The AlphaFold Protein Structure Database contains over 200,000,000 three-dimensional protein structures. These structures are the results of predictions of how the protein will fold using AlphaFold, a neural network model. David Baker, Demis Hassabis, and John M. Jumper won the 2024 Nobel Prize in Chemistry for this technology and for using computational approaches to protein design. (JUMPER et al., 2021) (VARADI et al., 2022) (NOBEL PRIZE OUTREACH AB, 2024)
In the present research, this tool was used to obtain the three-dimensional structures of exotoxin U and exoenzymes T and S. Only then was it possible to perform reverse docking .
The procedure in AutoDock 4 (FORLI, S. et al., 2016) (MORRIS et al., 2009) for reverse docking was similar to that of other dockings , with the best compounds in the exotoxin A tests being tested in exotoxin U and exoenzymes S and T. The difference between the data collection procedure of standard docking and that of reverse docking is in the preparation of the grid , since the size of this grid is different for different proteins, but the spacing in all macromolecules was the same: 1 angstrom. In exotoxin U, whose three-dimensional structure identified by AF-A0A221LFV8-F1-v4 was obtained from the AlphaFold Protein Structure Database (JUMPER et al., 2021) (VARADI et al., 2022), the grid was made with 90 points in the X dimension with a center of 5.828, 126 points in the Y dimension with a center of -7.843, and 96 points in the Z dimension with a center of 0.615. The grid of exoenzyme S, whose three-dimensional structure identified by AF-Q51451-F1-v4 was obtained from the AlphaFold Protein Structure Database (JUMPER et al., 2021) (VARADI et al., 2022), was made with 98 points in the X dimension with a center of 0.886, 126 points in the Y dimension with a center of 5.318, and 96 points in the Z dimension with a center of 0.691. In exoenzyme T, whose three-dimensional structure identified by AF-A0A3S3VVG0-F1-v4 was obtained from the AlphaFold Protein Structure Database (JUMPER et al., 2021) (VARADI et al., 2022), the grid was established with 126 points in the X dimension with a center of –10.178, 120 points in the Y dimension with a center of –3.675 and 124 points in the Z dimension with a center -2.092.
For reverse docking , the four exotoxin A inhibitor compounds with the best therapeutic potential were selected, i.e., Samples 4, 12, and 11, and PJ34, the control group. Thus, by performing molecular docking of these compounds with other toxic substances, such as exotoxin U and exoenzymes S and T, it is possible to evaluate whether drugs developed from these small molecules can act as inhibitors for more than one virulence factor of Pseudomonas aeruginosa . Therefore, this step is important to ascertain the general efficacy of these compounds against nosocomial infections of this multidrug-resistant bacterium.
Source: Authors
In the graph above, it is possible to analyze the performance of each compound in terms of binding energy for each virulence factor of Pseudomonas aeruginosa tested in the reverse docking stage. Thus, it is clear that Sample 11 had the best results overall, followed by Sample 4. PJ34 had the third best performance, with Sample 12 having a poor performance when compared to the other compounds.
Sample 11 was the positive highlight for most of the proteins tested, having only a slightly lower result than Sample 4 for exotoxin U. Furthermore, it is important to note that Sample 11 had the only binding energy value lower than -7, with a result of -7.18 for exoenzyme T. That said, Sample 11 performed considerably worse on these virulence factors than on exotoxin A. For example, when comparing Sample 11's best result in reverse docking , which was for exoenzyme T, with that of exotoxin A (-9.75), there was a difference of -2.57, which is therefore a significant drop in performance.
Sample 4 also performed relatively well when compared to this data set, as it actually had the best binding energy on exotoxin U at –6.91 and consistently outperformed PJ34 and Sample 12. Furthermore, Sample 4’s binding energy was lower than –6 on all proteins tested. However, there was again a significant drop in performance when comparing these results to those of exotoxin A. In fact, comparing Sample 4’s best performance in reverse docking to its best performance on exotoxin A, there was a difference of –2.09.
PJ34, despite maintaining a constant performance superior to that of Sample 12, did not stand out in this reverse docking parameter. In fact, the best result of PJ34 in this aspect was -6.02 in exoenzyme T, while the other values were not lower than -6. Thus, there was also a difference of more than -2 between the performance of PJ34 in exotoxin A and in exoenzyme T, and the binding energy of the control group in exotoxin A was -2.21 lower than in reverse docking .
Sample 12 clearly had the worst performance when compared to the other compounds and did not excel in any of the proteins tested, with its best result being –5.04. Thus, Sample 12 had the greatest reduction in performance, with a large difference of –3.4.
Source: Authors
In the graph above, it is possible to analyze the performance of each compound in terms of ligand efficiency for each virulence factor of Pseudomonas aeruginosa tested in the reverse docking stage. Thus, it is clear that Sample 11 had the best results overall, followed by PJ34 and Sample 4. In addition, Sample 12 had a considerably worse performance than the other compounds.
Sample 11 was the standout performer for most of the proteins tested, with only one result equal to Sample 4 for exotoxin U and the best performance for the rest. Furthermore, it is important to note that Sample 11 had the only ligand efficiency value lower than -0.29, with a result of -0.3 for exoenzyme T. That said, Sample 11 performed worse on these virulence factors than on exotoxin A, with a difference of -0.11 between the yield on exotoxin A and that on exoenzyme T.
PJ34 also performed well in this data set, with results very similar to those of Sample 4. Thus, the best yields of this compound are -0.27 in exotoxin U and exoenzyme T, with a slightly worse performance in exoenzyme S, with -0.26. Furthermore, there was also a drop in performance compared to exotoxin A, with a difference of -0.1.
Sample 4 had a very similar result to PJ34, performing better on exotoxin U with -0.28 and worse on exoenzyme T with -0.25. Furthermore, comparing the lowest value of this data set with that of exotoxin A, there was a performance reduction of -0.08. Furthermore, it is important to highlight that this was the smallest performance drop among the compounds tested.
Sample 12 clearly performed the worst when compared to the other compounds and did not excel in any of the proteins tested, with its best result being -0.19. Thus, there is a difference of -0.06 between the best result of Sample 12 and the worst result of Sample 4. Thus, it is not surprising that Sample 12 had the greatest reduction in performance, with a large difference of -0.13.
Source: Authors
In the graph above, it is possible to analyze the performance of each compound in terms of the binding constant for each protein tested in the reverse docking step. Thus, it is clear that Sample 11 had the best performance in this data set, followed by Sample 4. While PJ34 had the third best performance, Sample 12 had a much worse performance than the other compounds.
Sample 11 was the positive highlight in most of the proteins tested, but had a lower result than Sample 4 in exotoxin U. Furthermore, it is important to highlight that Sample 11 had the only inhibition constant lower than 6 uM, with a result of 5.46 in exoenzyme T. The drop in performance in this aspect is even more evident, with the inhibition constant in this indicator in exoenzyme T being higher than that of exotoxin A by approximately 5.389 uM.
Sample 4 performed worse than Sample 11 in both exoenzymes, with its best result being exotoxin U, with 8.64 uM. Again, the reduction in performance compared to exotoxin U was significant, with a difference of approximately 8.386 uM.
PJ34 failed to stand out in this indicator, with its best performance being 38.91 uM in exoenzyme T. Thus, there was a drop in performance in relation to exotoxin A, with a large difference of around 37.986 uM.
Finally, Sample 4 clearly performed worst, with its best performance being 203.13 for exotoxin U. Thus, there is the greatest reduction in performance among the compounds tested in relation to exotoxin A, with a difference of approximately 202.479 uM.
FORLI, Stefano et al. Computational protein–ligand docking and virtual drug screening with the AutoDock suite. Nature protocols , vol. 11, no. 5, p. 905-919, 2016. Available at: < https://www.nature.com/articles/nprot.2016.051 >. Accessed on: 5 Aug. 2024.
MORRIS, GM et al. AutoDock4 and AutoDockTools4: automated docking with selective receptor flexibility. J. Computational Chemistry , v. 30, n. 16, p. 2785-2791, 2009. Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2760638/ . Accessed on: August 11, 2024.
JUMPER, John et al. Highly accurate protein structure prediction with AlphaFold. nature , vol. 596, n. 7873, p. 583-589, 2021. Available at: < https://www.nature.com/articles/s41586- 021-03819-2 >. Accessed on: 13 Aug. 2024.
NOBEL PRIZE OUTREACH AB. The Nobel Prize in Chemistry 2024. NobelPrize.org , 2024. Available at: < https://www.nobelprize.org/prizes/chemistry/2024/summary/ >. Accessed on: October 14, 2024.
VARADI, Mihaly et al. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic acids research , vol. 50, no. D1, p. D439-D444, 2022. Available at: < https://academic.oup.com/nar/article/50/D1/D439/6430488?login=false >. Accessed on: 5 Aug. 2024.
Comentários