text
stringlengths 4
179k
| predicted_class
stringclasses 5
values | confidence
float64 0
100
|
|---|---|---|
The last core criterion of the EvAAL framework is followed as well, as the third quartile of the point error is used as the final score. The reason behind using a point error as opposed to comparing trajectories using, for example, the Fréchet distance is that the latter is less adequate to navigation purposes, for which the real-time identification of the position is more important than the path followed. | other | 34.44 |
In tracks 1 and 2, the path is kept secret only until one hour before the competition begins, because it would be impractical to keep it hidden from the competitors after the first one in a public environment. However, competitors could not add this knowledge to their systems. In Track 3, the competitors work with blind datasets (logfiles) in the evaluation so the path can be kept secret. In Track 4, a black cover is used to avoid any visual reference of the path and other visual markers. | other | 30.78 |
The logging system is only independent in Track 1. An exception was added in Track 2 for the logging system, which was done by the competitors themselves rather than by an independent application. In Track 3, competitors submitted the results via email before a deadline. In Track 4, the competitors had to submit the results via email within a 2-min window after finishing the evaluation track. | other | 32.34 |
The path and timing was identical for all competitors in Track 3. The path and timing was also identical for all competitors in Track 4. The paths are slightly different in tracks 1 and 2, which involved positioning people in real-time, because the path was so long that it would have been impossible to force the actors to follow exactly the same path with the same timing many times. | other | 32.9 |
In Track 1, actors were not trained to use the competing system. Generally, actors are people from the conference audience or people from the organizing committee willing to support the competition. In Track 1, competitors had to use standard smartphones, with the possibility of using any sensor available on the device: GPS, accelerometer, gyroscope, compass, Wi-Fi radio and barometer. Competitors were allowed to only exploit the existing Wi-Fi access points available on the competition area. Teams in Track 1 were allowed to perform two runs each during the competition day, and consider only the best result. During each run, the actor tested one competing application by using a measurement application developed by the organizers, called StepLogger. | other | 33.44 |
Competitors in Track 2 were allowed to use any IMU sensor module, whether off-the-shelf or self-developed, without any limit concerning the number of devices and the mounting position on the body. Laptops, tablets and smartphones could be used to process data gathered from the Pedestrian Dead Reckoning (PDR) system in real time. Competitors were requested to provide the organizing committee with the sensor raw data, the estimated positions of the system and the key point indexes according to a specific log format. Competitors were not requested to use an external measurement application and could generate the log themselves. | other | 33.22 |
Competitors of both tracks received detailed geo-referenced maps of the competing area and were able to survey the area during the days before the competition day. They used the survey time to check how their system performed in the area and were able to perform algorithm tune-up. | other | 33.84 |
During the survey, competitors in Track 1 were mostly interested in taking Wi-Fi measurements, both fingerprints in various locations (a long and tedious task) and checks on the position of access points in the area. Most access points were, in fact, indicated on the maps, but some were not, and up-to-date ids (MAC, SSID) were not provided to competitors. After the competition, many competitors in Track 1 informally said their system would have performed much better if they had spent more time in taking Wi-Fi measurements during the survey, because the area was so big and the hours they dedicated to the task were not enough. Most notably, the winning Track 1 team needed very little survey time. To the amazement of bystanders, including some other competitors, after slightly more than half an hour spent in a specific small area of the building, they claimed that they had collected all the data required by their system. | other | 31.22 |
A reference path is necessary to measure the accuracy of the competing systems, and was used as the ground truth. In practice, some dozen markers (keypoints) were stuck on the floor and their coordinates were measured in advance. The actor followed the markers sequentially stepping over them. For Track 1, the synchronization between the real position of the actor and the estimated position was guaranteed by the StepLogger app, running on the same smartphone where the competing app was running. When the actor steps over a mark on the floor, he pushes a button on the smartphone’s screen, and the app records the time. Since the markers are walked in a predefined order, the time stamps are easily associated with the corresponding mark and logged to persistent memory. The log thus produced is then compared with the log of continuously estimated positions provided by the competitor. | other | 33.62 |
In Track 2, the same time-stamp log can be provided by the competitor if he/she uses the StepLogger app. Since most of PDR systems are implemented on laptops, it is not easy to have a communication interface with the StepLogger app, which runs on Android. Therefore, competitors generated logs including time-stamp, sensor raw data, estimated position and marker indexes indicating whether the position value is calculated at the reference point or not. | other | 32.38 |
In both Tracks 1 and 2, the organisers provided an application for producing the results from the time-stamp log and the log of competitor estimates. The applications read the space-time logs and computed the errors as xy distance plus floor penalty. The third quartile of error is then computed and presented as the final score. A dedicated application, available on the EvAAL website, produces a graphical representation of the map including the ground truth and the estimated path. | other | 32.16 |
The path was defined with the goal of realistically reproducing the way people move within wide indoor environments. In fact, path complexity is a distinguishing feature of the IPIN competition. For this purpose, the following rules were considered:stairs (for both Tracks 1 and 2) and a lift (Track 1 only) are used to move between floors;the path traverses four floors and includes the patio, for a total of 56 key points marked on the floor, 6000 m2 indoor and 1000 m2 outdoor;actors stay still for few seconds in six locations and for about 1 min in three locations; this cadence is intended to reproduce the natural behaviour of humans while moving in an indoor environment;actors move at a natural pace, typically at a speed of around 1 m/s;total length and duration are 600 m, 15’ ± 2’, which allows to stress the competing apps in realistic conditions. | other | 32.25 |
As mentioned before, the setting was the Polytechnic School of the University of Alcalá (EPS-UAH). EPS-UAH is hosted in a square building composed of four floors connected with stairs and lifts. Floors all have a similar layout, with a side of about 140 m. The central big and open round hall gives access to four wings where medium sized rooms are located. The structure is mostly made of concrete with many pillars in the central hall. Glass walls separate the patio from the ground floor indoor areas. Wi-Fi is available inside and immediately outside the building. | other | 29.44 |
The path is shown in Figure 4. It is composed of 56 key points, five of which were placed in the patio. The path starts from ground level, up to floors 1, 2 and 3 by means of stairs, then goes to a terrace about 50 m long and proceeds to the ground level, goes to the patio and indoors again. When going to the lower floors, the lift is used for Track 1, the stairs for Track 2. | other | 31.56 |
Key points were labelled with a tag in the form [building ID, floor, marker ID] (Key point labels were written on the button displayed by the StepLogger app, to reduce errors on the actor’s part which would require him to restart the path from the beginning). They were placed on the floor following these criteria:key points were placed in easily accessible places where people usually step over;distance between key points ranged from about 3 to 35 m, with a median of 8 m;each key point was visible from the previous one, to ease the movement of the actor and reduce random paths between two consecutive key points. | other | 33.44 |
NavIndoor and Samsung teams obtained the best results, respectively with a third quartile score of 5.4 m and 8.2 m, while the remaining teams had scores higher than 15 m, as shown in Figure 5. Note that the localization errors are in the order of a few metres, which is acceptable for navigating an indoor environment and consistent with the EvAAL criteria. Again, we stress that we tested real working systems in a realistic environment with a realistic usage pattern and rigorous criteria: no simulations, no small or controlled environments, and no simplifying assumptions. | other | 35.22 |
The results of Track 2 show a high deviation of positioning errors among the six competing teams. NESL produced the best result with a third quartile score of 1.5 m. The second ranked team, SYSNAV, showed very good performance during the initial phase, but the final score was 26.2 m. Such negative performance is due to the sunlight, an unexpected source of error, encountered when the competitor walked along a terrace on the fourth floor. All the other teams produced higher scores than 40 m, as shown in Figure 6, the main cause for this being inaccurate initial heading evaluation. Considering PDR competitors were not allowed to exploit any information gathered from the competition area, such as the Wi-Fi networks, handling of initial heading was a key factor for obtaining reasonable performance. | other | 31.3 |
Competitors were asked to design and implement their localization systems exploiting a total of 17 calibration log files. Logs were publicly available since the very beginning of the call to guarantee that all competitors had the same time to set up their systems. | other | 32.1 |
Competitors used the log files generated by the GetSensorData application (http://lopsi.weebly.com/downloads.html) in which a user walks along a predefined path. Logs are text files where each row contains data received from a different sensor available on the the phone at a given time. Each row begins with the sensor identifier, as four capital letters, that determines the kind of sensor data (Wi-Fi, Accelerometer, Magnetometer, etc.). Then, the timestamps and sensor readings follow, separated by a semicolon. The reference data to train/generate the IPSs was geo-referenced. So, the position of well-known landmarks trough the path were provided as special entries in the log file. For the evaluation log files, explicit location of landmarks was not provided. A simple Matlab-based application was provided to competitors in order to parse and re-arrange the log files (http://indoorloc.uji.es/ipin2016track3/). | other | 32.56 |
The competitors had to use databases provided by organizers to train and calibrate their systems. Those databases contained data from any sensor available on the smartphone used to log the track performed by an external actor (competition organizers and students). The competitors were not allowed to record any additional measurements on-site. Similarly, data for evaluation was also provided by organizers. | other | 33.62 |
Participants received a set of calibration log files with data from smartphone sensors which were recorded on different paths in a total of four different buildings. Different scenarios (four buildings), smartphones (four device models) and actors (eight people) were involved at this stage in order to maximize the heterogeneity of data. A total of 10 routes, considering the four buildings, were selected according to the same realistic-based goals established for the first and second track (see Section 4.1.2), which included 1025 calibration points (key points). | other | 32.97 |
Nine evaluation trajectories, or paths, were defined with the goal of realistically reproducing the way the people move within wide indoor environments, as was done for the evaluation in Tracks 1 and 2. As mentioned before, the high path complexity is a distinguishing feature of the IPIN competition. For this purpose, the following rules were considered:the stairs and the lifts could be used to move between floors;the paths traverse four floors in the UAH building, one floor in the CAR building, six floors in the UJIUB building and four floors in the UJITI building;paths in the CAR building also include an external patio;the nine paths cover a total of 578 key points;total duration is 2 h and 24 min, which allows to stress the competing applications in realistic conditions;actors may stay still for few seconds in a few locations; this rhythm is intended to reproduce the natural behaviour of humans while moving in an indoor environment;actors move at a natural pace, typically at a speed of around 1 m/s;phoning and lateral movements were allowed occasionally to reproduce a real situation;competitors have all the same data for calibrating and competing. | other | 31.95 |
As already mentioned, the setting was composed of four different buildings in different cities. The UAH building is in a university environment with corridors, classrooms, laboratories, among other typical facilities. The UJITI building has similar facilities but the Wi-Fi network infrastructures are different, since UAH and UJI have different strategies to deploy Wi-Fi. Unlike the two previous buildings, CSIC is a research institution and the UJIUB building has many offices hosting technological companies. In general, each of the selected buildings has unique features (occupancy, construction materials, architectural elements, Wi-Fi network policies, among many others). | other | 33.34 |
Competitors tested their localisation algorithms with nine evaluation log files and they sent us the estimated locations in a predefined format. The evaluation log files were not geo-referenced so competitors had to exploit the information collected by all sensors in order to provide an accurate position estimation. It is important to mention that once the evaluation log files were released, all competing teams had strict timing to provide their results. | other | 30.22 |
Since the competitors had just one chance to submit results without feedback to readjust their systems on-the-fly, up to three different results from each participant were accepted in this track, but only the best result was considered for the final ranking. In Tracks 1 and 2, the competitors had two chances, but they had feedback about their accuracy after the first run. The evaluation was performed off-line before the conference. | other | 32.56 |
The HFTS and UMinho teams obtained the best results, with 5.9 m and 7.3 m third quartile. BlockDox and FHWS provided slightly higher scores, at 7.8 m and 8.8 m respectively. The remaining team provided a score higher than 40 m. Results of Track 3 are shown in Figure 7. | other | 31.73 |
Track 4 is significantly different from other tracks due to three main facts:The tracked element is an industrial robot;The task not only needs discrete and usually well separated key positions to be estimated, but a detailed tracking of the actual precise and unknown robot trajectory;Competitors could put sensors on board as well as locate them on given poles around the navigation area. | other | 31.3 |
Participants decided both the nature and the number of sensors, and had up to 45 min to assemble, mount and calibrate their systems. Three reference points in the scenario were given to allow sensor calibration; one of them was the start/stop point for the mobile unit. Competitors had to report the consecutive points of the robot trajectory at least every 100 ms (minimum frame rate). The required information was the x,y coordinates (in millimetres) with respect to the starting point, and the corresponding time stamp. | other | 34.38 |
The chosen mobile unit was a Standard Easybot from the ASTI international robotic company (see Figure 8 bottom-right). The Easybot is an automatically guided vehicle implementing a magnetic guidance and its nominal velocity is 0.6 m/s, tracking the designated magnetic tape thanks to the feedback provided by a magnetic detector and the corresponding closed-loop control. | other | 31.34 |
The magnetic path was a closed route including circular and straight sections alternately arranged (see Figure 8 top). The magnetic tape was placed on a trajectory that was previously defined according to a mathematical description; this way, we assured a known ground-truth, identical for every experiment (see Figure 8 bottom-left). The path length was 32.84 m; the trip time was around 3 min, including programmed linear velocities changes, which were equal for all participants. A black cover avoided any visual reference of the magnetic path (see Figure 8 bottom-right). A mark placed on the robot defined the reference point to be tracked by competitors. | other | 34.44 |
Track 4 received six requests for participation from teams using different technologies (UWB, visible light, acoustics and laser), but only five of them formally registered . Two of them were local teams from the hosting university (University of Alcalá) and were considered as off-track competitors, three of the remaining groups formally registered, but only two teams competed and were evaluated: the ATLAS team and the TopoRTLS company . | other | 30.9 |
In smartphone-based Track 1, we observed that tests done on a realistic and long path stress the competing applications, which require a good degree of maturity to run regularly for long time periods. The independent actor may behave differently to how the competitors had anticipated. We think that these are key features for assessing the performance in realistic conditions. | other | 33.28 |
In PDR Track 2, we observed that a correct initial heading is important for obtaining good performance: for three teams, a bad initial heading produced a path rotated by some degrees with respect to the ground truth, resulting in bad performance even in the presence of a correctly-shaped path. Therefore, it could be considered to provide competitors not only with the starting point coordinates, but additionally with initial heading information for Track 2. As a second note, quality of sensors turned out to be an important factor. Three teams used commercially available sensors; the others used self-developed sensor modules: the lack of engineering in the latter was sometimes crucial. In one case, a sensor module misbehaved, so the team missed the first trial; in another case, the Bluetooth communication between the sensor module and the user mobile device was unreliable, resulting in incomplete log files. | other | 30.52 |
We have shown how diverse current indoor location solutions are and how the different environments, different test areas, different sensor technology, etc., have a significant impact on how location results are processed and evaluated, which makes it difficult to directly compare performance. | other | 31.12 |
We gave a review of the ways to approach this problem in the literature, either by using the performance measure indicated by the authors themselves, or by means of a competition. One additional approach to compare algorithms in controlled conditions has been to set up measurement databases; however, differences in formats, recording procedure and range of sensor used again makes this approach not straightforward. | other | 27.94 |
In the off-line Track 3, we observed that the best results are similar to Track 1, which is a compelling argument for the usefulness of off-line tests. On the other side, the NavIndoor team from FHWS participated in both tracks with the same system, and it reported lower accuracy off-line than in real time. It can be argued that the presence of both real-time and off-line tracks with similar characteristics has enriched the IPIN competition. Furthermore, the data sets are open-access and they can be reused by teams that did not participate in the competition. Therefore, the research community can make new comparisons of state-of-art systems. | other | 33.7 |
In robotics Track 4, the resulting localization accuracy is better than in the other tracks due to the fact that the navigation area is much smaller, the ground truth path has a much higher measurement accuracy and the teams had the possibility of installing their specific equipment on-board and off-board the robot. A consistent and sufficiently precise ground-truth can be achieved by using a mathematically modelled trajectory and considering time sequence information. Similar to the smartphone-based Track 1, time necessary for equipment placement and calibration, which depends on system complexity, has been demonstrated to be an important performance-limiting factor. On the other hand, the competition set-up has to consider enough space for an industrial robot performing a diverse and complex trajectory and competition rules must provide enough description about physical elements available for participants to deploy their systems, together with size, weight and safety constraints. | other | 29.6 |
As a possible solution, we propose the use of the EvAAL benchmarking framework, which we claim has the potential to become a standard way to compare systems in different application areas. It has been used since 2011 and lately in the 2016 IPIN competition, where four different tracks were focused on different use cases with significant variation along several dimensions: person vs. robot, smartphone vs. custom hardware, single vs. multi-storey building, single vs. multi-building environments, on-line vs. off-line processing. Our claim is that applying the EvAAL rigorous criteria makes it possible to directly compare the performance of heterogeneous systems in a more significant way than with other existing methods. We back this claim by presenting the details of the different tracks in the IPIN competition and showing how, in their diversity, they are still comparable. | other | 32.97 |
Experience gained during the IPIN 2016 competition and personal feedback from track organizers and competitors encourages the idea that the EvAAL framework is flexible and robust enough to successfully fit very different use cases when comparing indoor positioning systems. | other | 32.9 |
Finally, one of the main challenges of evaluating indoor positioning systems is considering the diversity of contexts and scenarios during the evaluation. In the 2016 IPIN competition, as it has been already mentioned, a few different contexts and scenarios were considered through the four tracks with the assumption of the normal actor’s movement. The spirit of EvAAL is to be open and integrate new tracks whenever it is possible. For instance, one difference with respect to the previous 2015 EvAAL-ETRI competition was the inclusion of robot-based Track 4. Databases to cover many multi-tier diverse buildings were introduced in the 2015 edition. As future work, the EvAAL community is open to discussing the inclusion of new tracks where the actor’s movement might be erratic (e.g., building evacuation in the case of a fire alarm), non-constant (e.g., sport tracking) or presenting non-walking patterns (e.g., people with mobility impairments) which are realistic movements in some contexts. | other | 29.16 |
The genus Listeria is currently comprised of 17 species, including 11 Listeria species described since 2009 (1). However, the genus Listeria sensu stricto includes six species: Listeria innocua, Listeria ivanovii, Listeria grayii, Listeria monocytogenes, Listeria seeligeri, and Listeria welshimeri. These species are well documented and are known to be commonly found in different environments throughout the world (2–6). Among all the Listeria species, L. monocytogenes is recognized as one of the most important foodborne pathogens in many industrialized countries. This pathogen is responsible for listeriosis, a potentially fatal disease that may lead to abortion or serious cases of meningitis, encephalitis, and septicemia (7, 8). Although listeriosis infections are uncommon, mortality rates can reach 30% in at-risk population groups (9–11). In 2015 in the United States, L. monocytogenes was responsible for an estimated 116 cases of listeriosis, 111 hospitalizations, and 15 deaths (12). | other | 30.83 |
While commercial, conventional production represents the majority of the U.S. poultry market, alternative production systems (e.g., organic, all-natural) are becoming more prevalent and there is very limited information related to the prevalence of Listeria spp. within this type of farm environment (35). Therefore, the goal of this work was twofold: (1) determine the prevalence and distribution of Listeria spp. within poultry-related environmental samples (feces and soil) during live production on pastured poultry farms and (2) evaluate whether the distribution of recovered Listeria spp. could be explained by a differential growth in the enrichment broth used in this study, or accurately reflected the native species distribution in the environment. | other | 36.34 |
Ten farms within the southeastern United States were sampled over a period of 3 years (from 2014 to 2016), representing 37 pasture-raised broiler flocks. Farm descriptions are available in Table 1. Soil and feces samples were collected from the pasture where the flock was currently residing at the time of sampling. Samplings occurred three times during grow-out: (i) within a few days of being placed in the pasture, (ii) halfway through their time on pasture, and (iii) on the day the flock was processed. At each sampling time, the pasture area was divided into five separate sections, and five subsamples in each section were pooled into a single sample for each section (a total of five soil samples and five feces samples were collected on each sampling day). Soil samples were collected from the surface (0–7 cm) with sterile scoops, and feces samples were collected from fresh droppings on the soil surface. Gloves and scoops were changed between sample types and between sampling areas. Samples were transported back to the lab on ice and processed within 2 h of collection. A total of 1,110 samples (555 feces samples and 555 soil samples) were collected over the 3-year study period. | other | 32.97 |
Listeria species have been isolated from a wide variety of environments including nature (13) and urban areas (14), agricultural environments (15), food processing plants (16, 17), and retail food (18, 19). The occurrence of Listeria species is of special interest in the food production chain due to the significant threat that L. monocytogenes represents to public health (20, 21). Numerous studies have investigated the occurrence of L. monocytogenes in final products and in food-processing and retail environments, thought to be the main source of contamination for the final product (21, 22). However, limited information is available on Listeria prevalence in poultry farm production environments. In a farm-to-fork approach, it is necessary to assess the incidence of L. monocytogenes along the entire production chain and particularly at the primary production step (the farm environment), taking into account that it could be a potential source of this pathogen into food processing plants. Few studies have investigated and characterized Listeria species in the farm environment, with L. innocua being the predominant species found on grow-out farms, representing ≤78% of all isolated Listeria species (15, 23–26). Other species such as L. ivanovii, L. monocytogenes, L. welshimeri, and L. seeligeri have also been identified in environmental farm samples or chicken feces, but their detection remains infrequent (15, 23). | other | 30.83 |
The initial isolation of L. monocytogenes may be difficult due to its low cell number within the larger indigenous microflora of environmental samples. Thus, the detection of Listeria species involves selective enrichment procedures. Numerous one-step and two-step enrichment broths have been described during the past 50 years (27), with the three most commonly used procedures being the (1) modified ISO 11290-1, (2) USDA-Food Safety Inspection Service (FSIS) Microbiology Laboratory Guide (MLG) method 8.10, and (3) U.S. Food and Drug Administration Bacteriological Analytical Method (FDA-BAM) method #10. Several studies have shown that the enrichment procedure can result in L. monocytogenes being overgrown by other non-pathogenic Listeria species in samples where multiple species are present (28–32). This has especially been demonstrated with L. innocua, whose presence may mask L. monocytogenes and lead to false negative results (33, 34). In addition, several studies have shown that among Listeria strains of food origins, L. innocua grows faster than L. monocytogenes in enrichment media cocultures or food matrices (28–32). These observations raised the question whether the higher prevalence of L. innocua observed in samples from the farm environment is due to a differential growth of Listeria species during the enrichment process or reflect their true distribution in the environment. | other | 32.66 |
Enrichment and isolation of Listeria from these environmental samples were performed using a modified version of the USDA-FSIS MLG 8.10 method (36). Three grams of fresh soil or feces were added to 9 ml of buffered peptone water (Acumedia, Lansing, MI, USA) in a filtered stomacher bag and were vigorously shaken for 30 s. As a pre-enrichment step, the stomached homogenates remained in the filtered stomacher bag and were incubated overnight at 35°C. This pre-enrichment step was followed by two enrichments in University of Vermont Modified Listeria Enrichment Broth (UVM; Remel, Lenexa, KS, USA) and Fraser Broth (Oxoid CM0895, Basingstoke, UK), both requiring overnight incubation for 24 h at 30°C. One loopful of the Fraser’s enrichment culture was streaked on Listeria selective agar (LSA, Oxoid CM0856, Basingstoke, UK) for the isolation of Listeria colonies. These plates were incubated overnight at 30°C, and on each plate three Listeria-like colonies per positive samples were picked and kept for further identification tests. Stock cultures were prepared by growing Listeria strains in tripticase soy broth (TSB; Acumedia, Lansing, MI, USA) at 37°C. After washing in sterile water, the cell pellet was suspended in a brain heart infusion (BHI) broth (Acumedia, Lansing, MI, USA) with 25% of glycerol, aliquoted (300 µl in microtubes) and frozen at −80°C until further utilization. | other | 35.16 |
The species of presumptive Listeria colonies recovered on LSA were determined by multiplex PCR (37). In short, speciation occurred using two multiplex PCR reactions, based on the size of PCR amplicons. Pool 1 contained the primers for the identification of L. ivanovii, L. grayi, and L. innocua, and pool two contained primers for the identification of L. welshimeri, L. monocytogenes, and L. seeligeri. A 25 µl PCR reaction was composed of 1× EconoTaq PLUS 2× Master Mix (Lucigen Corporation, Middleton, WI, USA), 1 µM of each livN, Igr, and lin2 reverse and forward primers (for Pool 1) or 1 µM of each lwe, Lmo, and lse reverse and forward primers (for Pool 2) and quantity sufficient (qs) of water. For negative controls, sterile water was added instead of template DNA. The cycling program consisted of 1 cycle at 95°C for 9 min; 30 cycles at 94°C for 30 s, at 60°C for 30 s, and at 72°C for 1 min; and 1 cycle at 72°C for 7 min. The serovar group of isolates classified as L. monocytogenes was determined by multiplex PCR using five sets of primers (38). Briefly, one colony of L. monocytogenes isolates was thoroughly mixed in a 25 µl PCR reaction containing: 1× EconoTaq PLUS 2× Master Mix (Lucigen Corporation, Middleton, WI, USA), 1 µM of each Lmo0737, ORF2819, and ORF2110 reverse and forward primers, 1.5 µM of Lmo1118 reverse and forward primers, 0.2 µM of prs reverse and forward primers and qs water. For negative controls, sterile water was added instead of template DNA. PCR was performed with an initial denaturation step at 94°C for 3 min; 35 cycles of 94°C for 0.40 min, 53°C for 1.15 min and 72°C for 1.15 min; and 1 final cycle for 72°C for 7 min. PCR reactions were performed in an Eppendorf Mastercycler EP Gradient S (Eppendorf). After the completion of all cycles, 18 µl of PCR product was mixed with 3 µl of BlueJuice™ loading buffer (Invitrogen, Carlsbad, CA, USA) and separated on a 2% E-gel® with SYBR-safe™ (Invitrogen) along with 12 µl of E-Gel™ 1 kb Plus DNA Ladder (Invitrogen). | other | 37.34 |
Based on the two main Listeria species found from farm distribution data, three L. monocytogenes strains representing the different recovered serovar groups (1/2a-3a, 1/2b-3b-7, and 4b-4d-4c) and one L. innocua were selected for the growth experiments. Pre-cultures were prepared by inoculating 60 ml of TSB with 100 µl of the thawed stock culture and incubated for 24 h at 30°C while shaking (150 rpm). After 24 h, cell density was estimated spectrophotometrically by measuring the optical density (OD) at 600 nm (OD600nm) with the Thermo Scientific Spectronic 200™ (Fisher Scientific). Pre-cultures were initially diluted in TSB or UVM to a concentration of 106 CFU/ml, and then serially diluted in TSB or UVM to obtain final inoculum concentrations of 105 and 102 CFU/ml. | other | 38.06 |
A volume of 0.4 ml of each culture (105 and 102 CFU/ml) for the four Listeria strains was aliquoted into wells of a microplate (Honeycomb 2 cuvette plate; Labsystems, Inc., Franklin, MA, USA), with five repeats of each culture condition (strain × medium × concentration) per plate. Negative controls consisted on 0.4 ml of uninoculated TSB and UVM (five repeats) incubated along the cultures. For each culture, two independent plate repeats containing all treatment combinations were performed. The inoculated microplate was placed in a Bioscreen C microbiology reader (Thermo Electron Corp., West Palm Beach, FL, USA), which was operated by a computer with Growth Curves Software, v 2.28 (Transgalactic Ltd., Helsinki, Finland). The microbiology reader recorded the OD values of cultures at 20-min intervals after a plate shaking of 10 s at a medium speed (30 shakes/min). Three incubation temperatures were chosen, and the corresponding incubation times were adjusted to make sure that all growth curves reach the stationary phase by the end of the experiment. Plates were incubated at (i) 20°C, estimated soil temperature calculated upon the average of the atmospheric temperatures encountered during the sampling period (from March to August), for 48 h, (ii) 30°C, recommended temperature used for the enrichment procedure of Listeria spp. in UVM medium, for 24 h, and (iii) 42°C, expected temperature inside the chicken intestine, for 24 h. | other | 39.03 |
The raw data for each growth curve were graphed along with the resulting fit, and the R2 value (coefficient of determination) for each resulting fit was calculated. A four-way analysis of variance (ANOVA) was performed separately on each growth parameter (λ, μmax, and A), followed by Tukey’s post hoc test in R software v3.2.1. Factors included in the model were the Listeria strains, the culture medium, the inoculum concentration, and the incubation temperature. For the coculture experiments, μmax and stationary phase cell densities (equivalent to ODmax) were log10-transformed before ANOVA, and a Tukey’s post hoc test was used to group treatments. For all analyses, differences among groups were considered significant if p ≤ 0.05. | other | 37.97 |
A total of 1,110 samples (555 feces samples and 555 soil samples) were collected from 37 flocks on 10 pastured poultry farms over a 3-year period, and the distribution of Listeria species varied according to the sampling year (Figure 1A), broiler farm (Figure 1B), and sample type (Figure 1C). Overall, Listeria species were detected on all the farms and isolated in 15% of samples (83 from feces and 85 from soils), which is in the range of Listeria species prevalences reported in poultry-related environmental samples (from 1.4 to 53%) such as broiler litter, farm feed, farm drinking water, soil, and grass (25, 26, 39, 40) as well as in poultry feces (4.7–17%) (23, 25). In our study, three species were isolated including L. innocua (65.7%), L. monocytogenes (17.4%), and L. welshimeri (15.1%), and each of these species were recovered from at least half of the broiler farms (80, 50, and 90%, respectively; Figure 1B). Although different Listeria species distributions were observed between farms, at least two Listeria species were recovered from all but one farm (Farm M), and all three species were recovered from soil samples in all 3 years of the study (Figure 1C). Listeria innocua has been previously shown to be the predominant species isolated from the broiler farm environment (23–26, 40), while the detection of other non-pathogenic Listeria species, such as L. welshimeri, remains infrequent (23, 40), mostly because studies only focus on L. monocytogenes (41–43). | other | 30.72 |
Since no significant growth differences were observed among the three L. monocytogenes isolates in the monoculture growth study, subsequent coculture growth studies with the L. innocua isolate were only performed with the L. monocytogenes 1/2a-3a isolate (the most prevalent serovar group found within the farm data). Three different coculture mixtures were used to observe coculture growth effects: (1) L. monocytogenes 1/2a-3a to L. innocua ratio of 102:102 CFU/ml in UVM, (2) L. monocytogenes 1/2a-3a to L. innocua ratio of 105:105 CFU/ml in UVM, and (3) L. monocytogenes 1/2a-3a to L. innocua ratio of 102:102 CFU/ml in TSB. As positive controls, monocultures of L. monocytogenes 1/2a-3a and L. innocua were tested under the same conditions (medium × concentration) as the coculture mixtures, while negative controls consisted of uninoculated TSB or UVM, respectively. Each coculture mixture was inoculated individually into a microplate along with positive and negative controls with a final inoculum volume of 0.4 ml for each condition. The OD was recorded at 30-min intervals after a brief plate shaking of 10 s at a medium speed (30 shakes/min) during the incubation at 30°C for 24 h. To quantify the growth of L. monocytogenes 1/2a-3a and L. innocua in coculture, 100 µl aliquots were sampled every hour from the microwell plate and serially 10-fold diluted. Appropriate dilutions were plated on Rapid’L.mono medium (Bio-Rad, Hercules, CA, USA). Blue and white colonies were enumerated as L. monocytogenes 1/2a-3a and L. innocua, respectively. | other | 41.3 |
Growth curves were plotted based on OD values over time. Each bacterial growth curve was fitted to a modified Gompertz model using Matlab 2007b. The model equation is as follows: y=Ae{−e[μmax⋅eA(λ−t)+1]}, where y is the OD value measured, t is the time (h), μmax is the maximum specific growth rate (h−1), A is the maximum OD value attained, and λ is the lag time (h). Within the m-file written in Matlab, the lsqcurvefit function (a nonlinear least-squares solver for data fitting) was utilized to fit the growth curves by first using the following Gompertz equation: y=Ae−e(B−Cx). | other | 40.06 |
Relative abundance (% total Listeria species isolated) and distribution of the isolated Listeria innocua, Listeria monocytogenes, and Listeria welshimeri (A) according to the sampling year, (B) the broiler farm, and (C) the sample type over the 3-year sampling period. The number of Listeria species isolated per year/farm/sample type is indicated to the right of the bar. | other | 30.81 |
Listeria monocytogenes was isolated from 5.8, 0.3, and 1.0% of all samples collected in 2014, 2015, and 2016, respectively, with 87% (26/30) recovered during 2014 (Figure 1A) and 57% (17/30) of all L. monocytogenes isolates coming from the only flock sampled on Farm D in 2014 (Figure 1B). Overall, three L. monocytogenes serovar groups were identified: 1/2a-3a (70%), 1/2b-3b-7 (20%), and 4b-4d-4e (10%). Interestingly, over the 3-year sampling period, only one L. monocytogenes-positive flock (Farm I, 2014) harbored more than one serotype, demonstrating the potential clonal nature of L. monocytogenes within a flock or on a farm (44). The overall prevalence of L. monocytogenes on these 10 farms was low compared with other grow-out farm environments where 0–46.2% of the environmental and feces samples were L. monocytogenes positive (45), but the distribution of the serotypes was consistent with other studies that have characterized L. monocytogenes serotypes in broiler flocks (26, 41, 43). The prevalence of L. monocytogenes contamination may be dependent on the type of production system. A significant difference between caged- and floor-reared hens was observed with a greater detection of L. monocytogenes in dust samples from floor-reared hens in L. monocytogenes-positive flocks (41). In alternative systems, broilers are raised in less controlled environments than conventional systems and are more likely to be in contact with L. monocytogenes known to be widely spread in soil and vegetation (35). | other | 30.42 |
Poultry farms frequently have other animals (beef cattle, sheep, goats, or swine) and pets present on the production site (35). These animals can be reservoirs for and play a role in the proliferation and deposition of L. monocytogenes into the environment. In our study, all but two farms had other animals raised in close proximity to the broiler flocks during the sampling period, but we did not investigate the possible genotype matching between animal species. Generally, the presence of other animals on the farm increase the risk factor associated with pathogenic bacteria contamination of poultry flocks (42, 46). This has been shown with Campylobacter spp. where adjacent broiler flocks and cattle appear to be the most frequently identified animals with broiler-flock matching Campylobacter spp. isolates (47). Another study has reported an increased risk of L. monocytogenes contamination in laying hen flocks when pets were present on the production site (42). | other | 32.12 |
Our field results showed a higher prevalence of L. innocua compared with L. monocytogenes on pastured poultry grow-out farms, which has been supported by other studies reporting the incidence and characterization of Listeria species in the commercial poultry farm environment (15, 23, 24, 26). In terms of food safety interests, there is a question as to whether there is any physiological basis for the dominance of L. innocua over L. monocytogenes within the poultry farm environment, and whether this dominance related to preferential growth. To determine if this environmental dominance of L. innocua over L. monocytogenes may be linked to growth conditions (e.g., initial concentration, growth temperature, and growth medium), monoculture and coculture growth studies were performed. Three L. monocytogenes isolates (one strain of each serovar groups: 1/2a-3a, 1/2b-3b-7, and 4b-4d-4e) and one L. innocua isolate were selected to compare their growth capacity in liquid media. Bacterial growth was monitored by recording the OD of a culture in growth media (TSB and UVM) inoculated at different initial concentrations (102 and 105 CFU/ml) and incubated at 20°C (average environmental temperature), 30°C (UVM enrichment temperature according to USDA-FSIS MLG 8.10), or 42°C (broiler body temperature). Curve modeling was performed with the Gompertz function that fits the data with R2 values ranging from 0.674 to 0.998, indicating a good fit. From the modified Gompertz equation, three relevant parameters [lag time (λ), maximum specific growth rate (μmax), and maximum OD (ODmax)] were determined for each curve and subsequently used to statistically compare the bacterial growth of the Listeria strains under the different cultural conditions. Using a four-way ANOVA (Tables S1–S3 in Supplementary Material for λ, μmax, and ODmax, respectively), we investigated whether the culture medium, the inoculum concentration, and the incubation temperature could explain the global variation observed between the growth curves. In the same model, we more specifically examine the growth differences between the four Listeria isolates for a single culture condition. The parameters λ, μmax, and ODmax representing bacterial growth characteristics were used in the model. | other | 36.53 |
Unsurprisingly, λ was significantly shorter for the higher inoculum concentrations for all strains, enrichment media, and incubation temperatures (F = 801, p < 0.0001; Figure 2). This is in agreement with other studies that have evidenced the importance of inoculum concentration on the ability of a microbial population to initiate growth (48, 49). While Baranyi et al. showed that as the cell numbers in the inoculum decrease, λ increases (50, 51), other studies have reported an effect of the inoculum size only under stressful conditions (49, 52). Increasing incubation temperature significantly decreased λ (F = 174, p < 0.0001) in both TSB and UVM enrichment media for all Listeria strains, as has been showed in other growth media for both L. monocytogenes and L. innocua (28, 49, 52, 53). The temperature-dependent effect was significantly greater in the low initial concentration treatments compared with the higher initial inoculum treatments (F = 77, p < 0.0001). While lag time was significantly shorter in TSB compared with UVM, this was the weakest association of the major growth variables tested (F = 60, p < 0.05). | other | 30.66 |
Average of lag time (λ) length of Listeria monocytogenes and Listeria innocua strains inoculated at low (102 CFU/ml) and high (105 CFU/ml) concentrations in TSB and UVM incubated at (A) 20°C, (B) 30°C, and (C) 42°C. *Significant differences between strains (p < 0.05). No growth observed for L. innocua in TSBLow and UVMLow at 42°C. | clinical case | 29.73 |
While many of the growth variables tested significantly effected μmax, by far the strongest association was to the enrichment medium (F = 2431, p < 0.0001), where the four Listeria strains grew faster in TSB than UVM (Figure 3). This result is in agreement with the general trend of Listeria strains from food origin showing a faster growth in general growth media (e.g., BHI, TSB + yeast extract) than Listeria enrichment media (UVM, Fraser, and Half-Fraser) (30, 33). We also observed that Listeria strains grew significantly faster at lower incubation temperatures, peaking at 30°C, and this effect was amplified in TSB medium (F = 81, p < 0.0001). This is in agreement with Duh and Schaffner (28), who showed that Listeria strains grew faster at temperatures below 41°C in general growth media (28). The growth variable with the weakest significant association to μmax was initial inoculum concentration, where its effect were only observed in the 42°C treatments (F = 25, p > 0.0001). | other | 31.73 |
Average of maximum specific growth rate (μmax) of Listeria monocytogenes and Listeria innocua strains inoculated at low (102 CFU/ml) and high (105 CFU/ml) concentrations in TSB and UVM incubated at (A) 20°C, (B) 30°C, and (C) 42°C. *Significant differences between strains (p < 0.05). No growth observed for L. innocua in TSBLow and UVMLow at 42°C. | clinical case | 30.11 |
Growth curves of cocultures of Listeria monocytogenes serovar groups 1/2a-3a and Listeria innocua at 30°C inoculated at (A) high initial concentrations (105 CFU/ml) in UVM, (B) low initial concentrations in UVM (102 CFU/ml), and (C) low initial concentrations (102 CFU/ml) in TSB. | other | 34.38 |
Growth parameters lag time (λ), maximum specific growth rate (μmax) and stationary phase cell density for triplicate cocultures studies of Listeria monocytogenes and Listeria innocua grown at 30°C and inoculated at two inoculum ratios (Low:Low and High:High) in two different growth media (UVM and TSB). | other | 32.3 |
As was observed for μmax, the enrichment medium was the dominant growth variable affecting ODmax (F = 1481, p < 0.0001) with significantly higher maximum optical densities found in the treatments grown in TSB (Figure 4). This is consistent with other data reporting a higher final cell density in the non-selective culture medium BHI than in selective enrichment media UVM or Half-Fraser (34, 54). For all Listeria strains, the ODmax significantly increased with decreasing incubation temperatures, especially in the UVM treatments (F = 193, p < 0.0001). Unlike λ or μmax, initial inoculum concentrations did not have a significant effect of ODmax overall (F = 0.01, p > 0.05), although limited effects were observed at treatments incubated at 42°C. | other | 30.39 |
Average of maximum optical density (ODmax) of Listeria monocytogenes and Listeria innocua strains inoculated at low (102 CFU/ml) and high (105 CFU/ml) concentrations in TSB and UVM incubated at (A) 20°C, (B) 30°C, and (C) 42°C. *Significant differences between strains (p < 0.05). No growth observed for L. innocua in TSBLow and UVMLow at 42°C. | clinical case | 30.2 |
While the general effects of the above variables on growth of Listeria spp. overall, the question of the differential effect between specific Listeria species still remained. While there were some exceptions, generally there were no significant differences between the three L. monocytogenes strains in terms of λ, μmax, or ODmax, and regardless of enrichment media, the L. innocua strain was unable to grow at broiler body temperature (42°C) when the initial inoculum concentrations was 102 CFU/ml (TSBLow and UVMLow). Significant differences in lag time between the three L. monocytogenes strains and the L. innocua strain varied based on the growth variables (Figure 2). At the average environmental and UVM enrichment temperatures (20 and 30°C, respectively), the lag time of L. innocua was significantly shorter than the L. monocytogenes strains in UVM, especially for low inoculum concentrations (p < 0.05; Figures 2A,B, respectively). However, at broiler body temperatures (42°C), L. innocua only grew in the high initial inoculum treatments (TSBHigh and UVMHigh), where there were no significant differences between the four Listeria strains (Figure 2C). No significant differences in λ between L. innocua and L. monocytogenes were found in any of the TSB treatments. Previous studies comparing the growth of L. monocytogenes and L. innocua strains mostly from food origins have reported shorter λ for L. innocua in Fraser (incubated at 30°C), and Half-Fraser (incubated at 37°C) enrichment media (31) and at lower incubation temperatures (≤8°C) (28). | other | 29.73 |
While significant differences in μmax were observed between the four Listeria strains used in this study, there were no consistent trends based on initial inoculum concentration, incubation temperature or enrichment medium (Figure 3). Only in two treatment combinations were there species-specific significant differences in μmax, with L. monocytogenes growing faster than L. innocua in TSBLow at 30°C (Figure 3B) and L. innocua growing faster than L. monocytogenes at 42°C in the UVMHigh treatment (Figure 3C). In contrast to previously reported findings, there were no significant differences found in μmax between the L. monocytogenes and L. innocua isolates in UVM at 30°C, conditions used for the Listeria enrichment process (28, 31, 33). However, the studies comparing the generation time or the growth rate have shown a faster growth of L. innocua compared with L. monocytogenes at temperatures below 40°C only in certain culture media (28, 31, 33), which may explain the similar μmax between L. innocua and L. monocytogenes in our study. In addition, a high level of heterogeneity in growth behavior within L. innocua and L. monocytogenes strains can lead to equivalent μmax between the slowest L. innocua and the fastest L. monocytogenes (30, 55). | other | 29.52 |
There were very few strain-specific differences in the maximum OD (ODmax) for any of the growth variables, with the significant differences found at 42°C (Figure 4C). Among those difference, only under one treatment condition (TSBHigh) were there significant differences between L. monocytogenes and L. innocua, so overall the maximum cell density in culture was unaffected by the Listeria species. No differences were observed between the ODmax of L. innocua and L. monocytogenes species in UVM at 30°C as reported in studies using Half-Fraser and Fraser (30, 31). However, these results are highly dependent on the experiment and opposite trends are also reported showing an higher final population density of L. innocua than L. monocytogenes in enrichment media (29, 54). | other | 30.86 |
Using the cultural conditions for the initial enrichment step for the USDA-FSIS MLG 8.10 L. monocytogenes enrichment method (UVM, 30°C) we found that L. innocua exhibited a significantly shorter lag time than the L. monocytogenes strains in monocultures, especially for low initial inoculum concentrations (102 CFU/ml). To determine if L. innocua has any direct competitive growth advantages over L. monocytogenes in UVM, coculture experiments were performed using the L. innocua strain and the L. monocytogenes 1/2a-3a strain (the most prevalent serovar group from the farm surveys). When both strains were inoculated into the coculture at 105 CFU/ml (Figure 5A), there were no significant differences in λ, μmax, or stationary phase cell density (similar to ODmax), although L. monocytogenes densities did begin to exceed L. innocua cell densities after 24 h. When both strains started at the lower inoculum level (102 CFU/ml), while λ and μmax were similar, L. innocua reached a significantly higher stationary phase cell density compared with L. monocytogenes (F = 31, p < 0.01; Figure 5B). Conversely, when the cocultures inoculated at the lower initial concentrations were grown in TSB, L. monocytogenes demonstrated significantly higher stationary phase cell densities compared with L. innocua (F = 19, p < 0.05), with L. monocytogenes cell densities being ~3× greater than L. innocua (Figure 5C). When comparing the growth curve parameters among the three coculture experiments, only the stationary phase cell density was significantly effected at the species-level (Table 2). | other | 31.77 |
Unlike the monoculture results using the UVM protocol from the USDA-FSIS MLG 8.10 method, no significant lag time difference was observed between L. innocua and L. monocytogenes at low initial inoculum concentrations; however, L. innocua still maintained a competitive growth advantage under these cultural conditions represented by significantly higher stationary phase cell densities (Table 2; Figure 5B). When grown under the same conditions in TSB (Figure 5C), L. monocytogenes grew at significantly higher levels than L. innocua, indicating that there is a enrichment media-specific effect on Listeria growth within these cocultures. When looking at the differences in stationary phase cell densities across all cocultures, there was no significant difference for L. innocua between UVM and TSB in the Low:Low cocultures, but there was over a 1 log reduction in stationary phase cell density for L. monocytogenes grown in TSB (7.17 ± 0.06 log10 CFU/ml) compared with UVM (6.38 ± 0.05 log10 CFU/ml) under those growth conditions (Table 2). While it appears that L. innocua has a competitive growth advantage in UVM with low initial inoculum concentrations, it is possible that this advantage comes more from a disadvantage of L. monocytogenes growing under these conditions, rather than a specific advantage that L. innocua possesses, and previous studies have shown that enrichment/culture media can differentially effect L. innocua and L. monocytogenes (28, 31, 33). | other | 30.2 |
However, considering the UVM enrichment is the first of two enrichments in the USDA-FSIS MLG 8.10 protocol, having significantly higher densities of L. innocua compared with L. monocytogenes would artificially increase the likelihood of isolating L. innocua from samples with equivalent levels of L. innocua and L. monocytogenes. Considering the UVM enrichment is used within the USDA-FSIS MLG 8.10 method, and USDA-FSIS is responsible for the testing of foodborne pathogens from broiler production farms and processing plants, this enrichment bias could potentially or partially explain the prevalence of L. innocua as the dominant Listeria spp. found on poultry farms (15, 23–26). | other | 31.38 |
In our study, we found that L. innocua is more prevalent than the foodborne pathogen L. monocytogenes in soil and feces samples collected from pastured poultry farms, which is consistent with conventional poultry farms. Mono- and coculture growth experiments showed that under cultural conditions used in the first enrichment step of the USDA-FSIS MLG 8.10 L. monocytogenes method (UVM, 30°C), L. innocua had a significantly shorter lag phase (monoculture) and a significantly higher stationary phase cell density (coculture) compared with L. monocytogenes; these growth advantages occurred at low initial inoculum concentrations simulating the low levels of Listeria species encountered in the environment. Based on these results, it is possible that UVM enrichment medium either preferentially supports L. innocua growth over L. monocytogenes, or preferentially restricts L. monocytogenes growth, and that this enrichment step may be biasing the recovery of L. innocua over L. monocytogenes from live production samples. Considering the public health importance of accurately identifying the source of L. monocytogenes outbreaks, future work will need to understand the cultural and molecular mechanisms of this preferential L. innocua growth in UVM, and alternative enrichment methods for L. monocytogenes may need to be considered. | other | 30.12 |
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer KG declared a shared affiliation, with no collaboration, with the authors to the handling editor. | clinical case | 35.22 |
Numerous mutations in oncogenes and tumor suppressors, as well as the alterations in gene expression profiles, result in the deregulations of cell metabolism in cancer cells. The Warburg effect, also named aerobic glycolysis, which is observed by Otto Heinrich Warburg in 1920s, is one of most prominent hallmarks of cancer cells . In Warburg effect, even in the presence of sufficient oxygen, glucose is converted to lactate, instead of totally oxidized via Krebs cycle to generate ATP . Tumor glycolysis supplies abundant energy to sustain the rapid tumor growth, moreover, the product of glycolysis such as lactate also provides an appropriate microenvironment to promote tumor metastasis . So far, various regulatory pathways involved in the conversion from the Krebs cycle to tumor glycolysis have been well investigated, HKs were considered to be one of the most important effectors [4, 5]. | other | 31.47 |
As important glycolytic enzymes, HKs are responsible for the first rate-limiting step in the process of glucose metabolism, in which glucose is phosphrylated to glucose-6-phosphate. Four different isoforms of HKs, named as HK1-4 have been identified so far. HK-1 and HK-2 are mainly located on the outer membrane of mitochondria, HK-3 is positioned in a perinuclear compartment, and HK-4 is in the cytoplasm. Localization to the outer membrane of mitochondria confers HK-2 the advantage to escape product inhibition and gain preferential access to ATP in mitochondrion . Among these different HKs, HK-2 is found to be expressed of high rate in malignant tumors and plays a key role in the development of Warburg phenotype. Overexpression of HK-2 was observed in various cancers, such as gastric , ovarian , breast cancer , cervical carcinoma , esophageal adenocarcinoma and nasopharyngeal carcinoma . | other | 31.67 |
Chrysin is a bioactive flavone derived from plant extracts such as blue passion flower, propolis, and honey, which are widely used as herb medicine in China. Except its multiple bioactivities in antioxidant , anti-inflammatory and antibacterial , its antitumor potential is also well validated in a variety of human cancer cell lines. Chrysin is demonstrated to exert antitumor effect by inducing cell cycle arrest and apoptosis through different mechanisms, for instance, activation of extrinsic apoptosis pathway , alteration of cyclins and CDKs . Moreover, multiple signaling pathways in cancer cells, such as Ras-Raf-MAPKs, PI3K-Akt, STAT, NF-κB, Wnt-β-catenin and Notch signaling pathways, were proved to be modulated by chrysin to inhibit cell proliferation, angiogenesis, invasion and metastasis [18–20]. However, the effect of chrysin on tumor glycolysis is largely unknown. Herein, we showed that chrysin inhibited glycolysis and induced apoptosis in HCC cells in vitro and in vivo. Mechanism investigations revealed that the biological activities exerted by chrysin were mainly attributed to its effect on HK-2. With the decrease of HK-2, chrysin inhibited tumor glycolysis and activated mitochondria-associated apoptosis. Given HK-2 was found to be overexpressed in the majority of HCC tissue, the results of present study suggested chrysin, or its analogues, may serve as effective candidates for HCC management. | other | 29.69 |
Chrysin and other chemical reagents such as Tris, NaCl, SDS and DMSO were purchased from Sigma-Aldrich (St. Louis, MO). The normal human hepatic cell LO2 and HepG2, Hep3B, Huh-7, HCC-LM3, Bel-7402 and SMMC-7721 were obtained from the Cell Bank of Chinese Academy of Sciences and cultured with Dulbecco’s Modified Eagle Medium containing 10% FBS and 1% antibiotics in a 37 °C incubator with 5% CO2. Anti-HK2, anti-VDAC1, anti-cleaved-caspase3, anti-cleaved-PARP, anti-cytochrome C, anti-Bax, anti-Bak, anti-Bcl-2, anti-Bcl-xL, anti-rabbit IgG-HRP and anti-mouse IgG-HRP antibodies were products of Cell Signaling Technology, Inc. (Danvers, MA). Anti-α-Tubulin antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Anti-β-actin (A5316) was product of Sigma (St. Louis, MO, USA). HK2 (ORF004940) construct was purchased from Applied Biological Materials (ABM) Inc. (Richmond, BC, Canada). Lipofectamin 2000 was purchased from Invitrogen (Carlsbad, CA). | other | 32.62 |
Human HCC cells (3× 103 per well) were seed in 96-well plate and then treated with or without different concentrations of chrysin for 0, 24, 48 and 72 h. Cell viability was measured with Cell Titer-Glo Luminescent Cell Viability Assay kit from Promega Corp. (Madison, WI) according to the manufacturer’s protocol. | other | 41.1 |
Cells were harvested by trypsinization and pelleted by centrifugation. The pellets were lysed with RIPA buffer supplemented with protease cocktail (Roche, Germany). Protein concentrations in cell lysates were determined with the Bradford assay (Bio-Rad, Philadelphia, PA, USA). The lysates were subjected to SDS-PAGE and then electrically transferred to PVDF membrane (Millipore, Billerica, MA, USA). After blocked with 5% non-fat dry milk, the membranes were incubated with specific primary antibodies overnight at 4 °C. After washed with TBS-Tween 20 three times, HRP-conjugated secondary antibodies were incubated at room temperature for 1 h. The membranes were washed with TBS-Tween 20 and the bands were visualized using ECL chemiluminescence reagents (Pierce Chemical Co., Rockford, lllinois, USA). | other | 44.5 |
Cells were seeded in six-well plate and then treated with chrysin for 24 h. After treatment, attached and floating cells were harvested by centrifugation. For apoptosis analysis, the cells were re-suspended with PBS and adjusted to 1 × 106 cells/ml, then 5 μl Annexin V and Propidium Iodide staining solution were added to 300 μl of the cell suspension. After incubated 10–15 min at room temperature in the dark, stained cells were assayed and quantified using a FACSort Flow Cytometer (BD, San Jose, CA, USA). | other | 42.97 |
Glucose uptake and lactate production measurement were performed as previously described . Briefly, Cells were trypsinized and seeded in 6-well plates (5 × 105), after incubation for 10 h, cell culture medium was discarded and replaced with fresh culture medium containing different concentrations of chrysin for 8 h. Glucose and lactate levels were measured by using the Automatic Biochemical Analyzer (7170A, HITACHI, Tokyo, Japan). The relative glucose consumption rate and lactate production rate were normalized by the protein concentration of samples. | other | 45.06 |
The cells (5 × 106) from a 10 cm dish were harvested by trypsinization and centrifuged at 800 rpm for 5 min at 4 °C. The cell pellets were washed once with cold PBS and then resuspended with isolation buffer (20 mM Hepes, pH 7.4, 10 mM KCl, 1.5 mM MgCl2, 1 mM sodium EDTA, 1 mM dithiothreitol, 10 mM phenylmethylsulfonyl fluoride, 10 mM leupeptin and 10 mM aprotinin). After chilling on ice for 3 min, the cells were disrupted by 60 strokes of a glass homogenizer. The homogenate was centrifuged at 2,000 rpm for 10 min at 4 °C to remove unbroken cells and nuclei. The mitochondria-enriched fraction (supernatant) was then pelleted by centrifugation at 13,000 rpm for 30 min. The pellets was lysed in RIPA buffer (10 mM Tris-Cl (pH 8.0), 1 mM EDTA, 0.5 mM EGTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS. 140 mM NaCl) and analyzed by western blot. For Immunoprecipitation, 500 μg extractions were pre-cleared with 30 μl (50% slurry) agarose A/G beads for 2 h rocking at 4 °C. The beads were removed, 30 μl (50% slurry) fresh agarose A/G beads and appropriate antibodies (4 μg) were added to the precleared lysate overnight at 4 °C. The beads were washed, mixed with 4 × SDS sample buffer, boiled, and then subjected to Western blotting. | other | 41.38 |
Six-week-old female Nu/nu athymic nude mice were maintained under specific pathogen free (SPF) condition in accordance with Institutional Animal Care and Use Committee. HCC-LM3 cells (5 × 106 cell/mice) were subcutaneously injected into the right flank of nude mice. When the tumor volume reached about 100 mm3, the mice were randomly grouped, five mice per group. The vehicle group was given 0.5% sodium carboxymethylcellulose, and the other group received 30 mg/kg chrysin. Chrysin was administrated three times weekly via intraperitoneal injection. The tumors were measured twice per week using microcalipers and tumor volume(V) was calculated as following: V = (length × width2)/2. After the experiment, mice were sacrificed and the tumors are weighed and photographed. | other | 45.4 |
A human HCC tissue array (HlivH150CS03), which contains 75 cases of malignant HCC biopsy and 75 cases of matched adjacent normal tissue, was purchased from Shanghai Outdo Biotech Co., Itd. (Shanghai, China). Tumor tissues from xenograft model were fixed in 4% paraformaldehyde and then embedded in paraffin. IHC staining was performed as previously described . Briefly, Paraffin sections were deparaffinized and hydrated, the endogenous peroxidase activity was blocked with 3% H2O2. Antigen retrieval was performed in citric acid solution using a microwave oven. The tissues were blocked with serum from the host of the secondary antibody and incubated with primary antibody of anti-HK2 (1:200) or anti-Ki67 (1:250) respectively at 4 °C overnight. Biotinylated secondary antibodies were added at a 1:100 dilution and followed by Vectastin ABC solution. Finally, the binding of the antibodies was visualized with 3, 3-diaminobenzidine (DAB) solution. Tissues were counterstained with harris’ hematoxylin and then dehydrated. Slides were viewed and photographed under a light microscope and analyzed using Image-Pro Plus software (version 6.2) program (Media Cybernetics). | other | 44.03 |
All statistical analysis was performed by SPSS software (version 13.0). All the quantitative data were expressed as mean values ± standard deviation, the significant difference between two groups was assessed by a two-tailed Student’s t test. p < 0.05 represented a statistically significant difference. | other | 34.3 |
First, the expression of HK-2 was assessed by western blotting in six HCC cell lines. As the results shown in Fig. 1a, comparing to the normal hepatic cell LO2 in which no HK-2 expression was detected, at least five HCC cell lines (HepG2, Hep3B, HCC-LM3, SMMC-7721, BEL-7402) expressed high levels of HK-2. We next sought to examine HK-2 expression in HCC tissue and paired adjacent normal tissue. Consistent with the result in HCC cell lines, in matched normal adjacent samples, HK-2 was not detectable or at a relatively low level. In contrast, HK-2 was substantially overexpressed in tumor samples (n = 75, p < 0.001) (Fig. 1b). These results suggested that HK-2 might have a role in HCC development.Fig. 1Aberrant expression of HK-2 in hepatocellular carcinoma(HCC). a, HK-2 was highly expressed in HCC cell lines. Western blotting was performed to examine HK-2 expression in several HCC cell lines and normal hepatic cell LO2. b, HK-2 was highly expressed in HCC tissue. Representative figures of immunohistochemical staining for HK-2 in HCC tissues and paired adjacent normal tissue (left panel), statistical results of HK-2 staining in 75 different HCC tissues and matched adjacent normal tissue (right panel). The asterisk (***, p < 0.001) indicated a significant difference of HK-2 expression between tumor and paired adjacent normal tissue | other | 33.44 |
Aberrant expression of HK-2 in hepatocellular carcinoma(HCC). a, HK-2 was highly expressed in HCC cell lines. Western blotting was performed to examine HK-2 expression in several HCC cell lines and normal hepatic cell LO2. b, HK-2 was highly expressed in HCC tissue. Representative figures of immunohistochemical staining for HK-2 in HCC tissues and paired adjacent normal tissue (left panel), statistical results of HK-2 staining in 75 different HCC tissues and matched adjacent normal tissue (right panel). The asterisk (***, p < 0.001) indicated a significant difference of HK-2 expression between tumor and paired adjacent normal tissue | other | 31.69 |
Since HK-2 was found to be of high expression in the majority of tested HCC cells, we examined the activity of chrysin in HCC cells with high HK-2 expression. As shown in Fig. 2b, after chrysin treatment, cell proliferation in LM-3, SMMC-7721 and Bel-7402 was substantially inhibited, and more than 50% cell growth inhibition was observed after 72 h treatment. Previous studies reported that tumor cells with high HK-2 expression often displayed high glycolytic phenomenon, we also investigated the effect of chrysin on tumor glycolysis. HCC cells exposed to chrysin (30 μM) showed significantly lower glucose consumption than the untreated. Along with the decrease of glucose uptake, the secretion of lactate, which is the product of tumor glycolysis, was also dramatically decreased (Fig. 2c-e). In accordance with the suppression of tumor glycolysis, in all tested HCC cells, the expression of HK-2 was markedly decreased in a dose-dependent manner (Fig. 2c-e). All these data demonstrated that chrysin displayed an inhibitory effect against cell proliferation and glycolysis in HCC cells via reducing HK-2 expression.Fig. 2Chrysin inhibited cell proliferation and glycolysis in HCC cells. a, The chemical structure of chrysin. b, HCC cells were treated with indicated concentration of chrysin for indicated times, cell proliferation was measured as described in Material and Methods. The asterisk (*, p < 0.05) indicated a significant decrease of HCC cell proliferation after chrysin treatment. c-e, chrysin suppressed glycolysis in HCC-LM3 (top), SMMC-7721 (middle) and Bel-7402 (bottom) cells. HCC cells were treated with various concentrations of chrysin for 8 h and the cell lysates were subjected to SDS-PAGE to examine the change of indicated protein (left panels). Glucose consumption (middle panels) and lactate production (right panels) in cell culture medium were analyzed. The graph showed the data of at least three independent experiments expressed as means ± SD, the asterisks (*, p < 0.05, **, p < 0.01, ***, p < 0.001, Student’s t test) indicated significant inhibition of glucose consumption and lactate secretion after chrysin treatment | other | 32.94 |
Chrysin inhibited cell proliferation and glycolysis in HCC cells. a, The chemical structure of chrysin. b, HCC cells were treated with indicated concentration of chrysin for indicated times, cell proliferation was measured as described in Material and Methods. The asterisk (*, p < 0.05) indicated a significant decrease of HCC cell proliferation after chrysin treatment. c-e, chrysin suppressed glycolysis in HCC-LM3 (top), SMMC-7721 (middle) and Bel-7402 (bottom) cells. HCC cells were treated with various concentrations of chrysin for 8 h and the cell lysates were subjected to SDS-PAGE to examine the change of indicated protein (left panels). Glucose consumption (middle panels) and lactate production (right panels) in cell culture medium were analyzed. The graph showed the data of at least three independent experiments expressed as means ± SD, the asterisks (*, p < 0.05, **, p < 0.01, ***, p < 0.001, Student’s t test) indicated significant inhibition of glucose consumption and lactate secretion after chrysin treatment | other | 31.42 |
Generally, HK-2 is located on the out membrane of mitochondria to exert it biological function. In order to further confirm the effect of chrysin on HK-2, we examined the change of HK-2 in mitochondria fractions. As expected, HK-2 in mitochondria was substantially decreased in a dose-dependent manner after exposure to chrysin (Fig. 3a). On the outer membrane of mitochondria, HK-2 interacts with VDAC-1 to form complex to prevent cancer cell apoptosis. Results of immunoprecipitation assay showed that, with the reduction of HK-2 in mitochondria, the HK-2 which bound to VDAC-1 was significantly decreased accordingly (Fig. 3b). Furthermore, as shown in Fig. 3c, cleaved caspase-3 and PARP, which are important markers indicating cell apoptosis, were dramatically elevated, suggesting that the decrease of HK-2 and disruption of HK-2/VDAC-1 interaction caused by chrysin resulted in HCC apoptosis.Fig. 3Chrysin induced HCC cell apoptosis by reducing HK-2 in mitochondria. a, HCC cells were treated with chrysin for 24 h, the mitochondria fractions was extracted and examined by western blotting to detect the change of indicated protein. b, HCC-LM3 cell was treated with chrysin for 24 h and the lysates of mitochondria fractions were immunoprecipitated with HK-2 or VDAC-1 antibodies, then the binding affinity was analyzed with western blotting analysis. c, HCC cells were treated with various chrysin for 24 h and cell lysates were probed with indicated antibodies | study | 29.73 |
Chrysin induced HCC cell apoptosis by reducing HK-2 in mitochondria. a, HCC cells were treated with chrysin for 24 h, the mitochondria fractions was extracted and examined by western blotting to detect the change of indicated protein. b, HCC-LM3 cell was treated with chrysin for 24 h and the lysates of mitochondria fractions were immunoprecipitated with HK-2 or VDAC-1 antibodies, then the binding affinity was analyzed with western blotting analysis. c, HCC cells were treated with various chrysin for 24 h and cell lysates were probed with indicated antibodies | study | 33.44 |
In order to further illustrate the role of HK-2 played in chrysin-mediated activities, HCC cells were transfected with pORF-HK-2 to overexpress HK-2 and then investigated chrysin’s activity. As shown in Fig. 4a, after transfection, reduction of HK-2 caused by chrysin was substantially recovered. With the increase of HK-2, chrysin-mediated suppression of tumor glycolysis was significantly impaired in HK-2-overexpression cells (Fig. 4b). Moreover, in contrast with the mock group, cleaved caspase-3 and PARP were significantly decreased in HK-2 overexpression group, suggesting cell apoptosis induced by chrysin was attenuated (Fig. 4a). Flow cytometry analysis also demonstrated that over 20% HCC cells were undergone apoptosis after treated with 60 μM chrysin, however, the ratio of cell induced apoptosis was significantly decreased in HK-2 overexpression cells. All these results verified chrysin-mediated suppression of glycolysis and induction of apoptosis in HCC cells was closely related to its effect on HK-2.Fig. 4Overexpression of HK-2 impaired the effect of chrysin on apoptosis and tumor glycolysis. a, activated caspase-3 and PARP in HCC cell with exogenous HK-2 expression. HCC cells were transfected with pORF-HK-2, 24 h later, the cells were treated with 60 μM chrysin, cell lysates were subjected to SDS-PAGE and probed with indicated protein. b and c, the effect of chrysin on glycolysis and apoptosis in HK-2 overexpressed HCC-LM3 cells. HCC-LM3 cell was transfected with pORF-HK-2 for 24 h and seeded in 6 well plates for 10 h, then exposed to 60 μM chrysin, tumor glycolysis (b) and apoptosis (c) was examined at 8 h and 24 h, respectively. b, The graph showed the data of at least three independent experiments expressed as means ± SD, the asterisks (*, p < 0.05, **, p < 0.01, ***, p < 0.001, Student’s t test) indicated significant difference between different groups. c, Representative FACS results of Annexin V-PI double staining were shown (left panels), and the graph (right panel) showed the data of at least three independent experiments expressed as means ± SD, the asterisks (***, p < 0.001, Student’s t test) indicated a significant difference | other | 31.22 |
Overexpression of HK-2 impaired the effect of chrysin on apoptosis and tumor glycolysis. a, activated caspase-3 and PARP in HCC cell with exogenous HK-2 expression. HCC cells were transfected with pORF-HK-2, 24 h later, the cells were treated with 60 μM chrysin, cell lysates were subjected to SDS-PAGE and probed with indicated protein. b and c, the effect of chrysin on glycolysis and apoptosis in HK-2 overexpressed HCC-LM3 cells. HCC-LM3 cell was transfected with pORF-HK-2 for 24 h and seeded in 6 well plates for 10 h, then exposed to 60 μM chrysin, tumor glycolysis (b) and apoptosis (c) was examined at 8 h and 24 h, respectively. b, The graph showed the data of at least three independent experiments expressed as means ± SD, the asterisks (*, p < 0.05, **, p < 0.01, ***, p < 0.001, Student’s t test) indicated significant difference between different groups. c, Representative FACS results of Annexin V-PI double staining were shown (left panels), and the graph (right panel) showed the data of at least three independent experiments expressed as means ± SD, the asterisks (***, p < 0.001, Student’s t test) indicated a significant difference | other | 31.98 |
Involvement of mitochondria in cell apoptosis is considered to be the common mechanism. In order to clarify the mechanism by which chrysin induced HCC cell apoptosis, we investigated the change of pro-apoptotic and anti-apoptotic protein in cytosolic and mitochondria fractions respectively. In HCC-LM3 and SMMC-7721 cells, cytochrome C was significantly increased in the cytosolic fractions after chrysin treatment (Fig. 5a and b). Conversely, Bax, which is a pro-apoptotic protein, was found to be translocated to mitochondria. With the decrease in cytosolic, its expression in mitochondria was dramatically increased. Other proteins involved in apoptosis regulation such as Bak, Bcl-2, Bcl-xL had no obvious change in chrysin treated cells (Fig. 5a-d). In HK-2 overexpression cells, the translocation of Bax from cytosolic to mitochondria was substantially attenuated, and the cytochrome C released from mitochondria into cytoplasm was also impaired, indicating that cell apoptosis induced by chrysin was attributed to the Bax activation on mitochondria.Fig. 5Chrysin induced Bax activation on mitochondrial. HCC cells were transfected with pORF-HK-2, 24 h later, the cells were treated with 60 μM chrysin for another 24 h, and subcellular fractions were prepared and subjected to SDS-PAGE and probed with indicated protein. a and b, cytosolic Bax, Bak and chrome c expressions in HCC-LM3 cell (a) and SMMC-7721 cell (b) were tested by western blotting. c and d, mitochondrial Bax, Bak, Bcl2, Bcl-xl, chrome c, and HK2 in HCC-LM3 cell (c) and SMMC-7721 cell (d) were tested by western blotting | study | 29.3 |
Chrysin induced Bax activation on mitochondrial. HCC cells were transfected with pORF-HK-2, 24 h later, the cells were treated with 60 μM chrysin for another 24 h, and subcellular fractions were prepared and subjected to SDS-PAGE and probed with indicated protein. a and b, cytosolic Bax, Bak and chrome c expressions in HCC-LM3 cell (a) and SMMC-7721 cell (b) were tested by western blotting. c and d, mitochondrial Bax, Bak, Bcl2, Bcl-xl, chrome c, and HK2 in HCC-LM3 cell (c) and SMMC-7721 cell (d) were tested by western blotting | study | 29.81 |
To validate the antitumor activity of chrysin against HCC, the efficacy of chrysin was assessed in HCC-LM3 xenograft model. As shown in Fig. 6a-d, comparing with the vehicle group, tumor growth in chrysin group was significantly suppressed. In the end of experiment, the tumor volume of vehicle group had reached about 500 mm3, while the average tumor volume of chrysin group was around 200 mm3. Meanwhile, no obvious toxicity was observed as evaluating the change of body weight. Immunohistochemistry analysis of chrysin-treated tumor tissue demonstrated the expression of HK-2 in tumor tissue was substantially decreased after chrysin treatment, which confirmed the effect of chrysin on HK-2 in vivo (Fig. 6e). Ki-67 is a well-known marker to represent the potential of cell proliferation, Ki67 in chrysin group was substantially decreased in contrast with the vehicle group. With the suppression of glycolysis by reducing HK-2 in tumor tissue, energy supply to sustain tumor growth was blocked, and the proliferative capability of tumor cell was weakened.Fig. 6Chrysin inhibited HCC-LM3 xenograft growth in vivo. Nude mice with HCC-LM3 xenograft were randomly divided to groups when tumor volume reached 50 to 100 mm3. 30 mg/kg chrysin was administrated three times weekly by intraperitoneal injection. a, photograph of tumors in vehicle and chrysin-treated group; b, the change of body weight of tumor bearing mice; c, tumor growth curve in vehicle and treated group; d, tumor weight in vehicle and chrysin group; e, tumor tissues were subjected to immunohistochemistry staining with indicated antibodies to detect the change of HK-2, Ki67 after chrysin treatment | other | 34.3 |
The expression of HK-2 was reported to be elevated in many cancers and was considered a predictive marker of poor prognosis of HCC [23, 24], breast , and gastric cancer . Consistent with previous report, HK-2 expression was detected to be of high rate in most tested HCC cell lines and tumor tissue in contrast with the normal hepatic cell and tissue in our studies (Fig. 1). Dai W et al. demonstrated that HCC cells with high HK-2 expression displayed high aerobic glycolysis, as indicated by increased glucose uptake and lactate production . After chrysin treatment, with the reduction of HK-2, glucose consumption and lactate production in HCC cells were dramatically decreased (Fig. 2). Further investigations revealed that the suppression caused by chrysin was substantially attenuated after overexpression of HK-2, suggesting HK-2 played an important role in chrysin-mediated glycolysis suppression. Metabolic control analysis of HCC cell lines demonstrated that the main control of glycolytic flux was exerted by HKs . Along with the increase of HK-2, altered glucose metabolism often conferred cancer cells resistance to chemotherapy, it was evidenced that the sensitivity of tumor cells to chemotherapy were substantially enhanced via glycolysis inhibition by targeting HK-2 [29, 30]. Given the activity of chrysin against HK-2 and glycolysis, we thought chrysin had the potential to strengthen the efficacy of other chemotherapies. | other | 30.73 |
Chrysin inhibited HCC-LM3 xenograft growth in vivo. Nude mice with HCC-LM3 xenograft were randomly divided to groups when tumor volume reached 50 to 100 mm3. 30 mg/kg chrysin was administrated three times weekly by intraperitoneal injection. a, photograph of tumors in vehicle and chrysin-treated group; b, the change of body weight of tumor bearing mice; c, tumor growth curve in vehicle and treated group; d, tumor weight in vehicle and chrysin group; e, tumor tissues were subjected to immunohistochemistry staining with indicated antibodies to detect the change of HK-2, Ki67 after chrysin treatment | other | 29.1 |
In present study, our results demonstrated that chrysin had profound potency against HCC in vitro and in vivo. Mechanism investigations revealed that HK-2 played a pivotal role for chrysin to exert its activity in HCC. With the reduction of HK-2 after chrysin treatment, the glycolysis in HCC was markedly inhibited. Except the effect on HCC glycolysis, HK-2 was also proved to be involved in chrysin-induced apoptosis. Exposure to chrysin resulted in the decrease of the interaction between HK-2 and VDAC-1, which caused the release of pro-apoptotic proteins from mitochondrial and induced tumor cells to undergo apoptosis. | other | 32.53 |
In conclusion, the results of this study demonstrated chrysin had profound antitumor activity against HCC via inhibiting tumor glycolysis and inducing cell apoptosis. Different from the mechanism reported by previous studies, reduction of HK-2 was an important underlying mechanism for chrysin to exert its effect on cell metabolism and cell apoptosis. This study provided a novel preclinical rational for chrysin, or its analogue, to be applied for HCC management. | other | 31.95 |
Subsets and Splits