Remodeling of Mycobacterium tuberculosis lipids regulates prpCD during acid growth arrest

Mycobacterium tuberculosis (Mtb) establishes a state of non-replicating persistence when it is cultured at acidic pH with glycerol as a sole carbon source. Growth can be restored by spontaneous mutations in the ppe51 gene or supplementation with pyruvate, supporting that acid growth arrests is a genetically controlled, adaptive process and not simply a physiological limitation associated with acidic pH. Transcriptional profiling identified the methylcitrate synthase and methylcitrate dehydratase genes (prpC and prpD, respectively) as being selectively induced during acid growth arrest. prpCD along with isocitrate lyase (icl) enable Mtb to detoxify propionyl-CoA through the methylcitrate cycle. The goal of this study was to examine mechanisms underlying the regulation of prpCD during acid growth arrest. Induction of prpCD during acid growth arrest was reduced when the medium was supplemented with vitamin B12 (which enables an alternative propionate detoxification pathway) and enhanced in an icl mutant (which is required for the propionate detoxification), suggesting that Mtb is responding to elevated levels of propionyl-CoA during acidic growth arrest. We hypothesized that an endogenous source of propionyl-CoA generated during metabolism of methyl-branched lipids may be regulating prpCD. Using Mtb radiolabeled with 14C-propionate or 14C-acetate, it was observed that lipids are remodeled during acid growth arrest, with triacylglycerol being catabolized and sulfolipid and trehalose dimycolate being synthesized. Blocking TAG lipolysis using the lipase inhibitor tetrahydrolipstatin, resulted in enhanced prpC induction during acid growth arrest, suggesting that lipid remodeling may function, in part, to detoxify propionate. Notably, prpC was not induced during acid growth arrest when using lactate instead of glycerol. We propose that metabolism of glycerol at acidic pH may result in the accumulation of propionyl-CoA and that lipid remodeling may function as a detoxification mechanism. Importance During infection, Mycobacterium tuberculosis (Mtb) colonizes acidic environments, such as the macrophage phagosome and granuloma. Understanding regulatory and metabolic adaptations that occur in response to acidic pH can provide insights int0 mechanisms used by the bacterium to adapt to the host. We have previously shown that Mtb exhibits pH-dependent metabolic adaptations and requires anaplerotic enzymes, such as Icl1/2 and PckA, to grow optimally at acidic pH. Additionally, we have observed that Mtb can only grow on specific carbon sources at acidic pH. Together these findings show that Mtb integrates environmental pH and carbon source to regulate its metabolism. In this study, it is shown that Mtb remodels its lipids and modulates the expression of propionyl-CoA detoxifying genes prpCD when grown on glycerol at acidic pH. This finding suggests that lipid remodeling at acidic pH may contribute to detoxification of propionyl-CoA, by incorporating the metabolite into methyl-branched cell envelope lipids.

Introduction mutations in PPE51 enable Mtb to grow in acidic medium with glycerol as a sole carbon source, demonstrating that acid growth arrest is a regulated process. This acidic pH-and 101 carbon source-dependent NRP is a new model of Mtb persistence, that is referred to as mechanisms underlying how pH regulates Mtb growth, metabolism, persistence and 104 transcriptional networks.

105
Transcriptional profiling identified several genes with strong induction at acidic 106 pH, independent of carbon source, including genes associated with anaplerotic 107 metabolism (e.g. icl1, pckA, ppdK, and mez), respiration (e.g. type 2 NADH 108 dehydrogenase, ndh), and amino acid/nitrogen metabolism (e.g. arginine biosynthesis, 109 argCDFGHR). Icl1/2 and pckA are induced at acidic pH in both carbon sources and we 110 have previously shown that they are required for optimal growth and metabolic 111 remodeling at acidic pH(11). In addition to inducing genes involved in the anaplerotic 112 node, we also observed that Mtb induces a subset of genes specifically during acidic pH 113 growth arrest(10). Notably, these genes were not induced at acidic pH when 114 supplemented with the growth permissive carbon source pyruvate. Among these acidic 115 pH growth arrest-induced genes were two genes involved in the methylcitrate cycle, 116 encoding methylcitrate synthase and methylcitrate dehydratase (prpC and prpD, 117 respectively). prpCD, along with icl1 which is also a methylisocitrate lyase, have been 118 characterized for their role in the detoxification of propionyl-CoA intermediates generated 119 from the catabolism of cholesterol as well as branched-and odd-chain fatty acids (12-120 16). The prpCD operon is induced by cholesterol, propionate and hypoxia and this 121 induction is dependent on the SigE-regulated Rv1129c regulator (16, 17). However, Mtb 122 cultured at acidic pH in a defined minimal medium does not have an exogenous source 123 of propionyl-CoA, cholesterol or other branched chain precursors. Given the lack of an 124 obvious propionyl-CoA source in the minimal medium conditions of acidic pH growth arrest, understanding the mechanisms of induction of prpCD at acidic pH could provide 126 further insights into the metabolic state of Mtb at acidic pH. Mtb remodels its lipids at 127 acidic pH, including the induction of sulfolipid, 2,3-diacyltrehaloses (DAT) and penta-128 acyltrehaloses (PAT)(10). These lipids and other long-chain fatty acids contain methyl-129 branched lipids, which are preferentially labelled when Mtb is provided 14 C-

130
propionate (18,19). The goal of this study was to test the hypothesis that prpCD is 131 induced during acid growth arrest due to presence of endogenous propionyl-CoA 132 released during lipid remodeling.

137
Induction of prpCD at acidic pH in minimal medium with glycerol as a sole carbon 138 source was verified by quantitative PCR ( Figure 1A). We hypothesized that the observed  Figure 1B). Similarly, in the ∆icl1/2 mutant that lacks methylisocitrate lyase 145 activity, prpC expression was increased ~2-fold at pH 5.7 as compared to WT Mtb 146 ( Figure 1C). Together, these results support the proposal that prpCD induction at acidic 147 pH is linked to propionyl-CoA metabolism.

148
Because prpCD expression at acidic pH responds to changes in propionyl-CoA 149 metabolism, we sought to identify the source of propionyl-CoA that could lead to prpCD 150 induction. The rich medium 7H9+OADC may contain some propionyl-CoA sources from carbon sources into the minimal medium culture. After washing Mtb cultures 3 times in minimal medium prior to transfer to acid growth arrest medium, prpC was still induced 154 following 3 days of acidic growth arrest ( Figure 1D). Furthermore, Mtb grown from a 155 frozen stock grown exclusively in minimal medium containing glycerol as a single carbon 156 source still induced prpC at pH 5.7 with glycerol as a sole carbon source ( Figure 1D).

157
Together, these results suggest that the induction of prpCD is not due to an exogenous despite blocking lipid remodeling, treatment with THL increased prpC expression during 181 pH 5.7 growth arrest 3-fold compared to DMSO treated Mtb ( Figure 3D). This result 182 suggests that lipid remodeling of TAG to SL and TDM is not a source of prpCD induction 183 at acidic pH; instead, the increase in prpC induction with addition of THL suggests that 184 lipid remodeling at acidic pH may act as a mechanism to relieve propionyl-CoA stress.

185
Given that endogenous lipid remodeling does not appear to be the source of 186 prpC induction at acidic pH, we sought to better understand the conditions leading to 187 prpCD induction. Interestingly, although Mtb arrests growth at pH 5.7 with lactate as a 188 single carbon source(10), induction of prpCD is not observed in Mtb cultured with lactate 189 as a single carbon source ( Figure 3E). This observation suggests that prpCD induction 190 during pH 5.7 growth arrest is glycerol-dependent rather than growth arrest dependent.

191
The observation that prpCD induction at pH 5.7 does not appear to be required for Mtb 192 growth arrest is consistent with the previous finding that the role of icl1/2 in growth 193 regulation appears to be independent of the methylcitrate cycle (11).

197
Transcriptional profiling experiments first indicated that prpCD were induced at acidic

211
Our initial hypothesis was that prpCD is induced during acid growth arrest due to

223
The induction of prpCD during acid growth arrest was shown to be dependent on 224 glycerol being present in the media, suggesting that the induction of prpCD is dependent 225 on metabolism of glycerol at acidic pH ( Figure 4A). To our knowledge, glycerol 226 metabolism leading to the production of propionyl-CoA has not been documented in Mtb.

227
However, in other bacterial species, three separate pathways for the production of 228 propionyl-CoA have been described (23). One of these pathways, the propanediol prpCD induction at acidic pH could be secondary to production of propionyl-CoA via this 233 propanediol pathway or a related pathway ( Figure 4B). However, whether Mtb contains 234 enzymes capable of performing this metabolism is not known. Endogenous production of

Figure 2. Mtb utilizes endogenous TAG for the synthesis of TDM and SL at acidic pH.
Mtb was grown in the presence of 14 C-acetate or 14 C-propionate for 3 weeks prior to transferring to minimal medium containing glycerol as a single carbon source buffered to pH 5.7. A) Total radioactivity of Mtb whole cells over time. Over 12 days a ~10% reduction in radioactivity was observed. B) Relative lipid species abundance of triacylglycerol (TAG), sulfolipid (SL), and trehalose dimycolate (TDM) over time in Mtb labelled with 14 C-acetate (Ace) or 14 C-propionate (Prop). C-H) Thin Layer Chromatography (TLC) images showing relative abundance of TAG, SL and TDM at 0, 6, and 12 days after transfer of 14 C-acetate-(C-E) or 14 C-propionate-(F-H) labeled Mtb to acidic pH growth arrest (D0, D6, and D12, respectively).

Figure 3. Inhibition of lipid remodeling at acidic pH increases prpC induction. A-C)
Remodeling of radiolabeled TAG, TDM, and SL at day 0 (D0) and after incubation for 6 days at pH 7.0 or pH 5.7 (D6 7.0 or D6 5.7, respectively) in minimal medium with glycerol as a single carbon source with or without the addition of the lipase inhibitor tetrahydrolipstatin (WT or +THL, respectively). Addition of THL blocks the ability of Mtb to undergo lipid remodeling. D) Addition of THL increases prpC expression at pH 7.0 and pH 5.7. *Significantly induced (p<0.05) at pH 5.7 relative to pH 7.0 (t test). E) prpC is not induced at pH 5.7 with lactate as a single carbon source. Fold change is relative to Mtb grown in glycerol as a sole carbon source at pH 7.0. . Models for prpCD induction and metabolism remodeling during acidic pH growth arrest. A) prpCD is induced during acidic pH growth arrest when glycerol is present in the medium, but not when lactate is the single carbon source. It is speculated that the metabolism of glycerol may lead to de novo synthesis of propionyl-CoA. The increased expression of prpCD in the absence of Mtb lipid remodeling of triacyglycerol (TAG) to sulfolipid (SL) and trehalose dimycolate (TDM) suggests that Mtb uses lipid remodeling as a sink for propionyl-CoA. B) Speculative metabolic pathway for the generation of propionate from glycerol that has been observed in microbes found in the human gut (23).