Deep learning–driven bacterial cytological profiling to determine antimicrobial mechanisms in Mycobacterium tuberculosis
Edited by Carl Nathan, Weill Cornell Medicine, New York, NY; received October 22, 2024; accepted January 8, 2025
Significance
Bacterial cytological profiling (BCP) is a well-established method to determine the mechanism of action (MOA) for antibiotics by examining the morphological changes that occur when bacteria are treated with a compound of interest. This study demonstrates the application of convolutional neural networks (CNN) to overcome technical challenges with a traditional approach to BCP, creating a robust platform to rapidly determine MOA for Mycobacterium tuberculosis. We demonstrate the capability of this platform by using it to confirm the MOA of several compounds that target M. tuberculosis. Our findings underscore the potential of CNN-based BCP to enhance the accuracy and efficiency of MOA determination, particularly for challenging pathogens like M. tuberculosis.
Abstract
Tuberculosis (TB), caused by Mycobacterium tuberculosis, remains a significant global health threat, affecting an estimated 10.6 million people in 2022. The emergence of multidrug resistant and extensively drug resistant strains necessitates the development of novel and effective drugs. Accelerating the determination of mechanisms of action (MOAs) for these drugs is crucial for advancing TB treatment. This study introduces MycoBCP, a unique adaptation of bacterial cytological profiling (BCP) tailored to M. tuberculosis, utilizing the application of convolutional neural networks (CNNs) within BCP to overcome challenges posed by traditional image analysis techniques. Using MycoBCP, we analyzed the morphological effects of various antimicrobial compounds on M. tuberculosis, capturing broad patterns rather than relying on precise cell segmentation. This approach circumvented issues such as cell clumping and uneven staining, which are prevalent in M. tuberculosis. In a blind test, MycoBCP accurately identified the MOA for 96% of the compounds, with a single misclassification of rifabutin, which was incorrectly categorized as affecting translation rather than transcription. The similar morphologies resulting from transcription and translation inhibition indicate a need for further refinement to distinguish them more effectively. Application of MycoBCP to a series of antitubercular agents successfully identified known MOAs and revealed unique effects, demonstrating its utility in early drug discovery and development. Our findings underscore the potential of CNN-based BCP to enhance the accuracy and efficiency of MOA determination, particularly for challenging pathogens like M. tuberculosis. MycoBCP represents a significant advancement in TB drug development, offering a robust and adaptable method for high-throughput screening of antimicrobial compounds.
Get full access to this article
Purchase, subscribe or recommend this article to your librarian.
Data, Materials, and Software Availability
Code and example files (27) can be found at github.com/MycoBCP/MycoBCP. Some study data are available. Training data for the CNN described in this manuscript exceeds 40,000 high resolution fluorescence microscopy images. Instead, we provide a checkpoint of the trained network that can be used on similar images.
Acknowledgments
Research was supported by the Bill and Melinda Gates Foundation (INV-040479 Joseph Sugie) for the purpose of supporting M. tuberculosis drug development.
Author contributions
D.Q., T.P., J.P., and J.S. designed research; D.Q., M.S., S.A., L.A., A.B., A.D., and J.S. performed research; J.S. contributed new reagents/analytic tools; D.Q., M.S., J.P., and J.S. analyzed data; and J.P. and J.S. wrote the paper.
Competing interests
D.Q., M.S., and J.S. are employed by Linnaeus Bioscience, Inc. J.P. has an equity interest in Linnaeus Bioscience Incorporated and receive consulting income from the company. The terms of this arrangement have been reviewed and approved by the University of California, San Diego in accordance with its conflict-of-interest policies.
Supporting Information
Appendix 01 (PDF)
- Download
- 4.10 MB
References
1
World Health Organization, Global tuberculosis report 2023 (World Health Organization, Geneva, Switzerland, 2023).
2
World Health Organization, WHO consolidated guidelines on tuberculosis: Module 4: Treatment: drug-susceptible tuberculosis treatment (World Health Organization, Geneva, Switzerland, 2022).
3
A. Lamsa, W.-T. Liu, P. C. Dorrestein, K. Pogliano, The Bacillus subtilis cannibalism toxin SDP collapses the proton motive force and induces autolysis. Mol. Microbiol. 84, 486–500 (2012).
4
P. Nonejuie, M. Burkart, K. Pogliano, J. Pogliano, Bacterial cytological profiling rapidly identifies the cellular pathways targeted by antibacterial molecules. Proc. Natl. Acad. Sci. U.S.A. 110, 16169–16174 (2013).
5
D. T. Quach, G. Sakoulas, V. Nizet, J. Pogliano, K. Pogliano, Bacterial Cytological Profiling (BCP) as a rapid and accurate antimicrobial susceptibility testing method for Staphylococcus aureus. EBioMedicine 4, 95–103 (2016).
6
P. Nonejuie et al., Application of bacterial cytological profiling to crude natural product extracts reveals the antibacterial arsenal of Bacillus subtilis. J. Antibiot. (Tokyo) 69, 353–361 (2016).
7
A. Lamsa et al., Rapid inhibition profiling in Bacillus subtilis to identify the mechanism of action of new antimicrobials. ACS Chem. Biol. 11, 2222–2231 (2016).
8
H. H. Htoo et al., Bacterial cytological profiling as a tool to study mechanisms of action of antibiotics that are active against Acinetobacter baumannii. Antimicrob. Agents Chemother. 63, e02310-18 (2019), https://doi.org/10.1128/aac.02310-18.
9
S. Stylianidou, C. Brennan, S. B. Nissen, N. J. Kuwada, P. A. Wiggins, SuperSegger: Robust image segmentation, analysis and lineage tracking of bacterial cells. Mol. Microbiol. 102, 690–700 (2016).
10
A. Paintdakhi et al., Oufti: An integrated software package for high-accuracy, high-throughput quantitative microscopy analysis. Mol. Microbiol. 99, 767–777 (2016).
11
T. Ursell et al., Rapid, precise quantification of bacterial cellular dimensions across a genomic-scale knockout library. BMC Biol. 15, 17 (2017).
12
C. McQuin et al., Cell Profiler 3.0: Next-generation image processing for biology. PLoS Biol. 16, e2005970 (2018).
13
T. C. Smith et al., Morphological profiling of tubercle bacilli identifies drug pathways of action. Proc. Natl. Acad. Sci. U.S.A. 117, 18744–18753 (2020).
14
E. Moen et al., Deep learning for cellular image analysis. Nat. Methods 16, 1233–1246 (2019).
15
A. Zagajewski et al., Deep learning and single-cell phenotyping for rapid antimicrobial susceptibility detection in Escherichia coli. Commun. Biol. 6, 1–12 (2023).
16
C. Shelton et al., Triazolopyrimidines target aerobic respiration in Mycobacterium tuberculosis. Antimicrob. Agents Chemother. 66, e0204121 (2022).
17
L. A. T. Cleghorn et al., Identification of morpholino thiophenes as novel Mycobacterium tuberculosis inhibitors, targeting QcrB. J. Med. Chem. 61, 6592–6608 (2018).
18
E. S. Zuniga et al., The synthesis and evaluation of triazolopyrimidines as anti-tubercular agents. Bioorg. Med. Chem. 25, 3922–3946 (2017).
19
K. A. Abrahams et al., Identification of novel imidazo[1,2-a]pyridine inhibitors targeting M. tuberculosis QcrB. PLoS One 7, e52951 (2012).
20
T. O’Malley et al., Imidazopyridine compounds inhibit mycobacterial growth by depleting ATP levels. Antimicrob. Agents Chemother. 62, e02439-17 (2018), https://doi.org/10.1128/aac.02439-17.
21
M. B. McNeil et al., Multiple mutations in Mycobacterium tuberculosis MmpL3 increase resistance to MmpL3 inhibitors. mSphere 5, e00985-20 (2020), https://doi.org/10.1128/msphere.00985-20.
22
J. Ollinger et al., A dual read-out assay to evaluate the potency of compounds active against Mycobacterium tuberculosis. PLoS One 8, e60531 (2013).
23
J. V. Early, S. Mullen, T. Parish, A rapid, low pH, nutrient stress, assay to determine the bactericidal activity of compounds against non-replicating Mycobacterium tuberculosis. PLoS One 14, e0222970 (2019).
24
J. Ollinger et al., A high-throughput whole cell screen to identify inhibitors of Mycobacterium tuberculosis. PLoS One 14, e0205479 (2019).
25
A. Krizhevsky, I. Sutskever, G. Hinton, ImageNet classification with deep convolutional neural networks. Commun. ACM 60, 84–90 (2017).
26
K. O’Shea, R. Nash, An introduction to convolutional neural networks. arXiv. [Preprint] (2015). https://doi.org/10.48550/arXiv.1511.08458 (Accessed 17 May 2024).
27
J. Sugie, MycoBCP sample code. GitHub. https://github.com/MycoBCP/MycoBCP. Deposited 16 September 2024.
Information & Authors
Information
Published in
Classifications
Copyright
Copyright © 2025 the Author(s). Published by PNAS. This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).
Data, Materials, and Software Availability
Code and example files (27) can be found at github.com/MycoBCP/MycoBCP. Some study data are available. Training data for the CNN described in this manuscript exceeds 40,000 high resolution fluorescence microscopy images. Instead, we provide a checkpoint of the trained network that can be used on similar images.
Submission history
Received: October 22, 2024
Accepted: January 8, 2025
Published online: February 6, 2025
Published in issue: February 11, 2025
Change history
March 17, 2025: The text of this article has been updated; please see accompanying Correction for details. Previous version (February 6, 2025)
Keywords
Acknowledgments
Research was supported by the Bill and Melinda Gates Foundation (INV-040479 Joseph Sugie) for the purpose of supporting M. tuberculosis drug development.
Author contributions
D.Q., T.P., J.P., and J.S. designed research; D.Q., M.S., S.A., L.A., A.B., A.D., and J.S. performed research; J.S. contributed new reagents/analytic tools; D.Q., M.S., J.P., and J.S. analyzed data; and J.P. and J.S. wrote the paper.
Competing interests
D.Q., M.S., and J.S. are employed by Linnaeus Bioscience, Inc. J.P. has an equity interest in Linnaeus Bioscience Incorporated and receive consulting income from the company. The terms of this arrangement have been reviewed and approved by the University of California, San Diego in accordance with its conflict-of-interest policies.
Notes
This article is a PNAS Direct Submission.
Authors
Metrics & Citations
Metrics
Altmetrics
Citations
Cite this article
Deep learning–driven bacterial cytological profiling to determine antimicrobial mechanisms in Mycobacterium tuberculosis, Proc. Natl. Acad. Sci. U.S.A.
122 (6) e2419813122,
https://doi.org/10.1073/pnas.2419813122
(2025).
Copied!
Copying failed.
Export the article citation data by selecting a format from the list below and clicking Export.
View Options
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginRecommend to a librarian
Recommend PNAS to a LibrarianPurchase options
Purchase this article to access the full text.
Restore content access
View options
PDF format
Download this article as a PDF file
DOWNLOAD PDF