Chemical approaches to unraveling the biology of mycobacteria

Abstract

Mycobacterium tuberculosis (Mtb), perhaps more than any other organism, is intrinsically appealing to chemical biologists. Not only does the cell envelope feature one of the most complex heteropolymers found in nature¹ but many of the interactions between Mtb and its primary host (we humans) rely on lipid and not protein mediators.^2,3 Many of the complex lipids, glycolipids and carbohydrates biosynthesized by the bacterium still have unknown functions and the complexity of the pathological processes by which tuberculosis (TB) disease progresses offers many opportunities for these molecules to influence the human response. Because of the importance of TB in global public health, chemical biologists have applied a wide-ranging array of techniques to better understand the disease and improve interventions.

Graphical Abstract

eTOC blurb

Mycobacterium tuberculosis is the largest single infectious cause of death globally and the community of chemical biologists has made significant strides in providing new tools and insights into this important pathogen. Finin et al. highlight progress made over the past ten years by this community.

This review will focus on advances in the chemical biology community over the past ten years as they relate to improving our understanding of the pathogen and the host-pathogen dynamic. We broke this literature up into four sections: (1) small-molecule probes of cytosolic and periplasmic processes, (2) probes based on the mycomembrane of Mtb, (3) genetically encoded probes, and (4) techniques that multiplex chemical biology with imaging and genome scale genetics. The latter two sections might not conventionally be considered chemical biology but the rapid expansion of techniques in these sections is moving towards asking fundamentally chemical questions and they are therefore central to the future of chemical biology in mycobacterial systems. We will highlight some examples including using genetically encoded fluorescent proteins and mass spectrometry to understand the molecular basis of drug susceptibility of bacterial subpopulations. As well as exciting new studies using genome wide CRISPR based knockdown libraries to understand the mechanism of action of both old and new drugs. Other techniques, focused solely on detecting bacteria in sputum to improve diagnostics such as quantum dot nanobeacons and isothermal amplification coupled to such nanobeacons, have been covered by other recent reviews and will not be discussed here.^4,5

Small-molecule probes of cytosolic and periplasmic processes

Universal activity-based probes (ABPs)

Many synthetic small molecules have been used for the design of chemical probes to explore the biology of mycobacterial cells and investigate Mtb-host interactions. These have enabled not only general protein profiling and target identification / validation in drug discovery programs but also cell imaging and screening of drugs. In this section, we focus on studies in which probes based on synthetic small molecules were actively used for the study of mycobacterial pathogens. Chemical probes introduced here include the application of active site-directed small molecules (activity-based probes, ABPs), substrates with affinity binding to known target proteins and enzyme-activated fluorogenic probes, many of which were supported by synthetic chemistry approaches for labeling the targets especially using bioorthogonal click chemistry.^6-9

Probes based on ATP have allowed the identification of ATP-binding proteins in mycobacteria. Desthiobiotin-ATP (Figure 1) has been used to biotinylate lysine in the nucleotide binding sites of diverse proteins in Mtb to identify differences in the nucleotide-binding proteome under different growth states.¹⁰ A click-chemistry compatible ATP-ABP (Figure 1) replaced the bulky biotin analog with an alkyne moiety enabling detection of a wider range of nucleotide-binding proteins further modified by click-chemistry to various detectable tags in the subsequent liquid chromatography tandem mass spectrometry (LC-MS/MS) based approach.¹¹ Competition of the ATP-ABP with staurosporine for binding to nucleotide-binding proteins in the proteome identified the serine/threonine kinases PknF, PknD and PknB as the primary targets of this broad kinase inhibitor demonstrating the power of ABP on drug target identification.¹²

Figure 1. A summary of small-molecule probes incorporated into Mtb cells.

(A) Universal activity-based probes (ABPs) have been produced that are based upon the inherent reactivity of fluorophosphonates towards esterases and ATP towards ATP-binding proteins. (B) Antibiotic-based ABPs have been produced that are based upon the inherent reactivity of THL and lalistat towards lipid hydrolases, carbapenems towards penicillin binding proteins, and EZ120P and BMB034 targeted serine hydrolases. (C) Fluorescent analogs of benzothiazinones, a novel class of covalent DprE1 inhibitors, have shown promise as whole cell labeling agents. (D) DDAO-OME and DCF-AME whose fluorescence is unmasked by serine esterases have been used for protein profiling of Mtb cells. Sal-AMS probe was designed to do protein profiling to study Sal-AMS inhibition in Mtb. (E) Several fluorogenic or luminogenic were reported for bacterial specific labeling, which are based on enzyme specificity of CDG series towards β-lactamase (and DprE1), FLASH towards Hip1 and DDAO-sulfate towards sulfatases. (F) Scanning electron micrograph of Mtb cells in mid log phase (false color applied).

Fluorophosphonate (FP)-based ABPs label the active site of serine hydrolases (SH). These have been used to label proteins on the basis of changes in their activity under different environmental conditions and to identify potential pharmaceutical targets in human cells.¹³ In the TB field, two independent protein profiling studies using serine-reactive FP-based ABP have been performed in both replicating and dormant cells. Grundner et al., used a click-chemistry-enabled FP probe, specifically alkyne-PEG-FP (Figure 1), and identified 78 proteins as potential candidates with SH activity expressed under conditions of active bacterial replication. Among them, only 34 proteins were found to be still active in nonreplicating cells, while 3 new SH were selectively labeled in these cells.¹⁴ In another study, Beatty et al., used desthiobiotin-FP (Figure 1) as an ABP and identified 22 esterases in replicating Mtb and confirmed that 5 of them retained activity in hypoxic Mtb cells. Additionally, they compared the protein profiling pattern obtained with the fluorogenic probes, 2’,7’-dichlorofluoresecein diacetyoxymethyl ether (DCF-AME) or 7-hydroxy-9H-(1,3-dichloro-9,9-dimethylacridin-2-one)-based probes (DDAOs),^15,16 to confirm esterase functionality and to detect esterases that were missed by FP-based ABPs, such as Culp1 (Rv1984c) and Rv3036c.^16,17 DCF-AME and DDAO-OME (Figure 1) are ester-masked fluorogenic probes that produce fluorescence upon serine-mediated hydrolysis. Therefore, these “turn-on” substrates and FP-based ABPs can be used as complementary tools for detecting and tracking esterase activities in different metabolic stages of Mtb.

The 1,2,3-triazole urea scaffold covalently modifies the active site of serine hydrolases and some have shown potent inhibition of Mtb growth and survival.¹⁸ The targets of two 1,2,3-triazole ureas was identified by competitive ABP in a method wherein ¹⁴N-labeled Mtb cultures are treated with the inhibitor followed by FP-biotin labeling of proteins in the cell lysates which are then compared to FP-biotin labeling of ¹⁵N-isotope-labeled Mtb proteins from vehicle control (or less active 1,2,3-triazole urea analog) treated cultures to identify differences in serine hydrolases of streptavidin-affinity purified proteomes. The targets of JCP276, a 7-urea chloroisocoumarin electrophile with selective inhibition of Mtb growth compared to other bacteria and a wide selectivity window above mammalian cells was similarly identified by competitive ABP.¹⁹ Biotin-FP revealed differences in serine hydrolase labeling in the presence or absence of this electrophile and competition with clickable JCP276 analog (BMB034, Figure 1) defined a set of multiple potential targets mainly in lipid metabolism which was further corroborated by lipidomic analyses.

Antibiotic-based ABPs and affinity probes

Tetrahydrolipstatin (THL), an FDA-approved anti-obesity drug acting as a covalent inhibitor of pancreatic lipases, was used for the design of ABP to identify new targets due to its antitumor and antimicrobial activities.^13,20 Lipid metabolism is critical for the pathogenesis of Mtb and to identify key players in this pathway, protein profiling using a THL-based ABP (Figure 1) was performed in cell lysates of Mycobacterium bovis BCG leading to the specific labeling of 14 targets including the lipid esterases TesA and LipH.²¹ Based on antimicrobial activity of another mammalian lipase inhibitor, lalistat, in depth target analysis was performed with an alkynylated lalistat ABP (Figure 1) and revealed a cluster of 20 hydrolytic proteins including members of the lipase family.²²

Another good example of ABP design was EZ120, which was selected as a hit from a panel of β-lactones for antimycobacterial activity. To identify molecular targets of EZ120, alkyne-modified probe, EZ120P (Figure 1), was used for activity-based protein profiling paired with metabolic labelling studies, and then confirmed that Pks13 and Ag85 serine hydrolases as major targets.²³

β-Lactam antibiotics including the cephalosporins and carbapenems are broad-spectrum drugs that inhibit cell wall biosynthesis by covalently modifying penicillin-binding proteins (PBPs) and l,d-transpeptidases (LDTs) involved in peptidoglycan (PG) layer biogenesis of bacterial cell wall. Although many β-lactams have limited clinical utility in treating TB with the exception of carbapenems, their ability to form covalent adducts with PBPs and LDTs has been used to perform protein profiling to identify PG-modifying enzymes in Mtb. To identify active PBPs, LDTs, and β-lactamases in Mtb, three different ABPs including aztreonam-Cy5, cephalexin-Cy5, and meropenem-Cy5 selected from three classes of β-lactams, monobactams, cephalosporins, and carbapenems respectively, were synthesized by reacting the alkyne probe of each drug with sulfocyanine5 (Cy5) azide.²⁴ One noteworthy discovery of this study was that they detected significant changes in PG-modifying protein expression between dormant and actively replicating cells using meropenem-Cy5 (Figure 1).²⁴ This finding supports the previous results of transcriptional analyses that indicated dynamic regulation of expression of PBPs, LDTs, and other PG-modifying enzymes during exposure of pathogens to environmental changes. This was the first study that demonstrated differences in PBPs and LDTs as a function of the replicative status of Mtb.

In a different application of β-lactam antibiotics, cephalosporins were used for the design of probes to label single cells of the pathogen. The resistance of Mtb to β-lactam antibiotics is largely due to a bacterially produced β-lactamase (BlaC) that hydrolyzes the β-lactam ring. BlaC is an Ambler class A β-lactamase that is highly conserved in Mtb clinical isolates. This enzyme has been used to engineer chemical probes for TB diagnostics or prodrugs that release active moieties upon BlaC hydrolysis. Highly specific cephalosporin-based fluorogenic probes have been developed for BlaC-mediated detection of Mtb. Cephalosporin-based molecules bearing a near-infrared (IR) fluorophore quenched by fluorescence resonance energy transfer through a cleavable quencher have shown considerable promise. BlaC-mediated hydrolysis releases the fluorophore with resultant dequenching allowing detection of as few as 500 Mtb cells. Most importantly, this wavelength penetrates tissue effectively allowing detection of Mtb in infected mice so that pulmonary infection can be monitored in real time in live animals.²⁵

Another series of cephalosporin-based fluorogenic probes (CDGs, Figure 1) were developed with high selectivity to BlaC compared with the TEM-1 β-lactamase that is the most common β-lactamase in Gram-negative bacteria. This enables the successful detection of live Mtb in human sputum clinical samples without interference from competing microbes.^26,27 CDG-DNB3 (Figure 1), published in 2018, is a dual-targeting fluorogenic probe with cephalosporin core.²⁸ CDG-DNB3 consists of three functional units, a cephalosporin core as a BlaC-sensing unit, a caged fluorescent reporter, and a dinitrobenzene (DNB) moiety as a decaprenylphosphoryl-β-D-ribose 2′-epimerase (DprE1)-binding unit for signal trapping. Since DprE1 is a critical enzyme for mycobacterial cell wall biosynthesis and reduction of the nitro groups of DNB forms a covalent complex in the active site, DprE1-mediated signal retention made it possible to image BCG phagocytosis in real time as well as Mtb in sputum samples.²⁸

DprE1-mediated labeling of Mycobacteria was also realized with a chemical probe based upon a specific inhibitor, an analog of the 8-nitro-benzothiazinones (BTZ).²⁹ BTZ are highly potent bactericidal inhibitors of DprE1. PBTZ169, the most advanced analogue of BTZ series, is currently in clinical trials.³⁰ Taking advantage of the binding specificity of BTZ to DprE1, Cole et. al., confirmed the subcellular localization of DprE1 in Mtb treated with BTZ-TAMRA (Figure 1).³¹ After structural modification of BTZ-based probes, JN108 (Figure 1) was created as the most efficient derivative of these fluorescent imaging probes, which allowed efficient labeling of its target DprE1 in vitro and at the poles of live Mtb in culture.³²

The Mtb enzyme known as “Hydrolase Important for Pathogenesis” 1 (Hip1) is a cell surface-associated serine protease that promotes pathogenesis by impairing host immune responses. Due to its localization and immunomodulatory functions, Hip1 has been considered as a potential drug target.³³ The Bogyo group previously developed CSL173, a selective fluorogenic substrate probe of Hip1, by positional scanning-synthetic combinatorial libraries (PS-SCL) and multiplex substrate profiling by mass spectrometry (MSP-MS). They identified novel irreversible inhibitors of Hip1 and designed ABPs based on these structures using the fluorogenic substrate assay as a readout.³⁴ Enabled by this search for inhibitors as drug candidates, they reported an enzyme-based chemical probe, FLASH, that contained a Hip1 substrate based on CSL173 with a p-amino-benzyl-alcohol self-eliminating linker and a phenoxy-dioxetane luminophore. Upon enzymatic cleavage, the aniline linker underwent spontaneous elimination followed by release of the activated phenoxy-dioxetane moiety. FLASH (Figure 1) was reported to have promising quick and easy diagnostic properties in quantitative measurement of live Mtb cells in culture and human sputum samples including differentiation of live from dead cells.³⁵

Other strategies

7-Hydroxy-9H-(1,3-dichloro-9,9-dimethylacridin-2-one) sulfate (DDAO-sulfate, Figure 1) is an enzyme-activated fluorogenic probe commonly used for the determination of specific enzymatic activity in native protein gels.³⁶ This probe is hydrolyzed to DDAO by sulfatases, resulting in red-shifted excitation and emission with a substantial increase in quantum yield. Since sulfatases are conserved across mycobacterial genomes, this probe has been used for the rapid detection of sulfatases in mycobacterial lysates.³⁶

An inhibitor, Sal-AMS, based on the structure of the acyl-adenylate intermediate Sal-AMP formed by the MbtA adenylating enzyme of mycobactin biosynthesis was further modified with a photoaffinity moiety to generate a probe that could be used to understand the expression of the MbtA target as well as off-targets in lysates of Mtb.³⁷ This bisubstrate probe (Figure 1) mimicking an enzyme intermediate could serve as a template for the development of probes for other acyl-adenylating enzymes in mycobacteria.

Chemical probes targeting the mycomembrane.

Trehalose-based chemical probes

The Mtb cell envelope is comprised of an outer lipid layer (consisting of very long chain fatty acids also termed mycolic acids), underlaid by a cross-linked network of PG that is coupled to highly branched chain arabinogalactan (AG) (Figure 2).^38,39 Trehalose, a non-mammalian disaccharide, is synthesized by Mtb. It is non-covalently associated with the outer leaflet of the Mtb cell envelope along with the glycolipids trehalose dimycolate (TDM) and trehalose monomycolate (TMM).³⁹ Trehalose is anchored into the Mtb cell wall as TMM or TDM by the three enzymes Ag85A, Ag85B and Ag85C. These Ag85A-C enzymes catalyze the transesterification reaction between two units of TMM to yield TDM and trehalose.⁴⁰ These glycolipids have been the target for designing chemical tools to investigate mycomembrane biosynthesis and evaluate trehalose metabolism by exploiting metabolic pathways such as SugABC – LpqY and Ag85.^40,41

Figure 2. A summary of mycomembrane targeting probes incorporation into the Mtb cell envelope.

A wide variety of such probes have been incorporated into the mycomembrane of Mtb. Among the various strategies employed for probe incorporation, trehalose-based probes have been widely used. Trehalose in Mtb is found in the outer portion of the cell envelope as its corresponding mycolate lipids. Trehalose and its analogues are processed by Ag85 family enzymes and can be metabolically incorporated into outer membrane of Mtb for wider applications. (A) Cell wall schematic showing only the portions relevant for labeling and excludes many other molecules present in the mycobacterial outer membrane (B) Fluorescent probes based on modified trehalose e.g., FITC-Tre, FIT-Tre, DMN-Tre and 3HC-3-Tre. (C) azido-trehalose(s), fluoro-deoxy trehalose(s) and other alkyne functionalized trehalose probes for interrogation of the mycomembrane and detection of live Mtb in various samples, including potential positron emission tomography integrated with computed tomography (PET-CT) imaging agents and probes for protein profiling (D) fluorogenic probes provide the potential to turn-on after metabolism by Mtb i.e. FRET-TDM & QTF. (E and F) Probes based on unnatural d- amino acids and azido sugars have been widely applied to label mycobacterial peptidoglycan. (G) Bidentate boronic acids also bind to the lipoglycan components of mycomembrane.

Trehalose plays a key role in Mtb’s metabolism and cellular function. Its absence in human biology creates a ‘blank’ (or low) biochemical background activity that could allow a valuable role in a) specific TB diagnosis, b) an anti-TB drug discovery, and c) identifying novel protein-protein interactions. Trehalose is synthesized by Mtb through three independent pathways and the most well-known among them is via the trehalose-6-phosphate synthase and phosphatase pathway.⁴² An analogue of this pathway has been exploited to synthesize potential imaging probes for TB.⁴³ These probes include trehalose-based fluorescently-labeled imaging agents and radioactive ¹⁸F-labeled PET imaging agents.⁴³

Backus et al. designed the first fluorescent trehalose probe, fluorescein isothiocyanate (FITC) trehalose and reported the specific one-step labeling of Mtb both in vitro and in macrophages.⁴³ They showed that the FITC-trehalose incorporation was reduced in a deletion mutant of one of the three isoforms of Ag85. This mutant incorporated 30% less FITC-trehalose as compared to wild type.⁴³ They synthesized a broad library of trehalose analogues and showed that many of these analogues were efficient substrates of the Ag85 enzymes.⁴³ This further demonstrated that Ag85 proteins could process a variety of trehalose analogues, and hence could be used to introduce a variety of reporter molecules in the Mtb cell wall. While this probe, and rather more importantly the methodology, opened new research avenues, it had some limitations. For example, this probe showed poor efficiency when labeling different actinobacterial species other than Mtb, such as Mycobacterium smegmatis and Corynebacterium glutamicum for labeling glycolipids.⁴⁴

Rodriguez-Rivera et al. expanded this library by introducing regioisomeric versions of fluorescein trehalose analogues. They speculated that the anomeric methyl group in FITC-trehalose may compromise the processing of this probe by the Ag85 complex.⁴⁴ They used the fluorescein trehalose analogues with slight modification to FITC-trehalose and reported that these molecules are recognized by mycolyltransferases of diverse mycobacterial species. They reasoned that the reagents used in this work showed improved metabolic efficiency compared to FITC-trehalose.⁴⁴ These analogues showed strong metabolic labeling with the highest labeling observed with 6-fluorescein trehalose (6-FITre). This probe enabled the demonstration that the fluidity of the mycomembrane across Actinobacteria decreased with increasing chain length of mycolic acids. This approach was also limited as unincorporated probe needed to be removed to eliminate background fluorescence.^44,45

In order to address the obstacle associated with FITre, Kamariza et al. hypothesized that a trehalose probe whose fluorescence signal is specifically activated by metabolic incorporation into the mycomembrane would overcome this limitation.⁴⁶ They developed 4-N,N-dimethylamino-1,8-naphthalimide (DMN), conjugated to trehalose, collectively called DMN-Tre for the detection of M. tuberculosis from the sputum samples of TB patients. They showed that this probe selectively labels live Mtb in TB-positive sputum samples and observed reduction in the fluorescence signal when the Mtb was exposed to front line TB drugs.⁴⁶

While DMN-Tre is an attractive probe for point of care TB diagnosis, it has several limitations, such as lower brightness and low quantum yield as well as the need to image for one hour to obtain an adequate fluorescence signal.⁴⁷ To address these limitations, Kamariza et al. developed a trehalose probe based on 3-hydroxychromone (3HC-3) dye.⁴⁷ This dye when conjugated to trehalose, showed 10-fold higher fluorescence intensity to that of DMN-Tre. In addition, higher signal to noise ratio allowed the rapid detection of labeled Mtb cells within 10 min as opposed to one hour with DMN-Tre and this may well present a major step forward towards point of care TB diagnosis in low resource settings.⁴⁷ Dinkele and coworkers used metabolic incorporation of DMN-Tre into cell walls to characterize the growth kinetics of Mtb captured from patient bioaerosols by serial imaging of the labeled cells in arrayed nanowells.⁴⁸

Hodges et al. have developed a quencher-trehalose-fluorophore (QTF), a fluorogenic probe for imaging mycolyltransferase, Ag85 activity in mycobacteria.⁴⁹ This probe is an analogue of the mycolyltransferase donor, TMM. It is hydrolyzed by mycolyltransferases to provide a real-time fluorescent signal of these enzymes’ activity and mycomembrane assembly. Unlike several other fluorescent agents, this new probe does not require washing steps and reports on mycobacterial cell wall assembly in live bacteria in real time.⁴⁹ The Swarts group⁵⁰ synthesized a FRET-TDM, a fluorescence-quenched analogue of TDM and evaluated its performance in vitro and in vivo. TDM is broken down by the mycobacteria-specific enzyme TDM hydrolase (Tdmh).⁵⁰ This study was trying to address the lack of methods available to probe Tdmh activity. The probe showed high quenching efficiency (98.7%) and good hydrolytic stability. Comparison of the kinetic parameters of purified Tdmh and Ag85C showed that the probe binds stronger to Ag85C but Tdmh catalyzes its hydrolysis more rapidly and activates the probe more efficiently.⁵⁰ In vitro cell assays showed that the probe was activated by live mycobacterial cells including M. smegmatis and Mtb. The protein profiling study using native gel fluorescence assay in whole cell lysates from M. smegmatis showed several unknown bands pointing towards the possibility of novel hydrolases that are yet to be explored.⁵⁰ To interrogate individual cells of Mtb within a macrophage phagosome, a two-stage activation probe consisting of a cephalosporin linked green trehalose (CDG-Tre) has also been used to label live M. bovis BCG cells.⁵¹ This probe allows efficient labeling of these bacteria at very low concentration of probe (2 μM) via both the BlaC and Ag85 enzymes.⁵¹

Metabolic labeling of azido-sugars followed by post-labeling visualization has also been used for the incorporation of trehalose analogues into the Mtb cell envelope by the Ag85 complex. Swarts et al. reported the synthesis and metabolic labeling of glycolipids by azide-modified trehalose (TreAz) analogues in live mycobacteria.⁵² After Cu-click reaction and subsequent visualization under fluorescent microscopy, they observed azide specific fluorescence in mycobacteria. In addition, they also confirmed the TreAz-labeled glycolipids by high resolution mass spectrometry. Flow cytometry based assays of Mtb and M. bovis BCG showed TreAz-dependent fluorescence and this labeling was significantly reduced in the M. bovis BCG ΔpqY-sugC mutant.⁵² Swarts and coworkers synthesized O-AlkTMM and N-AlkTMM as a reporter for in situ labeling of arabinogalactan linked mycolates and trehalose glycolipids respectively. By altering the linker, they controlled the probe incorporation mechanism and labeling target, with amide linked N-AlkTMM labeling TDM and ester linked O-AlkTMM labeling arabinogalactan-linked mycolate (AGM) and TDM.⁵³ To further extend the scope of this work, Pohane et al. developed a bifunctional probe 6-Azido-6-deoxy-6′-O-(6′-heptynoyl)-α,α-D-trehalose (O-AzAlkTMM).⁵⁴ Through a series of experiments they demonstrated that this bifunctional probe first labels the TDM/AGM with an alkyne via Ag85, then labels TMM using the azide functionality, hence allowing for visualizing multiple steps of mycomembrane metabolism.⁵⁴

To determine the protein content of the mycomembrane, Swarts and coworkers extended their previous work and developed bifunctional N- and O-x-AlkTMM-C15 probes. Both these probes have a mycomembrane-targeting TMM moiety which has a photoactivable diazirine for covalent cross-linking to lipid binding proteins and a clickable alkyne to facilitate in-gel proteomic target identification.⁵⁵

In a recently concluded study, we showed that one of the simplest ¹⁸F-analogs of trehalose 2-[¹⁸F]fluoro-2-deoxy-D-trehalose ([¹⁸F]FDT) can be used as an in vivo TB reporter using a chemoenzymatic synthesis from readily available 2-[¹⁸F]fluoro-2-deoxy-D-glucose ([¹⁸F]FDG).⁵⁶ Toxicity studies in rats and dogs model show that the probe is safe. Extensive pre-clinical studies in non-human primates infected with TB model showed that the probe was specifically taken up by TB lesions in the lung. In conjunction with [¹⁸F]FDG, we also used this tracer to monitor the response to treatment in marmosets. While [¹⁸F]FDG showed mixed, refractory and inconsistent uptake, [¹⁸F]FDT, showed a significant reduction in the standardized uptake value (SUV) and a lower total tracer uptake consistent with treatment outcome, suggestive of a more accurate probe for monitoring disease burden.⁵⁶

Probes targeting other mycomembrane sugars

In addition to trehalose, d-arabinose is also an essential sugar component of the Mtb cell membrane through the polymeric arabinan which bridges the PG and the galactan. Arabinose occurs in two isomeric forms: d–arabinopyranose and d- arabinofuranose.⁵⁷ Kolbe et al. synthesized azide-labeled arabinose derivatives i.e. 3-azido-3-deoxy-d-arabinose (3AraAz), 3-azido-3-deoxy-d-ribose (3RiboAz) and 5-azido-5-deoxy-d-arabinofuranose (5AraAz) as substrates for metabolic labeling of Mtb clinical isolates. They observed similar labeling using 3AraAz and 3RiboAz, however, 5AraAz showed labeling 4 times higher to that of corresponding 3-azido derivatives.⁵⁸ Incorporation of 5AraAz was unexpected and further experiments failed to conclusively elucidate the exact mechanism by which 5AraAz is metabolized. This, however, points to the presence of an unknown carbohydrate biosynthetic pathway in Mtb that still needs to be fully elucidated.⁵⁸ When they tested azido pentoses against clinical isolates of the Mtb complex, they observed that the labeling intensity was sugar and bacterial strain dependent.⁵⁸

A recent study from the Swarts group has developed synthetic azido inositol (InoAz) analogues for metabolic labeling of a cell envelope glycolipid, phosphatidylinositol mannoside (PIM) and their downstream products lipomannan (LM) and lipoarabinomannan (LAM).⁵⁹ These probes when coupled with biorthogonal click chemistry, will be valuable tools to investigate the biosynthetic pathways of PIM, LM and LAM.⁵⁹

Boronic acids, with their high affinity for diols, have been used widely to detect many carbohydrates. Based on the antimycobacterial activity of extracellular cell envelope glycan-targeting boronic acid agents,^60,61 dimeric boronic acid probes, B2-N₃ and B2-alkyne, were designed from multimeric glycan-chelating dimeric boronic acid with a linker containing a clickable moiety which can be linked to a dye of choice as new imaging tool. Unlike the traditional approaches of targeting specific intracellular pathways, they targeted the extracellular mycobacterial cell surface which eliminates the mycomembrane permeability barrier.⁶²

Probes of the outer mycomembrane lipid leaflet

The Mtb cell envelope contains an outer non-covalently associated lipid layer that contributes to membrane impermeability and provides protection against environmental stresses, including the host immune response.⁶³ Extractable lipids make almost 40% of the cell envelope weight and among those lipids phthiocerol dimycocerosate (PDIM, Figure 2) makes up 46% of the total lipid content.^1,64 Despite this abundance few efforts have been made to target this important interface at the host-pathogen contact zone. Bertozzi and coworkers synthesized a clickable analogue of PDIM by converting the methyl ether of PDIM to an alkyl halide with trimethylsilyl iodide.⁶⁵ Addition of sodium azide provided clickable azido-DIM. With this approach, they generated bacteria with chemically functionalized PDIM without affecting its pathogenic function.⁶⁵

Unnatural Amino Acid Based Probes

Besides the use of carbohydrate and lipid based chemical probes, unnatural amino acids have also been used to visualize PG biosynthesis or protein–based bacterial surface labeling in Mtb as well as other bacterial species.^66-68 Bertozzi and coworkers have employed d-alanine analogues bearing biorthogonal functional probes for metabolic labeling of PG both in vitro and during macrophage infection, primarily in L. monocytogenes.⁶⁹ They compared the alkDala labelling of WT L. monocytogenes to that of an d-alanine auxotroph which showed stronger labelling. When they added the exogenous d-alanine, it suppressed azDala labelling, suggesting that both alkDala and azDala access the same metabolic pathway as the natural d-alanine.⁶⁹ To demonstrate the selective labelling of d-amino acids inside host cells, they infected macrophages with L. monocytogenes and after removing extracellular bacteria and subsequent treatment of cells with alkDala followed by click reaction, they observed alkyne dependent signal variation dependent on alkDala incubation time.⁶⁹ While the method developed in this study is based on L. monocytogenes, d-alanine analogues incorporate into various species of bacteria and can be used to investigate PG dynamics both in vitro and in vivo.⁶⁹ A similar methodology has been successfully used in several other bacterial species with d-amino acids of various sizes and functionalities.⁷⁰ A further study from the same research group developed a library of fluorogenic azido Si-rhodamine probes in conjunction with cyclooctyne d-amino acids to visualize bacterial PG with low background.⁷¹ In a recent study, a new fluorescent imaging probe was designed to bind to the terminal amino acid dimer d-Ala–d–Ala present in the PG precursor.⁷² This probe allows the labeling of the PG of the Mtb cell wall and was able to detect Mtb in both actively replicating and non-replicating bacteria. This probe was used to screen Mtb genes that are critical for cell invasion in vitro.⁷²

The third position of the peptide stem is variable depending on bacterial species but consists of meso-diaminopimelate (mDAP) in mycobacteria, has also been labeled with modified analogs. Given the importance of mDAP-containing peptides in immune-recognition and potential resuscitation of non-replicating bacteria, labeling of this bacterial-specific metabolite offers opportunities to interrogate aspects of PG biosynthesis and turnover that may not be addressed by labeling of the distal residues. Selenolanthionine (SeLAN) serves as an mDAP bioisostere and fluorescently labeled SeLAN-containing tripeptides which can only serve as acyl acceptors and therefore retained during crosslinking in the PG were incorporated by mycobacteria into the PG.⁷³ The SeLAN metabolite could also be directly incorporated into the PG although an mDAP auxotroph is required for this limiting wider applicability.

In a recent study, Liu et al. developed a novel assay, PG Accessibility Click – Mediated Assessment (PAC – MAN) to quantify the penetration of small molecules across the outer mycomembrane.⁷⁴ The design of the assay was based on a D-amino acid modified with a biorthogonal handle via the addition of dibenzo cyclooctyne (DBCO). They synthesized DBCO conjugated d-Dap (d-DapD) and a DBCO conjugated tetrapeptide (TetD) and tested the efficiency of labeling of live cells. Cells were washed to remove unlabeled probe molecules and reacted with azide-modified fluorescein and measured using flow cytometry. PG labeling with TetD showed highest fluorescence increase when compared with vehicle and more than double with single amino acid (d-DapD), making TetD the primary PG tagging method. By testing with a panel of azide-tagged molecules they demonstrated that the assay workflow was robust and compatible with screening in Mtb.⁷⁴ Genetically encoded methods to measure compound permeability have also been developed (vide infra).

Genetically Encoded Probes

In addition to small molecule-based techniques previously described, many tools have been developed that make use of genetically encoded probes. These have been used for many purposes, including imaging, querying mycobacterial microenvironments and metabolic parameters, investigating protein-protein interactions and protein localization within the mycobacterial cell, and investigating pathways of susceptibility and resistance to various antimycobacterial drugs. In this section, we focus on recent chemical biology studies making use of genetically encoded probes in mycobacteria. We discuss fluorescent proteins, the photocrosslinkable unnatural amino acid pBpa, and proximity labeling with biotin or clickable tyramide derivatives.

The ability to express FPs stably without concerns about genetic modifications elsewhere in the chromosome is crucial to the use of FP-expressing Mtb strains for imaging studies in infection models that require weeks or months. Episomal plasmids and integrating vectors can be lost if antibiotic selection is not maintained. Loss of integrating vectors can occur due to ongoing expression of the integrase, which can catalyze not only integration but also excision of the vector.^75,76 One previously developed technique to improve the stability of integrating vectors is to first transform the mycobacterium with an integrating plasmid in which the integrase gene is floxed, then to transform with a second episomal plasmid expressing Cre that causes the integrase to be excised, followed by a SacB-sucrose counterselection step to remove the episomal plasmid.⁷⁶ Although effective, the multiple steps involved are laborious and time-intensive. Kolbe et al. developed a series of integrating plasmid backbones that use the alternate technique of expressing the integrase in trans on a suicide plasmid that is electroporated together with the payload.⁷⁷ This can create in a single step an exceptionally stable insertion of the genetic payload without a requirement for ongoing antibiotic selection. One downside of integrating vectors is that they are single-copy, and thus typically result in lower expression levels of their payload than do multi-copy episomal plasmids. Kolbe et al. additionally addressed this issue by optimizing the ribosomal binding site of the strong constitutive promoter pLeft.⁷⁷ This allowed for stable integrating plasmids to express fluorescent proteins (FPs) with brightness at least as good as is achieved by traditional high-copy number episomal plasmids using typical promoters.⁷⁷

Fluorescent protein-based techniques

Fluorescent protein (FP) based techniques largely fall into several groups. One large group consists of the expression of one or more FPs either constitutively or under the control of a promoter or riboswitch which responds to a known stimulus. A related technique is to fuse an FP to another protein of known function. This allows the level of fluorescence to be interpreted as indicative of the level of expression of the fused protein,⁷⁸ or even for localization of the fused protein to be visualized via fluorescence microscopy.⁷⁹ Another technique is to construct a FP for which fluorescence varies in intensity or color based on local conditions, such as the concentration of a metabolite of interest.^80-82 Genetically encoded fluorescent proteins are particularly interesting in that they enable analysis at the level of single cells via fluorescence microscopy or flow cytometry, allowing an appreciation for the heterogeneity of cellular states and microenvironments. A more in depth discussion of issues relating to use of fluorescent proteins in mycobacteria can be found in a recent review by MacGilvary and Tan.⁸³

There is a large and growing library of promoters and riboswitches used to drive expression of fluorescent proteins as reporters of a variety of conditions. This includes reporters responsive to extracellular pH,^84,85 cell envelope stress,^86,87 cholesterol and propionate,⁸⁸ hypoxia and nitric oxide(NO),⁸⁴ identity as a persister cell,⁸⁹ and electron transport chain stress.⁹⁰ More developed FP reporters include one using the pfurA promoter to report oxidative stress,⁹¹ and one using the magnesium-sensitive riboswitch upstream of the Mtb gene rv1806 to report intracellular magnesium concentration.⁷⁷ These constructs all result in expression of an FP being driven by the input of interest. They are often paired with constitutive expression of another FP with non-overlapping excitation/emission wavelengths, to control for parameters such as cell size, density, and global expression level of proteins. In this case, the readout will typically be the ratio of the fluorescence of the inducible FP to the constitutive FP.

One powerful example of a FP fusion with an endogenous protein is the single stranded binding protein GFP fusion, SSB-GFP.⁷⁸ The single stranded binding protein binds to single stranded DNA at the replication fork in a replicating cell. This allows for identification of mycobacterial cells which are actively replicating their DNA. Another recent example is of Wag31-GFP.⁹² Wag31 binds to mycobacterial cell poles, where it acts as a scaffold for recruitment of cell wall biosynthesis enzymes. Time-lapse fluorescence microscopy of Wag31-GFP in conjunction with atomic force microscopy revealed that cells have biphasic growth consisting of a lag phase followed by a phase of rapid polar growth and elongation.

The third category of FP reporters consists of FPs whose fluorescence varies depending on the local environment. Previously developed examples of this include Mrx1-roGFP2, which fuses a redox-sensitive GFP with Mtb mycoredoxin-1 to produce a ratiometric reporter of the redox status of the intracellular mycothiol pool,⁸⁰ and Peredox-mCherry, which fuses a circularly permuted T-Sapphire with the NADH:NAD⁺ sensor Rex protein domain from Thermus aquaticus, producing a ratiometric reporter of the NADH/NAD⁺ ratio.^81,82

Akela and Kumar adapted the ATP/ADP reporter PercevarHR (PHR) for use in mycobacteria.⁹³ PHR was developed by fusing a circularly permuted mVenus YFP with Glnk1 from Methanococcus jannaschii with subsequent mutagenesis for optimization.^94,95 The ratio of fluorescence excitation peaks at 500 and 420 nm varies with the ATP:ADP ratio. This needed to be adapted for use in mycobacteria, which have significant autofluorescence at excitation wavelengths of 420 nm.⁹³ By fusing PHR with the red fluorescent protein mCherry, it became possible to obtain a fluorescent ratiometric readout using the nucleotide sensitive PHR domain at 500nm and the mCherry domain at 587nm in mycobacteria. This was used to obtain information on ATP:ADP ratios in individual Mtb cells in infected macrophages using fluorescence microscopy and flow cytometry. Significant heterogeneity was seen, with much of it being driven by differences in the energy balance of Mtb cells in different macrophage compartments.⁹³

Several recent investigations have used genetically encoded fluorescent sensors in combination with each other or with other techniques to powerful effect. Lavin and Tam combined fluorescent reporters to shed significant light on the heterogeneity in the microenvironment that Mtb experiences in and around a caseous necrotic lesion in a C3HeB/FeJ mouse model of infection.⁹⁶ They used strains expressing mCherry constitutively alongside either GFP under the pH and [Cl⁻] sensitive rv2390c’ promoter, SSB-GFP (described above), or monomeric Kusabira Orange (mKO) under a tetracycline inducible promoter. mKO expression allows a readout of access to that specific compartment by a tetracycline. By using fluorescence microscopy of lung tissue sections, they were thus able to assess on a single-cell level the acidity of each mycobacterium’s local microenvironment, its replication status, or its ability to express new proteins in response to stimulus. They found that Mtb in the caseous necrotic core (Figure 3) experienced a more neutral pH and lower [Cl⁻], and had a higher proportion of replicating cells than the Mtb in the macrophage dominated cuff surrounding the lesion. Although doxycycline did not penetrate all the way to the core of the caseous necrotic core, it was apparent that Mtb cells in the outer part of the necrotic core were better able to respond to doxycycline by producing mKO than were Mtb cells in the cuff, suggesting greater transcriptional and translational capability. Differential effects of rifampin, isoniazid, and pyrazinamide on cells across different regions of the lesion were also characterized.⁹⁶

Figure 3. The utility of genetically encoded probes and advanced imaging in studying TB disease.

A. Both chemical and genetically encoded probes can be used simultaneously to probe the microenvironment of Mtb cells growing in various cells including various sub-compartments within macrophages. In the schematic two different phagosomes containing Mtb have been shown to be different by differential expression of environmentally sensitive protein fusions (for example, pH-responsive) combined with probes applied externally that are incorporated differentially (for example trehalose which is incorporated by replicating bacilli only). B. Use of real time positron emission tomography (PET) imaging using [¹⁸F]-fluorodeoxytrehalose to monitor disease progression in different mouse models of disease. C. [¹⁸F]-fluorodeoxyglucose PET-computed tomography (CT) scan of a TB patient showing the diversity of lesion types that are present. The feature at the top (red arrow) represents a cavity in the left apical region of the lung and the cartoon depicts the fibrotic wall of the cavity with some material remaining accompanied by an influx of neutrophils (blue cells with trilobed nuclei). The cartoon below shows an alternative lesion seen in this patient, a large granuloma with a caseous necrotic core. In both cartoons the Mtb bacilli are drawn in red and the blue, more spherical cells represent lymphocytes while the more elongated epitheloid cells are directly surrounding the necrotic core. Cavities are believed to result from necrosis of such lesions into the airways allowing rapid proliferation and dissemination.

Another recent example by Pisu et al. used Mtb expressing GFP under the hspX’ promoter driven by DosR activity alongside constitutive mCherry.⁹⁷ As DosR is activated by a variety of stresses, the GFP/mCherry fluorescence ratio was interpreted as a marker of mycobacterial stress in general. Flow cytometry was used to study immune cells from the lungs of infected mice using antibody-derived tags for surface marker expression, and correlations were found with mycobacterial stress. Both mouse immune cells and mycobacterial cells were subjected to single cell RNA sequencing (scRNA-seq). A rich picture emerged detailing the interactions between macrophage type and activation status, and bacterial stress and drug tolerance. Alveolar macrophages tended to cause higher GFP and thus higher mycobacterial stress than interstitial macrophages, and M1 activated macrophages more than M2. By sorting immune cells and subsequently treating with antibiotics, they were further able to demonstrate that high GFP Mtb is enriched in phenotypically drug tolerant persister cells.⁹⁷

The rapidly proliferating set of FP-based probes is a powerful, flexible, and facile resource. It allows measurement, at a single-cell level if desired, of a variety of parameters about the mycobacterial microenvironment or internal state both in vitro and in vivo. Its flexibility makes it well suited for use in combination with other non-genetically encoded techniques, especially at the intersection of fluorescence microscopy with other imaging techniques.

Genetically encoded incorporation of unnatural amino acids

Wang et al. evolved Methanococcus jannaschii orthogonal mutant tRNATyr/tyrosyl-tRNA synthetase pairs that could incorporate different unnatural amino acids at the amber codon in M. smegmatis and demonstrated the applicability of one of these pairs in Mtb both during growth in vitro and in macrophages.⁹⁸ Despite the high frequency of the amber stop codon in mycobacteria, the efficiency of unnatural amino acid incorporation was low which proved ideal since growth rates in Mtb were unaffected in the presence of these precursors allowing minimal interference with cellular metabolism during labelling studies. Touchette et al. utilized this system to study the mycobacterial lipoprotein LprG protein interaction network by incorporating the photocrosslinkable unnatural amino acid p-benzoyl-L-phenylalanine (pBpa) into the protein and FLAG tagging it.⁹⁹ After UV irradiation, pBpa residues cross-linked with neighboring proteins allowing identification of LprG binding partners by LC-MS/MS. This technique seems likely to be generalizable to investigate protein-protein interactions of many other proteins besides LprG.

Peroxidase mediated labeling

Ganapathy et al. demonstrated another technique involving genetic encoding of the peroxidase APEX2.¹⁰⁰ In the presence of hydrogen peroxide and phenol linked to a detection moiety, APEX2 generates phenoxyl radicals that covalently bind to nearby proteins. By targeting APEX to the cytosol or to the periplasm of M. smegmatis, proteins in each compartment can be selectively labeled. They initially used biotin-phenol to biotinylate proteins, however labeling was confounded by a high background of natively biotinylated proteins. They subsequently used tyramide azide or tyramide alkyne, which covalently labeled proteins and allowed for downstream Cu-catalyzed “click” chemistry. In this case, labeled proteins were reacted with fluorescein-conjugated alkyne, and visualized via in-gel fluorescence. This technique was later demonstrated to work in Mtb with a codon optimized APEX2 gene.¹⁰¹

Biotin ligase mediated labeling

Proximity-dependent protein modification has also been performed in mycobacteria by generating fusions of biotin ligase (BirA) to heparin-binding hemagglutinin to determine proteins that interact with this virulence factor and deduce the metabolic pathways that intersect with its function.¹⁰² Fusions with BirA that do not interfere with protein function allow BirA-mediated biotinylation of interacting partners upon biotin addition offering several advantages above APEX2-mediated methodology in that labeling is not transient but occurs during the duration of biotin exposure and does not require exposure to agents that can result in oxidative stress.

HaloTag-mediated measurement of compound permeability

An ingenious method termed BaCAPA, for Bacterial ChloroAlkane Penetration Assay, to measure compound permeation to the cytosol was developed by the Pires group wherein a modified bacterial haloalkane dehalogenase, HaloTag, is cytosolically expressed.¹⁰³ Treatment of the mycobacteria with a chloroalkane results in essentially irreversible covalent modification of the enzyme. The permeation of diverse molecules containing a chloroalkane group can be measured by subsequent competitive labelling of a fluorescently labeled chloroalkane with known permeation where lower labeling indicates lower levels of unlabeled HaloTag and thus higher permeation of the test molecule. This method was also used to measure permeation of molecules to the Mtb cytosol during growth in host macrophages.

Multiplexing chemical biology with recent advances in imaging and genome-scale genetics

Building topographical maps of granulomas.

Although in vitro and ex vivo tracking of mycobacterial growth and metabolism has benefited from the toolkit of metabolic labels and probes, in vivo tracking of bacterial physiology is more challenging. Even in animal models, the labeling of the pathogen with probes is hampered by restrictions such as appropriate PK/PD of the compounds especially at the site of disease, the granuloma. Nevertheless, major technological advances have accelerated our ability to interrogate pathogen and host during disease progression and at defined stages of disease. Multiomic analysis of TB granulomas entails integrating outputs from different modalities to generate a 3D image of host and pathogen transcriptome, proteome and lipidome. Marakalala and colleagues used laser-capture microdissection of tissue sections from different types of granulomas from rabbit lesions and TB patient lung resections to identify the proteome through LC-MS/MS of tryptic fragments of each microenvironment.¹⁰⁴ The proteome data was combined with phospholipid and eicosanoid mapping performed by matrix assisted laser desorption/ionization mass spectrometry imaging on tissue sections which revealed the geographic landscape of these molecules within the different granulomas. Finally, immunohistochemistry of the tissue sections allowed the inflammatory signatures to be mapped on top of the proteome, phospholipid and eicosanoid profiling data. To further incorporate dynamics of inflammatory status of the lesions during progression of disease, Gideon and coworkers performed serial positron emission tomography and computed tomography (PET-CT) imaging of Mtb-infected Cynomolgus Macaques which captures structural changes as well as glucose-uptake as a proxy for inflammation.¹⁰⁵ The serial PET-CT imaging was combined with post-necropsy analysis of single-cell RNA sequencing, single-cell FACS analysis of cell type and cytokine expression along with total and viable bacterial number analysis of individual granulomas to identify features that influence control of bacterial pathogenesis. This study correlated bacterial control for each type of granuloma instead of within the microenvironment of each lesion. In more recent work, multiplexed ion beam imaging by time-of-flight mass spectroscopy of TB patient tissue sections labeled with metal-conjugated antibodies was used to spatially map the types of immune cells combined with their functional states based on expression of proteins involved in key regulatory cascade such as IDO1 and PD-L1 expression within the granulomas.¹⁰⁶

Lewis et al. applied direct analyte-probed nanoextraction of different granulomas wherein 100-160 μm² sections of a granuloma section is extracted with solvent using a microscopically guided nanoinjector and aspirator followed by liquid chromatography-mass spectrometry analysis of the lipids and drugs to generate spatial resolution of drug and lipid species and concentration across lesions.¹⁰⁷ The same tissue sections were also used for particle-induced X-ray emission analysis to overlay elemental distribution maps on the images.

The structural resolution of the above imaging approaches can be further refined to the subcellular level by nanoscale secondary ion mass spectrometry (nanoSIMS) where a focused primary ion source is used to elicit secondary ions such as ¹²C¹⁴N⁻, ³²S⁻, ³¹P⁻, and ⁷⁹Br⁻ corresponding to protein, lipid (due to lipid-bound stain), nucleic acid/ polyphosphate and bromine containing compounds such as bedaquiline, respectively, which are detected by a mass spectrometer. The superposition of nanoSIMS with other imaging approaches such as light microscopy, confocal fluorescence microscopy, and electron microscopy (Correlative light electron microscopy; CLEM) allows the integration and correlation of data from all imaging modalities to provide a detailed 3-dimensional map of subcellular structures, proteins, host cell features, infecting pathogen and drugs. CLEM was used to track the replication of fluorescently labeled Mtb in the cytosol and membrane-associated compartments of functionally diverse human macrophages where staining with fluorescent dyes that measured membrane permeability and nuclear DNA along with fluorescently labeled Mtb in live as well as subsequently fixed cells allowed the correlation of the electron microscopy images of the same cells.¹⁰⁸ The fluorescent readout is not limited to fluorescent stains during confocal live cell imaging but can be further interrogated on fixed cells using labeled antibodies to proteins of interest and laser-scanning confocal fixed cell imaging prior to transmission electron microscopy. Combination of CLEM with nanoSIMS (CLEIM) provided a detailed dynamic image of the accumulation of bedaquiline in macrophage lipid droplets from where the drug transferred to the Mtb cells via host cell organelles.¹⁰⁹ Similar approaches were used to define the localization of pyrazinamide/pyrazinoic acid as detected by nanoSIMS in Mtb residing in acidified versus neutral pH compartments in host macrophages.¹¹⁰ CLEIM can also be used to explore antibiotic action in heterogeneous host cell and infecting pathogen populations and was used to image the disruption of bacterial homeostasis in infected macrophages as a result in cytosolic acidification during pyrazinoic acid formation in Mtb at a subcellular level in macrophages.¹¹¹ CLEIM has even been applied to image host, pathogen, and drug distribution in animal tissues.¹¹² In this approach coined CLEIMiT, sections from lungs from Mtb-infected animals were visualized followed by 3D fluorescence imaging of the sections to detect fluorescently labeled Mtb and nuclei and lipid droplets through fluorescent stains yielding a 3D map of distribution of these components in 30 μm slices followed by micro-computed tomography (μCT) to yield structural information on 100 μm sections. Finally scanning electron microscopy and nanoSIMs on granuloma-containing sections identified by the prior 2 techniques yielded ultrastructural and drug/ macromolecular distribution on 500 nm sections. The data obtained from the mCT imaging allowed precise correlation of the 3D fluorescence microscopy with the electron microscopy and nanoSIMS data.

Imaging of protein secretion and metabolic activity at the single cell level

Understanding cell responses to environmental and/ or stress-related stimuli often rely on measurement of protein expression at a cell population level. Tools to analyze protein secretion in response to stimuli within single mammalian cells include the ELISpot and FluoroSpot assays. ELISpot provides low-resolution images of secreted proteins of interest localized around a responsive cell while FlouroDot gives high-resolution images including spatial information about directional secretion at the single cell level.¹¹³ The FluoroDot assay achieves a high level of resolution and sensitivity by plasmon-enhanced fluorescent labels where the emission of a fluorophore is amplified when near gold or silver nanoparticles (plasmonic nanostructures) allowing single-nanoparticle detection. The FluoroDot assay requires a glass substrate with bound antibodies to capture secreted proteins of interest on which adherent cells are seeded. After stimulation, protein secretion is detected with a plasmonic-fluor conjugated antibody and a digital readout with the multiplexed detection of secreted proteins enabled by a palette of plasmonic fluor conjugates. The FluoroDot assay has also been used to show cellular heterogeneity in cytokine expression in co-culture experiments of Mtb-infected macrophages and T cells.

Labeling of cells with stable isotopically modified substrates may reveal biases in incorporation depending on metabolic status. Nanoscale resolution secondary ion mass spectrometry (NanoSIMS) or Raman microspectroscopy can provide a chemical fingerprint of a cell. Growth of cells in heavy water (D₂O) results in incorporation of deuterium into macromolecules, a large fraction being fixed during reductive hydrogenation in lipid biosynthesis, allowing metabolically active cells to be distinguished from metabolically inert cells by the intensity of the C-H compared to C-D peaks. D₂O Raman Imaging Spectroscopy was utilized by Ueno et al. to understand the metabolic status of single cells in a mycobacterial population that persisted during exposure to cidal concentrations of rifampicin.¹¹⁴

CRISPRi

The past several years have seen an explosion in work done with the clustered regularly interspaced short palindromic repeats – CRISPR associated proteins (CRISPR-Cas) system in mycobacteria. Previous work in other bacteria mostly used the type IIA system from Streptococcus pyogenes. By including a single guide RNA (sgRNA), with complementarity to a target piece of DNA, S. pyogenes Cas9 protein can be guided to the target and cause a double strand break in it.¹¹⁵ Mutating Cas9 to remove its nuclease activity results in nuclease-dead Cas9 (dCas9_Spy), which does not degrade DNA, but rather blocks transcription. This technique is called CRISPR interference (CRISPRi).^116,117 Unfortunately, this system works poorly in mycobacteria. Rock et al screened 11 CRISPR-Cas systems and found that the Streptococcus thermophilus system works best in mycobacteria.¹¹⁸ They developed a nuclease-dead Cas9 (dCas9_Sth1), and made available a set plasmids that allow insertion of a sgRNA of interest for inducible and tunable CRISPR knockout or CRISPRi knockdown of almost any gene of interest. These constructs have been used in many subsequent studies by numerous researchers.^119-128 Although an in-depth discussion of recent work using CRISPR and CRISPRi in mycobacteria is beyond the scope of this review, we would like to discuss several recent investigations prominently using genome-scale CRISPRi.

Rock developed a genome-scale CRISPRi library consisting of 96,700 sgRNAs targeting 98.2% of annotated Mtb genes, 1,658 non-targeting control sgRNAs and tetracycline inducible expression of dCas9_Sth1.¹²⁹ Most genes are targeted by multiple sgRNAs with varied level of strength. They transformed the library into Mtb, generating a mixed culture of inducible knockdown strains. When grown in liquid media and induced with tetracycline, strains became depleted to varying degrees, demonstrating the degree of vulnerability of Mtb to knockdown of each gene. This same library and technique was used again with co-treatment with seven selected antibiotics.¹³⁰ This identified 1,373 genes for which knockdown led to sensitization, and 775 genes for which it led to resistance to at least one drug. Encouragingly, for non-essential genes this data was generally in agreement with a previous similar study using transposon mutant library,¹³¹ and several previously established phenomena were redemonstrated, such as the importance of the waxy cell envelope in acting as a barrier to various drugs. An interesting finding involves the essential gene ettA, knockdowns of which resulted in resistance to streptomycin. Further investigation along this line suggested that ettA knockdown caused upregulation of whiB7, resulting in low level resistance to multiple antibiotics. Excitingly, a search of a clinical genome database demonstrated that all of Mtb lineage 1.2.1, common in the Philippines and Southeast Asia, have a nonsense mutation in whiB7. This line of investigation further led to the discovery that the drug clarithromycin is effective against this large Mtb sublineage, despite Mtb’s usual intrinsic resistance to it. The unexpected discovery that the cheap, safe, and widely available drug clarithromycin could be repurposed to help treat an estimated 600,000 cases of active TB annually is an example of the type of serendipitous discovery that can come from an unbiased genome wide test such as this.

A very similar investigation was described by Yan et al.,¹³² using a genome-scale library of sgRNAs with both CRISPRi and CRISPR-KO in Mtb. They assessed which sgRNAs were enriched or depleted under bedaquiline treatment. Their results are very similar to those described by Rock’s group, providing reassurance that this type of investigation is likely to produce repeatable and reliable data. While both investigations uncovered much that was already known about the included antimycobacterial drugs, these are drugs that have already been extensively studied. It seems likely that a similar investigation of a new drug about which less is known would yield a high proportion of novel findings that would shed light on the mechanism of action of the drug and mechanisms of resistance to it by the bacterium.

Conclusions

Recent years have seen an explosion of activity across all aspects of chemical biology and its application to the study of Mtb. Improvements continue to be made to many of the tools available to diagnose disease, identify new important drug targets in treatment-recalcitrant bacilli, and identify the targets and mechanisms of action of new drugs currently in development. In addition, applying new genetic and analytic technologies to TB disease is rapidly expanding our understanding of the disease pathogenesis and focus our attention on new promising areas of vulnerability. This review intentionally included overviews of progress in fields not traditionally considered “Chemical Biology”, namely genetically encoded reporter systems and imaging, including at the single-cell level. These tools have developed to the point where the information obtained is at the chemical level and the future of mycobacterial chemical biology will be shaped by the integration of these various tools in unified experiments that combine the strengths of all these approaches. The broad community of chemical biologists has been working hard to make this long and tragic epidemic a thing of the past.