The anti-tubercular activity of simvastatin is mediated by cholesterol-dependent regulation of autophagy via the AMPK-mTORC1-TFEB axis

Statins, which inhibit both cholesterol biosynthesis and protein prenylation branches of the mevalonate pathway, increase anti-tubercular antibiotic efficacy in animal models. We investigated the mechanism of anti-tubercular action of simvastatin in Mycobacterium tuberculosis-infected human monocytic cells. We found that the anti-tubercular activity of statins was phenocopied by cholesterol-branch but not prenylation-branch inhibitors. Moreover, statin treatment blocked activation of mechanistic target of rapamycin complex 1 (mTORC1), activated AMP-activated protein kinase (AMPK) through increased intracellular AMP:ATP ratios, and favored nuclear translocation of transcription factor EB (TFEB). These mechanisms all induce autophagy, which is anti-mycobacterial. The biological effects of simvastatin on the AMPK-mTORC1-TFEB-autophagy axis were reversed by adding exogenous cholesterol to the cells. Overall, our data demonstrate that the anti-tubercular activity of simvastatin requires inhibiting cholesterol biosynthesis, reveal novel links between cholesterol homeostasis, AMPK-mTORC1-TFEB axis, and intracellular infection control, and uncover new anti-tubercular therapy targets.


37
As our knowledge of host-pathogen interactions grows for many infectious agents, pharmacologically 38 manipulating the host response has emerged as a key approach to control or treat disease-causing infections.

39
This approach may be best suited for infections caused by intracellular pathogens, given the intimate 40 relationship between these pathogens and their host cells. For example, Mycobacterium tuberculosis, the 41 intracellular pathogen causing tuberculosis, can subvert and disable the antimicrobial mechanisms of the 42 host macrophage while adapting to the environmental conditions created by these mechanisms (1, 2). The 43 treatment of active tuberculosis, which typically presents with lung tissue damage, is prolonged and requires 44 multiple drugs, presumably due to the presence of phenotypically diverse M. tuberculosis subpopulations 45 that exhibit antibiotic tolerance and poor penetration of antibiotics into infected tissues (3)(4)(5). The prolonged 46 duration of anti-tubercular antibiotic therapy (a minimum of six months for drug-susceptible, uncomplicated tuberculosis) poses logistical difficulties for treatment providers and leads to poor patient

89
The mevalonate pathway, which is the mechanistic target of statins, is multi-branched (Fig. 1A). Identifying 90 the pathway branch(es) associated with the observed anti-tubercular activity of these compounds is the first 91 step toward elucidating the underlying mechanism of action. Treatment of M. tuberculosis-infected THP1 92 cells (a human monocytic cell line) with simvastatin, demonstrated that this drug decreased M. tuberculosis 93 intracellular burden ( Fig. 1 B), consistent with previous reports (11,12). The anti-mycobacterial effect of 94 simvastatin varied with the dose, with concentrations of 50-100 nM being the most effective (Fig. 1B).

95
Additional experiments determined that these anti-mycobacterial doses were non-toxic for THP1 cells 96 (supplementary Fig. 1A) and that simvastatin had no direct antimicrobial activity against M. tuberculosis 97 in axenic cultures (supplementary Fig. 1B). We next tested the effect on intracellular M. tuberculosis burden 98 of inhibitors specifically targeting each branch of the mevalonate pathway. As shown in Fig. 1A, the 99 mevalonate pathway branches at farnesyl pyrophosphate (FPP) synthesis, which is used for protein 100 prenylation, including farnesylation and geranylgeranylation, or is converted to squalene, which is required 101 for de novo cholesterol synthesis. We treated M. tuberculosis-infected THP1 cells with inhibitors of farnesyl 102 transferase (FTI-277), geranylgeranyl transferase -type 1 (GGTI-298) and 7-dehydrocholesterol reductase 103 (DHCR7) (BM 15766 sulfate) (colored boxes in Fig. 1A). Moreover, since no inhibitor of geranylgeranyl 104 transferase type 2 (GGTase-II) is commercially available, we used siRNA against the gene encoding the 105 Rab escort protein 1 (REP-1), a chaperone protein that presents Rab proteins for prenylation by GGTase-106 II. We found that only cells treated with the cholesterol-branch inhibitor BM 15766, but not other branch-107 specific inhibitors, showed a reduced mycobacterial burden (35% inhibition relative to solvent-treated cells) 108 ( Fig. 1C). In addition, the inhibition of intracellular M. tuberculosis growth by simvastatin was reversed by 109 exogenous addition of water-soluble cholesterol (Fig. 1D). We verified the link between simvastatin anti-110 mycobacterial activity and blockage of the cholesterol biosynthetic branch of the mevalonate pathway, as 111 treatment with M. tuberculosis-inhibiting doses of simvastatin (50-100 nM) (Fig. 1 B) reduced the amount 6 of free cholesterol in THP1 cells (Fig. 1E) while inducing no detectable alteration of the cellular ability to 113 prenylate proteins (sentinel targets of each prenylation pathway are shown in Fig. 1F). Taken together, these 114 results clearly demonstrate that the anti-tubercular activity of simvastatin specifically targets the cholesterol 115 biosynthetic branch rather than the prenylation branches of the mevalonate pathway.

153
To test our hypothesis, we used M. tuberculosis wild-type (WT) and a mutant strain genetically inactivated 154 in rv1129c, which encodes a transcriptional factor required to induce the methyl citrate cycle genes (32, 155 33). When we infected THP1 cells with M. tuberculosis WT and Δrv1129c mutant strains, we found that 156 the mutant strain was attenuated for growth in macrophages, as expected (up to three-fold reduction of CFU 157 relative to WT) ( Fig. 2A, dark blue bars). Treating infected cells with BM 15766 sulfate (which specifically 158 inhibits de novo cholesterol synthesis, Fig. 1A) drastically reduced intracellular growth of the WT strain 159 (45%), as expected, but increased growth in the Δrv1129c mutant strain by 50% ( Fig. 2A). This result 160 supports our initial hypothesis. In contrast, simvastatin treatment inhibited WT and mutant growth to a 161 similar extent (>40% growth inhibition) (   We next sought to identify the mTORC1-dependent pathway(s) through which simvastatin induces 244 autophagy and controls M. tuberculosis infection. A prime candidate is TFEB, a transcription factor 245 involved in lysosomal biogenesis and autophagy induction (44, 45). mTORC1 inhibits the nuclear 246 translocation of TFEB, which is required for its activation (46, 47). We found that simvastatin treatment 247 increased TFEB abundance in the nuclear fraction but not in the total extracts obtained from M.

289
Our pathway analysis of simvastatin-treated PBMC also identified upregulation of signaling pathways 290 regulated by AMPK, a serine-threonine kinase that inhibits mTORC1 signaling and induces autophagy (55) 291 and upregulated Liver kinase B1 (LKB1), a serine-threonine kinase that activates AMPK (56) ( Table 2), as 292 previously observed in squamous cell carcinoma (57). Since these two proteins participate in the energy-293 sensing cascade activated by an increased AMP:ATP ratio (56), we first asked whether simvastatin 294 treatment alters cellular AMP:ATP ratios in M. tuberculosis-infected cells. We did find that simvastatin 295 reversed the infection-induced decrease in AMP:ATP ratio (Fig. 6A). In accord with these results, treating 296 THP1 cells with compound C, an AMPK inhibitor (58), reversed the reduction of intracellular M.