Supplementary Information Supplementary Figures

Supplementary Figure 1: The absolute OCR profiles for the data presented in Figure 1. The vertical dotted lines indicate the time points at which the glucose (Glc), the specific drug (at the respective concentrations) and CCCP was injected. Mtb H37Rv treated with BDQ (3x, 30x and 300x the MIC50), Q203 (3x, 30x and 300x the MIC50) and CFZ (only at 3x and 30x the MIC50 due to lack of solubility at 300x) in the presence of glucose as carbon source. DMSO was used as a vehicle control. One representative experiment is shown, three replicate experiments were performed. Supplementary Figure 2: The OCR profiles of Mtb H37Rv treated with BDQ (x30 and x300 the MIC50), Q203 (x30 and x300 the MIC50) and CFZ (x30 the MIC50) in the presence of lactate and palmitate as carbon sources. The vertical dotted lines indicate the time points at which the carbon source (CS), the specific drug and CCCP was injected. It is clear from the OCR profiles that the increase in OCR after BDQ and Q203 addition is not carbon source dependent. One representative experiment is shown, three replicate experiments were performed.

in the presence of glucose as a carbon source. The vertical dotted lines indicate the time points at which the specific drug and CCCP was injected. The profiles indicate that increase in both OCR and ECAR, after BDQ and Q203 addition, is maintained for at least 16 hrs after addition, and after the addition of CCCP the bacilli were still able to increase their OCR and ECAR.
One representative experiment is shown, two replicate experiments were performed. although the non-Cell-Tak bacilli OCR measurements standard deviation is on average 50% higher than that of the Cell-Tak bacilli OCR measurements. This demonstrates the advantage of using Cell-Tak to adhere the bacilli to the XF cell culture microplate. The experiment was done in duplicate with the mean and standard deviations calculated from nine replicate wells in each experiment (n=9). Unpaired t-test (using GraphPad Prism 6.05) was used to analyze the statistical difference between each of the four basal respiration readings.

Mtb ETC targeting drug combination cytotoxicity
Previous studies indicate that BDQ, Q203 or CFZ as monotherapy did not have any cytotoxic effects on mammalian cells when they are treated with the same concentrations used in our study [1][2][3] , but the combined effect of these Mtb ETC targeting drugs on mammalian respiration has not yet been determined. XF technology has been frequently used to assess drug toxicity and is being regarded as superior to purified mitochondria as it provides a cellular context 4-6 . Here, in addition to the conventional approach that measures the effects of drugs on unstressed mitochondria, we determine if the drug combinations had an effect on mammalian bioenergetics by examining four different bioenergetic parameters in HepG2 ( Figure 6, panels a-c) and RAW 264.7 ( Figure 6, panels d-e) cell lines. Firstly, the drug combinations had no effect on basal respiration (BR) above that of the media, DMSO and RIF/INH controls; indicating that when these drugs are used in combination they do not affect normal cellular respiration. BR of drug treated cells does not deviate from the controls' basal respiration even after extended drug treatment ( Supplementary Fig. 6).
Secondly, ATP turnover (AT) is the decrease observed in OCR upon addition of oligomycin, which blocks the proton channel of the F 0 -subunit of mammalian ATP synthase and inhibits ATP synthesis.
Although the CFZ drug combinations induced a slightly smaller AT in the HepG2 cell line than the non-CFZ drug combinations, the difference was minute. One possible explanation for the smaller Extrinsic uncoupling could take several forms. One might be for BDQ to directly cause an increase in proton conductance, by binding to the subunit a-c interface of ATP synthase subunit F 0 , allowing it to act as a proton channel, as has recently been proposed by other authors 9 . We argue that this is unlikely for several reasons. First, our direct measurements of proton conductance in IMVs show no change with BDQ ( Figure 2c). Second, BDQ does not reduce membrane potential in live cells, unlike the protonophore CCCP (Figure 2h). Third, DCCD increases OCR similarly (Figure 3c), suggesting that increased OCR is a common downstream response to varied types of Complex V inhibition. Fourth, the recently completed crystal structure of the F 0 c-ring with BDQ shows BDQ binds on the H + binding site and prevents the c-ring from turning 10 . It is unlikely that BDQ has two binding sites on ATP