H) and discovered that this restored the saltation frequency but has no effect on saltation length of TS interneurons. Of note, saltation
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. |
All rights reserved. No reuse allowed without permission. Figure 4. Altered gene expression in TS interneurons. (A) Schematic illustrating the RNA-sequencing experimental design. (B) Heatmap of expression of the top differentially expressed gene (DEG) based on hierarchical clustering in hSS samples. Genes encoding for the GABA receptor subunits are in bold. (C) Boxplot plots showing differential expressed GABA receptor subunits in hSS samples (FDR <0.1). (D) Select genes from gene set enrichment analysis (GSEA) in hSS samples showing enriched terms in TS. Line indicates FDR = 0.05. (E) Dendrogram illustrating the significant Weighted Gene Co-expression Network Analysis (WGCNA) modules (Module M1-M7). (F) Enrichment for the genome-wide association study (GWAS) signal from ASD, ADHD, BD, MDD and SCZ as well as ASD- associated SFARI genes. Numbers and colors represent the enrichment score (FDR< 0.05). (G) Module M2 GO terms enriched in TS. Line indicates FDR = 0.05. (H) Module M4 GO terms enriched in TS. Line indicates FDR = 0.05. Boxplots: center, median; lower hinge, 25% quantile; upper hinge, 75% quantile; whiskers extend to ±1.5× interquartile range. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. frequency in Ctrl interneurons was also reduced (Fig 5I, Ctrl saltation frequency: P< 0.05; TS saltation is expected given the role of GABA receptors in interneuron migration. To further validate GABA receptor-mediated rescue of saltation frequency, we also used a second GABA-A blocker– bicuculine (50 μM; Figure S6A), and found a similar rescue of saltation frequency. but no effect on saltation length (Figure S6B, Ctrl saltation frequency: P< 0.01; TS saltation frequency: P< 0.05). In conclusion, we find that two distinct pathways underlie the inefficient migration phenotype in TS interneurons: (i) calcium influx through LTCC acutely modulates actomyosin function and regulates saltation length, and receptor signaling mediates saltation frequency and overall motility. frequency: P< 0.001), which
(ii) GABA
DISCUSSION
Various genetic and environmental factors have been proposed to perturb cortical GABAergic and development interneuron to neuropsychiatric including ASD, disease schizophrenia and epilepsy (Marín, 2012). The unique features of interneurons in primates (Silbereis et al., 2016; Krienen et al., 2019; Hodge et al., 2019) and the lack of patient- derived tissue preparations have limited our mechanistic understanding of the role of cortical interneurons in disease. We have previously developed forebrain assembloids (Birey et al., 2017) – a hiPSC-derived 3D culture platform where region-specific spheroids can be assembled to model the migration and functional integration of cortical interneurons. We have also in sought forebrain assembloids derived from patients with TS – a rare and highly penetrant form of monogenic ASD and epilepsy. |
TS is caused by the a gain-of-function mutation G406R alternatively spliced exon 8a of CACNA1C, which encodes for the α subunit of Cav1.2. Voltage-gated calcium channels have been implicated in regulating interneuron migration (Bortone and Polleux, 2009; Kamijo et al., 2018). Using TS forebrain assembloids, we discovered that the TS Cav1.2 channel leads to a compound the migration phenotype associated with saltatory mode of migration in cortical interneurons, namely decreased saltation length and increased saltation frequency (Birey et al., 2017). We now discovered that the decreased
lead
to study
interneuron migration
in
saltation length in TS interneurons is driven by an acute increase in calcium influx through LTCCs, followed by aberrant actomyosin function at the soma rear and an uncoupling of during soma increased nucleokinesis; saltation is mediated by enhanced GABA receptor signaling (Figure 6). rear-front
coordination
in contrast, in TS
the interneurons
frequency
Although significant advances have been made in deciphering both cell-extrinsic (e.g., chemoattractants, ambient neurotransmitters, activity) and (e.g., genetic cell-intrinsic programs) factors that coordinate interneuron migration (Peyre et al., 2015), a mechanistic understanding of how actively migrating interneurons integrate external cues with signal cytoskeletal transduction rearrangements relatively underexplored. Signaling localized to specific cellular compartments (e.g., growth cone and soma) may differentially affect the migratory behavior. For example, coordinated interaction between actin/myosin and microtubules is essential for a variety of dynamic cellular processes such as motility, division and adhesion (Dogterom and Koenderink, 2019). In the neuronal growth cone, actomyosin contractility is tightly coupled to microtubule assembly and disassembly to promote efficient turning and outgrowth (Coles and Bradke, 2015). A similar coordinated crosstalk has been extensively cortical interneuron migration, where a balance of pulling forces generated by microtubules in the leading branch followed by the pushing forces generated by the actomyosin contractility in the cell rear coordinates nucleokinetic movement (Bellion et al., 2005). We find that abnormal calcium influx mediated by TS Cav1.2 disrupts the coupling between pulling and pushing forces during nucleokinesis. This is possibly related to excess phosphorylation of MLC in the cell soma and aberrant cell rear contractility, ultimately resulting in decreased saltation length. Calcium signaling through LTCCs has been previously shown in neuronal growth cone turning and outgrowth (Forbes et al., 2012). pMLCs19 is also enriched in the cortical interneuron growth cone and it is possible that alterations in the growth cone dynamics contribute to TS migration phenotype. Additional aspects of intracellular calcium signaling might also contribute to the phenotype. pathways
and
remains
during
characterized
to regulate actomyosin
function
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. |
The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 5. Blocking GABAA receptors in TS interneurons restores saltation frequency but not saltation length. (A) Schematic illustrating Fura-2 calcium imaging and GABA pharmacology in plated hSS cells. (B) Changes in Fura-2 ratio in response to 100 μM GABA in plated hSS cells (day 99-d104). (C) Quantification of peak amplitudes corresponding to changes in Fura-2 ratio in response to 100 μM GABA (left: pooled, right: separated by line. Ctrl, n = 27 responding cells from 3 hiPSCs; TS, n = 35 responding cells. 2-3 hSS pooled per line. Mann-Whitney test, **P< 0.01). (D) Schematic illustrating the GABA puffing experiment on plated hSS cells (day 98-d102). (E) GABA current vs. Puff duration plots (Ctrl, n = 16 cells from 3 hiPSC lines; TS, n = 17 cells from 3 hiPSC lines. 2-3 hSS pooled per line. Two-tailed t-test per puff duration, *P <0.05, **P< 0.01. Two-way ANOVA for puff times (20 ms puff values excluded due to unequal number of cells in Ctrl vs TS groups), F15, 105 = 12.74, ****P< 0.0001. Curve was fitted using the Boltzmann sigmoid equation). (F) Representative GABA current traces in response to 10 ms GABA puff. (G) Quantification of GABA current amplitudes in in response to 10 ms GABA puff. (left: pooled, right: separated by line. Ctrl, n = 16 cells from 3 hiPSC lines; TS, n = 17 cells from 3 hiPSC lines. 2-3 hSS pooled per line. Mann-Whitney test, **P< 0.01) (H) Schematic illustrating the migration imaging and picrotoxin pharmacology. (I) Quantification of saltation length and frequency following picrotoxin (PTX) administration (Ctrl, n = 9 cells from 3 hiPSC lines, 1–2 assembloid per line; TS, n = 17 cells from 3 hiPSC lines, 1–2 assembloid per line. Paired t-test, *P< 0.05, ***P< 0.001). Bar charts: mean ± s.e.m. Boxplots: center, median; lower hinge, 25% quantile; upper hinge, 75% quantile; whiskers; minimum and maximum values. n.s.= not significant. For (5C), (5G), (5I), different shades of gray represent individual Ctrl hiPSC lines. Different shades of cyan represent individual TS hiPSC lines. Calcium-induced calcium release (CICR), where calcium from internal stores is released in response to calcium influx (Roderick et al., 2003), is critical in controlling repulsion/attraction in the growth cone (Gasperini et al., 2017). Aberrant CICR and/or ER calcium handling or altered physical interactions with ER-bound receptors in TS could also play a role in the cell rear contractility defect. Remarkably, we
acute modulation of either LTCCs (nimodipine, low
found
that
extracellular calcium levels, FPL 64176) or downstream cytoskeletal elements (ML-7) rescues the saltation length but not the saltation frequency phenotype in TS interneurons. This prompted us to explore alternative pathways that might frequency phenotype. Intracellular calcium entry through Cav1.2 gene expression changes type-specific manner (Greenberg et al., 1986; Morgan and Curran, 1986; Murphy et al., 1991). |
The TS
underlie
the
saltation
induces
activity-dependent
in a cell
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Cav1.2 channel has been previously shown to cause changes in the activity-dependent gene expression program (Paşca et al., 2011; Tian et al., 2014; Servili et al., 2020). In line with this, we observed enhanced phospho-CREB levels in TS interneurons. Indeed, RNA-sequencing of hSS revealed a robust enrichment of genes related to neurotransmission, membrane GABAergic potential and ion channel activity. The expression of GABA receptors has previously been shown to be regulated by LTCC activity (Saliba et al., 2009), CREB (Yu et al., 2017) and activity-dependent pathways (Roberts et al., 2005), suggesting that their upregulation in TS interneurons may follow CaV1.2 gain of function. GABA, GABA-A receptor activation and changes in excitability of interneurons have been associated with migration (Cuzon et al., 2006; Heck et al., 2007). In development, GABA can depolarize neurons leading to calcium transients (Bortone and Polleux, 2009). We found that relatively migrating TS depolarized membrane potentials, increased calcium transients, enhanced calcium influx in response to GABA and larger GABA currents. Moreover, we found that blocking GABA-A receptors frequency phenotype but left saltation length unchanged. Together, these data suggest a model in which ambient GABA acts in a paracrine/autocrine fashion in migrating interneurons and provides a pro-migratory signal; in TS, GABA sensitivity is enhanced due transcriptional transmission and programs excitability increased saltation frequency. interneurons had
rescued
the saltation
to altered
potassium transporter (KCC2), GABA becomes hyperpolarizing and promotes the termination of interneuron migration (Bortone and Polleux, that 2009). it developmental upregulation of KCC2 is regulated interneurons differentially in TS leading integration of TS to abnormal interneurons into cortical networks. Thus,
is
conceivable
In summary, our work provides novel regulation of insights human cortical interneuron development in the context of disease and suggest strategies for restoring these defects. into LTCC-mediated
in GABAergic
leading
to
Moving forward, a number of questions remain to be explored. Does the TS Cav1.2 directly induce upregulation of GABA receptors via calcium-mediated transcriptional programs, or is it an indirect consequence of other electrical changes or homeostatic mechanisms in TS? Do other by interneurons also play a role in the TS migration phenotype? Glycine, for example, has previously been shown to fine-tune actomyosin contractility in migrating interneurons (Avila et al., 2013). What are the consequences of the TS migration deficit at later stages of development? Would this defect in TS lead to changes in network integration cortical and maturation interneurons? |
As GABAergic cells undergo a shift in their GABA reversal potential later in the chloride development by upregulating
neurotransmitters
secreted
of
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ACKNOWLEDGMENTS
AUTHOR CONTRIBUTIONS
We thank the members of the Pasca lab, in particular J. Andersen, R. Agoglia, N. Thom, S. Kanton and T. Khan for insightful discussions, advice and support. This work was supported by National (R01 MH115012 (to S.P.P), K99 MH119319 (to F.B. ), respectively), National Institute of Mental Health Convergent Neuroscience Consortium (U01 MH115745) (to D.H.G. and S.P.P. ), the Stanford Human Brain Organogenesis Program in the Wu Tsai Neuroscience Institute (to S.P.P. ), the Kwan Funds (to S.P.P. ), the Senkut Research Fund (to S.P.P. ), the Stanford Maternal & Child Health Research (MCHRI) Postdoctoral Fellowship (to F.B. and to O.R.) the American Epilepsy Society Postdoctoral Research Fellowship (to F.B. ), the Stanford Science Fellows Program (to A.M.V. ), the Autism Science Foundation (ASF) and the Brain and Behavior Young Research Investigator award (to A.G.). S.P.P. is a New York Stem Cell Foundation (NYSCF) Robertson Stem Cell Investigator and a Chan Zuckerberg Initiative (CZI) Ben Barres Investigator. Institute of Mental Health
Institute
Foundation
(BBRF)
F.B. and S.P.P. conceived the project and designed experiments. F.B. and M.V.T. performed hCS and hSS differentiations. F.B. generated forebrain assembloids, performed migration and calcium imaging experiments. F.B. implemented the DeepLabCut pipeline and downstream analysis. F.B. and M.V.T. tracking analysis, performed migration and immunofluorescence staining quantifications. M.V.T. performed the RNA extraction and qPCRs. M.-Y.L. conducted and analyzed electrophysiology experiments. A.G. and D.H.G analyzed and interpreted the RNA- sequencing experiments. M.V.T and A.M.V conducted and analyzed western blotting experiments. O.R. developed MATLAB code for calcium transient detection and provided input for the development of other MATLAB routines and electrophysiology experiments. A.M.P. assisted with tissue preparation. F.B. wrote the manuscript with input from all authors. S.P.P. supervised all aspects of the work. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 6. Proposed model for distinct pathways mediating the migration phenotype of TS cortical interneurons. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. |
All rights reserved. No reuse allowed without permission. METHODS
Characterization and maintenance of hiPSCs hiPSC lines used in this study were validated using standardized methods as previously described (Paşca et al., 2011; Birey et al., 2017). Cultures were frequently tested for Mycoplasma and maintained free of Mycoplasma. A total of 8 control hiPS cell lines derived from fibroblasts collected from 8 healthy individuals and 3 hiPS cell lines derived from fibroblasts collected from 3 patients with TS were used for experiments (Supplementary Table 1). Approval was obtained from the Stanford IRB panel, and informed consent was obtained from all participants. Differentiation and assembly of hCS and hSS hCS and hSS differentiations from hiPSCs grown on MEF feeders were performed as previously described (Birey et al., 2017; Sloan et al., 2018). For hCS and hSS differentiations from feeder-free maintained hiPSCs, cells were maintained on vitronectin-coated plates (5 µg/mL, Thermo Fisher Scientific, A14700) in Essential 8 (E8) medium (Thermo Fisher Scientific, A1517001). Cells were passaged every 4-5 days with UltraPure 0.5 mM EDTA, pH 8.0 (Thermo Fisher Scientific, 15575020). Two days prior to aggregation, hiPS cells were exposed to 1% DMSO (Sigma-Aldrich, 472301) in E8 medium. For the generation of 3D neural spheroids, hiPSCs were incubated with Accutase (Innovative Cell Technologies, AT104) at 37 °C for 7 minutes and then single-cell dissociated. For aggregation into spheroids, 2.5-3 × 106 single cells were added per AggreWell 800 plate well in E8 medium supplemented with the ROCK inhibitor Y27632 (10 µM, Selleck Chemicals, S1049), centrifuged at 100g for 3 minutes and then incubated overnight at 37°C with 5% CO2. Next day, spheroids consisting of approximately 10,000 cells were dislodged from each microwell by pipetting the E8 medium up and down in the well with a P1000 pipette with a cut tip and transferred into ultra-low attachment plastic dishes (Corning, 3262) in Essential 6 (E6) medium (Thermo Fisher Scientific, A1516401) supplemented with the SMAD pathway inhibitors dorsomorphin (2.5 µM, Sigma-Aldrich, P5499) and SB-431542 (10 µM, R&D Systems, 1614). Feeder-free hCS differentiation was performed as previously described (Yoon et al., 2019) using the recipe variant without XAV299. Feeder-free hSS differentiation protocol was based on the Feeder-free hCS differentiation protocol with the following modifications (added on top of small molecules/growth factors specified in the hCS recipe): day 3–6; XAV-939 (2.5μM, Tocris, 3748), day 7–24; IWP-2 (2.5μM, Selleck Chemicals, S7085), day 13–24; SAG (100nM, EMD Millipore, 566660). To promote progenitor differentiation in hCS and hSS, BDNF (20 ng/mL) and NT-3 (20 ng/mL) were added starting at day 25 with media changes every other day. After day 43, media changes every 4-5 days were performed only with neural media (NM) without growth factors. Assembly of hCS and hSS to generate forebrain assembloids was performed as previously described (Birey et al., 2017; Sloan et al., 2018). |
Human primary tissue Human brain specimens were obtained under a protocol approved by the Research Compliance Office at Stanford University. PCW18 forebrain tissue was processed immediately after being received. hSS dissociation for monolayer culture hSS dissociation for monolayer culture was performed as previously described (Miura et al., 2020) with minor modifications. Briefly, 2–3 randomly selected hSS per hiPSC line were pooled in a 1.5-ml Eppendorf tube in the dissociation solution (10U/mL papain (Worthington Biochemical, LS003119), 1 mM EDTA, 10 mM HEPES (pH 7.4), 100 µg/mL BSA, 5 mM L-cysteine and 500 µg/mL DNase I (Roche, 10104159001)). Cells were incubated at 37°C for 20-25 minutes and gently shaken once midway through. Papain was inactivated with 10% FBS in NM, and hSS were then gently triturated with a P1000 pipette. Samples were centrifuged once at 1,200 rpm for 4 minutes, filtered with a 70-µm Flowmi Cell Strainer (Bel-Art, H13680-0070) and then suspended in NM with BDNF (20 ng/mL) and NT-3 (20 ng/mL). 0.75- 1 × 105 cells were seeded on 15-mm round coverslips (Warber Instruments, 64-0713). Coverslips were coated with approximately 0.001875% polyethylenimine (Sigma-Aldrich, 03880) for 0.5-1 hour at room temperature, washed 3-4 times and air-dried prior to plating. Plated cells were cultured in NM supplemented with BDNF (20 ng/mL) and NT-3 (20 ng/mL) with media changes every other day. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Real-time qPCR RNA extraction was performed using RNeasy Mini Kit (Qiagen, 74106). Genomic DNA was removed prior to cDNA synthesis using DNase I, Amplification Grade (Thermo Fisher Scientific, 18068-015). Reverse transcription was performed using the SuperScript III First-Strand Synthesis SuperMix for qRT– PCR (Thermo Fisher Scientific, 11752250). qPCR was performed using the SYBR Green PCR Master Mix (Thermo Fisher Scientific, 4312704) with a ViiA7 Real-Time PCR System (Thermo Fisher Scientific, 4453545). Primers used in this experiment are listed in Supplementary Table 2. Activity-dependent gene expression assay hSS were incubated in low-potassium Tyrode’s solution (5 mM KCl: 129 mM NaCl, 2 mM CaCl2, 1 mM MgCl2, 30 mM glucose, 25 mM HEPES, pH 7.4) supplemented either with DMSO (vehicle) or Nimodipine (5μM) for 5 days. On day 4, A subset of hSS in each group was depolarized for 24 hours with high- potassium Tyrode’s solution (high-KCl) (67 mM KCl: 67 mM NaCl, 2 mM CaCl2, 1 mM MgCl2, 30 mM glucose and 25 mM HEPES, pH 7.4). RNA extraction, cDNA amplification and qPCR were performed as described above. Primers used in this experiment are listed in
Immunofluorescence staining and quantification hSS were fixed in 4% PFA–PBS for 2 hours at 4°C, washed in PBS once and transferred to 30% sucrose– PBS for 2–3 days. |
They were next embedded in optimal cutting temperature (OCT) compound (Tissue- Tek OCT Compound 4583, Sakura Finetek) and 30% sucrose–PBS (1:1) for cryosectioning (18-20 µm- thick sections) using a Leica Cryostat (Leica, CM1850). Plated hSS cultures on glass coverslips were fixed in 4% PFA at room temperature for 10 minutes and washed with then rinsed twice for 5 minutes with PBS. For immunofluorescence staining, cryosections were washed with PBS and blocked in 10% normal donkey serum (NDS, Millipore Sigma, S30-M) and 0.3% Triton X-100 (Millipore Sigma, T9284-100ML) diluted in PBS for 1 hour at room temperature. Sections were then incubated overnight at 4 °C with primary antibodies diluted in PBS containing 10% NDS. PBS was used to wash away excess primary antibodies, and the cryosections were incubated with secondary antibodies in PBS containing 10% NDS for 1 h. For live staining of Cav1.2, plated hSS cells infected with LV-Dlxi1/2b-mScarlet (Vector Builder) were incubated in Anti-Cav1.2 antibody conjugated with DyLight-488 (1:100, StressMarq Biosciences, SMC-300D-DY488) for 30minutes, rinsed three times with NM and fixed in 4% PFA at room temperature for 10 minutes. The following primary antibodies were used for staining: anti-phospho-CREB (ser133) antibody (1:1000 dilution, rabbit, Millipore Sigma, 06-519), Phospho-Myosin Light Chain 2 (Ser19) antibody (1:200 dilution, rabbit, Cell Signaling, 3671S), anti-MAP2 antibody (1:10,000 dilution, guinea pig, Synaptic Systems, 188 004), anti-GFP antibody (1:10,000 dilution, chicken, GeneTex, GTX13970) and anti-Cav1.2 antibody conjugated with DyLight-488 (1:100 dilution, StressMarq Biosciences, SMC-300D-DY488). For fluorescence intensity quantifications, 16-bit confocal images (Leica TCS SP8 confocal microscope) were collected and processed in Fiji (ImageJ, version 2.1.0, NIH) using the same parameters across conditions per experiment. Regions-of-interest (ROIs) were defined using ROI manager in Fiji either by DAPI+ nuclei (pCREBs133 quantifications; Figure S5C) or MAP2+ soma (pMLC2s19 quantifications; Figure 3G) and mean gray values were collected per ROI. Gray values were collected from a cell-absent ROI and subtracted from cell-ROI values for background normalization per field. All values were subsequently normalized to mean Ctrl mean gray value. Viral labeling of hSS hSS were transferred to a 1.5-ml Eppendorf tube containing 200 µl NM and incubated with virus overnight at 37°C with 5% CO2. The next day, fresh NM (800 µl) was added. The following day, hSS were transferred into fresh culture medium in ultra-low attachment plates (Corning, 3471, 3261). The expression was evident 1–2 weeks after infection across viral reporters used throughout the study. Migration imaging and analysis Live-cell migration analysis and pharmacology were performed as previously described (Birey et al., 2017; Sloan et al., 2018). Briefly, hSS labeled with LV-Dlxi1/2b-eGFP (gift from J. Rubenstein) was
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. |
The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. assembled with hCS between d60 and d100 to generate forebrain assembloids. Forebrain assembloids after 15-45 days after assembly were transferred to a Corning 96-well microplate (Corning, 4580; one assembloid per well) in 200µL NM and incubated in an environmentally controlled chamber (Oko Lab, H201 T Unit-BL with 5% CO2/air perfusion) for 15–30 minutes before imaging on a Leica TCS SP8 confocal microscope. GFP+ interneurons from 6-8 assembloids were imaged (10x objective, 0.75x zoom) in a given session at a rate of 18-20 minutes per volume (~200µm volume per assembloid) for 18-22 hours. For pharmacology experiments, following the baseline imaging, half-volume media changes with fresh media containing the drug at twice the working concentration were carefully performed without disturbing the assembloids under the hood and transferred back to the confocal. The fields were adjusted for minor shifts pre-acquisition. For DeepLabCut analysis, small-volume, high-resolution migration imaging was performed (20x objective, 2-3x zoom) at a rate of approximately 45 seconds per volume for 3-6 hours. For migration imaging in primary fetal cortical slices, 400μm vibrotome-section cortical slices were placed on culture inserts with 0.4 mm pore size (Corning, 353090), infected with LV-Dlxi1/2-eGFP and imaged 10-14 days after infection. 4-5 fields were imaged per slice (10x objective, 0.75x zoom). Saltation length and saltation frequency quantifications were performed as previously described (Birey et al., 2017). For DeepLabCut analysis, imaging planes were cropped to contain individual saltations from single interneurons. Following drift correction (Linear Stack Alignment plugin) and smoothing (Gaussian Blur 3D plugin) in Fiji, timeseries stacks (250-400 frames in total, 45 seconds per frame) were converted to .avi files in Fiji. DeepLabCut (v2.16, v2.2b6) with GPU support (NVIDIA GeForce GTX 1080 Ti) was installed as per developers’ instructions. Config files were edited to label ROIs. 19 training frames that were most distinct across each timeseries was automatically extracted using the k-means algorithm (cluster step = 1, network = resnet_50, augmentation = imgaug) and ROIs were manually annotated in each training frame per video. Network was then trained for up to 100,000 iterations or until the loss plateaued. Network was evaluated and videos were analyzed with filtered predictions. Occasional ROI mismatch was detected by outlier likelihood measure and smoothened by the moving median function in MATLAB (k= 1000-2000). Tracking fidelity was then manually validated by converting X and Y pixel coordinates to timeseries ROIs per movie in Fiji. X and Y pixel coordinates were converted to Euclidian distances in microns to calculate saltation length as validation of the TS phenotype. Pearson’s correlation coefficient between soma front and rear was computed using corrcoef function in MATLAB. |
Calcium imaging For GCaMP calcium imaging in intact hSS, hSS were labeled with AAV-DJ-mDlx-GCaMP6f-Fishell-2 (Addgene, 83899) and placed in a well of a Corning 96-well microplate (Corning, 4580) in NM and imaged using a ×10 objective on a Leica TCS SP8 confocal microscope. GCaMP6 was imaged at a frame rate of 2 frames per second. Mean gray values were collected from GCaMP6+ soma with Fiji (ImageJ, version 2.1.0, NIH). Spontaneous calcium transients were detected using a custom-written MATLAB routine (version R2019b, 9.4.0, MathWorks). Mean gray values were transformed to relative changes in fluorescence: dF/F(t) = (F(t)−F0)/F0, where F0 represents average grey values of the time series of each ROI. Calcium transients were detected as dF/F(t) crossed the threshold of 2 median absolute deviations. For the GCaMP calcium imaging in hSS coupled with GABA application, intact hSS labeled with Dlxi1/2b-mScarlet-P2A-GCaMP6s (Vectorbuilder) was placed in a perfusion chamber (RC-20, Warner instruments) in Tyrode’s solution (5 mM KCl: 129 mM NaCl, 2 mM CaCl2, 1 mM MgCl2, 30 mM glucose, 25 mM HEPES, pH 7.4). After baseline imaging of mScarlet and GCaMP signals using a 10x objective on a Leica TCS SP8 confocal microscope at the frame rate of 1 frame per second, Tyrode’s solution supplemented with 1 mM GABA was perfused. GCaMP signal was normalized to mScarlet signal from selected ROIs. dF/F was calculated where F represents the 5th percentile of mean grey values per ROI. Fura-2 calcium imaging on plated hSS was performed as previously described (Khan et al., 2020). Briefly, cells were loaded with 1 μM Fura-2 acetoxymethyl ester (Invitrogen, F1221) for 30 minutes at 37°C in NM medium, washed with NM medium for 5 minutes and then transferred to a perfusion chamber (RC-20, Warner instruments) in Tyrode’s solution (5 mM KCl: 129 mM NaCl, 2 mM CaCl2, 1 mM MgCl2, 30 mM glucose, 25 mM HEPES, pH 7.4) on the stage of an inverted fluorescence microscope (Eclipse TE2000U; Nikon). After 0.5 minute of baseline imaging, low-potassium Tyrode’s solution supplemented with 100μM GABA was perfused for 1 minute. Imaging was performed at room temperature (∼25 °C) on
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. an epifluorescence microscope equipped with an excitation filter wheel and an automated stage. Openlab software (PerkinElmer) and IGOR Pro (version 5.1, WaveMetrics) were used to collect and quantify time- lapse excitation 340 nm/380 nm ratio images, as previously described (Barreto-Chang and Dolmetsch, 2009). Only the cells that had a response higher than 0.1 340 nm/380 nm ratio were included in the analysis. Western blotting Whole cell protein lysates from hCS and hSS samples from control and TS hiPSCs were prepared using SDS Buffer (1.5% SDS, 25 mM Tris pH 7.5). |
Briefly, 75 µL of SDS Buffer were added to two spheres in a 1.5 mL tube. Samples were sonicated for a total of 7-9 seconds using an ultrasonicator (Qsonica Q500 sonicator; pulse: 3 seconds on, 3 seconds off; Amplitude: 20%) using a small tip attachment until samples were no longer viscous. Protein concentrations were quantified using the bicinchoninic Acid (BCA) assay (Pierce, Thermo Fisher 23225). Protein samples were normalized and prepared in 1X LDS Sample Buffer (NuPAGE LDS Sample Buffer, Thermo Fisher NP007) and 0.1M DTT. Samples (12-15 µg per sample per lane) were loaded and run on a 10%-20% tricine gel (Novex 10%-20% Tricine Protein Gel, Thermo Fisher Scientific), and transferred onto a polyvinylidene fluoride (PVDF) membrane (Immobilon-PSQ, EMD Millipore). Membranes were blocked with 5% BSA (Sigma) in TBST for 1 h at room temperature (RT) and incubated with primary antibodies against β-actin (mouse, 1:50,000, Sigma, A5316) and Myosin Light Chain 2 (rabbit, 1:1,000, Cell Signaling, 8505S), and Phospho-Myosin Light Chain 2 (Ser19) (rabbit, Cell Signaling, 1:1,000, 3671S) antibodies for 72 h at 4 °C. Membranes were washed three times with TBST and then incubated with near-infrared fluorophore-conjugated species-specific secondary antibodies: Goat Anti-Mouse IgG Polyclonal Antibody (IRDye 680RD, 1:10,000, LI-COR Biosciences, 926-68070) or Goat Anti-Rabbit IgG Polyclonal Antibody (IRDye 800CW, 1:10,000, LI-COR Biosciences, 926-32211) for 1 h at RT. Following secondary antibody application, membranes were washed three times with TBST, once with TBS, and then imaged using a LI-COR Odyssey CLx imaging system (LI- COR). Protein band intensities were quantified using Image Studio Lite (LI-COR) with built-in background correction and normalization to β-actin controls. Whole-cell patch clamping For patch-clamp recordings, cells were identified as Dlx1/2::EGFP+ with an upright slice scope microscope (Scientifica) equipped with Infinity2 CCD camera and Infinity Capture software (Teledyne Lumenera). Recordings were done with borosilicate glass electrodes with a resistance of 7–10 MΩ. For all the experiments, external solution of BrainPhys neuronal medium (STEMCELL Technologies, 05790) was used. Data were acquired with a MultiClamp 700B Amplifier (Molecular Devices) and a Digidata 1550B Digitizer (Molecular Devices), low-pass filtered at 2 kHz, digitized at 20 kHz and analyzed with pCLAMP software (version 10.6, Molecular Devices). Cells were given a -10mV hyperpolarization (100ms) every 10s to monitor input resistance and access resistance, and cells were not included for analysis if they had a change > 30%. The liquid junction potential was calculated using JPCalc66, and membrane voltage was manually corrected with an estimated −15-mV liquid junction potential for current clamp recordings. For calcium current recording, calcium was replaced by barium in the external solutions. Monolayer neurons at day 140 were recorded after lentivirus-Dlx1/2::EGFP labeling and EGFP+ neurons were recorded. |
Cells were held at –70 mV in voltage-clamp and depolarizing voltage steps (500 ms) were given with an increment of 10 mV. The external solution contained 100 mM NaCl, 3 mM KCl, 2 mM MgCl2, 20 mM BaCl2, 25 mM TEA-Cl, 4 mM 4-Aminopyridine, 10 mM HEPES, 20 mM glucose, pH 7.4 with NaOH, 300 mOsm. Internal solution contained 110 mM CsMethylSO3, 30 mM TEA-Cl, 10 mM EGTA, 4 mM MgATP, 0.3 mM Na2GTP, 10 mM HEPES, 5 mM QX314-Cl, pH 7.2 with CsOH, 290 mOsm. For recordings from migrating interneurons, forebrain assembloids were placed on cell culture inserts with 0.4 mm pore size (Corning, 353090). Fusions on inserts were carefully dislodged and transferred into the recording chamber. Dlx1/2-eGFP+ cells migrated into hCS part and near the fusion border of hCS and hSS were recorded. Resting membrane potential was recorded immediately after whole cell. Brainphys was used as external solution. The internal solution contained 127 mM K-gluconate, 8 mM NaCl, 4 mM MgATP, 0.3 mM Na2GTP, 10 mM HEPES and 0.6 mM EGTA, pH adjusted to 7.2 with KOH (290 mOsm). bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. For GABA puffing experiment, plated hSS cells were recorded in BrainPhys neuronal medium in the presence of TTX (1 μM), NQBX (10 μM) and APV (50 μM), with internal solution containing high chloride as follows: 135 mM CsCl, 4 mM MgATP, 0.3 mM Na2GTP, 10 mM HEPES and 0.6 mM EGTA, pH adjusted to 7.2 (290 mOsm). GABA (100 μM diluted in Brainphys, Tocris, Cat. No. 0344) was filled in a pipette (puffing pipette), placed at a distance of 20 μm away from the soma of the recorded cell, and puffed with different durations (3ms, 5ms, 10ms, 20ms, 50ms, 100ms, 200ms, 500ms) controlled by a picospritzer (General Valve, Picospritzer II). Cells were held at –60 mV under voltage clamp to record GABA-induced inward current. For puff duration of 3 ms, 5 ms, 10 ms, GABA current was recorded once every 10 s, and at least 5 trials were averaged; for puff durations over 20 ms, due to the large responses, cells were recorded at least 2 min after the previous puffing to prevent run down. One control line and one TS line were recorded for the same day using the same puffing pipette with the same aliquot of GABA to minimize the variability of puffing pipette tip size. RNA-sequencing RNA was extracted from hSS and hCS using the Quick-RNA Miniprep kit (Zymo Research, R1054). For library construction, non-stranded cDNA libraries were constructed using poly(A) mRNA enrichment and the NEBNext Ultra™ II RNA Library Prep Kit for Illumina following manufacturer’s instructions (NEB, E7770S). RNA and library qualities were confirmed using Qubit Flourometric Quantification (Invitrogen, Q33239) and fragment analysis (Agilent). 150bp pair-end sequencing was performed using NovaSeq 6000 Sequencing system (Illumina). |
Using STAR (v2.5.2b) (Dobin et al., 2013), reads were mapped to hg38 with Gencode v25 annotation. Sample identity was verified by descent (IBD) (PLINK v1.09) (Purcell et al., 2007) based on SNPs called from the aligned reads using the GATK Haplotype caller (v3.3) (McKenna et al., 2010). Sex was verified by expression of Y chromosome genes. Regional Identity of hCS and hSS was verified using regional marker genes (hSS: DLX, NKX2-1 and SLC32A1; hCS: SLC17A6, SLC17A7, and NEUROD2). Samples with marker genes belonging to different regions higher than two standard deviations from the mean were removed. Picard sequencing metrics (http://broadinstitute.github.io/picard/, v2.5.0) were summarized using the first 6 principal components (seqPCs) and were included in the model to control for technical variation resulting from the sequencing. Outlier samples (standardized sample network connectivity Z scores < −2) were removed. In total, 29 samples (hCS: 4 control and 6 TS; hSS: 8 control and 11 TS) with 17,963 expressed genes (>10 read in 50% of samples) were used for downstream analysis. Gene expression was normalized using the trimmed mean of M values (TMM) method from the edgeR package (Robinson et al., 2010). Differential expression was calculated using voom method from the limma package. The hiPSC line was used as a blocking factor in the model using the duplicateCorrelation function (Ritchie et al., 2015). The model used was ~ DX+Region+Differentiation+Sex+seqPC1:6. Genes with FDR < 0.1 were considered to be differentially expressed. 60 and 1830 differentially expressed genes were found in hSS and hCS respectively. Gene set enrichment analysis (GSEA) was performed on all genes ranked by log2 fold change using the fgsea package (v1.10.1) (Korotkevich et al., 2021) using with 1,000,000 permutations, a minimal set size of 30 and a maximal set size of 500. Gene ontology (GO) sets (v7.0) were downloaded from http://software.broadinstitute.org/gsea/msigdb/. GO terms with FDR < 0.05 were considered to be significant. from
the
limma package
Weighted gene network analysis (WGCNA; v.1.68) (Langfelder and Horvath, 2008) was performed on the hSS data using the following parameters: soft power = 9, minimal module size = 160, deep split = 4, cut height for creation of modules = 0.9999 and cut height for merging modules of 0.2. The module eigengene (first principal component of the module) were then tested for association with TS using a linear model. Modules were considered to be associated with TS when FDR < 0.05. For modules associated with TS, biological process and molecular function gene ontology enrichment, for modules associated with TS, was performed using cluterProfiler (v3.12.0) (Yu et al., 2012) with default parameters. Modules significantly associated with TS were tested for enrichment for common variation associated with ASD (Grove et al., 2019), SCZ (Pardiñas et al., 2018), ADHD (Demontis et al., 2019), MDD (Howard et al., 2019) and BD (Mullins et al., 2020). |
SNPs with 10 kb of a gene were assigned to that gene and were used. SNP heritability was calculated using a stratified LDscore regression (v1.0.0) (Finucane et al., 2015). Enrichment was calculated as the proportion of SNP heritability accounted for by the genes in
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. the module divided by the proportion of total SNPs within the module. Enrichment for ASD genes was performed using a fisher exact test on high confidence gene (gene score < 2 or syndromic genes) from SFARI (https://gene.sfari.org/ database/gene-scoring/). Enrichments with FDR < 0.05 were considered significant. Statistics Data are presented as mean ± s.e.m. or box plots showing maximum, third quartile, median, first quartile and minimum values. Raw data were tested for normality of distribution, and statistical analyses were performed using paired or unpaired t-tests (two-tailed), Mann–Whitney tests, one-way ANOVA tests with multiple comparison and two-way ANOVA tests. Sample sizes were estimated empirically. MATLAB routines and Fiji macros were used for migration and calcium imaging analysis. GraphPad Prism (version 8.4.2-9.0.0) and MATLAB (version R2019b, 9.4.0, MathWorks) were used for statistical analyses. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Top: representative image of two migrating Dlxi1/2-eGFP interneurons (dashed lines: Patch pipette; Asterisk: nucleokinesis. Figure S1. Cav1.2 expression and isoform differences in TS hSS. (A) Live staining of Dlxi1/2b-mScarlet-labeled interneurons with Cav1.2-DyLight-488 on plated hSS cells. (B) qPCR quantification of CACNA1C exon 8a vs 8 isoform expression in hSS over time (Ctrl, n = 7 (day 25), 4 (day 60), 5 (day 100) samples from 5 (day 25), 4 (day 60, day 100) hiPSC lines; TS, n = 7 (day 25), 3 (day 60), 5 (day 100) samples from 3 (day 25, day 60, day 100) hiPSC lines. 2-3 hSS pooled per sample. (C) Quantification of distance traveled by soma front and rear per saltation. (Ctrl, n = 16 cells from 3 hiPSC lines, 1-3 assembloid per line; TS, n = 21 cells from 3 hiPSC lines, 1-3 assembloid per line. Two-tail t-test, ****P<0.0001). Scale bar: 20 μm (S1A), insert, 10 μΜ. For (S1B), (S1C), different shades of gray represent individual Ctrl hiPSC lines. Different shades of cyan represent individual TS hiPSC lines. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. |
Cells undergoing nucleokinesis based on morphology was omitted for patching), Bottom: Trace from a typical single-spiking
Figure S2. Termination of interneuron migration at 0.5mM [Ca2+]e (A) Schematic illustrating the migration imaging experiment in 0.5 mM extracellular calcium [Ca2+]e. (B) Quantification of saltation length in 0.5 mM extracellular calcium [Ca2+]e (Ctrl, n = 7 cells from 3 hiPSC lines; TS, n = 4 cells from 1 hiPSC line, 1–2 assembloid per line. Paired t-test, *P< 0.05, ***P<0.001). Bar charts: mean ± s.e.m. For (S2B), different shades of gray represent individual Ctrl hiPSC lines. Different shades of cyan represent individual TS hiPSC lines. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. migrating interneuron. Figure S3. Western blot quantification split by run. (A) Values normalized to mean Ctrl value per run. Range: mean ± s.e.m. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure S4. Functional validation of RNA-sequencing. (A) Select genes from gene set enrichment analysis (GSEA) in hCS samples showing enriched terms in TS. Line indicates FDR = 0.05. (B) Coefficient of determination (adjusted R2) of WGCNA modules that are significantly associated with TS (FDR< 0.05). (C) Module M2 hub genes. (D) Schematic illustrating calcium imaging experiment in intact hSS using interneuron- specific calcium indicator Dlx5/6-GCaMP6f. (E) Heatmap showing representative calcium transients. (F) Quantification of calcium transient frequency (Ctrl, n = 12 cells from 4 hiPSC lines; TS, n = 19 cells from 3 hiPSC lines, 2–3 hSS per line. Man- Whitney test, **P< 0.01). (G) Schematic illustrating whole-cell patch clamping experiment of migrating interneurons in intact assembloids (day 97–108, at 19–30 days after assembly). (H) Representative images of assembloids cultured on trans-wells prior to patch clamping. (I) Top: representative image of two migrating Dlxi1/2-eGFP interneurons (dashed lines: Patch pipette; Asterisk: nucleokinesis. Cells undergoing nucleokinesis based on morphology was omitted for patching), Bottom: Trace from a typical single-spiking migrating interneuron. (J) Representative traces of repetitive-spiking migrating interneurons. (K) Quantification of resting membrane potential (RMP), capacitance and input resistance in migrating interneurons (Ctrl, n = 12 cells from 3 hiPSC lines, 1 assembloid per line; TS, n = 14 cells from 3 hiPSC lines, 1 assembloid per line. Two-tailed t-test, ****P< 0.0001). Bar charts: mean ±s.e.m. Boxplots: center, median; lower hinge, 25% quantile; upper hinge, 75% quantile; whiskers; minimum and maximum values. |
Scale bar: 1 mm (S4H); 20 μm (S4I)). For (S4D), (S4K), different shades of gray represent individual Ctrl hiPSC lines. Different shades of cyan represent individual TS hiPSC lines. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure S5. Alteration in activity-dependent gene expression in TS interneurons. (A) Schematic illustrating quantification of pCREBs133 levels in Dlxi1/2b-GFP+ hSS interneurons (day 90). (B) Representative imaging of pCREBs133 immunocytochemistry in in Dlxi1/2b-GFP+ hSS cyro-sections. (C) Quantification of pCREBs133 intensity of Dlxi1/2b-GFP+ hSS interneurons (left: aggerate, right: split by line. Ctrl, n = 55 cells from 2 hiPSC lines; TS, n = 70 cells from 3 hiPSC lines, 1 hSS per line. Mann-Whitney test, ***P<0.001). (D) Schematic illustrating chronic depolarization (24 hours) paradigm used to test LTCC-dependent immediate early gene (IEG) induction. (E) qPCR quantification of FOS and NPAS4 gene expression in response to chronic depolarization in the presence or absence of nimodipine (Ctrl, n = 5 samples from 4 hiPSC lines; TS, n = 4-5 samples, from 3 hiPSC lines, 2-3 hSS pooled per line per condition. One-way ANOVA, *P<0.05). Bar charts: mean ±s.e.m. Boxplots: center, median; lower hinge, 25% quantile; upper hinge, 75% quantile; whiskers; minimum and maximum values. Scale bars: S5B, 20μM, inset, 10μM. For (S5C), (S5E), different shades of gray represent individual Ctrl hiPSC lines. Different shades of cyan represent individual TS hiPSC lines. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure S6. Rescue of saltation frequency by bicuculine. (A) Schematic illustrating the migration imaging and bicuculine pharmacology. (B) Quantification of saltation length and frequency following bicuculine (BIC) administration (Ctrl, n = 18 cells from 3 hiPSC lines; TS, n = 16 cells from 3 hiPSC lines, 1 assembloid per line. Paired t-test, *P< 0.05, **P< 0.01). Bar charts: mean ±s.e.m. n.s.= not significant. For (S6B), different shades of gray represent individual Ctrl hiPSC lines. Different shades of cyan represent individual TS hiPSC. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Supplementary Table 1. hiPSC lines used in various experiments
Supplementary Table 2. Primer sequences. Supplementary Table 3. Samples used for the RNA-sequencing. Supplementary Table 4. List of differentially expressed genes (hCS and hSS, Ctrl vs. TS)
Supplementary Table 5 List of enriched GSEA terms (hCS and hSS, Ctrl vs. TS)
Supplementary Table 6. |
List of enriched WGCNA GO terms (hSS, Ctrl vs. TS)
Supplementary Video 1. DeepLabCut pipeline for markerless tracking of subcellular ROIs
Supplementary Video 2. Representative videos showing soma rear-front uncoupling in a TS interneuron. Supplementary Video 3. Representative video showing the depolarizing effect of 1 mM GABA on hSS interneurons labeled with Dlxi1/2b-mScarlet-P2A-GCaMP6s. DATA AVAILABILITY Gene expression data is available in the Gene Expression Omnibus (GEO) under accession number GSE175898. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. REFERENCES
Anderson, S.A., Eisenstat, D.D., Shi, L., and Rubenstein, J.L.R. (1997). Interneuron Migration from Basal Forebrain to Neocortex: Dependence on Dlx Genes. Science 278, 474–476. Avila, A., Vidal, P.M., Dear, T.N., Harvey, R.J., Rigo, J.-M., and Nguyen, L. (2013). Glycine Receptor α2 Subunit Activation Promotes Cortical Interneuron Migration. Cell Reports 4, 738–750. Barreto-Chang, O.L., and Dolmetsch, R.E. (2009). Calcium Imaging of Cortical Neurons using Fura-2 AM. JoVE (Journal of Visualized Experiments) e1067. Bellion, A., Baudoin, J.-P., Alvarez, C., Bornens, M., and Métin, C. (2005). Nucleokinesis in Tangentially Migrating Neurons Comprises Two Alternating Phases: Forward Migration of the Golgi/Centrosome Associated with Centrosome Splitting and Myosin Contraction at the Rear. J. Neurosci. 25, 5691–5699. Bers, D.M. (2002). Cardiac excitation–contraction coupling. Nature 415, 198–205. Bhat, S., Dao, D.T., Terrillion, C.E., Arad, M., Smith, R.J., Soldatov, N.M., and Gould, T.D. (2012). CACNA1C (Cav1.2) in the pathophysiology of psychiatric disease. Progress in Neurobiology 99, 1–14. Birey, F., Andersen, J., Makinson, C.D., Islam, S., Wei, W., Huber, N., Fan, H.C., Metzler, K.R.C., Panagiotakos, G., Thom, N., et al. (2017). Assembly of functionally integrated human forebrain spheroids. Nature 545, 54–59. Bortone, D., and Polleux, F. (2009). KCC2 Expression Promotes the Termination of Cortical Interneuron Migration in a Voltage-Sensitive Calcium-Dependent Manner. Neuron 62, 53–71. Buchsbaum, I.Y., and Cappello, S. (2019). Neuronal migration in the CNS during development and disease: insights from in vivo and in vitro models. Development 146, dev163766. Coles, C.H., and Bradke, F. (2015). Coordinating Neuronal Actin–Microtubule Dynamics. Current Biology 25, R677–R691. Cuzon, V.C., Yeh, P.W., Cheng, Q., and Yeh, H.H. (2006). Ambient GABA Promotes Cortical Entry of Tangentially Migrating Cells Derived from the Medial Ganglionic Eminence. Cerebral Cortex 16, 1377– 1388. Demontis, D., Walters, R.K., Martin, J., Mattheisen, M., Als, T.D., Agerbo, E., Baldursson, G., Belliveau, R., Bybjerg-Grauholm, J., Bækvad-Hansen, M., et al. (2019). Discovery of the first genome-wide significant risk loci for attention deficit/hyperactivity disorder. |
Nature Genetics 51, 63–75. Dobin, A., Davis, C.A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, P., Chaisson, M., and Gingeras, T.R. (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21. Dogterom, M., and Koenderink, G.H. (2019). Actin-microtubule crosstalk in cell biology. Nat. Rev. Mol. Cell Biol. 20, 38–54. Dolmetsch, R. (2003). Excitation-Transcription Coupling: Signaling by Ion Channels to the Nucleus. Sci. STKE 2003, pe4–pe4. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Finucane, H.K., Bulik-Sullivan, B., Gusev, A., Trynka, G., Reshef, Y., Loh, P.-R., Anttila, V., Xu, H., Zang, C., Farh, K., et al. (2015). Partitioning heritability by functional annotation using genome-wide association summary statistics. Nature Genetics 47, 1228–1235. Forbes, E.M., Thompson, A.W., Yuan, J., and Goodhill, G.J. (2012). Calcium and cAMP Levels Interact to Determine Attraction versus Repulsion in Axon Guidance. Neuron 74, 490–503. Gasperini, R.J., Pavez, M., Thompson, A.C., Mitchell, C.B., Hardy, H., Young, K.M., Chilton, J.K., and Foa, L. (2017). How does calcium interact with the cytoskeleton to regulate growth cone motility during axon pathfinding? Molecular and Cellular Neuroscience 84, 29–35. Greenberg, M.E., Ziff, E.B., and Greene, L.A. (1986). Stimulation of neuronal acetylcholine receptors induces rapid gene transcription. Science 234, 80–83. Grove, J., Ripke, S., Als, T.D., Mattheisen, M., Walters, R.K., Won, H., Pallesen, J., Agerbo, E., Andreassen, O.A., Anney, R., et al. (2019). Identification of common genetic risk variants for autism spectrum disorder. Nature Genetics 51, 431–444. Heck, N., Kilb, W., Reiprich, P., Kubota, H., Furukawa, T., Fukuda, A., and Luhmann, H.J. (2007). GABA-A Receptors Regulate Neocortical Neuronal Migration In Vitro and In Vivo. Cerebral Cortex 17, 138–148. Hodge, R.D., Bakken, T.E., Miller, J.A., Smith, K.A., Barkan, E.R., Graybuck, L.T., Close, J.L., Long, B., Johansen, N., Penn, O., et al. (2019). Conserved cell types with divergent features in human versus mouse cortex. Nature 573, 61–68. Howard, D.M., Adams, M.J., Clarke, T.-K., Hafferty, J.D., Gibson, J., Shirali, M., Coleman, J.R.I., Hagenaars, S.P., Ward, J., Wigmore, E.M., et al. (2019). Genome-wide meta-analysis of depression identifies 102 independent variants and highlights the importance of the prefrontal brain regions. Nature Neuroscience 22, 343–352. Inamura, N., Kimura, T., Tada, S., Kurahashi, T., Yanagida, M., Yanagawa, Y., Ikenaka, K., and Murakami, F. (2012). Intrinsic and Extrinsic Mechanisms Control the Termination of Cortical Interneuron Migration. J. Neurosci. 32, 6032–6042. Kamijo, S., Ishii, Y., Horigane, S.-I., Suzuki, K., Ohkura, M., Nakai, J., Fujii, H., Takemoto-Kimura, S., and Bito, H. (2018). |
A Critical Neurodevelopmental Role for L-Type Voltage-Gated Calcium Channels in Neurite Extension and Radial Migration. J. Neurosci. 38, 5551–5566. Khan, T.A., Revah, O., Gordon, A., Yoon, S.-J., Krawisz, A.K., Goold, C., Sun, Y., Kim, C.H., Tian, Y., Li, M.-Y., et al. (2020). Neuronal defects in a human cellular model of 22q11.2 deletion syndrome. Nature Medicine 26, 1888–1898. Korotkevich, G., Sukhov, V., Budin, N., Shpak, B., Artyomov, M.N., and Sergushichev, A. (2021). Fast gene set enrichment analysis. BioRxiv 060012. Krey, J.F., Paşca, S.P., Shcheglovitov, A., Yazawa, M., Schwemberger, R., Rasmusson, R., and Dolmetsch, R.E. (2013). Timothy syndrome is associated with activity-dependent dendritic retraction in rodent and human neurons. Nat Neurosci 16, 201–209. Krienen, F.M., Goldman, M., Zhang, Q., Rosario, R. del, Florio, M., Machold, R., Saunders, A., Levandowski, K., Zaniewski, H., Schuman, B., et al. (2019). Innovations in Primate Interneuron Repertoire. BioRxiv 709501.
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Kuo, I.Y., and Ehrlich, B.E. (2015). Signaling in Muscle Contraction. Cold Spring Harb Perspect Biol 7. Langfelder, P., and Horvath, S. (2008). WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 9, 559. Manent, J.-B., Demarque, M., Jorquera, I., Pellegrino, C., Ben-Ari, Y., Aniksztejn, L., and Represa, A. (2005). A Noncanonical Release of GABA and Glutamate Modulates Neuronal Migration. J. Neurosci. 25, 4755–4765. Marín, O. (2012). Interneuron dysfunction in psychiatric disorders. Nat Rev Neurosci 13, 107–120. Marín, O. (2013). Cellular and molecular mechanisms controlling the migration of neocortical interneurons. Eur J Neurosci 38, 2019–2029. Martini, F.J., and Valdeolmillos, M. (2010). Actomyosin contraction at the cell rear drives nuclear translocation in migrating cortical interneurons. J. Neurosci. 30, 8660–8670. Maset, A., Galla, L., Francia, S., Cozzolino, O., Capasso, P., Goisis, R.C., Losi, G., Lombardo, A., Ratto, G.M., and Lodovichi, C. (2021). Altered Cl− homeostasis hinders forebrain GABAergic interneuron migration in a mouse model of intellectual disability. PNAS 118. Mathis, M.W., and Mathis, A. (2020). Deep learning tools for the measurement of animal behavior in neuroscience. Current Opinion in Neurobiology 60, 1–11. Mathis, A., Mamidanna, P., Cury, K.M., Abe, T., Murthy, V.N., Mathis, M.W., and Bethge, M. (2018). DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nature Neuroscience 21, 1281. McKenna, A., Hanna, M., Banks, E., Sivachenko, A., Cibulskis, K., Kernytsky, A., Garimella, K., Altshuler, D., Gabriel, S., Daly, M., et al. (2010). The Genome Analysis Toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data. |
Genome Res. 20, 1297–1303. Meechan, D.W., Tucker, E.S., Maynard, T.M., and LaMantia, A.-S. (2012). Cxcr4 regulation of interneuron migration is disrupted in 22q11.2 deletion syndrome. Proc Natl Acad Sci U S A 109, 18601–18606. Miura, Y., Li, M.-Y., Birey, F., Ikeda, K., Revah, O., Thete, M.V., Park, J.-Y., Puno, A., Lee, S.H., Porteus, M.H., et al. (2020). Generation of human striatal organoids and cortico-striatal assembloids from human pluripotent stem cells. Nature Biotechnology 38, 1421–1430. Morgan, J.I., and Curran, T. (1986). Role of ion flux in the control of c- fos expression. Nature 322, 552–555. Mullins, N., Forstner, A.J., O’Connell, K.S., Coombes, B., Coleman, J.R.I., Qiao, Z., Als, T.D., Bigdeli, T.B., Børte, S., Bryois, J., et al. (2020). Genome-wide association study of over 40,000 bipolar disorder cases provides novel biological insights. MedRxiv 2020.09.17.20187054. Muraki, K., and Tanigaki, K. (2015). Neuronal migration abnormalities and its possible implications for schizophrenia. Front. Neurosci. 9. Murphy, T.H., Worley, P.F., and Baraban, J.M. (1991). L-type voltage-sensitive calcium channels mediate synaptic activation of immediate early genes. Neuron 7, 625–635. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Panagiotakos, G., Haveles, C., Arjun, A., Petrova, R., Rana, A., Portmann, T., Paşca, S.P., Palmer, T.D., and Dolmetsch, R.E. (2019). Aberrant calcium channel splicing drives defects in cortical differentiation in Timothy syndrome. ELife 8, e51037. Pardiñas, A.F., Holmans, P., Pocklington, A.J., Escott-Price, V., Ripke, S., Carrera, N., Legge, S.E., Bishop, S., Cameron, D., Hamshere, M.L., et al. (2018). Common schizophrenia alleles are enriched in mutation-intolerant genes and in regions under strong background selection. Nature Genetics 50, 381– 389. Parikshak, N.N., Gandal, M.J., and Geschwind, D.H. (2015). Systems biology and gene networks in neurodevelopmental and neurodegenerative disorders. Nature Reviews Genetics 16, 441–458. Paşca, S.P., Portmann, T., Voineagu, I., Yazawa, M., Shcheglovitov, A., Paşca, A.M., Cord, B., Palmer, T.D., Chikahisa, S., Nishino, S., et al. (2011). Using iPSC-derived neurons to uncover cellular phenotypes associated with Timothy syndrome. Nat Med 17, 1657–1662. Peyre, E., Silva, C.G., and Nguyen, L. (2015). Crosstalk between intracellular and extracellular signals regulating interneuron production, migration and integration into the cortex. Front. Cell. Neurosci 9, 129. Powell, E.M. (2013). Interneuron Development and Epilepsy: Early Genetic Defects Cause Long-Term Consequences in Seizures and Susceptibility. Epilepsy Curr 13, 172–176. Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M.A.R., Bender, D., Maller, J., Sklar, P., de Bakker, P.I.W., Daly, M.J., et al. |
(2007). PLINK: A Tool Set for Whole-Genome Association and Population-Based Linkage Analyses. The American Journal of Human Genetics 81, 559–575. Purcell, S.M., Moran, J.L., Fromer, M., Ruderfer, D., Solovieff, N., Roussos, P., O’Dushlaine, C., Chambert, K., Bergen, S.E., Kähler, A., et al. (2014). A polygenic burden of rare disruptive mutations in schizophrenia. Nature 506, 185–190. Ripke, S., Neale, B.M., Corvin, A., Walters, J.T.R., Farh, K.-H., Holmans, P.A., Lee, P., Bulik-Sullivan, B., Collier, D.A., Huang, H., et al. (2014). Biological insights from 108 schizophrenia-associated genetic loci. Nature 511, 421–427. Ritchie, M.E., Phipson, B., Wu, D., Hu, Y., Law, C.W., Shi, W., and Smyth, G.K. (2015). limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Research 43, e47–e47. Roberts, D.S., Raol, Y.H., Bandyopadhyay, S., Lund, I.V., Budreck, E.C., Passini, M.A., Passini, M.J., Wolfe, J.H., Brooks-Kayal, A.R., and Russek, S.J. (2005). Egr3 stimulation of GABRA4 promoter activity as a mechanism for seizure-induced up-regulation of GABA(A) receptor alpha4 subunit expression. Proc Natl Acad Sci U S A 102, 11894–11899. Robinson, M.D., McCarthy, D.J., and Smyth, G.K. (2010). edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140. Roderick, H.L., Berridge, M.J., and Bootman, M.D. (2003). Calcium-induced calcium release. Current Biology 13, R425. Saliba, R.S., Gu, Z., Yan, Z., and Moss, S.J. (2009). Blocking L-type Voltage-gated Ca2+ Channels with Dihydropyridines Reduces γ-Aminobutyric Acid Type A Receptor Expression and Synaptic Inhibition*. Journal of Biological Chemistry 284, 32544–32550. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Schaar, B.T., and McConnell, S.K. (2005). Cytoskeletal coordination during neuronal migration. PNAS 102, 13652–13657. Servili, E., Trus, M., Sajman, J., Sherman, E., and Atlas, D. (2020). Elevated basal transcription can underlie timothy channel association with autism related disorders. Progress in Neurobiology 101820. Silbereis, J.C., Pochareddy, S., Zhu, Y., Li, M., and Sestan, N. (2016). The Cellular and Molecular Landscapes of the Developing Human Central Nervous System. Neuron 89, 248–268. Silva, C.G., Peyre, E., and Nguyen, L. (2019). Cell migration promotes dynamic cellular interactions to control cerebral cortex morphogenesis. Nature Reviews Neuroscience 1. Sloan, S.A., Andersen, J., Pașca, A.M., Birey, F., and Pașca, S.P. (2018). Generation and assembly of human brain region–specific three-dimensional cultures. Nature Protocols 13, 2062. Splawski, I., Timothy, K.W., Sharpe, L.M., Decher, N., Kumar, P., Bloise, R., Napolitano, C., Schwartz, P.J., Joseph, R.M., Condouris, K., et al. |
(2004). Ca(V)1.2 calcium channel dysfunction causes a multisystem disorder including arrhythmia and autism. Cell 119, 19–31. Tan, J.L., Ravid, S., and Spudich, J.A. (1992). Control of nonmuscle myosins by phosphorylation. Annu. Rev. Biochem. 61, 721–759. Tian, Y., Voineagu, I., Paşca, S.P., Won, H., Chandran, V., Horvath, S., Dolmetsch, R.E., and Geschwind, D.H. (2014). Alteration in basal and depolarization induced transcriptional network in iPSC derived neurons from Timothy syndrome. Genome Medicine 6, 75. Voineagu, I., Wang, X., Johnston, P., Lowe, J.K., Tian, Y., Horvath, S., Mill, J., Cantor, R.M., Blencowe, B.J., and Geschwind, D.H. (2011). Transcriptomic analysis of autistic brain reveals convergent molecular pathology. Nature 474, 380–384. Wamsley, B., and Fishell, G. (2017). Genetic and activity-dependent mechanisms underlying interneuron diversity. Nature Reviews Neuroscience 18, 299–309. Wheeler, D.G., Groth, R.D., Ma, H., Barrett, C.F., Owen, S.F., Safa, P., and Tsien, R.W. (2012). CaV1 and CaV2 Channels Engage Distinct Modes of Ca2+ Signaling to Control CREB-Dependent Gene Expression. Cell 149, 1112–1124. Yap, E.-L., and Greenberg, M.E. (2018). Activity-Regulated Transcription: Bridging the Gap between Neural Activity and Behavior. Neuron 100, 330–348. Yoon, S.-J., Elahi, L.S., Pașca, A.M., Marton, R.M., Gordon, A., Revah, O., Miura, Y., Walczak, E.M., Holdgate, G.M., Fan, H.C., et al. (2019). Reliability of human cortical organoid generation. Nature Methods 16, 75. Yu, G., Wang, L.-G., Han, Y., and He, Q.-Y. (2012). clusterProfiler: an R Package for Comparing Biological Themes Among Gene Clusters. OMICS: A Journal of Integrative Biology 16, 284–287. Yu, J., Liu, Z., Wang, L., Wu, G., and Wu, M. (2017). Increased GABA (A) Receptors alpha1, gamma2, delta Subunits might be Associated with the Activation of the CREB Gene in Low Mg2+ Model of Epilepsy. Neuropsychiatry 7, 398–405. Zhang, B., and Horvath, S. (2005). A General Framework for Weighted Gene Co-Expression Network Analysis. Statistical Applications in Genetics and Molecular Biology 4.
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.06.14.448277
;
this version posted June 15, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. |
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 1
A pseudorabies virus serine/threonine kinase, US3, promotes retrograde transport in axons via
2
Akt/mToRC1
3
Running title: PRV US3 activates Akt/mToRC1 to promote axonal transport
4
Andrew D. Estevesa#, Orkide O. Koyuncua*, Lynn W. Enquista#
5
aDepartment of Molecular Biology, Princeton University, Princeton, NJ, USA
6
Correspondence: Andrew D. Esteves ([email protected]), Lynn W. Enquist
7
Present address: Department of Microbiology and Molecular Genetics, School of Medicine,
8
University of California, Irvine, Irvine, CA, USA
9
([email protected])
10
Word Count: Abstract (224),
11
Main Text: (4909)
12
Key words: Akt, PRV, axon, retrograde transport, US3, mToRC1, translation, viral entry, kinase,
13
intracellular transport
14
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 15
Abstract
16
Infection of peripheral axons by alpha herpesviruses (AHVs) is a critical stage in
17
establishing a life-long infection in the host. Upon entering the cytoplasm of axons, AHV
18
nucleocapsids and associated inner-tegument proteins must engage the cellular retrograde
19
transport machinery to promote the long-distance movement of virion components to the
20
nucleus. The current model outlining this process is incomplete and further investigation is
21
required to discover all viral and cellular determinants involved as well as the temporality of the
22
events. Using a modified tri-chamber system, we have discovered a novel role of the
23
pseudorabies virus (PRV) serine/threonine kinase, US3, in promoting efficient retrograde
24
transport of nucleocapsids. We discovered that transporting nucleocapsids move at similar
25
velocities both in the presence and absence of a functional US3 kinase; however fewer
26
nucleocapsids are moving when US3 is absent and move for shorter periods of time before
27
stopping, suggesting US3 is required for efficient nucleocapsid engagement with the retrograde
28
transport machinery. This led to fewer nucleocapsids reaching the cell bodies to produce a
29
productive infection 12hr later. Furthermore, US3 was responsible for the induction of local
30
translation in axons as early as 1hpi through the stimulation of a PI3K/Akt-mToRC1. These data
31
describe a novel role for US3 in the induction of local translation in axons during AHV infection,
32
a critical step in transport of nucleocapsids to the cell body. |
33
Importance
34
Neurons are highly polarized cells with axons that can reach centimeters in length. 35
Communication between axons at the periphery and the distant cell body is a relatively slow
36
process involving the active transport of chemical messengers. There’s a need for axons to
37
respond rapidly to extracellular stimuli. Translation of repressed mRNAs present within the axon
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 38
occurs to enable rapid, localized responses independently of the cell body. AHVs have evolved a
39
way to hijack local translation in the axons to promote their transport to the nucleus. We have
40
determined the cellular mechanism and viral components involved in the induction of axonal
41
translation. The US3 serine/threonine kinase of PRV activates Akt-mToRC1 signaling pathways
42
early during infection to promote axonal translation. When US3 is not present, the number of
43
moving nucleocapsids and their processivity are reduced, suggesting that US3 activity is required
44
for efficient engagement of nucleocapsids with the retrograde transport machinery. 45
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 46
Introduction
47
Members of the alpha herpesvirus (AHV) subfamily, including the human pathogens,
48
herpes simplex virus type-1 and 2 (HSV-1 and HSV-2), as well as the animal pathogen,
49
pseudorabies virus (PRV), are pantropic viruses capable of infecting the peripheral (PNS) and
50
central (CNS) nervous system of their hosts. Infection of the highly polarized PNS at axon
51
terminals occurs after infection of the epithelial layer. Once in the axonal cytoplasm,
52
nucleocapsids undergo efficient retrograde transport to the nucleus where the viral DNA is
53
transcribed and either a productive or a life-long latent infection is established. In natural hosts,
54
AHVs tend to establish a latent infection. Reactivation from latency results in the anterograde
55
transport of progeny virion particles in the axon to re-infect the epithelial layer that promotes
56
spread to new hosts. In rare events, progeny virion particles can spread in the opposite direction
57
and transsynaptically invade the CNS, often leading to death of the organism. In non-natural
58
hosts, a productive infection followed by invasion of the CNS and death are the most common 1. 59
The recruitment of the retrograde transport machinery to mediate the active transport of
60
virion particles is facilitated by the viral nucleocapsid and inner-tegument proteins (UL36, UL37,
61
and US3) 2,3. |
This process, while dispensable in non-polarized cells, is essential for neuronal
62
infection via axons due to the large distance between the axon terminal and the cell body 3–5. 63
Despite this, the viral and cellular factors involved, and the temporality of the events are not well
64
understood. Following fusion of the viral and cellular membranes, a breach in the actin
65
cytoskeleton is created through the activation of cofilin by the US3 protein 6. The microtubule
66
plus-tip proteins, EB-1 and CLIP170 also have been shown to aid in this process during HSV-1
67
infection 7. The UL36 inner-tegument protein then interacts with dynactin, to recruit the
68
nucleocapsid complex to dynein, the retrograde-directed motor protein 8,9. UL37, even though it
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 69
has not been shown to interact with dynein, is capable of modulating its activity 4. We have
70
previously shown that upon infection of axons with PRV, translation of a subset of axonally-
71
localized mRNA occurs, producing a subset of proteins related to intracellular transport and
72
cytoskeletal remodeling 10. Among these was the dynein regulator, Lis1. Local translation in
73
axons was shown to be essential for the efficient transport of virion components through the
74
axon; however the mechanism that regulates this is unknown. 75
Translation of cellular mRNA is a tightly regulated process, with the initiation stage
76
being the most rate-limiting. The PI3K/Akt-mToRC1 signaling pathway is the canonical route by
77
which translation initiation occurs in eukaryotic cells 11. AHVs have been demonstrated to
78
manipulate this signaling pathway to support infection and replication. HSV-2 infection induces
79
Akt phosphorylation upon binding of gB, on the virion envelope, to αvβ3 integrins leading to
80
release of intracellular calcium stores to promote entry of nucleocapsids into the cell 12,13. In
81
HSV-1 infected cells activation of mToRC1 was induced by the phosphorylation of Akt
82
substrates in an Akt independent manner. This work demonstrated that a viral kinase could act as
83
an Akt surrogate to bypass cellular signal pathway control mechanisms to promote constitutive
84
viral replication 14. In VZV infected cells Akt phosphorylation is increased and required for
85
efficient replication15–17. PRV infection induces Akt phosphorylation to mediate anti-apoptosis
86
effects on infected cells 18. Despite these findings in non-neuronal cells, it is not known if AHV
87
infection of axons induces Akt signaling pathways. In this paper we investigated whether PRV
88
infection stimulates the Akt-mToRC1 signaling pathway to induce local translation in axons. |
89
US3 is a multifunctional, viral encoded serine/threonine kinase present in the inner-
90
tegument layer of the virion and one of only two protein kinases conserved by all AHVs 19. 91
Although US3 has been shown to be dispensable for virus replication in cell culture, it is vital for
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 92
viral fitness in vivo 20–25. Some of the most notable functions of US3 include the inhibition of
93
apoptosis in infected cells through the activation of Akt and NF-κB signaling pathways,
94
promotion of nuclear egress of newly-made nucleocapsids, and disassembly of actin stress fibers
95
by cofilin activation 26–28. Due to its serine/threonine kinase function, known interactions with
96
Akt, and its presence in the tegument layer, and thus delivered directly to the cytoplasm, we
97
hypothesized that US3 stimulates Akt-mToRC1 signaling pathways in axons early after
98
infection, to induce local translation. 99
In this study, we established a novel role for PRV US3 in the induction of translation in
100
axons via a PI3K/Akt-mToRC1 signaling pathway early after virion entry to promote the
101
efficient retrograde transport of nucleocapsids to the cell body. In the absence of US3 or Akt
102
phosphorylation, the number of transporting nucleocapsids and their processivity were
103
significantly reduced. These events led to a reduction in the number of infected neuronal cell
104
bodies later in infection. Together, these findings suggest a role for US3 in the continuous
105
engagement of PRV nucleocapsids with the retrograde transport machinery. Due to the
106
significance this stage of infection plays in AHV infection of neurons at the axon, US3 may
107
serve as a target for drugs aiming to prevent the life-long, reactivatable infection caused by
108
AHVs. 109
Results
110
Akt is phosphorylated in axons early after PRV infection
111
To determine if Akt is phosphorylated in axons early after PRV infection, we cultured
112
primary rat superior cervical ganglion (SCG) neurons in vitro in Campenot tri-chambers
113
(Fig.1A). This cell culture system allows for the fluidic separation of neuronal cell bodies (in the
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 114
S compartment) from axons (in the N compartment), enabling us to infect pure populations of
115
axons and monitor responses independently of the cell bodies 29. N compartment axons were
116
infected with 106 plaque forming units (PFU) of a PRV-Becker recombinant expressing a
117
monomeric red fluorescent protein (mRFP)-tagged capsid protein (VP26) termed PRV 180. |
Akt
118
phosphorylation was monitored using a phospho (p-) S473-Akt antibody. Phosphorylation at
119
serine 473 is the most well-known mechanism of Akt activation 30 and can be observed as early
120
as 30 minutes post infection (mpi) in axons and continues to at least 180mpi (Fig.1B), indicating
121
that PRV infection does induce Akt phosphorylation in axons. Furthermore, when axons in the N
122
compartment were pretreated with LY294,002 (a potent and selective PI3K inhibitor) prior to
123
infection, no Akt phosphorylation was observed, suggesting that PRV-induced Akt
124
phosphorylation is PI3K-dependent (Fig.1C). When axons were pretreated with the mToRC1
125
inhibitor rapamycin, or the translation inhibitor cycloheximide, Akt phosphorylation did occur
126
(Fig.1C). No change in Akt phosphorylation occurred in S compartment cell bodies 1hour post
127
infection (hpi) when the N compartment was infected, demonstrating that the infection and
128
intracellular signals do not reach the cell body in that time (Fig.1D). 129
Akt phosphorylation in axons is required for efficient retrograde transport of PRV
130
nucleocapsids
131
Next, we determined whether infection-induced Akt phosphorylation in axons affected
132
the spread of PRV to the distant cell body. N compartments were treated with a green lipophilic
133
dye, Fast-DiO, to label the membranes of all axons in the N compartment and their attached cell
134
bodies in the S compartment such that only the cell bodies with axons extended into the N
135
compartment were labeled. 12 hours post DiO staining, N compartments were infected with PRV
136
180 (Fig.2A). At 12hpi, infected cell bodies were visualized by the presence of red fluorescence
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 137
from the mRFP-VP26 fusion protein (Fig.2B). The presence of dual-colored cell bodies
138
(expressing red and green fluorescence) were those who were directly infected via axons in the N
139
compartment. The ratio of dual-colored cell bodies to total green cell bodies represents the
140
efficiency of retrograde infection. When N compartment axons were pretreated with Akt
141
inhibitor VIII for 1 hour prior to infection, we observed a ~56.4% ± 14.7% reduction in the
142
number of dual-colored cell bodies compared to PRV 180 infection alone. Additionally, when
143
axons were pretreated with LY294,002, rapamycin, or cycloheximide, dual-colored cell bodies
144
were reduced by ~48% ± 16.2%, ~58% ± 17.6%, and ~65.7 ± 9.5% respectively. Ras/MAPK is
145
also known to promote translation by signaling through mToRC131–36, however pretreatment of
146
axons with the Erk1/2 inhibitor, U0126, had no significant effect on retrograde infection (~8.3%
147
± 14%) (Fig.2C). |
148
Local translation of axonal mRNA after PRV infection promoted the efficient retrograde
149
transport of nucleocapsids through the axon 10. To determine if Akt-mToRC1 is also required for
150
efficient transport, N compartment axons were infected with PRV 180 and at 2hpi, videos of
151
fluorescent nucleocapsids trafficking through the M (middle) compartment were recorded
152
(Fig.3A). Maximum intensity projections and kymographs were created from the videos of
153
moving nucleocapsids to analyze transport kinetics (Fig.3B). Moving nucleocapsids were
154
represented as continuous “tracks”. When N compartments were pretreated with LY294,002, Akt
155
inhibitor VIII, rapamycin, or cycloheximide the number of moving nucleocapsids was reduced
156
by ~75.8% ± 24.4%, ~76.3% ± 24.4%, ~79% ± 22.8%, ~75.8% ± 40.8% respectively compared
157
to PRV 180 infection alone (Fig.3C). The net displacement of moving nucleocapsids (length of
158
tracks) was also reduced by ~50% when these inhibitors were present (Fig.3D). Nevertheless, the
159
velocity of nucleocapsids that were moving in each condition was the same (Fig.3E), suggesting
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 160
that engagement with the transport machinery, not the transport speed was disrupted. U0126
161
treatment had no significant effect on nucleocapsid transport or processivity. Taken together,
162
these data demonstrate that Akt-mToRC1 signaling is induced by PRV infection in axons and is
163
required for efficient retrograde transport of nucleocapsids. These effects were comparable to
164
those observed when translation was blocked, suggesting Akt-mToRC1 activation acts early after
165
PRV infection to promote translation in axons. 166
Akt phosphorylation in axons requires PRV US3 ser/thr kinase
167
HSV-2 induces Akt phosphorylation upon binding of the glycoprotein gB on the virion
168
envelope with αvβ3 integrins leading to release of intracellular calcium stores to promote entry of
169
nucleocapsids into the cytoplasm 13. We determined if virion binding to the cell surface or fusion
170
of the viral and cellular membranes were responsible for Akt phosphorylation and if Akt
171
phosphorylation affected transport directly or indirectly by regulating entry of nucleocapsids into
172
the cytoplasm. 173
N compartment axons were infected with PRV-Becker mutants lacking either
174
glycoprotein D (gD) (PRV GS442), the viral envelope protein required for specific PRV binding
175
to the nectin-1 cell receptor, or glycoprotein B (gB) (PRV 233), the viral envelope protein
176
required for fusion of the viral and cell membranes. Akt phosphorylation was monitored via
177
western blot (Fig.4) and was unchanged after infection with either of these mutants suggesting
178
that both binding and membrane fusion are required for Akt phosphorylation. |
179
To determine if Akt phosphorylation was induced by PRV after entry into the cell, we
180
investigated the role of US3. US3 is an AHV inner-tegument protein with ser/thr kinase
181
function19. US3 induces phosphorylation of Akt (in PRV) and downstream Akt substrates (in
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 182
HSV-1) 14,18. When we infected axons with a PRV-Becker mutant lacking US3 (PRV 813NS) or
183
a mutant lacking a functional kinase domain (PRV 815KD), no Akt phosphorylation was
184
observed (Fig.5). When axons were infected with a revertant of PRV 813NS, termed PRV 813R,
185
Akt phosphorylation was restored (Fig.5). These data indicate that US3 is responsible for
186
inducing Akt phosphorylation after PRV infection in axons. 187
To determine if US3, and by extension Akt phosphorylation, are required for entry of
188
PRV into cells, rat fibroblasts (Rat2) were infected with multiplicity of infection (MOI) of 5 of
189
either PRV 180 or a ΔUS3 mutant with the mRFP-VP26 fusion (PRV823) at 4C, a temperature
190
permissive to binding, but not entry. 1hpi cultures were washed to remove inoculum and the
191
temperature was brought up to 37ºC to allow for entry. After 15 minutes cells were washed with
192
a low pH citrate buffer (pH 3) to inactivate any un-entered virus particles and the infection was
193
permitted to continue for another hour. Cells were fixed and nucleocapsid fluorescence was
194
visualized by fluorescence microscopy (Fig.6). Nucleocapsids present within the cytoplasm were
195
counted manually and no significant difference was seen indicating US3 and Akt
196
phosphorylation do not affect PRV entry in these cells. 197
Next, we determined if US3 is required for efficient retrograde infection of neurons using
198
the technique described in Fig.2A. Compared to infection with PRV 180, infection with PRV
199
823 in axons led to a ~44.7% ± 23.8% reduction in dual-colored cell bodies in the S
200
compartment (Fig.2B). PRV 823 infection in axons in the N compartment also led to a ~47.3% ±
201
38.4% reduction in the number of trafficking nucleocapsids in the M compartment when
202
compared to PRV 180 (Fig.3C). PRV 823 transport kinetics were comparable to those of PRV
203
180 when Akt-mToRC1 signaling or translation was disrupted (Fig.3D, E,); net displacement
204
was significantly reduced (~37% ± 31.2%), but not transport velocity. Taken together, these data
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. |
It is made
available under a
CC-BY-NC-ND 4.0 International license . 205
suggest US3 mediates efficient retrograde transport of PRV nucleocapsids through axons by
206
inducing an Akt-mToRC1 signaling pathway. 207
US3 and Akt phosphorylation are required for virus-induced local translation in axons
208
We directly tested whether US3 and Akt phosphorylation were required for translation by
209
performing a surface sensing of translation (SUnSET) assay. Axons in the N compartment were
210
either untreated or treated with Akt inhibitor VIII, cycloheximide, or U0126 then infected with
211
PRV 180 or PRV 813NS. Puromycin was added to both N and S compartments to label nascent
212
peptides 15 minutes prior to harvesting for western blot (Fig. 7A). Nascent peptides were
213
visualized using a monoclonal puromycin antibody. Infection of axons with PRV 180 but not
214
PRV 813NS led to an increase in puromycin incorporation when compared to mock, indicating
215
that US3 is required for PRV-induced translation in axons. When Akt phosphorylation and
216
translation were inhibited, puromycin incorporation did not increase past mock level indicating
217
Akt phosphorylation is also required for translation to occur. Pretreatment with U0126 had no
218
effect on puromycin incorporation showing the Ras/MAPK pathway does not affect PRV-
219
induced translation in axons. (Fig.7B). 220
Discussion
221
PNS neurons are highly polarized; with axon terminals extending centimeters away from
222
their cell bodies. Accordingly, the transport of molecular messengers over these long distances is
223
highly regulated 37. Axons contain within them the complete set of protein synthesis machinery,
224
including multiple subsets of localized mRNA that are used to synthesize proteins in response to
225
changes in the extracellular environment and send messages to the cell body to produce a global
226
response to stimulus 38,39. PRV hijacks these processes to promote its efficient retrograde
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 227
transport from the site of infection, in the axon, to the cell body 10. In this study, we have
228
discovered that PRV utilizes an Akt-mToRC1 signaling pathway to induce translation in axons at
229
early time points after infection. The inner-tegument protein kinase, US3, is required for the
230
induction of this signaling pathway and acts upstream of PI3K to promote Akt phosphorylation
231
(Fig.8). 232
Infection of axons with PRV 180 in the presence of the PI3K inhibitor, LY294,002,
233
significantly reduced the number of moving nucleocapsids (Fig. 3C), and reduced the distance
234
nucleocapsids traveled before stopping (Fig.3D). US3 must act upstream of PI3K to promote
235
efficient transport, but it is unknown if PI3K is the direct kinase target of US3. |
Earlier studies
236
have shown that US3 is responsible for the reorganization of the actin cytoskeleton via cofilin
237
activation 6. US3 interacts directly with group 1 PAKs (p-21 activated kinases) to promote
238
cofilin dephosphorylation (activation), leading to disassembly of filamentous actin and the entry
239
of virion tegument and nucleocapsid into the cytoplasm 40. Group 1 PAKs play an important role
240
in cytoskeleton rearrangement and apoptosis signal transduction 41–48 and have been shown to
241
interact with various members of PI3K and Akt signaling pathways 49. It is possible that the
242
cross-talk between group1 PAKs and the PI3K-Akt signaling pathway could be utilized by US3
243
to promote cytoskeletal rearrangements and local translation. 244
When US3 was absent or PI3K/Akt-mToRC1signaling was disrupted, we observed a
245
significant reduction in the number of transporting nucleocapsids and their processivity (Fig.3). 246
Under these conditions, nucleocapsids moved shorter distances before stopping; however the
247
velocity of these nucleocapsids while in motion was similar to wild-type infection without
248
inhibitor present. This suggests nucleocapsids had difficulty engaging with the transport
249
machinery rather than difficulty transporting once engaged. These results were similar to what
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 250
was seen for PRV and HSV-1 mutants when the R2 domain of UL37 inner-tegument protein was
251
altered 4. The UL37-R2 mutant replicates well in peripheral tissue but is unable to move
252
efficiently in peripheral neurons. As a result, the mutant cannot establish life-long infection in
253
the host, a property that would provide a new possibility for vaccine design 4. It’s possible that
254
the use of a US3-null-UL37-R2 mutant would provide a similar or additive effect by further
255
reducing the chance of retrograde infection in peripheral neurons. 256
Long-distance transport in axons requires the microtubule network and associated motor
257
proteins. The kinesin motor proteins mediate anterograde (plus-end directed) transport and
258
dynein motor proteins mediate retrograde (minus-end directed) transport 50. Dynein is a
259
multisubunit complex composed of two heavy-chains, two intermediate chains, and two light
260
chains that regulate motor activity and interactions with cellular cargo 51–54. To achieve
261
spatiotemporal regulation of transport, various accessory factors associate with the dynein
262
complex; one such factor is dynactin50. The UL36 inner-tegument protein interacts with dynactin
263
to recruit nucleocapsids to dynein8,9; however this interaction alone is not sufficient to promote
264
efficient transport. |
Dynein must be phosphorylated to obtain an active conformation50,51,55–57. 265
Recently, it was shown that upon infection of epithelial cells with HSV-1, dynein intermediate
266
chain 1B was phosphorylated at position S80 in an Akt and PKC independent manner 58. We
267
have also observed dynein phosphorylation (at position T88) during PRV infection that was US3
268
dependent (A. D. Esteves and L. W. Enquist, unpublished data). US3’s kinase function could be
269
mediating the activation of dynein throughout the retrograde transport process. More work needs
270
to be done to determine the cause and function of dynein phosphorylation during AHV infection. 271
Axonal infection with PRV mutants lacking either gD or gB, the glycoproteins that
272
mediate virion binding to the plasma membrane and carry out membrane fusion, were unable to
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 273
stimulate Akt phosphorylation (Fig.4). These observations suggested that Akt is phosphorylated
274
after entry of the nucleocapsid and tegument proteins in the cytoplasm. These findings are
275
different from what was observed after HSV-1 and HSV-2 infection. Infection with these AHVs
276
lead to the stimulation of a low-level Ca2+ fluctuation following binding of gD or gB to the
277
nectin-1 co-receptor and heparin sulfate proteoglycans on the cell surface. This is followed by an
278
interaction between Akt and gB leading to Akt phosphorylation, a larger Ca2+ fluctuation and
279
entry of the nucleocapsid and tegument into the cytoplasm 13. We have shown that PRV
280
nucleocapsid entry does not depend on US3, and by extension, Akt phosphorylation (Fig.6). 281
However, we have not investigated the potential for the induction of Ca2+ fluctuation during
282
infection and whether or not it participates in the retrograde transport process. Future work needs
283
to be done to determine the effects of Ca2+ signaling in PRV infection. 284
Our work has revealed a novel role for US3 in the promotion of efficient microtubule-
285
based transport of PRV nucleocapsids in axons. We determined that US3 signals through an Akt-
286
mToRC1 signaling pathway, upstream of PI3K, to induce translation of localized, repressed
287
mRNAs, a process required for efficient retrograde transport. The absence of US3 or the
288
disruption of Akt-mToRC1 not only reduced the number of transporting nucleocapsids but also
289
negatively impacted the processivity of nucleocapsids moving through the axon, suggesting
290
engagement with the dynein-motor complex was disrupted. This work helps to clarify the viral
291
and cellular factors involved in PRV entry and retrograde transport in axons and identifies US3
292
as a potential target for therapies aiming to prevent the spread of AHVs in the nervous system. |
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 293
Acknowledgements
294
This work was supported by the US National Institute of Health grant (5R37NS033506)
295
to Lynn W. Enquist. The authors declare that they have no competing interests. 296
We thank Kevin Pfister for providing the phospho-dynein antibodies used in this study
297
and Heath E. Johnson for assistance with particle tracking automation. 298
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 299
Materials and Methods
300
Primary neuronal culture
301
Superior cervical ganglia (SCG) neurons were isolated from embryonic day 17 Sprague-
302
Dawley rat embryos and cultured as previously described 29. Briefly, SCG were trypsinized and
303
mechanically dissociated. 35mm cell culture dishes were coated with poly-DL-ornithine (Sigma-
304
Aldrich) and laminin (Invitrogen) then fitted with a Campenot tri-chamber (CAMP320 isolator
305
rings, Tyler Research) using autoclaved silicone vacuum grease as an adherent. Dissociated SCG
306
were seeded in one compartment (the S-compartment) of the tri-chamber filled with neurobasal
307
medium (Life Technologies; 21103049) + 50x B-27 supplement (Life Technologies; 17504044)
308
+ 100x Penicillin-Streptomycin-Glutamine (ThermoFisher; 10378016) + 80ng/ml NGF
309
(ThermoFisher; 13257019) and left to grow for ~3weeks with media changes every 7 days. 310
Cell lines and virus stocks
311
Porcine kidney epithelial cells (PK15, ATCC), and Rat2 cells (ATCC) were maintained
312
in Dulbecco modified Eagle medium (DMEM, Hyclone) + 10% fetal bovine serum (FBS,
313
Hyclone) + 1% penicillin-streptomycin (Hyclone). Virus propagation and titer were performed in
314
PK15 cells unless otherwise specified. Rat2 cells were used for synchronized infection assays. 315
LP cells (gB complementing cells derived from PK15 cells) were used to grow PRV 233 virus
316
stocks (Lisa Pomeranz, personal communication). 100ug/ml Geneticin (G418) (InVivo Gen, ant-
317
gn-1) was added to culture every 5th passage to maintain gB expression. G5 cells (gD
318
complementing cells derived from PK15 cells) were used to grow PRV GS442 virus stocks 59. 319
2.5mM L-histidinol dihydrochloride (Sigma, H6647) was added to cultures every 5th passage to
320
maintain gD expression. All cells were incubated at 37ºC, and 5% CO2. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. |
The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 321
Unless otherwise specified, virus infections of PK15 and Rat2 cells were performed with
322
DMEM + 2% FBS. Titers of virus stocks were determined as Plaque forming units (PFU). For
323
live-cell imaging PRV 180 and PRV 823 stocks were used within 2 weeks of production for no
324
more than one freeze-thaw cycle to preserve the fluorophore intensity. Virus stocks used are as
325
follows:
Virus Strain
Genotype
Reference
PRV 180
PRV 813NS
PRV 813R
Becker wild-type strain with mRFP-VP26 fusion US3-null Becker strain; nonsense mutation Revertant of 813NS
60
20
20
PRV 823
PRV GS442
PRV 233
US3-null Becker strain, mRFP-VP26 fusion gD-null Becker strain, diffusible GFP expression gB-null Becker strain, diffusible GFP expression
20
59
61
326
327
Antibodies and chemicals
328
All antibodies were diluted in TBS-T (Tris buffered saline-Tween, 0.1%) treated with 3%
329
bovine serum albumin (BSA) and stored in -20ºC unless otherwise specified. Rabbit phospho-
330
Akt (ser473)(D9E) (Cell Signaling Technology, 4060) was used at 1:1,000 for western blot
331
(WB). Rabbit Akt antibody (Cell Signaling Technology, 9272) was used at 1:1,000 to detect total
332
Akt in WB. Monoclonal anti-beta-actin antibody (Sigma Aldrich, A1978) was used at 1:10,000
333
for WB. Mouse monoclonal US3 7H10.21 20 was used at 1:1,000 for WB. Rabbit phospho-
334
dynein-threonine88 was a gift from Kevin Pfister and was used at 1:300 for WB. Mouse
335
recombinant anti-cytoplasmic dynein intermediate chain (74.1) (Abcam, ab23905) was used at
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 336
1:500 for WB. Mouse monoclonal anti-puromycin clone 4G11 (EMD Millipore, MABE342) was
337
used at 1:2,000 for WB. 338
All drug treatments occurred 1hr prior to infection, concentrations represent final
339
experimental concentrations, and drugs were resuspended in DMSO unless otherwise specified. 340
LY294,002 (Cell Signaling Technology, 9901) was used at a concentration of 20uM. Akt
341
inhibitor VIII (Sigma, 124018) was used at a concentration of 5uM. Rapamycin (Sigma, 553210)
342
was used at a concentration of 100uM. U0126 (Sigma, 662009) was used at a concentration of
343
10uM. Puromycin (AG Scientific, P-1033-SOL) was solubilized in deionized H2O and used at a
344
concentration of 1ug/ml. FAST-DiO (ThermoFisher, D3898) was used a concentration of 5ug/ml
345
and incubated for 12hr prior to infection (24hr prior to imaging). 346
Western Blotting
347
SCG neurons in tri-chambers were seeded at a density of 1 SCG per S compartment. |
348
Dishes were washed 3x with warm PBS (phosphor buffered saline) then incubated with
349
neurobasal media + 100x penicillin-streptomycin-glutamine (note the lack of B-27 supplement)
350
for 12hr prior to infection. Either axons in the N compartment or cell bodies in the S
351
compartment were lysed in the chamber with 40ul of 2x Laemmli buffer prepared from a dilution
352
of 5x Laemmli buffer (10% SDS, 300mM Tris-Cl pH = 6.8, 0.05% bromophenol blue,100mM
353
DTT, and 50% Glycerol in ddH2O). Lysates were boiled at 90ºC for 5min then cooled on ice. 354
Proteins were separated by SDS-PAGE and 4-12% gradient NuPAGE Bis/Tris gels. Proteins
355
were transferred to nitrocellulose membranes (GE Healthcare, 45-0040002) using a Trans-Blot
356
SD semi-dry transfer cell (Bio Rad). Membranes were blocked using 1x Pierce Clear Milk
357
Blocking Buffer (ThermoFisher, 37587) for 30min at room temperature (RT) then washed 3x
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 358
with TBS-T. Membranes were incubated in primary antibody dilution over night at 4ºC followed
359
by 3x TBS-T washes then incubation with horseradish peroxidase-conjugated secondary
360
antibody (1:10,000 dilutions in 3% BSA-TBS-T) for 45min at RT and a final 3x washes with
361
TBS-T. Chemiluminescent substrate, Supersignal West Pico (ThermoFisher, 34080), West Dura
362
Extended Duration Substrate (ThermoFisher, 34075), West Femto Maximum Sensitivity
363
Substrate (ThermoFisher, 34094), or West Atto Ultimate Sensitivity Substrate (ThermoFisher,
364
A38554) were added to the membranes for 5min at RT. Protein bands were visualized by
365
exposing the membranes on HyBlot CL autoradiography film (Denville scientific, E3018). 366
SUnSET assay for labeling nascent peptides
367
Puromycin was added to the chamber compartment to be analyzed 15 minutes prior to
368
harvest. The rest of the protocol follows the same as for WB. Puromycin-labeled peptides were
369
detected with puromycin antibodies. See “Antibodies and chemicals” for the primary antibody
370
used. 371
Synchronized infection assay
372
Rat2 cells in culture were cooled to 4ºC. PRV 180 or PRV823 inoculumns were added at
373
an MOI of 5 and virion absorption to cells was allowed to occur for 1hr at 4ºC before media was
374
aspirated and cells were washed with chilled PBS 3x. Chilled 2% FBS DMEM was added to the
375
dish and temperature was brought up to 37ºC for 15 minutes to enable virion entry. Media was
376
removed and infected cells were washed once with a low pH citrate buffer for 2min to inactivate
377
virions that had not entered cells. Citrate buffer was removed and cells were washed 3x with
378
PBS at RT before warm 2% FBS DMEM was added. |
Infected cells were incubated for 1 hour at
379
37ºC. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 380
Cells were fixed using 4% paraformaldehyde (PFA) for 10min at RT and permeabilized
381
with 0.1% Triton x-100 (Sigma, T8787) for 10min at RT. After permeabilization, plates were
382
blocked for 1hr at RT with 1% BSA in PBS, washed 3x with PBS then treated with DAPI
383
(ThermoFisher, 62248) to stain cell nuclei for 5min at RT in the dark with 1ug/ml DAPI in 0.1%
384
BSA PBS. Cells were washed 3x with PBS then imaged. Total and average number of red-
385
fluorescent nucleocapsids per cell in a single field of view (FOV) were manually counted using
386
NIS Elements Advanced Research software (Nikon). 387
Retrograde transport assay
388
SCGs were seeded at 2/3 of an SCG per S compartment. Fast- DiO was added to axons in
389
the N compartment 12hr prior to infection. N compartments were either untreated or treated with
390
LY294,002, Akt inhibitor VIII, rapamycin, cycloheximide, or U0126 1hr prior to infection. 391
Axons in N compartments were infected with 106 PFU of PRV. At 6hr post infection unabsorbed
392
virus inoculum and drug in N compartment were washed out and replaced with fresh neurobasal
393
medium. At 12hr post infection cell bodies in the S compartment were tile-imaged for mRFP-
394
VP26 (red) and DiO (green). Total numbers of green and dual-colored particles (red and green)
395
were manually counted (see synchronized infection assay). The ratio of dual-colored cell bodies
396
to total green cell bodies was determined per S compartment and averaged for all conditions. 397
Single particle tracking in tri-chambers
398
SCGs were seeded at 2/3 of an SCG per S compartment with optical plastic tissue culture
399
dishes (Ibidi, 81156-400). N compartment axons were either untreated or treated with drug for
400
1hr prior to infection. Infections were initiated at 106 PFU for 2hr before time-lapse imaging was
401
used to visualize moving fluorescent virus particles in the M compartment. 30 second recordings
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 402
of a single FOV were completed at approximately 2 frames per second. The total number of
403
moving virus particles were counted. For instantaneous velocity measurements the TrackMate
404
plugin for imageJ was used. Velocities were parsed into bins in a histogram ranging from -
405
4um/sec (the maximum anterograde velocity recorded) and +4um/sec (the maximum retrograde
406
velocity recorded) with intervals of 1um/sec. |
Instantaneous velocities in bins 1um/sec to 4um/sec
407
were averaged across all conditions to determine the average velocity of all moving virion
408
particles. Velocities in the 0um/sec bin were assumed to be stationary particles and were not
409
included in the average velocity measurements. 410
Imaging processing and analysis
411
All imaging was conducted on a Nikon Eclipse Ti inverted epifluorescence microscope
412
using a CoolSNAP ES2 CCD camera. Images and movies were processed using ImageJ 62 and
413
NIS Elements Advanced Research software (Nikon). Comparative images were all captured with
414
the same exposure times, brightness and contrast adjustments were applied to the entire image
415
and alterations were applied equally across conditions. 416
To analyze trafficking virus particles in the M compartment, imageJ was used to create
417
maximum intensity projections of all videos to visualize the path of moving particles. Particle
418
trajectories were manually counted and taken to represent a single moving particle. The
419
directionality and velocity of a moving particle was visualized using the imageJ multi
420
kymograph tool. Particle lengths were calculated by measuring the total particle distance traveled
421
over the course of the 30sec movie using the segmented line tool in imageJ. In order to be
422
counted, particles must not enter or exit the FOV during the entire 30sec recording time. 423
Statistical analysis
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 424
All data were analyzed using GraphPad Prism 9 (GaphPad Software, La Jolla California
425
USA, www.graphpad.com). Figure legends provide the statistical test used for each analysis. 426
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 427
References
428
1. Pomeranz, L. E., Reynolds, A. E. & Hengartner, C. J. Molecular Biology of Pseudorabies
429
Virus: Impact on Neurovirology and Veterinary Medicine. Microbiol. Mol. Biol. Rev. 69,
430
462–500 (2005). 431
2. Döhner, K. et al. Function of dynein and dynactin in herpes simplex virus capsid
432
transport. Mol. Biol. Cell 13, 2795–2809 (2002). 433
3. Sodeik, B., Ebersold, M. & Helenius, A. Microtubule-mediated transport of incoming
434
herpes simplex virus 1 capsids to the nucleus. J. Cell Biol. 136, 1007–1021 (1997). 435
4. Richards, A. L. et al. The pUL37 tegument protein guides alpha-herpesvirus retrograde
436
axonal transport to promote neuroinvasion. |
PLOS Pathog. 13, e1006741 (2017). 437
5. Sekine, E., Schmidt, N., Gaboriau, D. & O’Hare, P. Spatiotemporal dynamics of HSV
438
genome nuclear entry and compaction state transitions using bioorthogonal chemistry and
439
super-resolution microscopy. PLOS Pathog. 13, e1006721 (2017). 440
6. Jacob, T. et al. Alphaherpesviral US3 Kinase Induces Cofilin Dephosphorylation To
441
Reorganize the Actin Cytoskeleton. J. Virol. 87, 4121 (2013). 442
7. Jovasevic, V., Naghavi, M. H. & Walsh, D. Microtubule plus end–associated CLIP-170
443
initiates HSV-1 retrograde transport in primary human cells. J. Cell Biol. 211, 323 (2015). 444
8. Radtke, K. et al. Plus- and minus-end directed microtubule motors bind simultaneously to
445
herpes simplex virus capsids using different inner tegument structures. PLoS Pathog. 6, 1–
446
20 (2010). bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 447
9. Zaichick, K. et al. The herpesvirus VP1/2 protein is an effector of dynein-mediated capsid
448
transport and neuroinvasion. Cell Host Microbe 13, 193–203 (2013). 449
10. Koyuncu, O., Perlman, D. & Enquist, L. W. Efficient retrograde transport of pseudorabies
450
virus within neurons requires local protein synthesis in axons. Cell Host Microbe 13, 54–
451
66 (2013). 452
11. Sonenberg, N. & Hinnebusch, A. Regulation of translation initiation in eukaryotes:
453
mechanisms and biological targets. Cell 136, 731–745 (2009). 454
12. Cheshenko, N., Liu, W., Satlin, L. M. & Herold, B. C. Multiple Receptor Interactions
455
Trigger Release of Membrane and Intracellular Calcium Stores Critical for Herpes
456
Simplex Virus Entry. Mol. Biol. Cell 18, 3119 (2007). 457
13. Cheshenko, N. et al. Herpes simplex virus type 2 glycoprotein H interacts with integrin
458
αvβ3 to facilitate viral entry and calcium signaling in human genital tract epithelial cells. 459
J. Virol. 88, 10026–10038 (2014). 460
14. Chuluunbaatar, U. & Mohr, I. A herpesvirus kinase that masquerades as Akt: You don’t
461
have to look like Akt, to act like it. Cell Cycle 10, 2064–2068 (2011). 462
15. Liu, X. & Cohen, J. Varicella-zoster virus ORF12 protein activates the
463
phosphatidylinositol 3-kinase/Akt pathway to regulate cell cycle progression. J. Virol. 87,
464
1842–1848 (2013). 465
16. Nandini, S. et al. Single-cell mass cytometry analysis of human tonsil T cell remodeling
466
by varicella zoster virus. Cell Rep. 8, 633–645 (2014). 467
17. Rahaus, M., N, D. & MH, W. Varicella-zoster virus requires a functional PI3K/Akt/GSK-
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. |
It is made
available under a
CC-BY-NC-ND 4.0 International license . 468
3alpha/beta signaling cascade for efficient replication. Cell. Signal. 19, 312–320 (2007). 469
18. Ching-Dong, C. et al. Suppression of apoptosis by pseudorabies virus Us3 protein kinase
470
through the activation of PI3-K/Akt and NF-κB pathways. Res. Vet. Sci. 95, 764–774
471
(2013). 472
19. Deruelle, M. & Favoreel, H. Keep it in the subfamily: the conserved alphaherpesvirus
473
US3 protein kinase. J. Gen. Virol. 92, 18–30 (2011). 474
20. Olsen, L. M., Ch’ng, T. H., Card, J. P. & Enquist, L. W. Role of Pseudorabies Virus Us3
475
Protein Kinase during Neuronal Infection. J. Virol. 80, 6387–6398 (2006). 476
21. Frame, M., FC, P., DJ, M., HS, M. & DP, L. Identification of the herpes simplex virus
477
protein kinase as the product of viral gene US3. J. Gen. Virol. 68 ( Pt 10, 2699–2704
478
(1987). 479
22. Hanks, S., AM, Q. & T, H. The protein kinase family: conserved features and deduced
480
phylogeny of the catalytic domains. Science 241, 42–52 (1988). 481
23. Heineman, T. C., Seidel, K. & Cohen, J. I. The varicella-zoster virus ORF66 protein
482
induces kinase activity and is dispensable for viral replication. J. Virol. 70, 7312 (1996). 483
24. McGeoch, D. & Davison, A. Alphaherpesviruses possess a gene homologous to the
484
protein kinase gene family of eukaryotes and retroviruses. Nucleic Acids Res. 14, 1765–
485
1777 (1986). 486
25. Purves, F., RM, L., DP, L. & B, R. Herpes simplex virus 1 protein kinase is encoded by
487
open reading frame US3 which is not essential for virus growth in cell culture. J. Virol. 488
61, 2896–2901 (1987). bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 489
26. Brzozowska, A., M, R., AD, L. & K, B.-S. Point mutations in BHV-1 Us3 gene abolish its
490
ability to induce cytoskeletal changes in various cell types. Vet. Microbiol. 143, 8–13
491
(2010). 492
27. Granstedt, A. E., Bosse, J. B., Thiberge, S. Y. & Enquist, L. W. In vivo imaging of
493
alphaherpesvirus infection reveals synchronized activity dependent on axonal sorting of
494
viral proteins. Proc. Natl. Acad. Sci. U. S. A. 110, (2013). 495
28. Mettenleiter, T., BG, K. & H, G. Herpesvirus assembly: an update. Virus Res. 143, 222–
496
234 (2009). 497
29. Curanović, D., Ch’ng, T. H., Szpara, M. & Enquist, L. Compartmented Neuron Cultures
498
for Directional Infection by Alpha Herpesviruses. Curr. Protoc. Cell Biol. CHAPTER,
499
Unit (2009). 500
30. Chan, T., Rittenhouse, S. & Tsichlis, P. AKT/PKB and other D3 phosphoinositide-
501
regulated kinases: kinase activation by phosphoinositide-dependent phosphorylation. 502
Annu. Rev. Biochem. 68, 965–1014 (1999). 503
31. Ballif, B. A. et al. Quantitative phosphorylation profiling of the ERK/p90 ribosomal S6
504
kinase-signaling cassette and its targets, the tuberous sclerosis tumor suppressors. |
Proc. 505
Natl. Acad. Sci. 102, 667–672 (2005). 506
32. Carriere, A. et al. ERK1/2 Phosphorylate Raptor to Promote Ras-dependent Activation of
507
mTOR Complex 1 (mTORC1). J. Biol. Chem. 286, 567 (2011). 508
33. Hassan, B., Akcakanat, A., AM, H. & F, M.-B. Targeting the PI3-kinase/Akt/mTOR
509
signaling pathway. Surg. Oncol. Clin. N. Am. 22, 641–664 (2013). bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 510
34. Hemmings, B. & Restuccia, D. PI3K-PKB/Akt pathway. Cold Spring Harb. Perspect. 511
Biol. 4, (2012). 512
35. Ma, L., Z, C., H, E.-B., P, T. & PP, P. Phosphorylation and functional inactivation of
513
TSC2 by Erk implications for tuberous sclerosis and cancer pathogenesis. Cell 121, 179–
514
193 (2005). 515
36. Roux, P., BA, B., R, A., SP, G. & J, B. Tumor-promoting phorbol esters and activated Ras
516
inactivate the tuberous sclerosis tumor suppressor complex via p90 ribosomal S6 kinase. 517
Proc. Natl. Acad. Sci. U. S. A. 101, 13489–13494 (2004). 518
37. Holt, C. E., Martin, K. C. & Schuman, E. M. Local translation in neurons: visualization
519
and function. Nat. Struct. Mol. Biol. 2019 267 26, 557–566 (2019). 520
38. Merianda, T. T. et al. A functional equivalent of endoplasmic reticulum and Golgi in
521
axons for secretion of locally synthesized proteins. Mol. Cell. Neurosci. 40, 128–142
522
(2008). 523
39. Murashov, A. K. et al. RNAi pathway is functional in peripheral nerve axons. FASEB J. 524
21, 656–670 (2007). 525
40. Broeke, C. Van den et al. Alphaherpesvirus US3-mediated reorganization of the actin
526
cytoskeleton is mediated by group A p21-activated kinases. Proc. Natl. Acad. Sci. 106,
527
8707–8712 (2009). 528
41. Frost, J. A. et al. Stimulation of NFB Activity by Multiple Signaling Pathways Requires
529
PAK1*. (2000) doi:10.1074/jbc.M909860199. 530
42. Bagrodia, S., Dé Rijard, B., Davis, R. J. & Cerione, R. A. Cdc42 and PAK-mediated
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 531
Signaling Leads to Jun Kinase and p38 Mitogen-activated Protein Kinase Activation*. 532
(1995). 533
43. Zhang, S. et al. Rho Family GTPases Regulate p38 Mitogen-activated Protein Kinase
534
through the Downstream Mediator Pak1*. (1995). 535
44. Polverino, A. et al. Activation of Mitogen-activated Protein Kinase Cascades by p21-
536
activated Protein Kinases in Cell-free Extracts of Xenopus Oocytes*. (1995). 537
45. Sells, M. A. et al. Human p21-activated kinase (Pak1) regulates actin organization in
538
mammalian cells. |
(1997). 539
46. Frost, J. A. et al. Cross-cascade activation of ERKs and ternary complex factors by Rho
540
family proteins. EMBO J. 16, 6426–6438 (1997). 541
47. Tang, Y. I. et al. Kinase-Deficient Pak1 Mutants Inhibit Ras Transformation of Rat-1
542
Fibroblasts. MOLECULAR AND CELLULAR BIOLOGY vol. 17
543
https://journals.asm.org/journal/mcb (1997). 544
48. Frost, J. A., Xu, S., Hutchison, M. R., Marcus, S. & Cobb, M. H. Actions of Rho Family
545
Small G Proteins and p21-Activated Protein Kinases on Mitogen-Activated Protein
546
Kinase Family Members. MOLECULAR AND CELLULAR BIOLOGY vol. 16
547
https://journals.asm.org/journal/mcb (1996). 548
49. Thillai, K., Lam, H., Sarker, D. & Wells, C. M. Deciphering the link between PI3K and
549
PAK: An opportunity to target key pathways in pancreatic cancer? Oncotarget 8, 14173
550
(2017). 551
50. Liu, J.-J. Regulation of dynein-dynactin-driven vesicular transport. (2017)
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 552
doi:10.1111/tra.12475. 553
51. Höök, P. & Vallee, R. B. The dynein family at a glance. J. Cell Sci. 119, 4369–4371
554
(2006). 555
52. Pfister, K. et al. Genetic analysis of the cytoplasmic dynein subunit families. PLoS Genet. 556
2, 11–26 (2006). 557
53. Pfister, K. et al. Cytoplasmic dynein nomenclature. J. Cell Biol. 171, 411 (2005). 558
54. Nyarko, A. & Barbar, E. Light chain-dependent self-association of dynein intermediate
559
chain. J. Biol. Chem. 286, 1556–1566 (2011). 560
55. Kardon, J. & Vale, R. Regulators of the cytoplasmic dynein motor. Nat. Rev. Mol. Cell
561
Biol. 10, 854–865 (2009). 562
56. Chowdhury, S., Ketcham, S., Schroer, T. & Lander, G. Structural organization of the
563
dynein-dynactin complex bound to microtubules. Nat. Struct. Mol. Biol. 22, 345–347
564
(2015). 565
57. Olenick, M. & Holzbaur, E. Dynein activators and adaptors at a glance. J. Cell Sci. 132,
566
(2019). 567
58. Musarrat, F. Cellular and Viral Determinants of Hsv-1 Entry and Transport. (2019). 568
59. Peeters, B., de Wind, N., Broer, R., Gielkens, A. & Moormann, R. Glycoprotein H of
569
pseudorabies virus is essential for entry and cell-to-cell spread of the virus. J. Virol. 66,
570
3888–3892 (1992). 571
60. Del Rio, T., Ch’ng, T. H., Flood, E. A., Gross, S. P. & Enquist, L. W. Heterogeneity of a
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 572
Fluorescent Tegument Component in Single Pseudorabies Virus Virions and Enveloped
573
Axonal Assemblies. |
J. Virol. 79, 3903 (2005). 574
61. Curanovic, D. & Enquist, L. W. Virion-incorporated glycoprotein B mediates
575
transneuronal spread of pseudorabies virus. J. Virol. 83, 7796–7804 (2009). 576
62. Abramoff, M., Magalhaes, P. & Ram, S. Image Processing with ImageJ. Biophotonics Int. 577
11, 36–42 (2004). 578
579
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 580
Figure Legends
581
Fig 1 Akt is phosphorylated in axons during PRV infection. (A) A Campenot tri-chamber
582
neuronal culture system divided into soma (S), middle (M), and neurite (N) compartments used
583
to separate neuronal cell bodies from axons. Addition of virus or drug into the N compartment
584
allows for the study of axonal responses independent of the cell body. (B and C) Immunoblot of
585
p-S473-Akt in axons infected with PRV 180 in the N compartment. (B) Infections continued for
586
0min, 15min, 30min, 60min, and 180min. (C and D) N compartments were pretreated with
587
LY294,002, Akt inhibitor VIII, rapamycin, or cycloheximide prior to infection. (C) N
588
compartments were harvested 1hpi. (D) S compartments were harvested 1hpi in the N
589
compartment. (C and D) p-S473-Akt bands were normalized to total-Akt bands using a
590
densitometry assay. Mean ± Std Dev with n = 3 for each condition are plotted with ** p ≤ 0.01,
591
*** p ≤ 0.001, using a one-way ANOVA (ns = not significant). 592
Fig 2 Disruption of Akt signaling pathways reduced PRV retrograde infection. (A) N
593
compartment axons were treated with FAST-DiO 12hr prior to infection. PRV 180 or PRV 823
594
were added to axons and at 12hpi S compartment cell bodes were tile-imaged. For conditions
595
involving inhibitor treatment, inhibitor was added to N compartments 1hr prior to infection and
596
at 6hpi unabsorbed virus inoculum and inhibitor were removed from the N compartment. (B)
597
Tile-images of neuron cell bodies in S compartments. Scale bar represents 500um. (C)
598
Quantification of primarily infected cells. The ratio of dual-colored to total green cell bodies for
599
each condition was calculated. Mean ± Std Dev with n = 10 chambers for each condition are
600
plotted with *** p ≤ 0.001, using a one-way ANOVA (ns = not significant). bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 601
Fig 3 Quantification of PRV transport kinetics in axons. (A) N compartments were infected
602
with PRV 180 or PRV 823. 2hpi 30sec videos of moving nucleocapsids in M compartments were
603
recorded. |
Inhibitor was added to N compartments 1hr prior to infection when specified. (B)
604
Maximum intensity projections (bottom) were created from the videos to visualize nucleocapsid
605
displacement. Moving nucleocapsids are represented as tracks in the image (red box). 606
Kymographs (top) were made from the maximum intensity projections to visualize nucleocapsid
607
velocity throughout the recording process. Diagonal lines starting from the upper right corner
608
represent retrograde movement. Horizontal lines represent stationary nucleocapsids. (C)
609
Quantification of the number of moving nucleocapsids in M compartments. (D) The
610
displacement of individual nucleocapsids over the 30sec recordings were measured for each
611
condition. (E) The average velocity of the nucleocapsids moving in the retrograde direction were
612
calculated by acquiring the mean of all instantaneous velocities ≥ 1um/sec. Data represent mean
613
± Std Dev with n = 7 chambers and 5 fields of view per chamber. **** p ≤ 0.0001, *** p ≤
614
0.001, ** p ≤ 0.01, * p ≤0.05 using a one-way ANOVA (ns= not significant). 615
616
Fig 4 Akt phosphorylation in axons occurred after entry of PRV into the cytoplasm. 617
Immunoblot of p-S473-Akt in axons infected with PRV 180, PRV 233, or PRV GS442 in N
618
compartments. p-S473-Akt bands were normalized to total-Akt bands using a densitometry assay. 619
Mean ± Std Dev with n = 3 for each condition are plotted with * p ≤ 0.05, using a one-way
620
ANOVA. 621
Fig 5 US3 induced Akt phosphorylation in axons. Immunoblot of p-S473-Akt in axons infected
622
with PRV 180, PRV 813NS, PRV 815KD or PRV 813R in N compartments. p-S473-Akt bands
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 623
were normalized to total-Akt bands using a densitometry assay. Mean ± Std Dev with n = 3 for
624
each condition are plotted with * p ≤ 0.05, using a one-way ANOVA. 625
Fig 6 US3 does not affect entry of PRV into cells. Fluorescence imaging of PRV 180 and PRV
626
823 nucleocapsids in Rat2 fibroblasts. A synchronized infection assay was used to allow entry of
627
all bound virion particles to occur simultaneously. Nucleocapsids that entered the cytoplasm
628
were manually counted. Data represents mean ± Std Dev with 5 replicates and 2-5 cells per
629
replicate for each condition using an unpaired t-test (ns = not significant). 630
Fig 7 US3 and Akt phosphorylation are required for virus-induced translation in axons. A)
631
N compartments were infected with PRV 180 for 1hr prior to harvest. Puromycin was added to N
632
compartments at 45mpi to label nascent peptides. B) Immunoblot of puromycin incorporated
633
peptides from axons in the N compartment. PRV 180 or PRV 813NS were added to N
634
compartments for 1hr. |
Puromycin was added to N compartments 45mpi. Inhibitor was added 1hr
635
prior to infection when specified. A single band is visible for each condition in this blot,
636
representing the most abundant peptide synthesized at the time of incubation. Other peptide
637
bands become visible at higher exposure times. 638
Fig 8 The model PRV-induced translation in axons. 1) PRV virions bind to nectin-1 receptors
639
on the host-cell membrane. 2) Entry is mediated by fusion of the viral envelope with the cell’s
640
plasma membrane allowing nucleocapsid and tegument proteins to enter the cytoplasm. Inner
641
tegument proteins (US3, UL36, and UL37) stay bound to nucleocapsids. 3) US3 stimulates a
642
PI3K/Akt-mToRC1 signaling pathway to induce 4) translation of axonal mRNAs leading to Lis1
643
expression. 5) Nucleocapsid engagement with the retrograde transport machinery is mediated by
644
the inner tegument and Lis1 and 6) subsequent retrograde transport through the axon occurs. bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . 645
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. |
It is made
available under a
CC-BY-NC-ND 4.0 International license . bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.12.464169
;
this version posted October 13, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY-NC-ND 4.0 International license . |
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.05.463293
;
this version posted October 7, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY 4.0 International license . Memory B cells in Ocular Herpes
Antiviral CD19+CD27+ Memory B Cells Are Associated with Protection from Recurrent Asymptomatic
Ocular Herpes Infection
Nisha R. Dhanushkodi1, Swayam Prakash1, Ruchi Srivastava1, Pierre-Gregoire A. Coulon1, Danielle Arellano1, Rayomand V. Kapadia1, Raian Fahim1, Berfin Suzer1, Leila Jamal1, Lbachir BenMohamed1, 2, 3, ‡
1Laboratory of Cellular and Molecular Immunology, Gavin Herbert Eye Institute, University of California Irvine, School of Medicine, Irvine, CA 92697; 2Department of Molecular Biology & Biochemistry; and 3Institute for Immunology; University of California Irvine, School of Medicine, Irvine,
CA 92697
Running Title: Memory B cells in ocular herpes
Keywords: B memory cells, plasma cells, ocular herpes, asymptomatic herpes, virus-specific B cell. Corresponding author: Dr. Lbachir BenMohamed, Laboratory of Cellular and Molecular Immunology,
Gavin Herbert Institute; Hewitt Hall, Room 232; 843 Health Sciences Rd; Irvine, CA 92697-4390;
Phone: 949-824-8937; Fax: 949-824-9626; E-mail: [email protected]
Conflict of interest: The authors have declared that no conflict of interest exists
$Footnotes: This work is supported by Public Health Service Research R01 Grants EY026103,
EY019896 and EY024618 from National Eye Institute (NEI) and R21 Grant AI110902 from National
Institutes of allergy and Infectious Diseases (NIAID) (to L.BM. ), and in part by The Discovery Center
for Eye Research (DCER) and the Research to Prevent Blindness (RPB) grant. 1
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.05.463293
;
this version posted October 7, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY 4.0 International license . Memory B cells in Ocular Herpes
ABSTRACT
Reactivation of herpes simplex virus 1 (HSV-1) from latently infected neurons of the trigeminal ganglia
(TG) leads to blinding recurrent herpetic disease in symptomatic (SYMP) individuals. Although the
role of T cells in herpes immunity seen in asymptomatic (ASYMP) individuals is heavily explored, the
role of B cells is less investigated. In the present study, we evaluated whether B cells are associated
with protective immunity against recurrent ocular herpes. The frequencies of circulating HSV-specific
memory B cells and of memory follicular helper T cells (CD4+ Tfh cells), that help B cells produce
antibodies, were compared between HSV-1 infected SYMP and ASYMP individuals. |
The levels of
IgG/IgA and neutralizing antibodies were compared in SYMP and ASYMP individuals. We found that:
(i) the ASYMP individuals had increased frequencies of HSV-specific CD19+CD27+ memory B cells;
and (ii) high frequencies of HSV-specific switched IgG+CD19+CD27+ memory B cells detected in
ASYMP individuals were directly proportional to high frequencies of CD45R0+CXCR5+CD4+ memory
Tfh cells. However, no differences were detected in the level of HSV-specific IgG/IgA antibodies in
SYMP and ASYMP individuals. Using the UV-B-induced HSV-1 reactivation mouse model, we found
increased frequencies of HSV-specific antibody-secreting plasma HSV-1 gD+CD138+ B cells within
the TG and circulation of ASYMP mice compared to SYMP mice. In contrast, no significant
differences in the frequencies of B cells were found in the cornea, spleen, and bone-marrow. Our
findings suggest that circulating antibody-producing HSV-specific memory B cells recruited locally to
the TG may contribute to protection from symptomatic recurrent ocular herpes. 2
58
59
60
61
62
63
64
65
66
67
68
69
70
71
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.05.463293
;
this version posted October 7, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY 4.0 International license . Memory B cells in Ocular Herpes
IMPORTANCE
Reactivation of herpes simplex virus 1 (HSV-1) from latently infected neurons of the trigeminal ganglia
(TG) leads to blinding recurrent herpetic disease in symptomatic (SYMP) individuals. Although the
role of T cells in herpes immunity against blinding recurrent herpetic disease is heavily explored, the
role of B cells is less investigated. In the present study, we found that in both asymptomatic (ASYMP)
individuals and ASYMP mice there was increased frequencies of HSV-specific memory B cells that
were directly proportional to high frequencies of memory Tfh cells. Moreover, following UV-B induce
reactivation, we found increased frequencies of HSV-specific antibody-secreting plasma B cells within
the TG and circulation of ASYMP mice, compared to SYMP mice. Our findings suggest that circulating
antibody-producing HSV-specific memory B cells recruited locally to the TG may contribute to
protection from recurrent ocular herpes. 3
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.05.463293
;
this version posted October 7, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY 4.0 International license . Memory B cells in Ocular Herpes
INTRODUCTION
With over a billion individuals worldwide currently infected with herpes simplex virus type 1
(HSV-1), herpes remains one of the most prevalent viral eye infections (1-3). |
Ocular herpes is mainly
caused by HSV-1, which infects the cornea and then establishes latency in sensory neurons of the
trigeminal ganglia (TG). Sporadic spontaneous reactivation of HSV-1 from latently infected neurons
leads to viral shedding in saliva and tears which can ultimately cause symptomatic recurrent Herpes
Stromal Keratitis (HSK), a blinding corneal disease. Despite the availability of many intervention
strategies, the global picture for ocular herpes continues to deteriorate (4). Current anti-viral drug
therapies (e.g., Acyclovir and derivatives) do not eliminate the virus and reduce recurrent herpetic
disease by only ~45% (5, 6). The challenge in developing an effective herpes treatment or vaccine is
to determine the immune correlates of protection (7-13). Profiling humoral immunity in asymptomatic
(ASYMP) and symptomatic (SYMP) HSV-1 infected individuals will further help in the 1) understanding
if SYMP individuals have dampened humoral immune response in natural infection, 2) knowing
immune correlates of protection that will help to understand vaccine efficacy, and 3) standardizing
methods to explore vaccine efficacy. Although the role of CD4+ and CD8+ T cells against HSV-1 reactivation is heavily explored, the
role of B cells is less investigated. Unlike T cell memory which can be tissue-resident, recent studies
suggest a migratory role for memory B cells (14, 15). Memory B cells (MBCs) are potential antibody
secreting immune cells that differentiate following exposure to the virus. Following differentiation,
MBCs remain in the peripheral circulation and secrete antibodies when they are re-exposed to their
cognate antigen (16). MBCs in HSV-1 infected humans is not feasible to explore due to the low
abundance of MBCs in peripheral blood. In fact, the low abundance of MBCs for any specific virus
makes it challenging to study frequency, specificity, and breadth for the virus of interest. MBCs can be
identified by their B-cell receptor (BCR), a membrane bound immunoglobulin (Ig) identical to the
4
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.05.463293
;
this version posted October 7, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY 4.0 International license . Memory B cells in Ocular Herpes
antibody they secrete upon activation (17, 18). In this study, we explored the frequency of MBCs
found in the circulation of asymptomatic and symptomatic HSV-1 infected individuals by two different
approaches –antigen-specific flow cytometry and ELISPOT. To understand the mechanisms that cause a differential humoral response between SYMP
and ASYMP herpes individuals, we studied the levels of several B cell associated ligands and
cytokines like APRIL, BAFF, IL-21, IL-10, IL-17 and TNF- in serum. |
APRIL (a proliferation inducing
ligand) and BAFF (B cell activating factor) are ligands that can be released by proteolytic cleavage to
form active, soluble homotrimers (19-21). They are expressed by immune cells other than B cells and
their receptors BCMA (B cell maturation antigen) and TACI (TNFR homolog transmembrane activator
and Ca2+ modulator and CAML interactor), are expressed exclusively by B cell lineage (22-24). IL-21,
apart from its regulatory function plays a role in maintenance of antigen-specific memory B cells (25)
and IL-10 is known to be involved in germinal center B cell responses (26). Using the UV-B-induced HSV-1 reactivation mouse model, the frequencies of HSV-1 specific
memory B and plasma cells were compared between SYMP and ASYMP mice by flow cytometry and
ELISpot within cornea, TG, spleen and bone marrow and circulation. We found a significant increase
in the frequencies of HSV-specific antibody-secreting plasma B cells within the TG of ASYMP mice
compared to SYMP mice. In contrast, no significant differences were found in the cornea, spleen and
bone-marrow. Our findings suggest that circulating HSV-specific antibody-producing memory B cells
recruited locally at the TG site could contribute to protection from symptomatic recurrent ocular
herpes. 5
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
bioRxiv preprint
doi:
https://doi.org/10.1101/2021.10.05.463293
;
this version posted October 7, 2021. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under a
CC-BY 4.0 International license . Memory B cells in Ocular Herpes
MATERIALS AND METHODS
Human study population and PBMC isolation: Twenty-two volunteers (10 ASYMP, 10
SYMP and 2 HSV-1& HSV-2 negative) were enrolled in this study (Table 1). Peripheral blood was
collected and PBMCs (peripheral blood mononuclear cells) were isolated by Ficol-Paque density
gradient. The cells were then washed in PBS and re-suspended in complete culture medium
consisting of RPMI-1640 medium containing 10% FBS (Bio-Products, Woodland, CA) supplemented
with 1x penicillin/L-glutamine/streptomycin, 1x sodium pyruvate, 1x non-essential amino acids. The
blood plasma was stored in -20ºC for evaluation of herpes-specific antibodies later. HSV-1 infection
was confirmed using a commercially available kit that detects anti-gG antigen of HSV-1
(HerpeSelect1 IgG ELISA, Focus Diagnostics, Cypress, CA). Flow cytometry: PBMCs of HSV-1 infected individuals were stained ex-vivo with the following
flow antibodies listed below. Additionally, HSV-1 gD antigen (Virusys Corporation, Taneytown, MD)
was coupled using FITC or AP lightning kit (Novus Biologicals, Littleton, CO) to detect HSV-1 specific
B memory cells ex-vivo (27). ASYMP and SYMP mice were euthanized and immune cells from
peripheral blood, spleen, bone marrow, TG and cornea were collected for flow cytometry staining for
memory B cells and T follicular helper (Tfh) cells. |
Subsets and Splits