Integration of multi-modal measurements identifies critical mechanisms of tuberculosis drug action

Treatments for tuberculosis remain lengthy, motivating a search for new drugs with novel mechanisms of action. However, it remains challenging to elucidate the direct targets of a drug, and even more so, to determine which disrupted cellular processes lead to bacterial killing. We developed a computational tool, DECIPHAER (Decoding Cross-modal Information of Pharmacologies via AutoEncodeRs), to select correlated transcriptional and morphological responses of Mycobacterium tuberculosis to drug treatments. By finding a latent space from these measurements, DECIPHAER highlighted important features of Mtb cellular damage such as phosphosugar stress and inhibition of translation and DNA replication. After training, DECIPHAER provides cell-death-relevant insight into even single-modal datasets, enabling interrogation of drug treatment responses for which transcriptional data are not available. Using only morphological data with DECIPHAER, we discovered that respiration inhibition by the poly-pharmacological drugs, SQ109 and BM212, can be their primary mechanism influencing cell death, not their effect on the cell wall. This study demonstrates that DECIPHAER can extract the critical shared information from multi-modal measurements to identify cell death-relevant mechanisms of TB drugs. Mtb strain Erdman with plasmid pMV306hsp+LuxG13 were adapted to a cholesterol in-vitro growth condition prior to drug exposure and RNA extraction. Briefly, Mtb were first cultured in standard growth medium consisting of 7H9 broth, (ThermoFisher; DF0713-17-9), 0.2% glycerol, 10% OADC (0.5g/L oleic acid, 50g/L albumin, 20g/L dextrose and 0.04g/L catalase), and 0.05% Tween-80 and then subcultured at least once prior to cholesterol adaptation. Once the bacterial cultures reached an OD600nm ~0.5, the cultures were diluted to an OD600nm of 0.05 in cholesterol growth medium consisting of 7H9 powder (4.7g/L), fatty acid-free BSA (0.5g/L), NaCl (100mM) and tyloxapol (0.05%). Cultures were then subcultured once more in the same media after reaching mid-log phase (OD600nm ∼0.5-0.7). Then, the cultures were subcultured to 0.05 OD in 50ml conical tubes containing the cholesterol growth media and drug treatment at a concentration of 1x the IC90. Drug-treated cultures and untreated controls were then harvested by centrifugation at timepoints 4, and 24 hours post drug treatment. All cultures were buffered with 100 mM 3-(N-morpholino)propanesulfonic acid at pH 7.