id
stringlengths 24
24
| idx
int64 0
402
| paragraph
stringlengths 106
17.2k
|
|---|---|---|
617e93c2f9f05b6743e75dca | 26 | relative concentration to the S = 3/2 species. In order to understand this difference, we used the ORCA quantum chemistry program in conjunction with a published isomer shift correlation to calculate the Mössbauer parameters of 8, 9, and 10 using DFT at the B3LYP/ZORA-def2-TZVP level of theory. All initial Mössbauer computations were carried out on BP86-optimized structures from X-ray diffraction data in the quartet spin state in accordance with experimental evidence. No counterions were included for anionic 9 and 10, as previous studies have indicated that Mössbauer predictions are largely independent of the counterion. These potassium-free calculated parameters were in agreement with the experimental Mössbauer parameters of thiolate complex 8 and ammonia complex 10, but both δ and |ΔEQ| of amide complex 9 were overestimated by 0.11 mm/s and 3.18 mm/s, respectively (Table ). Since the crystal structure of 9 shows a potassium cation only 3.13 Å from one S atom in the SCS pincer, we hypothesized that including K + may be necessary to properly describe its electronic structure. Thus, we started from the crystallographic structure of 9 and performed Mössbauer calculations on structures that included K + : one with geometry optimization of all atoms (model A) and one with only H atoms optimized (model B) (Figure ). Model C in Figure is the optimized structure of potassium-free 9, whose predicted Mössbauer parameters are listed in Table . In the crystal structure, the potassium is also coordinated to a Dipp aryl group of a neighboring molecule of 9, and this second arene was modeled as benzene in the calculations. |
617e93c2f9f05b6743e75dca | 27 | A is predicted to have δ = 0.19 mm/s and ΔEQ = 0.97 mm/s while B is predicted to have δ = 0.22 mm/s and ΔEQ = 0.84 mm/s. The Mössbauer calculations on model C give a very different quadrupole splitting, and these are listed in Table as the calculated parameters of 9. These results are summarized in Figure , and the drastic difference will be discussed below. all-atom optimization without K + . Potassium is represented as a purple sphere. |
617e93c2f9f05b6743e75dca | 28 | Analogously, we performed Mössbauer calculations on all-atom-optimized 9-crown with potassium included. This structure gave δ = 0.16 mm/s and ΔEQ = -0.91 mm/s, in agreement with experimental values for 9-crown despite the potassium being formally outer sphere with an Fe•••K distance of 8.026 Å. The Löwdin charge on potassium was found to be -0.06 in this structure, suggesting significant delocalization of positive charge onto the coordinated THF and crown ether components. To test the isolated effect of a distant and free potassium ion, we removed the two THF molecules and the 18-crown-6 that were coordinated to potassium in optimized 9-crown and repeated the Mössbauer calculations (Figure ). This model gave δ = 0.33 mm/s and ΔEQ = -4.08 mm/s, which is not close to the experimental spectrum of 9-crown and instead is much closer to that calculated for model C (anionic 9). In contrast to the structure containing THF and 18-crown-6, the Löwdin charge on potassium was found to be +0.99 in this structure. These results indicate that the THF and 18-crown-6 may be accepting most of the positive charge of the potassium cation, which then places enough positive charge close to the iron to alter its electronic structure. Charge reorganization, albeit in varying degrees, has been implicated as a stabilizing factor in the formation of alkali-crown complexes in previous theoretical studies. Additionally, although the anionic model of thiolate complex 8 was in agreement with experiment, we calculated Mössbauer parameters for optimized 8-crown with potassium included to evaluate its consistency with experiment. This calculation gave δ = 0.28 mm/s and ΔEQ = 3.96 mm/s, also in agreement with the spectrum of 8-crown and essentially unchanged from the potassium-free predictions for 8 that are shown in Table . Taken together, the experimental and computed Mössbauer spectra for the models described above suggest that the amide donor in 9 leads to an unusual situation in which its electronic structure is dependent on the presence of a nearby cation. |
617e93c2f9f05b6743e75dca | 29 | To compare the electronic structures of the computational models whose Mössbauer parameters match experiment to those that do not, we plotted the frontier quasi-restricted orbitals (QROs) for models A (reproduces experimental Mössbauer parameters; all atoms optimized with K + ) and C (does not reproduce experimental Mössbauer parameters; all atoms optimized without K + ) in Figure . The QROs show that the shape of the doubly-occupied d orbital (DOMO) for model A resembles dz 2 while the doubly-occupied d orbital for model C resembles dyz. The dz 2 DOMO has 3.7% Fe s character, while the dyz DOMO has 0.0% Fe s character, explaining the significant change in the isomer shift. The ordering of the dxz and dxy orbitals is also switched, but the dx 2 -y 2 orbital is always higher in energy than the remaining four d orbitals due to its s antibonding interactions with the SCS pincer. QRO analysis of optimized 9-crown shows the same d orbital ordering as model A (all atoms optimized with K + ), though dz 2 is 0.90 eV more stabilized than dxy in 9-crown. (Figure ) Additional experiments were performed to probe the influence of the cation on the electronic structure of 9. First, we computationally sampled a wide range of potential locations for a positive charge to learn about the orientational dependence, but the results were not conclusive (Figure ). In addition, we experimentally sought to "remove" the potassium completely by reacting starting material 5-Et2O with N(TMS)2 in the presence of excess (two equiv) of for free thioamides (90-100 kcal/mol), and its pKa of ~21 indicates that the N-H proton is many orders of magnitude less acidic than free thioamides, which lie in the range 11-15 (Figure ). It is seen from the Bordwell equation that the high pKa of 2 raises the BDFE relative to organic thioamides, but the very negative reduction potential (E1/2 = -1.42 V vs. Fc + /Fc) of the iron(II) center lowers the BDFE to a much larger extent. This potential is 1.0 V more negative than E1/2 |
617e93c2f9f05b6743e75dca | 30 | for the Fe 3+/2+ couple in a four-coordinate SNS pincer complex with an NHC as the fourth ligand, and 1.4 V more negative than the analogous five-coordinate SNS scaffold with two THF molecules. The more negative reduction potential in our system may result from the negativelycharged aryl pincer in place of the pyridine pincer, and also from the coordination of three strong phosphine donors. Although the N-H BDFE of 3 is significantly lower than those of free thioamides, it is comparable to other iron systems with N-H bonds in the supporting ligand. In a recent example, the N-H BDFEs of aminobenzenethiolate iron(II) complexes supported by a TACN ligand were determined by measurement of pKa and E1/2. In the parent system, the Fe 3+/2+ redox couple was found at E1/2 = -0.84 V, and the pKa was determined to be ≥ 19.5, giving a BDFE of ≥ 62 kcal/mol; bracketing showed an exact BDFE near 68 kcal/mol. Studies of nickel, platinum, and ruthenium SCS compounds also corroborate decreased N-H bond acidity and more negative reduction potential upon metalation, though these examples do not report bond energies. There are interesting implications of our determination that deprotonating a distant thioamide site gives a highly-reducing iron(II) species, because this could be a mechanism for generating electron-rich active sites in metalloenzymes. This fits into a growing body of evidence that reduction potentials can may be modulated by the protonation state of sites outside the vicinity of metals in metallocofactors like the nitrogenase FeMoco. Electronic Structure. The EPR spectrum of 9 is notable for its high rhombicity. While the spectrum resembles a few other S = 3/2 examples of well-defined iron complexes with large E/D, the most notable comparison is to a square planar S = 3/2 cobalt(II) ONO pincer complex with a chloride fourth donor. Structurally, this dianionic complex is reminiscent of 9, with the ONO pincer coordinating through a central amido N donor and lithium counterions flanking the alkoxide arms of the pincer ligand. The EPR spectrum of this complex was also highly rhombic, with E/D ≥ 0. Fe (~10 -8 s) but slower than the ~10 -12 s that would give negligible paramagnetic broadening. The smaller line widths of 9 relative to those in 8 and 10 indicate that paramagnetic relaxation is faster, which is attributable to its larger E/D that mixes the Kramers doublets. To our knowledge, the quadrupole splitting values of 9 and 9-crown are the lowest reported for any intermediate-spin iron(III) complex, though anomalously small quadrupole splitting has also been reported in a few S = 2 iron(II) complexes with planar ligand environments. Planar ligand scaffolds can induce large positive contributions to the electric field gradient (EFG) on the axis perpendicular to the ligand plane (usually assigned as the z axis), and the Mössbauer quadrupole splitting is proportional to the z-component of the EFG. Interestingly, computational analysis of the cited systems with anomalously small ΔEQ also revealed a doubly-occupied dz 2 orbital, as we found for 9 (Figure ). Thus, the small ΔEQ values imply that the large positive contribution to the z-component of the electric field gradient by the planar ligand is counteracted by a large negative contribution from the dz 2 ground state. Double occupation of the dz 2 orbital also accounts for the decreased isomer shift (which decreases with increasing iron s-electron density) of 9 by greater mixing of the doubly-occupied 3dz 2 with the 4s orbital, which is corroborated by our DFT calculations. |
617e93c2f9f05b6743e75dca | 31 | The question that arises is then: why do amide complexes 9 and 9-crown show evidence for a dz 2 ground state electronic structure while thiolate complexes 8 and 8-crown as well as ammonia complex 10 do not? First, we propose that the π-donating orbitals of the N(TMS)2 ligand in 9 exert a stronger destabilizing influence on the π-symmetry iron dxy and dyz orbitals than the thiolate donor in 8 and certainly more than the ammine donor in 10. Our proposition of a stronger interaction in 9 than 8 is tied to its structure, which shows a short Fe-N bond and a nearly 90° angle between the N(TMS)2 plane and the pincer plane, permitting excellent overlap between dxy and the N px orbital and the dyz and sp 2 -like orbitals aligned roughly along the N-Si bonds. This highly π-donating ligand field closes the energy gap between dz 2 and dxz/dyz in the typical square planar splitting pattern. This explanation alone does not account, however, for the extreme variability in the calculated Mössbauer parameters for 9 depending on whether there is a positive charge near iron. On the basis of our DFT results, we therefore also suggest that the dz 2 orbital can be further stabilized by the presence of a nearby positive charge. Though the complete description of the cation interaction with each of the d orbitals is not elucidated by our work, selective stabilization of the dz 2 orbital may be a result of its shape. A positive charge in the plane of the pincer could interact with the torus-shaped portion of the dz 2 orbital, and one along the z axis can interact with the lobes pointing above and below the pincer plane. The stabilizing interaction of charge with the lobes of dz 2 is illustrated by the comparison between the QRO energies of model A in Figure with optimized 9-crown (Figure ). Model A has positive charge with non-zero projections both onto the z axis and xy-plane while 9-crown has positive charge that is along the z axis. Accordingly, the dz 2 orbital is 0.35 eV lower in energy than the DOMO+1 in 9-crown versus model A. |
617e93c2f9f05b6743e75dca | 32 | The influence of the alkali metal cation is a distinctive aspect of the work reported here, and it pushes forward the growing understanding of the potential tuning role of cations near a transition-metal center. In a relevant example, Yang has shown that appended cations cause substantial electric fields, which can have beneficial influences on catalysis. Electronic structure calculations in these systems did not show that the orbital energies were differentially affected by the cation, leading to the conclusion that the cation mainly exerts an electrostatic effect. |
617e93c2f9f05b6743e75dca | 33 | Tolman has shown the influences of nearby cations on O-H bond dissociation energies. In more recent work, Tomson observed shifts in redox potentials that were attributed to stabilization of a dz 2 orbital, and supported this idea with DFT calculations. Redox potentials were primarily used to probe cation effects in these examples, but incorporation of iron in our complexes enabled us to use Mössbauer spectra that can be directly related to the electronic structure of iron. Through the combination of spectroscopy and DFT, we identified cation-dependent changes in orbital energetics. While the overall spin state is conserved throughout, a b electron shifts to a cationstabilized dz 2 orbital, manifested in Mössbauer spectra by low δ and |ΔEQ|. In the future, this effect may be utilized to control reactivity that depends on the ordering of orbital energies. |
617e93c2f9f05b6743e75dca | 34 | The conclusions from our electronic structure studies of complexes 8 (thiolate), 9 (amide), versus an atypical small |ΔEQ| is observed. Independently, the relatively low distortion from planarity, strong π donation from the amide ligand, and short Fe-N bond in 9 are hypothesized to enact relative orbital energies that are amenable to changing order in the presence of a cation. |
617e93c2f9f05b6743e75dca | 35 | Comparisons to the Nitrogenase FeMoco. The C and S-ligated iron complexes have the same donor atoms as the iron sites in the FeMoco, and we briefly explore the comparison of molecular and electronic structures. The most thoroughly-characterized state of FeMoco is the enzyme resting state (E0), and in this structure, the high-spin belt irons are bridged by sulfide ligands and share a central carbide, leading to a 3S/1C environment. Our complex 8 likewise has an iron center ligated exclusively by three S donors and one C donor. Its Fe-C bond length of 2.00 Å is very close to the 2.01 Å Fe-C distance in FeMoco. The Fe-S lengths, which range from 2.24 to 2.25 Å, are also highly similar to Fe-S bond lengths in FeMoco (2.25 to 2.27 Å). |
617e93c2f9f05b6743e75dca | 36 | Despite being in an intermediate spin state, these bond distances are closer to those in resting state FeMoco than the two previously-reported (SCS)iron complexes, which were in lower oxidation states and thus had longer Fe-C/S bonds. The final state of the enzyme during catalysis is an iron-ammonia adduct (E8). Because turnover from E8 to E0 is redox neutral, ammonia would be expected to bind to either the iron(II) |
66d6fe0b12ff75c3a143ad98 | 0 | Plastics are good examples of the "take-make-waste" production model, with the vast majority of its production coming from fossil petrochemical feedstocks and most of its disposal relying on landfilling or incineration. These practices lead to severe, negative environmental and economic impacts due to their accumulation and biodegradation recalcitrance. Due to the non-renewable origin of plastics, such a paradigm also raises questions about its scarcity and potential impact on production chains worldwide in the near future. Polyethylene terephthalate (PET) is one of the most single-used plastics produced worldwide, with wide applicability in the textile fiber and packing industries. The most common recycling process applied to PET is mechanical grinding followed by melting and reforming. Such processes lead to a downcycling: each cycle produces a material with fewer mechanical properties of interest, which eventually hampers further recycling. In recent years, plastic recycling research has gained new momentum with the identification of a PET hydrolase isolated from I. sakaiensis capable of degrading PET close to room temperature. Many other PET hydrolases have been characterized since then, improving catalysis rates, enzyme thermostability, and reaction conditions. Still, most enzymatic protocols struggle with degrading high-crystallinity substrate or performing well under low-pH conditions, which are the main challenges in reducing the cost of enzymatic PET degradation. More recently, molecular dynamics (MD) simulations have provided many structural insights into the interactions between PET and many different esterases. In addition, other studies have modeled the interaction between carbohydrate-binding modules and PET crystal films to understand the forces driving such interactions, aiming to improve the binding of PET hydrolases onto high-crystallinity regions of PET. At the core of those studies, force field parameters modeling PET nonbonded interactions will determine the accuracy of any prediction. Current force field models available for PET do not capture the complex electronic polarization effects of PET residues, which likely influence the behavior of PET in specific electronic microenvironments, such as enzyme active sites or at PET-solvent interfaces. |
66d6fe0b12ff75c3a143ad98 | 1 | The parametrization protocol used to develop our PET model was based on previously published methods used for the development of protein, nucleic acid, lipid, and carbohydrate Drude parameters. Briefly, molecular polarizabilities, dipole moment, water interactions, and dihedral potential scans were used as target data. A detailed description of our protocol is provided as Supplementary Information. |
66d6fe0b12ff75c3a143ad98 | 2 | To build an amorphous PET material, we used CHARMM 31 to arrange an 18 × 13 array of PET polymer composed by 9-mer chains spaced laterally and vertically to fill a box of 100 Å × 108 Å × 52 Å. CGenFF parameters were used to build the crystal PET chain topology and terminal methylation at both ends of each chain. In total, our simulation box had 234 PET polymer chains. The energy of the system was minimized via 500 steps of steepest descent minimization followed by 500 steps of adopted-basis Newton-Raphson minimization. |
66d6fe0b12ff75c3a143ad98 | 3 | To induce glass transition amorphization of our system, we performed a series of 1-ns simulations in OpenMM 34 to increase the temperature of the system from 300 K to 800 K in an NVT ensemble, in increments of 50 K every simulation. To cool down the system, the reverse strategy was applied under an NPT ensemble while applying a pressure of 50 bar via a Monte Carlo Membrane barostat in which x-and y-axes were scaled anisotropically while the z-axis was allowed to vary independently. A final NPT simulation was carried out at 300 K for 10 ns to allow for equilibration and relaxation of the amorphous PET material model. |
66d6fe0b12ff75c3a143ad98 | 4 | The equilibrated coordinates were then converted to the Drude model in CHARMM and further equilibrated for 500 ps. Position restraint forces were applied to heavy atoms with a force constant of 1000 kJ mol -1 nm 2 . Periodic boundary conditions were applied in all dimensions and pressure was kept at 1 bar via the Monte Carlo Membrane barostat described above. Unbiased production simulations were carried out for both the additive and Drude systems for 500 ns under an NPT ensemble for data collection, saving coordinates every 10 ps. A complete description of Drude conversion and simulation parameters used for all polarizable systems is provided below. |
66d6fe0b12ff75c3a143ad98 | 5 | Additionally, the final coordinates of our Drude PET model material were used to further study the electronic and structural properties of an amorphous PET slab in vacuum and water. To create our slab model in vacuum, we increased the z-axis dimensions of the simulation box by 20 Å on both sides, while keeping the x-and y-axis box dimensions constant to preserve interactions across the periodic boundaries. To build the solvated system, we converted the system to its additive counterpart by deleting the Drude oscillators and lone pairs and solvated the enlarged box with CHARMM-modified TIP3P water to fill the open volume. The energy of the additive system was minimized, followed by equilibration for 500 ps using the protocol described above. The solvated, additive system was converted to Drude following the protocol described below. Equilibration was carried for both in vacuo and solvated Drude systems for 500 ps using the semiisotropic Monte Carlo Membrane barostat mentioned above. For the in vacuo system, the Monte Carlo Membrane barostat with a fixed z-axis was used to impose a vacuum layer. Production runs for each system were carried for 100 ns and coordinates were saved every 10 ps. |
66d6fe0b12ff75c3a143ad98 | 6 | To build a crystal PET film, we generated the coordinates of a small crystal sample by replicating the crystal packing information obtained by Tse and Mak 37 using Avogadro. The system comprised 72 PET polymer chains, arranged into 6 crystal layers such that each layer contained 12 units of hexameric chains. In doing so, we sought to model a thin PET crystal film while preserving the structural and spatial organization between polymer chains. |
66d6fe0b12ff75c3a143ad98 | 7 | The system was then parsed by CHARMM and both ends of each PET chain were methylated. An initial box of 90 Å× 90 Å × 90 Å was built around the crystal film to generate the in vacuo model. TIP3P water molecules were inserted into the box using CHARMM to generate the solvated system. The energy of both systems was minimized via 500 steps of steepest descent minimization followed by 500 steps of adopted-basis Newton-Raphson minimization. Equilibration was carried in OpenMM for 1 ns while a position restraint was applied on all heavy atoms with a force of 500 kJ mol -1 nm 2 . |
66d6fe0b12ff75c3a143ad98 | 8 | The equilibrated coordinates were then converted to the Drude model in CHARMM as described below. Finally, Drude systems were equilibrated for 1 ns. An isotropic Monte Carlo barostat was used to maintain the pressure at 1 bar by attempting box scaling every 25 integration steps. Position restraints were applied to heavy atoms during equilibration with a force constant of 800 kJ mol -1 nm 2 . For the solvated and in vacuo systems of both additive and Drude models, unbiased simulations were performed for 100 ns, saving coordinates every 10 ps. While the production simulations of solvated systems were carried out under an NPT ensemble, the in vacuo simulations employed an NVT ensemble. In all systems, a weak position restraining force of 50 kJ mol -1 nm 2 was applied to the bottom layer of the crystal film to facilitate post-processing reimaging and analyses. |
66d6fe0b12ff75c3a143ad98 | 9 | We tested the performance of our PET parameters when modeling protein binding events by simulating the binding of a hevein, a lectin with chitin-binding specificity from Hevea brasiliensis, to the PET crystal film described above. To do so, we simply placed the hevein such that its center-of-mass was about 14 Å away from the film surface, with its chitin-binding domain facing the PET residues. The protein topology was generated using CHARMM36m force field. The system was subsequently solvated in CHARMM with TIP3P water molecules and neutralized with a total ionic strength of 0.15 M of KCl, including counterions. Equilibration of the additive system, conversion to Drude, and production runs of additive and Drude systems were carried out as described below. For the Drude systems, the protein topology was generated using the latest Drude-2019 parameters. |
66d6fe0b12ff75c3a143ad98 | 10 | We used the same additive and polarizable simulation parameters for the different systems detailed above. After the additive systems were built, all simulations were carried out in OpenMM 7.7. The short-range Lennard-Jones forces were switched smoothly to zero from 10 to 12 Å and electrostatic forces were calculated beyond a real-space cutoff of 12 Å via PME. An integration timestep of 2 fs was used with a Langevin integrator maintaining a temperature of 300 K. Different schemes of the Monte Carlo barostat algorithm were applied as described above for each system. Coordinates were saved every 10 ps. |
66d6fe0b12ff75c3a143ad98 | 11 | For our Drude simulations, equilibrated coordinates of each system was converted to the Drude model in CHARMM. After, Drude oscillators were relaxed via energy minimization in CHARMM by employing 1000 steps of steepest descent minimization, followed by 500 steps of adopted-basis Newton-Raphson minimization. For solvated systems, TIP3P water models were converted to SWM4-NDP model. Equilibration of each system was subsequently carried out using OpenMM 45 using the Langevin integration algorithm 46 and a 1-fs time step. A "hard wall" constraint 47 was enforced on Drude oscillators at 0.2 Å to prevent polarization catastrophe. A dual Langevin thermostat was used to maintain the average temperature of real particles at 300 K while Drude oscillators were kept at a relative temperature of 1 K. Different schemes of the Monte-Carlo barostat were applied to each system as described above. The Lennard-Jones potential was also switched to zero from 10 to 12 Å while electrostatic forces beyond a real-space cutoff of 12 Å was calculated via PME, as is conventional for the Drude force field. Coordinates were saved every 10 ps. |
66d6fe0b12ff75c3a143ad98 | 12 | For each of the systems, we used the last frame of production simulation and relaxed the Drude oscillators via energy minimization in CHARMM, as described above for Drude conversion. We then generated the Connolly surface of the PET molecules (either amorphous slab or crystal array) using the MSMS software with a probe radius of 1.5 Å. Finally, we calculated the electrostatic potential (V (r p )) exerted by PET residues at each vertex of the Connolly surface (r p ) defined by MSMS according to Equation : |
66d6fe0b12ff75c3a143ad98 | 13 | in which q i is the partial atomic charge (Drude oscillators and lone pairs included) and r p is the position of the particle. The vacuum permittivity ( 0 ) was set to 8.8541878128 × 10 -12 F m -1 . By using this approach, we were able to account for heterogeneous electronic microenvironments at the PET surfaces, allowing for a better description of the electrostatic potential. |
66d6fe0b12ff75c3a143ad98 | 14 | Accurate modeling of molecular electrostatic properties is one of the most important aspects of the Drude force field. As shown in Table , the optimized parameters yielded gas-phase total dipole moments and their components that were in good agreement with their QM target values. An interesting challenge arose when fitting the dipole components of MBOA and 2HEB due to the asymmetric electron density distribution on the ring carbons as a consequence of their proximity to the ester group. Given the rotation about the ester carbon-ring carbon bond, we averaged the ESP charge density between ortho and para ring carbons, yielding a smaller dipole moment, which was used as target in our fitting protocol. |
66d6fe0b12ff75c3a143ad98 | 15 | As their parameters were transposed directly from 2HEB and 3CB, we can also assess the quality of these parameters by evaluating MHET, which was also built by transposing parameters from 2HEB and 3CB. As shown in Table for α iso for 12ED and EGDA, respectively. This result is most likely due to the absence of these molecules in the dataset used to fit electronic parameters. For EGDA, specifically, parameters were transposed from 2HEB, which probably influenced the polarizability due to its bulky aromatic ring close to the ester group. For more details, see the Supporting Methods. |
66d6fe0b12ff75c3a143ad98 | 16 | Water interactions were evaluated for all model compounds and PET-derived molecules using geometries shown in Figure . Values of the minimum interaction distance and interaction energy are listed in Tables and. For both BHET and MHET, interaction energies and distances for carbonyl oxygen O4 and the hydroxyl oxygen OE were in very good agreement with QM target data. In case of the ester oxygen O3, interactions were slightly off for the interaction bisecting the lone pairs (O3_180) due to a steric clash caused by a In addition, we evaluated the anisotropic polarizabilities by scanning water interactions along arcs circling the lone pair axis of the carbonyl oxygen and the ester oxygen using MBOA as reference (Figure ). For the majority of the data points, interaction energies and distances were in good agreement with QM target data, although the ester oxygen showed larger deviations of ∼1 kcal/mol for some data points and could be improved in the future by more exhaustively targeting QM polarization anisotropies. |
66d6fe0b12ff75c3a143ad98 | 17 | As a final assessment of nonbonded interactions of our PET model, we calculated the BHET dimer interaction based on PET crystal packing geometry. As shown in Figure , our parameters yielded an interaction energy of -10.54 kcal/mol while the QM target value was -11.54 kcal/mol. For comparison, the same interaction using the additive CGenFF coun-terpart of the BHET model yielded an energy of -9.64 kcal/mol, suggesting that electronic polarization might play a role in PET crystal packing energies. |
66d6fe0b12ff75c3a143ad98 | 18 | Given the importance of interactions between PET and enzymes, we evaluated the interactions between BHET and common protein functional groups to guarantee self-consistency between our parameters and the Drude force field. As shown in Figure and 2C, benzene (BENZ) stacking interactions with BHET are slightly underestimated, with maximum errors of ∼1.5 and ∼1.4 kcal/mol for sandwich and T-shape stacking interactions, respectively. Ethane (ETHA) interactions (Figure ) were also well captured, with a similar interaction energy minimum distance, and well depth difference of less than 0.5 kcal/mol. For methylguanidinium (MGUAN) interactions (Figure ), the minimum distance was shifted slightly outward in our Drude model, suggesting a slight repulsion in the MM model relative to the QM calculations for distances shorter than 3.5 Å. Nevertheless, the magnitude of the interaction energy minimum was well captured, suggesting that cation-π interactions with our PET model should be reasonably represented. In addition, we considered N-methylacetamide (NMA) interactions with the carbonyl oxygen as a model for the well-known σ-hole interactions observed in PET hydrolases. To do so, we scanned the interaction of NMA along both the carbonyl bond axis and at an angle similar to that observed for σ-hole interactions. In both cases, the energy profiles yielded by Drude were able to reproduce both the QM minimum distance and the well depth accurately (Figure and). Finally, we also scanned the interaction of a water molecule perpendicular to the BHET ring plane to assess whether solvent interactions would be correctly captured. |
66d6fe0b12ff75c3a143ad98 | 19 | Finally, we derived new dihedral parameters for torsions that had not previously been optimized for the Drude force field. Careful tuning was required to guarantee that our model would capture the conformational dynamics of PET polymer and its derived molecules. As shown in Figure and 3B, we used 2HEB to derive parameters for the bond between the ester carbon and the aromatic ring and the ester rotatable bond (defined in Figures and, respectively). For both torsions, our Drude parameters were able to reproduce the energy minima and barrier heights of ∼6 kcal/mol and ∼11 kcal/mol. For torsions associated to the ethylene glycol (EG) linkage between PET residues, we used EGDA to scan the dihedrals CX2-O3-CD-CA and O3-CD-CA-O1 as defined in Figure and S3D and shown in Figure and 3D, respectively. The torsion O3-CD-CA-O1 is commonly defined as Ψ and is directly tied to the gauche (Ψ g ∼ 70°) and trans (Ψ t ∼ 180°) torsional populations in crystalline or amorphous PET. In our torsional profiles, two equivalent minima are located at the gauche angles with the trans configurations ∼0.85 kcal/mol higher, in line with relative energetics proposed by Schmidt-Rohr et al. |
66d6fe0b12ff75c3a143ad98 | 20 | We tested our parameters by modeling an amorphous PET material to study its structural and electronic properties under our additive and polarizable models. At room temperature, amorphous PET has a density of 1.34 g/mL. In our production runs, the Drude and CGenFF systems produced average densities of 1.218 ± 0.003 and 1.233 ± 0.002 g/mL, respectively (Table ). A complete density time series are presented in Figure . Although we observed a better in vacuo BHET dimer interaction energy using our Drude model relative to CGenFF, these results suggest a slight overestimation of Lennard-Jones repulsion for some of the ring atoms, the cumulative effects of which may be leading to an overall increase in intermolecular distances in condensed-phase applications. ). To build upon the well-modeled PET structural properties, future work should test our model in terms of mechanical properties such as elasticity, stress response, and pore formation under the influence of applied electric fields. |
66d6fe0b12ff75c3a143ad98 | 21 | To better understand the dynamics and electronic properties, we performed simulations of our amorphous PET slab model in vacuum and in water (Figure ). We calculated the induced dipole moment by subtracting the permanent dipole moment from the total dipole moment of each PET residue in the slab and binned them based on the z-axis coordinate of their centers-of-mass. As shown in Figure , the induced dipole distribution at the center of the slab is close to zero in both systems, most likely reflecting the cancellation on average of polarization influences due to the unorganized structure of an amorphous phase. Interestingly, the induced dipole moments in vacuum trended towards slightly negative values at slab-vacuum interface (|Z| > 25 Å), reaching approximately -1 D in the most exposed monomers (Figure ). These results are consistent with the dipole moment decomposition for model compounds shown in Table , suggesting that PET residues tend to depolarize when exposed to vacuum conditions. |
66d6fe0b12ff75c3a143ad98 | 22 | In contrast, PET residues at the slab interfaces have a positive induced polarization response when exposed to water; that is, their dipole moments increase rather than decrease as in the case of the vacuum systems (Figure ). As shown in the broad induced dipole distributions at the interfaces (|Z| > 25 Å), PET residues sampled induced dipole values up to 3 D, with an average close to 1 D per residue. The cumulative effect of this polarization response among all residues at the slab-water interface is likely to be relevant when modeling solvent interactions or protein-PET binding interactions. Based on these findings, we set out to further characterize the electrostatic potential surface of the amorphous slab. As shown in Figure , the hydration of the slab surface increases the electrostatic potential |
66d6fe0b12ff75c3a143ad98 | 23 | To further test our model, we also built and studied crystal arrays of PET polymer chains in vacuum and in water (Figure ). In crystalline PET, the repetitive linear pattern of the polymer chains (Figure ) forces a Ψ t of approximately 180 • in all PET monomers, yielding nearly 100% trans content. This same pattern was observed in our simulations in vacuum and in water, with average |Ψ t | values of 172 ± 7 • and 172 ± 9 • , respectively (Figure ). |
66d6fe0b12ff75c3a143ad98 | 24 | As with the amorphous PET slab, we also examined the electronic polarization behavior of PET chains in the crystal film as a function of their exposure to the solvent. As shown in Figure , the total dipole moments of PET residues in the additive system were insensitive to the their position in the crystal film, producing an average dipole moment of ∼0.5 D in all layers. Meanwhile, in the Drude system, the total dipole moments were increased among the residues at the crystal surfaces, with layers ±3 manifesting a total dipole moment of ∼1 D. As shown in Figure , the increased induced dipole moment of the PET residues at the crystal-water interface (i.e., layers ±3) arises from the water exposure and this effect is confined to the layer in direct contact with the solvent. Moreover, the small induced dipoles observed for the PET residues in the crystal film in vacuum suggest the inner layers of the PET film produce a very small, localized electronic polarization response on the interfacial layers. |
66d6fe0b12ff75c3a143ad98 | 25 | We also calculated the electrostatic potential at the film surfaces for both additive and Drude systems (Figure ). For all systems, regions of negative potential were mostly coincident with oxygen atoms, whereas regions of positive potentials were closer to the aromatic ring atoms. A comparison between the distribution of electrostatic potential values of all systems can be found in Figure . For the additive systems, the electrostatic potential maps were similar for the systems in vacuum and in water (Figure and 7B) which is expected due to the lack of explicit electronic polarizability. That is, the partial charges assigned to each atom are invariant and therefore are not sensitive to environmental changes. |
66d6fe0b12ff75c3a143ad98 | 26 | Interestingly, the Drude system in vacuum produced a surface potential very similar to the additive systems (Figure ). An additional shoulder peak at ∼1 V was observed, mostly arising from the aromatic ring atoms. Visually, however, the electrostatic potential maps show small differences, mostly associated with locations of positive potential. In contrast, the Drude system in water produced an electrostatic potential clearly shifted towards positive values, leading to a larger distribution of positive potentials across the crystal surface (Figure ). At the same time, the potential was more negative in constricted regions closer to oxygen atoms, reaching values of -2 V, while the strongest potential produced by the Drude system in vacuum was -1.3 V. For comparison, the strongest negative potential produced by the in vacuo and solvated additive systems were -1.4 and -1.5 V, respectively. Overall, such electrostatic effects are likely to impact the water dynamics at the film surface and the desolvation penalty associated with protein binding events to PET crystal films. |
66d6fe0b12ff75c3a143ad98 | 27 | Degrading high-crystallinity PET is still a challenge for current PET hydrolases due to the enzyme binding site structure and poor enzyme adsortion to the PET surface. Recently, carbohydrate-binding modules (CBMs) were evaluated as possible anchoring proteins that could enhance the concentration of PET hydrolases at the crystal film surfaces and, consequently, facilitate catalysis. As such, accurate modeling of protein-PET interactions in multiple media is essential for designing better PET hydrolases or new technologies that could enhance PET degradation. |
66d6fe0b12ff75c3a143ad98 | 28 | We evaluated the performance of our parameters by simulating the binding of hevein, a chitin-binding protein, to our PET crystal film (Figure ). Throughout both additive and Drude trajectories, the residues comprising the chitin-binding domain interacted closely with the PET surface. Residues W21, W23 and H35 engaged in sandwich and T-shaped π-π stacking interactions with PET residues (Figure ), while other polar and charged residues transiently interacted with PET residues through water-bridged interactions. Over the course of our additive simulation, the hevein rapidly bound to the PET surface, although the same hydrophobic interactions were more intermittent. In our Drude simulation, however, the chitin-binding domain rearranged itself in the first 30 ns of the simulation, better positioning W21 and W23 to promote interactions with the PET film. Such dynamic events can be drawn from the time series of interaction energies between the hevein and the PET film (Figure ). By the end of the simulations, the Drude systems yielded an interaction energy more than 20 kcal/mol stronger than the additive system, reflecting the role of electronic polarization in dictating such protein-PET binding events. |
66d6fe0b12ff75c3a143ad98 | 29 | Finally, driven by the difference in interaction energies, we calculated the electrostatic potential at the PET film surface when bound to the hevein protein. As shown in Figure , the PET residues bound to the hevein were polarized, increasing the electrostatic potential at the PET surface directly in contact with the protein. This increase is clearly demonstrated in Figure by the positive shift of the electrostatic potential distribution in the system bound to the hevein protein relative to the PET crystal film in water (Figure ). These results suggest that electrostatic interactions between PET crystal films and CBMs could be further explored to design better PET-binding modules. |
66d6fe0b12ff75c3a143ad98 | 30 | Here, we have described the development of a Drude polarizable force field for PET polymers and its derived molecules. The parameter fitting protocol employed here led to accurate modeling of molecular polarizabilities, dipole moments, and water interactions by targeting QM data. Interactions between BHET and protein functional groups were evaluated, showing good agreement between gas-phase QM interaction energies and our Drude model. Lastly, new dihedral parameters were fit to QM configurational energy profiles, allowing an accurate reproduction of energy minima and barriers separating important configurations. |
66d6fe0b12ff75c3a143ad98 | 31 | Our parameter set was further tested by modeling amorphous PET. Structural properties of our amorphous model reveald a slight underestimation of PET density, but a very good agreement of gauche and trans populations in relation to experimental data. By further studying the electronic properties of a PET amorphous slab, we demonstrated the electronic polarization response of PET residues as a function of their exposure to aqueous solvent. We observed an average induced dipole moment of ∼1 D per residue for the areas most exposed to the solvent, with a maximum value of ∼3 D being sampled. |
66d6fe0b12ff75c3a143ad98 | 32 | We then tested our parameter set by modeling PET crystal films in different environments. We demonstrated how the most exposed layers of a PET crystal electronically respond to water, yielding an average total dipole moment of ∼1 D per residue in the interfacial layers. In contrast, the inner layers contribute very little to any electronic modulation of the most exposed layers. Moreover, we calculated the electrostatic potential surface of the crystal PET film, allowing us to map and quantify how solvent molecules modulate the elecrostatic properties of the PET film surface. |
66d6fe0b12ff75c3a143ad98 | 33 | Finally, we showcase the binding of a hevein to a crystal PET film, allowing us to identify interacting protein residues and demonstrate the role of electronic polarization in driving such interactions, which led to an average interaction energy between hevein and PET film almost 20 kcal/mol stronger than in the additive model. Lastly, we mapped and quantified how hevein binding to the PET film modulates the electrostatic potential surface, which could be a strategy used to design better PET binders. Together, these results represent the first extension of the Drude polarizable force field to synthetic polymers. We demonstrated how important the inclusion of explicit electronic polarization of materials is in describing PET material properties and how these outcomes can help designing more efficient (bio)technologies to tackle the plastic pollution crisis. |
61f74f310716a8f5d341dceb | 0 | Carbon monoxide (CO) is an essential feedstock that is widely used in the synthesis of commodity chemicals such as alcohols, carboxylic acids, and polycarbonates . CO has also been used for metal-refining processes, e.g., the Mond process, which consists of the carbonylation of crude Ni(0) at around 50 °C and thermolysis of gaseous Ni(CO)4 at 180-280 °C . Huge amounts of high-purity CO are thus produced during the removal of contaminants such as H2, N2, CO2, and CH4 from crude materials obtained from the gasification processes of hydrocarbon resources and the steel production industry . In these cases, cryogenic distillation technology is typically applied for CO purification, although technologies based on adsorption, absorption, and membranes have also been explored intensively . In terms of the purity of the produced CO, repeatable adsorption/desorption sequences based on the coordination/dissociation of CO on metal ions such as Fe(II) , Co(II) , Ni(II) , Cu(I) , Cu(II) , and Ir(III) incorporated in solid-state adsorbents have shown exceptional results (Fig. ). For example, Kirschner et al. reported the use of the crystalline solid of a Fe(II)-carbonyl complex that bears a PNP pincer-type ligand for the reversible chemisorption of CO . In this reaction, the adsorption proceeded smoothly under ambient conditions, while the efficient desorption required heating (100 °C) under reduced pressure. The crystalline coordination polymers known as metal-organic frameworks (MOFs) have also been used for the purification of CO . Matsuda and Kitagawa et al. demonstrated the separation of CO from a gaseous mixture including nitrogen (N2), which is the most competitive gas for CO in physisorption-based separation processes due to its similar molecular size, using a Cu(II)-based nanoporous crystalline material; this separation was enabled by the coordination of CO to Cu(II) ions under cryogenic conditions . In this report, desorption was carried out by raising the temperature to 27 °C in a closed system that had undergone a single degassing cycle. Similarly, Long et al. proposed the application of Fe(II)-, Co(II)-, and Ni(II)based MOFs for the purification of CO based on their high susceptibility to adsorb CO . |
61f74f310716a8f5d341dceb | 1 | Various metal-containing sorbents dispersed in activated carbons, zeolite, and silica have also been proven to be potential materials for CO purification ; however, these are usually associated with higher costs and lower metal density compared to molecular-based systems and MOFs. Given the significant impact of purification processes on capital and operating costs in industry (accounting for 40-70%) and on global energy consumption (accounting for 10-15%) , the establishment of a novel strategy for less-energy-consuming and sustainable chemisorption systems is desirable. |
61f74f310716a8f5d341dceb | 2 | To design such systems for CO purification, the affinity between the metal and CO is a central point of consideration (Fig. ). In principle, more rapid and selective adsorption of CO can be achieved by the introduction of low-(or zero-) valent, d-electron-rich metals, as such metals can form thermodynamically favorable metal-CO interactions through their stronger metal-to-CO -backdonation compared to higher-valent metals . However, the strong interaction between low-valent metals and CO greatly affects the desorption efficiency. In fact, the reversible chemisorption of CO with zero-valent transition metals has been achieved under the extreme conditions used in the Mond process (vide supra). Thus, hitherto reported adsorption technologies have predominantly relied on the chemisorption of CO by higher-valent metals to minimize the influence of metal-to-CO -backdonation. Temperature-swing operations, i.e., the use of a higher operation temperature during CO desorption than during CO adsorption, are frequently applied, and desorption is often carried out under reduced pressure (pressure-swing operation) . Against this background, we envisioned the development of a process for the reversible chemisorption of CO using a zero-valent transition metal that could potentially exhibit strong metal-to-CO -backdonation to showcase a novel strategy for the controlled purification and long-term storage of CO. |
61f74f310716a8f5d341dceb | 3 | Herein, we present a method for the reversible pressure-swing chemisorption of CO on a Ni(0) complex at room temperature (rt, indicating a temperature of around 22-27 °C in this work) via ligand substitution (Fig. ). This mechanism for CO adsorption/desorption is based on using a multifunctional carbene ligand with a hemi-labile coordination site, whereas previously reported systems rely on the simple coordination/dissociation of CO on coordinatively unsaturated metal centers (6,9,10,13) (Fig. ). Furthermore, we demonstrate that the use of an ionic liquid (IL) as the reaction medium, i.e., as a dispersant and/or solvent for the adsorbents enhances the desorption effectively, which stands in sharp contrast to the typical use of ILs for the absorption of CO . For the design of the Ni(0)-based system, the choice of ancillary ligand is critical. This ligand should be equipped with a hemi-labile coordination moiety that can compete with the coordination of CO to the Ni(0) centers even in the solid (crystalline) state. We thus focused on the use of N-phosphine-oxide-substituted imidazolylidenes (PoxIms; 1a-b) and the corresponding imidazolinylidene (SPoxIm; 1c), as the N-phosphinoyl group can serve as a hemi-labile ligand to coordinate Ni(0) in addition to the diaminocarbene moiety (Fig. ) . To date, (S)PoxIms have demonstrated various coordination modes toward metals, including coordination by only the carbene atom (-C) , by only the N-phosphinoyl oxygen atom (-O) , and by both the carbene and oxygen atoms (-C,O) ; however, dynamic coordination exchange between the -C and -C,O modes remains unknown. |
61f74f310716a8f5d341dceb | 4 | The molecular structures of 2a, 2c, 3a and 3c obtained from SC-XRD analysis are shown in Fig. . In these cases, the C1 and O1 atoms adopt a syn-orientation with respect to the N-P bonds (C1-N2-P-O1 torsion angle: 5.8(2)° in 2a; 8.5(1)° in 2c; 1.3(3)° in 3a; 6.5(2)° in 3c), indicating that the complexation proceeded via the rotation of the N-phosphinoyl group in free 1, wherein the C1 and O1 atoms adopt an anti-orientation (C-N-P-O: 175.9(1)° in 1a; 179.7(2)° in 1c) . The interatomic distances between Ni and O1 suggest the absence of a bonding interaction between these atoms in 3a During the preparation of the aforementioned complexes, we noticed the partial formation of 2c when a solution of 3c was concentrated in vacuo. In fact, stirring the crystalline powder of 3c at rt for 10 h in vacuo (0.3 mmHg) resulted in the formation of 2c in 50% yield (Fig. ). Prolonging the reaction time resulted in a slight improvement in the efficiency of CO desorption from 3c (20 h, 59%). Nevertheless, further desorption was not expected, as the solids adhered to the inner surface of the reaction vessel, limiting the surface area of 3c exposed to the reduced pressure even under stirring conditions (Fig. ). To promote the desorption, we explored the use of a dispersant. Dispersing 3c into tetradecane (C14H30) in the reaction flask (V = 50 mL) resulted in a significant improvement of the desorption, and 2c was obtained in >99% yield after 2 h at rt with concomitant loss of the crystallinity (Fig. ); however, ca. 2 wt% of C14H30 was removed under the applied reaction conditions. Desorption also proceeded quantitatively within 30 minutes when 3c was fully dissolved in 1,3-dimethoxybenzene (DMB), albeit that the partial removal of DMB (ca. 2 wt%) was again inevitable (Fig. ). |
61f74f310716a8f5d341dceb | 5 | To achieve a fully reusable and reversible chemisorption system that can produce CO of high purity, the concomitant removal of reaction media should be avoided. We thus turned our attention to the use of ionic liquids (ILs), which exhibit negligible vapor pressure. ILs including Cu(I) ions have been explored as potential CO absorbents ; however, these have not yet been used as the dispersant/solvent in the desorption process, probably because ILs can occupy the pores of nanoporous materials. The dispersion of the crystalline powder of 3c in imidazolium-based IL-1 with the anion OSO2CF3 -(OTf -) (350 mg) under reduced pressure resulted in obvious improvement of the desorption to afford 2c in 90% yield (run 1, Fig. ). It should be noted that the yield of 2c was calculated via NMR analysis after the addition of THF-d8 to the resulting mixture. We experimentally confirmed that the addition of THF-d8 causes negligible changes to 2c over a short period at rt; however, after 24 h, 2c partially decomposed to give 1c•HOTf (36%) via deprotonation of the proton at the C2 position in IL-1 (Fig. ). In contrast, 3c did not show any decomposition under identical conditions. Subsequently, we explored the optimization of the desorption conditions. The use of C2methylated IL-2 with the anion NTf2 -resulted in the formation of 2c in 98% yield by preventing the aforementioned decomposition (run 2). IL-3 with the anion PF6 -also exhibited good compatibility with the applied conditions, although a slight decrease in the desorption efficiency was observed (90%; run 3). In contrast, a black precipitate was immediately generated after mixing 3c and IL-4 with the anion CH3SO4 -, and 2c was not formed (run 4). Thus, IL-2 was used in the following experiments. It should be noted that up to 2.1 ×10 -2 M of 3c can be dissolved in IL-2 at 25 °C, which corresponds to a 20% loading of 3c, while 2c shows higher solubility (6.7 ×10 -2 M at 25 °C) (Fig. ). Thus, a significant amount of solid 3c remained during the initial stage of the CO desorption, while little solid was observed after the reaction had completed, as most of the formed 2c was dissolved in IL-2 (vide infra). |
61f74f310716a8f5d341dceb | 6 | The loading amount of IL-2 was optimized by comparing the average yields of 2c obtained in five independent experiments under each loading condition (runs 5-8). When 350 mg of IL-2 was used, an average yield of 65(3)% was confirmed (run 5), while using 100 mg of IL-2 (run 6) furnished a yield of 54(3)%; however, this difference was not confirmed to be statistically significant. Nevertheless, the desorption efficiency significantly decreased when 500 mg (39(3)%; run 7) or 1000 mg (45(3)%; run 8) of IL-2 was employed, even though more 3c could be dissolved in IL-2 under these conditions. Based on the aforementioned results, the CO desorption should occur predominantly on the dispersed solids of 3c, and the amount of IL should influence the efficiency of its dispersion. The amount of IL should be optimized based on the reaction apparatus; accordingly, we employed 350 mg of IL-2 in the reaction vial (V = 2.0 mL) in subsequent experiments. |
61f74f310716a8f5d341dceb | 7 | Under the optimal desorption conditions using IL-2 as the reaction medium, 3c was converted into 2c in 97% yield after 4 h at rt under reduced pressure (Fig. ). Interestingly, only 9% desorption of CO from 3a proceeded under identical conditions , even though the geometric and electronic features of 3a and 3c are almost identical (vide supra). This significant difference in the desorption rate could be interpreted in terms of the structural flexibility of the ancillary ligands. Complex 3b was also subjected to identical desorption conditions, but the resulting yield of 2b was only 19%. No reaction occurred for 3d or 3e. To evaluate the role of the N-phosphinoyl oxygen atom in 3c, we synthesized Ni(-C-1f)(CO)3 (3f), which underwent desorption of CO to generate Ni(-C,P-1f)(CO)2 (2f) in 39% yield, where 1f is a N-phosphanylsubstituted imidazolidin-2-ylidene. These results demonstrate that the hemi-labile behavior of the N-phosphinoyl moiety (3c vs 3e-f) and the structural flexibility derived from the ethylene moiety in the imidazolidin-2-ylidene ring (3c vs 3a) are both essential to achieve the efficient desorption of CO from the Ni(0) center under the applied conditions. |
61f74f310716a8f5d341dceb | 8 | Then, we explored the adsorption of CO by 2 in the presence of IL-2 at rt; this adsorption should predominantly occur in the solvated state given the sufficient solubility of complexes 2 in IL-2 (Fig. ). Stirring 2a/2c and 350 mg of IL-2 under a CO/N2 (1 atm each) atmosphere afforded 3a/3c in 98% yields via the selective adsorption of CO with concomitant precipitation of fine crystals of 3 (3c is shown as an example in Fig. ). The reversible coordination of the Nphosphinoyl oxygen atom was again confirmed to be effective, as treatment of 2f with CO/N2 resulted in the formation of 3f in 47% after 30 minutes. In addition, CO was directly stored in 3c from gaseous mixtures of CO/CH4/N2 (1 atm each) and CO/H2/N2 (1 atm each) in excellent yields through adsorption by 2c. Thus, the present system could also be used for the purification of CH4 and H2 through the removal of the accompanying CO. Ni(0) dicarbonyl complexes 2g and 2h, which bear a bidentate carbene (1g) or phosphine (1h) ligand, respectively, did not react with excess CO even after being dissolved in THF-d8 (Fig. ). We also explored the preparation of Ni(0) dicarbonyl complexes that bear the bidentate ligands with a phosphine oxide group 1i and 1j. As a result, Ni(-C-1i)2(CO)2 (4i) and Ni(-P-1j)2(CO)2 (4j) were obtained even in the presence of excess CO (Fig. ). The results of these experiments show that our strategy based on the use of (S)PoxIm ligands and IL-based media is effective for the precise separation of CO from gaseous mixtures including N2, H2, and CH4. We then investigated the reusability of the present chemisorption system and found a significant acceleration of the rate of CO desorption between the first and second cycles (Fig. ). In fact, 2c was afforded in 87-88% yield within 2 h from 3c prepared under the optimized conditions via either the adsorption of CO on 2c or sequential CO desorption-adsorption reactions from crystalline 3c, whereas 2c was obtained in 65(3)% yield from crystalline 3c (run 5, Fig. ). This result can be explained by the increase in the total surface area of 3c exposed to the reduced pressure, as the crystals of 3c that were re-precipitated after CO adsorption were significantly smaller than those used in the first desorption process (Fig. ). Furthermore, CO was effectively desorbed from 3c even after five desorption-adsorption cycles. It should be noteworthy that 3c can be used for the purpose of CO-storage, as solid-state 3c and the mixture of 3c/IL-2 were stable for at least 7 days at rt and -30 °C, respectively (Fig. ). In the aforementioned five cycles, the mixture of 3c/IL-2 was stored at -30 °C for 14-16 h after each desorption/adsorption cycle completed. These results shed light on the key features of the Ni(0)-based reversible chemisorption of CO, i.e., this system can be reused without the removal/addition of the IL, and crystallization is not essential for the preparation of the adsorbents. To clarify the reason for the obvious difference in the CO desorption rates of 3a and 3c, density functional theory (DFT) calculations were carried out at the B97X-D/Def2-TZVPD//M06-L/Def2-SVPD (for Ni and O) and Def2-SVP for others//gas phase level of theory. First, we identified two plausible pathways that connect 3c and 2c; in the first, the C2≡O2 moiety at the distal position with respect to the N-phosphinoyl oxygen atom dissociates from 3c, while in the second, the C3≡O3 moiety at the proximal position dissociates (for atomic labels, see Fig. . For details of these two pathways, see Fig. ). A significant difference in the activation energy barriers (G ‡ ) of these pathways was observed (+13.3 kcal mol -1 for the former; +17.0 kcal mol -1 for the latter), indicating that the dissociation of CO from 3c should proceed via cleavage of the C2-Ni bond (Fig. ). In the optimized structure of TS1c shown in Fig. , the interatomic distance between Ni and O1 is shortened to 2.53 Å from the 3.14 Å found in the optimized structure of 3c, while the distance between Ni and C2 is elongated to 2.73 Å from 1.82 Å in 3c. Although these results are based on the structures optimized in the gas phase, the transformation of 3c to 2c should proceed via ligand substitution even under the applied experimental conditions. This ligand substitution results in the formation of the intermediate [2c•••CO] (G ° = +12.9 kcal mol -1 with respect to 3c). The 2c moiety in [2c•••CO] exhibits a geometry that is nearly identical to that of the optimized 2c, e.g., the Ni-O1 lengths are 2.31 Å in [2c•••CO] and 2.30 Å in 2c. |
61f74f310716a8f5d341dceb | 9 | Next, the activation energy barriers for the dissociation of CO from the Ni(0) centers of 3a, 3c, and 3f were compared (Fig. ). The values of G ‡ (with respect to that of 3) increase in the order TS1c (+13.3 kcal mol -1 ) < TS1f (+14.2 kcal mol -1 ) < TS1a (+14.9 kcal mol -1 ). This trend is consistent with the experimental results that show that the efficiency of CO desorption increases in the order 3a (9%) < 3f (39%) < 3c (97%) under the applied experimental conditions (Fig. ). The presence of the N-phosphinoyl oxygen atom in 3c minimizes the change in the spatial environment around the Ni(0) center during the CO substitution process, which was evaluated using the percent buried volume (%Vbur) calculated based on the geometrical parameters obtained from DFT calculations . The change in %Vbur (%Vbur) (26) was found to be 3.3 when the %Vbur values of the 1c moieties in 3c, TS1c, and [2c•••CO] were compared; this value is obviously smaller than the %Vbur of 7.5 calculated for the transformation of 3f into [2f•••CO] via TS1f, thus rationalizing the faster interconversion in the former case compared to that of the latter. |
61f74f310716a8f5d341dceb | 10 | A comparison of the coordinates of the Ni, C5, C6, P, and O1 atoms between 3 and TS1 reveals that larger deviations are generated in these atoms during the transformation from 3c to TS1c than during that from 3a to TS1a, highlighting the enhanced flexibility of the skeleton of 1c (Fig. ). The C2, Ni, and O1 atoms in 3c can thus smoothly adopt a suitable orientation for the ligand substitution by reducing the Ni•••O1 distance by 0.61 Å to reach TS1c. In contrast, in the case of the formation of TS1a from 3a, the Ni•••O1 distance must be shortened by 0.69 Å under more structurally restricted conditions, resulting in a larger G ‡ to reach TS1a. The reported monodentate NHCs yielded either nickel dicarbonyl (e.g., 2k-l) or tricarbonyl (e.g., 3d-e and 3m-p) complexes, depending on their steric demand when a single molecule of NHC was treated with a Ni(0) species (Fig. ) . Interestingly, %Vbur values of around 39.5-40.0 seem to represent a plausible boundary that determines whether di-or tri-carbonyl complexes are generated as isolable species. In this context, (S)PoxIms 1a-c and N-phosphanylsubstituted 1f demonstrate unprecedented reactivity to afford both di-and tri-carbonyl complexes and realize their interconversion beyond the possible boundary of %Vbur by effectively scaling the spatial volume around the Ni center. from the SC-XRD analysis, reported in this work (2a-c, 2f, 3a-c and 3f) and the previous works (2k-l, 3d-e and 3m-p) . |
61f74f310716a8f5d341dceb | 11 | The presented preliminary results serve as proof-of-concept for a reusable and reversible chemisorption system for CO based on the use of zero-valent transition-metal complexes at room temperature driven only by pressure-swing manipulation. We believe that the strategy shown in this work, i.e., (i) the construction of a ligand system that functions even in the solid-state for the reversible CO substitution using flexible multifunctional ligands and (ii) the use of an ionic liquid as the reaction medium, will pave the way for the design of an unprecedented molecular-based chemisorption system that can effectively purify (or remove) CO in a low-energy-consuming and sustainable manner. |
65b7c7fb9138d23161faff06 | 0 | 3D bioprinting is a technology with applications in critical biomedical fields, including tissue engineering 1 and regenerative medicine. Bioprinted 3D tissue biomimetics are attractive options for replacing traditional 2D culture and animal models. 3D printed biomimetics can reproduce the physical and biochemical environment of human tissue and better capture signalling pathways and drug responses. In addition, these systems lend themselves to dynamic stimuli that mimic those experienced by cells during regular tissue function. For example, mechanical stimulation of lung cells through cyclic stretching of the lung during breathing is a crucial feature to consider when modelling this tissue, which provides a more accurate model over conventional static cultures performed in 2D. Thus, 3D bioprinting opens the door to the creation of dynamic biomimetic models that resemble the mechanical and biochemical microenvironment of in situ tissues. Several technologies are used for 3D bioprinting, with inkjet, 7-9 laser, and extrusion bioprinting being the most popular. In the latter approach, a bioink is placed in a container (e.g., cartridge, syringe) and extruded through a nozzle, producing filaments and allowing the fabrication of cell-laden 3D structures in a layer-by-layer fashion. The bioinks used in extrusion-based bioprinting are formulations containing hydrogels and embedded cells, and must meet specific requirements to simultaneously result in good printability and cell viability. To form stable filaments and consequently yield high structural integrity and shape fidelity structures, it is necessary to use bioinks with high viscosity, shear thinning behaviour, and fast thixotropic recovery. The elevated viscosity and thixotropic behavior avoid the formation of droplets after extruding the ink through the nozzle and cell sedimentation in the cartridge during printing, while the shear thinning behaviour allows the extrusion to occur at low pressure, protecting cells from high shear stress. Gelatin is the material of choice for many tissue engineering and bioprinting applications due to its natural origin, biocompatibility, low immunogenicity, biodegradability, gelation properties, and low cost, along with the presence of RGD (Arg-Gly-Asp) domains that promote cell adhesion and spreading. Gelatin is fabricated through the acidic (Type A) or basic (Type B) hydrolysis of collagen, which produce polypeptides with average molecular weights between 15 and 400 kDa that undergo reversible gelation at low temperatures (Tg = 20-30 ºC). Gelatin gels are thermoreversible because the noncovalent interactions between the protein chains can be disrupted by increasing the temperature. Gelatin methacryloyl (GelMA) is a modified form of gelatin that can undergo rapid initiator mediated photo-crosslinking. This permits its use in light-assisted 3D bioprinting, resulting in hydrogels that are covalently crosslinked and retain their shape independent of temperature, which is not the case with unmodified gelatin. GelMA retains many of the desirable properties of unmodified gelatin (i.e., biocompatibility, biodegradability, low immunogenicity, low cost, RGD motifs), which has made it a preferred material for bioink formulation. Importantly, the hydrogels obtained from crosslinked GelMA also promote cell adhesion, growth, and proliferation. Despite the numerous advantages that GelMA offers, there are challenges in direct 3D bioprinting of this material via extrusion. At low concentrations and/or physiological temperatures suitable for cell culture (i.e., 37 ºC), GelMA solutions have low viscosity and cannot be successfully printed by themselves with extrusion printers. Therefore, low temperatures (to promote physical gelation) and high total solids content are needed to print constructs with structural stability and good shape fidelity. Yet, using a high concentration of GelMA during printing leads to lower cell viability, adhesion, and migration. In addition, decreasing the temperature to improve the viscosity of the ink complicates the printing due to clogging of the extrusion nozzle, and tighter control over temperature is needed, as small fluctuations can lead to significant changes in viscosity that impact the pressure needed for optimal extrusion. The need for higher GelMA concentrations and tight temperature control affect the printing reproducibility and complicate the optimization of printing workflows. GelMA 3D bioprinting through direct extrusion. (A) Graph of total solids vs rheological modifier content for reported for GelMA 3D bioprinting. The total solid is defined as the sum of the GelMA and modifier content. (B) Temperatures used to 3D bioprint GelMA bioinks with the respective theological modifiers. Data extracted from References . |
65b7c7fb9138d23161faff06 | 1 | Another strategy to improve the viscosity and printability of GelMA-based bioinks is to introduce materials that enhance their rheological properties. Some examples of rheological modifiers used are methylcellulose, alginate, hyaluronic acid, collagen, nanosilicate, gelatin, and gellan gum. Despite the improvement of the printability with the inclusion of these modifiers, in most cases, it is necessary to use a high concentration of at least one of the components of the ink (modifier or GelMA), which leads to a high total solids content in the bioink, which can affect normal cell growth (Figure ). Furthermore, printing of GelMA using rheological modifiers is not usually done at a physiological temperature (37 ºC), making the printing process more challenging and potentially impacting the cells (Figure ). Laponite and gellan gum are notable because low concentrations of these modifiers result in suitable viscosity for extrusion -printed constructs were reported from inks containing 0.5% gellan gum and 3% GelMA at 21 ºC. Another approach reported to decrease the GelMA content in the bioink is using gelatin as an additive and taking advantage of its thermal-crosslinking. Using this strategy, a bioink containing 5% GelMA and 8% gelatin exhibited good printability, and the unmodified gelatin could be removed post-printing by dissolving through a gradual increase of the temperature. Although these approaches have reported some success, developing bioinks that contain low concentrations of GelMA and rheology modifier, and achieve high printability and cell viability at 37 ºC is still an outstanding challenge. |
65b7c7fb9138d23161faff06 | 2 | Recently, our lab developed a versatile strategy to formulate inks for high-fidelity 3D printing of non-extrudable polymers using low percentages of Carbopol (CBP) as a rheology modifier. CBP is a partially crosslinked polyacrylic acid with a high molecular mass that at low concentrations forms high viscosity gels with shear thinning and thixotropic behaviour. Our previous work showed that the formulation strategy works for biopolymers with different crosslinking mechanisms and is amenable to the incorporation of cells and various additives, resulting in reproducible 3D printed structures with high cell viability. In this manuscript, we build on our previous work by developing GelMA-CBP bioinks for extrusion 3D bioprinting of tissue biomimetic constructs. The bioinks developed present many advantages, including excellent rheological properties for extrusion 3D printing (e.g., shear thinning and thixotropy), biocompatibility, biodegradability, and the ability to 3D print cell-laden structures (Figure ). These new bioinks allow GelMA bioprinting at low concentrations, using low amounts of rheological modifier, and printing at a temperature of 37 ºC while maintaining outstanding printability and cell viability, filling a gap in the 3D bioprinting workflows for GelMA-based bioinks. When culturing 3T3 fibroblasts, high cell viability, cell adhesion, proliferation and spreading were observed. The materials were then used to develop a stretchable lung model incorporating primary human lung fibroblasts to study fibroblast-to-myofibroblast transition. Our work solves the three key issues of GelMA-based bioinks for extrusion 3D bioprinting, by allowing the printing of complex constructs using i) low concentrations of GelMA, ii) low concentrations of additive, and iii) physiological temperatures. Our work lays the foundation for using 3D printable GelMA materials in tissue engineering, regenerative medicine, and implantable medical device applications. |
65b7c7fb9138d23161faff06 | 3 | The introduction of methacrylamide groups in the gelatin was conducted according to the method reported by Li, 47 with modifications. First, 5 g of gelatin were dissolved in 45 mL of PBS at 50 °C under constant magnetic stirring. Then, 167 μL of methacrylic anhydride (94%) were added dropwise every 30 minutes for 2.5 h (6 additions, 1 mL total). After the last addition, the mixture was left to react for another 30 minutes. Next, the reaction was stopped by adding 100 mL of PBS, and the product was dialyzed against ultrapure water (18.2mΩ•cm from a Reference A10+ Millipore System, Z00QSVC01 Millipore S.AS., Molsheim, Bas-Rhib, France, subsequently referred to simply as water) for 7 days at 37 °C, with two daily water changes. |
65b7c7fb9138d23161faff06 | 4 | Degree of substitution via NMR spectroscopy and ninhydrin test. H-NMR spectra of unmodified gelatin and GelMA were obtained using a Bruker AV 500 spectrometer. The samples were dissolved in deuterium oxide and the experiment was performed at 35 °C. To obtain the degree of substitution (DoS), the spectra of gelatin and GelMA were normalized against the signal of phenylalanine (6.9 -7.5 ppm). Then, the signal of lysine methylene groups in both spectra was integrated. The calculation was done according to Equation . |
65b7c7fb9138d23161faff06 | 5 | The ninhydrin assay was also performed in order to obtain the DoS of GelMA as the percent of modified amines, using the method reported by Zatorski et al. For this, a calibration curve of unmodified gelatin was obtained using concentrations between 1 and 10 mg/mL. The gelatin was dissolved in PBS and 87.5 μL of each solution were placed in a 96-well plate. Then, 12.5 μL of a 20 mg/mL ninhydrin solution in ethanol was added to each sample. Next, the well plate was sealed with an optical sealing tape and was kept at 70 °C for 30 minutes. Finally, the absorbance was determined at 560 nm with a plate reader (Tecan Infinite M200, Männedorf, Switzerland). A solution of 10% GelMA was prepared in triplicate and read under the same conditions as the calibration curve. The DoS was determined as: |
65b7c7fb9138d23161faff06 | 6 | Ink formulation. Stock solutions of 1.5 wt% CBP (prepared by neutralization of a suspension of CBP in water using 25 wt% NaOH), 15 wt% GelMA (dissolved in water at 60 ºC) and 4 wt% LAP (dissolved in water) were used for the formulation of the inks (Table ). The CBP gel was weighed in a 20 mL conical tube and centrifuged at 1050g for 3 min. Next, the GelMA, water, and LAP were added, and the inks were heated for approximately 5 min at 60 ºC and vortexed. Finally, the inks were centrifuged at 2900g for 5 min to remove bubbles introduced during the mixing process. |
65b7c7fb9138d23161faff06 | 7 | Table . Final composition of the GelMA-CBP inks. All values are reported as wt%. Rheological measurements. Rheological measurements were conducted on the inks and the crosslinked materials in a TA rheometer (HR20, TA instruments, New Castle, DE, USA) using an 8 mm parallel plate geometry. Plots of viscosity of the ink as a function of the shear rate (0.1 to 100 Hz) were obtained to assess the shear thinning behaviour. To determine the thixotropy and fast recovery behaviour of the inks, they were exposed to a constant shear of 1 Hz for 15 s, then the shear was increased to 500 Hz for 10 s and finally decreased to 1 Hz for another 15 s. The storage (G') and loss (G") modulus were measured through oscillation frequency sweeps (angular frequencies between 0.1 -100 rad/s and shear strain of 0.6%). The values of the plateau region (between 1 and 10 rad/s) were averaged to obtain a single G' and G" value for each sample. The inks were then crosslinked in a UV chamber (Anycubic Wash and Cure, Shenzhen, China, λ = 365 nm & 405 nm) for 4 minutes to form puck-like constructs (10 mm diameter, 1 mm thickness) followed by measurement of the rheological properties of the crosslinked materials. Compressive moduli were obtained by compressing the crosslinked samples 100 µm (10% strain) at a constant linear rate of 5 µm/s. The values reported were obtained from the slope of the linear region of the strain vs stress curve. A minimum of three independently prepared replicate samples were measured per condition. |
65b7c7fb9138d23161faff06 | 8 | Tissue processing and characterization. Storage and compressive modulus of samples of porcine brain, lung, liver, and heart were obtained as described above. Fresh tissues kindly donated from a local butcher (J. Waldron Butchers, Hamilton, ON, Canada) were sliced, cut into 8 mm pucklike samples, and soaked in cell culture media for 1 h before the characterization. Similar data was also acquired without submerging the tissue in cell culture media. At least ten replicate samples were prepared and measured per tissue type. |
65b7c7fb9138d23161faff06 | 9 | 3D printing and printability. The inks described in Table were printed using an Allevi 3 bioprinter (Allevi, Philadelphia, PA, USA). The inks were pre-heated to 37 °C, placed in a syringe and centrifuged at 2900g for 5 minutes to remove bubbles before printing. The printing parameters used are shown in Table . Mesh-like and cylinder constructs were printed to assess the printability and shape fidelity. The printability was evaluated from the mesh constructs using: |
65b7c7fb9138d23161faff06 | 10 | Where L is the experimentally measured perimeter of one square and A0 is the designed square area (taking into consideration the decrease in the area due to the experimentally measured printed fibre width). The mesh constructs were designed with dimensions of 1.2 cm × 1.2 cm width, 2 mm height, and 3 mm ´ 3 mm internal squares. At least three 3D printed replicates were measured per condition. |
65b7c7fb9138d23161faff06 | 11 | and LAP (sterilized by filtration using 0.2 µm filters) were mixed to form a pre-ink. The pre-ink was heated to 37 ºC and weighed in a biosafety cabinet to maintain sterile conditions. Next, cells suspended in culture media were added (pre-ink to cell culture media at 4:1 mass to mass ratio). |
65b7c7fb9138d23161faff06 | 12 | 3D bioprinting of bioinks containing 3T3 fibroblasts. The protocol for formulating the bioinks for 3D bioprinting was similar to that used for preparing the bioinks for 3T3 fibroblast-laden cast constructs. After adding the cells, the bioink was heated at 37 ºC, placed in a syringe and centrifuged at 500g. Mesh constructs were bioprinted using the conditions described in Table , and the 3D printed constructs were incubated at 37 ºC and 5% CO2, with culture media changes every 3 days. Two experiments were performed with a final composition of the bioinks of 5%GelMA-0.5%CBP-0.5%LAP and cell concentrations of 0.5×10 6 cells/mL or 5×10 6 cells/mL. |
65b7c7fb9138d23161faff06 | 13 | For experiments with cells seeded on top of the constructs, meshes of 1 cm × 1 cm width, 1.4 mm height, and internal squares of 1 mm side were printed. The samples were sterilized in a UV chamber and soaked in cell culture media for at least 4 h prior to cell seeding. Following this preconditioning of the construct, the media was removed and 50,000 cells per construct were seeded (50 µL of 1×10 6 cells/mL dispersion) and left to adhere for 4 h. After allowing time for cell adhesion, 2mL of cell culture media was added, the samples incubated at 37 ºC and 5% CO2, with culture media changes every 3 days. |
65b7c7fb9138d23161faff06 | 14 | Cell viability and imaging. The viability of 3T3 fibroblasts embedded within cast constructs and bioprinted constructs was assessed on days 1, 3, 7, and 14 using a live/dead assay (LIVE/DEAD TM Cell Imaging Kit, 488/570, Life Technologies Corp., CA, USA) following the manufacturer's instructions. Fluorescence images were taken with an inverted confocal microscope (Nikon A1R HD25, Nikon Canada, Mississauga, ON, Canada) using a 10X/0.45NA objective. Image J (National Institute of Health, Bethesda, MD, USA) software was used to process the images and calculate the cell viability. |
65b7c7fb9138d23161faff06 | 15 | The samples were sterilized in a UV chamber and pre-conditioned in cell culture media for at least 4 h prior to cell seeding. Then, 1×10 6 3T3 fibroblasts were seeded on each construct and incubated at 37 ºC and 5% CO2 for 24 h. After that, the samples were clamped in a previously reported cell and tissue (CaT) stretcher, and subjected to uniaxial cyclic stretching for 24 h at 0-20% strain and 0.33 Hz. A single stretch cycle included a ramp up for a duration of 1 s, hold at desired strain for 0.5 s, ramp down to original baseline position for a duration of 1 s, and hold at baseline for 0.5 s. |
65b7c7fb9138d23161faff06 | 16 | . Dog-bone shape constructs with internal square openings of 1 mm side were printed using 3%GelMA-0.5%CBP-0.5%LAP and the conditions shown in Table . The samples were sterilized in a UV chamber and soaked in cell culture media for at least 4 h prior to cell seeding. Then, 2.5×10 5 hLFs were seeded on each construct and incubated (DMEM-10% FBS-1% A/A) at 37 ºC and 5% CO2 for 4 days. The cells were then starved overnight with FBS-free media (DMEM-1% A/A). Next, the samples were clamped in the CaT stretcher, and subjected to cyclic uniaxial stretching for 24 h at 20% strain and 0.2 Hz in DMEM-2% FBS-1% A/A. To model breathing without pauses, a single stretch cycle included a ramp up to desired level of strain over a duration of 2.5 s that was followed by a ramp down to baseline level of strain over a duration of 2.5 s. Thereafter, the samples were fixed and labeled for αSMA and F-actin quantification. Control samples incubated under static conditions were also analyzed. |
65b7c7fb9138d23161faff06 | 17 | The solution was removed, anti-αSMA primary antibody-Alexa Fluor 488 (1/1000, v/v, 0.5 µg/mL) diluted in 1% BSA was added, and the samples incubated at 4 ºC for 24 h. Then, the solution was removed, and the samples were washed three times with PBS for 5 min each. A solution of phalloidin-Alexa Fluor 647 in DMSO diluted in 1% BSA (1/300 v/v, 0.22 µM) and |
65b7c7fb9138d23161faff06 | 18 | Fluorescence images were acquired on an upright confocal microscope (Nikon A1R HD25, Nikon Canada, Mississauga, ON, Canada) using a 10X/0.45NA objective for quantification images and a 25X/1.1NA objective (water immersion) for high-resolution images. The channels were optimized independently based on the most intense samples to avoid pixel saturation. All the images from experimental and control samples were acquired using the same conditions on the microscope. Z-stack images of the center area of the dog-bone shape constructs were taken and the mean intensity per pixel of the cells was obtained using Image J software |
65b7c7fb9138d23161faff06 | 19 | Synthesis of GelMA. The methacrylation of gelatin was performed using methacrylic anhydride (Supplementary Information Figure ). The successful modification of the gelatin was confirmed through 1 H-NMR (Supplementary Information, Figure ). The GelMA spectra presented two signals of vinylic protons at 5.8 ppm and 5.55 ppm and a signal of methylene protons at 2.03 ppm, which corresponded to the methacrylic groups grafted and were not found in the unmodified gelatin. In addition, the signal of the lysine methylene groups (3.12 ppm) decreased considerably in the spectra of GelMA compared to that of gelatin, indicating that these residues were modified during the reaction. The DoS, defined as the percentage of modified amino groups, of the product obtained after purification was determined via 1 H-NMR and a ninhydrin assay (Supplementary Information, Figure ). There were no statistically significant differences between the DoS obtained from 1 H-NMR (67 ± 3%, n = 5) and the ninhydrin test (68 ± 3%, n = 3). |
65b7c7fb9138d23161faff06 | 20 | Rheological properties of GelMA-CBP biomaterial inks. GelMA is a popular biomaterial in the tissue engineering and bioprinting fields because it is derived from a natural biopolymer, contains cell adhesion (RGD) motifs, is biodegradable, biocompatible, affordable, and the presence of methacrylate groups allows fast and controllable photo-crosslinking. However, the direct extrusion of inks containing low concentrations of GelMA at physiological temperature is challenging without introducing high concentrations of rheological modifiers (Figure ). In this work, we focused on developing a simple strategy to formulate GelMA-based bioinks that overcomes these challenges while achieving high printability and cell viability. The first step was the formulation of GelMA-CBP inks with low total solids concentrations that exhibited adequate rheological properties for extrusion 3D bioprinting at 37 ºC. To do so, the concentration of the photoinitiator (LAP) was fixed at 0.5% and the contents of GelMA and CBP were varied (Figure ). As expected, the GelMA inks without CBP were liquid at 37 ºC at all the tested concentrations. Although the viscosity significantly increased with the addition of 0.1% CBP, such inks were flowing liquids and, therefore, inadequate for 3D extrusion bioprinting. However, when CBP percentages were increased to 0.25% and 0.5%, most of the formulations behaved as gels that did not flow after inverting the vials, indicating they were promising candidates for extrusion bioprinting. The only exception was the ink containing 3% GelMA and 0.25% CBP, where the viscosity was not high enough prevent the gel from slowly flowing after inverting the vial. It is worth noting that 0.25% CPB dissolved in water behaves as a gel; however, when 0.5% LAP is included, the increase in the ionic strength partially disrupts the gel structure and causes it to flow (Figure , panel 3, vial 1). Nevertheless, we observed that at 0.25% CBP, the inks containing GelMA showed a higher viscosity than CBP-LAP alone and passed the inversion test. The fact that both GelMA-0.5%LAP and 0.25%CBP-0.5%LAP solutions are liquids that flow at 37 ºC, but the GelMA-0.25%CBP-0.5%LAP inks were gels, suggests that there are GelMA-CBP interactions that favor gelation. We hypothesize that the free amino groups of GelMA could electrostatically interact with the carboxylic acid groups present in the CBP microparticles, leading to an increase in the viscosity of the formulation prior to any photo-initiated crosslinking. (I) G' and G" of crosslinked pucks made from 3 (blue), 5 (red), and 7% (green) GelMA inks with 0.25% CBP. (J) G' and G" of crosslinked pucks made from 5% GelMA inks with 0.1, 0.25, and 0.5% of CBP. All the experiments were conducted at 37 ºC. Data is presented as the mean and standard deviation for n ≥ 3 independently prepared samples. |
65b7c7fb9138d23161faff06 | 21 | or CBP (Figures ) concentrations while fixing the other component. The storage (G') and loss (G") moduli of the inks showed a dependency on the concentration of either component (Figures ). Increasing the CBP and GelMA concentrations resulted in a proportional increase of the storage and loss moduli. The trend was more pronounced in the case of CBP, where G' increased ~60-fold with a slight variation in the total concentration of solids (0.1-0.5%), indicating that a minimum CBP concentration (i.e., 0.25%) is needed for the ink to present high viscosity and gel-like behaviour. |
65b7c7fb9138d23161faff06 | 22 | Two key requirements to develop a functional bioink for high fidelity extrusion bioprinting are that the formulation presents a shear thinning behaviour that reduces the viscosity significantly during the extrusion process, helping preserve the cell viability, and that the formulation presents thixotropic behaviour with fast recovery that favors the formation of a stable filament after the ink is extruded through the nozzle. The curves of viscosity vs. shear rate for GelMA-CBP formulations indicated that all the inks possessed shear thinning behavior at 37 ºC (Figures ). |
65b7c7fb9138d23161faff06 | 23 | Increased shear rate (such as that experienced by the ink during extrusion through a nozzle) resulted in a reduction in viscosity, which implies a more favorable conditions for cells included in a bioink during extrusion. In addition, the viscosity vs. shear curves shifted to higher values as the concentrations of GelMA or CBP were increased, with a more significant influence of the latter (Figures ), indicating that the CBP concentration largely dictates the resulting viscosity of the inks. To assess the thixotropic behaviour of the GelMA-CBP inks, the process of extrusion through a nozzle was mimicked by alternating shear rates between 1 and 500 Hz (Figures ). |
65b7c7fb9138d23161faff06 | 24 | To evaluate the influence of temperature on the rheological properties of the inks, the shear thinning and thixotropy experiments were repeated at 22 ºC (Supplementary Information, Figure ). At this temperature, all the samples (even those with 0% CBP) produced well-defined gels that passed the inversion test (Supplementary Information, Figure ), although samples without CBP did not form robust filaments when extruded. The trends for G', G", viscosity, and thixotropy were similar at both temperatures (Supplementary Information, Figure ). Higher concentrations of GelMA or CBP resulted in increased G', G" and viscosity while the inks also presented thixotropy and fast recovery. As expected, given the gelation (physical crosslinking) of GelMA at 22 ºC, when evaluated at this temperature the inks showed higher modulus and viscosity. |
65b7c7fb9138d23161faff06 | 25 | While the formulations could be extruded at 22 ºC, printing with inks containing low concentrations of GelMA at 37 ºC is more relevant for cell-laden bioinks. Therefore, the printing experiments going forward were all conducted at this temperature that approximates human physiology. Overall, these results show that GelMA-CBP inks present the required rheological properties for bioink formulation for extrusion 3D printing at physiological temperatures. |
65b7c7fb9138d23161faff06 | 26 | GelMA-CBP inks were made to determine the influence of both components on the storage and loss moduli of the crosslinked materials. Our data showed that the values of G' and G" directly correlated with the concentration of GelMA and CBP (Figures ). Changing the GelMA concentration from 3% to 7%, while keeping the CBP concentration at 0.25% allowed tuning the mechanical properties of the crosslinked hydrogels across an order of magnitude, from G' values of 410 ± 50 to 4700 ± 600 Pa, respectively (Figure ). These G' values show that it is possible to bioprint constructs from GelMA-CBP bioinks capable of mimicking the mechanical properties of a range of soft tissues (e.g., brain, or lung). In addition, we observed that G' and G" values plateaued, showing no statistically significant differences, for pucks made from inks containing 5% GelMA and CBP concentrations > 0.1% (Figure ). However, when CBP was absent, the G' value was almost half that of the samples containing CBP, and the differences were statistically significant. We further compared the G' and G" of the crosslinked materials at 22 and 37 ºC (Supplementary Information, Figure ), and observed no statistically significant differences between values, indicating a full crosslinking of GelMA under the experimental conditions used. |
65b7c7fb9138d23161faff06 | 27 | Unlike the inks, after photo-crosslinking the physical gelation of GelMA does not play a significant role in the mechanical properties of the final constructs. Altogether these results suggest that, even at low concentrations, CBP plays a relevant role in the mechanical properties of the crosslinked materials, and the mechanical properties do not change within the CBP concentration range tested. |
65b7c7fb9138d23161faff06 | 28 | This can be explained by CBP serving the role of an agent that fills in the pores of the GelMA gels, leading to increased mechanical properties of the crosslinked materials. This is supported by images of crosslinked GelMA with and without CBP (Figure ), where it can be observed that he incorporation of CBP into the structure of the pucks considerably improved their transparency. This is relevant if cells incorporated within 3D printed constructs made from GelMA-CBP bioinks need to be imaged through optical microscopy. |
65b7c7fb9138d23161faff06 | 29 | Printability of GelMA-CBP inks. The effect of GelMA and CBP concentrations on ink printability was assessed at 37 ºC by printing cylinder and mesh constructs (Figure ). The cylinders were used to qualitatively evaluate the ability of the inks to produce structurally robust constructs, while the meshes were used to quantitatively assess the printability figure of merit for each ink (Figure ). In general, all inks printed with high shape fidelity, structural stability, and printability at physiological temperature. Figure shows that, at a fixed 0.25% CBP, cylinder-like constructs with 1 cm height were successfully printed at all the concentrations of GelMA tested. The cylinders were able to support their weight and did not collapse, indicating that the mechanical properties were sufficient to maintain structural stability. Furthermore, the 3D printed constructs showed the ability to be stretched and return to original dimensions (Supplementary Information, Figure ), which indicates that these materials are suitable for dynamic tissue biomimetic constructs, e.g., for lung and muscle studies, where the stretching of the cell-laden structure is desired. In addition, the thickness of the cylinders decreased with increasing GelMA concentrations, which implies that at lower concentrations of the crosslinkable polymer the filaments can deform, resulting in broader structures that deviated from the designed shape; thinner, better-defined, and more accurate constructs were obtained from the inks with 7% |
65b7c7fb9138d23161faff06 | 30 | GelMA. This observation was supported by the numerical values calculated for the printability parameter (Figure ). A significant increase in printability from 0.3 ± 0.1 to 0.72 ± 0.03 (with a value of 1 representing a perfect reproduction of the designed shape) was seen when the concentration of GelMA increased from 3 to 7%. This was expected, considering the improved rheological properties of the inks and the formation of a higher density crosslinked polymer network after photo-crosslinking. Qualitatively, the images of the center square openings of the mesh structures show the influence of the GelMA concentration on printability, where better defined openings were obtained at higher GelMA concentrations. The squares obtained with 3% |
65b7c7fb9138d23161faff06 | 31 | GelMA were poorly defined due to the significant deformation and broadening of the extruded filaments prior to crosslinking, while higher GelMA contents produced less deformable fibers, which resulted in well-defined squares. The width of the extruded fiber was also correlated with the printing speed, with a tendency to decrease at higher speeds (Figure ). This result also highlights the importance of the printing speed on the fiber width and, therefore, on the shape fidelity of the bioprinted construct. experiments were conducted at 37 ºC, using 0.5% LAP, and crosslinking with 15% 405 nm light. |
65b7c7fb9138d23161faff06 | 32 | The influence of CBP on the printability of GelMA-CBP inks was assessed at 0.1, 0.25, and 0.5% CBP, while keeping the concentration of GelMA at 5% (Figures ). The cylinders printed with 0.25 and 0.5% CBP showed good shape fidelity and structural integrity, and both inks had good printability values. On the other hand, we found that 0.1% CBP was not enough to print 5% GelMA under the experimental conditions used (15% 405 nm light and 0.5% LAP). The cylinder-like construct collapsed, and it was not possible to print a mesh without a considerable deformation of the filaments. Therefore, the printability of 5%GelMA-0.1%CBP ink was assigned the value of 0. |
65b7c7fb9138d23161faff06 | 33 | This was an expected result since this ink formulation flowed (Figure ) and formed droplets when extruded. Our data indicates that the influence of CBP concentration on printability was considerably stronger than that of GelMA. Varying the CBP concentration from 0.1 to 0.5% resulted in inks that were not printable (0.1%) to ones that presented outstanding printability (0.5%). This trend was also observed in the rheological characterizations of the inks and confirmed that CBP concentration is critical to achieve good printability and shape fidelity through extrusion 3D bioprinting. |
65b7c7fb9138d23161faff06 | 34 | We further evaluated the influence of LAP concentration, and the intensity and wavelength of light used for crosslinking on the inks' printability (Supplementary Information, Figure ). A 5%GelMA-0.1%CBP ink with 0.05% LAP, 10-fold lower concentration than the amount used in the inks described above, presented adequate viscosity for extrusion. This highlights the impact of the ionic strength on the properties of CBP-based gels, given that the ink with the same concentration of GelMA and CBP but 0.5% LAP flowed and was not extrudable (Figure ). |
65b7c7fb9138d23161faff06 | 35 | Because LAP is an ionic compound, its concentration directly affects the hydrogen bonding between CBP microparticles, which is critical for the formation of a gel. Although the formulation with low LAP content presented suitable viscosity for extrusion, the constructs obtained at 15% of 405 nm light (condition used for the printing previously described) had very poor shape fidelity (Supplementary Information, Figure ). The material did not crosslink fast enough to retain the shape during printing, indicating that a higher concentration of LAP is needed under these irradiation conditions. However, when 50% intensity for 365 nm (UV) light was used, constructs with good shape fidelity and integrity were obtained (Supplementary Information, Figure ). |
65b7c7fb9138d23161faff06 | 36 | These results show that 0.1% CBP and 0.05% LAP can be used for the extrusion 3D printing of GelMA inks if high intensity UV light is employed. However, the experimental protocol we used to 3D print the inks keeps a constant irradiation throughout the print, meaning that the prolonged exposure to high intensity UV light could lower the viability of cells in the printed construct. Thus, we elected to keep LAP at 0.5% and used 0.5% CBP as the optimal concentration for the 3D printing of cell-laden inks (bioinks) using low intensity 405 nm light and printing at 37 ºC. |
65b7c7fb9138d23161faff06 | 37 | 3D printing of complex constructs. After demonstrating the high printability and the tunability of the mechanical properties of GelMA-CBP constructs, we explored the ability to print complex structures and tested the limits of each formulation. A largely untapped and exciting application of 3D bioprinting is to use materials with properties resembling those of ECM to build structures, that mimic tissues and organs. To assess the suitability of our bioinks for this application, formulations composed of the optimal concentration of 0.5% CBP and 1, 3, 5, or 7% GelMA were used to print structures at 37 ºC with increasing levels of complexity. While 0.5% CBP concentration is not the minimum that can be used for each formulation, it is a good compromise because it guarantees high printability for this range of GelMA concentrations at a constant CBP content, which simplifies the analysis and allows direct comparisons between inks printed at 37 ºC. |
65b7c7fb9138d23161faff06 | 38 | The first step was to print our standard cylinder and mesh where we found that all these formulations had acceptable shape fidelity and reproducible printability (Figure , Supplementary Information, Figure ). The ink with the lowest GelMA concentration (1%) was used to test the lower limit of total solids that provided adequate printability, since this ink had a water content of 98%. We observed that with this formulation it was possible to print mesh-like constructs and achieve acceptable shape integrity and printability. In the case of the cylinder, the construct could be printed but deformed under its weight when we tried to handle it, showing that tall structures (1 cm) were not structurally sound and easily deformed. Nevertheless, the construct was stretchable (Supplementary Information, Figure ), indicating that there was enough crosslinking in the GelMA network to provide cohesion to the soft material and to allow printing of low-profile structures. |
65b7c7fb9138d23161faff06 | 39 | After confirming the high printability of the four formulations selected, we printed a 5 mm-tall maple leaf outline, an anatomically sized solid human ear, an anatomically sized hollow nose, and a hollow ghost figurine using 405 nm light and at 37 ºC (Figure ). Increasingly more complex models could be printed as concentration of GelMA was increased, since this improved the robustness of the printed materials. There was a very high fidelity of the printed structures and the models for the biomaterial inks with GelMA contents between 3 and 7%. On the other hand, the maple leaf outline was the only structure accessible to the softest ink (1%GelMA-0.5%CBP-0.5%LAP, Figure ), where the final construct presented acceptable shape fidelity and stability. |
65b7c7fb9138d23161faff06 | 40 | All other constructs printed with 3, 5, and 7% GelMA were structurally sound and could be handled without damage (Supplementary Information, Figure ). The crosslinked constructs showcase the ability of the biomaterial inks to print complex models that are challenging, or impossible, to achieve with other protocols by direct extrusion 3D printing. The human ear shape, printed with the 3% GelMA ink, presented layers that are not fully supported by underlying layers, producing slight overhangs that lead to hollow spaces (Figure ). The nose, printed with the 5% GelMA ink, increased the complexity by presenting a completely hollow dome-like structure with soft angles (Figure ), while the hollow ghost figurine, printed with the 7% GelMA ink, presented more pronounced angles and overhanging features (Figure ). Overall, these results demonstrate that complex structures can be printed with GelMA-CBP inks and achieve outstanding printability and shape fidelity while maintaining the desired properties for bioprinting: low concentration of GelMA and additive, and printing at physiological temperature of 37 ºC. The ability to print complex constructs can be critical when using these materials for tissue engineering applications. Crosslinked GelMA-CBP materials can mimic the mechanical properties of tissue. |
65b7c7fb9138d23161faff06 | 41 | Recapitulating the mechanical properties of human tissue is key when developing biomimetic models for basic research and drug development. To determine the potential of GelMA-CBP inks to mimic tissue, we measured the mechanical properties of porcine tissues and compared them with those of 1-7% GelMA formulations. Fresh porcine brain, lung, liver, and heart were sliced and cut to form 8 mm puck-like pieces that match the size of our GelMA cast pucks. The storage (G'), loss (G") and compressive (E) modulus were obtained from the naïve tissue samples and after conditioning in cell culture media (Supplementary Information, Figure ). We found statistically significant differences between the naïve and the cell culture media conditioned samples for the softer tissues (brain, lung, and liver), with the conditioned tissues showing higher moduli than fresh naïve ones. In the case of the heart, the stiffest tissue analyzed, there were no differences between naïve and conditioned samples in all cases. These differences may point to interactions between the tissue of interest and the cell culture media components (e.g., proteins, divalent cations) that lead to stiffening of soft tissues, where the extracellular matrix network is not strong enough to overcome these interactions. In addition, we observed that the samples measured without a previous conditioning step showed a larger replicate to replicate variability, suggesting that the tissues could be drying during processing and data acquisition, yielding variable and progressively higher modulus values. These are key observations that, together with the use of different techniques for the determination of the mechanical properties, could explain some of the high variability observed in the moduli reported in the literature for fresh tissues (the interested reader is referred to Reference 51, where the mechanical properties of human tissue as reported by different techniques is discussed). To ameliorate these issues and obtain the most reproducible and comparable data, we elected to do conditioning step in cell culture media for the GelMA-CBP materials prior to performing the mechanical property measurements. |
65b7c7fb9138d23161faff06 | 42 | A comparison between the mechanical properties of samples with and without conditioning in cell culture media was also performed (Supplemental Information, Figure ). The comparisons of the storage, loss, and compressive moduli for porcine tissues and the crosslinked GelMA-CBP materials are shown in Figure -H (shaded areas present the minimum and maximum values for the tissues measured). It can be observed that compressive moduli, and largely the storage moduli, of the crosslinked GelMA-CBP constructs closely matched those of the porcine organs, indicating that they could be used for building representative biomimetic models. However, when a deeper analysis is conducted, it is evident that the GelMA-CBP materials do not exactly match the three moduli simultaneously. For example, despite the match of the E values between the heart tissue and 7% GelMA, the range of values for G' and G" of the heart were well above that of any of the crosslinked materials. Furthermore, most of the G" values of the crosslinked materials were lower than those of the tissues, indicating a limitation in capturing the viscous properties of porcine tissues. This mismatch of the viscoelastic properties between the GelMA-CBP materials and the porcine tissues depending on the type of rheological measurement points to the importance of performing a complete analysis of the samples of interest and biomaterial inks when developing biomimetic models. It has been previously reported that tissues, including the brain, 52 lung, liver, and heart, present viscoelastic properties that result in a time-dependent mechanical response and stress-relaxation in response to a deformation. This means that to mimic tissues accurately, it is necessary to use hydrogels with the appropriate viscoelastic characteristics. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.