Introduction

Cell division is a critical event in the life of a bacterium as it is the prime mode of bacterial growth and reproduction. Hence, the process of their division is tightly knitted. The key process in bacterial cell division, i.e., assembly of the FtsZ ring at the mid-cell (1), is tightly regulated by the Min and Nucleoid occlusion systems (2–6). Mycobacterium, which comprises clinically relevant pathogens-Mycobacterium tuberculosis (Mtb), Mycobacterium leprae, and Mycobacterium abscessus lacks both the regulators of FtsZ ring placement (7). Instead, they encode for a DivIVA protein, usually encoded by Firmicutes and other Actinobacteria (8–11). It is necessary that the bacterial shape, crucial for normal cellular physiology, is maintained during cell division. Rod-shaped bacteria such as Escherichia coli and Bacillus subtilis encode MreB and multiple MreB-like proteins that sustain the rod morphology of cells (12). Mycobacterial spp. lack MreB homologs (13, 14). The DivIVA protein of Mtb called Wag31 is known to regulate both cellular growth and morphology (10, 15), which makes it an exciting candidate to investigate.

DivIVA proteins were discovered three decades ago during the study of cell division mutants of B. subtilis (16). DivIVA proteins are rich in coiled-coil domains (17–21), which enable them to form higher oligomeric structures and, hence, scaffolds, making them suitable adaptor proteins (22, 23). They are known to play diverse roles in various species (24), e.g., they play dual roles of septum positioning and chromosome partitioning in B. subtilis (9, 25); cell-separation/resolution post-cell-division in Listeria monocytogenes (26, 27); pole formation and maturation in Streptococcus spp. (Fadda, 2007) In Mycobacterium spp., Wag31 is an essential protein that senses negative membrane curvature found at poles and septum and localizes there (28). It plays roles in cellular elongation and septation in concert with other divisome and elongasome components (10, 15, 29). It is specifically found in higher amounts at the old pole, where it directs polar elongation by recruiting the acyl CoA carboxylase (ACCase) complex involved in fatty acid, mycolic acid, and methyl-branched chain lipid precursor synthesis (28, 30–32). It localizes to the septum after cytokinesis is over, perhaps for septal synthesis via its interactions with FtsI (32–34), though the process remains largely unknown. Consequently, loss of Wag31 leads to cell division defects-loss of polar elongation, which renders the cells ’round’ in shape that occurs due to affected peptidoglycan (PG) synthesis (10, 15, 28). Though Wag31 participates in coordinating PG metabolism, the mechanism or the protein network by which it does so remains unknown.

Wag31 has also been associated with the maintenance of the intracellular membrane domain (IMD), a specialized compartment for lipid synthesis, localized predominantly at the old pole in consonance with pole elongation and spottily all around the plasma membrane where it may function to repair membrane defects (35–37). The loss of Wag31 leads to the delocalization of IMD and its protein markers, such as MurG and GlfT2, but it is not understood whether Wag31 plays a direct role in membrane-partitioning or it is just a consequence of localized membrane synthesis (34, 37). The role of Wag31in PG and lipid synthesis that drives cell division remains unclear. Despite numerous studies, the details of how Wag31 orchestrates mega-complexes at the pole or partitions the membrane into two to govern cellular elongation remain elusive.

Here, we endeavoured to address the lacunae in the functionality of Wag31 by asking the following questions-i) what are the ultrastructural changes that occur in mycobacteria due to the absence of Wag31, ii) do Wag31 deficient cells undergo transcriptional rewiring to adjust to cell division defects, iii) is the change in cellular morphology from ‘rod-to-round’ accompanied by altered lipid levels, iv) what is the interactome of Wag31, v) what are the functions of its N and C-terminal domains and finally vi) what is its relationship with the mycobacterial membrane. The present study shows that Wag31 levels are critical for maintaining lipid homeostasis in mycobacteria. Our results demonstrate that Wag31 binds and tethers the membrane together to orchestrate membrane-localized processes essential for maintaining normal cellular morphology and physiology.

Results

Loss of Wag31 leads to the formation of intracellular lipid inclusions (ILIs)

In the pursuit of bridging the knowledge gap existing in the functionality of Wag31 (Fig. 1a), we generated a Wag31 conditional mutant in Msm, by integrating a copy of wag31 under tetracycline (ATc-tet off) regulation at the L5 locus and replacing the native copy with a hygromycin cassette (Fig. S1a). The mutant, Δwag31, thus generated, was confirmed by PCRs (Fig. S1b) and Western blotting (Fig. 1b). The lack of Wag31 led to a drop in the turbidity of culture (Fig. 1c). Analysis of bacillary survival in vitro by enumerating CFUs 12h post-ATc addition, revealed a 2 log fold difference in the survival of Δwag31 (Fig. 1d). Given the essential role of Wag31 in causing polar growth of the cell, we investigated the effect of depletion of Wag31 on bacterial cellular morphology using electron microscopy (EM). Scanning EM (SEM) suggested that the absence of Wag31 caused a transition in the morphology of the cells from rod to round, as previously documented (10, 15, 38). Apart from round morphology (∼36%), we also observed a rod-to-round transformation stage in which cells were bulged at one pole with the other pole intact (Fig. 1e-g). To obtain detailed insights, we performed transmission EM (TEM), which divulges information on the internal structure of cells. We discovered spherical, electron-lucent bodies (indicated by an arrowhead in Fig.1h) in the cytosol of all the samples. While Msm and Δwag31-ATc harbored a few (∼2-3) electron-lucent bodies, their numbers were drastically higher in the case of Δwag31+ATc (Fig. 1h). The cytosol of Δwag31+ATc was filled with multiple variable-sized bodies which resembled Lipid bodies that Mycobacterium tuberculosis (Mtb) accumulates inside the host (39, 40). As the term ’lipid bodies/lipid droplets’ is used specifically for the lipids acquired by Mtb from the host cell, we referred to them as intracellular lipid inclusions (ILIs). We categorized the cells based on the number of ILIs they contained. Their distribution remained reasonably similar within Msm and Δwag31-ATc cells for all the classes (Fig. 1i). Whereas Δwag31 treated with ATc had as high as 85% cells harboring more than 5 ILIs and very few cells with less than 5 ILIs (Fig. 1i). Our data, therefore, reports a novel observation of lipid accumulation when Wag31 is depleted from mycobacteria.

Fig. 1.

Wag31 levels influence lipid homeostasis

Next, we set out to analyze if the ILIs are indeed lipidic in nature by staining cells with BODIPY (493/503), a dye that preferentially partitions into neutral lipid droplets (41, 42). We stained either untreated or ATc-treated Msm and Δwag31 with BODIPY and visualized them on a microscope. ATc-treated Δwag31 cells displayed the highest level of BODIPY fluorescence, indicating the maximum abundance of lipid bodies among the three strains (Fig. 2a, c). Further, we examined whether the appearance of ILIs is associated just with the loss of Wag31 or is a consequence of deviations from the native expression level of Wag31. Towards this, we overexpressed Wag31 episomally using an isovaleronitrile (IVN) inducible system (Msm::wag31) (Fig. S2a) and stained the cells with BODIPY. Unlike Wag31 depletion, which resulted in spherical cells (Fig. 1e), Wag31 overexpression resulted in the bulging of cells from one pole (Fig. 2b). Additionally, the overexpression of Wag31 also led to lipid aggregates in the cell with a few larger ‘donut-like’ aggregates in the bulged pole (Fig 2b,d).

Fig. 2.

Lipid accumulation in both cases led us to inspect whether lipid metabolism is closely tied to Wag31 expression levels in mycobacteria. To investigate that possibility, total cellular lipids were extracted from Msm, Wag31 depletion, and overexpression conditions and subjected to semi-quantitative LC-MS analysis (illustrated in Fig. 2e). Lipidomic analysis of Δwag31+ATc revealed an imbalance in the levels of neutral lipids, phospholipids, and free fatty acids in comparison to Msm-ATc (Fig. 2f-g). Δwag31+ATc cells had a significant accumulation of unsaturated diacylglycerols (∼3-5 fold) and a very long chain containing (>C36) phosphatidylethanolamines (∼5-32 fold). This was accompanied by increased levels of different chain lengths (C18-C24) of free fatty acids (∼1.7-4 fold) and reduced levels of cardiolipins C70:0 & C70:1 (∼3 fold). Wag31 overexpression, on the other hand, led to a 2-fold increase in the level of different species of alpha-mycolic acids (C74-C78) as compared to Msm+IVN (Fig. 2h), while other tested lipids remained unchanged (Fig. S2). Altogether, the results presented above demonstrate that Wag31 levels are crucial for maintaining lipid homeostasis in mycobacteria, the depletion and overexpression of which impact lipid metabolism differently.

Molecular interactions of Wag31 with membrane-associated proteins govern cellular homeostasis

We reasoned that Wag31, an adaptor protein, might maintain lipid homeostasis by interacting with proteins involved in these pathways. To examine the protein interaction network of Wag31, we performed an interactome analysis by utilizing Δwag31, which expresses FLAG-Wag31 integrated at the L5 locus, and used FLAG-GFP as the control to differentiate from non-specific interactions. FLAG-Wag31 and FLAG-GFP from biological triplicates were immunoprecipitated using whole cell lysates from the log phase cells and subjected to mass spectrometry analysis (Fig. S3a & Table S2). This study identified both direct and indirect protein-partners of Wag31. The proteins identified were categorized based on their functions. It was found that Wag31 interacts with proteins involved in cell-wall elongation and division, lipid metabolism, intermediary metabolism and respiration and information pathways (translation, replication) (Fig. 3a, Fig. S4). Interactions of Wag31 with AccA3 and AccD5, subunits of the ACCase complex, and Rne or RNaseE (Msm4626), a membrane-associated protein, have been reported previously (28, 43). Subunits of ACCase complex and Rne were among the top 5 hits in our MS/MS data, validating our approach (Fig. 3a, marked with asterisk). The classification of the top 20 hits is represented in Fig. 3a. The highest number of protein interactors, expectedly, fell in the cell wall and cell processes category (Fig. 3a). For the present study, we focussed on the interactors found in the cell wall elongation and cell division and lipid metabolism functional categories to probe the cause behind the lipid dysbiosis phenotype.

Fig. 3.

MurG, involved in PG metabolism (36, 44–46), and SepIVA, involved in regulating septation and the spatial regulator of MurG (47) also surfaced in the interactome of Wag31. Msm1813, or acyl CoA carboxylase (ACAD), crucial for nascent fatty acid precursor synthesis, was also enriched in the interactome. Other lipid metabolism interactors of Wag31 included acyl-CoA dehydrogenase (Msm2080), involved in lipid catabolism; GpsA, involved in phospholipid biosynthesis; and Msm0372, involved in fatty acid biosynthesis. To validate the high-throughput data, we chose MurG, SepIVA and Msm2092 and performed pull-down assays with His-tagged Wag31. Briefly, His-tagged Wag31 (Fig. 3b) was incubated with Msm lysates expressing 3X-FLAG tagged versions of the tested proteins (Fig. 3c) and pulled-down with the help of cobalt beads followed by a western blot analysis. Results obtained confirmed the interaction of Wag31 with MurG, SepIVA and Msm2092, thus validating our data (Fig. 3d).

Since Wag31 is a spatiotemporal regulator of other proteins, we were intrigued to discover the localisation of its interacting partners. We used the data curated by Zhu et al. (48) and determined that Wag31 majorly interacts with membrane-associated proteins with a few exceptions (Fig. 3e, Fig. S3b). The mycobacterial plasma membrane has been recently shown to have a cell wall-associated compartment called PM-CW and a cell wall-free compartment called PM_f or IMD, which is the site for lipid synthesis. Based on the available data (36, 48, 49), we found that Wag31 interacts more with IMD-proteins than with PM-CW proteins (Fig. 3f.) In accordance with assigned roles (36), PM-CW residing proteins in the interactome are involved in cell wall and cell processes and intermediary metabolism and respiration. Similarly, IMD-homing proteins clustered in the lipid metabolism and cell wall synthesis category. These results suggest that Wag31 maintains lipid homeostasis and cellular morphology primarily through its interactions with the membrane proteins participating in such pathways.

Wag31 binds and tethers cardiolipin-containing membranes

Wag31 is a cytoplasmic protein that recognizes negative curvature regions in the cell, i.e., poles and septum (28). It also forms higher-order structural assemblies at the poles (21, 50). However, the precise mechanism by which Wag31 assemblies localize to the poles remains elusive. The predominant interactions of Wag31 with membrane-associated proteins (Fig 3) were intriguing and led us to investigate if it can directly bind to membranes. We reasoned that such membrane binding activity might render it capable of acting as a scaffold that anchors and organizes other proteins, thereby modulating their functions. To test this, we performed a lipid crosslinking-based PLiMAP assay (Fig. 4a). Purified Wag31 was incubated with ∼100 nm extruded liposomes containing each of the abundant classes of bacterial lipids phosphatidylethanolamine (PE), phosphatidylglycerol (PG), or cardiolipin (CL). Liposomes contained trace amounts of a reactive fluorescent lipid (RFL), which crosslinks with membrane-bound proteins upon photoactivation. These assays revealed that Wag31 preferentially binds CL-but not PG-or PE-containing liposomes (Fig. 4b). Both CL and PG are anionic lipids, and the preferential binding of Wag31 to CL indicates that the binding cannot be attributed to a charge-based interaction. Remarkably, the sensitive fluorescence-based detection revealed the presence of SDS-resistant higher-order oligomers of membrane-bound Wag31 (Fig. 4b).

Fig. 4.

Wag31 is a homolog of the DivIVA protein from B. subtilis. The latter has been shown to bind to the membrane via the N-terminal DivIVA domain (N-DivIVA_BS) and localize other cell division proteins through its interactions with the C-terminal region (51, 52). However, the functional significance of the N-& C-terminal domains of mycobacterial Wag31 has not been dissected. Considering that the Wag31 N-terminal region shares 42% sequence identity with N-DivIVA_BS (50), we sought to inspect the effect of disruption of a particular residue (K20) in the N-terminus and the deletion of the entire N-terminus, encompassing the DivIVA domain (1-60 amino acids) on membrane binding (Fig. S5a). The K20 residue lies within a positively charged patch, which likely constitutes a membrane-associating surface in the intertwined loop region of Wag31(50). Moreover, mutating the K20 residue has been shown to decrease the cell length and elongation rate by 18% and 22%, respectively (34). The K20A mutation showed a marginal defect in binding to CL-liposomes (Fig. 4c-d), indicating that K20 alone is insufficient for membrane binding. However, Wag31_K20A showed a marked defect in forming SDS-resistant higher-order oligomers, as seen in the binding assays. Interestingly, deletion of the entire N-terminal domain caused a significant defect in both membrane binding and forming SDS-resistant higher-order oligomers (Fig. 4c-d). Surprisingly, binding was not entirely abolished, suggesting that the C-terminus of Wag31 may also contribute to membrane binding. Taken together, our results indicate that Wag31 preferentially binds CL-containing membranes and that regions other than the N-terminal domain contribute to membrane binding. The latter feature distinguishes Wag31 from DivIVA_BS, and Wag31 likely contains membrane binding sites located at both the N-and C-terminal regions, which might lead to differences in the manner in which it interacts with the membrane and partner proteins compared to DivIVA_BS.

To further confirm membrane binding, we imaged 100 nm-extruded CL-containing liposomes with a membrane dye DiI in the presence of Wag31. We used a GFP-tagged Wag31 mixed with an untagged Wag31 at a 1:10 molar ratio for these experiments. In the absence of Wag31, liposomes appeared as faint but distinct fluorescent puncta (Fig. 4e). Surprisingly, in reactions containing Wag31, we observed liposomes clustered into bright fluorescent puncta that were positive for GFP-Wag31 (Fig. 4e). The Wag31 tetramer has been modelled as a long rod-shaped protein filament (21, 50) and such clustering of liposomes falls in congruence with the model. It indicates that these long filaments of Wag31 act to tether liposomes. We then systematically evaluated this effect by testing liposomes of different compositions and Wag31 mutants. Again, in the absence of Wag31, both PE-and CL-containing liposomes appeared as faint but distinct fluorescent puncta (Fig. 4f). The addition of Wag31 caused clustering of CL-containing liposomes into bright foci (Fig. 4f), which is quantified by a particle intensity analysis (Fig. 4g). This clustering was caused by membrane binding of Wag31 because PE-containing liposomes that do not bind Wag31 (Fig. 4b) showed no clustering (Fig. 4f,g). As expected, Wag31_Δ1-60 was defective in membrane binding (Fig. 4c,d) and showed a severe tethering capacity defect (Fig. 4f,g). Surprisingly, Wag31_K20A, which showed a marginal defect in membrane binding (Fig. 4c,d), showed a marked defect in tethering capacity (Fig. 4f,g). Together, these results suggest that membrane-binding and membrane-tethering are non-overlapping attributes in the structure of Wag31, and the N-terminal DivIVA-domain is indispensable for membrane-tethering activity.

Tethering is crucial for the survival of mycobacteria

Results presented in Fig.4c indicate that Wag31 binds strongly to CL-containing membranes. CL is a negatively charged inverted cone-shaped lipid that has been shown to be concentrated at the sites of negative curvature in the bacterial cell, i.e., pole and septum (53, 54). Wag31 has been demonstrated previously to localize at sites of negative curvature (28). Their similar localization pattern, coupled with Wag31’s affinity to bind to CL, led us to examine the effect of this binding in vivo. We hypothesized that modulations in Wag31 levels might alter CL distribution in the cell. To examine this, we utilized a fluorophore 10-N-Nonyl-acridine orange (NAO) that binds to anionic phospholipids (54, 55). We stained Msm and Wag31 depleted and overexpressing cells with NAO and imaged them using a fluorescence microscope. While the localization pattern of NAO was polar in Msm+ATc and Δwag31-ATc, there was a redistribution of red fluorescence to the cytosol and along the perimeter of the cell in Δwag31+ATc (Fig.5a-b). We quantified the distribution of fluorescence using a line profile which yielded peaks of intensity at the poles and a basal level of fluorescence along the lateral cell body for both Msm+ATc and Δwag31-ATc (Fig.5b). Wag31 overexpression also led to a change in the distribution of NAO from the poles and septum to ‘donut-like’ fluorescent foci in the cytosol (Fig.5c-d). Since our in vitro results indicate that Wag31 prefers to bind CL over PG (Fig. 4b), the change in the distribution of NAO fluorescence implies that the loss or overexpression of Wag31 causes the mislocalization of CL in mycobacterium.

Subsequently, we investigated the impact of tethering-deficient Wag31 mutants on cell survival. Wag31 has an N-terminal domain joined to the C-terminal via a linker region. The N-terminal domain contains a positively charged patch in the intertwined region (the region between two helices in the N-terminus) and membrane binding has been attributed to this region (50). This patch is lined with conserved arginine and lysine residues, K15, K20, and R21 (50) (Fig. 5e). As our in vitro results suggest the importance of K20 and the DivIVA-domain in tethering, we performed complementation assays with the tethering-defective mutants to examine cell survival and morphology. Δwag31 was complemented with an integrative copy of either Wag31 or Wag31_K20A or Wag31_Δ1-60. Δwag31+ATc showed a two log-fold difference in survival and a 40 % decrease in the percentage of rod-shaped cells compared with Δwag31-ATc (Fig. 5f-g). Complementation with Wag31 and Wag31_K20A rescued the loss of cell viability and restored the rod shape of the cells (Fig. 5f-g), whereas Wag31_Δ1-60 behaved similar to Δwag31+ATc. It neither restored the rod shape of the cells nor reversed the survival defect (Fig. 5f-g). These results highlight the importance of tethering for sustaining mycobacterial morphology and survival.

Fig. 5.

Tethering is correlated to lipid levels and CL localization in mycobacteria

As described above, Wag31-mediated membrane tethering is an absolute for the survival of mycobacteria. Perturbations in lipid homeostasis and mislocalization of CL due to altered Wag31 levels hint at a correlation between its ability to function as a homotypic tether and maintain lipid composition and localization. To derive the mechanism behind it, we performed fluorescence imaging experiments with native and tethering defective mutants of Wag31. To investigate a role of tethering in lipid homeostasis in mycobacteria, we stained wildtype and complementation strains with BODIPY and quantified the fluorescence intensities. We observed comparable staining intensity in Msm +ATc and, Δwag31 -ATc (Fig. 6a-b). Δwag31+ATc expectedly had the highest BODIPY staining intensity, indicating higher than normal lipid levels (Fig. 6a-b). Complementation with Wag31 fully restored the lipid levels to normal, whereas complementation with Wag31_K20A couldn’t restore the lipid levels completely (Fig. 6a-b). Wag31_Δ1-60 exhibited higher lipid levels than Δwag31-ATc (Fig. 6a-b). These results highlight that the maintenance of lipid homeostasis by Wag31 is a consequence of its tethering activity.

Fig. 6.

Next, we examined the dependence of CL localisation on tethering by staining Msm +ATc, Δwag31 -ATc, Δwag31+ATc, Δwag31::wag31+ATc, Δwag31::wag31_K20A+ATc, and Δwag31:: wag31_Δ1-60 +ATc with NAO. NAO staining showed a polar and septal (in dividing cells) localization pattern in the case of Msm and Δwag31-ATc, whereas it stained the cytosol and perimeter of the Wag31 depleted cells (Fig. 6c-d). Complementation with Wag31 restored NAO localization completely, whereas Wag31_K20A restored it partially, which could be attributed to its limited tethering activity (Fig. 6c-d). Importantly, we observed a mis-localized pattern of NAO with Wag31_Δ1-60 similar to Wag31 depletion (Fig. 6c-d). These results inform of the dependence of CL localization on membrane tethering activity of Wag31. Taken together, the data suggests that Wag31 regulates lipid homeostasis and localisation by the virtue of its tethering activity.

The N and C-terminal domains of Wag31 have distinct functions

From the results presented above, we established a role for the N-terminal region of Wag31 that comprises DivIVA-domain in membrane-tethering (Fig. 4). We also identified the protein-interaction network of Wag31 with the help of MS/MS analysis (Fig. 3 & Fig. S4). This implored us to investigate the functionality of the C-terminal region of Wag31. We hypothesized that it might be responsible for mediating interactions of Wag31 with other proteins. We performed a pull-down assay utilizing E. coli lysates expressing either Wag31 or Wag31_Δ1-60 (Fig. 7a). We chose novel interactors of Wag31 viz. MurG, SepIVA, and Msm2092 from our interactome database (Fig.3a) and a known interactor of Wag31, AccA3. The Msm lysates expressing 3x-FLAG tagged versions of the above-mentioned proteins (Fig.7b) were incubated with E. coli lysate or E. coli lysates expressing either Wag31 or Wag31_Δ1-60, followed by pulldown with Cobalt beads. The western blot analysis revealed that all the tested interactors bound to Wag31 (Fig. 7c). Interestingly, interactions with all the proteins were retained in the case of Wag31_Δ1-60 as well, thus indicating that Wag31 interacts with other proteins, largely via its C-terminal. Taken together, the results suggest that the N-terminal region of Wag31 is involved in membrane-tethering through interactions with CL, and the C-terminal region is involved in modulating interactions with other membrane-associated proteins. Thus N-terminal and C-terminal of Wag31 have distinct molecular functions that together are crucial for maintaining cell shape, lipid homeostasis and survival of mycobacteria.

Fig. 7.

Discussion

Wag31 modulates mycobacterial cell division by acting as an adaptor protein that recruits the elongation machinery to the old pole for unipolar growth of cells (10, 15, 28). In the absence of Wag31, rod-shaped cells advance into a spheroplast-like stage, which is thought to arise from dysregulated PG metabolism (10, 11, 28). Recently, Wag31 has been implicated in maintaining the lipid-rich IMD compartment (37). But the molecular mechanisms by which Wag31 manages these functions has remained unclear. In this study, we discover an additional function of Wag31 as a membrane tether, which likely reconciles many of the previously ascribed functions of this protein into a unified model. To gain insights into defects associated with a change in cell shape, we imaged Wag31-deficient cells and found a dramatic accumulation of lipids in the form of ILIs. Interestingly, both Msm and ΔWag31-ATc showed the presence of ILIs, but in much less abundance, indicating that ILIs are a normal occurrence that may function to regulate cellular functions (Fig. 1g-h).

The loss of Wag31 leads to the accumulation of neutral lipids while its overexpression increase the levels of alpha-mycolic acid (a-MA), implying that the maintenance of native Wag31 levels is critical for the proper balance of lipid composition (Fig. 2f-h). A protein’s interactome is akin to its functional fingerprint. Here, we identify 116 novel interacting partners of Wag31 through mass spectrometry and validate three of these interactions, namely with MurG, SepIVA and Msm2092, through pull-down studies (Fig. 3 & 5i). Strikingly, the bulk of Wag31 interactors are membrane proteins, which might be attributed to the formation of Wag31 oligomers on the membrane. Previous studies have reported that the depletion of Wag31 delocalize MurG, but if this was caused by loss of direct interaction or was a consequence of change in cell shape was not known (34, 37), which our results clarify. Furthermore, our results identify SepIVA, which regulates MurG (47), the D-aminopeptidase Msm2092 and AmiB, both of which are PG remodellers, as Wag31 interactors. Together, these results contribute towards our understanding of how Wag31 affects PG synthesis (Fig 3a,d). AccA3 is an interactor of Wag31 (28) (34) and catalyses the first committed step in the synthesis of long-chain fatty acids by forming malonyl-CoA. In addition, Long-chain specific acyl dehydrogenase (LCAD), which we report here to be another interactor of Wag31, catalyses the first step in the β-oxidation of fatty acids, which might explain the accumulation of long chain PEs and long-chain fatty acids.

As mentioned above, the overexpression of Wag31 leads to an upregulation in a-MA levels (Fig. 2h). The localization of the a-MA transporter MmpL3 depends on polar PG levels (56, 57). MurG is an essential enzyme for PG synthesis (58) and we report here that it strongly interacts with Wag31 (Fig. 3a,d). It is tempting to hypothesize that homeostatic control on lipid composition arises from the ability of Wag31 to modulate the activity of enzymes involved in lipid metabolism and PG synthesis. That both depletion and overexpression of Wag31 affects lipid composition is likely also a manifestation of its ability to function as a homotypic tether, where the tethering ability is inherently sensitive to the tether concentration.

Our results are the first to establish a membrane tethering function of Wag31. DivIVA proteins possess an N-terminal membrane-binding domain and a C-terminal coiled-coil domain (50, 51). Our data show that disruption of the N-terminal domain (Wag31_Δ1-60) markedly reduces membrane binding and completely abolishes membrane tethering (Fig. 4f, g). The N-terminal domain binds CL while the C-terminal coiled-coil domain facilitates multimerization, which together could explain the molecular basis for Wag31 to function in homotypic tethering of CL-containing membranes. Results from testing mutants defective in membrane tethering indicate that this function is important for the proper localisation of CL and eventually the survival of mycobacteria (Fig. 5f & 6c-d). Thus, alterations in the native levels of Wag31 results in a displacement of CL from the poles (Fig. 5a-d). However, given NAO’s ability to bind to anionic phospholipids, the observed staining could also typify a redistribution of anionic phospholipids in the cytosol and membrane. In the absence of Wag31 binding to the poles (due to gene silencing or deletion of the DivIVA-domain), the multiprotein scaffold required for cell envelope biosynthesis and cell elongation is not assembled properly, leading to changes in lipid and PG metabolism and consequently to cells failing to form poles where CL would normally concentrate. The failure of a Wag31-centered multiprotein scaffold formation likely cause of this phenotype is further supported by the wag31 overexpression study, which shows similar changes in NAO staining. By affecting the stoichiometry of Wag31 relative to other interacting protein partners, Wag31 overexpression negatively impacts the formation and activity of the multiprotein scaffold.

While, complementation with Wag31 reverses the phenotypic changes observed upon Wag31 depletion, complementation with the membrane tethering mutant (Wag31_Δ1-60) fails to do so (Fig. 5f-g & 6a-d). These results not only highlight the dependence of CL localization on membrane-tethering but also underscore the importance of tethering for maintaining cellular morphology, physiology and survival. CL micro-domains at the poles (53, 54) have been shown to be an important hub for protein localisation and activity. Several studies have implicated CL in modulating the activities of membrane-resident proteins involved in electron-transport, protein translocation, bacterial two-component system etc. (59–62).

Our work delineates the roles of the N-and C-terminal domains of Wag31. The N-terminal DivIVA-domain facilitates lipid binding and membrane tethering, and the C-terminal coiled-coil domain engages in protein-protein interactions and facilitates the multimerization of Wag31. This is evident from findings that deletion of the N-terminal DivIVA-domain affects lipid binding and tethering but not its ability to bind interacting partners (Fig. 4c-d, f-g & 7c). Based on our results, we propose a model in which the N-terminal of Wag31 binds and tethers CL ensuring its proper localization while the C-terminal is free to interact and recruit proteins to the membrane to facilitate cell growth and division (Fig. 8). We reason that the loss of cell-shape and viability in the absence of tethering occurs due to mis-localization of CL which in turn has a negative impact on membrane-localized processes at a global scale.

Fig. 8.

Materials and Methods

All the illustrations were made using BioRender. List of constructs and strains used in the study are given in Table S1. Mass spectrometry samples were processed and performed by Valerian Chem.

Bacterial strains and culturing

Middlebrook 7H9 medium (BD Biosciences) or 7H11 agar (BD Biosciences) was used to cultivate mycobacterial strains. 7H9 medium was supplemented with 10% ADC (albumin, dextrose, NaCl, and catalase) and 0.2% glycerol (Sigma) and, 0.05% Tween-80 (Sigma) or 7H11 agar (BD Biosciences) with 10% OADC (oleic acid added to ADC) and 0.2% glycerol were added to the media.

Generation of wag31 conditional mutant and in vitro Growth kinetics

Recombineering method, as described in (63), was used to generate a Wag31 conditional gene replacement mutant. Detailed methodology is provided in the supplementary text. To perform in vitro growth kinetics, secondary cultures of Msm and Δwag31 were seeded at a density of A₆₀₀∼ 0.05 and either left untreated or treated with 100 ng/mL ATc for 12 h. At 0 and 12 h, appropriate serial dilutions were plated on 7H11 agar plate and incubated at 37^°C for 2-3 days, and CFUs were enumerated. CFUs were plotted using GraphPad Prism 9 and analysed by performing Two-way ANOVA (0.0021 (**), 0.0002 (***).

Western blot analysis

Msm, Msm::wag31 and Δwag31 transformants grown in the presence or absence of IVN/ ATc for 12 h were resuspended in Phosphate Buffer Saline containing 5% glycerol and 1mM protease inhibitor-phenymethylsulfonylfluoride, PMSF (Sigma) and whole cell lysates (WCL) were prepared with the help of bead-beating. Concentration of protein in each sample was estimated by Pierce BCA protein Assay kit (Thermo Fischer Scientific) and 40 mg WCLs were resolved on 10-12% SDS-PAGE gels, transferred onto nitrocellulose membrane, probed with desired antibodies and developed using ECL reagent (Millipore).

Sample preparation for SEM and TEM

Msm and Δwag31 were incubated in the absence of the presence of ATc for 12 h, and the samples were prepared for both SEM and TEM as described previously in (64). For SEM, the samples were imaged at 20,000X using FEI Nova NanoSEM 450. TEM sections were examined on a 120 kV transmission electron microscope (L120C, Talos, Thermo Scientific) at 17,500 magnification.

BODIPY labelling, imaging and analysis

1 mL cells of Msm and Δwag31 either untreated or treated with 100 ng/mL ATc for 12 h were stained with 20 μg/mL BODIPY (Invitrogen). For examining lipids in Wag31 overexpression strain, Msm and Msm::wag31 were treated with 5 mM IVN for 12 h and 1 mL cells of both strains were stained with 2.5 mg BODIPY. Staining was performed for 30 min protected from light in an incubator shaker followed by washing with sterilized Phosphate buffer saline (PBS).

Sample preparation for Lipidomics and analysis

Secondary cultures of Msm and Δwag31 were seeded at a density of A₆₀₀∼ 0.05 and either left untreated or treated with 100 ng/mL ATc for 12 h until A₆₀₀∼ 0.8-1.0. Post 12 h incubation with ATc, cells corresponding to A₆₀₀∼20 were processed for lipid extraction as described (64). Detailed methodology is provided in supplementary text.

Immunoprecipitation (IP) and MS/MS analysis

250 mL secondary cultures of Msm::gfp and Δwag31 were inoculated at an A₆₀₀∼ 0.05 and incubated in a shaker incubator until A₆₀₀∼ 0.8-1.0. Cells were collected and lysed using bead-beating to make Whole cell lysate (WCL) and passed through a 0.2 μM filter. 3 mg lysate from each group was immunoprecipitated overnight with anti-FLAG^® M2 magnetic beads (Millipore), washed thrice, and heated in 4X Laemmli Buffer. One-fifth of the IP supernatant was resolved on a 10% gel, transferred onto a Nitrocellulose membrane, and probed with anti-FLAG Antibody (Sigma). Lysates from confirmed IPs (biological quadruplets) were taken further for mass spectrometry sample preparation and analysis as described in supplementary text.

Proximity-Based Labelling of Membrane-Associated Proteins (PLiMAP)

For carrying out the PLiMAP assay, Wag31 and its mutants were purified and desired liposomes were prepared as described in supplementary. PLiMAP was carried out and analysed as described (65). Briefly, 2 µM Wag31 or its mutants were incubated with liposomes containing 30 mol% CL, DOPG, or DOPE (200 µM total lipid) in a final volume of 30μL. The reaction was incubated for 30 min in the dark at room temperature followed by exposure to a 1 min long pulse of 365 nm UV light (UVP crosslinker CL-1000L) at an intensity of 200mJ cm⁻². The reaction was mixed with Laemmli sample buffer, boiled, and equal volumes for all reactions were loaded onto a 10% polyacrylamide gel and resolved using SDS-PAGE. iBright1500 (Thermo Fischer Scientific) or Amersham Typhoon (GE) were used for imaging gels. Gels were first imaged for BODIPY fluorescence and later fixed and stained with Coomassie Brilliant Blue.

Tethering assays, fluorescence imaging, and image analysis

µM of Wag31 and its mutants were mixed with 10 µM of 100 nm extruded liposomes in Eppendorf. The mixture was then transferred to a LabTek chamber and imaged using 100x 1.4 NA oil-immersion objective on an Olympus IX73 inverted microscope connected to an LED light source (CoolLED) and an Evolve 512 EMCCD camera (Photometrics). Image acquisition was controlled by μManager, and images were analysed using Fiji. For single liposome intensity analysis, all images from the same condition from two independent experiments were combined into one montage. The montage was duplicated and thresholded using Fiji’s Otsu algorithm to generate a binary mask. The mask was overlaid onto the original montage to obtain the maximum intensity of liposomes.

NAO labelling, imaging and analysis

Msm, Msm::wag31 were treated with 5 mM IVN for 12 h and Msm and Δwag31 were either left untreated or treated with 100 ng/mL ATc for 12 h. For complementation experiment, 1 mL of each strain was stained with 2 mM NAO for 1 h protected from light in an incubator shaker and washed thrice with sterilised 1X PBS. 2 ml of cells were spotted on a 1% agar pad (prepared in 7H9) and visualized using a 100X/1.32 NA oil immersion objective on a Leica THUNDER Imaging systems at 475nm:535nm (Excitation: Emission). About 20 Z-stack images, each spaced 0.1 mM apart, were acquired using a scientific CMOS K8 camera (Leica microsystems). For CL specific labelling, collection was also performed at 642nm.

Complementation experiment

Secondary cultures of Msm, Δwag31, Δwag31::wag31, Δwag31::wag31_K20A, and Δwag31::wag31_D1-60 were set up at A₆₀₀ ∼0.05. For each strain, two types of cultures were set up in triplicates-untreated and treated with 100ng/mL ATc and grown for 12 h. At 0 and 12 h, appropriate serial dilutions were plated on 7H11 agar plates and incubated at 37^°C for 2-3 days. CFUs were enumerated and plotted using GraphPad Prism 9. Statistical analysis was performed using two-way RM ANOVA followed by Tukey’s multiple comparison test, a=0.01, GP: 0.1234 (ns), 0.0332 (*), 0.0021 (**), 0.0002 (***), <0.0001 (****). For BODIPY-and NAO-labelling, the strains were grown as described above. At 12 h, 1 mL cultures of Msm+ATc, Δwag31, Δwag31 +ATc, Δwag31::wag31+ATc , Δwag31::wag31_K20A +ATc, and Δwag31::wag31_D1-60 +ATc were taken and labelled with either BODIPY or NAO. The labelling, imaging and analysis for both fluorophores was performed as described in the respective sections above.

His pulldown assay

E.coli and Msm lysate preparation for the assay is described in supplementary. Briefly, 500 mg of E. coli lysates expressing Wag31 or mutant were mixed with 300 mg Msm lysates expressing Wag31 interactors in a total reaction volume of 700 mL and the samples were twirled at 4°C for 2 h. 50 mL cobalt beads/ sample (GoldBio) were washed twice with lysis buffer and incubated with 5% BSA solution for 2 h at 4°C. Subsequently, beads were washed with lysis buffer and incubated with samples for 2 h (twirled at 4°C). The pull downs were washed thrice in lysis buffer containing 1% Triton X-100 and finally resuspended in 2X-SDS sample buffer. Pulldown samples were resolved on 10-12% SDS-PAGE and probed with a-His and a-FLAG antibodies.

Supplementary material

Materials and methods

Bacterial culturing

Antibiotic concentrations used: Hygromycin (Hyg) was used at 100 μg/mL, Kanamycin (Kan) at 25 μg/mL, and Apramycin (Apra) at 30 μg/mL for Msm. E. coli was cultured in Miller Luria Bertani Broth (Himedia) or Miller Luria Bertani Agar. Antibiotic concentrations used: Hyg was used at 150 μg/mL, Kan at 50 μg/mL, Apra at 30 μg/mL, Ampicillin (Amp) at 100 μg/mL for E. coli.

Generation of wag31 conditional mutant

First, wag31 was cloned under a tet regulatable promoter in pST-KiT_off (kan^res ; 68), which turns of transcription upon the addition of anhydrotetracycline. pST-wag31 construct thus generated was electroporated into a recombineering proficient strain, Msm::pJV53. An allelic exchange substrate was generated by cloning 675 bp from upstream and 700 bp from downstream of wag31 (including ∼150 bp at both ends) with hyg^res cassette and oriE+cosλ and digested with SnaBI to release the donor LHS-hyg^res-RHS. The linear donor was electroporated into Msm::wag31 expressing gp60 and gp61 recombinases. The colonies were selected on Hyg and ten colonies were screened with multiple primer pairs to confirm recombination at the native locus. Confirmed recombinants were passaged multiple times to cure them of pJV53 to generate conditional mutant Δwag31.

BODIPY imaging and analysis

For Msm and Δwag31, ∼10 ml of cells were spotted on a 1% agar pad (prepared in 7H9) and visualized at 4X zoom through Confocal microscopy (LSM880) at 488 nm/510 nm (Excitation: Emission). Background correction in the collected images followed by Corrected Total Cell Fluorescence analysis was performed using Fiji (69) and plotted using Graph Pad Prism 9. For examining lipids in BODIPY-stained Msm and Msm::wag31, 2 ml of cells were spotted on a 1% agar pad (prepared in 7H9) and visualized using a 100X/1.32 NA oil immersion objective on a Leica THUNDER Imaging systems at 475nm:519nm (Excitation: Emission). About 20 Z-stack images, each spaced 0.1 mM apart, were acquired using a scientific CMOS K8 camera (Leica microsystems). Background correction and Thunder image processing was performed on the images using the Las-X software module (Leica microsystems). Processed images were then analysed for CTCF using Fiji (69) and plotted using Graph Pad prism 9.

Lipid extraction and analysis

The dried lipid extracts were re-solubilized in 200 μL of 2:1 CHCl₃: MeOH, and 10 μL was injected into an Agilent 6545 QTOF (quadrupole-time-of-flight) instrument for semi-quantitative analysis using high-resolution auto MS/MS methods and chromatography techniques. A Gemini 5U C-18 column (Phenomenex) coupled with a Gemini guard column (Phenomenex, 4x3 mm, Phenomenex security cartridge) was used for LC separation. The solvents used for negative ion mode were buffer A: 95:5 H₂O:MeOH + 0.1% ammonium hydroxide and buffer B: 60:35:5 Isopropanol:MeOH:H₂O + 0.1% ammonium hydroxide. The 0.1% ammonium hydroxide in each buffer was replaced by 0.1% Formic acid + 10 mM Ammonium Formate for positive ion mode runs. All LC/MS runs were for 60 min, starting with 0.3 mL/min 100% buffer A for 5 minutes, 0.5 mL/min linear gradient to 100% buffer B over 40 minutes, 0.5 mL/min 100% buffer B for 10 minutes, and equilibration with 0.5 mL/min 100% buffer A for 5 minutes. The following settings were used for the ESI-MS analysis: drying gas and sheath gas temperature: 320°C, drying gas and sheath gas flow rate: 10 L/min, fragment or voltage: 150 V, capillary voltage: 4000 V, nebulizer (ion source gas) pressure: 45 psi and nozzle voltage: 1000 V. For analysis, a lipid library was employed in the form of a Personal Compound Database Library (PCDL), and the peaks were validated based on relative retention times and fragments obtained. This library was selectively curated from the MycoMass database and the LIPID MAPS Structure Database (LMSD). All robustly detected lipid species were quantified by normalizing areas under the curve (AUC) to the AUC of the relevant internal standard added and by normalizing to the total cell number. Log₂ fold changes were then plotted for each lipid compared to the Msm-ATc samples.

Immunoprecipitation (IP) sample preparation and MS/MS analysis

Sample preparation and, mass spectrometry analysis were outsourced to VProteomics, New Delhi. Briefly, the IP samples were reduced with 5 mM TCEP, followed by alkylation with 50 mM iodoacetamide. The samples were digested with Trypsin (1:50, Trypsin/lysate ratio) for 16 h at 37 °C. Digests were cleaned using a C18 silica cartridge and resuspended in a buffer containing 2% acetonitrile and 0.1% formic acid. 1 mg of peptide was used for analysis in an Easy-nlc-1000 system coupled to an Orbitrap Exploris mass spectrometer. The gradients were run for 110 min. MS1 spectra were acquired in the Orbitrap (Max IT = 60 ms, AGQ target = 300%; RF Lens = 70%; R=60K, mass range = 375−1500; Profile data). Dynamic exclusion was employed for 30 seconds, excluding all charge states for a given precursor. MS2 spectra were collected for the top 20 peptides. MS2 (Max IT=60ms, R= 15K, AGC target 100%). Proteome Discoverer (v2.5) was used to analyse the RAW data against the Uniprot Msm database. The precursor and fragment mass limitations for the dual Sequest and Amanda searches were established at 10 ppm and 0.02 Da, respectively. The false discovery rate for proteins and peptide spectrum matches was adjusted to 0.01 FDR. Proteins found in at least three replicates with minimum unique peptides #2 were considered interactors. To identify top hits, PSM cut-off was set to 18 and unique peptides to 5.

Protein expression and purification

BL21(DE3) cells transformed with pET-wag31, or pET-wag31K20A or pET-wag31_D1-60 were cultured at 37°C until the A₆₀₀ ∼0.6. 0.1 mM IPTG was added to the culture to induce protein expression, followed by incubating the culture at 18°C for 12 h. Cells were pelleted and stored at -40 °C. The bacterial pellet was resuspended in a lysis buffer (20 mM HEPES pH 7.4, 150 mM NaCl, and 20 mM imidazole with 1 mM phenylmethylsulphonyl fluoride (PMSF)) and lysed by sonication in an ice-water bath. Lysate was spun at 30,000 g for 20 min, and the supernatant obtained was incubated with TALON^® Metal Affinity Resin (Takara Bio) 4 °C for 30 min. The resin was extensively washed with lysis buffer, and the bound protein was eluted with 20 mM HEPES pH 7.4, 150 mM NaCl, and 250 mM imidazole. The elution was dialyzed against 20 mM HEPES pH 7.4 150 mM NaCl and stored at 4 °C for the duration of the experiments. Aggregates were removed by centrifugation at 100,000 g. Concentration of the proteins was quantified by measuring A₂₈₀ using the molar extinction coefficient predicted by the Expasy ProtParam tool.

Liposome preparation

1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) and1′,3′-bis[1,2-dioleoyl-sn-glycero-3-phospho]-glycerol (sodium salt), CL were obtained from Avanti Polar Lipids. 1,1’-dilinoleyl-3,3,3’,3’-tetramethylindocarbocyanine, 4-chlorobenzenesulfonate (FAST DiI™), and 3,3’-dilinoleyloxacarbocyanine perchlorate (FAST DiO™) were obtained from Invitrogen. The UV-activable, diazirine-containing reactive fluorescent lipid probe BODIPY-diazirine phosphatidylethanolamine (BDPE) was prepared as described earlier (68, 69). For liposome preparation, lipids were aliquoted from chloroform stocks at desired ratios in a clean glass tube and dried under a high vacuum for an hour to a thin film. The dried lipids were used at a final concentration of 1 mM (deionised water was added to them). After hydrating at 50 °C for 30 minutes, lipids were vigorously vortexed and then extruded through polycarbonate filters with pore sizes of 100 nm (Whatman).. For PLiMAP assays, liposomes were CL, DOPG, or DOPE with 69 mol% DOPC and 1 mol% BDPE. Liposomes used in tethering assays were made similarly with 30 mol% CL or DOPE and 1 mol% of the fluorescent lipid DiI or DiO) in the DOPC background.

NAO imaging analysis

Background correction and Thunder image processing was performed on the images using the Las-X software module (Leica microsystems). Processed images were then analysed using Fiji (69) and plotted using Graph Pad prism 9. Briefly, N=100 cells across two independent experiments were examined for their profile of CL distribution by normalizing the cell length and assigning pole with brighter NAO fluorescence intensity as 0 and the other pole as 100. The fluorescence intensity was normalised to the highest fluorescent value for each cell. Corresponding fluorescence intensity value for all cells for every point between 0 to 100 were calculated from the LOOKUP function of excel. The values representing CL fluorescence across 100 cells were averaged out and plotted against normalised cell length in GraphPad prism 9. 0^th order smoothening with 4 neighbours on each size was performed on the curves obtained.

His pulldown assay-E.coli and Msm lysate preparation

BL21 (DE3) cells transformed with no plasmid, or pET-wag31 or pET-wag31_D1-60 were cultured as described above and cell lysates were prepared. The expression of Wag31 or Wag31_D1-60 was assessed by resolving the samples on 10% SDS-PAGE followed by Coomassie staining. The interacting partners, namely murG, sepIVA, msm2092 and accA3 were amplified from Msm genomic DNA using gene specific primers harbouring NdeI-HindIII sites and amplicons were digested and cloned into the corresponding sites in pNit-3X-FLAG shuttle vector. The constructs thus generated were electroporated in Msm to generate Msm::murG, Msm::SepIVA, Msm::msm2092 and Msm::accA3. The cultures grown till A₆₀₀ ∼0.8 were used for fresh inoculation at A₆₀₀ of 0.05 in 7H9 media containing 0.5 mM IVN. Cultures were grown for 12 h and lysates were prepared by bead-beating and 40 mg were resolved on 10% gels and probed with a-FLAG antibody.

Acknowledgements

We thank Dharani Kumar, SEST, University of Hyderabad for helping with imaging of SEM stubs; CCMB for access to its Genomics, Advanced Microscopy, and Electron Microscopy facilities; VIDO, USask for providing infrastructure to carry out some part of the work; Asis Kumar Khuntia for the generation of pET28b-wag31_K20A and pET28b-wag31_1-60. YK was supported by a Research Associateship from NIH (Grant 6044-SC23-26). YK and ND acknowledge MITACS for the Globalink Research Award (FR118569) which supported the work carried out at VIDO. ND also acknowledges support from Canadian Institutes of Health Research (CIHR, funding reference number 185715). VIDO receives operational funding from the Government of Saskatchewan through Innovation Saskatchewan and the Ministry of Agriculture and from the Canada Foundation for Innovation through the Major Science Initiatives Fund. Research in this publication was supported by SERB grant (CRG/2018/001294), Govt. of India, and JC Bose Award (JCB/2019/000015), Govt. of India to VKN. Work in the SSK laboratory was supported by the Swarna Jayanti Fellowship to SSK by SERB, Govt. of India (Grant: SB/SJF/2021-22/01). AC was supported by a Senior research fellowship from CSIR, India. HK thanks the Indian Institute of Science Education and Research, Pune, for a graduate fellowship. TP thanks the Howard Hughes Medical Institute for funding support. We thank Dr. Christopher Sassetti for his critical comments.