While Gram-negative bacteria can release outer membrane vesicles directly into the extracellular environment from their outermost compartment, Gram-positive bacteria and mycobacteria have thick cell envelopes surrounding the cytoplasmic membrane that acts as a permeability barrier. The mycobacterial cell envelope is composed of PG covalently linked to AG, which in turn is decorated with exceptionally long-chain mycolic acids (MA). These MA, together with intercalating glycolipids, form the unique mycobacterial outer membrane (OM) (Kalscheuer et al., 2019). Previous analysis of mycobacterial vesicle associated lipids showed predominantly polar lipids, consistent with the cytoplasmic membrane being the likely origin of the vesicles (Prados-Rosales et al., 2011). This is also true in EV produced by Fe-limited Mtb whose lipid content consists mainly of polar lipids and the lipidic siderophore, mycobactin (Prados-Rosales et al., 2014b). Given that these vesicles must traffic from their point of origin in the plasma membrane through the cell envelope, we hypothesized that local remodeling of the cell envelope is necessary for EV release, and that this process may be exacerbated in a virR mutant (virR^mut). To test this, we examined the cell envelope structure of virR^mut grown in minimal medium (MM), as well as wild type (WT) and virR^mut complemented (virR^mut-C) strains (Rath et al., 2013) (Fig 1). Using cryo-electron microscopy (cryo-EM) on whole cells, which allows the study of mycobacterial cell surface-associated compartments in a close-to-native state (Sani et al., 2010), we could easily discern three layers comprising the mycobacterial cell envelope: i) the outer membrane (OM); ii) an intermediate layer composed of two sublayers (L1 and L2, previously assigned to the mycolate-PG-AG (mAGP) network (Sani et al., 2010)); and iii) the cytoplasmic membrane (CM) (Fig 1). We observed a significant increase in the thickness of the overall cell envelope (OM-CM) in the virR^mut strain compared to the WT. This difference was largely explained by the expansion of the CM-L1 layer, which appeared also more granulated, suggesting cell wall alterations closer to the cell membrane and indicating aberrant production of either AG, PG or both. Complementation of virR restored normal cell envelope structure (Fig 1A-C).

In Gram-negative bacteria increased vesiculation has been observed in conditions associated with either loss or maintenance of periplasm integrity (McMillan and Kuehn, 2021). Moreover, a novel mechanism of vesicle production common to both Gram-negative and Gram-positive bacteria has recently been described, in which a cell explosion phenomenon driven by bacteriophages leads to the formation of EVs in culture when bacteria are under genotoxic stress (Nagakubo et al., 2021). Consequently, we tested whether the enhanced vesiculation described in virR^mut (Rath et al., 2013) could be linked to cell lysis events. To do this, we analyzed the presence of cytoplasmic IdeR, a transcription factor which is not present in EVs and is not detected in secreted proteins preparations, in culture supernatants of Mtb strains, assuming that its detection could be linked to a loss of envelope integrity (Gupta et al., 2023). We first confirmed that the virR^mut strain manifested an enhanced vesiculation by measuring EVs using nanoparticle tracking analysis (NTA) (Fig S2A). We did not detect differences in size distribution of isolated EVs between strains. The same supernatants and cell lysates were used to detect the presence of IdeR by immunoblot using a specific polyclonal antibody (Pandey and Rodriguez, 2014). As a control for normal release of an extracellular protein, we used a monoclonal antibody against Ag85b (Prados-Rosales et al., 2017). We could not detect IdeR in the supernatant of WT and virR^mut strain, but a faint band corresponding to IdeR could be observed in the SN of the virR complemented strain (virR^mut-Comp). These results suggest that the absence of virR does not lead to cell lysis but its overexpression may compromise cell envelope integrity (Fig S2B). We validated this approach by serially diluting WT, virR^mutand virR^mut-Comp strains on 7H10 solid media. As shown in Fig S2D, we did not observe differences in the number of cells across the diluted spots between strains. Taken together, the observed changes in the cell envelope of high EV producing bacteria suggest that cell envelope alterations in response to lack of VirR and increased membrane vesicle biogenesis may be connected in Mtb.

The collective cryoEM observations that indicated an enlargement of the compartment next to the cell membrane may be linked to a more permeable cell envelope. Supporting this notion is the recent finding that virR^mut or the mutant in the virR ortholog cei (Rv2700) manifest an enhanced uptake of ethidium bromide (EtBr) relative to WT strain (Ballister et al., 2019). We could reproduce these results and add that virR^mut-Comp strain restores the permeability levels of the WT strain (Fig 2A). This enhanced permeability was concomitant to an increased sensitivity to different antitubercular drugs, including meropenem and vancomycin (Ballister et al., 2019). In agreement with these results, we could measure a significantly enhanced sensitivity of virR^mutto the PG-targeting drug vancomycin relative to WT and virR^mut-Comp strains (Fig 2B). Interestingly, the magnitude of the sensitivity to cell wall targeting drugs was either moderate (ethambutol) or absent (isoniazid) in virR^mut (Ballister et al., 2019).

Furthermore, we observed enhanced incorporation of fluorescent D-amino acids (FDAAs) in virR^mut relative to the WT and complemented strains (Fig 2C). These probes can freely enter cells and may be incorporated into PG by at least three different mechanisms, depending on the species: through the cytoplasmic steps of PG biosynthesis and via two distinct transpeptidation reactions taking place in the periplasm (Kuru et al., 2019). Consequently, the differential labeling observed in virR^mut relative to WT strain may be a consequence of altered PG turnover in the mutant.

We then tested whether the enhanced permeability and vesiculation of virR^mut are also connected in an Mtb mutant lacking cei (Ballister et al., 2019). To do this, we generated cei conditional mutants using the CRISPR interference (CRISPRi) technology (Rock et al., 2017). Further, we also generated virR CRISPRi conditional mutants to properly compare both strains (Fig 2D). When these strains were cultured in the presence of anhydrotetracycline (ATc) to induce the transcriptional silencing of target genes, we could measure enhanced vesiculation by analyzing vesicle concentration in the supernatant by NTA (Fig 2E), relative to strains cultured in the absence of ATc. Overall, these results validate the enhanced permeability phenotype of Mtb mutants in the LytR_C solo domain proteins VirR and Cei and establish that this subfamily of proteins participates in the regulation of vesicle production in Mtb. Moreover, the higher susceptibility to PG-cross-linking inhibitors in both mutants and the enhanced incorporation of FDAAs in the virR mutant strain suggests defects in PG remodeling.

To identify virR-dependent mediators of vesicle production, we carried out a transcriptional profiling of virR^mutand compared it to those from a WT and virR^mut-Comp strains. Previous efforts to define the transcriptional profile of virR^mut showed no significant differences compared to WT (Rath et al., 2013). However, those experiments were carried out with bacteria grown in rich 7H9 medium, while here we use a defined high-Fe minimal medium (MM), which is typically used for vesiculation experiments (Prados-Rosales et al., 2011). We extracted RNA from four independent cultures (average optical densities (OD) of 0.40, 0.38 and 0.42 for WT, virR^mut, and virR^mut -Comp, respectively) one day before vesicle isolation, and conducted RNA-seq (Fig 3A). Principal component analysis (PCA) was then performed on the transcriptomic data, after filtering out lowly expressed genes, as well as genes containing outlier observations (N=3985 genes tested, see Methods), and revealed significant differences between strains. The first principal component, explaining 49% of the variance, separated virR^mut strain samples from strains bearing functional copies of virR (Fig 3B, WT and virR^mut-Comp vs virR^mut, one-tailed Mann-Whitney test p=2.02E-2). Furthermore, the second principal component (28.3% variance) captured virR-independent effects of mutagenesis on transcriptional profiles, as it separated WT samples from virR^mut-Comp and virR^mut (one-tailed Mann-Whitney test p=2.02E-2).

After PCA, we conducted differential expression analysis between strains to find 1808, and 2230 genes differentially expressed in virR^mutcompared to either WT, or virR^mut-Comp strains, respectively (Supplementary file S1, Fig 3C, Benjamini-Hochberg FDR<0.05), representing 45.4% and 56.0% of all genes tested. Although transcriptional profiles of the WT and virR^mut-Comp strains are also significantly diverse, with up to 2184 differentially expressed genes between them, the differences in expression between virR^mut when compared to either WT or virR^mut -Comp are significantly correlated (r=0.63, p<2.2e-16), and the sets of both upregulated and downregulated genes in virR^mut vs any of the other strains are mutually, directionally enriched (Fig 3D). Finally, we defined a set of N=399 genes that are simultaneously upregulated in virR^mut with respect to any of the two strains containing functional copies of virR (logFC>0 and FDR<0.05 for both virR^mut vs WT and virR^mut vs virR^mut-Comp), as well as a set of N=502 genes downregulated in virR^mut vs both virR-containing strains. This way, the differential expression signature found for these genes in the virR^mutstrain is most likely VirR-dependent and not-induced by non-specific consequences of the mutagenesis procedures.

Next, we looked for gene ontologies (GO) enriched among these virR-dependent genes, stratified by the direction of the effects (399 up-regulated, and 502 down-regulated in virR^mut relative to WT and virR^mut-Comp alike). A total of 51 GO terms related with biological processes and cell compartment appeared significantly enriched among genes upregulated in the virR^mut strain (FDR=10%), while 85 terms were enriched among downregulated genes (FDR=10%, see Methods, and supplementary file S2). When focusing on the most prominent of these enrichments (OR>2), the analyses highlighted several groups of biological processes and cell compartment ontologies that appeared altered in virR^mut with respect to both WT and virR^mut-Comp strains (Fig 3E).

First, among genes that were expressed at lower levels in virR^mut, we found terms related to several stresses associated with the intra-phagosomal environment (Schnappinger et al., 2003), such as response to hypoxia (FDR=1.7E-5), heat (FDR = 2.2E-2) or response to metal ions (FDR=3.4E-2), suggesting that responses to host-induced stress are compromised in the mutant, as it is also suggested by the weaker, albeit significant enrichments of responses to host immune response (OR=1.88, FDR=7.47e-02, not highlighted in Fig 3E). Furthermore, the mutant strain showed a reduction in transcripts related with secretion (FDR=7.6e-03), and, specifically, protein secretion by the type VII secretion system (FDR=5.2 e-03). Contributing to this enrichment, we found that the key antigen ESAT-6 was significantly down-regulated in virR^mut with respect to both WT (log2FC=−0.17, FDR=4.1E-3) and virR^mut-Comp strains (log2FC=−0.19, FDR=2.6E-3), while cfp10 (Rv3874), was similarly less expressed in virR^mut than in WT (log2FC=−0.24, FDR=2.6E-3).

Second, we found enriched terms related with lipid metabolism and transport both among the sets of genes up and downregulated in the virR^mut strain. On the one hand, genes downregulated in virR^mut appeared enriched in fatty acid biosynthetic processes (FDR=6.7E-2), suggesting that the regulation of both anabolic and catabolic pathways of lipids, key for bacterial survival and pathogenesis upon infection (Ghazaei, 2018), are, at least in part, VirR-dependent. Furthermore, the mutant strain also shows a significant induction of a number of GO terms related with cell wall composition (e.g. cell wall organization: FDR=3.9e-2), including terms associated with the biosynthesis key components of the mycobacterial cell wall such as PDIMs (DIM/DIP cell wall layer assembly enrichment FDR=9.5e-03) and importantly peptidoglycan (peptidoglycan-based cell wall biogenesis FDR=2.2e-2) (Fig 3F).

Third, we found a series of enrichments related with global regulation of gene expression (positive: FDR=4.4e-2, negative: FDR=5.4e-2), and RNA biosynthesis (positive: FDR=5.4e-2, negative: FDR=5.8e-2), among genes downregulated in virR^mut, suggesting a global alteration of the transcriptional landscape in the mutant.

Fourth, we found that metabolic reprogramming in the virR^mutstrain is not restricted to lipids but also involves amino acids (cellular amino acid metabolic processes FDR=2.1e-3), and carbohydrates (cellular carbohydrate biosynthetic process FDR=6.8e-3), enriched among genes upregulated in the mutant. Similarly, energy production also appears altered in virR^mut, with related GO terms enriched among downregulated genes, such as electron transfer activity, (FDR=4.2e-2), as well as terms enriched among upregulated genes in the mutant such as NADH dehydrogenase activity, (FDR=3.3e-3). All together, these results combined with the differences concerning lipid metabolism discussed above, suggest a global remodeling of metabolic networks in virR^mut.

To shed light on the specific transcriptional circuits whose disruption in the virR^mut strain may lead to the differential gene expression patterns described, we resorted to the regulon annotations previously reported (Minch et al., 2015). In this work, authors collected ChIP-seq data on binding events between a panel of transcription factors (TFs) and DNA to annotate interactions between 143 regulators and the promoters of 7248 target genes (peaks between −150 and +70 nucleotides from transcription staring sites). Capitalizing on these data, we interrogated whether the sets of genes up-(N=399) and downregulated in the virR^mut strain (N=502) were significantly enriched among each of the regulons reported by Minch et al. A total number of 10 regulons appeared enriched among either down (9) or upregulated genes (1) in virR^mut (Supplementary file S3 and Fig 4A, highlighted TFs at OR>2 FDR<0.1). Furthermore, the expression of 7 out of 10 of the TFs controlling these regulons was itself found to be virR-dependent in our transcriptional data (Fig 4B). Genes more highly expressed in the virR^mut strain appeared significantly enriched among Rv0339 targets (OR=9.1, FDR=7.5E-2, Fig 4A), named IniR because of its role as key regulator of the isoniazid-induced operon iniBAC, a key element of the response to cell wall biosynthesis inhibition induced by isoniazid (Boot et al., 2017) and linked to vesicle biogenesis in Mtb (Gupta et al., 2023). Several additional transcription factors known to regulate cell wall biosynthesis and homeostasis were also enriched among genes downregulated in the virR^mut strain. These include SigF (OR=5.8, FDR=8.1E-2), a sigma factor controlling cell envelope organization (Williams et al., 2007), mtrA (OR=2.2, FDR=6.0E-2), a TF controlling peptidoglycan cleavage and alterations in the context of regulation of cell growth and division, whose disruption has recently been shown to lead to increased permeability (Peterson et al., 2023); as well as Rv0472c (OR=11.7, FDR=3.5E-2), a transcription factor involved in the regulation of mycolate remodeling pathways (Peterson et al., 2019), a process used by the pathogen upon phagocytosis to evade macrophage-induced stress and enter dormancy. Interestingly, two genes under the control of Rv0472c, that are key to enable this process, are the mycolate desaturases desA1 and desA2 (Peterson et al., 2019), which appear downregulated in the virR^mut strain with respect to WT (logFC=−0.4, −0.1; FDR=2.3E-11, 3.2E-2, for desA1, and desA2, respectively), as well as with respect to virR^mut-Comp (logFC=−0.4, −0.4; FDR=7.2E-12, 5.3E-21). Furthermore, Rv0472c has also been linked with the regulation of key proteins in the response to several sources of stress in Mtb, such as the putative nitroreductase Rv3131 (response to oxidative/nitrosative stress), as well as the triacylglycerol synthase tgs1, responsible for the synthesis and accumulation of triacylglycerol, a major energy source for the bacterium during dormancy (Chauhan and Tyagi, 2009), both of which are strongly downregulated in the virR^mutstrain vs both WT: (logFC=−1.2, −1.7; FDR=3.1E-19, 3.2E-14) and virR^mut-Comp (logFC=−0.4, −1.9; FDR=3.9E-5, 1.4E-17).

The adjacent genes Rv3131 and tgs1 (Rv3030c) appear under the control of other key regulators of the response to intracellular stress in Mtb and the transition to dormancy, whose regulons appear as well enriched among genes downregulated in virR^mut (Fig 4C). These include devR (OR=2.7, FDR=1.9E-5), the canonical regulator of the response to hypoxia and transition to dormancy in Mtb (Park et al., 2003), as well as Rv1985c (OR=4.1, FDR=4.5E-11), a key regulator of the homeostasis of the threonine biosynthesis pathway, a metabolic route needed by the bacterium for persistence in vivo (Hasenoehrl et al., 2019). Three of the four remaining TFs whose regulons we found enriched among virR^mut downregulated genes have been similarly linked to the response to assorted sources of stress that are cardinal features of the phagosome environment, such as oxidative stress, response to toxic metals, and low pH (Schnappinger et al., 2003). These include the regulator lsr2 (OR=2.3, FDR=5.1E-4), a global histone-like protein responsible from protecting the DNA physically from oxidative stress (Bartek et al., 2014), the zinc-uptake regulator Zur (furB) (Macia̧g et al., 2007) (OR=8.8, FDR=4.5E-2), and tcrX (OR=2.6, FDR=5.0E-2), part of the recently characterized acid-sensing two-component system tcrXY (Stupar et al., 2024). Notably, we observed that removing catalase from culture medium prior to submitting virR^mutto a source of oxidative stress like H202 for 8 h, significantly affected its growth relative to WT and virR^mut-C strains (Fig S3). This result validates the transcriptomics data reflecting a defect in regulating oxidative stress in virR^mut and encourage to investigate its connection with vesicle production.

If enhanced cell permeability is a global consequence of virR disruption that leads to enhanced vesiculation, we hypothesized that other conditions either inducing or decreasing cell envelope permeability could lead to alterations in vesicle production in Mtb. To test this, we treated Mtb with sublethal concentrations of thioridazine (TRZ), a known enhancer of cell permeability (De Keijzer et al., 2016); and, alternatively, we cultured the bacillus with cholesterol as a sole carbon source, a condition which reduces mycobacterial cell permeability (Brzostek et al., 2009) (Fig 5A). The enrichment of the DevR regulon among the genes that are downregulated in virR^mut is also suggestive of the existence of a convergent regulatory route controlling both cell permeability and vesiculation under control of this transcription factor, whose activation has been described as a key part of the pathogen’s transcriptomic response to cholesterol (Soto-Ramirez et al., 2017), concomitant to a reduction in cellular permeability (Brzostek et al., 2009). Importantly, neither cholesterol nor TRZ altered bacterial viability as measured by the absence of the release of the transcription factor IdeR, as an indication of cell lysis or by spotting serially diluted cultures on solid medium from the corresponding conditions (Fig S2C,D). We then measured vesicle levels in culture supernatants and found that Mtb significantly releases more EVs upon treatment with TRZ, and significantly reduces vesicle release in the presence of cholesterol as a sole carbon source (Fig 5B), as measured by dot blot after normalization to colony forming units (CFUs). These results strongly suggest that cell envelope permeability determines the magnitude of the vesiculation phenomenon in Mtb. We next investigated whether common transcriptional features underpin the regulation of cell envelope permeability in the tested conditions. To do that, we compared the transcriptional response of virR^mut with that of available datasets including Mtb cultured in the presence of 0.01% of cholesterol as a sole carbon source (Pawełczyk et al., 2021), and Mtb treated with 6 μg ml-1 of TRZ (Dutta et al., 2010). Experimentally, we retrieved the set of genes up and downregulated in Mtb grown in cholesterol vs glucose-rich culture media (Brzostek et al., 2009), and interrogated whether those genes were enriched among the sets of differentially expressed genes in virR^mut vs WT, and virR^mut vs virR^mut-Comp reported in this study (Fig 5C). Interestingly, we found that those genes whose expression changed in the same direction in conditions associated with low, or high permeability, in both studies were mutually enriched, while the cross-comparisons between sets with opposite responses to low-permeability-associated treatments in both studies led to significant depletion statistics. These results validate the initial hypothesis according to which changes in permeability induced by these two treatments are linked to the activation of a common transcriptional program. The inspection of genes that are simultaneously upregulated in treatments associated with low permeability in both studies identified DevR as a putative regulator associated with the orchestration of this transcriptional footprint of low cell permeability. The two-component system DevS-DevR (also known as DosS-DosR) has a well-known role as a global regulator of the transition to dormancy, and its regulon is consistently upregulated in non-permeable dormant bacteria (Honaker et al., 2009).

Interestingly, authors found that low-permeability bacteria cultured in a cholesterol medium, albeit featuring a regular growing phenotype, activated the DosR regulon, suggesting that cell wall remodeling leading to a decrease in permeability linked with the induction of DosR can be triggered independently to the transcriptional regulatory programs leading to growth arrest and dormancy in Mtb (Pawełczyk et al., 2021). Importantly, the integration of the genes that are downregulated in virR^mut validates this hypothesis (Fig 5D and FigS4), since not only the intersection between the DevR regulon and the set of genes downregulated in virR^mut are significantly larger than expected by chance (see Fig 4A, OR=2.67, p=2.7E-7), but also the intersection between DevR regulon and the genes that are activated by cholesterol in Brzostek et al. data (OR=1.89, p=3.5e-5). 30 out of the 173 genes annotated in the DosR regulon described in (Minch et al., 2015) are simultaneously upregulated in conditions linked to low permeability in both studies.

A similar analytic approach for the comparison of the response of virR^mut and the response to the drug TRZ, a treatment well-known for inducing an increase in permeability in Mtb-treated cells (Dutta et al., 2010) (GSE16626), led to conceptually similar results, with genes simultaneously downregulated in both TRZ and virR^mut (OR=1.96, p=9.1E-4) being enriched. Interestingly, among the 22 genes showing the largest, and most significant repression effects by both treatments (log2FC>0.25 & FDR<0.05 simultaneously, for TRZ effects, virR^mut vs WT), we found 4 genes whose functions are related with metal nutrients uptake and homeostasis, such as genes encoding the bacterioferritin BrfB, the metal exporter CtpJ, the cupper regulator CsoR and Moew, involved in Molibdenum-cofactors biosynthesis (Fig 5E). Taken together, these analyses highlight the transcriptional parallels between virR^mut and both TRZ and cholesterol conditions, which strongly suggests the existence of a common transcriptional signature associated with the control of cell permeability.

We next wondered about the effect of culturing virR^mut on both cholesterol or TRZ could have on cell growth, permeability and EV production. In the case of cholesterol, it has also been shown to affect other aspects of physiology (redox, respiration, ATP), which can directly affect permeability (Lu et al., 2017). We monitored virR^mutgrowth in MM supplemented with either glycerol, cholesterol as a sole carbon source, or TRZ at 3 ug ml⁻¹ for 20 days. While cholesterol significantly enhanced the growth virR^mut after 5 days relative to glycerol medium, supplementation of glycerol medium with TRZ restricted growth during the whole time-course (Fig S5A). The study of cell permeability in the same conditions indicated that the enhanced cell permeability observed in glycerol MM was reduced when virR^mut when cultured with cholesterol as sole carbon source. Conversely, the presence of TRZ increased cell permeability relative to the medium containing solely glycerol (Fig S5C). As we have previously observed for the WT strain, either condition (Chol or TRZ) also modified vesiculation levels in the mutant accordingly (Fig S5B). Interestingly, TLC analysis of cell membrane polar lipids showed that virR^mutgrown in cholesterol restores the previously observed higher Ac₂PIM₂ levels relative to WT when growing in glycerol (Fig S8A). These results strongly indicates that other aspects of mycobacterial physiology besides permeability are also affected in the absence of VirR and may contribute to the observed enhanced vesiculation.

To capture the impact of transcriptional adaptation to the proteome, and the degree of divergence between transcriptional shifts and proteins changes observed in the absence of virR, we next performed label-free peptide mass spectrometry (LF-MS) of whole cell lysates (WCLs) of virR^mut, WT, and virR^mut-Comp strains. Additionally, we isolated EVs from axenic cultures and obtained their proteomic composition to examine for potential VirR-dependent differences in protein cargo that may alter the functionality of EVs (Fig 6A). As a result, we obtained proteomic profiles across strains and cell compartments for a total of N=935 proteins. Upon PCA analysis of the proteomic data (Fig 6B), we found that the first principal component (PC1), explaining 36.9% of total variance, separated samples from WCLs and EVs in all strains (PC1 differs across cell compartments: one-tailed Mann Whitney test p=2.1e-05), while the second PC (PC2) (13.6% of total variance explained) captured strain effects on the proteomic footprint only among EVs (one tailed Mann Whitney t test for PC2 differences between virR^mut EVs and those from either WT or virR^mut -Comp strains (one-tailed Mann Whitney p=1.2 e-02). These results suggest that differences in protein abundances between EVs and WCLs are larger than differences found between strains, and that, concerning the differences across strains, these are more important among EVs. To verify this, we conducted differential protein expression analyses to find barely 27 proteins differentially expressed between virR^mutand any of the two virR-expressing strains, which represents 2.9% of the total of proteins under analyses (supplementary file S4). Despite the scarcity of significant differences in expression across strains in WCLs, some of the features most differentially expressed in virR^mut are, besides VirR itself (Fig 6C), proteins involved in key processes for bacterial survival and virulence, including key metabolic enzymes such as the Acetyl-coenzyme A synthetase Acs (logFC = −3.92 (FDR=4.5e-3) and −2.96 (FDR=5.2E-2) for the differences between virR^mut and WT, and between virR^mutand virR^mut-Comp, respectively), as well as FprA, (logFC= −3.12, FDR=4.5e-3 e-02 for virR^mut-WT and logFC= −2.16, FDR=5.9 e-02 for virR^mut-virR^mut-Comp), involved in the reduction of NADP+ into NADPH, both downregulated in virR^mut.

Conversely to the WCL dataset, we did find more frequent differences in protein expression across strains in EVs, with as many as 361, and 382 proteins differentially expressed in virR^mut EVs when compared to those from WT and virR^mut-Comp strains, respectively (supplementary file S4, Fig 6D). This represents 38.6 and 40.9% of all proteins tested, respectively. Similarly to what was observed at the level of RNA expression, differences in expression for the contrasts (virR^mut-WT) and (virR^mut-virR^mut-Comp) were directionally enriched (Fig 6E), thus motivating the definition of two sets of proteins that 1) were observed at higher levels in the EVs isolated from virR^mut than in those from both WT and virR^mut -Comp strains (N=117 virR^mut -upregulated proteins); or 2) were observed at lower levels in the EVs from virR^mutthan in those from any of the other two strains (N=108 virR^mut-downregulated proteins). Interestingly, as many as 21 out of the 117 proteins that appeared enriched in VirR^mut EVs were also upregulated in virR^mutat the transcriptional level (One-sided Fisher exact test of enrichment OR=2.02, p=7.0e-3) suggesting that at least a part of these differences in EVs’ protein cargo are a consequence of differences in gene expression already observed at a transcriptional level. Notwithstanding that, proteins appearing at lower concentrations in virR^mut EVs were not enriched among genes bearing significant transcriptional effects in the mutant. Regarding downregulated EV proteins in virR^mut we observed robust enrichments in GO terms related with protein biosynthesis, including translation (FDR=9.8 e-22), ribosome (FDR=5.3 e-29), and ribosome biogenesis (FDR=4.1 e-27, Fig 6F and supplementary file S5). Consistent with these results, 44 ribosomal proteins (out of 49 detected) were found to be significantly less expressed in the virR^mutstrain relative to either WT or virR^mut-Comp strains (FDR<0.05 for at least one comparison), while no ribosomal protein appeared significantly upregulated in virR^mut. Furthermore, 30 of these proteins were simultaneously downregulated in virR^mut when compared to both VirR-containing strains (Fig S6) implying a stark ribosomal depletion in the EVs in the absence of VirR. Furthermore, we found that the two bacterioferritin proteins BfrA and BfrB were simultaneously less abundant in virR^mut than in both WT and virR^mut-Comp strains (logFC<-1.5 and FDR<7.3E-4, for differences virR^mut vs WT and virR^mut vs virR^mut-Comp in both genes, Fig 6G, and supplementary file S4), together contributing to the enrichment of the GO term ferroxidase activity (FDR=6.4 e-02, Fig 6F). This result is suggestive of a possible defect of virR^mut EVs in iron sequestration (Mohammadzadeh et al., 2021; Prados-Rosales et al., 2014b). We also found a series of metabolic pathways enriched among proteins upregulated in the EVs of the virR^mut strain, including enzymes involved in metabolism of nucleotides (e.g. purine nucleotide biosynthetic process FDR=2.0e-02, amino acids (alpha amino acids FDR=8.3e-04), as well as carbohydrates (hexose metabolic process FDR=6.3e-2), whose catabolism appears also enriched among proteins downregulated in the mutant (e.g. glycolytic process FDR=2.2e-2 and carbohydrate catabolic process FDR=4.7E-2). The simultaneous enrichments of these metabolic pathways observed both among upregulated and downregulated vesicular proteins (Fig S7), indicates a complex, VirR-dependent, rewiring of metabolic routes in the vesicles from the mutant strain.

Further, proteins upregulated in virR^mut EVs (N=117) were enriched in terms related to mycolic acids metabolic processes (FDR=2.1E-2), and peptidoglycan-based cell-wall biogenesis (FDR=9.0 e-02). Proteins such as InhA (logFC=3.71, 2.26 FDR=2.20 e-05, and 1.4E-3 for comparisons virR^mut vs WT and virR^mut vs virR^mut-Comp, respectively), and KasB (logFC=1.83; 2.49 FDR=2.23 e-02; 4.4e-3), involved in mycolic acid metabolism were also upregulated in virR^mut (Fig 6H).

The upregulation of these proteins, along with the enrichments at the transcriptional level of DIM/DIP biosynthetic pathways (upregulated in virR^mut), suggests a shift in the control of the pool of propionyl-CoA, both at the level of its production, -as a byproduct of the catabolism of methyl-branched lipids, odd-chain-length fatty acids as well as cholesterol- and its detoxification by the bacterium through different routes, such as the methylmalonyl pathway, the glyoxylate cycle and the methylcitrate cycles (Lee et al., 2013). To shed light on these changes at the metabolic level, we conducted a metabolic profiling to quantify the concentration of metabolites that are central in these different detoxification routes including propionyl-CoA: acetyl-CoA, pyruvate, methylcitrate, and methylmalonyl-CoA (Fig 6I). Only methylmalonyl-CoA was found to be significantly repressed in virR^mut (p=0.028). We also observed differences between virR^mut and virR^mut -Comp strains in methylcitrate. This result, along with the trend towards increased protein, and especially gene expression that is observed for genes involved in the biosynthesis of PDIMs from methylmalonyl (including phenolpthiocerol synthesis type-I polyketide synthases A-E, (ppsA-E, Fig 6I), among others), suggests that the methylmalonyl pool is depleted in the mutant strain because of an increased activity of the PDIM biosynthetic pathway. Furthermore, most of the subunits of the acetyl/propionyl-CoA carboxylase Acc (AccA1, AccA2, AccA3, AccD1, AccD5 appear repressed in virR^mut EVs (Fig 6I), suggesting that the production of methylmalonyl Co-A from the degradation of propionyl CoA is inhibited within EVs of the mutant strain.

Taken together, the omics analyses align with the structural and functional differences observed in virR^mut. We reasoned that the increased expression of PDIMs biosynthetic genes and the reduced expression of genes related to protein synthesis may be triggered to compensate the enhanced permeability measured in virR^mut. In addition, this enhanced permeability leads to an increase in the release of EV, which seems to be depleted in ribosomes and iron storage proteins.

To elucidate the specific molecular defects that explain the periplasm enlargement, the enhanced permeability, and the antibiotic sensitivity phenotypes, we performed a biochemical analysis of WT and virR^mutcell envelopes. First, we analyzed the total lipid profile by liquid chromatography-mass spectrometry (LC-MS) and observed significant differences in prevalent polar lipids linked to the cell membrane including, lysophosphatidic and phosphatidic acid, species of monoacylated and diacylated glycerol, phosphatidylethanolamines and phosphatidyl-myo-inositol mannoside (PIM) species, being more abundant in virR^mutrelative to WT and complementing strain. (Fig 7A). To validate these results, we ran thin layer chromatography (TLC) experiments on the same lipid extracts and found that the mutant carries more phospholipids in the membrane including more AcPIM₂ and PE (Fig 7B). Moreover, we could not measure major differences in the level of mycolic acid species (Fig 7C), while significantly more PDIMs were detected in the mutant relative to the other strains (Fig 7D). This later result partially validates proteomics and metabolomic data and suggest that the methylmalonyl pool is depleted in the mutant strain because of an increased activity of the PDIM biosynthetic pathway. However, other lipids including SL1 or PAT/DAT that share a common precursor (propionyl-CoA) with PDIM may also explain the depletion of methylmalonyl. Notably, we observed a reduction in the level of these lipids (SL1 or PAT/DAT) in virR^mutcultured in glycerol relative to WT and complemented strains, suggesting that the excess of PDIM synthesis can take away methyl malonyl CoA from the biosynthesis of SL-1 and PAT/DAT in the absence of VirR (Fig S8B).

When analyzing apolar lipids, we also observed increased levels of free mycolic acids (FMA) and free fatty acids (FFA) upon virR deletion (Fig 7E). The enhanced release of these apolar lipids has also been documented in a C. glutamicum mutant in Cg0847 (Baumgart et al., 2016), a canonical LCP protein. Importantly, we could not detect any of these lipids in EVs from virR^mut (Fig 7F), while major polar phospholipids could be resolved in EVs from all strains (Fig 7G), as previously reported (Prados-Rosales et al., 2011). This result indicates that Mtb releases free apolar lipids independently from membrane associated EVs.

Next, we performed a glycosyl composition analysis of isolated cell walls (including mycolyl-arabinogalactan-peptidoglycan complex) and determined that there were no significant differences between strains in AG content as well as in the ratio between arabinose and galactose (A/G) (Fig 8A). Similarly, no changes were observed in the galactan length as measured by the ratio between galactose (G) and rhamnose (R) G/R) (Justen et al., 2020). We noticed that the ratio between N-acetylglucosamine (NAGc) and rhamnose (R) (NAGc/R) was significantly higher in virR^mut relative to WT, suggesting an increase in PG content in this strain (Fig 8A). The validation of these results was performed independently by measuring the amount of diaminopimelic acid (DAP) in isolated cell walls, as a surrogate of PG content, and determined that virR^mut carries two times more PG than WT or complemented strains (Fig 8B).

We next obtained pure PG preparations of each strain and performed an analysis by atomic force microscopy (AFM) (Fig 8C and Fig S9). This technique has been recently used to investigate the nanometric features of isolated Gram-positive bacterial cell walls (Pasquina-Lemonche et al., 2020). Semi-automated analysis of liquid AFM images from mycobacterial isolated PG samples revealed that: (i) the absence of VirR leads to a significant reduction in PG pore diameter (down to 9.2 ± 0.6 nm) relative to WT (16 ± 3.3 nm) and complemented strains (10 ± 0.8 nm) (Fig 8C, D and Fig S9); (ii) the number of pores is significantly higher in virR^mut compared to WT and complemented strain (Fig 8E and Fig S9); (iii) manual analysis of AFM images in ambient conditions revealed that virR^mut PG is significantly thicker than that of the other strains of the study (Fig 8F). These results show for the first time the nanometric architecture of isolated PG in mycobacteria by AFM and suggest a role for VirR in regulating PG thickness and porosity. Strikingly, we found that the PG enlargement observed in the virR^mut correlated with increased susceptibility to the muramidase lysozyme (Fig S10), indicating modifications in PG remodeling in virR^mut. In fact, a deeper MS analysis of PG showed a significant increase in the abundance of deacetylated muropeptides in virR^mut relative WT and virR^mut-Comp strain (Fig 8G, H). This PG feature has been linked to variability in lysozyme susceptibility in other Gram positive bacterial species (Bera et al., 2005). Collectively, although the structural changes at nanoscopic level of the peripheral areas of the cell observed by AFM leads to a more dense and thick structure harder to penetrate, the chemical changes in the peptidoglycan structure observed and biochemical assays suggest that changes associated with alterations in VirR-regulated PG remodeling makes the wall more permeable and support increased release of EV in Mtb.

As mentioned above VirR is a LytR_C solo domain protein with no catalytic LCP domain. Other members of the LCP protein family like Rv3267 and Rv3287 are the main PG-AG ligases (Grzegorzewicz et al., 2016; Harrison et al., 2016) and also carry a LytR_C domain. We have previously shown that VirR interacts with itself (Rath et al., 2013). Consequently, we reasoned that if VirR contributed to PG-AG ligation it could do it via interaction with canonical LCP proteins through the LytR_C domain. Although VirR is one of the most abundant proteins found in tuberculin preparations (Prasad et al., 2013), we determined that it localizes at sites of nascent cell wall by fluorescence microscopy (Fig S11). Therefore, its location could initially allow for interactions with other cell surface proteins like LCP. Both recombinant VirR and the main AG-PG ligase Lcp1 (Rv3267) oligomerize in a blue-native PAGE gel (Fig 9A). This observation suggests that they interact with proteins sharing similar domain features. We next performed a flotation assay where we examined the behavior of VirR on an iodixanol gradient in the presence or absence of Lcp1. Using a specific antibody raised against VirR, we observed that the presence of Lcp1 displaces the relative location of VirR toward upper fractions of the gradient, indicating a potential interaction of both proteins and the formation of higher molecular weight complexes (Fig 9B). To investigate the interaction of virR and Lcp1, along with other Lcp proteins, in a more physiological environment we used the bipartite splitGFP approach (Cabantous et al., 2013) (Fig 9C and Fig S12), which has been recently utilized to demonstrate the co-localization and interaction of the membrane-bound metal ATPase CtpC with the chaperon PacL1 (Boudehen et al., 2022). We first recapitulated the VirR self-interaction and added that such interaction implicates the LytR_C domain since no differences in fluorescence were measured between full and truncated versions of VirR (VirRsol) lacking the first 41 amino acids. As negative controls, we included interactions between VirR and the cell surface associated metal ATPase CtpC (Fig 9C). When testing interactions between VirR and other Lcp proteins (Rv0822c, Lcp1 (Rv3267) and Rv3284) we could not measure any GFP signal when full Lcp proteins were used in the assay. Conversely, a significant increase in fluorescence was measured for all combinations when truncated Lcp proteins including either the C-terminal domain or specifically the LytR_C domain were tested (Fig 9C). These results strongly suggest that VirR interacts with Lcp proteins and that these interactions occur via LytR_C terminal domains.

Mtb H37Rv (ATCC) and derivative strains virR^mut and virR^mut-C (complemented strain) (Rath et al., 2013) were used in this study. Bacteria were initially grown in Middlebrook 7H9 medium (7H9) supplemented with 10% (v/v) oleic acid-albumin-dextrose-catalase (OADC) enrichment (Becton Dickinson Microbiology Systems, Spark, MD, United States), 0.5% (v/v) glycerol and with Tyloxapol 0.05% (v/v; Sigma-Aldrich, Burlington, MA, United States) prior to inoculation in a minimal medium (MM) consisting of KH₂PO₄ 1 g/l, Na₂HPO₄ 2.5 g/l, asparagine 0.5 g/l, ferric ammonium citrate 50 mg/l, CaCl₂ 0.5 mg/l, ZnSO₄ 0.1 mg/l, without Tyloxapol 0.05% (v/v), containing 0.7% (v/v) glycerol, pH 7.0.

Mtb was also cultured in MM in the presence of sublethal concentrations of thirodazine (TRZ) of 6 µgml⁻¹ as previously described. We determined that this concentration does not affect cell growth. Mtb was also cultured in MM with cholesterol as a sole carbon source as previously described (Brzostek et al., 2009).

Mtb expressing fluorescent VirR was generated by cloning VirR into pMV261-Venus using fast cloning approach (Li et al., 2011).

M. smegmatis mc²155 (ATCC 700084) and recombinants derived from this strain were grown at 37^°C in 7H9 medium (Difco) supplemented with 10% albumin-dextrose-catalase (ADC, Difco) and 0.05% Tween-80 (Sigma-Aldrich), or on complete 7H11 solid medium (Difco) supplemented with 10% OADC(Difco). When needed, streptomycin (25 μg ml⁻¹) was added to the culture media. Escherichia coli strains StellarTM (Takara bio, San Jose, CA, United States) were grown at 37^°C in LB broth (Difco) or on L-agar plates (Difco) supplemented with streptomycin (25 μg ml⁻¹) when required.

Plasmids used for split GFP experiments were constructed by modifying the plasmid pGMCS-PpacL1-pacL1-Lk2-GFP11-ctpC(1-85)-Lk15-GFP(1-10) (Boudehen et al., 2022) (Table S1). Plasmids pGMCS, conferring resistance to streptomycin, are shuttle vectors, episomal in E. coli and integrative in mycobacteria through insertion at the attL5 mycobacteriophage insertion site in the glyV tRNA gene. Lk2 (LEGSGGGGSGGGS) and Lk15 (GPGLSGLGGGGGSLG) are flexible linkers separating the protein domains of interest from the last 11th b-sheet or the first ten b-sheets of GFP, respectively. First, the Zinc-inducible PpacL1 promoter was replaced by P1, a strong constitutive promoter from M. smegmatis (Ariyachaokun et al., 2020), resulting in pGMCS-P1-pacL1-Lk2-GFP11-ctpC(1-85)-Lk15-GFP(1-10). Second, the pacL1 (Rv3269) sequence was replaced by that of virR (encoding amino-acids 1 to 164) or virRsol (encoding amino-acids 42 to 164) by In-Fusion HD cloning (Takara). PCR fragments were amplified using appropriate primer pairs (Table S1) and Phusion High-Fidelity PCR Master Mix with GC Buffer (Thermo Scientific, Walthman, MA, United States). Linear fragments were purified on agarose gels and inserted into pGMCS backbone linearized by PCR amplification with appropriate primers, using the In-Fusion HD Cloning Kit (Takara), following the manufacturer’s instructions. Finally, the ctpC sequence was replaced on the resulting plasmids by the virR or virRsol sequences, or that encoding either the C-terminal part or only the LytR_C domains of the Rv0822c, Rv3267 or Rv3484 proteins from M. tuberculosis H37Rv. Plasmids were constructed by transformation into E. coli Stellar recipient cells. All plasmids were verified by sequencing before introduction by electroporation into M. smegmatis mc²155 strain.

Bacterial strains were initially cultured in 7H9+ADN until they reached an OD600nm of 0.5. Cells were washed in PBS and transferred to roller bottles including the same medium and 13 mM H₂0₂. After 8 h, bacterial cultures were serially diluted and spotted onto 7H10 solid medium.

To generate the conditional CRISPR mutants we took advantage of the CRISPR interference (CRISPRi) technology (Rock et al., 2017). Briefly, we designed sgRNA oligonucleotides containing the target Rv0431 (virR) and Rv2700 (cei) genes sequences (Table S1) located immediately upstream of the PAM sequence. The strengths of the PAM sequence were chosen following authors recommendation to get high and intermediate gene repression, 158.1 or 42.2-fold repression for virR and 145.2 and 47.3 for cei, respectively. The primers were annealed and ligated into the gel purified BsmBI-digested CRISPRi vector (plJR965 plasmid) followed by transformation of competent Mtb cells via electroporation. Transcriptional knockdown of genes were quantified by qRT-PCR; and protein levels were examined by immunoblot using specific antibodies raised against VirR (Rath et al., 2013) and Cei (Ballister et al., 2019). Complementation of virR mutant strain was performed using the plasmid previously generated by Gateway® Cloning Technology (Invitrogen) (Rath et al., 2013). Importantly, neither virR^mutnor the virR^mut-C showed altered expression of downstream genes including Rv0432 or Rv0433 relative to WT strain (Fig S13).

Mtb-EV were prepared and purified as previously described. Briefly, the culture filtrate of 1 L cultures of Mtb grown in low iron MM without detergent for 14 days was processed for vesicle isolation by differential centrifugation. In parallel a 2 ml culture in MM supplemented with 0.05% Tyloxapol to disperse bacterial clumps was used to determine viability by plating culture dilutions onto 7H10 agar plates and enumerating colony forming unit (CFU) at the time of culture filtrate collection.

The membranous pellet containing vesicles obtained after ultracentrifugation of the culture filtrate at 100,000 × g for 2 h at 4^°C, was resuspended in 1 ml sterile phosphate-buffered saline (PBS) and overlaid with a series of Optiprep (Sigma-Aldrich) gradient layers with concentrations ranging from 30% to 5% (w/v). The gradients were centrifuged at 100,000 × g for 16 h. At the end 1 ml fractions were removed from the top, diluted to 20 ml with PBS and purified vesicles recovered by sedimentation at 100,000 × g for 1 h. Vesicle pellets were suspended in PBS before analysis. Protein concentration was measured by Bradford assay (Bio Rad, Hercules, CA, United States) according to manufacturer instructions. For all conditions test from which EVs were isolated, normalization was performed by referring EVs concentration to colony forming units (CFUs).

NTA was conducted using ZetaView (Particle Metrix, Winning am Ammersee, Germany). Instrument calibration was performed prior to EV analysis using 102 nm polystyrene beads (Thermo Fisher Scientific), according to manufacturer instructions. Measurements were performed using a 405 nm 68 mW laser and CMOS camera by scanning 11 cell positions and capturing 60 frames per position at 25 C with camera sensitivity 85, shutter speed 100, autofocus and automatic scattering intensity. Samples were diluted in pre-filtered PBS to approximately 106-107 particles⋅ml⁻¹ in Millipore DI water. Analysis was performed using ZetaView Software version 8.05.12 SP1 with a minimum brightness of 30, maximum brightness of 255, minimum area of 5, maximum area of 1000, and s minimum trace length 15. Triplicate videos of each sample were taken in light scatter mode. Particle size and concentration were analyzed using the built-in EMV protocol and plotted using Prism software, version 8.0.1 (GraphPad Inc., San Diego, CA, United States).

Cells were fixed with 2% glutaraldehyde in 0.1 M cacodylate at room temperature for 24 h, and then incubated overnight in 4% formaldehyde, 1% glutaraldehyde, and 0.1% PBS. For scanning microscopy, samples were then dehydrated through a graded series of ethanol solutions before critical-point drying using liquid carbon dioxide in a Toumisis Samdri 795 device and sputter-coating with gold-palladium in a Denton Vacuum Desk-2 device. Samples were examined in a Zeiss Supra Field Emission Scanning Electron Microscope (Carl Zeiss Microscopy, LLC, North America), using an accelerating voltage of 5 KV.

For Cryo-EM, grids were prepared following standard procedures and observed at liquid nitrogen temperatures in a JEM-2200FS/CR transmission electron microscope (JEOL Europe, Croissy-sur-Seine, France) operated at 200 kV. An in-column omega energy filter helped to record images with improved signal/noise ratio by zero-loss filtering. The energy selecting slit width was set at 9 eV. Digital images were recorded on an UltraScan4000 CCD camera under low-dose conditions at a magnification of 55,058 obtaining a final pixel size of 2.7 Å/pixel.

Density profiles were calculated along rectangular selections with the ImageJ software (NIH, Bethesda, MD, United States). The number of cells analyzed varied from n = 20 (WT), n = 15 (virR^mut) and n = 23 (virR^mut-C). The comparisons extracted from this data is statistically relevant with t test with Welch’s correction performed. For each cell three representative measurements were obtained. Mean and standard error was then calculated for measurements pooled for each strain.

Two µl of EV isolates and two-fold serial dilutions were spotted onto a nitrocellulose membrane (Abcam, Cambridge, United Kingdom) and process for dot blot using anti-EV polyclonal murine serum (Prados-Rosales et al., 2014a) as primary antibody and goat anti-mouse-HRP conjugated as secondary antibody and ECL prime Western Blotting Detection chemiluminescent substrate (GE Healthcare, Chicago, IL, United States). The signal was visualized in a Chemidoc MP imaging system (Bio Rad)s).

Mtb strains were grown in MM supplemented with 0.05% Tyloxapol at 37°C until they reached an OD595nm of 0.3 and harvested by centrifugation. The cell pellets were resuspended in 1 ml Qiagen RNA protect reagent (Qiagen, Venlo, Netherlands) and incubated for 24 h at room temperature. Cells were disrupted by mechanical lysis in a FastPrep-24 instrument (MP Biomedicals, Santa Ana, CA, United States) in Lysing Matrix B tubes and RNA was purified with the Direct-zol RNA miniprep kit (Zymo Research, Irvine, CA, United States). The quantity and quality of the RNAs were evaluated using Qubit RNA HS Assay Kit (Thermo Fisher Scientific) and Agilent RNA 6000 Nano Chips (Agilent Technologies, Santa Clara, CA, United States), respectively. Sequencing libraries were prepared using the Ribo-Zero rRNA Removal Kit (Gram-positive Bacteria) (Illumina Inc., San Diego, CA, United Stated) and the TruSeqStranded mRNA library prep kit (Illumina Inc), following the Ribo-Zero rRNA Removal kit Reference guide and the “TruSeqStranded mRNA Sample Preparation Guide. Briefly, starting from 1 µg of total RNA, bacterial rRNA was removed and the remaining RNA was cleaned up using AgencourtRNAClean XP beads (Beckman Coulter). Purified RNA was fragmented and primed for cDNA synthesis. cDNA first strand was synthesized with SuperScript-II Reverse Transcriptase (Thermo Fisher Scientific) for 10 min at 25^°C, 15 min at 42^°C, 15 min at 70^°C and pause at 4^°C. cDNA second strand was synthesized with Illumina reagents at 16^°C for 1 hr. Then, A-tailing and adaptor ligation were performed. Libraries enrichment was achieved by PCR (30 sec at 98^°C; 15 cycles of 10 sec at 98^°C, 30 sec at 60^°C, 30 sec at 72^°C; 5 min at 72^°C and pause at 4^°C). Afterwards, libraries were visualized on an Agilent 2100 Bioanalyzer using Agilent High Sensitivity DNA kit (Agilent Technologies) and quantified using Qubit dsDNA HS DNA Kit (Thermo Fisher Scientific). Library sequencing was carried out on an Illumina HiSeq2500 sequencer with 50 nucleotides single end reads and at least 20 million reads per individual library were obtained.

Quality Control of sequenced samples was performed by FASTQC software (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were mapped against the M. tuberculosis H37Rv strain (GCF_000195955.2_ASM19595v2) reference genome using Tophat (Trapnell et al., 2009) with --bowtie1 option, to align 50 bp reads. The resulting BAM alignment files for the samples were the input to Rsubread’s (Liao et al., 2019) featureCounts function to generate a table of raw counts required for the Differential Expression (DE) analysis. Data preprocessing and differential expression analysis of the transcriptomic data was performed using DeSeq2 (Love et al., 2014). Genes with a median of expression lower than 1 CPM, and genes containing outlier observations were filtered before statistical modeling, and the remaining N=4004 genes were used for differential expression analysis. P-values were corrected for multiple testing using the R package, implementing the Storey-Tibshirani FDR correction method (Storey et al., 2023). Genes with an FDR<0.05 were deemed as differentially expressed. Data were deposited at Gene expression Omnibus (GEO); GSE143996.

For reverse transcription cDNA was obtained from 10 µl of isolated RNA. Each reaction included an RT control, where RNA was replaced by 10 µl of DEPC-treated water.

Reaction mixture 1 (volumes refer to a single sample) was prepared, containing 2 µl of random primers 500 ng/µl; 2 µl of dNTPs mi× 10 mM; and 12.5 µl of DEPC-treated water. A total of 16.5 µl of this mixture was added to 10 µl of each sample, and the samples were incubated for 5 minutes at 65 °C to remove secondary structures. After 2 minutes on ice, samples were centrifuged at 12,000 rpm for 5 minutes. Reaction mixture 2 (volumes refer to a single sample) was prepared by mixing 8 µl of 5X reverse transcription buffer; 2 µl of 0.1 M DTT; 2 µl of RNAseOUT inhibitor; and 1.5 µl of SuperScript IV reverse transcriptase 200 U/µl. This reverse transcriptase was chosen for its high specificity in synthesizing cDNA, particularly for nucleic acids with high G+C content, as is the case with M. tuberculosis. A total of 13.5 µl of mixture 2 was added to each sample, obtaining a final volume of 40 µl. The samples were incubated for 10 minutes at 25 °C and then at 55 °C for 30 minutes to synthesize cDNA. Finally, the reaction was stopped by inactivating the enzyme for 10 minutes at 80 °C and centrifuging briefly to recover the full volume. The samples were stored at −20 °C until use.

Amplification reactions were conducted by preparing a PCR reagent mixture containing, per sample: 1 µl of MgCl2; 1 µl of SYBR Green (previously mixed with L.C. Fast Start containing the polymerase); 0.25 µl of oligonucleotides specific to the sequences of interest (Table 1) at a concentration of 20 µM; and 4.5 µl of H2O. In this case, the sequences of interest are the Rv0432 (Forward primer: GCGTCTACAGTTCCGGGTAC; Reverse primer: AGGGTACTGGTCAGGCTCTG) and Rv0433 (Forward primer: GCTGATCTGGGGTGTACACG; Reverse primer: GCCAACAGATGCGGGTAGTA) genes. The reaction was performed in a Light Cycler using capillaries containing 7 µl of the PCR mixture and 3 µl of each sample. For the negative control, the 3 µl were replaced with water, and for the positive control, they were replaced with 0.1 ng/µl DNA from the reference strain of M. tuberculosis. After a brief centrifuge pulse, the volume was introduced into the capillaries before analysis in the Light Cycler. Gene fold induction normalized to 16S rRNA was calculated using the 2^−ΔΔCt method.

Total cell proteins and MEV-associated proteins were submitted to LC-MS label-free analysis using a Synapt G2Si ESI Q-Mobility-TOF spectrometer (Waters) coupled online to a NanoAcquity nano-HPLC (Waters), equipped with a Waters BEH C18 nano-column (200mm x 75 um ID, 1.8um). Samples were incubated and digested following the filter-aided sample preparation (FASP) protocol (Wiśniewski et al., 2009) Trypsin was added to a trypsin:protein ratio of 1:50, and the mixture was incubated overnight at 37^°C, dried out in a RVC2 25 speedvac concentrator (Christ), and resuspended in 0.1% FA. Peptides were desalted and resuspended in 0.1% FA using C18 stage tips (Millipore). Digested samples (500 ng) were loaded onto the LC system and resolved using a 60 min gradient (5 to 60% ACN). Data was acquired in HDDA mode that enhances signal intensities using the ion mobility separation step. Protein identification and quantification were carried out using Progenesis LC-MS (version 2.0.5556.29015, Nonlinear Dynamics). One of the runs was used as the reference to which the precursor masses in all other samples were aligned to. Only features comprising charges of 2+ and 3+ were selected. The raw abundances of each feature were automatically normalized and logarithmized against the reference run. Samples were grouped in accordance with the comparison being performed, and an ANOVA analysis was performed. A peak list containing the information of all the features was generated and exported to the Mascot search engine (Matrix Science Ltd.). This file was searched against a Uniprot/Swissprot database, and the list of identified peptides was imported back to Progenesis LC-MS. Protein quantitation was performed based on the three most intense non-conflicting peptides (peptides occurring in only one protein), except for proteins with only two non-conflicting peptides.

Normalization and imputation of the proteomics data was done using the DEP package for R (Zhang et al., 2018). Normalization was done using the vsn method (Huber et al., 2002), which transforms data tackling heteroskedasticity. Imputation of missing values was performed using the bpca method (Oba et al., 2003), which leans on principal component analysis to impute missing values.

Data was analyzed according to a design where differences between strains (virR^mut vs -WT, as well as virR^mut vs virR^mut-Comp) were estimated independently within each cell compartment, (whole cell lysates or EVs). according to a nested design: Expression ∼ Compartment + Strain:Compartment.

Differential expression analysis of the proteomics data was performed using limma (Ritchie et al., 2015) and p-values were corrected for multiple testing according to the Storey-Tibshirani method using the R package qvalue (Storey et al., 2003). Proteins with an FDR<0.05 were selected as differentially expressed.

Enrichment analyses of the sets of differentially expressed genes and proteins were performed in R using the terms of the Gene Ontology database for testing, version 11.02.2020. Gene sets corresponding to the biological process and cellular component ontologies between levels 4 to 6 (N=2245) were selected and filtered according to their size (minimum size =10 in the RNAseq analyses; and =3 in the proteomics analyses). The selected sets were tested for enrichment using Fisher’s exact test among genes, or proteins consistently up, or down regulated in virR^mut with respect to both virR-expressing strains in each case. Multiple testing correction was performed using the qvalue R package (Storey et al., 2023). Terms with an FDR < 0.1 and an odds ratio > 2 were selected for visualization in figures 3E, and 6F, where node sizes was proportional to the significance of the enrichments. In these visualizations, two terms were connected depending on the amount of sharing between the groups of genes contributing to the enrichments in each of them, drawing a link whenever the intersection between the groups of genes was larger than 50% of the smallest of the two gene sets involved. Ontology term clusters, represented by colors, were then assigned using the Louvain’s algorithm for modularity optimization (Blondel et al., 2008), as implemented in the R package igraph (Csardi and Nepusz, 2006). Visualization of the resulting networks was done using the open-source software Gephi (Bastian et al., 2009).

LC-MS analysis was carried out on a Schimazdu LC/MS-8030 coupled with a triple quadrupole mass analyser (QqQ) provided with an electrospray source in a negative ionization mode. Extracted metabolites were quantified in Multiple Reaction Monitoring (MRM) mode. A stock standard solution prepared in 17% (v/v) methanol in water and containing all metabolites: acetyl-CoA, methyl citrate, methylmalonyl-CoA, pyruvate, was used as external standard. Metabolites were identified by ion pairing liquid chromatography. Acetyl-CoA, methyl citrate and methylmalonyl-CoA were separated by reverse phase with a poroshell 120 Phenyl-Hexyl analytical column (2.1 x 50mm, 2.7 µm, Agilent) and a binary gradient (Table S2) consisting of a mobile phase A with water-methanol 97:3 (v/v), 10 mM tributylamine and 3 mM acetic acid, and mobile phase B composed of pure methanol. Pyruvate was separated by normal phase using a Kinetex HILIC analytical column (2.1 x 150 mm, 2.6 µm, Phenomenex) and a binary gradient (Table S3) consisting of pure water as mobile phase A and pure methanol as mobile phase B.

The flow rate was maintained at 0.4 ml/min and the injection volume was 20 µl. Nitrogen was used as nebulizing and drying gas at a flow rate of 1.5 and 15 ml/min, respectively. The desolvation line (DL) temperature was set at 250^°C and the ionization voltage was fixed at 4.5 kV. Argon was used as collision gas to perform collision-induced dissociation at a pressure of 230 kPa. The dwell time was set at 100 ms and the detection was performed in MRM mode (detector voltage of 1.8 Kv), monitoring three transitions for each compound. The transition with higher intensity was selected as quantitative purposes (Table S4), while the other two were used for confirmation of the identity.

The effect of virR on relative abundance of different cell wall lipids was characterized as had been done previously for MEVs (Prados-Rosales et al., 2014b). Briefly, cell wall lipids were extracted from triplicate 10 ml cultures of wild type, virR^mut, and virR^mut complemented (virR^mut-C) strains overnight into 3 ml 2:1 chloroform:methanol at room temperature, extracts were dried under nitrogen before being dissolved in 0.75 ml LC-MS grade 2:1:1 isopropanol:acetonitrile:water. Extracts were separated on a Waters ACQUITY BEH C18 column UPLC heated to 55°C and eluted at a flow rate of 0.4 ml/minute with 60:40 acetonitrile:water as mobile phase A and 90:10 isopropanol:acetonitrile as mobile phase B, both containing 10 mM ammonium formate and 0.1% formic acid. Initial condition was 40%, B which increased to 43% at 2 minutes, 50% at 2.1 minutes, 54% at 12 minutes, 70% at 12.1 minutes, 90% at 18 minutes before returning to initial condition at 18.1 minutes to complete the run at 20 minutes. The coupled Waters Synapt G2 q-TOF MS was operated in positive ion resolution mode with the following conditions: capillary voltage 2kV, cone voltage 30V, desolvation gas temperature 550^°C, gas flow 900L/hour, source temperature 120^°C. Mass spectra were acquired in centroid mode from 100-3000 m/z with scan times of 0.5 second. Leucine enkephaline was used as reference. Relative abundances of Mass-retention time pairs were normalized to total ion current and lipid species were identified using Mycomass (Layre et al., 2011) and Mtb LipidDB databases (Sartain et al., 2011).

Lipidomic analysis of both whole cells and EVs by TLC was performed as previously described (Boldrin et al., 2021). Mtb strains were grown at 37°C in standing Middlebrook 7H9 and when they reached an OD595 of 0.2, cells were washed and transferred to MM. Cells were grown until they reached an OD595 of 0.5. Further, cells were harvested by centrifugation at 3,500 rpm for 5 min, and pellets were heat-inactivated at 100^°C for 1 h and used for lipid extraction. Five milliliters 2:1 (vol/vol) chloroform-methanol was added to the cell pellet and incubated at room temperature (RT) for 12 h with constant stirring. The organic extract was separated by centrifugation (1,000 × g, 5 min, RT) and decantation. The obtained pellet was extracted with 5 ml 2:1 (vol/vol) chloroform-methanol for 1 h at RT with constant stirring and separated under the same conditions. The combined two organic extracts (total lipids) were dried under a stream of nitrogen gas at RT and saved for lipid analysis.

Polar lipids were separated by 2D-TLC by spotting 200 μg of total lipids on 60 F254 silica gel plates (Merck) using chloroform, methanol, water (50:30:6) as a mobile phase 1 (first dimension), and chloroform, acetic acid, methanol, and water (40:25:3:6) as mobile phase 2 (second dimension). Analysis of PDIMs and TAGs was performed in a 1D-TLC format. Two hundred μg of total lipids were spotted on the silica plates and separated using 9:1 petroleum ether-diethyl ether as mobile phase. Free mycolates were separated by 1D-TLC using chloroform/methanol/ammonium hydroxide (80:20:2) as mobile phase. Mycolic acids were extracted and separated from cells as previously described (Vilchèze and Jacobs, 2007). Sulfolipids (SLs) and diacyltrehaloses (DATs) were separated chloroform/methanol/water (100:14:0.8) (first dimension) and chloro-form/acetone/methanol/water (50:60:2.5:3) (second dimension). EV-associated lipids were separated and visualized following the same procedures as for WCLs. Development of TLC plates was performed by repeating the process of pulverizing the staining solution and heating the plate at 100°C 4 times. The staining solution contained 10.5 ml 15% ethanolic solution of 1-naphthol, 6.5 ml 97% sulfuric acid, 40.5 ml ethanol, and 4 ml water.

To obtain mycobacterial cell walls, whole cell pellets were resuspended in PBS and boiled for 15 min. Boiled cells were transferred to 2 ml Lysing matrix tubes with 0.1 mm silica beads (M.P. Biomedical, Santa Ana, CA, United States). And mechanically disrupted in a FastPrep-24TM (M.P. Biomedical) giving 12 cycles at predetermined speed of 6 during 30 sec. Samples were allowed to chill on ice for 3 min between cycles. Cell breakage was monitored by optical microscopy (using Methylene blue as stain), if 95% of the cells had not broken perform additional FastPrep-24TM cycles were performed. Lysed cells were centrifuged at 170 × g for 30 sec to separate them from beads and the supernatant were transferred to a clean 1.5 ml tube. The suspensions were further digested with 10 μg of DNase and RNase/ml for 1 h at 4^°C to obtain a cell wall-enriched fraction after centrifugation at 27,000 × g for 10 min. Cell walls were resuspended in PBS containing 2% sodium dodecyl sulfate (SDS) and the suspension was incubated for 1 h at 90 ^°C with constant stirring and recentrifuged at 27,000 × g for 30 min, and the supernatant was discarded. This process was repeated twice. The resulting pellet was washed trice in distilled water, 80% acetone and acetone and lyophilized. Pellets were resuspended in 0.9 ml Tris-HCL (50 mM, pH 7) buffer, plus 0.1 ml of protease K stock solution. The mixture was incubated at 60 ^°C for 90 min, to digest any remaining surface proteins attached to the CW and centrifuged again for 3 min at 20,000 × g and resuspend pellet in distilled water. The mixture was then heated at 90^°C for 1 h before centrifugation at 27,000 × g for 30 min. The supernatant was discarded and washed twice with PBS and four times with deionized water to remove SDS. Pure PG was obtained from isolated cell walls. Briefly, pure cell walls were resuspended in 0.5% KOH in methanol and stirred at 37^°C for 4 days to hydrolyze mycolic acids (MA). The mixture was centrifuged (27,000 × g 20 min) and the pellet was washed twice with methanol. MA were separated from cell walls using diethyl ether and recentrifuged at 27,000 × gfor 20 min. The supernatant contained MAMEs. The extraction process with diethyl ether was repeated twice and let dry. The resulting arabinogalactan (AG)-PG was digested with 0.2 N H₂SO₄ at 85^°C for 30 min and neutralized with BaCO₃ to remove the arabinogalactan. Treated AG-PG was centrifuged at 27,000 × g for 20 min. Supernatant contained soluble AG. The resulting insoluble PG was washed four times by centrifugation in deionized water and stored at 4^°C or room temperature.

For DAP quantification (Tsuruoka et al., 1985), samples were hydrolyzed overnight in 6N HCl at 100°C in 1 ml capacity-crystal ampoules (Wheaton). Samples were dried and resuspended in water and mixed in a 1:1:1 ratio with pure acetic acid and ninhydrin reagent (250 mg ninhydrin, 4 ml phosphoric acid 0.6 M, and 6 ml pure acetic acid). Samples were measured at OD 434nm, and the amounts of DAP were calculated based on a standard curve with pure DAP. The data displays the average ± standard deviations n=3.

500 μg of purified cell walls were mixed with 10 μg of inositol and hydrolyzed with 0.1 M trifluoroacetic acid (TFA) for 2 h at 121°C. The monosaccharides released were then converted into alditol acetates by reduction with NaBH₄ and acetylation with acetic anhydride in pyridine. The products were analyzed by gas chromatography-mass spectrometry in a 7890A/5975C system from Agilent, using a DB-5ht (30 m x 0.25 mm x 0.1 μm) fused silica capillary column. The compounds were identified by comparing the retention times of the peaks with those recorded for standards analyzed under identical conditions. Quantification was carried out considering the area of the peaks and the response factors of each compound with respect to inositol.

The dry pellets of purified peptidoglycan from the three strains were dissolved in pure water and further broken by tip sonication at 2 mA for 30 sec, three cycles. Then 10 µl of the PG suspension was added to standard AFM substrate made of mica coated with 0.01 % Poly-L-lysine by incubating the sample during 90 min to achieve good coverage. The sample was further rinsed with water and dried with Nitrogen flow. The AFM in ambient conditions (i.e. air) was performed using Tapping mode in a Bruker FastScan Bio machine (Santa Barbara, CA, United States) with Nunano SCOUT 350 - Silicon AFM probes (spring constant: 42 N/m, Resonance frequency: 350 kHz). The images were taken using a free amplitude of 10 nm with set point of 70-80 % of free amplitude (e.g. 7 nm). For AFM high resolution imaging in liquid the images were acquired in Peak force Tapping mode in imaging buffer made of 10 mM Tris and 150 mM KCl, pH: 8, with the FastScan Bio machine using Bruker Fastscan-D AFM probes (spring constant: 0.1 N/m, Resonance frequency: 70 kHz) at the force range of 1-3 nN peak force set point. The imaging parameters used are as follows; Scan rate: 1 Hz; Scan angle: 0 °; Peak force amplitude: 80-100 nm and with high pixel resolution (512-880).

The thickness measurements were performed with standard Gwyddion 1D height distribution tool (selecting an area containing background and a single leaflet of peptidoglycan). The number of cells analysed varied from n = 23 (for Mtb WT), n = 36 (for virR^mut) and n = 25 (for virR^mut-C). The comparisons extracted from this data is statistically relevant with t test with Welch’s correction performed. The analysis of the high-resolution images obtained in liquid was performed using the semi-automated custom-made routine from (Pasquina-Lemonche, et al. “Pore AFM”, 2024) where the pore area, later converted in diameter, and the pore number were measured from the 2D slice of each image containing the maximum number of pores (see Fig. S8). The number of images similar to images shown in Fig. 5 were n = 8 for all strains, analyzing both pore size and pore number, the statistical comparisons were also performed using a t test with Welch’s correction.

Oligomerization of VirR and Lcp1 was analyzed by Blue native PAGE (Invitrogen, Waltham, MA, United States). Recombinant VirR and Lcp1 proteins were prepared in 1xLB Native PAGE including DDM at 0.5% final concentration. After incubation at 4°C for 30 min, samples were centrifuged at 20,000 × g for 30 min at 4^°C. The supernatant was recovered and mixed with G-250 additive following manufacturer instructions (Invitrogen) and loaded onto a Novex® Bist-Tris 4-16% gradient gel and run following manufacturer instructions. Gel was first fixed and then stained with Coomassie.

Individual or combined VirR and Lcp1 were prepared at 2 μM in buffer A in a final volume of 50 µl and incubated for 30 min at 37°C. This volume was mixed with 100 µl of 60% optiprep (Sigma-Aldrich) to a final concentration of 40% and overlaid with 100 µl of subsequent optiprep solutions at 35%, 30%, 20% and 10% (optiprep solutions were prepared in Buffer A). The samples were centrifugated at 70000 rpm for 2 h at 4^°C in an Optima TLX Ultracentrifuge (Beckman Coulter, Indianapolis, IN United States). Eleven fractions of 50 µl were recovered and examined by dot blot (performed as previously described). Polyclonal antibodies raised against VirR and Lcp1 were used as primary antibodies. Quantification of dotblots was performed in ImageJ as previously described (Schirmer et al., 2022). Briefly, integrated densities were calculated after background subtraction, and referred to wild type values to obtain the relative integrated intensities.

Mtb WT and virR^mut strains were grown in 7H9 media supplemented with ADC and Tween 80 to an OD 540 of 0.2 and then diluted to an OD540 of 0.1 with 7H9 containing 0 or 50 µgml⁻¹ lysozyme (Sigma). The cultures were incubated with agitation at 37^°C and OD540 was monitored daily.

The Mtb strain bearing the plasmid VirR-pMV261-Venus was grown in MM to an OD595nm of 0.4. At this stage, cells were harvested, washed three times in 1 × PBS (pH 7.4), resuspended in 1/3 – 1/6 volume of 4% paraformaldehyde (PFA) and incubated for 2 h at 37 ^°C. Finally, the cells were washed twice in 1 × PBS and resuspended in 100–200 μl of PBS. Microscope slides were covered with 5 μl poly-L-lysine (Sigma) and excess poly-L-lysine was removed away with distillated water. A 20 – 40 μl aliquot was applied to pre-treated slides, allowed to air dry and covered with antifade solution (Prolong Gold, Invitrogen) and visualized under the fluorescence microscope using a 100x oil immersion objective on an Orca 12-ERG digital CCD camera (Hamamatsu Photonics) mounted on a Nikon E600 microscope.

M. smegmatis mc²155 transformed with the pGMCS derivatives were inoculated in complete 7H9 medium. After overnight incubation at 37°C, bacteria were collected and analyzed by fluorescence activated cell sorting using a BD FACS LSRFortessa X20 flow cytometer. Flow cytometry data analysis was performed using FlowJo software (Version 10; Becton, Dickinson and Company, Ashland, OR, United States). The gating strategy is displayed in supplementary figure 10 (Fig. S10).

The statistical significance of the difference between experimental groups was determined by the two-tailed Student’s test using PRISM 5.0. P values ≤ than 0.05 were considered significant. Statistical analysis of RNAseq is detailed in the corresponding section of Materials and Methods.