Predicting antimicrobial mechanism-of-action from transcriptomes: A generalizable explainable artificial intelligence approach

To better combat the expansion of antibiotic resistance in pathogens, new compounds, particularly those with novel mechanisms-of-action [MOA], represent a major research priority in biomedical science. However, rediscovery of known antibiotics demonstrates a need for approaches that accurately identify potential novelty with higher throughput and reduced labor. Here we describe an explainable artificial intelligence classification methodology that emphasizes prediction performance and human interpretability by using a Hierarchical Ensemble of Classifiers model optimized with a novel feature selection algorithm called Clairvoyance; collectively referred to as a CoHEC model. We evaluated our methods using whole transcriptome responses from Escherichia coli challenged with 41 known antibiotics and 9 crude extracts while depositing 122 transcriptomes unique to this study. Our CoHEC model can properly predict the primary MOA of previously unobserved compounds in both purified forms and crude extracts at an accuracy above 99%, while also correctly identifying darobactin, a newly discovered antibiotic, as having a novel MOA. In addition, we deploy our methods on a recent E. coli transcriptomics dataset from a different strain and a Mycobacterium smegmatis metabolomics timeseries dataset showcasing exceptionally high performance; improving upon the performance metrics of the original publications. We not only provide insight into the biological interpretation of our model but also that the concept of MOA is a non-discrete heuristic with diverse effects for different compounds within the same MOA, suggesting substantial antibiotic diversity awaiting discovery within existing MOA.

Novel Antimicrobials from Uncultured Bacteria Acting against Mycobacterium tuberculosis

Mycobacterium tuberculosis, which causes tuberculosis (TB), is estimated to infect one-third of the world's population. The overall burden and the emergence of drug-resistant strains of Mycobacterium tuberculosis underscore the need for new therapeutic options against this important human pathogen. Our recent work demonstrated the success of natural product discovery in identifying novel compounds with efficacy against Mycobacterium tuberculosis Here, we improve on these methods by combining improved isolation and Mycobacterium tuberculosis selective screening to identify three new anti-TB compounds: streptomycobactin, kitamycobactin, and amycobactin. We were unable to obtain mutants resistant to streptomycobactin, and its target remains to be elucidated. We identify the target of kitamycobactin to be the mycobacterial ClpP1P2C1 protease and confirm that kitamycobactin is an analog of the previously identified compound lassomycin. Further, we identify the target of amycobactin to be the essential protein secretion pore SecY. We show further that amycobactin inhibits protein secretion via the SecY translocon. Importantly, this inhibition is bactericidal to nonreplicating Mycobacterium tuberculosis This is the first compound, to our knowledge, that targets the Sec protein secretion machinery in Mycobacterium tuberculosis This work underscores the ability of natural product discovery to deliver not only new compounds with activity against Mycobacterium tuberculosis but also compounds with novel targets.IMPORTANCE Decreasing discovery rates and increasing resistance have underscored the need for novel therapeutic options to treat Mycobacterium tuberculosis infection. Here, we screen extracts from previously uncultured soil microbes for specific activity against Mycobacterium tuberculosis, identifying three novel compounds. We further define the mechanism of action of one compound, amycobactin, and demonstrate that it inhibits protein secretion through the Sec translocation machinery.

Reducing the Bottleneck in Discovery of Novel Antibiotics

Most antibiotics were discovered by screening soil actinomycetes, but the efficiency of the discovery platform collapsed in the 1960s. By now, more than 3000 antibiotics have been described and most of the current discovery effort is focused on the rediscovery of known compounds, making the approach impractical. The last marketed broad-spectrum antibiotics discovered were daptomycin, linezolid, and fidaxomicin. The current state of the art in the development of new anti-infectives is a non-existent pipeline in the absence of a discovery platform. This is particularly troubling given the emergence of pan-resistant pathogens. The current practice in dealing with the problem of the background of known compounds is to use chemical dereplication of extracts to assess the relative novelty of a compound it contains. Dereplication typically requires scale-up, extraction, and often fractionation before an accurate mass and structure can be produced by MS analysis in combination with 2D NMR. Here, we describe a transcriptome analysis approach using RNA sequencing (RNASeq) to identify promising novel antimicrobial compounds from microbial extracts. Our pipeline permits identification of antimicrobial compounds that produce distinct transcription profiles using unfractionated cell extracts. This efficient pipeline will eliminate the requirement for purification and structure determination of compounds from extracts and will facilitate high-throughput screen of cell extracts for identification of novel compounds.

Lassomycin, a ribosomally synthesized cyclic peptide, kills mycobacterium tuberculosis by targeting the ATP-dependent protease ClpC1P1P2

Languishing antibiotic discovery and flourishing antibiotic resistance have prompted the development of alternative untapped sources for antibiotic discovery, including previously uncultured bacteria. Here, we screen extracts from uncultured species against Mycobacterium tuberculosis and identify lassomycin, an antibiotic that exhibits potent bactericidal activity against both growing and dormant mycobacteria, including drug-resistant forms of M. tuberculosis, but little activity against other bacteria or mammalian cells. Lassomycin is a highly basic, ribosomally encoded cyclic peptide with an unusual structural fold that only partially resembles that of other lasso peptides. We show that lassomycin binds to a highly acidic region of the ClpC1 ATPase complex and markedly stimulates its ATPase activity without stimulating ClpP1P2-catalyzed protein breakdown, which is essential for viability of mycobacteria. This mechanism, uncoupling ATPase from proteolytic activity, accounts for the bactericidal activity of lassomycin.

Specialized persister cells and the mechanism of multidrug tolerance in Escherichia coli

Bacterial populations produce persisters, cells that neither grow nor die in the presence of bactericidal agents, and thus exhibit multidrug tolerance (MDT). The mechanisms of MDT and the nature of persisters have remained elusive. Our previous research has shown that persisters are largely responsible for the recalcitrance of biofilm infections. A general method for isolating persisters was developed, based on lysis of regular cells by ampicillin. A gene expression profile of persisters contained toxin-antitoxin (TA) modules and other genes that can block important cellular functions such as translation. Bactericidal antibiotics kill cells by corrupting the target function (for example, aminoglycosides interrupt translation, producing toxic peptides). We reasoned that inhibition of translation will lead to a shutdown of cellular functions, preventing antibiotics from corrupting their targets, giving rise to MDT persister cells. Overproduction of the RelE toxin, an inhibitor of translation, caused a sharp increase in persisters. Functional expression of a putative HipA toxin also increased persisters, while deletion of the hipBA module caused a sharp decrease in persisters in both stationary and biofilm populations. HipA is thus the first validated persister-MDT gene. We suggest that random fluctuation in the levels of MDT proteins leads to the formation of rare persister cells. The function of these specialized dormant cells is to ensure the survival of the population in the presence of lethal factors.

Persister cells and tolerance to antimicrobials

Bacterial populations produce persister cells that neither grow nor die in the presence of microbicidal antibiotics. Persisters are largely responsible for high levels of biofilm tolerance to antimicrobials, but virtually nothing was known about their biology. Tolerance of Escherichia coli to ampicillin and ofloxacin was tested at different growth stages to gain insight into the nature of persisters. The number of persisters did not change in lag or early exponential phase, and increased dramatically in mid-exponential phase. Similar dynamics were observed with Pseudomonas aeruginosa (ofloxacin) and Staphylococcus aureus (ciprofloxacin and penicillin). This shows that production of persisters depends on growth stage. Maintaining a culture of E. coli at early exponential phase by reinoculation eliminated persisters. This suggests that persisters are not at a particular stage in the cell cycle, neither are they defective cells nor cells created in response to antibiotics. Our data indicate that persisters are specialized survivor cells.

Effect of immunization on herpes simplex virus type 1 latent infection in the trigeminal ganglion

PURPOSE

To quantify and characterize immune protection from herpes simplex virus (HSV) latent infection in mice following corneal challenge.

METHODS

Mice immunized or mock-immunized and boosted in the flank with an HSV replication-deficient mutant were challenged by corneal inoculation with wild type (wt) or thymidine kinase-negative (TK(-)) HSV. At specified times post challenge, trigeminal ganglia were assayed for in vitro reactivation, latent and acute viral load (using quantitative PCR), acute infection, and cellular infiltration (hematoxylin and eosin stained sections).

RESULTS

With wt HSV challenge infection, immunization led to reduced reactivation, significantly less latent and acute viral DNA, and no acute viral replication in ganglia, and rapid infiltration of inflammatory cells. Immunization had little effect on viral load following challenge with replication-conditional TK(-) mutant virus.

CONCLUSION

These results indicate that immune protection from latent HSV infection in mouse trigeminal ganglia following ocular infection can act under these experimental conditions to block acute viral replication in ganglia and is directed to antigenic targets within the ganglia.