Selective Glycan Labeling of Mannose-Containing Glycolipids in Mycobacteria

Abstract

Mycobacterium tuberculosis (Mtb) is one of history’s most successful human pathogens. By subverting typical immune responses, Mtb can persist within a host until conditions become favorable for growth and proliferation. Virulence factors that enable mycobacteria to modulate host immune systems include a suite of mannose-containing glycolipids: phosphatidylinositol mannosides (PIMs), lipomannan (LM), and lipoarabinomannan (LAM). Despite their importance, tools for their covalent capture, modification, and imaging are limited. Here, we describe a chemical biology strategy to detect and visualize these glycans. Our approach, biosynthetic incorporation, is to synthesize a lipid-glycan precursor that can be incorporated at a late-stage step in glycolipid biosynthesis. We previously demonstrated selective mycobacterial arabinan modification by biosynthetic incorporation using an exogenous donor. This report reveals that biosynthetic labeling is general and selective: it allows for cell surface mannose-containing glycolipid modification without non-specific labeling of mannosylated glycoproteins. Specifically, we employed azido-(Z,Z)-farnesyl phosphoryl-β-d-mannose (AzFPM) probes and took advantage of the strain-promoted azide-alkyne cycloaddition to label and directly visualize the localization and dynamics of mycobacterial mannose-containing glycolipids. Our studies highlight the generality and utility of biosynthetic incorporation, as probe structure directs the selective labeling of distinct glycans. The disclosed agents allowed for direct tracking of the target immunomodulatory glycolipid dynamics in cellulo. We anticipate that these probes will facilitate investigating the diverse biological roles of these glycans.

INTRODUCTION

Mycobacterium tuberculosis (Mtb) causes tuberculosis, which kills over one million people each year.¹ An estimated 3 billion people are infected with latent Mtb worldwide.² The ability of Mtb to suppress the host immune system during infection and persist in a latent form severely exacerbates its impact.^3-6 Mannose-containing glycolipids produced by mycobacteria, including phosphatidylinositol mannosides (PIMs), lipomannan (LM), lipoarabinomannan (LAM), and mannosylated lipoarabinomannan (ManLAM), modulate the host immune response.^7-12 Host recognition of bacterial glycans is critical for immune responses and pathogen clearance, and in the case of Mtb, mannose-containing glycolipids participate in multiple immune evasion mechanisms. For instance, LAM inhibits lysosome-phagosome fusion, allowing Mtb to form an intracellular niche within host macrophages.^{13, 14} Mannose-containing glycolipids have also been implicated in maintaining cellular integrity.^{11, 13}

Despite the importance of these glycolipids, few direct methods exist to study their biosynthesis, trafficking, and biological function. Monoclonal antibodies (mAbs) that recognize LAM or LM can be used^{15, 16}, yet an antibody’s size (180kDa) can also limit its utility for visualizing the LAM (15kDa) as structural occlusion of other membrane components can complicate in cellulo analyses.^{17, 18} Thus, new strategies are needed for the selective structural perturbation of mannose-containing glycans to probe their composition, function, and dynamics within the context of live cells and disease pathogenesis.¹⁹

Our understanding of the roles of these mannose derivatives would be enhanced by methods to visualize them. Because glycan biosynthesis is not directly genetically encoded, installing labels through genetic manipulation is difficult.²⁰ Metabolic incorporation offers an incisive route to installing handles into glycans.^21-25 The process involves multi-step, intracellular processing of simple monosaccharide building blocks by endogenous metabolic machinery, culminating in the presentation of the end products on the cell surface.²³ Though several successful examples have been demonstrated in bacteria, applying this strategy in these organisms poses challenges due to the complexity of glycan metabolism and catabolism in bacteria.^{21, 26-33}

The only known probe of the PIMs, LM, and LAM is based on the metabolic incorporation of azide-functionalized inositol.³⁴ Still, competition with the endogenous substrate occurs at multiple steps, so selective probe incorporation is contingent upon deleting genes in the endogenous inositol biosynthetic pathway. Though this requirement prevents the use of this probe in wild-type cells, these metabolic incorporation probes led to the discovery of a previously unidentified inositol transporter. Still, no probes for labeling cell-surface mannose residues have been disclosed.

Two key challenges render mannose an elusive target. First, mannose plays a central role in mycobacterial metabolism. The mannose precursors for the relevant mycobacterial glycans are generated from the isomerization of fructose-6-phosphate by mannose-6-phosphate isomerase or through mannose import and elaboration.^{35, 36} Intracellular mannose can be used for energy production or converted to fucose, rhamnose, or mannose sugar donors. Consequently, a probe that depends on metabolic incorporation would have a higher risk of off-target labeling than one that intervenes at a late-stage in biosynthesis. Secondly, glycolipid donor forms of mannose are used in mycobacteria for the biosynthesis of mannose-containing glycolipids and mannosylated proteins. Thus, distinguishing between these two classes of glycans via metabolic incorporation would be challenging. Such distinctions are important, however, as mannosylated proteins in mycobacteria are also linked to virulence.⁷ Thus, dissecting the roles of different glycan structures depends on specific labeling strategies that differentiate between mannosylated glycoproteins and glycolipids.

We postulated that intervening late in the biosynthetic pathway could facilitate the labeling of the target structures with selectivity for glycolipids over glycoproteins. We define this late-stage metabolic engineering strategy as biosynthetic incorporation. To this end, we designed probes that could be used directly by glycosyltransferases, thereby limiting competition with endogenous substrates to a single step to increase labeling efficiency. Because most metabolic processes occur inside the cell, we postulated that agents that exploit extracellular enzymes would selectively modify specific cellular glycans, such as glycolipids over glycoproteins. Neither the locale nor structure of the donor for protein mannosylation are well characterized in mycobacteria. However, the proposed dolichyl donor for glycoprotein biosynthesis and polyprenyl glycolipid donor structures vary in the degree of β-γ saturation, and preservation of this motif in farnesyl-bearing probes may enable specific modification of glycolipid targets. Second, protein glycosylation is likely to occur intracellularly rather than at the cell surface and thus we hypothesized that selectivity could be dictated by probe localization and lack of uptake. This information gave us optimism that we could generate a selective probe of mycobacterial glycolipids.

We previously synthesized and applied lipid-linked arabinofuranose derivatives that were incorporated into the mycobacterial cell envelope. Specifically, synthetic azido-(Z,Z)-farnesyl phosphoryl-β-d-arabinofuranose (AzFPA) derivatives function as surrogates for the endogenous arabinofuranose donor decaprenyl phosphoryl-β-d-arabinofuranose (DPA).^{37, 38} The shorter lipid facilitated exogenous reagent delivery, such that the isomeric azide-substituted AzFPA probes led to an azide-containing arabinogalactan. We observed different modification levels depending on the bacterial species, and the data indicated that the azido-arabinose derivatives had been used by periplasmic glycosyltransferases to generate a modified arabinogalactan. LM and LAM synthesis also occurs in the periplasm mediated by the glycosyltransferase-C (GT-C) superfamily of mannosyltransferases that utilize a polyprenyl phosphoryl-β-d-mannose (PPM) as their donor.^{39, 40} Consequently, we set out to leverage this information to visualize and track these glycans.

Herein, we report the synthesis, validation, and application of probes to study mannose-containing glycolipids in model mycobacterial organisms and Mtb (Figure 1A). We prepared mannosylated lipid-linked donors with azide handles. These derivatives are successfully used by cells and, therefore, can be used to monitor these glycans within the cell envelope. Our results underscore the generality of late-stage biosynthetic incorporation with a lipid-linked donor as a glycan bioconjugation strategy. Beyond demonstrating the generality, however, our findings highlight the specificity of this approach. We show the probes label mannosylated glycolipids over glycoproteins. Indeed, we could even selectively label subsets of glycolipids, depending on the site of azide substitution. In validating the probes in Mtb, we provide new tools to study the roles of these glycans in disease.

Figure 1. Overview schematic of biosynthetic labeling strategy targeting glycolipids.

(A) Previous studies led to the identification of an effective lipid motif for membrane-associated glycosyltransferase recognition. We leverage this information to selectively label immunodulatory mycobacterial glycolipids with a novel suite of glycolipid probes (azido-(Z,Z)-farnesyl phosphoryl-β-D-mannose, AzFPM). (B) Mannose-containing glycolipids are elongated via 1,6-glycosidic bonds and feature branching at the 2-position.

RESULTS AND DISCUSSION

Design and Synthesis of Azido-Mannose Probes

Biosynthetic incorporation in mycobacteria has been applied to a single target, arabinose, within arabinogalactan.^{37, 41} We hypothesized the approach would be general and could be exploited to differentiate between glycoprotein and glycolipid modification. Based on our previous success using a farnesyl motif to mimic the endogenous polyprenyl donor, we designed a panel of azido-(Z,Z)-farnesyl phosphoryl-β-d-mannose (AzFPM) derivatives (Scheme 1A). We generated this panel because we reasoned that the efficiency of non-natural mannose incorporation could vary between different isomers and between model species, as seen previously with arabinose derivatives.³⁷ Specifically, the core of the LAM is a polymer consisting of α-1,6 linked mannose residues, and this backbone is elaborated with 2-substituted α-mannose. This arrangement suggests that the incorporation of mannose derivatives with azido groups 3- or 4-positions might be the least perturbing, as they would not lead to chain truncation (Figure 1B). However, the efficiency of incorporation could be impacted by enzyme-substrate complementarity, so we also tested substrate mimics with azide groups at the 2 and 6 positions (Scheme 1A).

Scheme 1. Design and synthesis of AzFPM probes.

(A) The azide-modified substrate surrogates were designed based on structural homology to the endogenous mannose donor, polyprenyl phosphoryl mannose. Four azide regioisomers were produced (1–4). (B) Synthetic routes were established to produce 1-4: a: i) acetic anhydride, pyridine (78%), ii) NaN₃, DMF (97%) b: i) NaBH₄, CeCl₃ (93%), ii) AgOAc, I₂, then MeOH, NH₄OH (88%) c: NaN₃, DMF (81%) d: TFA, acetic anhydride (84%, α:β=1:0.36) e: i) tBuOOH, DBU (58%) ii. NaBH₄ (63%) f. NaN₃, DMF (69%) g: TFA, acetic anhydride (84%, α:β=1:0.24) h: i) NH₂NH₃OAc or benzylamine (68-88%) ii) POI(Oallyl)₂ (46-68%) i: i) Pd(PPh3)₄, ii) Cl₃CCN, (Z, Z)-farnesol, iii) NH₃, MeOH (8-28% over three steps)

The site of azide group placement could also manifest in labeling differences between organisms, as the LAM varies significantly. For instance, Corynebacterium glutamicum produces AraLAM, in which a mannan core is terminated with arabinose residues, whereas Mycobacterium smegmatis generates a phosphoinositol-capped PILAM, and Mtb affords a mannose-capped ManLAM. Accordingly, we prepared and evaluated all four possible regioisomeric AzFPM probes.

The AzFPM regioisomers were synthesized from commercially available mannose derivatives and an abundant cellulose degradation product dihydrolevoglucosenone (Cyrene^™) (Scheme 1B). Reduction of the bicyclic unsaturated compound to the allylic alcohol afforded a compound that could be exploited to yield two diastereomeric epoxides. To generate the 4-azido derivative, we took advantage of halohydrin formation to afford an epoxide that could undergo reaction with azide to ultimately generate compound 9. In contrast, a nucleophilic epoxidation of the starting material afforded the diastereomeric epoxide that eventually led to the 3-azido analog. Thus, each route exploited the stereochemical control afforded by the bicyclic precursor to afford the desired derivatives in high yield. In all cases, the critical azide functionality was installed through nucleophilic displacement or nucleophilic epoxide opening. We appended the (Z,Z)-farnesyl recognition motif and removed the protecting groups to afford the target compounds (1-4). The last steps were rigorously optimized. Their modest yield is due to inconsequential counterion speciation and the challenges inherent in the purification of such compounds.

AzFPM Probes Label Mannose-Containing Glycans in cellulo

We first confirmed that AzFPM probes were incorporated into the cell envelope of model organisms C. glutamicum and M. smegmatis (Figure 2A). Bacterial cultures were supplemented with AzFPM in media and grown to the late-logarithmic phase. Probes at this concentration did not affect cell viability, though at higher concentrations 2-AzFPM inhibited growth in M. smegmatis (Figure SI-1). Following growth, the labeled cells were washed to remove any remaining free probe and then exposed to Alexa Fluor 647 (AF647)-conjugated dibenzocyclooctyne (DBCO). Thus, the fluorescent label was appended via a strain-promoted azide-alkyne cycloaddition (SPAAC)⁴². Fixed samples were analyzed by flow cytometry (Figure SI-2).

Figure 2. AzFPM labeling in cellulo.

(A) Flow cytometry analysis of AzFPM (125 μM) labeled C. glutamicum and M. smegmatis treated with DBCO-AF647. Mean fluorescence intensity (MFI) was calculated using the geometric mean and plotted relative to a dye-only control. Error bars denote the standard error of the mean of three replicate experiments. (B) Streptavidin-enrichment of labeled glycans from AzFPM (125 μM) labeled C. glutamicum and M. smegmatis reacted with DBCO-biotin. Load and eluent analyzed by SDS-PAGE using Pro-Q^™ Emerald 300 glycan staining kit. (C) Streptavidin-enrichment of labeled proteins from AzFPM (125 μM) labeled C. glutamicum and M. smegmatis reacted with DBCO-biotin. Load and eluent analyzed by SDS-PAGE using Silver Stain. L: ladder, 2M: 2-AzFPM, 3M: 3-AzFPM, 4M: 4-AzFPM, 6M: 6-AzFPM. Data is representative of two independent experiments.

AzFPM-treated cells had increased fluorescence relative to cells treated only with DBCO-dye conjugate. The degree of AzFPM incorporation, measured by flow cytometry-derived relative mean fluorescence intensity (MFI) levels, varied across organisms. In M. smegmatis, the 2-AzFPM probe was efficiently utilized, whereas the incorporation of the 6-AzFPM probe was negligible. In C. glutamicum, the 6-AzFPM probe afforded the highest levels of incorporation, whereas the signal from the 3-AzFPM probe was minor. This labeling was shown to be dose-dependent (Figure SI-3). The monosaccharide dependence indicates that AzFPM probes are substrates for mycobacterial mannosyltransferases. The organism-dependent labeling of each regioisomer highlights the value of testing different isomers. Moreover, the results could not have been predicted solely from the propensity of different probes to result in chain termination. We validated that the observed flow cytometry data correlated with membrane-localized labeling using super-resolution structured illumination microscopy (SIM) (Figure SI-4). These results demonstrate the generality of biosynthetic incorporation probes bearing farnesyl lipid motifs.

We then isolated the labeled structures to assess the specificity of our strategy. Cells grown in media supplemented with AzFPM were then labeled for enrichment by reaction with DBCO-biotin. To capture modified components, the cells were lysed, and beads modified with streptavidin were added. The bead-captured compounds were analyzed by gel electrophoresis using a glycan stain. We detected bands consistent with the expected molecular weight for LAM and LM (Figure 2B). In C. glutamicum, 4- and 6-AzFPM primarily label the LAM. In M. smegmatis, the 3- and 4-AzFPM result in robust LM labeling with some signal from the LAM. Although the LAM contains arabinose residues in addition to mannose residues, the AzFPA probes were not incorporated into the LAM but only into the arabinogalactan (Figure SI-5). This selectivity is unexpected as the arabinofuranose polymers in the LAM and the arabinogalactan are thought to be biosynthesized by the same enzymes using an identical lipid-linked donor.^43-47 Our probes indicate differences in the biosynthesis of these two cell envelope components that are under further investigation.

To investigate AzFPM labeling selectivity, we visualized the enriched samples using silver stain, given its high sensitivity relative to conventional Coomassie stains.⁴⁸ No enriched soluble proteins were detected (Figure 2C, SI-6, SI-7). This observation indicates that our probes are processed by mannosyltransferases responsible for glycolipid biosynthesis. This result is notable as there is a known mycobacterial membrane-associated protein-O-mannosyltransferase (Msmeg_5447) with sequence homology to eukaryotic protein-O-mannosyltransferases which accept polyprenyl lipid carrier substrates (dolicholphosphate). Thus, despite the similarity in these proposed biosynthetic precursors, AzFPM probes retain specificity for the desired glycolipid targets.

Azide Regioisomer Dictates the Selectivity of PIM Labeling

Although 2-AzFPM is incorporated into M. smegmatis, it does not appear to label LAM or LM. Specifically, we could not detect probe incorporation into the LAM or LM. We, therefore, hypothesized that this probe is labeling another mannose-containing structure–phosphatidylinositol mannosides (PIMs). PIMs are implicated as key virulence factors and are known ligands for C-type lectin dendritic cell (DC)-specific intercellular adhesion molecule 3-grabbing nonintegrin (DC-SIGN).⁴⁹ We hypothesized that 2-AzFPM might be used in the final steps of PIM biosynthesis. To test this possibility, we acquired an M. smegmatis deletion lacking the lipid-dependent glycosyltransferase PimE (Figure 3A). This enzyme acts at the first step in which extended PIM and LM/LAM biosynthesis diverge.⁵⁰ If 2-AzFPM is labeling PIMs, the knock-out cell line should exhibit a decrease in cellular staining.

Figure 3. Investigation of PIM labeling using a genetic model.

(A) The point of divergence in the biosynthesis of PIMs versus LM and LM indicates the relevance of the PimE knockout. (B) MFI from flow cytometry analysis of AzFPM (125 (μM) labeled M. smegmatis wild-type and ΔPimE cells treated with DBCO-AF647. An asterisk denotes a statistically significant differences by two-way ANOVA between samples (**** p < 0.0001).

We observed a stark decrease in staining with 2-AzFPM in M. smegmatis ΔpimE compared to wild-type cells. In contrast, we observed only a minor decrease in staining with 3-AzFPM (Figure 3B). We employed 7-hydroxycoumarin-3-carboxylic acid-3-amino-D-alanine (HADA), a fluorescent d-amino acid probe that labels the peptidoglycan,⁵¹ to benchmark the decrease in staining detected in the M. smegmatis ΔpimE cells. A similar degree of staining by HADA was detected in M. smegmatis ΔpimE and wild-type cells, demonstrating the metabolic competence of this cell line (Figure SI-8). C. glutamicum lacks extended PIMs or a PimE ortholog. As expected, there was no discrepancy between staining and LAM/LM pulldown upon 2-AzFPM exposure in this organism.^{28, 31} Thus, using a specific azide-substituted mannose isomer, we obtained highly selective labeling of PIMs in wild-type M. smegmatis. This finding emphasizes further the utility of biosynthetic incorporation probes for selective labeling of specific glycans.

Visualizing the in cellulo Biosynthesis of Mannose-Containing Glycans

We next applied these tools to investigate the sites of mannose-containing glycan biosynthesis in cellulo. Peptidoglycan, arabinogalactan, and mycolic acid biosynthesis all occur at the poles and septa of dividing cells. Limited information is available on the sites of LM and LAM biosynthesis or localization. To address this knowledge gap, we labeled C. glutamicum with the probe that stained to the greatest extent, 6-AzFPM, and visualized the sites of its localization (Figure 4, Figure SI-9). We used an incubation time to capture the initial sites of modification and opted for C. glutamicum as our model as it's brighter staining and larger cell width enables visualization of staining patterns more readily. We compared our results to those obtained with our previously described 5-AzFPA probe as a control for known biosynthesis patterns. Although 5-AzFPA labeled arabinose localized to the bacterial septa and poles at short time points, 6-AzFPM labeling occurred along the entire cell wall (see Figure SI-10 for quantification). This annular staining suggests that the localization of mannose-containing glycans is decoupled from known areas of cell envelope biosynthesis.

Figure 4. Analysis of localization of AzFPM and AzFPA probes visualized via confocal fluorescence microscopy.

Diagram of cell envelope components that are labeled by each probe. Arabinogalactan (AG) is covalently connected to the peptidoglycan, whereas LAM structures are non-covalently embedded in the mycolic acid (MA) layer. mAGP: mycolyl-arabinogalactan-peptidogalactan, PM: plasma membrane. Images taken after 2 h labeling period (Scale bar = 3 μm).

To assess whether this localization corresponds to that of relevant biosynthetic enzymes, we generated C-terminal mCherry fusions of two M. smegmatis mannosyltransferases, one which forms α-1,6 linkages, mptA (Msmeg_4241), and one which forms α-1,2 branching linkages mptC (Msmeg_4247) (Figure SI-11). We chose to perform these experiments in M. smegmatis as this model has a more elaborated LAM structure. The fluorescent proteins localized primarily at the septa rather than displaying homogenous fluorescence along the entire cell wall. (Figure SI-12). The decoupling of the localization between the labeled mannose-containing glycans and biosynthetic enzymes suggests that the mannosylated glycolipids freely diffuse.

Tracking Diffusion of Mannose-Containing Glycans

While the inner mycolate and arabinogalactan are covalently linked to the peptidoglycan, the PIMs, LM, and LAM is non-covalently embedded in the outer and inner lipid membranes. Our observations that the localization of the labeled glycolipids and the biosynthetic enzymes are decoupled indicate that the LM and LAM can rapidly diffuse throughout the cell membrane. We tested this hypothesis by employing our chemical probes and fluorescence recovery after photobleaching (FRAP). In FRAP, a portion of the labeled cell is photobleached, and the cell is visualized over a recovery period. If the fluorophore is attached to a diffusible entity, the surrounding fluorescent molecules will redistribute over the bleached spot to restore its signal. If the labeled structure is covalently tethered, it will not diffuse and the photobleached region will remain dark.

We performed FRAP in both C. glutamicum and M. smegmatis after AzFPM treatment and labeling. The results were compared to those obtained with AzFPA,³⁷ which our data indicated led to covalent labeling of the arabinogalactan (Figure 5). As anticipated, cells treated with AzFPA and labeled with a fluorophore did not recover fluorescence after photobleaching. In contrast, cells with modified mannosylated glycolipids afforded fluorescence recovery. Moreover, FRAP experiments revealed that the rate of diffusion varied between model organisms. Those with longer mycolic acids (e.g., M. smegmatis versus C. glutamicum) led to slower diffusion. This trend is consistent with previous studies employing mycolate probes.⁵² To provide another test that could distinguish movement of mannosylated lipids versus proteins, we employed formaldehyde fixation. Formaldehyde reacts with available amines, thereby immobilizing proteins. If our probes yielded mannosylated proteins, they would preserve the photobleached area. In contrast, formaldehyde treatment had no effect on fluorescence recovery, indicating that the mannosylated structures retain the ability to diffuse (Figure SI-13). These data are consistent with glycolipid rather than glycoprotein labeling. These findings provide a first glimpse of LAM or LM diffusion on the mycobacterial cell surface (Figure SI-14). Similar experiments using 2-AzFPM to label PIMs in M. smegmatis indicated the PIMs also readily diffuse (Figure SI-15). The rates of PIMs, LM and LAM diffusion in M. smegmatis were similar. Thus, the organisms’ cell wall structure and lipid anchor dictate the rate of diffusion and not the elaborated glycan.

Figure 5. FRAP analysis to track glycan diffusion.

Representative FRAP profiles of FPM- or FPA-treated C. glutamicum (A) or M. smegmatis cells (B). White dashed circles indicate the areas of photobleaching (Scale bar = 3 μm). On the right fluorescence recovery traces for the cells are shown. Error bars denote the standard error of the mean (n=50 cells).

Testing FPM Probes in Mtb

Though much informative mycobacterial research is derived from nonpathogenic species within the genus,⁵³ the importance of the LAM for disease led us to assess the utility of the probes in Mtb. Thus, we evaluated incorporation by flow cytometry (Figure 6A). The level of incorporation for different probe regioisomers in Mtb was similar to that seen in M. smegmatis (Figure SI-16). We next enriched the labeled structures by modifying them with DBCO-biotin and performing streptavidin pulldowns (Figure 6B). As with M. smegmatis, Mtb exposed to 2-AzFPM afforded bright staining, but we detected no probe incorporation into the LAM or LM. In contrast, 3-AzFPM and 4-AzFPM robustly labeled both the LAM and LM. Thus, as in M. smegmatis, 2-AzFPM likely labels the PIMs in Mtb. Finally, we treated cells with 3-AzFPM and appended a fluorophore, which labels the LAM and LM (Figure 6C). We observed annular staining with the brightest staining at the poles and septum. Together, our data highlight the versatility of these probes in different mycolic acid-containing species. We anticipate these new tools will provide further insight into the role of mannose-containing glycans in pathogenic mycobacteria.

Figure 6. AzFPM labeling in Mtb.

(A) Flow cytometry analysis of AzFPM (125 μM) labeled Mtb treated with DBCO-AF647. Mean fluorescence intensity (MFI) was calculated using the geometric mean and plotted relative to a dye-only control. Error bars denote the standard error of the mean of two replicate experiments. (B) Streptavidin-enriched labeled structures from AzFPM (125 μM) labeled Mtb reacted with DBCO-biotin. Load and eluent analyzed by SDS-PAGE using Pro-Q^™ Emerald 300 glycan staining kit. (C) Confocal fluorescence microscopy images of Mtb grown with 3-AzFPA (125 μM). (Scale bar = 3 μm)

CONCLUSIONS

We generated a suite of biosynthetic incorporation probes and showed they can be used to visualize mannose-containing glycans in mycobacteria. These probes can reveal the location of the PIMs, LM, and LAM on the cell surface of wild-type mycobacteria. Using late-stage biosynthetic intermediates, we could specifically target immunomodulatory structures on mycobacteria's cell surface without engaging the biochemical pathways involved in mannose metabolism. We observed that the structures labeled are dictated by the regioisomeric donor employed and that specific donors allow selective labeling of specific glycans (e.g., PIMs versus other mannosylated lipids). We used these features to track the diffusion of specific glycans on the cell surface. Our findings highlight the utility of a farnesyl-based scaffold to mimic lipids beyond a decaprenyl donor, including the polyprenyl donor used for mannose biosynthesis. Members of the targeted GT-C class of glycosyltransferases are found in humans, yeast, archaea, and bacteria. Thus, these glycolipid donor surrogates could install labels into a wide range of future glycan targets, making these previously invisible cellular components visible.