Transcriptomic signatures for identifying drug-susceptible mycobacterium tuberculosis complex

Abstract

Techniques are provided for determining whether a biological sample obtained from a patient contains bacteria, such as My cobacterium tuberculosis complex, susceptible to treatment by a particular antibiotic performed at least in part by using at least one computer hardware processor to obtain RNA expression data for the biological sample previously- obtained by profiling RNA from the biological sample after exposure to the at least one antibiotic, and for each particular antibiotic: determine whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some bacterial genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic, and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the bacteria in the biological sample is susceptible to treatment by the particular antibiotic.

Description

TRANSCRIPTOMIC SIGNATURES FOR IDENTIFYING DRUG-SUSCEPTIBLEMYCOBACTERIUM TUBERCULOSIS COMPLEXCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] The present application claims the benefit of priority under 35 U.S.C. § 119(e) to U.S. Provisional Application Serial No.: 63 / 582,509, titled “TRANSCRIPTOMIC SIGNATURES FOR IDENTIFYING DRUG-SUSCEPTIBLE MYCOBACTERIUM TUBERCULOSIS”, filed on September 13, 2023 (Attorney Docket No: H0776.70163US00) and to U.S.Provisional Application Serial No.: 63 / 590,452, titled “TRANSCRIPTOMIC SIGNATURES FOR IDENTIFYING DRUG-SUSCEPTIBLE MYCOBACTERIUM TUBERCULOSISCOMPLEX”, filed on October 15, 2023 (Attorney Docket No.: H0776.70163US01), each of which is incorporated by reference in its entirety.FEDERALLY SPONSORED RESEARCH

[0002] This invention was made with government support under AI176498 awarded by National Institutes of Health (NIH). The government has certain rights in the invention.BACKGROUND

[0003] Successful treatment of tuberculosis infections requires the prompt and accurate diagnosis of resistance to the full panel of antibiotics in clinical use. Mycobacterium tuberculosis complex (Mtb) has among the highest resistance rates of any pathogen globally, but Mtb resistance remains the most challenging to diagnose due to its very slow growth in in vitro culture and culture’s inordinate cost and biohazard.SUMMARY

[0004] According to some aspects, there is provided a method for determining whether a biological sample obtained from a patient contains mycobacterium tuberculosis complex (Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: using at least one computer hardware processor: obtaining RNA expression data for the biological sample previously-obtained by profiling RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether aparticular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some Mtb genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic.

[0005] According to some aspects, there is provided a system comprising: at least one computer hardware processor; and at least one non-transitory computer-readable storage medium storing instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform a method for determining whether a biological sample obtained from a patient contains mycobacterium tuberculosis complex (Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: obtaining RNA expression data for the biological sample previously-obtained by profiling RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some Mtb genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic.

[0006] According to some aspects, there is provided at least one non-transitory computer- readable storage medium storing instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform a method for determining whether a biological sample obtained from a patient contains mycobacterium tuberculosis complex (Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: obtaining RNA expression data for the biological sample previously-obtained by profiling RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some Mtb genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; andwhen it is determined that the transcriptomic signature is present in the RNA expression data, determining that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic.

[0007] According to some aspects, there is provided method for determining whether a biological sample obtained from a patient contains bacteria susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: using at least one computer hardware processor: obtaining RNA expression data for the biological sample previously-obtained by profiling RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some bacterial genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the bacteria in the biological sample is susceptible to treatment by the particular antibiotic.BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Various aspects and embodiments will be described with reference to the following figures. It should be appreciated that the figures are not necessarily drawn to scale. For purposes of clarity, not every component may be labeled in every drawing. In the drawings:

[0009] FIG. 1 is an example process for determining whether a biological sample obtained from a patient contains bacteria susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, in accordance with some embodiments of the technology described herein.

[0010] FIG. 2 is an illustrative implementation of a computer system that may be used in connection with some embodiments of the technology described herein.

[0011] FIG. 3 shows Mtb gene transcription under multiple stress (leftmost column), exposure to PA824 (center) and PZA (rightmost). Data reanalyzed from Boshoffet al. JBC 2004 and Deb et al. PloS One 2009. ‘Response to stress’ was the most upregulated pathway. ‘Translation’ or ribosomal function was the most downregulated pathway (blue square and blue in the heatmap) under stress and PA824 exposure. Only 3 conditions are displayed due to space.

[0012] FIG. 4 shows Mtb H37Rv transcriptional responses to drug exposure. CIP: ciprofloxacin, SM: streptomycin, INH: isoniazid. Log-phase cultures were exposed to antibiotic, lysed, and analyzed using nCounter probe sets. Counts were normalized by the mean of the middle 50% of all counts for a sample, and DE was determined by comparing drug-exposed and unexposed samples. Genes contributing to the signature are on the x axis. Shown are the responses of a panel of susceptible (black bars, Upper) or resistant (gray bars; Lower) strains. Error bars represent SD of three to seven isolates tested. Adapted from ref 5.

[0013] FIGs. 5A-5B show determination of antibiotic susceptibility is independent of the mechanism of resistance. A: Mtb laboratory and clinical isolates exposed to 1 pg / mL isoniazid. INH-R, isoniazid-resistant; INH-S, isoniazid-sensitive. B: Isoniazid-sensitive and high- or low-level resistant Mtb strains were exposed to isoniazid at 1 pg / mL or 0.2 pg / mL Dashed lines indicate three SDs from the mean of the resistant samples. From ref 5.

[0014] FIGs. 6A-6B show the ribosomal synthesis (RS) ratio correlates with Mtb replication.

[0015] FIGs. 7A-7C show RIF, BDQ and isoniazid in vitro exposure suppress the RS ratio (top) before reducing growth (bottom). The reduction in RS ratio was most rapid and profound for sterilizing drugs (orange) vs non-sterilizing drug (green). Adapted from ref-7. CFU = colony forming units.

[0016] FIG. 8 shows a schematic of RNA DR assay for each drug using NanoString to and Co are selected in Aiml for maximal signal. The assay will be run before (1) and after (2) culture of the prospectively collected sputum. Control unexposed runs will be conducted for each sample but may or may not be needed in the final workflow due to the use of internal controls.

[0017] FIG. 9 shows a method of determining the transcriptional signature of differential gene expression by Mycobacterium tuberculosis complex in response to antibiotics can be used as a marker of drug- sensitivity.

[0018] FIG. 10 shows a schematic of study isolates.

[0019] FIGs, 11A-11B show graphs depicting culture durations and antibiotic concentrations.

[0020] FIGs. 12A-12D show principal component analysis using log2fold values of top 10 genes from antibiotic and control TPs.

[0021] FIG. 13 shows a study flow for example 2, in accordance with some embodiments of the technology described herein.

[0022] FIGS. 14A-14D show example graphs indicating duration of culture and concentration ratio distribution.

[0023] FIG. 15 illustrates antibiotic heatmaps.

[0024] FIGS. 16A-16B illustrate temporal changes in select antibiotic-gene combinaitons.

[0025] FIGS. 17A-17B illustrates area under the receiver operating curve (AUC) used to determine the classification accuracy of a regression model in distinguishing antibiotic from control transcription profiles for antibiotics with at least four transcription profiles in different time periods.

[0026] FIGS. 18A-18D illustrates time category- specific heatmaps constructed using median log2fold values of upregulated and downregulated TARs for each antibiotic within a time category.DETAILED DESCRIPTION

[0027] Improved diagnosis and treatment are designated priorities by the World Health Organization and Centers for Disease Control for addressing the global antibiotic resistance challenge. For example, multidrug-resistant tuberculosis (MDR-TB) treatment is observed in 3.6% of new tuberculosis (TB) cases and 18% of previously treated cases. Treatment requires the concurrent use of five or more effective antibiotics and costs eight times as much as treatment of drug susceptible TB. The early diagnosis of TB and universal drug-susceptibility testing are important components of the World Health Organization’s END-TB strategy.

[0028] Conventional techniques for determining susceptibility of a sample of bacteria may include phenotypic testing and molecular testing. For example, phenotypic testing involves culturing cells containing the bacteria in vitro and administering a candidate antibiotic to the cultured sample. If the sample of bacteria is susceptible to the administered antibiotic, destruction of the bacteria and / or inhibition of further bacteria growth can be observed in the cultured sample. Such techniques are lengthy, requiring several days to reach a determination on antibiotic susceptibility of the sample.

[0029] Molecular testing includes DNA based tests. DNA based tests are available for detecting resistance to a few drugs but are limited by their need for extensive knowledge of DNA resistance determinants for interpretation, and concerns about limit of detection at the point-of-care. For novel antibiotics entering clinical use, genetic resistance determinants are particularly poorly understood and may be too dispersed across the genome to lend themselves to targeted sequencing and delays in understanding mechanisms may miss an opportunity to avoid more widespread resistance.

[0030] Molecular diagnostics that rely on the detection of targeted antibiotic resistance mutations in the Mycobacterium tuberculosis complex (Mtb) genome have been rapidly integrated into national TB programs and are currently standard-of-care. However, availablecommercial molecular assays have not yet been able to replace the traditional more costly and slow process of growth-based phenotypic drug susceptibility testing (pDST) due to the limited number of drugs tested. pDST is performed after growth is identified from a patient sample and requires several weeks (up to 10-12) to yield a result. Pathogen whole genome sequencing (WGS) is a promising replacement for pDST and is currently performed on culture-positive samples in several high resource clinical laboratories. The diagnostic accuracy of WGS is excellent for drugs like isoniazid (INH) and rifampicin (RIF), but still lags phenotypic methods for other key drugs like pyrazinamide (PZA) and the novel / repurposed TB drugs. The redundancy in resistance mechanisms combined with the diversity of background genetic variation can make it challenging to tease out causative from neutral variants for these agents. Mtb is also acquiring new mutations under antibiotic pressure including to novel antibiotics like bedaquiline and delamanid since being introduced recently into clinical care. Designing genotype-based diagnostics to these drugs currently requires more data on resistance determinants.

[0031] Individually, molecular and phenotypic testing have drawbacks, as described herein. The inventors have developed techniques that combine molecular and phenotypic approaches into a new technique that leverages the transcriptomic response of bacteria to drug exposure. The inventors have recognized that such techniques circumvent the challenges of phenotypic and molecular based methods while having the potential to be equally rapid as DNA based methods.

[0032] Like all cells, bacteria including Mtb demonstrate transcriptional changes in response to environmental stressors such as hypoxia or pH changes, and chemical pressure like antibiotic exposure. These responses may be the result of the direct antibiotic effect on a specific set of genes or pathway or may reflect complex or global changes in bacterial physiology. Because transcriptomic methods can detect both coding mutations and transcript abundance, they combine phenotypic and genotypic methods into a single test. The transcriptional antibiotic response (TAR) can be used to distinguish between antibioticsensitive and antibiotic -resistant strains of bacteria and is independent of the mechanism of resistance.

[0033] The inventors have developed new techniques for determining the susceptibility of bacteria to one or more antibiotics. For example, certain samples of bacteria can be resistant to particular antibiotics intended to target and treat the bacteria. As a result, administration of the particular antibiotic to resistant bacteria is ineffective or less effective in treating the bacteria (e.g., inhibiting growth of new bacteria, destroying existing bacteria). This may bedue to an antibiotic resistance mutation in the bacteria’s genome. On the other hand, administration of a particular antibiotic to a sample of bacteria that is susceptible to the particular antibiotic should be effective in treating the bacteria. Susceptibility tests are used to determine which antibiotics a particular sample of bacteria is sensitive to. The result of the susceptibility test (e.g., whether a particular sample of bacteria is susceptible to or resistant to a particular antibiotic) is useful in determining whether and / or how to administer the antibiotic to a patient.

[0034] The inventors have developed techniques which use RNA expression data to determine whether a sample of bacteria is susceptible to a particular antibiotic. For example, the techniques described herein include obtaining RNA expression data for a biological sample containing the bacteria. The RNA expression data is indicative of phenotypic and molecular information. For example, the response of the bacteria to the antibiotic (e.g., as reflected by certain genes being upregulated or down regulated) may be reflected in the RNA expression data (e.g., via increased or decreased expression levels for the upregulated or downregulated genes, respectively). In addition, genetic mutations in the DNA of the bacteria (e.g., mutations which may affect the bacteria’s resistance or susceptibility to a particular antibiotic) are also conveyed in the RNA expression data. The RNA expression data may be obtained by profiling RNA obtained from the biological sample after exposure to the antibiotic. The biological sample may be obtained from a patient, cultured, and treated with an antibiotic. The cultured sample may then be profiled (e.g., by performing molecular profiling, by performing sequencing) to obtain the RNA expression data. The RNA expression data may then be analyzed to determine whether the RNA obtained from the sample having been administered the particular antibiotic contains a particular transcriptomic signature. The particular transcriptomic signature comprises genes expressed in the RNA expression data that are differentially regulated in response to administration of the particular antibiotic to a sample of antibiotic- susceptible bacteria.

[0035] For example, when a particular antibiotic is administered to antibiotic-susceptible bacteria, the bacteria undergo a stress response which can be observed phenotypically in the bacteria’s RNA expression data. More specifically, one or more genes of the antibiotic- susceptible bacteria may be upregulated due to the bacteria’s stress response to administration of the antibiotic and one or more genes of the antibiotic-susceptible bacteria may be downregulated due to the bacteria’s stress response to administration of the antibiotic. The particular transcriptomic signature for a particular antibiotic comprises the one or more upregulated genes and / or the one or more downregulated genes observed in the antibiotic-susceptible bacteria’s RNA expression data. Therefore, the inventors have recognized that susceptibility of a new biological sample containing bacteria with unknown susceptibility can be determined by determining if the RNA expression data of the new biological sample includes the particular transcriptomic signature of one or more upregulated genes and / or one or more down regulated genes.

[0036] The inventors have recognized that Mtb susceptibility of a culture of bacteria to a particular antibiotic may be determined by observing the stress response of the bacteria to the particular antibiotic, as opposed to observing growth or lack thereof of the bacteria, which is a lengthy process that can take days to observe. The stress response of the bacteria, by contrast, can be observed by analyzing the differentially regulated genes of the bacteria in response to administration of the antibiotic, which can be completed more quickly than growth-based techniques.

[0037] A stress response indicates a change in cellular state in response to the action of an external factor, including antibiotic, that constitutes the earliest form of cellular stunting or death. The stress response may be defined and quantified using any of the following molecular cell measurements including messenger or other non-coding RNA levels, epigenetic markers, proteomics or metabolomics. Genes with altered transcription (either upregulated or downregulated) as an indicator of stress response include head shock response (e.g. the hsp gene), DNA damage (e.g. lexA), master transcriptional regulators (e.g. whib7), efflux pump and other membrane transporters, cell wall component biosynthesis enzymes (e.g. FAS-II genes), siderophore biosynthesis enzymes (e.g. the mbt genes), chaperone proteins, virulence effector genes (e.g. esx and nice genes), ribosomal subunit proteins (including 16S and 30S and others) and ribonuclease P protein, mycolic acid desaturation genes, resuscitation promoting factors (rpf genes), respiratory genes e.g. NADH dehydrogenase genes, and ATP synthetases, and hypoxic response genes (e.g. hspX). In some embodiments, at least two differentially regulated genes in a transcriptomic signature are associated with a stress response pathway. For example, the at least two differentially regulated genes associated with the stress response pathway may be upregulated in response to administration of one or more antibiotics. For example, the at least two differentially regulated genes may comprise at least one of a heat shock protein (hsp') or a / urA gene. In other embodiments, the at least two differentially regulated genes associated with the stress response pathway may be downregulated in response to administration of one or more antibiotics. For example, the at least two differentially regulated genes may comprise at leasttwo genes encoding for chaperone proteins involved in the stress response pathway (e.g., groEL2, groES).

[0038] Table S8 illustrates the impact of different antibiotics on stress response pathways. In particular, table S8 provides information regarding which antibiotics impact certain genes (e.g., hsp.furA. groEE2 and groES) in the stress response pathway, and whether the impact is upregulation or downregulation of those genes. Table S8 illustrates that response (e.g., up- or downregulation) to an antibiotic of genes belonging to the stress response pathway can be studied to determine susceptibility of the sample to an antibiotic (e.g., whether the sample is resistant or susceptible to the antibiotics).

[0039] The techniques described herein may be applied to Mycobacterium tuberculosis complex (Mtb). Mtb has among the highest resistance rates of any pathogen globally. Determining whether a sample of Mtb is susceptible or resistant to a particular antibiotic is challenging to diagnose due to Mtb’s very slow growth in in vitro culture and the inordinate cost and biohazard of culturing Mtb. Successful treatment of tuberculosis infections requires prompt and accurate diagnosis of susceptibility or resistance to the full panel of antibiotics in clinical use. Accordingly, alternative approaches to determining susceptibility of Mtb to a particular antibiotic are needed.

[0040] Compared with other bacteria such as E.coli, mycobacteria are notable for an RNA- polymerase that forms unstable open promoter complexes that appear more easily prone to regulation by repressors or stimulating factors. The Mtb RNA polymerase complex has a much shorter half-life (e.g., approximately 1-2 min) when compared to E. coli, implying a more rapid transcriptional response. At the same time, Mtb demonstrates longer mRNA halflife on average (e.g., approximately 9.5 min), with even the shortest half-lives longer than the average transcript half-life in E.coli. Hence, despite Mtb’s substantially slower growth in vitro (doubling time 16-22 hours) than E.coli (doubling time 20 minutes), its rapid transcriptional response offers a much more promising window to predict antibiotic effect than pDST.

[0041] The inventors have determined particular transcriptomic signatures comprising one or more upregulated genes, one or more downregulated genes, or both one or more upregulated genes and one or more downregulated genes for a plurality of antibiotics. The techniques developed by the inventors described herein (e.g., in the Examples) include using RNA expression data obtained from biological samples containing antibiotic- susceptible bacteria having been administered a particular antibiotic to test to train a logistic regression model. The logistic regression model can be used to determine susceptibility of a biological sampleto the particular antibiotic based on the presence of one or more differentially expressed genes in RNA expression data of the biological sample.

[0042] The techniques developed by the inventors improve existing technology for determining susceptibility of a biological sample to an antibiotic. For example, the techniques developed by the inventors determine susceptibility of a sample of Mtb more rapidly than existing phenotypic tests. For example, the techniques described herein can be performed within hours, while existing phenotypic tests can take multiple days to reach a determination of susceptibility. Rapid determination of susceptibility allows for faster treatment of a patient (e.g., with tuberculosis) which can provide for better patient outcomes.

[0043] The techniques described herein avoid the need to rely on growth-based phenotypic susceptibility testing alone to reach a determination of susceptibility. Conventionally, grow- based phenotypic susceptibility testing involves culturing a sample of Mtb for multiple days, which can present increased costs and biohazards. Accordingly, the techniques developed by the inventors present a cheaper and safer alternative to determining susceptibility of a biological sample to an antibiotic.

[0044] The techniques developed by the inventors also provide a more accurate method for determining susceptibility of a biological sample to an antibiotic than existing DNA-based testing, for example. As described herein, DNA-based testing has lower accuracy for certain key agents including BDQ, pyrazinamide (PZA), and fluoroquinolones due in part to the diversity of DNA resistance determinants, continued evolution of novel, rare and difficult to characterize mutations, or mutation-mutation interactions complicating resistance interpretation. In addition, drug resistance mutations can vary substantially by geography. This challenge is addressed by the techniques developed by the inventors which provide phenotypic information in addition to transcriptomic information, such that susceptibility of a sample of Mtb to an antibiotic can be determined even when drug resistance mutations to a certain antibiotic vary greatly or are difficult to characterize, and without knowledge of the precise mechanism of resistance of the sample. With the techniques developed by the inventors, the phenotypic response of the sample to the antibiotic can be used to determine susceptibility where genotypic information is less useful due to mutations in the genome.

[0045] According to some aspects, there is provided a method for determining whether a biological sample obtained from a patient contains mycobacterium tuberculosis complex (Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: using at least one computer hardware processor: obtaining RNA expression data (e.g., mRNA expression data, rRNA expression data, both mRNA expressiondata and rRNA expression data) previously-obtained by profiling (e.g., molecular profiling, sequencing) RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some Mtb genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic. In some embodiments, profiling the RNA comprises isolating and profiling the RNA.

[0046] In some embodiments, the method further comprises profiling the RNA obtained from the biological sample after exposure to the at least one antibiotic. RNA expression measurement may be performed using next generation sequencing, digital droplet polymerase chain reaction (PCR), quantitative reverse transcription polymerase chain reaction (RT- qPCR), and / or probe-based RNA sequencing (optionally, using nCounter NanoString technology).

[0047] In some embodiments, the method further comprises (1) culturing at least a portion of the biological sample to obtain a cultured sample; (2) exposing the cultured sample to the at least one antibiotic for a specified time period; and (3) extracting the RNA from the culture after the specified time period of exposure to the at least one antibiotic.

[0048] In some embodiments, the particular transcriptomic signature for the particular antibiotic comprises at least two (optionally, three, four, five, six, . . ., all of the upregulated Mtb genes for the particular antibiotic as specified in Table A3 and / or in one or more of Tables Bl -B7.

[0049] In some embodiments, the particular transcriptomic signature for the particular antibiotic comprises at least two (optionally, three, four, five, six, . . ., all of the downregulated Mtb genes for the particular antibiotic as specified in Table A3 and / or in one or more of Tables B 1-B7.

[0050] In some embodiments, the biological sample is exposed to the particular antibiotic for a specified time period, and wherein the particular transcriptomic signature for the particular antibiotic comprises at least two (optionally, three, four, five, six, . . ., all of the upregulated Mtb genes for the particular antibiotic as specified in Table A3 and / or in one or more of Tables B1-B7 with respect to the specified time period.

[0051] In some embodiments, the biological sample is exposed to the particular antibiotic for a specified time period, and wherein the particular transcriptomic signature for the particular antibiotic comprises at least two (optionally, three, four, five, six, . . ., all of the downregulated Mtb genes for the particular antibiotic as specified in Table A3 and / or in one or more of Tables B1-B7 with respect to the specified time period.

[0052] In some embodiments, at least one antibiotic includes at least one antibiotic selected from the list of antibiotics consisting of: Bedaquiline, Linezolid, Moxifloxacin, Pretomanid, and Pyrazinamide. In some embodiments, the at least one antibiotic includes at least one antibiotic selected from the list of antibiotics consisting of: Isoniazid, Pyrazinamide, Ethambutol, Moxifloxacin, Levofloxacin, Bedaquiline, Linezolid, Pretomanid, Clofazimine, Amikacin, Streptomycin, Delaminid, Ethionamide, Capreomycin, Kanamycin, and Rifapentine.

[0053] In some embodiments, the method further comprises, when it is determined that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic, providing a recommendation to administer the particular antibiotic. In some embodiments, the method further comprises, when it is determined that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic, administering the particular antibiotic to the patient. In some embodiments, the at least one antibiotic comprises multiple antibiotics, and when it is determined that the Mtb in the biological sample is susceptible to treatment by each of the multiple antibiotics, administering the multiple antibiotics to the patient.

[0054] According to some aspects, there is provided a system comprising: at least one computer hardware processor; and at least one non-transitory computer-readable storage medium storing instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform a method for determining whether a biological sample obtained from a patient contains mycobacterium tuberculosis complex (Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: obtaining RNA expression data for the biological sample previously-obtained by profiling (e.g., molecular profiling, sequencing) RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some Mtb genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomicsignature is present in the RNA expression data, determining that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic.

[0055] According to some aspects, there is provided at least one non-transitory computer- readable storage medium storing instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform a method for determining whether a biological sample obtained from a patient contains mycobacterium tuberculosis complex (Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: obtaining RNA expression data for the biological sample previously-obtained by profiling (e.g., molecular profiling, sequencing) RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some Mtb genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic.

[0056] According to some aspects, there is provided method for determining whether a biological sample obtained from a patient contains bacteria susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: using at least one computer hardware processor: obtaining RNA expression data (e.g., mRNA expression data, rRNA expression data, both mRNA expression data and rRNA expression data) for the biological sample previously-obtained by profiling (e.g., molecular profiling, sequencing) RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some bacterial genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the bacteria in the biological sample is susceptible to treatment by the particular antibiotic.

[0057] The aspects and embodiments described above, as well as additional aspects and embodiments, are described further below. These aspects and / or embodiments may be usedindividually, all together, or in any combination, as the application is not limited in this respect.

[0058] FIG. 1 illustrates an example process 100 for determining whether a biological sample obtained from a patient contains bacteria susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, in accordance with some embodiments of the technology described herein. In some embodiments, the bacteria may be Mycobacterium tuberculosis complex (Mtb). As described herein, Mtb has among the highest resistance rates of any pathogen globally. In addition, Mtb exhibits very slow growth in in vitro culture and the cost and biohazard of culturing Mtb is high. Accordingly, the inventors have recognized that applying the techniques described herein to determine whether a biological sample obtained from a patient contains Mtb susceptible to treatment by at least one antibiotic in a set of one or more antibiotics is beneficial.

[0059] The example process 100 begins at act 102 where RNA expression data for the biological sample is obtained. As described herein, RNA expression data comprises data obtained by profiling (e.g., molecular profiling, sequencing) RNA of a biological sample. In some embodiments, the RNA expression data comprises information (e.g., RNA expression levels) relating to the presence, absence, and / or relative amounts of particular genes. In some embodiments, the RNA expression data comprises RNA expression levels of genes listed in any of Tables A1-A3 of and / or Tables B1-B7. The RNA expression data may be obtained by profiling (e.g., molecular profiling, sequencing) all or a portion of the transcriptome. The RNA expression data may comprise information regarding all or a portion of the transcriptome.

[0060] The RNA expression data used in the example method for determining susceptibility of a biological sample to an antibiotic may be previously obtained RNA expression data obtained by profiling RNA from the biological sample after exposure to at least one antibiotic.

[0061] In some embodiments, the process optionally includes the act of profiling the RNA from the biological sample after exposure to the at least one antibiotic to obtain the RNA expression data that is used in the example method for determining susceptibility of the biological sample to an antibiotic. The skilled artisan will recognize that performing the profiling of the RNA of the biological sample is optional and is not always required to perform process 100. For example, in some instances, profiling has already been performed on the biological sample and processed to generate RNA expression data, and the RNA expression data exists prior to beginning process 100.

[0062] In some embodiments, profiling the RNA of a biological sample involves processing the biological sample using a sequencing platform to obtain RNA expression data which indicates RNA expression levels for one or more genes. The biological sample processed may be obtained from a subject having a particular disease, such as Tuberculosis. The biological sample may be obtained by performing a biopsy or by obtaining a blood sample, a salivary sample, or any other suitable biological sample from the subject, as described herein. The biological sample may include diseased tissue and / or healthy tissue.

[0063] In some embodiments, profiling the RNA may be performed using next generation sequencing (e.g., RNAseq). In some embodiments, profiling the RNA may be performed using digital droplet polymerase chain reaction (PCR). In some embodiments, profiling the RNA may be performed using quantitative reverse transcription polymerase chain reaction (RT-qPCR) In some embodiments, profiling the RNA is performed using probe-based RNA sequencing. For example, the probe-based RNA sequencing may be performed using nCounter NanoString technology, in some embodiments.

[0064] In some embodiments, the sequencing platform includes any suitable instrument and / or system configured to perform nucleic acid profiling , as aspects of the technology described herein are not limited to any particular type of sequencing system. In some embodiments, the biological sample may be prepared according to manufacturer’s protocols associated with the sequencing platform. In some embodiments, the biological sample may be prepared according to any suitable protocol, as embodiments of the technology described herein are not limited to any particular preparation protocol. As one illustrative example, in some embodiments, the data obtained by profiling the RNA of the biological sample may comprise bulk sequencing data. In some embodiments, bulk sequencing data comprises at least 1 million reads, at least 5 million reads, at least 10 million reads, at least 20 million reads, at least 50 million reads, or at least 100 million reads. In some embodiments, bulk sequencing data comprises between 1 million reads and 5 million reads, 3 million reads and 10 million reads, 5 million reads and 20 million reads, 10 million reads and 50 million reads, 30 million reads and 100 million reads, or 1 million reads and 100 million reads (or any number of reads including, and between). In some embodiments, the sequencing data comprises single cell RNA sequencing (scRNA-seq) data, or next generation sequencing (NGS) data.

[0065] In some embodiments, RNA expression data can include sequence data generated by a sequencing protocol (e.g., the series of nucleotides in a nucleic acid molecule identified by next-generation sequencing, sanger sequencing, etc.) as well as information contained therein(e.g., information indicative of source, tissue type, populations of cell types, etc.) which may also be considered information that can be inferred or determined from the sequence data. In some embodiments, expression data can include information included in a FASTA file, a description and / or quality scores included in a FASTQ file, an aligned position included in a BAM file, and / or any other suitable information obtained from any suitable file.

[0066] In some embodiments, prior to obtaining the RNA expression data, the process may include preparing the biological sample in one or more ways so as to enable obtaining the RNA expression data. For example, the process 100 may include an initial act of culturing at least a portion of the biological sample to obtain a cultured sample.

[0067] For example, at least a portion of a biological sample that has been obtained from a subject can be cultured to obtain the cultured sample. In some embodiments, a subject is a mammal (e.g., a human, a mouse, a cat, a rat, a dog, a horse, a hamster, a cow, a pig, a nonhuman primate, or other animal). A subject may be a human. The subject may be an adult human (e.g., of 18 years of age or older) or a child (e.g., less than 18 years of age). The human may be or may have been diagnosed with at least one condition, such as tuberculosis.

[0068] The biological sample may be any type of biological sample including, for example, a biological sample of a bodily fluid (e.g., sputum, blood, urine or cerebrospinal fluid), one or more cells (e.g., from a scraping or brushing such as a cheek swab or tracheal brushing), a piece of tissue (cheek tissue, muscle tissue, lung tissue, heart tissue, brain tissue, or skin tissue), or some or all of an organ (e.g., brain, lung, liver, bladder, kidney, pancreas, intestines, or muscle), or other types of biological samples (e.g., feces or hair).

[0069] Any of the biological samples described herein may be obtained from the subject using any known technique. See, for example, the following publications on collecting, processing, and storing biological samples, each of which are incorporated by reference herein in its entirety: Biospecimens and biorepositories: from afterthought to science by Vaught et al. (Cancer Epidemiol Biomarkers Prev. 2012 Feb;21(2):253-5), and Biological sample collection, processing, storage and information management by Vaught and Henderson (IARC Sci Publ. 2011;(163):23-42).

[0070] Methods of the present disclosure encompass obtaining one or more biological samples from a subject for analysis. In some embodiments, one biological sample is collected from a subject for analysis. In some embodiments, more than one (e.g., 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or more) biological samples are collected from a subject for analysis. In some embodiments, one biological sample from a subject will be analyzed. In some embodiments, more than one (e.g., 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,15, 16, 17, 18, 19, 20, or more) biological samples may be analyzed. If more than one biological sample from a subject is analyzed, the biological samples may be procured at the same time (e.g., more than one biological sample may be taken in the same procedure), or the biological samples may be taken at different times (e.g., during a different procedure including a procedure 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 days; 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 weeks; 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 months, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 years, or 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 decades after a first procedure).

[0071] In some embodiments, culturing the biological sample to obtain a cultured sample comprises preparing a vessel for culturing the sample. For example, the process may include preparing a surface of a vessel for receiving the biological sample. The vessel may comprise, for example, a petri dish. Preparing the surface of the vessel may comprise treating the surface with a substance, such as an agar. The substance may be introduced to the vessel in a liquid form, and subsequently cooled to form a gel on the surface of the vessel. The substance may comprise ingredients for facilitating growth of the biological sample.

[0072] Culturing the biological sample may further comprise introducing (e.g., seeding) the biological sample containing the bacteria into the vessel having the treated surface to obtain the cultured sample. In some embodiments, the biological sample may be grown in the treated vessel for a period of time before administering an antibiotic to the biological sample. The biological sample may be grown for the period of time according to a protocol that allows the bacteria in the biological sample to form a bacterial colony in the vessel.

[0073] The process 100 may optionally include exposing the cultured sample to at least one antibiotic for a specified time period. For example, exposing the cultured sample to the at least one antibiotic may comprise administering the at least one antibiotic whose susceptibility is desired to be determined to the cultured sample. The antibiotic may be a dissolved antibiotic in solution form. Aspects of administering an antibiotic are described herein.

[0074] The at least one antibiotic may be a particular antibiotic intended to induce a stress response in susceptible forms of the bacteria. The at least one antibiotic may be one or more of the antibiotics identified in any of Tables A1-A3 of and / or Tables B1-B7. For example, the at least one antibiotic may comprise at least one antibiotic selected from the list of antibiotics consisting of: Isoniazid, Rifampicin, Pyrazinamide, Ethambutol, Moxifloxacin, Levofloxacin, Bedaquiline, Linezolid, Pretomanid, Clofazimine, Amikacin, Streptomycin, Delaminid, Ethionamide, Capreomycin, Kanamycin, and Rifapentine.

[0075] The at least one antibiotic may be administered to the cultured sample for a specified time period. He specified time period may be a time period long enough to induce a stress response in susceptible forms of the bacteria. Administering the at least one antibiotic to the cultured sample for the specified time period can comprise injecting the at least one antibiotic into the cultured sample and allowing the cultured sample to incubate with the injected at least one antibiotic for the specified time period. The specified time period may be any of the specified time periods identified in any of Tables A1-A3 of and / or Tables B1-B7. For example, the specified time period be one of less than three hours, between three and six hours, less than six hours, greater than six hours, between six and twenty four hours, or greater than twenty four hours.

[0076] The process may subsequently include extracting the RNA from the cultured sample after the specified time period of exposure to the at least one antibiotic. In some embodiments, extracting the RNA may comprise lysing the sample to break down binding reagents of the bacterial cells in the cultured sample such that the RNA of the bacterial cells can be isolated and sequenced. Extraction of the RNA may be performed according to protocols known in the art.

[0077] The process may subsequently include profiling the RNA extracted from the cultured sample to obtain profiling data. Profiling the RNA may be performed according to any of the techniques described herein. RNA expression data can be derived from the RNA profiling data, according to any of the techniques described herein, for example.

[0078] Subsequent to obtaining the RNA expression data at act 102, the process 100 proceeds to act 104, it is determined whether a particular transcriptomic signature for a particular antibiotic is present in the RNA expression data.

[0079] As described herein, the transcriptomic signature may comprise one or more differentially expressed genes. For example, the transcriptomic signature may comprise one or more genes that are upregulated in an antibiotic- susceptible bacteria in response to administration of the particular antibiotic and / or one or more genes that are downregulated in an antibiotic- susceptible bacteria in response to administration of the particular antibiotic.

[0080] In particular, as described herein, when an antibiotic is administered to bacteria of a biological sample that is susceptible to the particular antibiotic, the bacteria undergoes a stress response. The stress response can be observed in the RNA expression data of the bacteria, for example, in the upregulation and / or downregulation of certain genes. For a particular antibiotic to which the bacteria is susceptible, particular genes of the bacteria are upregulated, and particular genes are downregulated. By observing the RNA expression dataof a new biological sample comprising bacteria for which susceptibility is desired to be determined, it may be determined whether the particular genes that would be upregulated if the bacteria is susceptible to the particular antibiotic are in fact upregulated and whether the particular genes that would be downregulated if the bacteria is susceptible to the particular antibiotic are in fact downregulated. If one or both of these conditions are met, it may be determined that the bacteria of the new biological sample is susceptible to the particular antibiotic.

[0081] In some embodiments, each antibiotic in the set of antibiotics may have its own particular transcriptomic signature with one or more upregulated genes and / or one or more downregulated genes. The inventors have identified transcriptomic signatures for a set of antibiotics for treating MtB, as described herein and shown in any of Tables Al -A3 of and / or Tables B1-B7. The transcriptomic signature for a particular antibiotic may comprise at least two (optionally, three, four, five, six . . . all) of the upregulated MtB genes for the particular antibiotic as specified in any of Tables A1-A3 and / or Tables B1-B7. In some embodiments, the transcriptomic signature for a particular antibiotic may comprise at least two (optionally, three, four, five, six . . . all) of the upregulated MtB genes for the particular antibiotic as specified in any of Tables A1-A3 and / or Tables B 1-B7.

[0082] Act 104 may be repeated for each antibiotic in the set of one or more antibiotics. The set of one or more antibiotics may include a single antibiotic or may include multiple antibiotics. Any suitable number of antibiotics may be tested.

[0083] When it is determined, at 104, that the particular transcriptomic signature for the particular antibiotic is not present in the RNA expression data obtained at act 102, the process X00 moves through the “NO” branch to act 106B, where it is determined that the bacteria in the biological sample is not susceptible to treatment by the particular antibiotic. The determination at act 106B can be helpful in developing a treatment plan for a patient, as it may be determined not to administer the particular antibiotic to a patient based on the determination at act 106B that the bacteria in the biological sample is not susceptible to the particular antibiotic.

[0084] When it is determined, at act 104, that the particular transcriptomic signature for the particular antibiotic is present in the RNA expression data obtained at act 102, the process 100 moves through the “YES” branch to act 106A, where it is determined that the bacteria in the biological sample is susceptible to treatment by the particular antibiotic. The determination at act 106 A can be helpful in developing a treatment plan for a patient, as it may be determined to administer the particular antibiotic to a patient based on thedetermination at act 106A that the bacteria in the biological sample is susceptible to the particular antibiotic.

[0085] In some embodiments, the process 100 further comprises administering the particular antibiotic to the patient. For example, when it is determined at act 106 A that the bacteria in the biological sample is susceptible to treatment by the particular antibiotic, the process may proceed to administer the particular antibiotic to the patient from which the biological sample was obtained. In some embodiments, “administering” or “administration” means providing the antibiotic to the patient in a manner that is pharmacologically useful.

[0086] Administering the particular antibiotic to the patient comprises delivering one or more antibiotics (e.g., one or more antibiotics to which a bacteria, such as Mtb, is susceptible to) to a patient. The one or more antibiotics may be administered by any suitable route. For example, an effective amount of the antibiotic can be administered to a patient by any mode that delivers the agent to the desired tissue. In some embodiments, the one or more antibiotics may be administered orally or topically. Other suitable routes of administration include but are not limited to vaginal (e.g., intravaginal administration), rectal, parenteral, intravenous, intraperitoneal, intranasal, intramuscular, sublingual, intratracheal, inhalation, and subcutaneous. Systemic routes include oral and parenteral. For oral administration, an antibiotic can be administered in the form of a tablet, pill, dragee, capsule, liquid, mouthwash, gel, syrup, slurry, suspension, and the like.

[0087] In some embodiments, an antibiotic comprises one or more pharmaceutically acceptable carriers. As used herein, the term “pharmaceutically acceptable carrier” is intended to include any and all solvents, dispersion media, coatings, isotonic and absorption delaying agents, and the like, compatible with the antibiotic administration. The use of such media and agents for pharmaceutically active substances is well known in the art. Except insofar as any conventional media or agent is incompatible with the active compound of an antibiotic, use thereof in the antibiotics is contemplated. Supplementary active compounds can also be incorporated into the antibiotics. Antibiotics can be prepared as described below. The active ingredients may be admixed or compounded with any conventional, pharmaceutically acceptable carrier or excipient. The antibiotic may be sterile.

[0088] An antibiotic is said to be a “pharmaceutically acceptable carrier” if its administration can be tolerated by a recipient patient. Sterile phosphate-buffered saline is one example of a pharmaceutically acceptable carrier. Other suitable carriers are well-known in the art. See, for example, REMINGTON'S PHARMACEUTICAL SCIENCES, 18th Ed. (1990).

[0089] An antibiotic may comprise an effective amount of a bioactive agent. An effective amount of a bioactive agent is an amount sufficient to inhibit growth of one or more bacteria (e.g., Mtb) and / or destroy existing bacteria in a patient. In some embodiments, an effective amount of a bioactive agent is an amount greater than the minimum inhibitory concentration (MIC) of the bioactive agent with respect to a particular target bacteria. The effective amount to be included in an antibiotic depends, in each case, upon several factors, e.g., the type, size and condition of the patient to be treated, the intended mode of administration, the capacity of the patient to incorporate the intended dosage form, etc. One of ordinary skill in the art would be able to determine empirically an appropriate effective amount. The effective amount for any particular application can vary depending on such factors as the disease or condition being treated, the particular antibiotic being administered, the size of the subject, or the severity of the disease or condition. One of ordinary skill in the art can empirically determine the effective amount of a particular bioactive agent, amino acid, and / or other therapeutic agent without necessitating undue experimentation.

[0090] An antibiotic may be administered alone (e.g., as a pill or powder, in saline or buffer, etc.) or using any delivery vehicle known in the art.

[0091] The antibiotics also may comprise suitable solid or gel phase carriers or excipients. Examples of such carriers or excipients include but are not limited to calcium carbonate, calcium phosphate, various sugars, starches, cellulose derivatives, gelatin, and polymers such as polyethylene glycols.

[0092] The antibiotics also include granules, powders, tablets, coated tablets, (micro)capsules, suppositories, syrups, emulsions, suspensions, creams, drops or preparations with protracted release of active compounds, in whose preparation excipients and additives and / or auxiliaries such as disintegrants, binders, coating agents, swelling agents, lubricants, flavorings, sweeteners or solubilizers are customarily used. The antibiotics are suitable for use in a variety of drug delivery systems. For a brief review of methods for drug delivery, see Langer R (1990) Science 249:1527-1533, which is incorporated herein by reference.

[0093] Other delivery systems can include time release, delayed release or sustained release delivery systems. Such systems can avoid repeated administrations of the antibiotics, increasing convenience to the patient and the physician. Many types of release delivery systems are available and known to those of ordinary skill in the art. They include polymer base systems such as poly(lactide glycolide), copolyoxalates, polycaprolactones, polyesteramides, poly orthoesters, polyhydroxybutyric acid, and polyanhydrides.

[0094] In some embodiments, an antibiotic is administered to a patient no more than once per calendar day (e.g., a 24-hour period). In some embodiments, an antibiotic is administered to a patient no more than once per 2, 3, 4, 5, 6, or 7 calendar days. In some embodiments, an antibiotic is administered to a patient no more than once per calendar week (e.g., 7 calendar days). In some embodiments, an antibiotic is administered to a patient no more than biweekly (e.g., once in a two calendar week period). In some embodiments, an antibiotic is administered to a patient no more than once per calendar month (e.g., once in 30 calendar days). In some embodiments, an antibiotic is administered to a patient no more than once per six calendar months. In some embodiments, an antibiotic is administered to a patient no more than once per calendar year (e.g., 365 days or 366 days in a leap year).

[0095] Preparation of compositions for administration to a subject are known in the art. For example, a dosage may be dissolved in 1 ml of isotonic NaCl solution and either added to 1000 ml of hypodermoclysis fluid or injected at the proposed site of infusion, (see for example, "Remington's Pharmaceutical Sciences" 15th Edition, pages 1035-1038 and 1570- 1580). Some variation in dosage will necessarily occur depending on the condition of the patient being treated. The person responsible for administration will, in any event, determine the appropriate dose for the individual subject. Moreover, for human administration, preparations should meet sterility, pyrogenicity, and the general safety and purity standards as required by, e.g., FDA Office of Biologies standards.

[0096] In some embodiments, a patient “having” a disease such as tuberculosis has been diagnosed as having the disease by a medical professional (e.g., a medical doctor). In some embodiments, the diagnosis has been confirmed by a laboratory assay. In some embodiments, a subject “suspected of having” a disease has been diagnosed as having the disease by a medical professional but the diagnosis has not been confirmed by a laboratory assay.

[0097] To "treat" a disease as the term is used herein, means to reduce the frequency or severity of at least one sign or symptom of a disease or disorder experienced by a subject. The antibiotics described above or elsewhere herein are typically administered to a subject in an effective amount, that is, an amount capable of producing a desirable result. The desirable result will depend upon the active agent being administered. A therapeutically acceptable amount may be an amount that is capable of treating a disease, e.g., tuberculosis. As is well known in the medical and veterinary arts, dosage for any one subject depends on many factors, including the subject's size, body surface area, age, the particular composition to be administered, the active ingredient(s) in the composition, time and route of administration, general health, and other drugs being administered concurrently.

[0098] In some embodiments, the at least one antibiotic may comprise multiple antibiotics. When it is determined that the bacteria (e.g., Mtb) in the biological sample is susceptible to treatment by each of the multiple antibiotics, at least one of the multiple antibiotics (e.g., all, one, more than one) may be administered to the patient.

[0099] As described herein, the process 100 may be performed for each antibiotic in a set of one or more antibiotics. In some embodiments, the process 100 may be performed until the bacteria in the biological sample is determined to be susceptible to at least one antibiotic in the set of one or more antibiotics. In some embodiments, the process 100 may be performed for each antibiotic in the set of one or more antibiotics and a set of candidate antibiotics may be formed comprising each antibiotic in the set of one or more antibiotics to which the bacteria in the biological sample is susceptible to. In some embodiments, a treatment may be determined based on the set of candidate antibiotics. For example, the treatment may comprise administration of one or more of the set of candidate antibiotics to the patient. The treatment may be based on a degree of susceptibility of the antibiotic(s) in the set of candidate antibiotics.

[0100] As described herein, the example process 100 for determining whether a biological sample obtained from a patient contains bacteria susceptible to treatment by at least one antibiotic comprises determining whether RNA expression data contains a transcriptomic signature. The inventors have developed techniques for determining transcriptomic signatures to which RNA expression data of a new biological sample can be compared to. For example, as shown in the examples described herein, RNA expression data obtained from biological samples containing antibiotic-susceptible bacteria having been administered a particular antibiotic were used to train a logistic regression model. The trained logistic regression model can be used to determine susceptibility of a new biological sample to the particular antibiotic based on the presence of one or more differentially expressed genes in RNA expression data of the biological sample.

[0101] The inventors have recognized that RNA expression data of a sample of bacteria can vary based on one or more factors. For example, different antibiotics can produce different stress responses in antibiotic-susceptible bacteria, and therefore can produce different upregulated and / or downregulated genes in response to administration of the particular antibiotic. Accordingly, transcriptomic signatures may be categorized for a particular antibiotic. The inventors have further recognized that different periods of time in which a particular antibiotic is administered to a sample of bacteria can produce different upregulated and / or downregulated genes (e.g., as the stress response of the bacteria evolves over time).Accordingly, transcriptomic signatures may be additionally or alternatively categorized by administration time.

[0102] Tables A1-A3 and Tables B1-B7 provide transcriptomic signatures for a plurality of antibiotics intended to treat tuberculosis. For each antibiotic, Tables A1-A3 and Tables B1-B7 provide a list of top ten upregulated genes and top ten downregulated genes. Tables Al -A3 and Tables B1-B7 further provide such transcriptomic signatures for a plurality of administration time periods.

[0103] The example processes described herein may be performed using at least one computer hardware processor. For example, aspects of the technology described herein include at least one non-transitory computer-readable storage medium storing instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform any of the processes described herein for determining whether a biological sample obtained from a patient contains bacteria (e.g., Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics. Aspects of the technology described herein further include a system comprising the at least one computer hardware processor and the at least one non-transitory computer-readable storage medium described herein.

[0104] An illustrative implementation of a computer system 200 that may be used in connection with any of the embodiments of the technology described herein (e.g., such as the process of FIG. 1, for example) is shown in FIG. 2. The computer system 200 includes one or more processors 210 and one or more articles of manufacture that comprise non-transitory computer-readable storage media (e.g., memory 220 and one or more non-volatile storage media 230). The processor 210 may control writing data to and reading data from the memory 220 and the non-volatile storage device 230 in any suitable manner, as the aspects of the technology described herein are not limited to any particular techniques for writing or reading data. To perform any of the functionality described herein, the processor 210 may execute one or more processor-executable instructions stored in one or more non-transitory computer-readable storage media (e.g., the memory 220), which may serve as non-transitory computer-readable storage media storing processor-executable instructions for execution by the processor 210.

[0105] Computing device 200 may also include a network input / output (I / O) interface 240 via which the computing device may communicate with other computing devices (e.g., over a network), and may also include one or more user I / O interfaces 250, via which the computing device may provide output to and receive input from a user. The user I / O interfaces mayinclude devices such as a keyboard, a mouse, a microphone, a display device (e.g., a monitor or touch screen), speakers, a camera, and / or various other types of I / O devices.

[0106] The above-described embodiments can be implemented in any of numerous ways. For example, the embodiments may be implemented using hardware, software, or a combination thereof. When implemented in software, the software code can be executed on any suitable processor (e.g., a microprocessor) or collection of processors, whether provided in a single computing device or distributed among multiple computing devices. It should be appreciated that any component or collection of components that perform the functions described above can be generically considered as one or more controllers that control the above-discussed functions. The one or more controllers can be implemented in numerous ways, such as with dedicated hardware, or with general purpose hardware (e.g., one or more processors) that is programmed using microcode or software to perform the functions recited above.

[0107] In this respect, it should be appreciated that one implementation of the embodiments described herein comprises at least one computer-readable storage medium (e.g., RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or other tangible, non-transitory computer-readable storage medium) encoded with a computer program (i.e., a plurality of executable instructions) that, when executed on one or more processors, performs the above-discussed functions of one or more embodiments. The computer-readable medium may be transportable such that the program stored thereon can be loaded onto any computing device to implement aspects of the techniques discussed herein. In addition, it should be appreciated that the reference to a computer program which, when executed, performs any of the above-discussed functions, is not limited to an application program running on a host computer. Rather, the terms computer program and software are used herein in a generic sense to reference any type of computer code (e.g., application software, firmware, microcode, or any other form of computer instruction) that can be employed to program one or more processors to implement aspects of the techniques discussed herein.

[0108] The foregoing description of implementations provides illustration and description but is not intended to be exhaustive or to limit the implementations to the precise form disclosed. Modifications and variations are possible in light of the above teachings or may be acquired from practice of the implementations. In other implementations the methods depicted in these figures may include fewer operations, different operations, differentlyordered operations, and / or additional operations. Further, non-dependent blocks may be performed in parallel.

[0109] It will be apparent that example aspects, as described above, may be implemented in many different forms of software, firmware, and hardware in the implementations illustrated in the figures. Further, certain portions of the implementations may be implemented as a “module” that performs one or more functions. This module may include hardware, such as a processor, an application- specific integrated circuit (ASIC), or a field-programmable gate array (FPGA), or a combination of hardware and software.

[0110] Having thus described several aspects and embodiments of the technology set forth in the disclosure, it is to be appreciated that various alterations, modifications, and improvements will readily occur to those skilled in the art. Such alterations, modifications, and improvements are intended to be within the spirit and scope of the technology described herein. For example, those of ordinary skill in the art will readily envision a variety of other means and / or structures for performing the function and / or obtaining the results and / or one or more of the advantages described herein, and each of such variations and / or modifications is deemed to be within the scope of the embodiments described herein. Those skilled in the art will recognize or be able to ascertain using no more than routine experimentation many equivalents to the specific embodiments described herein. It is, therefore, to be understood that the foregoing embodiments are presented by way of example only and that, within the scope of the appended claims and equivalents thereto, inventive embodiments may be practiced otherwise than as specifically described. In addition, any combination of two or more features, systems, articles, materials, kits, and / or methods described herein, if such features, systems, articles, materials, kits, and / or methods are not mutually inconsistent, is included within the scope of the present disclosure.

[0111] The above-described embodiments can be implemented in any of numerous ways. One or more aspects and embodiments of the present disclosure involving the performance of processes or methods may utilize program instructions executable by a device (e.g., a computer, a processor, or other device) to perform, or control performance of, the processes or methods. In this respect, various inventive concepts may be embodied as a computer readable storage medium (or multiple computer readable storage media) (e.g., a computer memory, one or more floppy discs, compact discs, optical discs, magnetic tapes, flash memories, circuit configurations in Field Programmable Gate Arrays or other semiconductor devices, or other tangible computer storage medium) encoded with one or more programs that, when executed on one or more computers or other processors, perform methods thatimplement one or more of the various embodiments described above. The computer readable medium or media can be transportable, such that the program or programs stored thereon can be loaded onto one or more different computers or other processors to implement various ones of the aspects described above. In some embodiments, computer readable media may be non-transitory media.

[0112] The terms “program” or “software” are used herein in a generic sense to refer to any type of computer code or set of computer-executable instructions that can be employed to program a computer or other processor to implement various aspects as described above. Additionally, it should be appreciated that according to one aspect, one or more computer programs that when executed perform methods of the present disclosure need not reside on a single computer or processor, but may be distributed in a modular fashion among a number of different computers or processors to implement various aspects of the present disclosure.

[0113] Computer-executable instructions may be in many forms, such as program modules, executed by one or more computers or other devices. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Typically the functionality of the program modules may be combined or distributed as desired in various embodiments.

[0114] Also, data structures may be stored in computer-readable media in any suitable form. For simplicity of illustration, data structures may be shown to have fields that are related through location in the data structure. Such relationships may likewise be achieved by assigning storage for the fields with locations in a computer-readable medium that convey relationship between the fields. However, any suitable mechanism may be used to establish a relationship between information in fields of a data structure, including through the use of pointers, tags or other mechanisms that establish relationship between data elements.

[0115] When implemented in software, the software code can be executed on any suitable processor or collection of processors, whether provided in a single computer or distributed among multiple computers.

[0116] Further, it should be appreciated that a computer may be embodied in any of a number of forms, such as a rack-mounted computer, a desktop computer, a laptop computer, or a tablet computer, as non-limiting examples. Additionally, a computer may be embedded in a device not generally regarded as a computer but with suitable processing capabilities, including a Personal Digital Assistant (PDA), a smartphone, a tablet, or any other suitable portable or fixed electronic device.Z1

[0117] Also, a computer may have one or more input and output devices. These devices can be used, among other things, to present a user interface. Examples of output devices that can be used to provide a user interface include printers or display screens for visual presentation of output and speakers or other sound generating devices for audible presentation of output. Examples of input devices that can be used for a user interface include keyboards, and pointing devices, such as mice, touch pads, and digitizing tablets. As another example, a computer may receive input information through speech recognition or in other audible formats.

[0118] Such computers may be interconnected by one or more networks in any suitable form, including a local area network or a wide area network, such as an enterprise network, and intelligent network (IN) or the Internet. Such networks may be based on any suitable technology and may operate according to any suitable protocol and may include wireless networks, wired networks or fiber optic networks.

[0119] Also, as described, some aspects may be embodied as one or more methods. The acts performed as part of the method may be ordered in any suitable way. Accordingly, embodiments may be constructed in which acts are performed in an order different than illustrated, which may include performing some acts simultaneously, even though shown as sequential acts in illustrative embodiments.

[0120] Various aspects of the present invention may be used alone, in combination, or in a variety of arrangements not specifically discussed in the embodiments described in the foregoing and is therefore not limited in its application to the details and arrangement of components set forth in the foregoing description or illustrated in the drawings. For example, aspects described in one embodiment may be combined in any manner with aspects described in other embodiments.

[0121] Various events / acts are described herein as occurring or being performed at a specified time. One of ordinary skill in the art would understand that such events / acts may occur or be performed at approximately the specified time.

[0122] Use of ordinal terms such as “first,” “second,” “third,” etc., in the claims to modify a claim element does not by itself connote any priority, precedence, or order of one claim element over another or the temporal order in which acts of a method are performed, but are used merely as labels to distinguish one claim element having a certain name from another element having a same name (but for use of the ordinal term) to distinguish the claim elements.

[0123] Also, the phraseology and terminology used herein is for the purpose of description and should not be regarded as limiting. The use of “including,” “comprising,” or “having,” “containing,” “involving,” and variations thereof herein, is meant to encompass the items listed thereafter and equivalents thereof as well as additional items.

[0124] The terms “substantially”, “approximately”, and “about” may be used to mean within ±20% of a target value in some embodiments, within ±10% of a target value in some embodiments, within ±5% of a target value in some embodiments, within ±2% of a target value in some embodiments. The terms “approximately” and “about” may include the target value.

[0125] Having thus described several aspects of at least one embodiment of this invention, it is to be appreciated that various alterations, modifications, and improvements will readily occur to those skilled in the art.

[0126] Such alterations, modifications, and improvements are intended to be part of this disclosure, and are intended to be within the spirit and scope of the invention. Further, though advantages of the present invention are indicated, it should be appreciated that not every embodiment of the invention will include every described advantage. Some embodiments may not implement any features described as advantageous herein and in some instances. Accordingly, the description herein and drawings are by way of example only.EXAMPLES

[0127] Example 1: An RNA-based assay for the diagnosis of antibiotic resistance in M. tuberculosis

[0128] Summary

[0129] Mycobacterium tuberculosis complex (Mtb) has among the highest resistance rates of any pathogen globally, but Mtb resistance remains the most challenging to diagnose due to its very slow growth in in vitro culture and culture’s inordinate cost and biohazard. Rapid and accurate alternative diagnostic approaches are urgently needed. DNA based tests are available for detecting resistance to a few drugs but are limited by their need for extensive knowledge of DNA resistance determinants for interpretation, and concerns about limit of detection at the point-of-care. For novel antibiotics entering clinical use, genetic resistance determinants are particularly poorly understood and may be too dispersed across the genome to lend themselves to targeted sequencing and delays in understanding mechanisms may miss an opportunity to avoid more widespread resistance. A work around is the development of afunctional or phenotypic assay that circumvents the need to target specific genetic mechanisms, however there is no such rapid assay currently commercially available largely because these assays have traditionally relied on observed bacterial growth in vitro under antibiotic pressure. Leveraging considerable preliminary data, the development of an innovative ‘functional’ RNA-based assay for resistance diagnosis in Mtb is proposed. It may be anticipated that increasing sensitivity of resistance detection, expanding the numbers of drugs to which resistance can be detected, while significantly shortening time-to-result. Specifically in this early phase development proposal key assay parameters that maximize the Mtb RNA susceptibility signal to five key agents in multi-drug resistant tuberculosis treatment (bedaquiline, pretomanid, linezolid, moxifloxacin and pyrazinamide) will be identified using RNA sequencing in a factorial design of experiments assessing drug exposure dose and exposure duration. In addition, the effect of key clinical variables including genetic resistance mechanism, background lineage (Aim 1) will be studied. In Aim 2, NanoString technologies will be utilized to develop hybridization probe sets (called nanosensors for short) to target antibiotic responsive and control genes in two iterative phases. This will be followed by assay performance assessment in vitro, and on sputum from participants newly diagnosed with rifampicin resistant TB before and after incubation in culture media. This work will build a strong foundation for an innovative resistance diagnostic that addresses an unmet need.

[0130] Narrative

[0131] Successful treatment of tuberculosis infections requires the prompt and accurate diagnosis of resistance to the full panel of antibiotics in clinical use. Here a novel approach to diagnose antibiotic resistance to bedaquiline, pretomanid, linezolid, pyrazinamide, and moxifloxacin using RNA hybridization detecting a transcriptomic signature of antibiotic response is proposed. This approach circumvents the delays and biohazard of bacterial growth-based tests and the difficulties in interpretation of DNA based methods.

[0132] Specific Aims

[0133] Treatment success rates for multi-drug resistant (MDR) tuberculosis (TB) remain poor at 59% of treated cases globally in 20181. An important driver of poor treatment outcome is the limited global access to comprehensive drug susceptibility testing (DST) needed to guide regimen composition in MDR-TB therapy. Only -50% of TB patients get tested for fluoroquinolone resistance globally, and although bedaquiline (BDQ) is now in use in over 109 countries there is currently no accessible assay to confirm BDQ susceptibility2. An urgent need exists for rapid diagnostics that accurately capture drug resistance (DR) to acomprehensive panel of anti-TB antibiotics without the cost, biohazard and delays incurred by conventional growth-based drug susceptibility tests. DNA-based tests including targeted and whole genome sequencing (WGS) have high accuracy for some first-line antibiotics such as rifampicin (RIF) but lower accuracy for other key agents including BDQ, pyrazinamide (PZA), and fluoroquinolones. This challenge is due in part to the diversity of DNA resistance determinants, continued evolution of novel, rare and difficult to characterize mutations, or mutation-mutation interactions complicating resistance interpretation3. Further DR mutations can vary substantially by geography4. The lack of an accessible diagnostic test for novel agents like BDQ is an urgent public health need that risks compromising the promised efficacy of regimens like BPaLM (BDQ, pretomanid [PA824], linezolid [LZD], and moxifloxacin [MXF]) that have taken decades to develop. Novel, rapid phenotypic resistance diagnostics that do not require knowledge of resistance mechanisms can address this challenge.

[0134] The early development of an innovative phenotypic resistance assay based on detecting an RNA signature of antibiotic response (AR) rapidly after in vitro Mtb exposure to antibiotic is proposed. Rationale: Antibiotic susceptible Mtb rapidly develops cellular stress after antibiotic exposure, including drug mechanism- specific and more broad responses such as heat shock response and decreases in rRNA synthesis for bactericidal agents5. These transcriptional changes can be captured within a few hours from antibiotic exposure6and demonstrate high discriminatory power in distinguishing antibiotic susceptibility from resistance in preliminary analysis. This approach can expedite resistance diagnosis to a few hours after early culture growth and has potential for use on direct sputum prior to in vitro culture. Non-ribosomal mRNA signatures that predict susceptibility to several Mtb antibiotics5have been developed, and rRNA signatures detectable directly on sputum to support treatment monitoring7. Detecting AR transcriptionally offers several advantages including: 1- independence from genetic mechanism, 2- lower limit of detection as RNA abundance is generally higher than DNA in the cell, 3- concomitant evaluation of Mtb viability, and 4- leveraging RNA detection technologies developed for CO VID- 19 diagnosis. Building on the combined expertise in mycobacterial molecular biology, RNA diagnostics, bioinformatics, and clinical field research, parameters of detection of the transcriptional AR were optimized, developed and benchmarked in an early prototype NanoString assay to diagnose DR in prospectively sampled Mtb clinical isolates.

[0135] Aim 1: Optimize detection of transcriptional antibiotic response to five drugs

[0136] BPaLM and PZA will be targeted due to their importance in TB treatment and due to challenges in developing accurate DNA-based diagnostics to these agents. In preliminary analysis, the RNA diagnostic approach for isoniazid was confirmed to be accurate, streptomycin and ciprofloxacin resistance5, and derived mRNA AR signatures for BPaLM and PZA. It may be hypothesized that combining rRNA and mRNA response with generate a more robust signal of AR and hypothesize that in vitro antibiotic concentrations higher than the critical cut-point for growth-based assays can shorten the time to detect transcriptional AR. RNA sequencing and rRNA targeted RT-qPCR and a factorial design of Mtb exposure in vitro to evaluate drug concentrations and exposure times that maximize the transcriptional AR signal may be utilized. Generalizability of the signature across Mtb genetic lineage and genetic mechanism of resistance using stored specimens from the South African biobank (n=40 clinical isolates) was confirmed.

[0137] Aim 2: Develop an early phase Mtb assay for susceptibility to five drugs

[0138] It may be hypothesized that the measurement of transcript level for <15 genes after antibiotic exposure can diagnose resistance to the five TB drugs with high sensitivity and specificity. A prototype diagnostic workflow based on NanoString RNA probe hybridization in two phases will be developed. Probes will be designed based on published5,7and independent preliminary data previously generated on transcriptional AR and may be refined in Aim 1. Preliminary performance (sensitivity and specificity) was evaluated and limit of detection (LOD) on archived culture (n=80) and performance on prospectively collected sputum directly and after early positive culture. 30 RIF resistant (RR) TB patients from two clinics and a hospital in Worcester, South Africa where there is an established site to donate sputum prior to treatment initiation. LOD will be benchmarked against RT-qPCR. Diagnostic sensitivity and specificity will be compared with Mtb WGS and growth-based phenotypic MIC.

[0139] This was the first attempt to develop an RNA phenotypic assay for drug resistance detection in Mtb and provides for an innovative, clinically useful and accurate diagnostic that captures resistance profiles to the novel and repurposed anti-TB drugs.

[0140] Significance

[0141] DNA based resistance diagnostics have inherent limitations including for the novel anti-TB drugs

[0142] Multidrug-resistant (MDR) TB treatment requires the concurrent use of five or more effective antibiotics and costs eight times as much as treatment of drug susceptible TB8. Improved diagnosis and treatment are designated priorities by the World Health Organization(WHO) and Centers for Disease Control (CDC) for addressing the antibiotic resistance challenge9 1 1. The knowledge of genetic mechanisms of antibiotic resistance has formed the basis of several commercial molecular diagnostics for TB that have had remarkable global uptake despite the fact that they only reliably test for one or two TB drugs9'12 l4. For guiding MDR-TB treatment, available commercial molecular assays have not yet been able to replace the traditional more costly and slow process of growth based phenotypic drug susceptibility testing (pDST) that relies on in vitro mycobacterial culture. pDST currently takes about 2-8 weeks from the time of initial sample collection to yield a result. The clinical use of pathogen whole genome sequencing (WGS) is promising as a replacement for pDST and is currently performed on early positive in vitro culture samples (usually 2-3 weeks from sample collection) in several high resource laboratories like the New York State Wadsworth Center and Public Health England15. The diagnostic accuracy of WGS is excellent for drugs like isoniazid (INH) and rifampicin (RIF) but still lags behind phenotypic methods for other key drugs like pyrazinamide (PZA) and moxifloxacin (MXF) and the novel / repurposed TB drugs16 l s. This is partly due to the diversity and rarity of certain resistance variants making assessments of their effect on the resistance phenotype more difficult16. It is also true that M. tuberculosis (Mtb) is continuously evolving under antibiotic pressure including to novel antibiotics as they are introduced into clinical care. The redundancy in resistance mechanisms combined with the diversity of background genetic variation makes it difficult to tease out causative from neutral variants. Rapid phenotypic assays can circumvent these challenges and can have wide applicability even for drug susceptible TB, given that MXF is now recommended as part of 4-month regimen.

[0143] Tackling the antibiotic resistance epidemic requires at least two tiers of resistance diagnostics

[0144] Tackling the antibiotic resistance epidemic requires at least two tiers of resistance diagnostics: (1) rapid point-of-care tests for initial screening directly on patient sputum or other clinical samples without in vitro culture such as the widely used GeneXpert-TB / RIF.(2) The second-tier of resistance diagnostics needed are laboratory-based confirmatory tests that complement the GeneXpert by offering superior speed, sensitivity and specificity and by testing for susceptibility to other key drugs needed to guide antibiotic therapy. For the second-tier, the Foundation for Innovative New Diagnostics (FIND) defines the minimal target product profile (TTP) at >90% sensitivity, >95% specificity and <72 hours turn around19. Here an early phase development work for a work-flow that will be first optimized as a second-tier diagnostic focusing on meeting the TTP for five antibiotics (Aims 1 and 2) isproposed. The assay also holds promise as a first-tier assay, and in Aim 2 preliminary data was collected on its potential for use in that capacity. Focus will be on five anti-TB drugs that either have a very limited knowledge base on resistance mutations due to their recent introduction into clinical use or because the drugs continue to have gaps in sensitivity with WGS.

[0145] Cellular mRNA levels are rapidly perturbed by antibiotics in susceptible Mtb

[0146] Like all cells, Mtb adapts to chemical pressure and other environmental stressors through transcriptional change20 22. Compared with E.coli, mycobacteria are notable for an RNA-polymerase that forms unstable open promoter complexes that appear more easily prone to regulation by repressors or stimulating factors23. This may imply a more rapid transcriptional response than other bacteria23. At the same time, Mtb demonstrates longer mRNA half-life on average (9.5 min) compared with E.coli (3-8 min)22. Whereas this may delay the observation of transcriptional changes after they occur, especially for abundant transcripts, the delay will be on the order of minutes rather than hours. Hence despite Mtb’s substantially slower growth in vitro than E.coli, its transcriptional response is similar in kinetics and offers a much more rapid window to predict antibiotic effect than growth inhibition. These observations have recently fueled innovations in Mtb treatment monitoring with development of the ribosomal synthesis metric (RS- using rRNA targets) that can be assayed using digital droplet-PCR (ddPCR) directly on sputum from patients within a few days of treatment initiation7. Similarly, a cell viability score was recently developed by an independent team applying support vector machine (SVM) algorithm to Mtb RNAseq data under stress conditions24. The viability score correlated with Mtb killing by antibiotic in culture, in macrophages and directly on sputum of patients receiving antibiotic treatment24. It was demonstrated that the transcriptional antibiotic response (AR) observed in susceptible isolates differs vastly from that in resistant isolates, with the AR in resistant isolates being comparable to an unexposed susceptible isolate. This supports that transcriptional AR has the needed characteristics to capture antibiotic resistance5.

[0147] Innovations in RNA diagnostics and their application to pathogens other than Mtb motivate this proposal

[0148] The COVID19 pandemic has rapidly accelerated innovation in molecular diagnostics for RNA detection25. Because of high analytical sensitivity, RT-qPCR remains the gold standard for RNA detection, but labor and reagent shortages and concerns about global access26accelerated the development of alternative simpler assays such as the isothermal amplification e.g. LAMP27,28or RPA29, and CRISPR RNA detection30,31. These assayscurrently have lower diagnostic sensitivity than qPCR28, and in some cases also lower specificity30, but active innovation in this area is promising to close these gaps32. One of the most sensitive and reproducible hybridization methods for detecting RNA without amplification is the nCounter technology from NanoString33. Using color-coded molecular barcodes and single molecule imaging, this technology can quantify up to 800 unique targets (lOObp in length) across a dynamic range of 10-50,000 counts in one reaction33. When compared against RT-qPCR, it has comparable to higher analytical sensitivity and higher reproducibility down to a target abundance of 10 counts (R2 > 0.99)6,33. nCounter can quantify transcripts in crude cell-lysates without RNA extraction or cDNA synthesis even on small input volumes6,33, and was recently used to diagnose resistance in E.coli and K. pneumonia with high sensitivity and specificity using an mRNA AR approach6. nCounter was also used for pathogen identification directly from blood34. Although further development is needed to advance the use of NanoString hybridization nCounter technology into a clinical assay, its versatility as a research tool supports the development of an RNA diagnostic prototype for Mtb resistance with high efficiency and low cost. Later stage development can then leverage burgeoning RNA detection technologies such as CRISPR.

[0149] Innovation

[0150] This proposal is innovative in one or more of the following ways described herein: its focus on transcriptomic antibiotic response i.e. a phenotypic + molecular approach. This presents a new concept for Mtb resistance diagnosis different from the mainstay approaches that focus on DNA mutations or growth assays. A diagnostic reliant on transcriptomic AR promises to be independent of the genetic resistance mechanism, i.e. a phenotypic assay that directly measures early signs of cellular stress.

[0151] This process is also innovative in the use of NanoString nCounter technology that vastly improves efficiency and decreases cost over RNA sequencing, and RT-qPCR, allowing us to develop an assay workflow and test it on prospectively collected sputum in one proposal.

[0152] Further, this proposal is innovative in the focus on novel Mtb drugs for which phenotypic assays are less accessible than first-line or older agents, and for which genetic mechanisms are still being fully elucidated. Rapid phenotypic assays may support the prevention of resistance before it emerges to these valuable agents for MDR-TB treatment.

[0153] Further, this proposal is innovative in the focus on both mRNA and rRNA transcriptional antibiotic response including experiments exploring attenuation NanoString probes for simultaneous abundance measurement of both RNA molecule types. The captureof both rRNA and mRNA AR promises to increase the sensitivity and accuracy of resistance diagnosis.

[0154] Approach

[0155] This example builds on potentially insurmountable gaps in predicting resistance from DNA sequence for PZA, BDQ and PA824. Over 30,000 Mtb whole genome sequences (WGS) and drug resistance phenotypes using both GWAS and machine learning have been analyzed by the inventors16,47’49. This work has highlighted that the genetic determinants of Mtb resistance are complex in number and distribution, spanning coding as well as noncoding variants49. This complexity is especially recognized for PZA to which upwards of 150 mutations have been characterized. Several instances of important interactions at the protein or pathway level for several antibiotics including MXF, the rifamycins and BDQ have been described that significantly complicate resistance interpretation arguing for the need for statistical models rather than simple catalogue methods to predict resistance3 16 18’37’48’51. Her lab has built machine learning models for predicting the resistance phenotype from genotype and these tools are used daily by clinical mycobacteriologists and researchers globally to predict antibiotic resistance3 16’47. Table 1 describes the cutting edge performance of these models. Notably for the novel and repurposed drugs, including BDQ, PA824, and LZD there is currently a lack of WGS and phenotypic data from resistant isolates to support GWAS for inference of clinically relevant resistance determinants. In a recent systematic review conducted on the resistance determinants to these drugs combining in vitro mutagenesis, spontaneous evolution and the available limited clinical studies18, it was found that similar to PZA, resistance to BDQ and PA824 is most often caused by mutations in non-essential genes with a large diversity of mutations identified in resistant isolates. More than 100 mutations were found across five genes for pretomanid alone, spanning 6,38 Ibp. Even if WGS is used to interrogate these genes, interpreting novel mutations will remain a challenge and building prediction models will require sequencing of hundreds to thousands of isolates developing resistance during care. Given expected rates of resistance development, 1-2% in one clinical trial, it was not possible to wait for resistance to develop in clinical settings before developing an accurate diagnostic. Resistance determinants appear more concentrated in genomic distribution for the LZD and MXF, the drug targets of which are the essential ribosomal genes and topoisomerases respectively. However, gaps remain in sensitivity of sequencing for these highly potent drugs (Table 1 for MXF), and phenotypic assays will remain difficult to scale globally due to availability of drug and laboratory infrastructure. A comprehensivephenotypic MDR-TB DR assay that can confirm susceptibility to the full BPaLM regimen (BDQ, PA824, LZD and MXF) is needed.

[0156] Transcriptional changes after antimicrobial exposure associate with cellular stressExp = exposure experiments from the same or a different source study. Each exp. had 2- 4 technical replicates, x CC = multiple of the critical concentration as defined by WHO handbook on TB DST. Controls were drug susceptible strains that were not exposed to antibiotic (no transcriptional data on DR strains exposed to antibiotic was available).Abased on average of DE across the exposure experiments for each drug. AUC = area under the ROC curve for classifying exposure vs. no exposure obtained by logistic regression using only the top 5 upregulated genes. *AUC is preliminary as the dataset is small and had a range of drug exposure concentrations and durations. Highlighted in grey are downregulated genes.

[0157] Transcriptional changes after antimicrobial exposure associate with cellular stress

[0158] In preliminary work, public Mtb microarray data20,52from antibiotic susceptible isolates after antibiotic exposure and multiple stress (hypoxia, high CO2, low nutrient and acidic pH) were analyzed, identifying that the heat shock response was the most significantly upregulated pathway across antibiotic exposure and multiple stress, and translation / ribosomal function is the most significantly down regulated pathway. This comparison highlights qualitative similarities in cellular response between stress and antibiotic effect among susceptible cells (Fig 3). The same comparison also highlighted differential expression of multiple genes in an antibiotic- specific manner. For example, PZA exposure results in many unique differences in gene regulation (Fig 3 compare PZA to PA824). One of the most upregulated genes is ATP-dependent protease subunit clpB (Table 2) consistent with PZA’s mechanism of action of tagging the PanD protein for degradation via ATP-dependent co proteases .

[0159] Feasibility of assaying transcriptional AR for resistance diagnosis

[0160] Drug specific transcripts that are extremely up or downregulated in response to ciprofloxacin, streptomycin or isoniazid have been characterized5. Five to nine transcripts were selected. NanoString probe sets were designed to capture these transcripts in Mtb reference strain H37Rv cell lysates after exposure to antibiotic for 6 hours, compared with the same strain grown in antibiotic free culture for the same duration (Fig 4 top). The largetranscriptional change observed in antibiotic susceptible isolates was not observed in exposed isolates that were resistant to this drug (FIG. 4, bottom). An assay based on this demonstrated 100% sensitivity and specificity in detecting resistance to isoniazid (FIG. 5A) and was independent of genetic resistance mechanism (FIG. 5B). This evidence makes a strong case for the feasibility of mRNA signature detection for the rapid diagnosis of antibiotic resistance in Mtb but requires RNA signature optimization for resistance detection across a larger number of clinically resistant genotypes and strain genetic backgrounds to guarantee generalizability. To identify AR genes for BDQ, PA824, PZA, LZD and MXF, a metaanalysis of transcriptional response after antibiotic exposure were performed. 16 source studies were identified that generated 26 micro-arrays or RNAseq datasets to assess transcription in drug exposed Mtb with at least two technical replicates with data meeting quality criteria24'54 64. The data spanned 346 experiments that used different technologies, drug exposure doses and duration. The majority were performed on the H37Rv Mtb reference strain. Differentially expressed genes (DEGs) were identified under antibiotic exposure compared to the control unexposed identical strain. For the purpose of homogeneity, experiments where the drug concentration was <0.4x the critical concentration (CC) in that medium, or were exposed for >6 hours were excluded. 16 transcriptomic profiles were generated across 5 antibiotics and an additional 57 conditions representing control conditions where there was 0-24 hour interval incubation in culture and no antibiotic exposure. Table 2 lists the top DEGs for each drug, and the preliminary predictive performance (measured by area under the receiver operating curve (AUC) for classifying exposure using the top 5 most upregulated genes). The high AUC although preliminary due to the small number of experiments demonstrates how different the exposed and unexposed transcriptional profiles are. It was noted that 2 genes overlap between the transcriptional AR to MXF and ciprofloxacin; this is expected given that ciprofloxacin has only partial cross-resistance with MXF and they have different potency against Mtb38 (FIG. 4 leftmost panel5, 2 of the 9 genes are in the top 50). The DEGs that were identified are consistent with the antibiotic mechanism of action69.

[0161] The addition ofrRNA targets could improve the accuracy of resistance diagnosis and bring the assay closer to the POC

[0162] The analysis above focuses on non-ribosomal mRNA response. Although mRNA is more abundant in the cell than DNA, the majority of cellular RNA is ribosomal estimated at 85-95% of total RNA70,71. Seminal work on an rRNA based signature was led to assess sterilizing drug effect for in vivo TB treatment monitoring. This focuses on detecting the ratiobetween the precursor rRNA (16S and 23S) that includes the short-lived spacer sequences ETS1 -external transcribed spacer 1 and ITS1 - internal transcribed spacer 1 (FIG. 6A) relative to mature rRNA with the ETS1 and ITS cleaved off. This ribosomal synthesis (RS) ratio is closely correlated to growth rate in an in vitro progressive oxygen depletion model (FIG. 6Bb) and to colony forming units-CFU in mice lung infection (not shown)7. In vitro exposure of Mtb to sterilizing antibiotics including RIF and BDQ rapidly decreases the RS ratio (FIG. 7). For BDQ, the RS ratio decreases by 6 fold in 6 hours (hr), providing a substantially earlier functional signal of AR than visible growth in culture (FIG. 7). The RS ratio has not yet been measured in BDQ resistant isolates, but given above observations on mRNA signals and expected normal growth signal in vitro for BDQ resistant isolates as assessed by traditional culture based drug susceptibility testing, it is highly likely that the RS can be used to differentiate antibiotic resistant from susceptible isolates for BDQ and potentially other antibiotics. The RS ratio detected by digital droplet PCR has a pending patent for treatment monitoring in Mtb in both human infection and in mice to facilitate drug development. The ability to accurately measure the RS ratio on direct sputum samples from patients undergoing TB therapy has been demonstrated,7demonstrating the promise of rRNA AR in diagnosing resistance at the POC.

[0163] Overview

[0164] In Aim 1, experimental exposure variables to detect RNA AR rapidly and reproducibly were planned to be optimized and assess the generalizability of the RNA AR signature across Mtb lineages and resistance mechanisms. This was followed in Aim 2 by the development of early phase diagnostic workflow based on NanoString probes and benchmark in parallel with RT-qPCR on limit of detection and operationalize the workflow on prospectively collected clinical isolates including direct sputum and early positive culture compared with traditional growth based phenotypic testing for 5 drugs. Table 3 gives an overview of the experiments.Table 3: Overview of Experimental # Sequencing / capture runs planned experiments Approach (unique clinical isolates)ART-qPCR will be run is triplicate and counts not shown*first batch will be validated with triplicate RT-qPCR, all LOD Nanostring experiments will be run in parallel with triplicate RT-qPCR. Nanoprobes will be designed for rRNA and mRNA targets.#30 patient samples directly and after two culture time points, each patient will be tested for PZA and one of the 4 other drugs

[0165] Clinical isolate archive and sample selection

[0166] All 55,000 RR isolates in the SU clinical Mtb isolate archive have phenotypes for the drugs isoniazid, RIF, amikacin and fluoroquinolones. Additionally, >400 of these isolates are confirmed to be phenotypically resistant to PZA and >50 isolates resistant to each of BDQ, LZD or PA824; 6,500 have available whole genome sequencing (WGS) data and a subset of the remaining isolates have MIRU-VNTR, spoligotyping or RFLP to assess mycobacterial genetic lineage. Resistant (R) and susceptible (S) isolates were selected from this archive that span the major 4 Mtb lineages (LI, L3, L2, L4) that are circulating in South Africa; -60% of the isolates will be L4, -25% will be L2 and the remaining -15% belong to LI or L3. 8 S and 8 R isolates per drug (expected total n across 5 drugs = 80) were selected. Isolates will be selected to represent five different resistance mechanisms for each drug of interest based on the DST and WGS data available e.g. frameshifts at ntl92 in Rv0678 vs. the mutation D28N in the gene atpE are two different resistance mechanisms for BDQ; the list of genetic mechanisms is derived from the WHO resistance mutation catalogue and the published literature18,50. Specific mechanism distribution will vary by lineage for each drug, but the aim was to have common mechanisms between lineages, and to match sub-lineages between susceptible and resistant to allow separation of effects of mechanism or lineage on RNA AR(Table 4). Susceptible isolates for each drug will span isolates that are pan-susceptible and resistant to other drugs (e.g. mono-resistant, MDR and XDR); specifically equal numbers of RIF resistant (RR) and susceptible isolate in this sampling were confirmed because mutations in the beta-subunit of RNA polymerase- rpoB, the main mechanism of resistance to RIF, are recognized to have pleotropic effects on transcription in Mtb72. L4 will be overrepresented given its high global prevalence and its recognized high genetic diversity. All 80 isolates will be used in Aim 2, and 40 isolates will be used in Aim 1 (marked in red in Table 4).Table 4: Approach to isolate selection for each drug. R: resistant, S: susceptible.Superscript designates sublineage (e.g. L2.1, 4.3 etc) to be matched between S and R isolates for each major lineage. Subscript: designates type of R genetic mechanism. Three L4, two L2, two L3, and one LI will be selected for each drug-susceptibility pattern. Isolates in red will be used in Aim-1. All isolates will be used in Aim-2.

[0167] Mycobacterial Culture

[0168] Archived culture will be revived by thawing and inoculating into MGIT (Mycobacteria Growth Indicator Tube) liquid culture media. After reaching early logarithmic growth at 107 CFU per ml expected in ~1 week, the isolates will be batched for each exposure experiment described below by drug.

[0169] Growth based phenotype drug susceptibility and Minimum inhibitory concentration (MIC) determination and DNA sequencing

[0170] The 80 isolates will have repeat DST to isoniazid and RIF at WHO recommended critical concentrations (CC) using MGIT culture (O.lug / ml and 0.5ug / ml respectively)73. MIC testing was performed at 6 concentrations in MGIT to the five drugs of interest to correlatewith the RNA based assay results. MIC concentrations will be chosen to include two doublings above the MGIT CC and three doublings under the CC as recently defined for each drug18’74~76. MICs for PZA will be determined using the BACTEC MGIT 960 PZA kit (Becton Dickinson Biosciences, Sparks). WGS will also be conducted on the 80 clinical isolates to capture any in vitro evolved variation during archiving and characterize any noncoding variation and lastly confirm resistance mechanisms. After culture, bulk aliquots will be lysed using lysozyme and proteinase K and DNA extracted using the CT AB / Chloroform method. Library preparation will be performed using NexteraXT or latest PCR-free method and sequencing on HiSeq X (Illumina, San Diego, CA), generating 150bp paired-end reads at 150x.

[0171] Aim 1: Optimize detection of transcriptional antibiotic response to five drugs

[0172] Overview

[0173] The preliminary data suggests that after a few hours of antimicrobial exposure at sufficient doses, bacterial gene regulation reflects both drug-specific and non-specific cellular stress that spans down regulation of ribosomal synthesis genes for bactericidal agents, and upregulation of heat-shock response and DNA repair machinery. In this aim, the detection of RNA AR for five key antibiotics in TB treatment were optimized, including an explicit focus on rRNA dynamics given their demonstrated promise in in vivo drug monitoring. It may be hypothesized that combining rRNA and mRNA response with generate a more robust signal of AR, and hypothesize that in vitro antibiotic concentrations higher than the critical cut-point for growth-based assays can shorten the time to detect transcriptional AR. Accordingly, rRNA and mRNA responses across a range of drug concentrations and durations of exposure were investigated. Generalizability of the AR across genetically diverse Mtb isolates that vary in lineage and drug resistance mechanism leveraging the unique SU clinical Mtb strain archive were evaluated.

[0174] Rationale

[0175] The choice of drug concentration is crucial in phenotypic DST. Resistant isolates can display growth inhibition at very high drug concentrations in vitro that are unachievable in a patient’s serum at safe doses. At the same time, higher drug concentrations can more rapidly and prominently illicit transcriptional AR (from the meta- analysis described in 3.2.3 and ref57). rRNA responses have not been studied in resistant isolates in vitro previously over a range of antibiotic exposure conditions. The generalizability of the AR is also important to establish. Thus far preliminary data has been generated largely on H37Rv, a pan-susceptible Mtb strain. Mtb is recognized to have low genomic diversity, but a substantial proportion ofdiversity is attributable to genetic ancestry a.k.a lineage77. Mtb lineages are recognized to have small differences in transcription, specifically 202 / 3906 (5%) of protein coding genes in Mtb were previously quantified to be differentially expressed (DE) between the four major Mtb lineages under standard in vitro non-drug exposed conditions78. In previous work spanning >10 years of study of genetic drug resistance mechanisms in Mtb, variants in 12 promoter regions or transcriptional regulators that cause resistance phenotype effects on transcription were characterized18,48’49. Overall the number of DEGs due to Mtb genomic diversity and drug resistance phenotypes is relatively small (~5% of Mtb genes), and this is consistent with the preliminary analysis of robustness of the RNA AR across resistance mechanisms. Nevertheless, the noted effects of genomic diversity on transcriptional AR have not been previously investigated, and their study will support assay development to confirm generalizability and performance across TB clinical settings treatment TB caused by different lineages and resistance types.

[0176] Choice of drug exposure doses and times for signal maximization

[0177] Mtb’s AR is dynamic over drug concentration and time with different effectors observed early vs. late after exposure. For example, BDQ acts through inhibition of mycobacterial ATP-synthase. Within the first few hours, BDQ exposure results in upregulation of heat shock response, and genes that generate ATP through alternate paths including glycolysis, followed by induction of drug tolerance / dormancy state around 48 hours including upregulation of DNA repair, and down regulation of ribosomal machinery and protein synthesis57,63. These preliminary observations justify the need to investigate the optimal drug dose and duration of exposure to maximize the signal. For each drug, susceptible isolates will be unexposed or exposed to drug at three concentrations (IxCC, 2xCC and 4xCC) and incubated for three time intervals (3 hr, 6 hr and 18 hr) in a factorial design of all concentrations vs all durations. The choice of concentrations and time points is informed by preliminary NanoString experiments for isoniazid, ciprofloxacin and streptomycin, and the analysis of micro-array and RNAseq data for the other drugs (discussion of Feasability of assaying transcriptional AR for resistance diagnosis), as well as evidence from gram negative bacteria given similar RNA polymerase kinetics described above. The preliminary data demonstrated that exposures at sub-CC concentrations for very short time intervals (< 3 hr) are unlikely to perturb the transcriptional response sufficiently across the five drugs of interest. Although it is tempting to extrapolate the CC from growthbased assays to transcriptional response, it is important to re-evaluate the optimal concentration of testing because this is a fundamentally different read out of susceptibility. Itwas tested if two doublings above the CC can elicit a more rapid and discriminatory response between susceptible and resistant isolates than the CC, This may be important with drugs with slower transcriptional response such as PZA (preliminary data not shown). The limit was drawn at 4xCC to avoid testing at concentrations that may elicit responses from resistant isolates. The specific concentrations and durations were modified based on the latest available public RNAseq data when the study starts. Shorter durations of exposure were tested if robust AR at 3 hours were found.

[0178] Antibiotic exposure experiments for maximization of AR

[0179] The AR genes summarized in Table 2 were derived by pooling drug exposure experiments over a range of exposure times and concentrations. Whereas this constitutes a robust AR that has been averaged over several replicates and platforms, ultimately a choice of specific time and dose needs to be made to optimize the detection of AR for diagnostic use. rRNA depleted RNAseq were used, and RT-qPCR targeting the RS ratio (Discussion of the addition of rRNA targets could improve the accuracy of resistance diagnosis and bring the assay closer to the POC(, Table 3). Exposure experiments were run in a factorial design of 3 drug concentrations and 3 durations on 1 S and 1 R isolates per drug. Experiments will be batched by drug for BDQ, PZA, PA824, MXF and LZD. This will result in 180 unique experiments that will be run in duplicate. Up to two additional time points were tested depending on the results at 3 hours and 6 hours. For example, if a robust signal is detected consistently for all drugs at 3 hours at higher drug concentration, then an alternative 1.5 hour incubation time was tested as this promises to shorten the assay run time. Previous studies suggests that BDQ generates a strong transcriptional response at 1.5 hours at 10xCC57, but this has not been tested for other drugs. After the drug concentration and time is chosen, 10 additional experiments were run in duplicate across the five drugs with two lower starter culture CFU counts (105 and 103 vs. 107 above). This will allow us to directly assess the impact of cell loading on AR, relevant for control transcript evaluation in Aim 2.

[0180] RNA preservation and cell lysis

[0181] When the drug exposure time is completed the cultures will be incubated for 5 minutes in a guanidine thiocyanate (GTC) RNA stabilization solution (5 M guanidine thiocyanate (Biosynth), 0.25 M sodium citrate, 1% Tween 80, 1% P-mercaptoethanol) followed by centrifugation. Compatibility of this solution was confirmed with downstream hybridization planned in Aim 2 (with NanoString application scientists). Supernatants will be discarded, and cell pellets immediately suspended in Trizol (Invitrogen). the Trizol suspended Mtb cells will be disrupted using 0.1 mm diameter zirconia / silica beads (BioSpec)with three 40 second pulses in a bead beater. Samples will be kept on ice for five minutes between pulses. Cellular debris will be separated from the cell lysate by centrifugation for 1 minute at 21,000xg.

[0182] Antibiotic exposure experiments for evaluation of AR generalizability across Mtb lineages and. resistance mechanisms

[0183] The 30 remaining isolates matched by lineage (Table 4 in red, 10 of the 40 were already exposed in 3.4.4), will be exposed to the optimal drug concentration, and duration of exposure for each drug (as determined in 3.4.9). Exposure experiments will be run in parallel with control experiments of same isolate incubated in culture for an identical duration but without drug added. Across the 30 isolates, 120 unique exposed / unexposed experiments will be conducted across the 5 drugs. Each experiment will be run in duplicate and batched by drug, RNA stabilization and cell lysis will follow as in (Discussion of RNA preservation and cell lysis').

[0184] RNA extraction, library preparation and sequencing

[0185] The protocol for RNA extraction from culture is similar to previously described with some modifications based on kit availability and cost in South Africa7. Briefly, total RNA will be extracted from the cell lysates using the RNeasy mini column system (Qiagen). The RNA integrity will be determined using the Agilent 2100 Bioanalyzer (Prokaryote Total RNA Nano Series II) version 2.5 (Agilent Technologies Inc., Santa Clara, CA, USA). Only RNA samples with a RIN value of between 6.5 and 10 were selected for further analysis, or otherwise the extraction or experiment will be repeated (expected for 5-10%). The total RNA extract will be split for rRNA RT-qPCR (Discussion of Ribosomal RT-qPCR), and for rRNA depletion and RNAseq; this will allow more accurate quantification of the lower abundance non-ribosomal mRNA during the sequencing step. 1 pg total RNA from each sample will be depleted of ribosomal RNA and packaged into a stranded cDNA library using the Illumina Stranded RNA Prep Ligation kit with Ribo-Zero Plus following manufacturer recommendations. After final library quality checks (Bioanalyzer), quantitation (Qubit fluorescence), and pooling, barcoded libraries will be sequenced to a depth of 10 million reads per sample using 2X150bp Illumina chemistry. RNA sequencing will be done in biological duplicates, i.e., starting with a separate culture aliquot replicating the full protocol of drug exposure to RNA sequencing.

[0186] RNAseq analysis and characterization of differential expression

[0187] The RNAseq reads will be quality controlled with FastQC79, trim low quality and adapter sequences with Trimmomatic80. The usual workflow is to align the trimmed reads tothe H37Rv reference genome using BWA81 and quantify transcripts using featureCounts 82 and identification of DEGs with DeSeq283 between the exposed and unexposed isolates at each time point. An alternative newly developed tool, Rockhopper, a tool specific to bacterial transcription that can annotate operon structure was tested and allows for high quality transcript mapping, normalization, transcript abundance quantification and analysis of differential expression in a nice user interface84,85. Quality of the transcriptomic analysis was ensured by checking the a) reads mapped to mRNA vs rRNA, b) correlation between biological replicates (R2 > 0.95), and c) global correlations to other RNAseq data that was previously collected (Discussion of Feasibility of assaying transcriptional AR for resistance diagnosis).

[0188] Ribosomal RT-qPCR

[0189] Three technical replicates of RT-qPCR will be performed for each capture reaction. From 0.5 pg total RNA extracted (Discussion ofRNA extraction, library preparation and sequencing), first strand cDNA synthesis was performed using random hexamer primers and Superscript II RT (Invitrogen Life Technologies). TaqMan FAM and HEX probes and primers for Mtb ETS1, ITS1, and 23S (FIG. 6A) will be synthesized following the a duplex design7. Real-time quantification will be carried out using the QuantStudio™ 5 (ThermoFisher) or CFX-96 Real-Time (Bio-Rad) PCR Instrument as described by the manufacturer.

[0190] Data analysis of maximal transcriptional AR

[0191] For signal maximization, the exposed transcriptome captured with RNAseq were compared with the paired unexposed transcriptome across concentrations and time points for susceptible isolates. Recent comparisons of tools for time course or other longitudinal analysis of DE (such as over a range of drug concentrations), supports the use of DESeq2 with pairwise comparisons between time and concentrations over more complex models86. The DESeq2 negative binomial (NB) generalized linear model (GLM) function was the first tool used for assessment if time and concentration are significantly associated with DE using the NB likelihood ratio test (LRT). Using Benjamini Hochberg false discovery rate (FDR) threshold of 0.05, DE were assessed at each time point and concentration relative to no exposure at the same time point using a two-sided Wald test for each transcript. The concentration-time point with the largest % of significant DE were identified among susceptibles (i.e., proportion of transcripts with significant DE at a false discovery rate of 5%) compared with no exposure and in addition the point of maximal amplitude of differential expression, mean Log-fold DE for upregulated and downregulated genes were respectively identified. Overlap between the DEGs captured by DESeq2 at the maximal pointwith the predefined AR genes (Table 2 lists upregulated genes) will be assessed. Similarly. It was determined if any AR is observed among resistant isolates. For the RS ratio, relative expression of ETS to 23S and ITS to 23S will be computed and averaged over the three replicates and compared pairwise between unexposed and exposed isolates across concentrations and time points using the non-parametric Wilcoxon rank sum test with correction for multiple testing using FDR<0.0587. These results were used to assess for the point of maximal significant amplitude of ribosomal transcription. If the 3 hr mark was found to be the maximal point for any drug under specific drug concentrations (expected to be at >CC), additional experiments were conducted at 1.5 hour for the highest drug concentration. If time points are similar with regards to signal strength, identifying a similar time point of testing across drugs in case a multiplex or array type of assay can be developed in later stages of assay development was prioritized.

[0192] Data analysis to assess the generalizability of the transcriptional AR

[0193] In the primary analysis, the pre-specified transcripts (partially described in Table 2) were used to assess if these transcripts are antibiotic responsive among resistant isolates (across resistance mechanisms / lineages) or if they are DE due to lineage / genetic mechanism of resistance. To do this, a pairwise Wald tests (at FDR of 0.05) was run using DESeq2 for (1) susceptible exposed / unexposed and (2) resistant exposed / unexposed and (3) susceptible exposed across different lineages, and (4) resistant exposed across different lineages and mechanisms at the chosen time point and concentration in 3.4.10. Given preliminary data, it may be anticipated that the signal to be robust and that minimal DE of these genes in exposed resistant isolates will be seen, and that the genes will not be significantly DE due to drug exposure but not lineage or genetic resistance mechanism. Similarly, a significant reduction in RS ratio for susceptible exposed / unexposed vs. resistant exposed / unexposed across lineages and resistant mechanisms was evaluated. If some putative AR transcripts are not consistently DE due to drug exposure among susceptible isolates, or are DE among resistant isolates then they will be removed from the list of AR genes. Secondarily, RNA AR signatures de novo were driven, using log2fold DE data for the 40 isolates at the optimal concentration and time, by running the feature selection algorithm reliefF (COREleam R package) using the resistance growth-based phenotype as a binary outcome variable.

[0194] Pitfalls and contingencies

[0195] A potential pitfail is the stability of the antibiotics in storage, accordingly control testing of standard strains with known MIC were conducted to quality control the compound concentration prior to each batch of experiments. The point of maximal AR may differbetween rRNA and mRNA, and the experiments here will help determine this and the feasibility of capturing both simultaneously or separately in assay development. Further because there is a risk of seeing AR under higher drug concentrations among resistant isolates, especially those with mutations that only cause low level increases in MIC, the highest concentration experiments on resistant isolates were started with, and abort further testing at that concentration on susceptible isolates AR is seen (similar to Table 2) among resistant isolates.

[0196] Aim 2: Develop an early phase RNA-based Mtb assay for susceptibility to five drugs

[0197] Overview: Use of the RNA AR signatures was aimed to build a diagnostic workflow using NanoString hybridization probes in two phases. A preliminary performance metrics was generated on the sensitivity, specificity and limit of detection of the prototype using archived and prospectively collected sputum from MDR-TB patients. The NanoString probe hybridization was benchmarked against RT-qPCR duplex assays for the same RNA targets in the limit of detection assessment and on the direct sputum samples (Table 3).

[0198] Aim 2A: Assay design and pilot performance in vitro

[0199] Identifying internal control targets: It has been shown in gram negative bacteria (GNR) that normalizing for cell loading, lysis efficiency, and other experimental factors that can affect absolute transcript abundance from a given sample is important. Such invariant transcripts (often referred to as “housekeeping” transcripts in qPCR) are needed for scaling candidate responsive genes and isolating the effect of antibiotic6. Candidate control genes will be identified as transcripts whose expression does not change in pooled public and Aim 1 RNAseq time course, either upon antibiotic treatment or over the untreated time course. To achieve this DESeq2 was used and an inverted hypothesis test across time that assesses for gene of temporally stable transcription as previously described6. 5-10 control genes were targeted per drug, as this was sufficient in GNRs6.

[0200] Initial NanoString Probe Design

[0201] 30-40 NanoString probes were designed targeting antibiotic responsive mRNA and rRNA transcripts (Table 2 and refined by, but not dependent on, Aim 1), or otherwise contain relevant additional signal meeting the criteria below. The 50bp paired probes will be synthesized by NanoString using standard procedures88.

[0202] 1- Up-regulated antibiotic response genes will be prioritized as their abundance will be on average more readily quantified than down-regulated transcripts due to lower intrinsic accuracy at low target ranges for both NanoString probes and RT-qPCR.

[0203] 2- Targets that are not differentially expressed across isolates due to lineage, and do not overlap with genetically diverse Mtb genomic regions, consulting existing literature77,78(refined by but not dependent on Aim 1)

[0204] 3- Targets that are not close to known coding or non-coding resistance mutations18,48,49 ,5Q, to limit any specificity to a particular resistance mechanism.

[0205] 4- rRNA targets within ETS, ITS and 23S previously validated for PCR. These regions have experimentally confirmed high specificity for Mtb over other bacteria and common non-tuberculous mycobacteria with the notable exception of M. canetti the closest ancestor of Mtb.

[0206] 5- Targets that are not in the same operon to maximize the information obtained from the signal.

[0207] Internal control ‘housekeeping’ targets

[0208] 7- Targets that have high predicted specificity to and within Mtb by careful evaluation for homology across the database of 20,000 assembled Mtbc genomes89and databases of other species using Primer-Blast90or Primer Express (Applied Biosystems).

[0209] Assess the feasibility of simultaneous capture of rRNA and mRNA targets with attenuation probes

[0210] The effect of antibiotics on precursor and processed rRNA abundance is a feature that can be leveraged to build a rapid / POC phenotypic DR test for Mtb. rRNA is -1000 more abundant than mRNA, and the difference in abundance will be most useful diagnostically for direct from sputum detection due to the lower bacillary abundance. At the same time, the substantially higher rRNA abundance relative to mRNA imposes technical challenges in accurate simultaneous quantification of either transcript type. Accordingly, the feasibility of simultaneous capture was tested using the three ETS / ITS / 23S probes and the mRNA probes and compare their capture to separate rRNA capture. For the simultaneous experiments, rRNA directed attenuation probes (directed to the same ETS / ITS / 23S targets) that cannot bind the nCounter cartridge and can be added at precise amounts effectively neutralizing or attenuating a proportion of the target were used. Attenuation probes can limit saturation of the cartridge allowing for quantification of lower abundance mRNA targets while precisely quantifying the rRNA targets. If this is not successful both types of RNA targets independently using two separate probe sets would be captured.

[0211] Assay experimental protocol

[0212] The planned assay overview is displayed in FIG. 8, experimental procedures for individual steps are otherwise similar to Aim 1 up to and including cell lysis. The assay wasdeveloped in two phases. In the first an assay will be performed for the full probe set designed above (40-50 targets) on the 4 R and 4 S isolates per drug (n=40) from Aim 1. The data from these experiments was used and phenotypic data to sub-select targets with the highest predictive power for susceptibility using a feature selection approach detailed below. In the second phase, a revised assay targeting only the selected features on the 40 remaining isolates was run. After each exposure (and unexposed control), cells will be lysed and the crude lysate frozen and transported on dry ice to the NanoString POP facility in Seattle where the capture will be performed using the nCounter Sprint or Pro Instrument depending on the batch size (up to 12, and up to 72 respectively). Shipping of similar samples for NanoString capture overseas, has been successfully done. Previous data supports high reproducibility of NanoString nCounter (R2>0.99) and hence technical duplicates are not needed6.

[0213] Assay statistical analysis

[0214] For each sample, the data analysis will follow manufacturer recommended positive and negative control corrections of the transcript counts using spiked in targets included with each kit6. The transcript count of each responsive gene will be further normalized by control gene expression as a proxy for cell loading using a variation on the geNorm algorithm91, then converted to fold-induction in treated compared with untreated strains. For rRNA targets, the ETS count will be scaled by the 23S count to generate the RS. The RS and normalized antibiotic responsive mRNA counts were used to predict resistance / susceptibly using an approach previously published6. For each phase (Discussion of Assay experimental protocol), the isolate data was split into a 50% train and 50% test set. Using the train set, the top transcripts that best distinguish phenotypically S from phenotypically R strains were selected using the feature selection algorithm refliefF (Phase- 1 -features, 5-10 transcripts are expected). A randomForests (R randomForest) was built using the selected features and geNorm normalized control transcripts. Classifier performance (AUC, F2, sensitivity, and specificity) will be measured on the hold-out test set from Phase 1. In Phase 2, the data was pooled from Phase 1 into one dataset and repeat feature selection (Phase-2-features) and rebuild the classifier using the Phase-2-features and the control features. NanoString capture was performed of only Phase-2-features and controls on new independent isolates (n=40) to measure the final performance of the full workflow including capture and statistical analysis.

[0215] Assay limit of detection (LOD) evaluation in vitro

[0216] An assay on four serial dilutions of the initial input culture (107 CFU) to 105, 103, 102, and 10 CFU was run respectively. This was done for 1 R and 1 S isolates for each drug, and compare the performance of the hybridization and the Phase-2 resistance classifier onthis serially lower input (80 experiments) with RT-qPCR. Performance will measured by dilution specific AUC, F2, sensitivity and specificity. The LOD will be the dilution below which a >10% decline in sensitivity was observed.

[0217] mRNA RT-qPCR primers / probes will be designed and tested to target the antibiotic responsive Phase-2-features using Primer Express (Applied Biosystems). RT-qPCR testing will be run in duplex including a common standard in each reaction and will use the instruments and procedure outlined in 3.4.9. RT-qPCR will be used in parallel in a subset of Phase-2 nanosensor runs for validation purposes and on the lowest two dilutions in the limit of detection experiments. Based on head-to-head comparisons by the manufacturer, the nCounter is expected to have higher sensitivity than RT-qPCR, but lower sensitivity than digital-droplet PCR33.

[0218] Aim 2B: Pilot performance characteristics on fresh sputum samples from RR-TB patients

[0219] Overview

[0220] The assay will be run directly on freshly collected sputum without culture (0-2 days from collection), and on early positive culture growth in MGIT (-1-3 weeks from collection, median of 7 days in previous studies), and after standard duration MGIT culture to 107CFU as in Aim 1 (-2-5 weeks from collection). Performance will be compared across the three time points for each participant and overall. Given the small numbers of participants recruited and the expected low rate of resistance to BDQ, PA824 and LZD, an evaluation of the prototype diagnostic sensitivity to these drugs was not expected. A preliminary evaluation of specificity of resistance diagnosis to these drugs was performed, and sensitivity and specificity to PZA and MXF. Extrapolating from the South African national drug resistance survey, it was expected that 59% of RR-TB in the catchment area to be resistant to PZA, and 13% to be resistant to fluoroquinolones. BDQ use began in Worcester at Brewelskloof Hospital in 2015 with recent surveys estimating -4% of RR-TB patients to have BDQ resistance92,93. Resistance to EZD and PA824 is expected to be rare and likely <2%.Participants were enrolled based on RR on GeneXpert Ultra, as RR triggers change to second line drug treatment. The majority (-75%) of participants with RR-TB will also have isoniazid resistance (hence MDR-TB, defined as resistance to isoniazid plus rifampicin) as has been the case in South Africa94. The full resistance panel was confirmed by conducting DST to the first-line drugs and MICs to the BPaLM and PZA.

[0221] Recruitment and study procedures

[0222] 30 RR-TB patients from two TB clinics in Worcester as well as the Worcester Brewelskloof TB hospital were prospectively recruited. Worcester is the main town of the Breede Valley municipality in the Western Cape Province, 90 minutes’ drive northeast from Cape Town, with an estimated population of 150,000. This clinical setting carries a high burden of TB at 0.9% of the population, and RR among TB occurs at a rate of 4.5%. This research side was established in 2016 (TRUST- 1 R01 All 19037). Clinical research projects are ongoing at these centers. Participant recruitment will benefit from resource sharing with three studies (TOTAL R01AI147316 PI Jacobson, TRUST-II R01AI155765 PI Farhat and K23152930 PI Dr. T. Bouton mentored by Dr. Jacobson). For recruitment, participants initiating RR-TB treatment, aged >15, and with GeneXpert-TB / RIF Ultra positive for MTB and RR, will be approached by study staff for screening and potential enrollment. Children <15 years were excluded as they may have much lower bacillary burden. If the individual is willing to participate, the study staff will obtain written informed consent to participate if >= 18 years or written assent and parental consent if < 18 years. Enrollment will occur before treatment initiation (within 3 days). Study staff will extract data on the participant’s TB, including prior TB episodes, HIV and physical health status, including known co-morbidities and active medications, from their clinic records. Enrollment will occur at Brewelskloof Hospital if the participant is hospitalized or at the Worcester CDC clinic if they are initiating treatment in the community.

[0223] Sample collection, initial processing, culture and MIC testing

[0224] Following standard procedures, two spontaneous sputum specimens will be collected by study staff from participants before initiating RR-TB treatment allowing a 2 day window and on two different days. Samples will be first morning specimens to the extent possible, and will be visually inspected to confirm quality of expectoration by trained study staff. Participants will be given a grocery voucher valued at 200 Rand (-USD 15) after their baseline appointment. Standard procedures were followed for collecting, storing, and transporting specimens twice weekly to a laboratory as has been done successfully for other studies95. The two sputum samples will be pooled upon receipt. For decontamination and liquefaction, NaOH / NALC incubation and centrifugation (standard processing) were compared vs. NALC alone to better preserve Mtb cell viability in experiment using RT-qPCR to rRNA targets. Liquefied sputum will undergo centrifugation to separate cellular material. The pellet will be resuspended in PBS and split into two aliquots, one will be used for direct NanoString assay (Discussion of Running the assay directly on sputum). The second will undergo NaOH incubation if not already done, followed by centrifugation and inoculationinto 7H9 BACTEC MGIT 960 liquid culture system (BD, Sparks, MD). pDST and MIC after culture will be performed as detailed above.

[0225] Running the assay directly on sputum

[0226] The sputum cellular suspension will be inoculated into 7H9 culture media with or without anti-TB drugs and incubated for the same time determined for culture in Aim 1 followed by RNA preservation and cell lysis as described above. The procedure will be repeated from a culture aliquot after early positive MGIT, and after standard culture duration. The lysates will be frozen at -80, and shipped on dry ice to Nanostring Technologies, WA for analysis. PZA was tested on all patient samples (n=30), and a second drug will be tested for each patient alternating between MXF (n=12), BPaL (n= 6 for each).

[0227] Power Calculation

[0228] With 60 hold-out in vitro samples, 50% of which are resistant and an expected assay sensitivity of 92%, an 95% CI for sensitivity of 76-99% was expected to be estimated. Similarly with an expected assay specificity of 95%, it could be estimated that 95% CI for specificity at 80-100% (estimated using PropCI R package). Among the 30 study participants, it may be expected 50% to be PZA R (n=15). The prospective evaluation performed here is only a pilot analysis and focused on assessing the potential of the assay for use at the POC. It may be expected 80% power to detect a reduction in sensitivity to PZA resistance from 92% in culture to <61% directly from sputum (-28%), assuming two sided test a<0.05 (estimated using power.prop.test() in R)

[0229] Pitfalls and Contingencies

[0230] A potential challenge is the low abundance of mRNA targets in sputum after antibiotic exposure. If Nanostring and RT-qPCR are not successful on sputum, ddPCR may be tested given the reported higher reproducibility of ddPCR96, and its success in capturing the RS ratio7. A ddPCR instrument is available at the SAMRC within 1 mile of SU (see equipments). Spike-in experiments of bacilli into sputum may be tried and variations on preprocessing e.g. pure mechanical sheering to liquefy rather than NALC to troubleshoot this step. Also, control unexposed isolates may not be needed in the final assay workflow, as the use of internal control transcripts may adequately normalize the signal; assessments may be made in secondary analysis if relative transcription for AR vs. control genes within each sample (i.e. without analysis of DE between exposed and unexposed isolates) has discriminatory power using an RF classifier of resistance vs. susceptibility.

[0231] Expected results and future directions

[0232] It may be expected to thoroughly study Mtb transcriptional responses to five drugs and foresee the ability to patent the AR signatures validated in Aim 1.

[0233] References

[0234] 1. Global Tuberculosis Report 2021. https: / / www.who.int / teams / global- tuberculosis-programme / tb-reports / global-tuberculosis-report-2021.

[0235] 2. Small, P. M. & Pai, M. Tuberculosis diagnosis— time for a game change. N. Engl. J. Med. 363, 1070-1071 (2010).

[0236] 3. Farhat, M. R. et al. Genetic Determinants of Drug Resistance in Mycobacterium tuberculosis and Their Diagnostic Value. Am. J. Respir. Crit. Care Med. (2016) doi:10.1164 / rccm.201510-20910C.

[0237] 4. Ektefaie, Y, Dixit, A., Freschi, L. & Farhat, M. R. Globally diverseMycobacterium tuberculosis resistance acquisition: a retrospective geographical and temporal analysis of whole genome sequences. Lancet Microbe 2, e96-el04 (2021).

[0238] 5. Barczak, A. K. et al. RNA signatures allow rapid identification of pathogens and antibiotic susceptibilities. Proc. Natl. Acad. Sci. U. S. A. 109, 6217-6222 (2012).

[0239] 6. Bhattacharyya, R. P. et al. Simultaneous detection of genotype and phenotype enables rapid and accurate antibiotic susceptibility determination. Nat. Med. 25, 1858-1864 (2019).

[0240] 7. Walter, N. D. et al. Mycobacterium tuberculosis precursor rRNA as a measure of treatment- shortening activity of drugs and regimens. Nat. Commun. 12, 2899 (2021).

[0241] 8. Marks, S. M. et al. Treatment Practices, Outcomes, and Costs of Multidrug- Resistant and Extensively Drug-Resistant Tuberculosis, United States, 2005-2007 - Volume 20, Number 5 — May 2014 - Emerging Infectious Diseases journal - CDC. doi:10.3201 / eid2005.131037.

[0242] 9. WHO | Global tuberculosis report 2018. WHO http: / / www.who.int / tb / publications / global_report / en / .

[0243] 10. National Strategy, https: / / www.cdc.gov / drugresistance / federal-engagement-in- ar / national- strategy / index .html .

[0244] 11. Progressing towards TB elimination. European Centre for Disease Prevention and Control http: / / ecdc.europa.eu / en / publications-data / progressing-towards-tb-elimination (2010).

[0245] 12. Boehme, C. C. et al. Rapid molecular detection of tuberculosis and rifampin resistance. N. Engl. J. Med. 363, 1005-1015 (2010).

[0246] 13. Miotto, P. et al. GenoType MTBDRsl performance on clinical samples with diverse genetic background. Eur. Respir. J. Off. J. Eur. Soc. Clin. Respir. Physiol. 40, 690-698 (2012).

[0247] 14. Tagliani, E. et al. Diagnostic Performance of the New Version (v2.0) of GenoType MTBDRsl Assay for Detection of Resistance to Fluoroquinolones and Second- Line Injectable Drugs: a Multicenter Study. J. Clin. Microbiol. 53, 2961-2969 (2015).

[0248] 15. The CRyPTIC Consortium and the 100, 000 Genomes Project. Prediction of Susceptibility to First-Line Tuberculosis Drugs by DNA Sequencing. N. Engl. J. Med. (2018) doi: 10.1056 / NEJMoal 800474.

[0249] 16. Chen, M. L. et al. Beyond multidrug resistance: Leveraging rare variants with machine and statistical learning models in Mycobacterium tuberculosis resistance prediction. EBioMedicine 43, 356-369 (2019).

[0250] 17. Meehan, C. J. et al. Whole genome sequencing of Mycobacterium tuberculosis : current standards and open issues. Vat. Rev. Microbiol. 1-13 (2019) doi:10.1038 / s41579-019-0214-5.

[0251] 18. Kadura, S. et al. Systematic review of mutations associated with resistance to the new and repurposed Mycobacterium tuberculosis drugs bedaquiline, clofazimine, linezolid, delamanid and pretomanid. J. Antimicrob. Chemother. dkaal36 (2020) doi:10.1093 / j ac / dkaa 136.

[0252] 19. Tuberculosis - FIND, https: / / www.finddx.org / tb / .

[0253] 20. Boshoff, H. I. M. et al. The Transcriptional Responses of Mycobacterium tuberculosis to Inhibitors of Metabolism NOVEL INSIGHTS INTO DRUG MECHANISMS OF ACTION. J. Biol. Chem. 279, 40174-40184 (2004).

[0254] 21. Galagan, J. E. et al. The Mycobacterium tuberculosis regulatory network and hypoxia. Nature 499, 178-183 (2013).

[0255] 22. Rustad, T. R. et al. Global analysis of mRNA stability in Mycobacterium tuberculosis. Nucleic Acids Res. 41, 509-517 (2013).

[0256] 23. Boyaci, H., Saecker, R. M. & Campbell, E. A. Transcription initiation in mycobacteria: a biophysical perspective. Transcription 11, 53-65.

[0257] 24. Srinivas, V. et al. Transcriptome signature of cell viability predicts drug response and drug interaction in Mycobacterium tuberculosis. Cell Rep Methods 1, None (2021).

[0258] 25. Oyewole, A. O. et al. COVID-19 Impact on Diagnostic Innovations: Emerging Trends and Implications. Diagnostics 11, 182 (2021).

[0259] 26. Mardian, Y, Kosasih, H., Karyana, M., Neal, A. & Lau, C.-Y. Review of Current COVID-19 Diagnostics and Opportunities for Further Development. Front. Med. 8, (2021).

[0260] 27. SARS-CoV-2 Rapid Colorimetric LAMP Assay Kit | NEB. https: / / www.neb.com / products / e2019-sars-cov-2-rapid-colorimetric-lamp-assay- kit?gclid=CjwKCAjw5s6WBhA4EiwACGncZbMGbdqrVrvOIiKKMbYcT7VwIF7wmRKdtk dYuEA59nnLlXqzXO8WjhoCI8AQAvD_BwE#Product%20Information.

[0261] 28. Moore, K. J. M. et al. Loop-Mediated Isothermal Amplification Detection of SARS-CoV-2 and Myriad Other Applications. J. Biomol. Tech. JBT32, 228-275 (2021).

[0262] 29. Kim, Y. et al. Single-strand RPA for rapid and sensitive detection of SARS- CoV-2 RNA. medRxiv 2020.08.17.20177006 (2020) doi: 10.1101 / 2020.08.17.20177006.

[0263] 30. Broughton, J. P. et al. CRISPR-Cas 12-based detection of SARS-CoV-2. Nat. Biotechnol. 38, 870-874 (2020).

[0264] 31. Kellner, M. J., Koob, J., Gootenberg, J. S., Abudayyeh, O. O. & Zhang, F. SHERLOCK: Nucleic acid detection with CRISPR nucleases. Nat. Protoc. 14, 2986-3012 (2019).

[0265] 32. Lu, S. et al. Fast and sensitive detection of SARS-CoV-2 RNA using suboptimal protospacer adjacent motifs for Casl2a. Nat. Biomed. Eng. 6, 286-297 (2022).

[0266] 33. nCounter Systems Overview. NanoString https: / / nanostring.com / products / ncounter-analysis-system / ncounter-systems-overview / .

[0267] 34. Bhattacharyya, R. P. et al. Rapid identification and phylogenetic classification of diverse bacterial pathogens in a multiplexed hybridization assay targeting ribosomal RNA. Sci. Rep. 9, 1-13 (2019).

[0268] 35. Farhat, M. R. et al. Genomic analysis identifies targets of convergent positive selection in drug-resistant Mycobacterium tuberculosis. Nat. Genet. (2013) doi:10.1038 / ng.2747.

[0269] 36. Farhat, M. R. et al. Fluoroquinolone Resistance Mutation Detection Is Equivalent to Culture-Based Drug Sensitivity Testing for Predicting Multidrug-Resistant Tuberculosis Treatment Outcome: A Retrospective Cohort Study. Clin. Infect. Dis. Off. Publ. Infect. Dis. Soc. Am. 65, 1364-1370 (2017).

[0270] 37. Nebenzahl-Guimaraes, H., Jacobson, K. R., Farhat, M. R. & Murray, M. B. Systematic review of allelic exchange experiments aimed at identifying mutations that confer drug resistance in Mycobacterium tuberculosis. J. Antimicrob. Chemother. (2013) doi:10.1093 / jac / dkt358.

[0271] 38. Farhat, M. R. et al. Concordance of Mycobacterium tuberculosis fluoroquinolone resistance testing: implications for treatment. Int. J. Tuberc. Lung Dis. 19, 339-341 (2015).

[0272] 39. Farhat, M. R. et al. Gyrase Mutations Are Associated with Variable Levels of Fluoroquinolone Resistance in Mycobacterium tuberculosis. J. Clin. Microbiol. 54, 727-733 (2016).

[0273] 40. de Vos, M. et al. Diagnostic Accuracy and Utility of FluoroType MTBDR, a New Molecular Assay for Multidrug-Resistant Tuberculosis. J. Clin. Microbiol. 56, e00531- 18 (2018).

[0274] 41. Dippenaar, A. et al. Diagnostic accuracy of the FluoroType MTB and MTBDR VER 2.0 assays for the centralized high-throughput detection of Mycobacterium tuberculosis complex DNA and isoniazid and rifampicin resistance. Clin. Microbiol. Infect. Off. Publ. Eur. Soc. Clin. Microbiol. Infect. Dis. 27, 1351.el-1351.e4 (2021).

[0275] 42. Derendinger, B. et al. Widespread use of incorrect PCR ramp rate negatively impacts multidrug-resistant tuberculosis diagnosis (MTBDRplus). Sci. Rep. 8, 3206 (2018).

[0276] 43. Barnard, M. et al. Genotype MTBDRsl line probe assay shortens time to diagnosis of extensively drug-resistant tuberculosis in a high-throughput diagnostic laboratory. Am. J. Respir. Crit. Care Med. 186, 1298-1305 (2012).

[0277] 44. Jacobson, K. R. et al. Implementation of genotype MTBDRplus reduces time to multidrug-resistant tuberculosis therapy initiation in South Africa. Clin. Infect. Dis. Off. Publ. Infect. Dis. Soc. Am. 56, 503-508 (2013).

[0278] 45. The diagnostic performance of the GenoType MTBDRplus version 2 line probe assay is equivalent to that of the Xpert MTB / RIF assay - PubMed. https: / / pubmed.ncbi.nlm.nih.gov / 22972826 / .

[0279] 46. Green, A. G. et al. A convolutional neural network highlights mutations relevant to antimicrobial resistance in Mycobacterium tuberculosis. http: / / biorxiv.org / lookup / doi / 10.1101 / 2021.12.06.471431 (2021) doi: 10.1101 / 2021.12.06.471431.

[0280] 47. Groschel, M. I. et al. GenTB: A user-friendly genome-based predictor for tuberculosis resistance powered by machine learning. Genome Med. 13, 138 (2021).

[0281] 48. Vargas, R. et al. Role of Epistasis in Amikacin, Kanamycin, Bedaquiline, and Clofazimine Resistance in Mycobacterium tuberculosis Complex. Antimicrob. Agents Chemother. 65, eO 1164-21 (2021).

[0282] 49. Farhat, M. R. et al. GWAS for quantitative resistance phenotypes in Mycobacterium tuberculosis reveals resistance genes and regulatory regions. Nat. Commun. 10, 1-11 (2019).

[0283] 50. Walker, T. M. et al. The 2021 WHO catalogue of Mycobacterium tuberculosis complex mutations associated with drug resistance: a genotypic analysis. Lancet Microbe 3, e265-e273 (2022).

[0284] 51. Yang, Y. et al. Machine learning for classifying tuberculosis drug-resistance from DNA sequencing data. Bioinforma. Oxf. Engl. 34, 1666-1671 (2018).

[0285] 52. Deb, C. et al. A novel in vitro multiple-stress dormancy model for Mycobacterium tuberculosis generates a lipid-loaded, drug-tolerant, dormant pathogen. PloS One 4, e6077 (2009).

[0286] 53. Gopal, P. et al. Pyrazinamide triggers degradation of its target aspartate decarboxylase. Nat. Commun. 11, 1661 (2020).

[0287] 54. Venkataraman, B., Vasudevan, M. & Gupta, A. A new microarray platform for whole-genome expression profiling of Mycobacterium tuberculosis. J Microbiol Methods 97, 34-43 (2014).

[0288] 55. Boot, M. et al. Accelerating Early Antituberculosis Drug Discovery by Creating Mycobacterial Indicator Strains That Predict Mode of Action. Antimicrob Agents Chemother 62, (2018).

[0289] 56. Karakousis, P. C., Williams, E. P. & Bishai, W. R. Altered expression of isoniazid-regulated genes in drug-treated dormant Mycobacterium tuberculosis. J Antimicrob Chemother 61, 323-31 (2008).

[0290] 57. Koul, A. et al. Delayed bactericidal response of Mycobacterium tuberculosis to bedaquiline involves remodelling of bacterial metabolism. Nat Commun 5, 3369 (2014).

[0291] 58. Jain, P. et al. Dual-Reporter Mycobacteriophages (Phi2DRMs) Reveal Preexisting Mycobacterium tuberculosis Persistent Cells in Human Sputum. mBio 7, (2016).

[0292] 59. Fu, L. M. & Shinnick, T. M. Genome-wide exploration of the drug action of capreomycin on Mycobacterium tuberculosis using Affymetrix oligonucleotide GeneChips. J Infect 54, 277-84 (2007).

[0293] 60. Liang, J. et al. Genome-wide expression profiling of the response to linezolid in Mycobacterium tuberculosis. Curr Microbiol 64, 530-8 (2012).

[0294] 61. Liu, Y. et al. Immune activation of the host cell induces drug tolerance in Mycobacterium tuberculosis both in vitro and in vivo. J Exp Med 213, 809-25 (2016).

[0295] 62. Ma, S. et al. Integrated Modeling of Gene Regulatory and Metabolic Networks in Mycobacterium tuberculosis. PLoS Comput Biol 11, el004543 (2015).

[0296] 63. Peterson, E. J. R., Ma, S., Sherman, D. R. & Baliga, N. S. Network analysis identifies Rv0324 and RvO88O as regulators of bedaquiline tolerance in Mycobacterium tuberculosis. Nat Microbiol 1, 16078 (2016).

[0297] 64. Immanuel, S. R. C. et al. Quantitative prediction of conditional vulnerabilities in regulatory and metabolic networks using PRIME. NPJ Sy st Biol Appl 7, 43 (2021).

[0298] 65. Kumar, P. et al. Synergistic Lethality of a Binary Inhibitor of Mycobacterium tuberculosis KasA. mBio 9, (2018).

[0299] 66. Boshoff, H. I. et al. The transcriptional responses of Mycobacterium tuberculosis to inhibitors of metabolism: novel insights into drug mechanisms of action. J Biol Chem 279, 40174-84 (2004).

[0300] 67. Van den Bossche, A. et al. Transcriptional profiling of a laboratory and clinical Mycobacterium tuberculosis strain suggests respiratory poisoning upon exposure to delamanid. Tuberc. Edinb 117, 18-23 (2019).

[0301] 68. Ma, S. et al. Transcriptomic Signatures Predict Regulators of Drug Synergy and Clinical Regimen Efficacy against Tuberculosis. mBio 10, (2019).

[0302] 69. Briffotaux, J., Liu, S. & Gicquel, B. Genome- Wide Transcriptional Responses of Mycobacterium to Antibiotics. Front. Microbiol. 10, 249 (2019).

[0303] 70. Rosenow, C., Saxena, R. M., Durst, M. & Gingeras, T. R. Prokaryotic RNA preparation methods useful for high density array analysis: comparison of two approaches. Nucleic Acids Res. 29, el 12 (2001).

[0304] 71. Scott, M., Gunderson, C. W., Mateescu, E. M., Zhang, Z. & Hwa, T.Interdependence of cell growth and gene expression: origins and consequences. Science 330, 1099-1102 (2010).

[0305] 72. Koch, A., Mizrahi, V. & Warner, D. F. The impact of drug resistance on Mycobacterium tuberculosis physiology: what can we learn from rifampicin? Emerg. Microbes Infect. 3, el7 (2014).

[0306] 73. WHO | Technical report on critical concentrations for TB drug susceptibility testing of medicines used in the treatment of drug-resistant TB. WHO http: / / www.who.int / tb / publications / 2018AVHO_technical_report_concentrations_TB_drug_s usceptibility / en / .

[0307] 74. Torrea, G. et al. Bedaquiline susceptibility testing of Mycobacterium tuberculosis in an automated liquid culture system. J. Antimicrob. Chemother. 7$, 2300-2305 (2015).

[0308] 75. Ismail, N. A. et al. A Multimethod, Multicountry Evaluation of Breakpoints for Bedaquiline Resistance Determination. Antimicrob. Agents Chemother. 64, e00479-20 (2020).

[0309] 76. Bateson, A. et al. Ancient and recent differences in the intrinsic susceptibility of Mycobacterium tuberculosis complex to pretomanid. J. Antimicrob. Chemother. 77, 1685— 1693 (2022).

[0310] 77. Freschi, L. et al. Population structure, biogeography and transmissibility of Mycobacterium tuberculosis. Nat. Commun. 12, 6099 (2021).

[0311] 78. Chiner-Oms, A. et al. Genome-wide mutational biases fuel transcriptional diversity in the Mycobacterium tuberculosis complex. Nat. Commun. 10, 3994 (2019).

[0312] 79. Babraham Bioinformatics - FastQC A Quality Control tool for High Throughput Sequence Data, https: / / www.bioinformatics.babraham.ac.uk / projects / fastqc / .

[0313] 80. Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinforma. Oxf. Engl. 30, 2114-2120 (2014).

[0314] 81. Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows- Wheeler transform. Bioinforma. Oxf. Engl. 25, 1754-1760 (2009).

[0315] 82. Liao, Y, Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923-930 (2014).

[0316] 83. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2 | Genome Biology | Full Text. https: / / genomebiology.biomedcentral.com / articles / 10.1186 / sl3059-014-0550-8.

[0317] 84. Tjaden, B. A computational system for identifying operons based on RNA-seq data. Methods 176, 62-70 (2020).

[0318] 85. McClure, R. et al. Computational analysis of bacterial RNA-Seq data. Nucleic Acids Res. 41, el40 (2013).

[0319] 86. Spies, D., Renz, P. F., Beyer, T. A. & Ciaudo, C. Comparative analysis of differential gene expression tools for RNA sequencing time course data. Brief. Bioinform. 20, 288-298 (2019).

[0320] 87. Yuan, J. S., Reed, A., Chen, F. & Stewart, C. N. Statistical analysis of realtime PCR data. BMC Bioinformatics 7, 85 (2006).

[0321] 88. Geiss, G. K. et al. Direct multiplexed measurement of gene expression with color-coded probe pairs. Nat. Biotechnol. 26, 317-325 (2008).

[0322] 89. Dixit, A. et al. Estimation of country-specific tuberculosis antibiograms using genomic data. 2021.09.23.21263991 https: / / www.medrxiv.org / content / 10.1101 / 2021.09.23.21263991vl (2021) doi:10.1101 / 2021.09.23.21263991.

[0323] 90. Ye, J. et al. Primer-BLAST: a tool to design target- specific primers for polymerase chain reaction. BMC Bioinformatics 13, 134 (2012).

[0324] 91. Vandesompele, J. et al. Accurate normalization of real-time quantitative RT- PCR data by geometric averaging of multiple internal control genes. Genome Biol. 3, research0034.1 (2002).

[0325] 92. Ismail, N. A. et al. Assessment of epidemiological and genetic characteristics and clinical outcomes of resistance to bedaquiline in patients treated for rifampicin-resistant tuberculosis: a cross-sectional and longitudinal study. Lancet Infect. Dis. 22, 496-506 (2022).

[0326] 93. Nimmo, C. et al. Population-level emergence of bedaquiline and clofazimine resistance-associated variants among patients with drug-resistant tuberculosis in southern Africa: a phenotypic and phylogenetic analysis. Lancet Microbe 1, el65-el74 (2020).

[0327] 94. McIntosh, A. I. et al. Using routinely collected laboratory data to identify high rifampicin-resistant tuberculosis burden communities in the Western Cape Province, South Africa: A retrospective spatiotemporal analysis. PLOS Med. 15, el002638 (2018).

[0328] 95. Myers, B. et al. Impact of alcohol consumption on tuberculosis treatment outcomes: a prospective longitudinal cohort study protocol. BMC Infect. Dis. 18, 488 (2018).

[0329] 96. Taylor, S. C., Laperriere, G. & Germain, H. Droplet Digital PCR versus qPCR for gene expression analysis with low abundant targets: from variable nonsense to publication quality data. Sci. Rep. 7, 2409 (2017).

[0330] Example 2: Transcrip tomic Signature for the identification of Drug- Susceptible Mycobacterium tuberculosis

[0331] Abstract

[0332] Transcriptional responses in bacteria following antibiotic exposure offer insights into antibiotic mechanism of action, bacterial responses, and characterization of antimicrobial resistance. It was an aim to define the transcriptional antibiotic response (TAR) in Mycobacterium tuberculosis (Mtb) isolates for clinically relevant drugs by pooling andanalyzing Mtb microarray and RNA-seq data sets. 99 antibiotic transcription profiles across 17 antibiotics were generated, with 76% of profiles generated using 3-24 hours of antibiotic exposure and 49% within one doubling of the WHO antibiotic critical concentration. TAR genes were time-dependent, and largely specific to the antibiotic mechanism of action. TAR signatures performed well at predicting antibiotic exposure, with the area under the receiver operating curve (AUC) ranging from 0.84-1.00 (TAR <6 hours of antibiotic exposure) and 0.76-1.00 (>6 hours of antibiotic exposure) for upregulated genes and 0.57-0.90 and 0.87- 1.00, respectfully, for downregulated genes. This example demonstrates that transcriptomics allows for the assessment of antibiotic activity in Mtb within 6 hours of exposure.

[0333] Introduction

[0334] Improved diagnosis and treatment are designated priorities by the World Health Organization (WHO) and Centers for Disease Control (CDC) for addressing the global antibiotic resistance challenge1'3. Multidrug-resistant tuberculosis (MDR-TB) treatment is observed in 3.6% of new tuberculosis (TB) cases and 18% of previously treated cases4. Treatment requires the concurrent use of five or more effective antibiotics and costs eight times as much as treatment of drug susceptible TB5. The early diagnosis of TB and universal drug- susceptibility testing are important components of the WHO’s END-TB strategy6.

[0335] Molecular diagnostics that rely on the detection of targeted antibiotic resistance mutations in the Mycobacterium tuberculosis (Mtb) genome have been rapidly integrated into national TB programs and are currently standard-of-care7'10. But available commercial molecular assays have not yet been able to replace the traditional more costly and slow process of growth-based phenotypic drug susceptibility testing (pDST) due to the limited number of drugs tested. pDST is performed after growth is identified from a patient sample and requires several weeks (up to 10-12) to yield a result. Pathogen whole genome sequencing (WGS) is a promising replacement for pDST and is currently performed on culture-positive samples in several high resource clinical laboratories11. The diagnostic accuracy of WGS is excellent for drugs like isoniazid (INH) and rifampicin (RIF), but still lags phenotypic methods for other key drugs like pyrazinamide (PZA) and the novel / repurposed TB drugs12'14. The redundancy in resistance mechanisms combined with the diversity of background genetic variation can make it challenging to tease out causative from neutral variants for these agents10 14. Mtb is also acquiring new mutations under antibiotic pressure including to novel antibiotics like bedaquiline and delamanid since being introduced recently into clinical care15 16. Designing genotype-based diagnostics to these drugs currently requires more data on resistance determinants.

[0336] Diagnosis based on the transcriptomic response to drug exposure can circumvent the challenges of DNA based methods and have the potential to be equally rapid17 18. Like all cells, bacteria including Mtb demonstrate transcriptional changes in response to environmental stressors such as hypoxia or pH changes, and chemical pressure like antibiotic exposure19'24. These responses may be the result of the direct antibiotic effect on a specific set of genes or pathway or may reflect complex or global changes in bacterial physiology21,24. Because transcriptomic methods can detect both coding mutations and transcript abundance, they have the potential to combine phenotypic and genotypic methods into a single test. The transcriptional antibiotic response (TAR) can be used to distinguish between antibioticsensitive and antibiotic -resistant strains of E. coli and is independent of the mechanism of resistance17,24,25. This approach was recently shown to have high sensitivity and specificity for detecting resistance in E. coli within 6 hours of blood cultures turning positive17.

[0337] Compared with E.coli, mycobacteria are notable for an RNA-polymerase that forms unstable open promoter complexes that appear more easily prone to regulation by repressors or stimulating factors26. The Mtb RNA polymerase complex has a much shorter half-life (~1- 2 min) when compared to E. coli, implying a more rapid transcriptional response26. At the same time, Mtb demonstrates longer mRNA half-life on average (9.5 min), with even the shortest half-lives longer than the average transcript half-life in E.coli.18Whereas this may delay the observation of transcriptional changes after they occur, especially for abundant transcripts, the delay will be on the order of minutes rather than hours. Hence, despite Mtb’s substantially slower growth in vitro (doubling time 16-22 hours) than E.coli (doubling time 20 minutes), its rapid transcriptional response offers a much more promising window to predict antibiotic effect than pDST. A major use of transcriptomics in bacteria has been to characterize the mechanism of action of drugs, study the impact of resistance mutations, and predict the response of individual and antibiotic combinations. The transcriptomic antibiotic response (TAR) may result from the direct antibiotic effect on a specific set of genes or pathways or reflect complex or global stress response. The TAR of antibiotic- sensitive bacteria differs from the TAR seen in antibiotic -resistant bacteria and can distinguish between clinical strains of antibiotic-sensitive and antibiotic -resistant E. coli, Klebsiella pneumoniae, Acinetobacter baumanii, Pseudomonas aeruginosa, and Mtb and is independent of the specific DNA mutation encoding resistance to a given drug. Barczak et al. demonstrated that the transcriptional antibiotic response (TAR) observed in susceptible Mtb isolates differs significantly from that in isolates resistant to ciprofloxacin, streptomycin, and isoniazid25. Additionally, the isoniazid TAR was able to differentiate between low-levelisoniazid resistance due to mutations in the inhA promoter vs high-level isoniazid resistance mutations in the katG mutant25. This supports that TAR has the needed characteristics to functionally capture antibiotic resistance in Mtb25.

[0338] Developing transcription-based antimicrobial susceptibility testing for patient care will require the identification of TAR profiles for a wider range of antibiotics, and the evaluation of the effect of antibiotic concentration, and duration of antibiotic exposure. In this work the first aim was to define antibiotic response signatures for 17 clinically important tuberculosis drugs by pooling publicly available M. tuberculosis transcriptomic data. The second aim was to study the extent of shared transcriptional response across drugs and determine the performance of the TAR signatures against control conditions for differentiating antibiotic exposed from unexposed M. tuberculosis to facilitate future diagnostic development using this approach.

[0339] Methods

[0340] Search strategy

[0341] NCBI’s Gene Expression Omnibus (ncbi.nlm.nih.gov / geo / ) and the European Nucleotide Archive were searched using the term “Mycobacterium tuberculosis” and identified microarray and / or RNA sequencing (RNA-Seq) studies datasets available until 30 June 2022 for further review. Studies were included that were performed using strains grown in liquid mycobacterial culture media (Middlebrook 7H9 media) and used both antibiotic- exposed and antibiotic-unexposed conditions. All antibiotic concentrations and duration of culture were included. Studies done using strains grown on solid culture media (Lowenstein- Jensen or Middlebrook agar), in vivo studies with cell lines or mouse studies, experiments that used knockout strains or stress conditions (such as hypoxia, low pH or UV light), and those that used experimental drugs or drugs not in current clinical use for TB treatment were excluded.

[0342] Metadata curation

[0343] The GSE accession ID were identified of studies of interest and the metadata was downloaded for each study (Table Al). Details were extracted on the isolates and their expression data including the accession ID, the Mtb strain name, the antibiotic and concentration of antibiotic, and the duration of culture, 2-channel vs 1 -channel microarray or single-end or pair-end RNA-seq data. Where necessary, the antibiotic concentration was converted from moles to microgram per milliliter (mcg / ml). For each isolate, the ratio of the antibiotic concentration used for that isolate to the WHO critical concentration27was calculated. This metadata was used to create antibiotic or control conditions, with eachcondition a unique permutation of strain, antibiotic (or control), antibiotic concentration, and duration of culture. These terms have been defined in the glossary (please see supplementary text).

[0344] Annotation

[0345] H37Rv reference sequence NC_000962.3 ncbi.nlm.nih.gov / nuccore / NC 000962.3 was used for annotation. For CDC1551 strains, modified H37Rv CDC1551 correspondence table was used in metadata of GSE1642 ncbi.nlm.nih.gov / geo / query / acc.cgi?acc=GSE1642 to convert CDC1551 locus tags to H37Rv locus tags. For those CDC1551 locus tags that mapped to multiple H37Rv locus tags or vice versa, the CDC1551 locus tags was retained. Transcriptomic data from clinical strains were aligned against H37Rv. 384 genes coding for PE / PPE genes, repetitive regions, and insertion sequences and phage related sequences28were excluded.

[0346] Bioinformatics pipelines and generation of transcription profiles

[0347] All analyses were done using R (v4.1.0) and RStudio (v 1.4.1717) with tidyverse (vl.3.1). Transcriptomic data are generated from biological replicates of a specific experimental condition or state. Downstream analysis requires aggregation of data from biological replicates, and differentially expressed genes are identified by comparing transcriptomic data from all biological replicates of two different conditions. Each study was analyzed independently using separate bioinformatics pipelines for microarray and RNA-Seq data to generate antibiotic and control transcriptional profiles (TPs) as detailed in the supplement. As data from individual samples cannot be used for downstream analysis or model building, the TP was used for all further analyses. The antibiotic transcription profile (TP) consisted of all differentially expressed genes (at a false-discovery rate cutoff of 0.1) obtained from comparison of multiple biological replicates of a specific strain of Mtb exposed to a particular antibiotic concentration for a defined duration, compared to the same strain incubated in culture without antibiotic exposure for identical duration. Two types of control TPs to contrast with antibiotic TPs were generated. The first was differential expression from baseline (time zero) comparison of a strain of Mtb exposed to antibiotic vs those not exposed to antibiotic that were expected to reflect noise in inoculum size, and the second was differential expression in antibiotic unexposed strains from two different time points that were expected to reflect temporal effect of growth in culture.

[0348] Microarray analysis

[0349] For microarray analysis R packages GEOquery (v2.6.0) and Limma (v3.48.3) were used. Processed microarray data from 1 -channel and 2-channel array experiments weredownloaded with GEOQuery. Probe intensities were converted to log-2-fold and the mean intensity for genes with multiple probes was used to create a single value for each gene. After removing the genes outlined above, Limma was used to perform quantile normalization across all arrays, followed by differential expression analysis to generate antibiotic and control transcription profiles.

[0350] RNA-Seq analysis

[0351] GSE and SRA accession IDs of identified studies were used to download metadata from SRA explorer (sra-explorer.info / ); FASTQ fdes were downloaded using Aspera, and md5sums were checked. FASTQC (v0.11.5) and multiQC (vl.5) were used for quality control. Reads were aligned to H37Rv with BWA (v0.7.15); Samtools (vl .9) was used to create sorted BAM files and calculate the percent of genome covered. Sorted BAM files were analyzed using featureCounts from Subread (v2.0.3) to count reads with separate flags for single-end and paired-end reads. Isolates with <80 % of genome coverage were excluded. Antibiotic and control transcription profiles were generated using DESeq2 (v 1.32.0), with quality control and analysis workflow adapted from the bulk-RNA Seq analysis by the Harvard Chan Bioinformatic Core29.

[0352] Transcription profile analysis and generation of transcriptional antibiotic response

[0353] TPs generated using single samples without replicates or those where the ratio of the antibiotic concentration to the WHO critical concentration (antibiotic concentration ratio) was under 0.5 were excluded. As the duration of antibiotic exposure was significantly associated with the number of differentially expressed genes in response to antibiotic exposure and because of the heterogeneity in the duration of culture across different studies, the duration of culture was categorized into <3 hours, 3-6 hours, 6-24 hours, and >24 hours, and in parallel to boost power categorized into <6 hours and >6 hours reasoning that <6 hours could be used as a same day assay, and > 6 hours as a next day assay. This allowed exploration of the impact of the duration of antibiotic exposure on gene expression. However, there were not enough TPs within each of these time categories (Table A3) to assess the ability to discriminate between antibiotic from control TPs using a logistic regression model. The <6 and >6 hours categories allowed development of such models. The antibiotic concentration ratio was categorized into <0.5, 1-2, 2-4 and >4 times the WHO critical concentration.

[0354] To control for the distributional skew and heterogeneity in log2fold values across transcriptomic technology and studies from different laboratories, genes within each transcription profile by the magnitude of log2fold expression, assigning the lowest rank to the gene with the highest log2fold value were ranked. The transcription profiles for an antibioticwithin each time category were used to calculate the median rank for each gene across those profiles, and the top 10 ranked genes were selected to create the transcriptional antibiotic response (TAR). This was done separately for upregulated and downregulated genes. Ten genes were chosen to develop the TAR based on previous studies that have used transcriptomics-based susceptibility testing methods.

[0355] On TARs across antibiotics and time categories, certain genes were noted to be common to antibiotics with similar mechanisms of action across multiple time categories or common to multiple antibiotics with different mechanisms of action within the same time category. To explore this further, the median rank for various combinations of antibiotics and genes across all time categories to determine temporal changes related to the mechanism of action for these antibiotic combinations was calculated. Heatmaps were created using R package pheatmap (vl.012) for all antibiotics within each time category. After assigning the value 0 to genes that were not differentially expressed, the median log2fold value for each gene for each antibiotic within a time category was calculated. The median log2fold value was obtained for upregulated and downregulated TARs for each antibiotic within a time category; genes that were significantly expressed in other antibiotics (but not within the TAR or not differentially expressed for that antibiotic) were also included.

[0356] Statistical analysis:

[0357] The number of genes differentially expressed between microarray and RNA-seq data for antibiotic and control TPs were compared using t tests. Linear regression was used to determine relationship between the number of genes differentially expressed as a function of duration of antibiotic exposure, adjusting for the transcriptomic technology used (microarray or RNA-Seq), study (batch effects), antibiotic, and the antibiotic drug concentration. The analysis for control TPs was repeated.

[0358] For antibiotics with four or more TPs within a time category, the classification accuracy of using the TAR signature for that time category to distinguish between antibiotic TPs within that category and all control TPs was determined. The antibiotic and control TPs were filtered to include only log2fold values of the 10 selected TAR genes for simplicity and due to the differences in the number of TPs, the proportion of genes differentially expressed, as well as the magnitude of log2fold value between studies. A random 75-25% split was performed and 75% of the data was used to train a logistic regression model and tested the model classification accuracy on the remaining hold-out 25% of the data. The area under the receiver operating curve (AUC) was calculated as well as the maximum sensitivity and specificity at the optimum decision threshold (Youden’s J-statistic) was calculated using theR package pROC. This cross-validation procedure was repeated ten-times to obtain the mean hold-out AUC and the 95% CI (using the / -distribution) for the AUC, sensitivity, and specificity. Classification accuracy was determined separately for upregulated and downregulated TARs, as well as for a model that combined upregulated and downregulated TARs.

[0359] Results

[0360] Data curation and quality control

[0361] 30 studies with publicly available transcriptomic data that met inclusion criteria (FIG. 13 and Table Al) were identified. Individual experiments were excluded that did not meet the inclusion or quality control criteria for both experimental replicates and microarray / RNA-Seq data quality (FIG. 13, Supplementary text provided below, Table Al). This left data from 562 experiments across 216 conditions (FIG. 13), of which 94% of experiments used laboratory reference strain H37Rv. Data from these experiments were used to generate 192 TPs. 18 TPs were further excluded where the antibiotic concentration used was < 0.5 the WHO critical concentration for that antibiotic, 5 TPs for antibiotics not recommended for treatment of TB, and 13 TPs that was supported by only one experiment per condition (FIG. 13 and Supplementary text provided below)27,3°. After all exclusions, 99 antibiotic TPs and 57 control TPs from 24 studies were retained for further analysis.

[0362] Transcription profiles

[0363] Of the 99 antibiotic TPs across 17 antibiotics, 33 profiles were from first-line antibiotics, while 46 were from WHO Grade A or B drugs from multidrug-resistant tuberculosis (MDR-TB) (including 8 pretomanid TPs) (Table Al). The duration of exposure ranged from 30 minutes to 96 hours. Most TPs were generated from antibiotic exposure ranging from 3-6 hours (41%) or 6-24 hours (35%) (Table Al, Tables B1-B7, Table S2 and Al, and FIGs. 14A - 14B). FIGS. 14A illustrates a scatterplot of the duration of antibiotic exposure for TPs of each antibiotic, with different colors for microarray (orange) and RNA- seq (blue), and shape of the point representing different strains. FIG. 14B illustrates a stacked bar chart of the count of TPs in each time category by antibiotic. FIG. 14C illustrates a scatterplot of the ratio of antibiotic concentration to the WHO concentration for TPs of each antibiotic, with different colors for microarray (orange) and RNA-seq (blue), and shape of the point representing different strains. FIG. 14D illustrates a stacked bar chart of the count of TPs within each category of the antibiotic concentration to the WHO critical concentration. The experimental drug concentration (expressed as a ratio to the WHO critical concentration) ranged from 0.5 to 50 (Figures 14A and 14C). Forty-nine profiles (49%) reflected antibioticexposure within one doubling of the WHO critical concentration (drug concentration ratio 0.5-2, Table Al and FIGS. 14C and 14D). 57 control TPs were generated: seven from time zero for either antibiotic exposure or unexposed inoculum, and 50 from longitudinal experiments of antibiotic unexposed strains at two different time points ranging from 1-96 hours (Tables B1-B7, Table S4). Tables S3 and S5 contain the antibiotic and control TPs, respectively. The content of the publication entitled “Transcriptomic responses to antibiotic exposure in Mycobacterium tuberculosis” by Poonawala et al. published April 8, 2024 by the American Society for Microbiology in the May 2024, Volume 28, Issue 5 of the Antimicrobial Agents and Chemotherapy Journal, including the supplemental tables and transcription profiles detailed therein, is hereby incorporated by reference in its entirety.

[0364] The mean number of genes differentially expressed was lower for microarray than for RNA-seq for both antibiotic (1,227 vs 2,646, respectively; P < 0.0001) and control TPs (589 vs 1,888, respectively; P < 0.0001). Regression demonstrated that after controlling for the transcriptomic technology used, study / batch effects, antibiotic and antibiotic concentration, a longer duration of exposure was associated with a larger number of differentially expressed genes (p< 0.001). Conditioning on the duration of exposure and the aforementioned experimental variables, no significant effect of antibiotic concentration on the number of differentially expressed genes within the range of evaluated concentrations (p =0.52) was measured. The number of genes differentially expressed in control TPs was also significantly associated with the time interval between control conditions used to generate the TP (P < 0.001) in a regression model controlling for the study / batch and type of transcription technology indicating an additional effect of time on growth or duration of culture independent of antibiotic exposure (p<0.001). The range of median log2fold values differed for microarray and RNA-seq in antibiotic (3.88 to -3.08 vs 6.06 to -4.43, respectively) but not control TPs (3.40 to -4.4 to vs 3.63 to -4.68).

[0365] Transcriptional antibiotic response

[0366] The top 10 most upregulated and the top 10 most downregulated genes were examined based on the pooled median rank of the log2fold values within a time category for each antibiotic and considered this subset the transcriptional antibiotic response (TAR) genes (Tables B1-B7. Below and in Table A2 and FIG. 18the timing of expression observed for some of the top transcripts and their known links to drug or drug resistance mechanisms was reported. FIGS. 18A-D illustrate time category-specific heatmaps constructed using median log2fold values of the upregulated and downregulated TARs for each antibiotic within a time category. Median log2fold values for genes that were significantly expressed in otherantibiotics (but not within the TAR or not differentially expressed for that antibiotic) were also used. The median log2fold values were calculated for each gene across all TPs for each antibiotic after assigning a value of 0 for genes that were not differentially expressed and scaled for each antibiotic within a time category. The heatmaps were constructed for TPs in the (a) less than 3 hours, (b) 3-6 hours, (c) 6-24 hours, and (d) >24 hours time categories. For brevity, the focus was on genes differentially expressed in response to two or more anti-TB drug classes. A detailed description of genes shared across time categories for individual drugs is provided in the supplementary text. (Tables B1-B4, FIGS. 15 and 16A-B, and Table Al) FIG. 15 illustrates antibiotic heatmaps. FIGS. 16A-B explore temporal changes in median ranges for various combinations of antibiotics and genes.

[0367] Upregulated genes shared across drug classes or mechanisms:

[0368] The TAR gene upregulated in response to the largest number of drugs was the heat shock protein hsp, upregulated after 3-6 hours of exposure to pyrazinamide, bedaquiline, streptomycin, ethambutol, amikacin, and after 6-24 hours for bedaquiline, ethambutol, and kanamycin, consistent with its response to multiple types of stress31. Transcripts associated with the lexA DNA damage response to fluoroquinolones and nitroimidazoles (moxifloxacin, levofloxacin, and delamanid, pretomanid respectively) were identified32,33. whiB7, a known transcriptional regulator of eis, that is known to inactivate kanamycin, was noted in the TARs for aminoglycosides, capreomycin, and linezolid, along with Rvl258c, erm (37), eis, and Rv0263c that are known to be regulated by whiB734,35. Transcripts coding for efflux pumps were noted for isoniazid and ethionamide (efpA), pretomanid (mmpL5), bedaquiline, clofazimine (mmpL5 and mmpS5), and linezolid (Rvl258c), suggesting the role of antibiotic efflux as a response to antibiotic exposure.

[0369] One or more transcripts from the iniBAC operon (iniA, iniB, iniC) of which iniA is linked to isoniazid and ethambutol resistance putatively by facilitating the activity of an efflux pump as well as by maintaining plasma membrane structure among the top 10 most upregulated genes in one or more time category for the cell wall inhibitors isoniazid, ethambutol, ethionamide, and pretomanid.36,37. In line with their role in inhibiting mycolic acid biosynthesis, isoniazid and ethionamide both upregulated FAS-II complex genes (kasA, kasB, acpM) and iniBAC (iniB, iniA) operons at 3-6 hours of exposure. One or more TARs for bedaquiline and clofazimine included members of the mbt-1 locus consisting of genes mbtA- mbtJ responsible for the synthesis of the siderophore mycobactin, a molecule necessary for the intracellular transport of extracellular iron and for intracellular survival of Mtb in host cells38,39. The 6-24 hour TAR for bedaquiline and 3-6 TAR for clofazimine also includedRvO678 which is transcriptional repressor of the Mmps5-Mmpl5 efflux pump that is involved in the export of mycobactins; mutations in Rv0678 confer resistance to azoles, bedaquiline, and clofazimine through drug export by the Mmps5-Mmpl5 efflux pump40'43. The primary activity of Mmpl5 and Mmps5 is export of the siderophores synthesized by the mbt-1 and mbt-2 loci,' interruption of its activity leads to toxicity from the intracellular accumulation of iron 44.

[0370] Downregulated genes shared across drug classes or mechanisms:

[0371] The genes identified in downregulated TARs were consistent with a global response to antibiotic effect rather than to a specific pathway or mechanism of resistance for that antibiotic. Genes from the ESX family and the mcel operon along with chaperone genes groES and groEL2 were seen in antibiotic TARs across multiple time categories. The ESX family is involved in virulence and six genes from this family (esxR, esxH, esxQ, esxG, esxS, esxM) are downregulated in response to levofloxacin, amikacin, streptomycin, capreomycin, rifampicin, rifapentine, pyrazinamide and linezolid45. mcel operon genes also code for proteins involved in virulence, and were noted in TARs for pyrazinamide, linezolid, pretomanid, clofazimine, streptomycin, ethionamide, and ethambutol46. Stress response chaperone proteins groES and groEL2 were present in TARs for pretomanid, delamanid, rifapentine, linezolid, ethambutol, and capreomycin. The presence of genes coding for 30S and 50S ribosomal subunit proteins and ribonuclease P protein (rplE, rplW, rplB, rplV, rplP, rplE, rplR, rpsNl, rpsH, rpsC, rpsS, rpsE, rpsQ, rpmD, rpmC, rnpA) in the signatures for pretomanid, delamanid, bedaquiline, and linezolid47were identified.

[0372] The downregulation in genes involved in respiration in all time categories was noted. Three NADH dehydrogenases (Ndh, NdhA, Nuo) responsible for NADH oxidation as part of the mycobacterial respiratory chain have been identified as potential drug targets48,49. The presence of genes from the nuo family (nuoM, nuoL, nuol, nuoH, nuoB, nuoC, nuoD, nuoF, nuoE) for isoniazid, ethionamide, linezolid, bedaquiline, moxifloxacin, capreomycin, kanamycin, streptomycin, amikacin, rifampicin, and genes involved in ATP synthase (atpB, atpC, atpD, alpF) for pyrazinamide, capreomycin, amikacin and levofloxacin were identified. The > 24 hour signature was striking for the presence of one or more genes involved hypoxic stress response and dormancy genes (hspX, tgsl, Rv3131, Rvl738, Rv2030c, hrpl, fdxA ) for six of the seven antibiotics (bedaquiline, isoniazid, moxifloxacin, pretomanid, linezolid and clofazimine)50'54. The delamanid TARs from earlier time periods (3-6 hours and 6-24 hours) also had multiple genes associated with the hypoxic response.

[0373] Transcripts from desAl and desA2 genes involved in mycolic acid desaturation werenoted for isoniazid and ethionamide55-57. Resuscitation-promoting factors (rpfA-E) are virulence factors that are required for Mtb recovering from a dormant state58. rpfB, rfpC, or rpfE are downregulated in response to the cell wall inhibitors: isoniazid, ethionamide, pretomanid, and delamanid, and were also noted in a single bedaquiline TAR Downregulation of kasB.

[0374] Antibiotic TAR performance

[0375] Given the strong temporality of the transcriptional signal and the lesser effect of drug concentration, two antibiotic TAR signatures were generated for prediction of antibiotic susceptibility by pooling across drug concentrations for the time intervals < 6 hours and > 6 hours respectively. (Table A3, FIGS. 17A-B). The focus was on the drugs isoniazid, bedaquiline, pyrazinamide, clofazimine, moxifloxacin, pretomanid, and linezolid that had at least 4 TPs for either time interval. There were a median of 7 TPs / drug for these agents for the <6 hour interval, and a median of 5 TPs / drug for the >6 hour time interval. A logistic regression model was built that used the log2fold expression values of the top 10 upregulated and downregulated genes for a particular time interval and drug to distinguish drug exposure from all pooled controls (i.e. both longitudinal and time zero controls) as a more extreme test of classification specificity. 10-fold 75%-25% splits of the data were used for training and cross-validation (Methods). The upregulated TAR signatures performed well with AUC for predicting exposure vs no exposure ranging from a minimum 0.84 to 1.00 across antibiotics for the < 6 hour time category, and 0.76-1.00 for the >6 hour time category (FIG. 17A, Table A3). The downregulated TAR had comparable AUC for most drugs in both time points (FIG. 17B, Table A3). Combining up- and downregulated TAR signatures did not consistently improve AUC over the upregulated or downregulated TAR alone across drugs and time categories. Using TARs of either 5 or 15 genes did not significantly change performance, (see Table S9, FIGS. 17A-B). FIGS. 17A-B illustrate performance of 5-gene and 15-gene signatures. The area under the receiver operating curve (AUC) used to determine the classification accuracy of a logistic regression model in distinguishing antibiotic from control TPs for antibiotics with at least four TPs in the (A) less than 6 hours and (B) greater than 6 hours time category is shown in FIGS. 17A-B. The position of the bars correspond to model constructed using genes from upregulated (left), downregulated (center), or combined (right) TAR.

[0376] For the antibiotics delamanid, streptomycin, and rifampicin there were fewer than 5 TPs in total per drug, so a model for resistance classification was built by pooling TPs across the <6 hours or > 6 hours time categories. The AUC for predicting antibiotic exposure was0.99-1, 0.82-0.85, and 0.68-0.71 for streptomycin, delamanid and rifampicin respectively (Table S9, Table A3).

[0377] Discussion

[0378] Microarray and RNA-Seq data from 562 isolates across 17 antibiotics were analyzed to identify Mtb gene expression signatures useful for identifying in vitro susceptibility to antibiotics. Transcriptionally responsive genes were demonstrated to vary by antibiotic and over time. Time appears to affect the number of differentially expressed genes to a larger extent than drug concentration in the available data. While genes common to TAR signatures across and between antibiotic classes were found, the majority of TAR genes were drug specific. TAR signatures performed well at discriminating gene expression due to antibiotic effect form Mtb growth in the absence of antibiotic. The results contribute significantly to the literature on this top as all publicly available data to date was pooled. Drugs currently approved for the treatment of TB were focused on, including many novel drugs for which the molecular mechanisms of resistance are unclear. The culture media, time, and antibiotic concentration categories studied also reflect the expected workflow of a clinical microbiology laboratory.

[0379] The upregulated TARs largely included genes related to the mechanism of action rather than genes associated with resistance mutations for antibiotics isoniazid (katG and inhA promoter), ethionamide (inhA promoter), and fluoroquionolones (gyrA and gyrB). Genes known to be associated with resistance were seen in the TARs for bedaquiline and clofazimine (Rv0678), kanamycin (whiB7). and capreomycin (eis and whiB7y however, these genes were also seen in pretomanid and pyrazinamide (Rv0678), delamanid (ezs), and linezolid (eis and whiB7y which are not known to be associated with Mtb resistance to these antibiotics. Similarly, whiB6. associated with Mtb resistance to amikacin, streptomycin, and capreomycin, was not present in the TAR for these drugs but was noted in the TARs for isoniazid, pretomanid, and linezolid. Mutations in the efflux pump coded by Rv 1258c are associated with resistance to isoniazid, streptomycin, and pyrazinamide; however, it was only identified in the TARs for linezolid along with other genes (eis and Rv0263c) induced by whiB7. Thus, the TAR likely reflects drug- specific mechanisms of action that result in characteristic perturbations in cellular processes. These data confirm that transcriptomic responses enable a better understanding of the drug’s mechanism of action. Although isolates with drug resistance mutations were not studied, the overlap of TAR genes with known drug resistance encoding genes suggests that the transcriptomic response to antibiotics will enable the characterization of phenotypic and genotypic resistance in M. tuberculosis.

[0380] The transcriptional response to antibiotic exposure had notable overlap across drug classes. Differential expression was observed, both upregulation and downregulation, of certain stress response genes including heat shock (hsp), DNA repair (lexA -regulated), stress response chaperone proteins {groES and groEL2) and hypoxic response and dormancy genes (hspX) across multiple antibiotics. These pathways demonstrated the most overlap across drug classes suggesting their value for diagnostic use. Downregulation of certain virulence genes including ESX and mcel genes, aerobic respiration (nuo and alp) and ribosomal protein subunits (rplL. rplW, rplB, rplV, rplP, rplE, rplR, rpsNl, rpsH, rpsC, rpsS, rpsE, rpsQ, rpmD, rpmC, mpA) across drug classes were also observed. Efflux pumps also featured highly among upregulated genes across classes, but the specific efflux genes differed by class. These may reflect global responses in bacterial physiology due to antibiotic effect. Use of a single TAR to demonstrate antibiotic activity among agents of the same class or multiple classes may be possible. The overall patterns of differential expression are consistent with the published literature for the effect of specific antibiotics including studies from which data was not available publicly for inclusion into this meta-analysis and in the sputum of patients treated for drug- susceptible tuberculosis5 U9’62.

[0381] The results suggest overlap in the TARs for isoniazid and ethionamide that both target the terminal step in mycolic acid biosynthesis catalyzed by the protein product of inhA. For both agents, upregulation of FAS-II and iniBAC genes in the H37Rv reference strain was observed. Consistent with the potential of these genes for use in resistance diagnosis, upregulation of FAS-II genes was observed in response to ethionamide but not to isoniazid in isolates with katG mutations causing isoniazid resistance63. katG encodes an enzyme needed to activate the prodrug isoniazid to its drug form and single point mutations in katG can inhibit isoniazid activation64. Isoniazid cannot inhibit inhA in these isolates, whereas ethionamide can63. Transcription based diagnostics hence hold promise in assessing crossresistance between agents of the same class, potentially with the same TAR signature.

[0382] The study was not without limitations. As this is a meta-analysis transcriptional changes were measured across a heterogeneous set of antibiotic concentrations and duration of time and some antibiotics were not well represented in the available data. There was not enough data to identify class-specific TARs for aminoglycosides and fluoroquinolones. The selection of 10 genes is arbitrary albeit based on previous work with gram- negative pathogens; the optimal number of genes may differ. Most experiments were performed using the reference strain H37Rv, and the impact of background genetic variation and evaluation the role of lineage in antibiotic response was not possible. Assessment of the performance oftranscriptional response for predicting antibiotic response is preliminary as it used cross- validation without the ability to validate on independent hold-out test dataset due to the dataset size available. Assessment of performance also assumes that the transcriptional profile of resistant isolates exposed to drug is similar to that of a susceptible isolate unexposed to the same drug. Although one prior study confirmed that this assumption holds for three drugs and mechanisms of resistance, it may not hold true for all drugs25. Transcriptional data on exposed resistant isolates were not available on NCBI / EBI at the time of the study. Despite this limitation, the approach to identifying the top ranked genes and creating antibiotic signatures produced results consistent with the published literature. In the future, the TAR signatures should be validated directly using transcriptional profiles generated from exposed resistant isolates.

[0383] In conclusion, the study demonstrates the potential of transcriptional response signatures for use as an accurate drug- susceptibility test for Mycobacterium tuberculosis within 6 hours of antibiotic exposure. The overall patterns of differential expression are consistent with the published literature for the effect of specific antibiotics, including studies from which data were not available publicly for inclusion in this meta-analysis and in the sputum of patients treated for drug- susceptible tuberculosis. This approach allows for the characterization of mechanisms of action of new compounds that affect a range of cellular processes. Because transcriptomic methods can detect both coding mutations and transcript abundance, they have the potential to combine phenotypic and genotypic methods into a single test. Compared with DNA-based diagnostics this approach lessens the dependence on a comprehensive catalogue of resistance mutations. Further work is needed to confirm the accuracy of this approach including the study of more genetically diverse clinical strains, the standardization of growth conditions and antibiotic concentrations, and refining and validating the signature using resistant strains of Mycobacterium tuberculosis.

[0384] References

[0385] 1. Organization WH. Global tuberculosis report 2018. Geneva: World Health Organization, 2018.

[0386] 2. <101111_SPR_Progressing_towards_TB_elimination.pdf>.

[0387] 3. <carb_national_strategy.pdf>.

[0388] 4. Organization WH. Global tuberculosis report 2022. Geneva: World Health Organization.

[0389] 5. Marks SM, Flood J, Seaworth B, et al. Treatment practices, outcomes, and costs of multidrug-resistant and extensively drug-resistant tuberculosis, United States, 2005- 2007. Emerg Infect Dis 2014;20(5):812-21. DOI: 10.3201 / eid2005.131037.

[0390] 6. Organization WH. Implementing the End TB Strategy: the essentials. Geneva: World Health Organization, 2015.

[0391] 7. Boehme CC, Nabeta P, Hillemann D, et al. Rapid molecular detection of tuberculosis and rifampin resistance. N Engl J Med 2010;363(l l):1005-15. DOI: 10.1056 / NEJMoa0907847.

[0392] 8. Miotto P, Cabibbe AM, Mantegani P, et al. GenoType MTBDRsl performance on clinical samples with diverse genetic background. Eur Respir J 2012;40(3):690-8. DOI: 10.1183 / 09031936.00164111.

[0393] 9. Tagliani E, Cabibbe AM, Miotto P, et al. Diagnostic Performance of the New Version (v2.0) of GenoType MTBDRsl Assay for Detection of Resistance to Fluoroquinolones and Second-Line Injectable Drugs: a Multicenter Study. J Clin Microbiol 2015;53(9):2961-9. DOI: 10.1128 / JCM.01257-15.

[0394] 10. Schon T, Miotto P, Koser CU, Viveiros M, Bottger E, Cambau E.Mycobacterium tuberculosis drug-resistance testing: challenges, recent developments and perspectives. Clin Microbiol Infect 2017;23(3):154-160. (In eng). DOI: 10.1016 / j.cmi.2016.10.022.

[0395] 11. The CRyPTIC Consortium and the 100 GP, Allix-Beguec C, Arandjelovic I, et al. Prediction of Susceptibility to First-Line Tuberculosis Drugs by DNA Sequencing. N Engl J Med 2018;379(15):1403-1415. DOI: 10.1056 / NEJMoal 800474.

[0396] 12. Meehan CJ, Goig GA, Kohl TA, et al. Whole genome sequencing of Mycobacterium tuberculosis: current standards and open issues. Nat Rev Microbiol 2019; 17(9):533-545. DOI: 10.1038 / s41579-019-0214-5.

[0397] 13. Kadura S, King N, Nakhoul M, et al. Systematic review of mutations associated with resistance to the new and repurposed Mycobacterium tuberculosis drugs bedaquiline, clofazimine, linezolid, delamanid and pretomanid. J Antimicrob Chemother 2020;75(8):2031-2043. (In eng). DOI: 10.1093 / jac / dkaal36.

[0398] 14. Chen ML, Doddi A, Royer J, et al. Beyond multidrug resistance: Leveraging rare variants with machine and statistical learning models in Mycobacterium tuberculosis resistance prediction. EBioMedicine 2019;43:356-369. DOI: 10.1016 / j.ebiom.2019.04.016.

[0399] 15. Omar SV, Ismail F, Ndjeka N, Kaniga K, Ismail NA. Bedaquiline-Resistant Tuberculosis Associated with Rv0678 Mutations. N Engl J Med 2022;386(l):93-94. DOI: 10.1056 / NEJMc2103049.

[0400] 16. Bloemberg GV, Keller PM, Stucki D, et al. Acquired Resistance to Bedaquiline and Delamanid in Therapy for Tuberculosis. N Engl J Med 2015;373(20):1986- 8. DOI: 10.1056 / NEJMcl505196.

[0401] 17. Bhattacharyya RP, Bandyopadhyay N, Ma P, et al. Simultaneous detection of genotype and phenotype enables rapid and accurate antibiotic susceptibility determination. Nat Med 2019;25(12): 1858-1864. DOI: 10.1038 / s41591-019-0650-9.

[0402] 18. Rustad TR, Minch KJ, Brabant W, et al. Global analysis of mRNA stability in Mycobacterium tuberculosis. Nucleic Acids Res 2013;41(l):509-17. DOI: 10.1093 / nar / gksl019.

[0403] 19. Khil PP, Camerini-Otero RD. Over 1000 genes are involved in the DNA damage response of Escherichia coli. Mol Microbiol 2002;44(l):89-105. DOI: 10.1046 / j.1365-2958.2002.02878.x.

[0404] 20. Lin JT, Connelly MB, Amolo C, Otani S, Yaver DS. Global transcriptional response of Bacillus subtilis to treatment with subinhibitory concentrations of antibiotics that inhibit protein synthesis. Antimicrob Agents Chemother 2005;49(5): 1915-26. DOI: 10.1128 / AAC.49.5.1915-1926.2005.

[0405] 21. Mitosch K, Bollenbach T. Bacterial responses to antibiotics and their combinations. Environ Microbiol Rep 2014;6(6):545-57. DOI: 10.1111 / 1758-2229.12190.

[0406] 22. Shaw KJ, Miller N, Liu X, et al. Comparison of the changes in global gene expression of Escherichia coli induced by four bactericidal agents. J Mol Microbiol Biotechnol 2003; 5 (2): 105-22. DOI: 10.1159 / 000069981.

[0407] 23. Boshoff HI, Myers TG, Copp BR, McNeil MR, Wilson MA, Barry CE, 3rd.The transcriptional responses of Mycobacterium tuberculosis to inhibitors of metabolism: novel insights into drug mechanisms of action. J Biol Chem 2004;279(38):40174-84. (In eng). DOI: 10.1074 / jbc.M406796200.

[0408] 24. Sangurdekar DP, Srienc F, Khodursky AB. A classification based framework for quantitative description of large-scale microarray data. Genome Biol 2006;7(4):R32. DOI: 10.1186 / gb-2006-7-4-r32.

[0409] 25. Barczak AK, Gomez JE, Kaufmann BB, et al. RNA signatures allow rapid identification of pathogens and antibiotic susceptibilities. Proc Natl Acad Sci U S A 2012;109(16):6217-22. (In eng). DOI: 10.1073 / pnas.1119540109.

[0410] 26. Boyaci H, Saecker RM, Campbell EA. Transcription initiation in mycobacteria: a biophysical perspective. Transcription 2020;l l(2):53-65. DOI: 10.1080 / 21541264.2019.1707612.

[0411] 27. Technical Report on critical concentrations for drug susceptibility testing of medicines used in the treatment of drug-resistant tuberculosis. Geneva: World Health Organization, 2018.

[0412] 28. Coscolla M, Copin R, Sutherland J, et al. M. tuberculosis T Cell Epitope Analysis Reveals Paucity of Antigenic Variation and Identifies Rare Variable TB Antigens. Cell Host Microbe 2015; 18(5):538-48. DOI: 10.1016 / j.chom.2015.10.008.

[0413] 29. Mistry M, Piper M, Liu J, Khetani R. hbctraining / DGE_workshop_salmon_online: Differential Gene Expression Workshop Lessons from HCBC (first release). Zenodo; 2021.

[0414] 30. Technical report on critical concentrations for drug susceptibility testing of isoniazid and the rifamycins (rifampicin, rifabutin and rifapentine). Geneva: World Health Organization, 2021.

[0415] 31. Stewart GR, Wemisch L, Stabler R, et al. Dissection of the heat-shock response in Mycobacterium tuberculosis using mutants and microarrays. Microbiology (Reading) 2002;148(Pt 10):3129-3138. DOI: 10.1099 / 00221287-148-10-3129.

[0416] 32. Davis EO, Dullaghan EM, Rand L. Definition of the mycobacterial SOS box and use to identify LexA -regulated genes in Mycobacterium tuberculosis. J Bacteriol 2002;184(12):3287-95. DOI: 10.1128 / JB.184.12.3287-3295.2002.

[0417] 33. Smollett KL, Smith KM, Kahramanoglou C, Amvig KB, Buxton RS, Davis EO. Global analysis of the regulon of the transcriptional repressor LexA, a key component of SOS response in Mycobacterium tuberculosis. J Biol Chem 2012;287(26):22004-14. DOI: 10.1074 / jbc.Ml 12.357715.

[0418] 34. Hurst-Hess K, Rudra P, Ghosh P. Mycobacterium abscessus WhiB7 Regulates a Species-Specific Repertoire of Genes To Confer Extreme Antibiotic Resistance.Antimicrob Agents Chemother 2017;61(l l). DOI: 10.1128 / AAC.01347-17.

[0419] 35. Morris RP, Nguyen L, Gatfield J, et al. Ancestral antibiotic resistance in Mycobacterium tuberculosis. Proc Natl Acad Sci U S A 2005;102(34):12200-5. DOI: 10.1073 / pnas.0505446102.

[0420] 36. Colangeli R, Helb D, Sridharan S, et al. The Mycobacterium tuberculosis iniA gene is essential for activity of an efflux pump that confers drug tolerance to both isoniazidand ethambutol. Mol Microbiol 2005;55(6): 1829-40. DOI: 10.1111 / j.1365- 2958.2005.04510.x.

[0421] 37. Wang M, Guo X, Yang X, et al. Mycobacterial dynamin-like protein IniA mediates membrane fission. Nat Commun 2019;10(l):3906. DOI: 10.1038 / s41467-019- 11860-z.

[0422] 38. Quadri LEN, Sello J, Keating TA, Weinreb PH, Walsh CT. Identification of a Mycobacterium tuberculosis gene cluster encoding the biosynthetic enzymes for assembly of the virulence-conferring siderophore mycobactin. Chemistry & Biology 1998;5(11):631-645.DOI: https: / / doi.org / 10.1016 / S1074-5521(98)90291-5.

[0423] 39. Baatjies L, Loxton AG, Williams MJ. Host and Bacterial Iron Homeostasis, an Underexplored Area in Tuberculosis Biomarker Research. Front Immunol 2021; 12:742059. DOI: 10.3389 / fimmu.2021.742059.

[0424] 40. Milano A, Pasca MR, Provvedi R, et al. Azole resistance in Mycobacterium tuberculosis is mediated by the MmpS5-MmpL5 efflux system. Tuberculosis (Edinb) 2009;89(l):84-90. (In eng). DOI: 10.1016 / j.tube.2008.08.003.

[0425] 41. Hartkoorn RC, Uplekar S, Cole ST. Cross-resistance between clofazimine and bedaquiline through upregulation of MmpL5 in Mycobacterium tuberculosis. Antimicrob Agents Chemother 2014;58(5):2979-81. DOI: 10.1128 / AAC.00037-14.

[0426] 42. Ismail NA, Omar SV, Joseph L, et al. Defining Bedaquiline Susceptibility, Resistance, Cross-Resistance and Associated Genetic Determinants: A Retrospective Cohort Study. EBioMedicine 2018;28:136-142. (In eng). DOI: 10.1016 / j.ebiom.2018.01.005.

[0427] 43. Andries K, Villellas C, Coeck N, et al. Acquired resistance of Mycobacterium tuberculosis to bedaquiline. PLoS One 2014;9(7):el02135. DOI:10.1371 / journal. pone.0102135.

[0428] 44. Jones CM, Wells RM, Madduri AV, et al. Self-poisoning of Mycobacterium tuberculosis by interrupting siderophore recycling. Proc Natl Acad Sci U S A 2014; 111(5): 1945-50. DOI: 10.1073 / pnas.l311402111.

[0429] 45. Grbschel MI, Sayes F, Simeone R, Majlessi L, Brosch R. ESX secretion systems: mycobacterial evolution to counter host immunity. Nature Reviews Microbiology 2016;14(l l):677-691. DOI: 10.1038 / nrmicro.2016.131.

[0430] 46. Arruda S, Bomfim G, Knights R, Huima-Byron T, Riley LW. Cloning of an M. tuberculosis DNA fragment associated with entry and survival inside cells. Science 1993;261(5127): 1454-7. (In eng). DOI: 10.1126 / science.8367727.

[0431] 47. Kapopoulou A, Lew JM, Cole ST. The MycoBrowser portal: a comprehensive and manually annotated resource for mycobacterial genomes. Tuberculosis (Edinb) 2011;91(1):8-13. (In eng). DOI: 10.1016 / j.tube.2010.09.006.

[0432] 48. Plocinska R, Wasik K, Plocinski P, et al. The Orphan Response RegulatorRv3143 Modulates the Activity of the NADH Dehydrogenase Complex (Nuo) in Mycobacterium tuberculosis via Protein-Protein Interactions. Front Cell Infect Microbiol 2022; 12:909507. DOI: 10.3389 / fcimb.2022.909507.

[0433] 49. Vilcheze C, Weinrick B, Leung LW, Jacobs WR, Jr. Plasticity of Mycobacterium tuberculosis NADH dehydrogenases and their role in virulence. Proc Natl Acad Sci U S A 2018;115(7):1599-1604. DOI: 10.1073 / pnas.1721545115.

[0434] 50. Deb C, Lee CM, Dubey VS, et al. A novel in vitro multiple-stress dormancy model for Mycobacterium tuberculosis generates a lipid-loaded, drug-tolerant, dormant pathogen. PLoS One 2009;4(6):e6077. DOI: 10.1371 / joumal.pone.0006077.

[0435] 51. Galagan JE, Minch K, Peterson M, et al. The Mycobacterium tuberculosis regulatory network and hypoxia. Nature 2013;499(7457): 178-83. (In eng). DOI: 10.1038 / nature 12337.

[0436] 52. Mushtaq K, Sheikh JA, Amir M, Khan N, Singh B, Agrewala JN. Rv2031c of Mycobacterium tuberculosis: a master regulator of Rv2028-Rv2031 (HspX) operon. Front Microbiol 2015;6:351. DOI: 10.3389 / fmicb.2015.00351.

[0437] 53. Walter ND, Dolganov GM, Garcia BJ, et al. Transcriptional Adaptation of Drug-tolerant Mycobacterium tuberculosis During Treatment of Human Tuberculosis. J Infect Dis 2015;212(6):990-8. DOI: 10.1093 / infdis / jivl49.

[0438] 54. Garton NJ, Waddell SJ, Sherratt AL, et al. Cytological and transcript analyses reveal fat and lazy persister-like bacilli in tuberculous sputum. PLoS Med 2008;5(4):e75.DOI: 10.1371 / journal.pmed.0050075.

[0439] 55. Chang Y, Fox BG. Identification of Rv3230c as the NADPH oxidoreductase of a two-protein DesA3 acyl-CoA desaturase in Mycobacterium tuberculosis H37Rv.Biochemistry 2006;45(45): 13476-86. DOI: 10.1021 / bi0615285.

[0440] 56. Singh A, Varela C, Bhatt K, et al. Identification of a Desaturase Involved in Mycolic Acid Biosynthesis in Mycobacterium smegmatis. PLoS One 2016;l l(10):e0164253. DOI: 10.1371 / journal.pone.0164253.

[0441] 57. Bailo R, Radhakrishnan A, Singh A, Nakaya M, Fujiwara N, Bhatt A. The mycobacterial desaturase DesA2 is associated with mycolic acid biosynthesis. Sci Rep 2022;12(l):6943. DOI: 10.1038 / s41598-022-10589-y.

[0442] 58. Kana BD, Gordhan BG, Downing KJ, et al. The resuscitation-promoting factors of Mycobacterium tuberculosis are required for virulence and resuscitation from dormancy but are collectively dispensable for growth in vitro. Mol Microbiol 2008;67(3):672-84. DOI: 10.1111 / j.l365-2958.2007.06078.x.

[0443] 59. Waddell SJ, Stabler RA, Laing K, Kremer L, Reynolds RC, Besra GS. The use of microarray analysis to determine the gene expression profiles of Mycobacterium tuberculosis in response to anti-bacterial compounds. Tuberculosis (Edinb) 2004;84(3- 4):263-74. (In eng). DOI: 10.1016 / j.tube.2003.12.005.

[0444] 60. Michael Wilson*f JD, Hans-Henrik Kristensen^[fi^, Paul Imboden^**,Sangeeta Ranc'][, Patrick O. Brown:;:, and Gary K. Schoolnik*^. Exploring drug-induced alterations in gene expression in Mycobacterium tuberculosis by microarray hybridization. PNAS 1999;96(22): 12833-12838.

[0445] 61. Fu LM, Shinnick TM. Understanding the action of INH on a highly INH- resistant Mycobacterium tuberculosis strain using Genechips. Tuberculosis (Edinb) 2007;87(l):63-70. DOI: 10.1016 / j.tube.2006.04.001.

[0446] 62. Betts JC, McLaren A, Lennon MG, et al. Signature gene expression profiles discriminate between isoniazid-, thiolactomycin-, and triclosan-treated Mycobacterium tuberculosis. Antimicrob Agents Chemother 2003;47(9):2903-13. (In eng). DOI:10.1128 / AAC.47.9.2903-2913.2003.

[0447] 63. Wilson M, DeRisi J, Kristensen HH, et al. Exploring drug-induced alterations in gene expression in Mycobacterium tuberculosis by microarray hybridization. Proc Natl Acad Sci U S A 1999;96(22): 12833-8. DOI: 10.1073 / pnas.96.22.12833.

[0448] 64. Zhang Y, Heym B, Allen B, Young D, Cole S. The catalase-peroxidase gene and isoniazid resistance of Mycobacterium tuberculosis. Nature 1992;358(6387):591-3. (In eng). DOI: 10.1038 / 358591a0.

[0449] Example 3: Transcrip tomic Signature for the identification of Drug- SusceptibleMycobacterium tuberculosis

[0450] Background

[0451] Early diagnosis of TB is Pillar 1 of WHO-END TB Strategy. Drug-resistant TB is associated with poor clinical outcomes and is a public health problem. Phenotypic drug susceptibility is slow and technically challenging. Genotypic tests rely on pre-existing knowledge of resistance mutations. Rapid drug- susceptibility test is a WHO research priority.Transcriptomic based diagnostics combine phenotypic and genotypic testing. Stress produces a transcriptomic response, antibiotic exposure results in stress, and drug-resistant bacteria do not show a stress response. The stress response does not depend on mechanism of resistance.

[0452] The hypothesis is that a transcriptional signature of differential gene expression by Mycobacterium tuberculosis in response to antibiotics can be used as a marker of drugsensitivity.

[0453] The study methods and results are illustrated in Figures 9, 10, 11 A, 11B, 12A, 12B, 12C, and 12D. Figures 12A-12D illustrate a Principal Component Analysis using log2fold values of top 10 genes from antibiotic and control TPs.

[0454] The Area under the Receiver Operative Curves (AUC) from logistic regression models distinguishing antibiotic from control transcription profiles is displayed below.

[0455] The candidate genes for diagnostic testing using top 10 genes from TPs < 6 hours are listed below.

[0456] Results

[0457] 562 replicates across 216 permutations of strain, antibiotic, antibiotic concentration, and duration of culture, 33 transcription profiles (TPs) from first-line antibiotics, and 46 TPs from WHO Grade A or B drugs for MDR-TB. Most (77%) TPs from antibiotic exposure ranged from 3-6 hours (41 TPs; 23 microarray) or 6-24 hours (35; 16 microarray). The gene signature for each antibiotic varied by time category with specificity to drug mechanism of action. kasA, kasB, fbpC, iniA, and iniB were upregulated for isoniazid, ethionamide, and ethambutol, and whiB7, lex A. and Rv3074 upregulated for aminoglycosides,fluoroquinolones, and newer drugs like linezolid, pretomanid and delaminid. ESX and the ribosomal family of proteins were the most downregulated across antibiotics.

[0458] Conclusions

[0459] There is significant heterogeneity in the publicly available data. The top 10 differentially expressed genes are able to distinguish between antibiotic and control transcription profiles across multiple antibiotics. Further research using clinical strains and drug-resistant strains is needed.

[0460] The following supplementary texts correspond to Example 2 provided herein.

[0461] Glossary

[0462] Sample: single biological experiment from which RNA was obtained with a unique and stable GEO accession number (GSMxxx).

[0463] Study: A group of related samples with a unique and stable GEO accessional number (GSExxx)

[0464] Replicate: 2 or more samples obtained under identical experimental conditions, each with its own GSE accession number.

[0465] Antibiotic condition: a combination of strain, antibiotic exposure, antibiotic concentration, and duration of culture. Eg. H37Rv exposed to O.lmg / L of isoniazid for 6 hours

[0466] Control condition: Mtb strain grown for defined duration in the absence of antibiotic exposure

[0467] Transcription profile: List of differentially expressed genes (at false-discovery rate cut-off of 0.1) obtained from comparison of all samples belonging to two different conditions.

[0468] Antibiotic transcription profile: Transcription profile obtained by comparing transcriptional data from an antibiotic condition and control condition where the strain and duration of culture were identical for both antibiotic and control

[0469] Control transcription profile: Transcription profile obtained by comparing transcriptional data from two different control conditions.

[0470] Transcriptional antibiotic response: List of 10 genes generated using the median ranks of all TPs for an antibiotic within a single time category

[0471] Details of excluded studies and experiments, including modifications to minimize discarded data

[0472] RNASeq data excluded:1) Experiments with only 1 FASTQ file:GSE165671GSM5047999 Pretomanid 0.7 mcg / ml 24 hours2) Experiments with < 80% genome covered:GSE166501:GSM5073319 Moxifloxacin 0.3 mcg / ml 72 hoursGSE107831:GSM2881210, GSM2881211, GSM2881212 Rifampicin 0.25mcg / ml 4 hours GSM2881228, GSM2881229, GSM2881230 Rifampicin 0.25 mcg / ml 24 hours3) Studies with control and antibiotic culture incubated for different duration: GSE7588, GSE62025, GSE100915Both Limma and DESeq require replicates to estimate the variance and dispersion, respectively, for every gene across all the conditions in a given study. All RNASeq studies had more than 1 experiment for all conditions. However, for microarray studies there were studies with only 1 experiment for all conditions or studies with only 1 experiment for some of the conditions, but more than 1 experiment for the other conditions. To minimize the number of studies and experiments that were discarded, we made certain assumptions about the studies.4) Studies with only 1 experiment per condition: a) We excluded studies GSE86141, GSE41582, GSE53843 as all conditions had only 1 experiment per condition and it would not be possible use Limma. b) For GSE71200, three concentrations (2-, 4-, and 8- fold the MIC) of each antibiotic were tested at a single time point, with only 1 experiment for each concentration. We used these three samples, each with a different concentration, to create a single TP, and assigned the median concentration as the antibiotic concentration for that antibiotic TP.5) Studies with only 1 experiment for certain conditions: a) For GSE1642 we rounded the time in specific experiments to allow them to be combined with experiments from nearby time points - GSM28332 Clofazimine 5 mcg / ml 6.5 hours to 6 hoursGSM27885 Pyrazinamide 120 mcg / ml 2.5 hours to 2 hours GSM27895 Pyrazinamide 120 mcg / ml 5.5 hours to 6 hours GSM27902 Pyrazinamide 1200 mcg / ml 10.5 hours to 11 hours GSM27903 Pyrazinamide 120 mcg / ml 10.5 hours to 11 hours b) For GSE1642 and GSE72459 there were 13 remaining conditions with only 1 experiment per condition. We used data from these experiments while creating the design matrix and contrast matrix to generate transcription profiles using Limma, but subsequently filtered out these 13 transcription profiles from further analysis. GSE1642:GSM28059 Amikacin 10 mcg / ml 6 hoursGSM28220 Clofazimine 10 mcg / ml 6 hours GSM28232 Ethambutol 20 mcg / ml 6 hours GSM28062 Isoniazid 0.2 mcg / ml 6 hours GSM27894 Pyrazinamide 120 mcg / ml 5 hours GSM27902 Pyrazinamide 120 mcg / ml 11 hours (originally 10.5 hours) GSM27905 Pyrazinamide 1200 mcg / ml 16 hours GSM27884 Pyrazinamide 1200 mcg / ml 2 hours GSM27893 Pyrazinamide 1200 mcg / ml 5 hoursGSE72459:GSM1862894 Bedaquiline 0.83 mcg / ml 48 hoursGSM 1862878 Bedaquiline 8.33 mcg / ml 0 hours GSM 1862893 Bedaquiline 8.33 mcg / ml 6 hours GSM 1862897 Bedaquiline 8.33 mcg / ml 96 hours

[0473] Supplemental Figure Legends

[0474] Figure 15: Antibiotic-specific heatmaps constructed using median log2fold values of upregulated and downregulated TARs for each time category; median log2fold values for genes that were significantly expressed in other time categories (but either not within the TAR or not differentially expressed for that time category) are also shown. The median log2fold values were calculated for each gene across all transcription profiles within a time category after assigning a value of 0 for genes that were not differentially expressed and scaled for each time category of an antibiotic. TAR genes are highlighted with an asterisk (*).The heatmaps were constructed for thirteen antibiotics with TARs in two or more time categories. TAR: transcriptional antibiotic response.

[0475] Figures 16A-B: Heatmap of changes in the combined median ranks for antibioticgene combinations across different time categories for A) Upregulated and B) Downregulated genes. Grey cell represent time categories for which antibiotic transcription profiles were not available. Please refer to Table S8 for details of individual genes. AG, aminoglycosides;CAP, capreomycin; PA824, pretomanid; INH, isoniazid; LZD, linezolid; ETA, ethionamide;ETH, ethambutol; BDQ, bedaquiline; CLZ, clofazimine; PZA, pyrazinamide; FQ, fluoroquinolone; MXF, moxifloxacin; LFX, levofloxacin; STP, streptomycin; RIF, rifampin; RPT, rifapentine, DEL, delamanid; FAS-II, fatty acid synthase-II complex; ESX, esx system genes.

[0476] Figures 17A-B: The area under the receiver operating curve (AUC) used to determine classification accuracy of a logistic regression model in distinguishing antibiotic from control transcription profiles using (A) 5 genes and (B) 15 genes, with the color of the bars corresponding to model constructed using genes from the upregulated (green), downregulated (red), or combined (purple) TAR. TAR, transcriptional antibiotic response.

[0477] Supplemental TablesS 1 : Study metadataS2: Antibiotic Transcription Profile MetadataS3: Antibiotic Transcription ProfilesS4: Control Transcription Profile MetadataS5: Control Transcription ProfilesAl: Antibiotic Transcription Profile CategoriesB1,B2,B3,B4: Transcription Antibiotic Response for <3 hours, 3-6 hours, 6-24, and >24 hours antibiotic exposureS8: Temporal changes in antibiotic-drug combinationsS9: Performance of 5-gene 15-gene signatures

[0478] Detailed discussion of upregulated and downregulated genes (please refer figures 3 and SI)

[0479] Upregulated genes

[0480] Cell wall agents'. The iniBAC operon is induced by multiple cell wall inhibitors including isoniaizid. While the precise function of these genes is unknown, iniA is linked toisoniazid and ethambutol resistance putatively by facilitating the activity of an efflux pump as well as by maintaining plasma membrane structure. Two or more transcripts from the iniBAC operon were noted in isoniazid TARs for all time categories. The FAS-II (fatty acid synthase) complex is responsible for the biosynthesis of mycolic acid component of the Mtb cell wall and is inhibited by isoniazid. At least two transcripts from the FAS-II complex operon (fabD, acp, kasA, kasB, accD6) were present in the <3 hours, 3-6 hours, and 6-24 hours category TAR for isoniazid, but absent in the > 24 hours category. Mutations in the iniBAC, fabD, acp, kasA and accD6 have been associated with isoniazid resistance. The TAR genes for ethionamide were notable for similarity to isoniazid, with genes from the iniBAC and FAS-II operons present in both TARs. Apart from iniB, the ethambutol TARs were notable for the presence of hsp (heat shock protein) and RNA polymerase sigma factors sigE and sigB known to be upregulated in response to stress.

[0481] Neither isoniazid nor ethionamide TARs included katG or inhA, both associated with resistance to isoniazid and encodes an enzyme needed to activate the prodrug isoniazid to its drug form, and single point mutations in katG can inhibit isoniazid activation (11). Isoniazid cannot inhibit inhA in these isolates, whereas ethionamide can (12). Upregulation of FAS-II genes was observed in response to ethionamide but not to isoniazid in isolates with katG mutations causing isoniazid resistance (12). Additionally, transcriptomic responses in isoniazid-resistant Mtb strains have previously been able to distinguish between low-level resistance caused by mutations in the inhA promoter and high-level resistance due to katG mutations (13).

[0482] The 3-6 hours and 6-24 hours TAR for delamanid included transcriptional regulatory proteins Rv2488c and ethR, as well as monoxygenese ethA. The latter is known to activate ethionamide from its pro-drug form. The 3-6 hour TAR was notable for the genes lexA, Rv3776, and Rv3074. lexA codes for a transcriptional repressor that mediates the SOS response to DNA damage, and Rv3074 and Rv3776 are regulated by lexA. The pretomanid TARs were notable for presence of iniB and Rv3074 in all time categories. The 3-6 hours TAR included lipid metabolism genes (fadE13, accA2), membrane transport proteins((gZn77, mmpL5) and regulatory proteins (Rv0678). The 6-24 hours TAR was notable for the presence of lexA and associated DNA damage genes (dnaE2. Rvl378c, Rv3776, Rv3074) and oxidoreductases (Rv3740c, Rv3741c, Rv3742c). The TAR from >24 hours also included lexA related genes Rv3370c and Rv3776, as well as transcriptional regulators Rv0196 and whiB6, genes that have been associated with the hypoxic response in Mtb.

[0483] One or more transcripts from the iniBAC operon (ini A, iniB, iniC) was among the top 10 most upregulated genes in one or more time categories for the cell wall acting agents isoniazid, ethambutol, ethionamide, and pretomanid. There was an overlap of 7 genes in the 3-6 hour time category TAR for isoniazid and ethionamide with genes from the FAS-II complex (kasA, kasB, acpM) and iniBAC (iniB, iniA) operons, as well as efpA (efflux pump) and Rvl592 (unknown function). These findings are consistent with previous studies demonstrating temporal differences in gene expression in response to isoniazid and ethionamide, as well as in sputa from patients treated with anti-TB medications including isoniazid. Both isoniazid and pretomanid contained transcripts from transcriptional regulators whiB6 and Rv0196 in their > 24 hours TAR. lexA and associated genes Rv3074, Rv3776, Rvl378c from the lexA DNA damage response were noted in both the pretomanid and delaminid response.

[0484] For pyrazinamide, the genes Rvl460, Rv0791c and Rv0792c were common to the TARs at <24hours (< 3hours, 3-6 hours, 6-24 hours), and Rv3173c common to <3 hours and 3-6 hours TARs. Rvl460, Rv3173, and Rv0792c are transcriptional regulators. clpB, a chaperone protein that protects against stress was identified in the 3-6hours and 6-24 hours TARs.Protein synthesis inhibitors

[0485] The 3-6 hours TAR for streptomycin was notable for genes hsp, dnaK, grpE, bfrB and dnaJL all involved in the stress response, as well as the transcriptional regulator furA. None of these genes were noted in the 6-24 hours TAR, with all 10 genes in this TAR being unannotated. Transcripts from furA, as well as ribosomal proteins rplU, rplY, rplX were noted in the 3-6 hours TAR for capreomycin. Transcriptional regulators whiB7 and higAB (also a toxin-antitoxin system) along with eis (associated with resistance to amikacin and kanamycin) were noted in the 6-24 hours TAR. Similar to streptomycin, there was no overlap in the genes present in the 3-6 hours and 6-24 hours Capreomycin TARs. The single amikacin 3-6 hours TAR included hsp,furA, bfrB, higB seen in the responses for streptomycin and capreomycin, as well as alkB, sthA, enzymes involved in lipid metabolism and intermediary metabolism and respiration. Kanamycin 6-24 hours TAR was notable for transcripts from genes hsp, whiB7, higB and icll. Icll isocitrate lyase is involved in fatty acid catabolism and mediates Mtb response to antibiotics.

[0486] Transcripts from the gene whiB7 were noted in the linezolid TARs from 3-6 hours, 6-24 hours and > 24 hours. whiB7 is a transcriptional regulator that has been linked to intrinsic and acquired antibiotic resistance in Mtb. It is induced by antibiotics that affect ribosomal activity such as erythromycin, clarithromycin, amikacin, spectinomycin, tetracycline and streptomycin, and acts by altering the composition of ribosome associated proteins that promote translation. One or more of transcripts from genes Rvl258c, erm(37), eis, and Rv0263c that have previously been demonstrated to be upregulated by whiB7 were seen in the 3-6 hours and 6-24 hours TARs. The >24 hours TAR also included whiB6 and Rv0196, transcriptional regulators induced by hypoxia that were also seen in the pretomanid s>24 hours TAR.Respiratory inhibitors

[0487] For bedaquiline, stress response gene hsp was present in TAR for 3-6 hours, 6-24 hours, and > 24hours. There was little overlap in the genes across TARs for other time points, with genes involved in stress response, cytochrome oxidases, and fatty acid metabolism noted in TARs across all three time points. mbtB, mbtC, mbtD, mbtE, mbtF seen in the 6-24 hours TAR are members of the mbt-1 locus consisting of genes mbtA-mbtJ responsible for the synthesis of siderophore mycobactin. The mycobactins are necessary for the intracellular transport of extracellular iron and for intracellular survival of Mtb in host cells. The 6-24 hour TAR also included Rv0678 which is transcriptional repressor of the Mmps5-Mmpl5 efflux pump that is involved in the export of mycobactins; mutations in Rv0678 confer resistance to azoles, bedaquiline and clofazimine through drug export by the Mmps5-Mmpl5 efflux pump. The primary activity of Mmpl5 and Mmps5 is export of the siderophores synthesized by the mbt-1 and mbt-2 loci', interruption of its activity leads to toxicity from the intracellular accumulation of iron.

[0488] The TARs for clofazimine were striking for the presence of 4 or more genes from the mbt-1 locus that codes for the siderophore mycobactin in all TARs. The 3-6 hours TAR also included transcription regulator furA, as well as mmpS5, mmpL5, and Rv0678.Fluoroquinolones

[0489] The moxifloxacin TARs were notable for including DNA damage SOS genes recA and lexA -regulated genes (Rv3074, Rv3776, Rv3395c, Rvl378c), but not lexA itself.Additional transcripts belonging to multiple ATP-dependent DNA helicases (Ihr, Rv3201c, Rv3202c) and DNA repair protein radA were also noted. The levofloxacin TAR included Rv3076, Rv3776, and Rv3395c, members of the recA / LexA SOS responses, as well as DNA repair enzyme alkA.

[0490] Fluoroquinolones act against Mtb by binding to DNA gyrase (coded by gyrA and gyrB) and resulting in DNA damage; mutations in gyrA and gyrB result in resistance to fluoroquinolones. The TARs for both moxifloxacin and levofloxacin included / cxA-rcgulon genes (induced by DNA damage) but did not include gyrA or gyrB. The TARs from pretomanid and delamanid (nitroimidazoles that inhibit mycolic acid synthesis in Mtb) also included DNA-damage genes from the / <?x4 -regulon. However, TARs for pretomanid included iniBAC genes, while TARs from delamanid did not.

[0491] Downregulated genesCell wall agents

[0492] Transcripts from desAl , desA2, and desA3 as well as from rpfE, rpfC (resuscitation promoting factors) were noted for isoniazid. desAl-3 are involved in mycolic acid desaturation, while the rpfE and rpfC are involved in virulence and recovery from dormancy. Additional genes included fbpB, which codes for antigen 85B, a cell wall agent and virulence factor. All these genes were also seen in the ethionamide TARs.

[0493] The delaminid TARs shared genes Rv2030c, hspX, hrpl, TB31.7,fdxA, as well as additional genes Rv0572c, tgsl, Rv3129 genes whose upregulation is associated with hypoxia and transition to dormancy as part of the DosR regulon. The pretomanid 3-6 hours and 6-24 hours TARs were notable for the presence of multiple genes coding for 30S and 50S ribosomal proteins (9 and 7 genes, respectively), with transcripts from rplB, rpsS, rpsC, rpsQ and rpfE common to both TARs. The 6-24 and > 24 hours TARs included genes coding for stress response such as chaperone proteins groEL2 and groES. membrane proteins rpfE and rpfC, as well as DosR regulon related genes tgsl, hspX, hrpl.

[0494] Both pretomanid and delaminid demonstrated downregulation of ribosomal proteins (rplW, rplB, rplV) as well as genes such as groES and groEE2, alongwith tgsl, Rv3131, hspX and hrpE The desAl, desA2,fbp, rpfE, rpfC, nuoO, Rv0516, Rv0950 and Rvl815 shared across the downregulated TARs for isoniazid and ethionamideTranscriptional inhibitors

[0495] The rifampicin TARs included transcripts from multiple different families. The single rifapentine TAR included groES and groEE2, ESX system transcripts esxM, esxH, and hupB, which codes for a polyfunctional protein involved in iron metabolism and other functions. We did notice downregulation of groES and gorEL2 for both rifampicin and rifapentine.

[0496] The >24 hours pyrazinamide TAR was notable for all 8 genes from the mcel (mammalian cell entry) operon, that are involved in infection of human cells.Protein synthesis inhibitors

[0497] The streptomycin TARs were notable for multiple members of the ESX family (esxH, esxR, esxQ, esxS) and NADH dehydrogenase enzyme- 1 enzymes InuoK. nuoM), with downregulation of transcriptional regulators ethA, Rv0324 and other genes involved in respiration and fatty acid synthesis. The amikacin TAR contained transcripts from multiple ESX family genes (esxR, esxG. esxS, esxH), ATP synthase genes (atpC and atpD), and nuoM. The Kanamycin TAR included transcripts involved in intermediary respiration and metabolism as well as cell wall proteins. The two Capreomycin TARs were notable for ESX family gcncs(c.sx / 7, esxG, esxR, esxQ, esxS), stress related {groES, groEE2), and intermediary respiration and metabolism (atpC, atpD, nuoM, nuoH, nuol).

[0498] For linezolid, transcripts from the genes groEE2 and groES were noted in the 3-6 and 6-24 hoursTAR, but with minimal overlap in the other time frames. We noted a transition from ESX family related genes () in the 3-6 hours TAR, to enxymes involved in intermediary respiratoy and metabolism in the 6-24 hours TAR, to hypoxic stress response genes in the >24 hours TAR.Respiratory inhibitors

[0499] There was limited overlap in the genes the comprised the clofazimine TARs. Most genes were involved in intermediary respiration and metabolism or lipid synthesis, with genes identified the hypoxic stress response noted across multiple time points (bfrB, hspX, tgsl, hrpl). For bedaquiline, there were only 2 genes seen in more than 1 TAR. Most genes involved in intermediary respiration and metabolism, with . The 3-6 hours TAR notable for transcripts from genes coding for ribosomal proteins (rpsH, rplL, rnpA)Fluoroquinolones

[0500] The moxifloxacin TARs showed a variety of different genes involved in each TAR, but was notable for the presence transcriptional regulator whiB2 and cell division proteins SepF and Ftsz in the 3-6 hours, hupB in the 6-24 hours TAR, and hypoxic stress response genes in the >24 hours TAR. The single levofloxacin TAR included 5 transcripts from the ESX family (esxR, esxH, esxQ- esxG. e.sxSp as well as hupB and whiB2 that were also noted in moxifloxacin TARs.

[0501] Exposure to some antibiotics can induce a drug tolerance state with a range of cellular responses. As expected, the TAR signatures include some drug tolerance transcripts involved in the transcriptional regulation (yvhiB7 and Sigma factors), metabolic slowdown (ATP synthase and NADH-1 dehydrogenase, ribosomal protein synthesis), metabolic shifting (zcZJ) and efflux pump activity (Rvl258c, iniBAC, efpA).

[0502] References

[0503] 1. Alland D, Steyn AJ, Weisbrod T, Aldrich K, Jacobs WR, Jr. Characterization of the Mycobacterium tuberculosis iniBAC promoter, a promoter that responds to cell wall biosynthesis inhibition. J Bacteriol 2000; 182(7): 1802-11. (In eng). DOI:10.1128 / jb.182.7.1802-1811.2000.

[0504] 2. Colangeli R, Helb D, Sridharan S, et al. The Mycobacterium tuberculosis iniA gene is essential for activity of an efflux pump that confers drug tolerance to both isoniazid and ethambutol. Mol Microbiol 2005;55(6): 1829-40. DOI: 10.1111 / j.1365-2958.2005.04510.x.

[0505] 3. Wang M, Guo X, Yang X, et al. Mycobacterial dynamin-like protein IniA mediates membrane fission. Nat Commun 2019;10(l):3906. DOI: 10.1038 / s41467-019-11860-z.

[0506] 4. Cantaloube S, Veyron-Churlet R, Haddache N, Daffe M, Zerbib D. The Mycobacterium tuberculosis FAS-II dehydratases and methyltransferases define the specificity of the mycolic acid elongation complexes. PLoS One 2011;6(12):e29564. DOI: 10.1371 / journal. pone.0029564.

[0507] 5. Duan X, Xiang X, Xie J. Crucial components of Mycobacterium type II fatty acid biosynthesis (Fas-II) and their inhibitors. FEMS Microbiol Lett 2014;360(2):87-99. DOI: 10.1111 / 1574-6968.12597.

[0508] 6. Mdluli K, Slayden RA, Zhu Y, et al. Inhibition of a Mycobacterium tuberculosis beta-ketoacyl ACP synthase by isoniazid. Science 1998;280(5369):1607-10. DOI: 10.1126 / science.280.5369.1607.

[0509] 7. Vilcheze C, Jacobs WR, Jr. Resistance to Isoniazid and Ethionamide in Mycobacterium tuberculosis: Genes, Mutations, and Causalities. Microbiol Spectr 2014;2(4):MGM2-0014-2013. DOI: 10.1128 / microbiolspec.MGM2-0014-2013.

[0510] 8. Manganelli R, Voskuil MI, Schoolnik GK, Smith I. The Mycobacterium tuberculosis ECF sigma factor sigmaE: role in global gene expression and survival in macrophages. Mol Microbiol 2001;41(2):423-37. (In eng). DOI: 10.1046 / j.1365- 2958.2001.02525.x.

[0511] 9. Stewart GR, Wemisch L, Stabler R, et al. Dissection of the heat-shock response in Mycobacterium tuberculosis using mutants and microarrays. Microbiology (Reading) 2002;148(Pt 10):3129-3138. DOI: 10.1099 / 00221287-148-10-3129.

[0512] 10. Lee JH, Karakousis PC, Bishai WR. Roles of SigB and SigF in the Mycobacterium tuberculosis sigma factor network. J Bacteriol 2008;190(2):699-707. DOI: 10.1128 / JB.01273-07.

[0513] 11 Zhang Y, Heym B, Allen B, Young D, Cole S. 1992. The catalase-peroxidase gene and isoniazid resistance of Mycobacterium tuberculosis. Nature 358:591-3.

[0514] 12. Wilson M, DeRisi J, Kristensen HH, Imboden P, Rane S, Brown PO, Schoolnik GK. 1999. Exploring drug-induced alterations in gene expression in Mycobacterium tuberculosis by microarray hybridization. Proc Natl Acad Sci U S A 96:12833-8.

[0515] 13. Barczak AK, Gomez JE, Kaufmann BB, Hinson ER, Cosimi L, Borowsky ML, Onderdonk AB, Stanley SA, Kaur D, Bryant KF, Knipe DM, Sloutsky A, Hung DT.2012. RNA signatures allow rapid identification of pathogens and antibiotic susceptibilities. Proc Natl Acad Sci U S A 109:6217-22.

[0516] 14. Baulard AR, Betts JC, Engohang-Ndong J, et al. Activation of the pro-drug ethionamide is regulated in mycobacteria. J Biol Chem 2000;275(36):28326-31. (In eng). DOI: 10.1074 / jbc.M003744200.

[0517] 15. Dover LG, Corsino PE, Daniels IR, et al. Crystal Structure of the TetR / CamR Family Repressor Mycobacterium tuberculosis EthR Implicated in Ethionamide Resistance. Journal of Molecular Biology 2004;340(5): 1095- 1105. DOI: h tip s : / 7doi , org / 10.1016 / j .j mb .2004.06 , 003.

[0518] 16. Davis EO, Dullaghan EM, Rand L. Definition of the mycobacterial SOS box and use to identify LexA -regulated genes in Mycobacterium tuberculosis. J Bacteriol 2002;184(12):3287-95. DOI: 10.1128 / JB.184.12.3287-3295.2002.

[0519] 17. Smollett KL, Smith KM, Kahramanoglou C, Amvig KB, Buxton RS, Davis EO. Global analysis of the regulon of the transcriptional repressor LexA, a key component of SOS response in Mycobacterium tuberculosis. J Biol Chem 2012;287(26):22004-14. DOI: 10.1074 / jbc.Ml 12.357715.

[0520] 18. Rustad TR, Harrell MI, Liao R, Sherman DR. The enduring hypoxic response of Mycobacterium tuberculosis. PLoS One 2008;3(l):el502. DOI:10.1371 / journal. pone.0001502.

[0521] 19. Smith TC, 2nd, Pullen KM, Olson MC, et al. Morphological profiling of tubercle bacilli identifies drug pathways of action. Proc Natl Acad Sci U S A 2020;l 17(31): 18744-18753. DOI: 10.1073 / pnas.2002738117.

[0522] 20. Michael Wilson* f JD, Hans-Henrik Kristensen^[fi^, Paul Imboden^**,Sangeeta Ranc'][, Patrick O. Brown:;:, and Gary K. Schoolnik*'][. Exploring drug-induced alterations in gene expression in Mycobacterium tuberculosis by microarray hybridization. PNAS 1999;96(22): 12833-12838.

[0523] 21. Walter ND, Dolganov GM, Garcia BJ, et al. Transcriptional Adaptation of Drug-tolerant Mycobacterium tuberculosis During Treatment of Human Tuberculosis. J Infect Dis 2015;212(6):990-8. DOI: 10.1093 / infdis / jivl49.

[0524] 22. Kapopoulou A, Lew JM, Cole ST. The MycoBrowser portal: a comprehensive and manually annotated resource for mycobacterial genomes. Tuberculosis (Edinb) 2011;91(1):8-13. (In eng). DOI: 10.1016 / j.tube.2010.09.006.

[0525] 23. Singh P, Khurana H, Yadav SP, et al. Biochemical characterization of ClpB protein from Mycobacterium tuberculosis and identification of its small-molecule inhibitors. Int J Biol Macromol 2020;165(Pt A):375-387. (In eng). DOI: 10.1016 / j.ijbiomac.2020.09.131.

[0526] 24. Tripathi P, Singh LK, Kumari S, Hakiem OR, Batra JK. ClpB is an essential stress regulator of Mycobacterium tuberculosis and endows survival advantage to dormant bacilli. Int J Med Microbiol 2020;310(3): 151402. (In eng). DOI:10.1016 / j.ijmm.2020.151402.

[0527] 25. Pym AS, Domenech P, Honore N, Song J, Deretic V, Cole ST. Regulation of catalase-peroxidase (KatG) expression, isoniazid sensitivity and virulence by furA of Mycobacterium tuberculosis. Mol Microbiol 2001;40(4):879-89. (In eng). DOI: 10.1046 / j.1365-2958.2001.02427.x.

[0528] 26. Bush MJ. The actinobacterial WhiB-like (Wbl) family of transcription factors. Mol Microbiol 2018;110(5):663-676. DOI: 10.1111 / mmi.l4117.

[0529] 27. Samuel LP, Song CH, Wei J, et al. Expression, production and release of the Eis protein by Mycobacterium tuberculosis during infection of macrophages and its effect on cytokine secretion. Microbiology (Reading) 2007;153(Pt 2):529-540. (In eng). DOI: 10.1099 / mic.0.2006 / 002642-0.

[0530] 28. Ramage HR, Connolly LE, Cox JS. Comprehensive functional analysis of Mycobacterium tuberculosis toxin- antitoxin systems: implications for pathogenesis, stress responses, and evolution. PLoS Genet 2009;5(12):el000767. DOI: 10.1371 / j ournal .pgen .1000767.

[0531] 29. Fivian-Hughes AS, Davis EO. Analyzing the regulatory role of the HigA antitoxin within Mycobacterium tuberculosis. J Bacteriol 2010;192(17):4348-56. DOI: 10.1128 / JB.00454-10.

[0532] 30. Nandakumar M, Nathan C, Rhee KY. Isocitrate lyase mediates broad antibiotic tolerance in Mycobacterium tuberculosis. Nat Commun 2014;5:4306. DOI: 10.1038 / ncomms5306.

[0533] 31. Munoz-Elias EJ, McKinney JD. Mycobacterium tuberculosis isocitrate lyases1 and 2 are jointly required for in vivo growth and virulence. Nat Med 2005;l l(6):638-44. (In eng). DOI: 10.1038 / nml252.

[0534] 32. Morris RP, Nguyen L, Gatfield J, et al. Ancestral antibiotic resistance in Mycobacterium tuberculosis. Proc Natl Acad Sci U S A 2005;102(34):12200-5. DOI: 10.1073 / pnas.0505446102.

[0535] 33. Hurst-Hess K, Rudra P, Ghosh P. Mycobacterium abscessus WhiB7 Regulates a Species-Specific Repertoire of Genes To Confer Extreme Antibiotic Resistance.Antimicrob Agents Chemother 2017;61(l l). DOI: 10.1128 / AAC.01347-17.

[0536] 34. Baatjies L, Loxton AG, Williams MJ. Host and Bacterial Iron Homeostasis, an Underexplored Area in Tuberculosis Biomarker Research. Front Immunol 2021; 12:742059.DOI: 10.3389 / fimmu.2021.742059.

[0537] 35. Quadri LEN, Sello J, Keating TA, Weinreb PH, Walsh CT. Identification of a Mycobacterium tuberculosis gene cluster encoding the biosynthetic enzymes for assembly of the virulence-conferring siderophore mycobactin. Chemistry & Biology 1998;5(11):631-645. DOI: http s : / / doi . org / 10.1016 / S 1074- 5521(98 ) 90291-5.

[0538] 36. Milano A, Pasca MR, Provvedi R, et al. Azole resistance in Mycobacterium tuberculosis is mediated by the MmpS5-MmpL5 efflux system. Tuberculosis (Edinb) 2009;89(l):84-90. (In eng). DOI: 10.1016 / j.tube.2008.08.003.

[0539] 37. Hartkoorn RC, Uplekar S, Cole ST. Cross-resistance between clofazimine and bedaquiline through upregulation of MmpL5 in Mycobacterium tuberculosis. Antimicrob Agents Chemother 2014;58(5):2979-81. DOI: 10.1128 / AAC.00037-14.

[0540] 38. Ismail NA, Omar SV, Joseph L, et al. Defining Bedaquiline Susceptibility, Resistance, Cross-Resistance and Associated Genetic Determinants: A Retrospective Cohort Study. EBioMedicine 2018;28:136-142. (In eng). DOI: 10.1016 / j.ebiom.2018.01.005.

[0541] 39. Andries K, Villellas C, Coeck N, et al. Acquired resistance of Mycobacterium tuberculosis to bedaquiline. PLoS One 2014;9(7):el02135. DOI:10.1371 / journal. pone.0102135.

[0542] 40. Jones CM, Wells RM, Madduri AV, et al. Self-poisoning of Mycobacterium tuberculosis by interrupting siderophore recycling. Proc Natl Acad Sci U S A 2014; 111(5): 1945-50. DOI: 10.1073 / pnas.l311402111.

[0543] 41. Takiff HE, Salazar L, Guerrero C, Philipp W, Huang WM, Kreiswirth B, Cole ST, Jacobs WR, Jr., Telenti A. 1994. Cloning and nucleotide sequence of Mycobacterium tuberculosis gyrA and gyrB genes and detection of quinolone resistance mutations.Antimicrob Agents Chemother 38:773-80.

[0544] 42. Aldred KJ, Blower TR, Kerns RJ, Berger JM, Osheroff N. 2016.Fluoroquinolone interactions with Mycobacterium tuberculosis gyrase: Enhancing drug activity against wild-type and resistant gyrase. Proc Natl Acad Sci U S A 113:E839-46.

[0545] 43. Mudde SE, Upton AM, Lenaerts A, Bax HI, De Steenwinkel JEM. 2022. Delamanid or pretomanid? A Solomonic judgement! J Antimicrob Chemother 77:880-902.

[0546] 44. Chang Y, Fox BG. Identification of Rv3230c as the NADPH oxidoreductase of a two-protein DesA3 acyl-CoA desaturase in Mycobacterium tuberculosis H37Rv. Biochemistry 2006;45(45): 13476-86. DOI: 10.1021 / bi0615285.

[0547] 45. Singh A, Varela C, Bhatt K, et al. Identification of a Desaturase Involved in Mycolic Acid Biosynthesis in Mycobacterium smegmatis. PLoS One 2016;l l(10):e0164253. DOI: 10.1371 / journal.pone.0164253.

[0548] 46. Bailo R, Radhakrishnan A, Singh A, Nakaya M, Fujiwara N, Bhatt A. The mycobacterial desaturase DesA2 is associated with mycolic acid biosynthesis. Sci Rep 2022;12(l):6943. DOI: 10.1038 / s41598-022-10589-y.

[0549] 47. Kana BD, Gordhan BG, Downing KJ, et al. The resuscitation-promoting factors of Mycobacterium tuberculosis are required for virulence and resuscitation from dormancy but are collectively dispensable for growth in vitro. Mol Microbiol 2008;67(3):672-84. DOI: 10.1111 / j.l365-2958.2007.06078.x.

[0550] 48. Wilkinson RJ, DesJardin LE, Islam N, et al. An increase in expression of a Mycobacterium tuberculosis mycolyl transferase gene (fbpB) occurs early after infection of human monocytes. Mol Microbiol 2001 ;39(3):813-21. DOI: 10.1046 / j.1365- 2958.2001.02280.x.

[0551] 49. Rosenkrands I, Slayden RA, Crawford J, Aagaard C, Barry CE, 3rd, Andersen P. Hypoxic response of Mycobacterium tuberculosis studied by metabolic labeling and proteome analysis of cellular and extracellular proteins. J Bacteriol 2002;184(13):3485-91. DOI: 10.1128 / JB.184.13.3485-3491.2002.

[0552] 50. Voskuil MI, Schnappinger D, Visconti KC, et al. Inhibition of respiration by nitric oxide induces a Mycobacterium tuberculosis dormancy program. J Exp Med 2003; 198(5):705-13. DOI: 10.1084 / jem.20030205.

[0553] 51. Arruda S, Bomfim G, Knights R, Huima-Byron T, Riley LW. Cloning of an M. tuberculosis DNA fragment associated with entry and survival inside cells. Science 1993;261(5127): 1454-7. (In eng). DOI: 10.1126 / science.8367727.

[0554] 52. Goossens SN, Sampson SL, Van Rie A. 2020. Mechanisms of Drug-Induced Tolerance in Mycobacterium tuberculosis. Clin Microbiol Rev 34.

[0555] Tables

Claims

CLAIMSWhat is claimed is:

1. A method for determining whether a biological sample obtained from a patient contains mycobacterium tuberculosis complex (Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: using at least one computer hardware processor: obtaining RNA expression data for the biological sample previously-obtained by profiling RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some Mtb genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic.

2. The method of claim 1, further comprising: profiling the RNA from the biological sample after exposure to the at least one antibiotic.

3. The method of claim 2, wherein profiling the RNA is performed using next generation sequencing.

4. The method of claim 2, wherein profiling the RNA is performed using digital droplet polymerase chain reaction (PCR).

5. The method of claim 2, wherein profiling the RNA is performed using quantitative reverse transcription polymerase chain reaction (RT-qPCR).

6. The method of claim 2, wherein profiling the RNA is performed using probe-based RNA sequencing (optionally, using nCounter NanoString technology).

7. The method of claim 2, further comprising: culturing at least a portion of the biological sample to obtain a cultured sample; exposing the cultured sample to the at least one antibiotic for a specified time period; and extracting the RNA from the culture after the specified time period of exposure to the at least one antibiotic.

8. The method of claim 1, wherein the particular transcriptomic signature for the particular antibiotic comprises at least two differentially regulated Mtb genes associated with a stress response pathway.

9. The method of claim 8, wherein the at least two differentially regulated Mtb genes associated with the stress response pathway comprise one or more of heat shock response genes, one or more deoxyribonucleic acid (DNA) damage genes, master transcriptional regulators, membrane transporters, cell wall component biosynthesis enzymes, siderophore biosynthesis enzymes, chaperone proteins, virulence effector genes, ribosomal subunit proteins, ribonuclease P proteins, mycolic acid desaturation genes, resuscitation promoting factors, respiratory genes, adenosine triphosphate (ATP) synthetases, and / or hypoxic response genes.

10. The method of claim 8, wherein the at least two differentially regulated Mtb genes are upregulated.

11. The method of claim 8, wherein the at least two differentially regulated Mtb genes are downregulated.

12. The method of claim 8, wherein the at least two differentially regulated genes comprise at least one of a gene encoding a heat shock protein (hsp) or a / u / A gene.

13. The method of claim 8, wherein the at least two differentially regulated genes comprise at least one of a gene encoding for a chaperone protein involved in the stress response pathway.

14. The method of claim 1 or any other preceding claim, wherein the RNA expression data comprises mRNA expression data, rRNA expression data, or both mRNA and rRNA expression data.

15. The method of claim 1 or any other preceding claim, wherein the RNA expression data is mRNA expression data.

16. The method of claim 1 or any other preceding claim, wherein the particular transcriptomic signature for the particular antibiotic comprises at least two (optionally, three, four, five, six, ... , all) of the upregulated Mtb genes for the particular antibiotic as specified in Table A3 and / or in one or more of Tables B1-B7.

17. The method of claim 16 or any other preceding claim, wherein the particular transcriptomic signature for the particular antibiotic comprises at least two (optionally, three, four, five, six, ..., all) of the downregulated Mtb genes for the particular antibiotic as specified in Table A3 and / or in one or more of Tables B 1-B7.

18. The method of claim 1 or any other preceding claim, wherein the biological sample is exposed to the particular antibiotic for a specified time period, and wherein the particular transcriptomic signature for the particular antibiotic comprises at least two (optionally, three, four, five, six, ... , all) of the upregulated Mtb genes for the particular antibiotic as specified in Table A3 and / or in one or more of Tables B1-B7 with respect to the specified time period.

19. The method of claim 12 or any other preceding claim, wherein the biological sample is exposed to the particular antibiotic for a specified time period, and wherein the particular transcriptomic signature for the particular antibiotic comprises at least two (optionally, three, four, five, six, ..., all) of the downregulated Mtb genes for the particular antibiotic as specified in Table A3 and / or in one or more of Tables B 1-B7 for the specified time period.

20. The method of claim 1 or any other preceding claim, wherein the at least one antibiotic includes at least one antibiotic selected from the list of antibiotics consisting of: Bedaquiline, Linezolid, Moxifloxacin, Pretomanid, and Pyrazinamide.

21. The method of claim 1 or any other preceding claim, wherein the at least one antibiotic includes at least one antibiotic selected from the list of antibiotics consisting of: Isoniazid, Rifampicin, Pyrazinamide, Ethambutol, Moxifloxacin, Levofloxacin, Bedaquiline, Linezolid, Pretomanid, Clofazimine, Amikacin, Streptomycin, Delaminid, Ethionamide, Capreomycin, Kanamycin, and Rifapentine.

22. The method of claim 1 or any other preceding claim, wherein when it is determined that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic, administering the particular antibiotic to the patient.

23. The method of claim 1 or any other preceding claim, wherein the at least one antibiotic comprises multiple antibiotics, and wherein when it is determined that the Mtb in the biological sample is susceptible to treatment by each of the multiple antibiotics, administering the multiple antibiotics to the patient.

24. A system comprising: at least one computer hardware processor; and at least one non-transitory computer-readable storage medium storing instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform a method for determining whether a biological sample obtained from a patient contains mycobacterium tuberculosis complex (Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: obtaining RNA expression data for the biological sample previously-obtained by profiling RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics,determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some Mtb genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic.

25. At least one non-transitory computer-readable storage medium storing instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform a method for determining whether a biological sample obtained from a patient contains mycobacterium tuberculosis complex (Mtb) susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: obtaining RNA expression data for the biological sample previously-obtained by profiling RNA from the biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some Mtb genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the Mtb in the biological sample is susceptible to treatment by the particular antibiotic.

26. A method for determining whether a biological sample obtained from a patient contains bacteria susceptible to treatment by at least one antibiotic in a set of one or more antibiotics, the method comprising: using at least one computer hardware processor:obtaining RNA expression data for the biological sample previously-obtained by profiling RNA from a biological sample after exposure to the at least one antibiotic; for each particular antibiotic of the at least one antibiotic in the set of one or more antibiotics, determining whether a particular transcriptomic signature for the particular antibiotic is present in the RNA expression data, wherein the particular transcriptomic signature includes at least some bacterial genes previously determined to be upregulated and / or downregulated in response to exposure to the particular antibiotic; and when it is determined that the transcriptomic signature is present in the RNA expression data, determining that the bacteria in the biological sample is susceptible to treatment by the particular antibiotic.

27. The method of claim 26, wherein the RNA expression data comprises rRNA expression data.

Images