Fluorescent Mycobacterium tuberculosis reporters: illuminating host–pathogen interactions

Abstract

The pathogenesis of Mycobacterium tuberculosis (Mtb) is intrinsically linked to its intimate and enduring interaction with its host, and understanding Mtb–host interactions at a molecular level is critical to attempts to decrease the significant burden of tuberculosis disease. The marked heterogeneity that exists in lesion progression and outcome during Mtb infection necessitates the development of methods that enable in situ analyses of Mtb biology and host response within the spatial context of tissue structure. Fluorescent reporter Mtb strains have thus come to the forefront as an approach with broad utility for the study of the Mtb–host interface, enabling visualization of the bacteria during infection, and contributing to the discovery of several facets such as non-uniformity in microenvironments and Mtb physiology in vivo, and their relation to the host immune response or therapeutic intervention. We review here the different types of fluorescent reporters and ways in which they have been utilized in Mtb studies, and expand on how they may further be exploited in combination with novel imaging and other methodologies to illuminate key aspects of Mtb–host interactions.

The authors review the utility of fluorescent reporter Mycobacterium tuberculosis strains for providing unique insight into multiple aspects of bacterial–host interactions, with single bacterium resolution.

INTRODUCTION

Mycobacterium tuberculosis (Mtb) continues to be a major global health problem, causing ∼1.7 million deaths per year (World Health Organization 2017). To better combat the health, population, and economic toll that Mtb infection exerts, it is vital that we understand how the bacterium interacts with its host, to identify potential Achilles' heels that may be targeted for therapeutic purposes. The chronic nature of Mtb infection manifests in a highly complex and regulated interaction between the bacterium and its host, and while much progress has been made in our understanding of these interactions, considerable questions still remain. In particular, there has been a rapidly growing appreciation of the heterogeneity observed during Mtb infection, which spans differences in local microenvironments experienced by the bacteria even within a single host cell, to non-uniformity within and between lesions (Tan et al.2013; Bhaskar et al.2014; Lin et al.2014; Sukumar et al.2014; Lenaerts, Barry and Dartois 2015; Manina, Dhar and McKinney 2015; Cadena, Fortune and Flynn 2017). Many outstanding questions in Mtb–host interactions are impacted by the presence of this non-uniformity; for example, how does the local microenvironment of Mtb differ at different lesion sites, what drives these differences, and how does it affect infection resolution at the individual lesion level versus the whole host? These questions span numerous aspects of the Mtb–host interface, including environmental cues, the host immune response, metabolism changes during infection, and Mtb physiology during latent infection. It would thus be beneficial for analyses of these questions to be carried out not just at a population level, but also at the level of the individual bacterium, ideally while maintaining the context of intact host tissue architecture.

To this end, the use of fluorescent reporter bacterial strains has emerged as a powerful and flexible tool for the visualization and analyses of multiple aspects of Mtb biology during infection. This has been aided by an ever-expanding genomics tool kit that has allowed not just genetic manipulation of Mtb, but importantly understanding of global bacterial transcriptomic changes during infection, knowledge which can be exploited for construction of novel reporters. We review here the types of fluorescent reporters that have been generated and examples of how they have been utilized in mycobacterial research. This spans (i) constitutive and inducible fluorescent protein reporters, (ii) reporters that exploit engineered properties of a given fluorescent protein, (iii) fluorescent reporters driven by Mtb promoters responsive to particular environmental or nutritional signals, and (iv) translational fusion protein reporters that provide insight not just into protein localization, but also bacterial replication status in situ. While this review focuses on fluorescent reporter use in host–pathogen interaction studies, it is important to note that these reporters also have extensive in vitro uses. Finally, we discuss how this technology can further be exploited and its potential maximized for unique insight into Mtb–host interactions.

CONSTITUTIVE AND INDUCIBLE FLUORESCENT PROTEIN EXPRESSION IN BACTERIA

The most straightforward use of fluorescent reporters for the study of bacterial–host interactions comes from constitutive expression of a fluorophore within the bacterium, allowing visualization of the bacteria during infection of host cells in culture or in whole animal infection models. This has been utilized in multiple mycobacterial species, including in studies with Mtb, M. bovis Bacillus Calmette-Guérin (BCG), and M. marinum (see for example Kremer et al.1995; Davis et al.2002; Tan et al.2013; Sukumar et al.2014; Kong et al.2016) (Table 1). Particularly given the difficulty of robust antibody staining in Mtb, the use of fluorescent bacteria has been vital to our ability to observe Mtb in situ during in vivo infection (Tan et al.2013; Sukumar et al.2014), and has also allowed live tracking of M. marinum during infection of its zebrafish host (Davis et al.2002; Cosma, Humbert and Ramakrishnan 2004). Such visualization provides a viewpoint not afforded by bulk/population-level assays and, as noted in the introduction, is especially important in the context of the marked heterogeneity in local environment and lesion progression and outcome present during Mtb infection. In addition to in vivo visualization during infection, constitutively fluorescent Mtb have also been utilized in several small molecule screens, in broth and in the context of macrophage infection (Queval et al.2014; Stanley et al.2014; Sorrentino et al.2015; VanderVen et al.2015; Rodrigues Felix et al. 2017). Use of constitutively fluorescent bacteria further enables the sorting of bacteria and infected cells by flow cytometry, with recovery of the cells for downstream applications. For example, using a smyc’::mCherry Mtb strain, Liu et al. (2016) recovered Mtb-infected myeloid cells from the lungs of C57BL/6J mice infected for 21 days, and showed that bacteria present in activated versus resting host cells were better able to survive exposure to isoniazid (INH) and rifampicin (Rif).

Table 1.

Summary of mycobacterial fluorescent reporters.

Reporter	Excitation max. (nm)	Emission max. (nm)	Function/notes	Mycobacteria species used	Example references
Fluorescent proteins used as constitutive reporters
CFP/Cerulean	433	475		M. marinum	Cosma, Humbert and Ramakrishnan (2004); Cronan et al. (2015)
GFP	488	510		Mtb, M. bovis BCG, M. smegmatis, M. marinum	Kremer et al. (1995); Davis et al. (2002); Queval et al. (2014); Stanley et al. (2014); Sorrentino et al. (2015); Rodrigues Felix et al. (2017)
mVenus	515	528		Mtb	Jain et al. (2016)
tdTomato	554	581		Mtb, M. bovis BCG, M. smegmatis	Carroll et al. (2010); Kong et al. (2016)
DsRed2	563	582		M. marinum	Cosma, Humbert and Ramakrishnan (2004)
mCherry	587	610		Mtb, M. bovis BCG, M. smegmatis, M. marinum	Carroll et al. (2010); Tan et al. (2013); VanderVen et al. (2015); Kong et al. (2016); Liu et al. (2016); Rodrigues Felix et al. (2017)
TurboFP635	588	635		Mtb, M. smegmatis, M. marinum	Carroll et al. (2010)
Fluorescent proteins used as inducible reporters
GFP/TurboFP635	488/588	510/635	Constitutive GFP, theophylline-inducible TurboFP635	Mtb, M. smegmatis	Mouton et al. (2016)
mCherry/GFP	587/488	610/510	Constitutive mCherry, tet-inducible GFP	Mtb	Martin et al. (2012)
mKO	548	561	Tet-inducible	Mtb	This study
Engineered fluorescent protein reporters
pH-GFP	395/475	510	Ratiometric reporter of intrabacterial pH	Mtb	Vandal et al. (2008)
Mrx1-GFP	390/490	510	Ratiometric reporter of intrabacterial mycothiol redox potential	Mtb	Bhaskar et al. (2014); Mehta, Rajmani and Singh (2016); Palde et al. (2016)
Peredox	400/587	510/615	Ratiometric reporter of intrabacterial NADH/NAD+ ratio	Mtb, M. smegmatis	Bhat, Iqbal and Kumar (2016)
ATeam1.03YEMK	435	475/527	FRET-based reporter of intrabacterial ATP levels	M. smegmatis	Maglica, Ozdemir and McKinney (2015)
Mtb promoter-based fluorescent reporters
aprA’::GFP	488	510	Fluoresces on exposure to low pH or upon macrophage infection; also used as a dual reporter, in combination with a constitutively expressed mCherry	Mtb	Abramovitch et al. (2011); Johnson et al. (2015)
rv2390c’::GFP	488	510	Fluoresces on exposure to low pH or high chloride concentrations; also used as a dual reporter, in combination with a constitutively expressed mCherry	Mtb	Tan et al. (2013); Sukumar et al. (2014)
hspX’::GFP	488	510	Fluoresces on exposure to hypoxia or nitric oxide stress; also used as a dual reporter, in combination with a constitutively expressed mCherry	Mtb	Tan et al. (2013); Sukumar et al. (2014); Zheng et al. (2017)
prpD’::GFP	488	510	Fluoresces in response to cholesterol or propionate, or upon macrophage infection; also used as a dual reporter, in combination with a constitutively expressed mCherry	Mtb	VanderVen et al. (2016); Nazarova et al. (2017)
dnaK’::tdTomato	554	581	Increased fluorescence in persister cells (generated via exposure of Mtb to high levels of INH for 4 days); used as part of a dual reporter, with a constitutively expressed mVenus	Mtb	Jain et al. (2016)
cydA’::mCherry	587	610	Fluorescence increases during host infection, and upon treatment with drugs directed against oxidative phosphorylation	M. marinum	Boot et al. (2017)
Translational fusion reporters
Replisome component-fluorophore fusion (e.g. SSB-GFP, mCherry-DnaN)	488 (GFP) 587 (mCherry)	510 (GFP) 610 (mCherry)	Fluorescent foci present during active DNA replication; SSB-GFP has been used as part of a dual reporter, with a constitutively expressed mCherry	Mtb, M. bovis BCG, M. smegmatis	Sukumar et al. (2014); Santi and McKinney (2015); Richardson et al. (2016); Logsdon et al. (2017)
Fluorophore fusions to cell wall synthesis/division dynamics machinery (e.g. ParB-mCherry, DivIVA-GFP)	488 (GFP) 587 (mCherry)	510 (GFP) 610 (mCherry)	RFP and Dendra versions also used in some cases	Mtb, M. smegmatis	Joyce et al. (2012); Meniche et al. (2014); Santi and McKinney (2015); Rego, Audette and Rubin (2017)
KatG-DsRed2	563	582		M. smegmatis	Wakamoto et al. (2013)
LucA-GFP	488	510		Mtb	Nazarova et al. (2017)

An added layer of control comes from the use of inducible promoter systems for driving expression of the fluorescent proteins, enabling expression to be turned on or off at desired time points. Several inducible systems have been utilized in the Mtb field, including a tetracycline-based system, with both tet-on and tet-off options, and a theophylline riboswitch system (Schnappinger and Ehrt 2014). With some of these systems, the amount of inducer added can serve as a second control mechanism for induction levels. A gateway cloning-based system for use in mycobacteria developed by Blumenthal et al. (2010) further allows for easy incorporation of promoters of varying strength, providing an additional layer of transcriptional control. While these inducible systems have mainly been utilized in the context of regulated expression or silencing of Mtb genes of interest, they can similarly be exploited for controlling fluorescent protein expression in Mtb. The dynamic range that can be obtained from the combination of promoters and fluorophores can be very large, for example exceeding 100-fold induction in the case of a tet-controlled codon optimized monomeric Kusabira Orange (mKO) fluorescent protein. At least two studies have utilized Mtb expressing both a constitutive fluorescent protein and an inducible fluorescent protein in a different color (Table 1). Martin et al (2012) generated an Mtb strain constitutively expressing mCherry, with green fluorescent protein (GFP) expression under the control of a tetracycline-inducible promoter. Addition of the tetracycline inducer at particular time points after infection allowed them to analyze the percentage of bacteria that were mCherry positive only versus mCherry + GFP positive, to determine the subcellular location of bacteria that were transcriptionally active in macrophages, and the impact of efferocytosis (phagocytosis of apoptotic cells) on Mtb survival and growth in vivo. Ultimately, use of this reporter contributed to their studies ascertaining that efferocytosis is able to control Mtb infection, with apoptosis itself not ‘intrinsically bactericidal’ (Martin et al.2012). Recently, Mouton et al. (2016) similarly combined a constitutively expressed GFP with a theophylline-inducible TurboFP635 fluorescent protein to generate a reporter that they used for analyses of replication rate via fluorescence dilution. In their study, the bacteria were preinduced for expression of TurboFP635 and subsequently grown in media lacking theophylline; bacterial growth was thus reflected by decreasing levels of TurboFP635.

These studies illustrate the two ways in which such constitutive/inducible fluorescent protein dual reporters can be utilized, with induction either carried out prior to use of the bacteria in the assay, or initiated at later time points as desired to mark bacteria that are continuing to undergo active transcription and translation. By combining animal model and microscopy methods, the latter approach holds potential for shedding light on heterogeneity in Mtb physiology during infection in vivo. In particular, immunofluorescence marking of host proteins, and genetic or chemical modulation of the host or bacterium, would allow such studies to aid in elucidating both host and Mtb factors that influence the non-uniformity in bacterial physiology observed. A second scenario of interest would be in drug screens, where the inducer could be added after a given period of drug exposure and subsequent washout, to test bactericidal versus bacteriostatic phenotypes. Bactericidal treatment would be expected to result in minimal signal of the induced fluorescent protein, even after compound washout and subsequent ‘outgrowth’, while fluorescent signal would be expected to increase during this outgrowth period if the tested compound/dose were bacteriostatic. Such a system is more amenable for use in high-throughput assays versus enumeration of colony forming units, and well suited for tests in the context of macrophage infection.

EXPLOITING ENGINEERED PROPERTIES OF FLUORESCENT PROTEINS FOR TRACKING Mtb PHYSIOLOGY DURING INFECTION

Constitutive expression of fluorescent proteins continues to be used extensively for visualizing and tracking Mtb; at the same time, an increasing understanding of fluorescent protein biology has led to a growing collection of engineered versions of fluorescent proteins that are now able to report on particular environmental or biochemical changes (Table 1). This has been achieved either by enhancing inherent properties of a given fluorescent protein, or by engineering fusions of fluorescent proteins to particular bacterial genes to confer novel fluorescence properties. In particular, GFP derivatives that respond to pH, mycothiol redox potential, and most recently NADH/NAD⁺ ratios, have been expressed in Mtb and used to track changes in these parameters within the bacterium itself, in different broth conditions, or during infection of macrophages (Vandal et al.2008; Bhaskar et al.2014; Bhat, Iqbal and Kumar 2016). The use of a pH-sensitive GFP enabled Vandal et al. (2008) to show that Mtb robustly maintains neutral intrabacterial pH even when present in the phagosomes of interferon-γ (IFNγ) activated macrophages, and led to the discovery of Rv3671c as a protein essential for the bacterium's ability to resist acid stress. To track intrabacterial mycothiol redox potential, Bhaskar et al. (2014) fused a mycoredoxin (Rv3198A) to roGFP2, a redox-sensitive variant of GFP, such that an oxidative shift is reflected by an increase in fluorescence excitation ratio at 405/488 nm (fixed emission at 510 nm). Among their findings was striking heterogeneity in mycothiol redox potential in Mtb during infection of macrophages, which was not observed during growth in rich broth. Host immune activation led to an oxidative shift in the mycothiol redox potential, and such a shift was similarly seen upon treatment of Mtb-infected macrophages with several anti-TB drugs. They further reported a correlation between mycothiol redox potential and drug susceptibility, with oxidation linked to increased susceptibility and reduction linked to increased tolerance (Bhaskar et al.2014). This reporter has since been utilized in several other studies, including work elucidating a role for WhiB3 in the regulation of the mycothiol redox potential of Mtb during macrophage infection, and in the validation of novel compounds that target sulfur metabolism (APS reductase) in Mtb (Mehta, Rajmani and Singh 2016; Palde et al.2016). In the same vein, heterogeneity in intrabacterial NADH/NAD⁺ ratios during macrophage infection has recently been described by Bhat, Iqbal and Kumar (2016) with a Peredox fluorescent reporter. Peredox combines an NADH-binding protein Rex from Thermus aquaticus with a circularly permutated GFP T-Sapphire; NADH binding leads to a conformational shift that results in increased GFP T-Sapphire fluorescence, with a tandemly expressed mCherry serving to normalize signal (Hung et al.2011).

The metabolism of Mtb during infection has been intensely studied, with great interest in uncovering possible novel drug targets that can exploit this critical aspect of Mtb biology (Warner 2014; Baer, Rubin and Sassetti 2015; Abuhammad 2017; Bald et al.2017). In addition to the mycothiol redox potential and NADH/NAD⁺ reporters described above, an engineered fluorescent protein reporter with direct relevance to metabolic state that has been successfully expressed in mycobacteria is an ATP-responsive biosensor (Maglica, Ozdemir and McKinney 2015). Specifically, Maglica, Ozdemir and McKinney (2015) utilized the ATP-responsive biosensor ATeam1.03^YEMK in M. smegmatis to analyze changes in intrabacterial ATP levels upon exposure to different antibiotics. This ATeam1.03^YEMK ATP reporter was originally developed by Imamura et al. (2009), and exploits coupling of the epsilon subunit of the Bacillus subtilis F₀F₁-ATP synthase with a variant of cyan fluorescent protein (CFP) and monomeric Venus (mVenus) on either end of the protein. Binding of ATP to this reporter leads to a conformational shift and resultant FRET (fluorescence resonance energy transfer) signal. With the single bacterium resolution afforded by use of this reporter, Maglica, Ozdemir and McKinney (2015) showed that treatment with bedaquiline (BDQ), an inhibitor of mycobacterial ATP synthase, led to a sudden and irreversible decrease in intrabacterial ATP levels, with this event ‘correlated with immediate growth arrest at the single cell level’. Their finding that the carbon source present affected the drop in ATP levels observed upon BDQ treatment further emphasizes the importance of understanding Mtb metabolism during infection, to ensure pertinent conditions are utilized in drug studies. The pattern of ATP decrease varied with the antibiotic used, a reflection of the mechanism of action of the drug (Maglica, Ozdemir and McKinney 2015). As is often the case with mycobacteria, this analysis at the single-cell level also revealed heterogeneity in bacterial response, with a small subset of bacteria identified that retained high levels of ATP upon BDQ exposure, but failed to resume growth after removal of the antibiotic (Maglica, Ozdemir and McKinney 2015). We note that Imamura's group has since described a new fluorescent ATP reporter QUEEN based on a circularly permutated enhanced GFP that is ratiometric, not based on FRET signal readout, and importantly unaffected by changes in bacterial growth rate (Yaginuma et al.2014). QUEEN’s dual excitation (peaks at 400/494 nm) and single emission (513 nm maximum) spectra should theoretically also aid in reducing background observed due to F420 auto-fluorescence in the CFP channel by Maglica et al (Yaginuma et al.2014; Maglica, Ozdemir and McKinney 2015).

While the physiologic dynamic range of these engineered fluorescent proteins tends to be relatively small (commonly on the order of 2 to 4-fold), careful quantification can provide valuable information, as illustrated by the studies above. By providing a direct readout of key metabolic parameters within the bacteria, these reporters enable insight into fundamental aspects of Mtb physiology during infection and in response to drug exposure. Combining the use of these engineered fluorescent protein reporters with animal infection models and thick tissue imaging methods holds promise for revealing the varied physiological status of individual Mtb bacilli that likely exists during host colonization, and the impact of therapeutic intervention on these key parameters in vivo at the level of the single bacterium.

Mtb PROMOTER-BASED FLUORESCENT REPORTERS ILLUMINATE MULTIPLE FACETS OF THE BACTERIAL–HOST INTERFACE

While the reporters described above rely on inherent or engineered properties of fluorescent proteins, a critical expansion in the repertoire of fluorescent reporters has more recently come from the use of carefully selected Mtb promoters to drive expression of a given fluorophore. Promoter-based reporters driving expression of lacZ or luciferase have long been used for analyses of the response profile of a given mycobacterial gene of interest, with levels of β-galactosidase activity and light output respectively indicating expression levels (see for example Gordon et al.1994; Timm, Lim and Gicquel 1994; Yuan et al.1998; Alland et al.2000). These reporters have contributed immense insight into Mtb biology, but their utility for studies of Mtb–host interactions is constrained by the need for substrate addition and/or lysis/permeabilization of the bacterial cell. The emergence of fluorescent proteins such as GFP changed the questions that could be addressed, and was rapidly applied and developed for mycobacterial studies (Dhandayuthapani et al.1995; Kremer et al.1995). Importantly, our growing understanding of the strong and distinct transcriptional response of Mtb to different environments has enabled the construction of promoter-based fluorescent reporters that provide a readout not just of expression of a particular gene, but serve as a proxy for the larger picture environmental cue being sensed and responded to (Table 1). In this way, such reporters provide a unique window to observe characteristics of the local microenvironment experienced by an individual bacterium during colonization of its host. In addition to tracking the expression profile of a given gene during infection, the possible facets that such reporters can inform on are thus vast, spanning immune and ionic environmental signals, to nutritional aspects of the infection focus.

On the immune and ionic environmental signal front, Abramovitch et al (2011) generated an aprA’::GFP reporter that fluoresces when Mtb is exposed to low pH or upon infection of macrophages. We have since described reporters that fluoresce when Mtb is exposed to low pH and/or high chloride (rv2390c’::GFP), or to nitric oxide stress and/or hypoxic conditions (hspX’::GFP) (Tan et al.2013). Addition of a constitutively expressed smyc’::mCherry to these promoter-based GFP reporters to generate dual fluorescent reporter strains allows all bacteria to be visualized regardless of GFP signal, and thus importantly quantification of GFP signal during infection both in macrophages and in vivo. We further developed an imaging methodology that minimizes post-processing of the lung tissue after fixation, with the fixed lungs sectioned and stained as is without any further embedding of the tissue in paraffin or OCT compound (used in cryosectioning) (Tan et al.2013). This both preserves tissue structure and prevents loss of epitopes for immunofluorescence staining of host proteins. The tissue is subsequently imaged in 0.5 μm steps out to a depth of 10–20 μm by confocal microscopy, reconstructed in three dimensions, and analyzed. These reporters, in combination with our imaging and analyses methods, have enabled several discoveries. This includes the finding of a synergistic response of Mtb to pH and chloride, which was linked to the immune status of the host, with rv2390c’::GFP reporter signal significantly lower during Mtb infection of IFNγ^−/− mice as compared to wild-type mice (Tan et al.2013). Interestingly, it was also observed that while boosting of the immune response via vaccination with heat-killed Mtb decreased bacterial load and accelerated the adaptive immune response (as shown by robust hspX’::GFP signal at 14 days post-infection), vaccination also simultaneously resulted in lower rv2390c’::GFP signal at early time points (Fig. 1) (Sukumar et al.2014). This suggests that surviving Mtb present in the vaccinated animals either adapted quicker to innate immune stresses, or that selection for the best-adapted Mtb was hastened, and raises intriguing questions about the long-term effect on individual bacteria of non-sterilizing vaccination (Sukumar et al.2014).

Figure 1.

rv2390c’::GFP reporter expression profile changes with vaccination. Mock-treated or vaccinated C57BL/6J mice were challenged with Erdman (rv2390c’::GFP, smyc’::mCherry) for 14, 28, 42, or 56 days. (A) 3D confocal images from a 14-day infection—Mtb (red), rv2390c’::GFP (green), nuclei (grayscale), F-actin (blue). Scale bar 10 μm. (B) Quantification of rv2390c’::GFP signal within each bacterium, measured as GFP/μm³, from multiple images. Each point represents a bacterium or a tight cluster of bacteria, and horizontal line indicates the median value. P-values were obtained from a Mann–Whitney statistical test. This figure is adapted from Sukumar et al. (2014).

As noted above, Mtb metabolism is a significant area of interest in the field of Mtb–host interactions. In particular, multiple studies have demonstrated the importance of Mtb utilization of cholesterol during infection (Van der Geize et al.2007; Pandey and Sassetti 2008; Yam et al.2009; VanderVen et al.2015; Lovewell, Sassetti and VanderVen 2016). To this end, Nazarova et al. (2017) recently described a reporter based on the promoter of prpD, which codes for an enzyme involved in the methyl citrate cycle that is upregulated upon Mtb exposure to cholesterol or propionate. This prpD’::GFP reporter fluoresces in response to cholesterol or propionate as expected, and further exhibits an increase in signal during Mtb infection of macrophages, reinforcing the significance of cholesterol metabolism for the bacteria during host colonization (VanderVen et al.2016; Nazarova et al.2017). A second metabolism-related promoter-based reporter that has been described is a cydA’::mCherry reporter in M. marinum (Boot et al.2017). CydA is a subunit of the cytochrome BD terminal oxidase, and Boot et al. (2017) found that its expression was increased during M. marinum growth in macrophages or its zebrafish host, as compared to growth in rich broth. Treatment particularly with drugs directed against oxidative phosphorylation also induced expression of the cydA’::mCherry reporter.

In addition to providing insight into the local microenvironment experienced by a single bacterium during infection, promoter-based reporters have further recently been utilized to inform on the presence of Mtb persister cells (Jain et al.2016). Jain et al. constructed several reporters driven by the promoters of genes found to be upregulated in persister cells (generated via exposure of Mtb to high levels of INH for 4 days) such as hspX, dnaK, and hsp. These promoters were used to drive expression of tdTomato, with a constitutively expressed mVenus present to mark all bacteria that had been successfully transduced with the constructs. Focusing on the dnaK’::tdTomato reporter, they found prior to INH exposure that a sub-population of Mtb both in broth and human sputa exhibited upregulation of reporter expression; this sub-population subsequently appeared to be selected for in the presence of INH (Jain et al.2016).

While this review is focused on ways in which fluorescent reporter Mtb strains can be directly used in infection studies, it is important to note that many of these reporters also have extensive utility for broth assays, whose results can subsequently be used in furthering understanding of the Mtb–host interface. For example, by using the aprA’::GFP and hspX’::GFP reporters in chemical biology screens, novel inhibitors of the PhoPR and DosRST two-component systems respectively were identified by searching for compounds that inhibited reporter GFP expression upon exposure of reporter Mtb to the inducing condition (low pH and hypoxia respectively) (Johnson et al.2015; Zheng et al.2017). Compounds identified from such screens serve as potential tools not just in drug discovery, but also for shedding light on fundamental aspects of Mtb–host interactions.

The studies described above illustrate the wide applicability of promoter-based fluorescent reporters for investigations on multiple aspects of Mtb biology and analyses of Mtb–host interactions. There is extensive literature on the transcriptional response of Mtb to various signals (see for example Sherman et al.2001; Rodriguez et al.2002; Park et al.2003; Rohde, Abramovitch and Russell 2007; Voskuil et al.2011; Tan et al.2013; Liu et al.2016), and the growing widespread use of high-throughput sequencing is adding to this knowledge base constantly. We are actively using such transcriptional information to expand the panel of promoter-based fluorescent Mtb reporters, to encompass additional immune and environmental signals.

TRANSLATIONAL FUSION REPORTERS—INSIGHT INTO PROTEIN LOCALIZATION AND BACTERIAL REPLICATION IN SITU

Finally, translational fusion reporters represent yet another class of reporters, where the mycobacterial protein of interest is translationally fused to a fluorescent protein. These translational fusion reporters have most commonly been used in protein localization studies (Table 1). For example, in M. smegmatis, translational fusions have been described for analyses of cell wall synthesis and division machinery dynamics (Joyce et al.2012; Meniche et al.2014; Rego, Audette and Rubin 2017), and for KatG expression during INH exposure (Wakamoto et al.2013). In Mtb, Nazarova et al. (2017) used a fusion of Rv3723 to GFP to demonstrate localization of Rv3723 to the bacterial cell membrane/envelope. Recently, Belardinelli and Jackson (2017) have also described a system that takes advantage of N- versus C-terminal fluorescent protein fusions, and the differential folding efficiency of GFP in the cytoplasm (high) versus periplasm (low), to determine not just protein localization, but also topology, in M. smegmatis.

While the use of protein-fluorophore fusions to examine localization of a protein of interest has been an established tool in the fluorescent toolkit for bacterial research, an exciting recent addition to such reporters has come from exploiting localization changes undergone by members of the replisome complex during active DNA replication, as a representation of Mtb replication status in situ (Table 1). Replisome component-fluorophore fusions have been shown in other bacteria such as Escherichia coli and B. subtilis to faithfully mark active DNA replication with fluorescent foci (Berkmen and Grossman 2006; Reyes-Lamothe et al.2008; Costes et al.2010). Taking advantage of this, we have shown that fusion of the Mtb single stranded binding protein to GFP (SSB-GFP), driven by the native Mtb ssb promoter, results in green foci during active DNA replication in the bacteria (Sukumar et al.2014). This replication reporter provides insight not possible from bacterial load information alone, as colony forming units do not distinguish between differences arising from growth rates versus death rates versus clearance rates. Use of this SSB-GFP reporter indicated considerable heterogeneity in the percentage of bacteria undergoing active DNA replication in infected C57BL/6J mice at an early time point (2 weeks post-infection, prior to the onset of adaptive immunity) (Fig. 2). In contrast, the range was much tighter at 4 weeks post-infection, with vaccinated mice showing a clear decrease in percentage of Mtb with SSB-GFP foci versus mock-treated mice, while an increase in the percentage of Mtb exhibiting SSB-GFP foci was observed in IFNγ^−/− mice (Fig. 2) (Sukumar et al.2014).

Figure 2.

SSB-GFP reporter marks Mtb replication in vivo. Mock-treated or vaccinated C57BL/6J mice were challenged with Erdman(SSB-GFP, smyc’::mCherry) for 14 or 28 days. (A) 3D confocal images from a 28-day infection—Mtb (red), SSB-GFP (green), nuclei (grayscale), F-actin (blue). For clarity, SSB-GFP signal is shown in extended focus, overlaid on the 3D image. Scale bar 10 μm. (B) Quantification of percentage of bacteria with SSB-GFP foci, measured from multiple images. Each point represents a mouse, and horizontal line indicates the median value. P-values were obtained from a Mann–Whitney statistical test. This figure is adapted from Sukumar et al. (2014).

Using live-cell imaging in combination with fluorescent marking of the origin of DNA replication and chromosomal terminus, the SSB-GFP reporter has since also been used in studies addressing how bacterial cell size control is achieved in M. smegmatis and M. bovis BCG (Logsdon et al.2017), and how growth properties impact Rif tolerance in M. smegmatis (Richardson et al.2016). Santi and McKinney (2015) have further described mCherry-DnaN and GFP-DnaX reporters in M. smegmatis, utilizing them in combination with other protein-fluorophore fusions (e.g. SSB-GFP, ParB-mCherry) to analyze the link between chromosome organization and dynamics of DNA replication.

In addition to tracking protein localization, these replisome component-fluorophore fusion reporters thus add a novel dimension to the use of translational fusion reporters. Critically, they provide a means for analyses of Mtb replication status in situ during infection in a whole animal model, in the context of intact tissue architecture and at a resolution (single bacterium/lesion) not previously possible. These reporters therefore open up multiple possibilities for in vivo studies, such as those interrogating how Mtb replication differs within a single host at different infection foci, and the bacterial and host determinants that may impact this vital facet.

COMBINING REPORTERS WITH INNOVATIVE DELIVERY AND IMAGING METHODOLOGIES

It is clear that reporters such as those described in the preceding sections have immense value in illuminating multiple critical aspects of Mtb–host interactions. We propose that the power of such reporters in Mtb pathogenesis research can further be amplified by the use of different methods for expressing these constructs in Mtb without extensive manipulation, and by combining them with other techniques, including novel imaging methodology. With regard to expression, use of these reporters has in general been predicated on introduction of the constructs via electroporation into Mtb (or other mycobacteria). A possible alternative to this is delivery of the reporter constructs via a bacteriophage (Jacobs et al.1993; Jain et al.2012, 2016). Such use of phages as a delivery system for fluorescent reporters has potential for broadening the settings in which the reporters have utility. Indeed, O’Donnell et al. (2015) have described a clinical trial in Durban, South Africa, where use of the ϕ²GFP10 reporter phage was tested as a diagnostic tool for detection of Mtb and bacterial resistance to Rif. Promising results were described, with the reporter phage showing overall high sensitivity for detection of both Mtb and Rif resistance, including in samples that were negative for acid fast bacillus staining. Using phages to deliver various reporter constructs may thus be a particularly intriguing option for enabling use of the constructs in a clinical or field setting.

As noted above, one of the greatest strengths of reporter Mtb strains is their ability to shed light on various aspects of Mtb–host interactions at the level of the individual bacterium. The expanding panel of reporter Mtb strains provides an opportunity to obtain a unique window into understanding what, where, and when Mtb experiences particular signals of interest during infection in a whole animal model, with the complete complexity of the host immune response present. For example, use of the hspX’::GFP reporter in combination with flow cytometry and a murine infection model indicates that Mtb present in alveolar macrophages exhibit lower levels of hspX’::GFP reporter signal versus bacteria present in interstitial macrophages or neutrophils (Huang and Russell 2017). hspX’::GFP reporter expression levels further correlated with host inducible nitric oxide synthase (NOS2) expression, and together these data suggest that alveolar macrophages are more permissive for Mtb colonization, in agreement with the finding that depletion of alveolar macrophages is host protective (Leemans et al.2001; Huang and Russell 2017).

Studies in a murine model of infection pursued with the rv2390c’::GFP, hspX’::GFP, and SSB-GFP reporters illustrate the power of combining use of the reporters with confocal imaging and analyses methods, to provide insight into the local environment and replication status of Mtb while retaining the spatial context of infection foci (Tan et al.2013; Sukumar et al.2014). This is particularly important given the marked heterogeneity that is observed during Mtb infection, which has significant impact on disease progression and treatment outcomes. The palette of colors available for fluorescent proteins has expanded significantly in recent years, opening the way for diverse possibilities for multicolor imaging, in combination with antibodies directed against host proteins. Even within mycobacteria, colors spanning the spectrum from blue (e.g. cerulean) to far red (e.g. TurboFP635) have been reported (Carroll et al.2010; Cronan et al.2015; Mouton et al.2016). This diversity makes it technically feasible to carefully select fluorophores that are spectrally distinct for simultaneous visualization of multiple aspects of interest, both on the bacterial and host side. In combination with this multicolor imaging, we are building on our analyses methods for infected tissue, utilizing several confocal-based imaging methodologies to enable visualization and analyses of Mtb–host interactions at the individual bacterium level not just within a single lesion, but also between lesions, all within the context of intact host tissue architecture. This requires expansion of imaging not just across the xy-plane, but ideally into the z-plane as well, to provide a greater depth of visualization and analysis of intact lesions. One way to achieve broad coverage across a sample is by mosaic image tiling across the xy-plane, where multiple images are taken and stitched together with appropriate software. The strength of such an approach is illustrated by elegant work from Mattila et al. (2013, 2015), who examined lung tissue from Mtb-infected cynomolgus macaques with antibodies against various immune markers, to define differences in host cell distribution and phenotype within a granuloma structure. As demonstrated in Fig. 3, the use of such methodology now in combination with reporter Mtb strains excitingly allows visualization of individual bacteria in infection foci relative to tissue structure, adding an extra layer of information not available when imaging a narrow field of view. Given the growing range of phenotypes addressed by reporter Mtb strains, this combination of approaches holds great promise for providing insight into the factors (e.g. environmental cues, immune signals, host cell types etc.) that distinguish an infection focus as conducive versus restrictive for bacterial growth in vivo, to further delineate how microenvironments may differ spatially within a lesion and define the resultant impact of such variation on Mtb growth.

Figure 3.

Mosaic imaging with retention of single bacterium resolution. (A) Six-week mCherry Mtb-infected C3HeB/FeJ murine lung section was imaged in a 4x4 tile, stitched together via Leica LAS X software, and reconstructed in 3D with Volocity software. Scale bar 100 μm. Inset is magnified image of boxed region. (B) Six-week mCherry Mtb-infected C3HeB/FeJ murine lung section was imaged in a 2x2 tile, stitched and 3D reconstructed as in (A). Bodipy 493/503 (green) marks neutral lipids, observed predominantly in CD68⁺ cells found within the fibrous cuff of the lesion. Scale bar 50 μm. Inset is magnified image of boxed region.

In the field of neurobiology, imaging deep into the brain has been accomplished by the use of techniques that optically clear the tissue (see for example Erturk et al.2012; Chung et al.2013; Susaki et al.2014; Yang et al.2014). CLARITY and PACT (passive CLARITY technique) are two such methods, with membrane lipid bilayers stripped away by the use of detergents, and structure integrity maintained via an acrylamide hydrogel (Chung et al.2013; Yang et al.2014). We propose that the application of such methods to infectious disease study has exciting potential, and indeed a very small number of reports are just beginning to emerge on this. In particular, DePas et al. (2016) used PACT, in combination with hybridization chain reaction for detection of specific rRNAs, in their analyses of Staphylococcus aureus, Pseudomonas aeruginosa, Streptococcus, and Achromobacter xylosoxidans in sputum samples from patients with cystic fibrosis. Among other findings, they found that the different bacterial species were present in aggregates of varied sizes, with Streptococcus also closely associated with host cells (DePas et al.2016). Cronan et al. (2015) have utilized CLARITY and PACT in adult zebrafish expressing different fluorescent host markers and infected with M. marinum expressing either the cerulean or tdTomato fluorescent proteins. They describe the presence of blood vessels encircling granulomas in adult zebrafish, and heterogeneity in tumor necrosis factor expression within the M. marinum granulomas. They were also able to visualize tdTomato-expressing Mtb in the lungs of infected C57BL/6 mice that were optically cleared by PACT (Cronan et al.2015). We have utilized our fluorescent reporter Mtb strains in infections of C3HeB/FeJ mice that exhibit diverse lesion pathology, including caseating granulomas, more similar to those observed in human infection (Irwin et al.2015; Kramnik and Beamer 2016). As illustrated in Fig. 4 and Video 1, PACT enables greater depth of imaging and thus visualization of an entire lesion, providing the ability to analyze Mtb and host cells in a complete lesion in three dimensions, and retaining spatial information that may otherwise be lost when observing a single z-plane image.

Figure 4.

PACT treatment of 6-week mCherry Mtb-infected C3HeB/FeJ murine lung section. (A) Optical clearing of tissue after PACT. (B) Confocal images (xyz-planes) of untreated and PACT-treated samples. The z-plane shown is at 80 μm depth in each case (last z-step taken in images). Scale bar 20 μm.

The imaging methods described above directly pair with and exploit the use of fluorescent reporter Mtb strains in animal infection models. At the same time, the path remains open for the combination of such analyses with other procedures on the same tissue, including both conventional histology and emerging imaging modalities such as matrix-assisted laser desorption/ionization imaging mass spectrometry (MALDI-IMS) (Moore, Caprioli and Skaar 2014; Spraggins et al.2016). The latter technology brings different strengths, such as its label-free nature, and has been utilized in a study examining moxifloxacin distribution within the lungs of Mtb-infected rabbits (Prideaux et al.2011). This study strikingly showed the non-uniform spatial distribution of moxifloxacin within the lung, with penetration of the drug into the very center of caseating granulomas observed to be low (Prideaux et al.2011). However, MALDI-IMS does not currently have the single bacterium resolution required for study of some of the aspects of the Mtb–host interface discussed above. Nonetheless, a single tissue sample could be divided into facing sections, or serially sectioned with varying depth for the different procedures. For example, thick sections >500 μm depth could be cut for optical clearing and confocal microscopy, with an immediately following thin section <10 μm for MALDI-IMS. This could be repeated through the depth of the entire tissue, with the images obtained from each approach then mapped onto each other and reconstructed in three dimensions for analyses. We believe that the utilization of reporter Mtb strains in combination with such methods holds great promise for the continued development of a molecular understanding of the Mtb–host interface.

CONCLUDING REMARKS

Fluorescent reporter Mtb strains have enabled the illumination of aspects of Mtb–host interactions at a resolution not previously feasible, and inroads into difficult questions that bulk/population-level assays are not suited to answer. We have highlighted here the array of fluorescent reporters that have been successfully utilized in mycobacterial research, and the breadth of key questions that they can help address, spanning heterogeneity in microenvironments and bacterial metabolism during infection, to the effects of vaccination and therapeutic intervention on Mtb growth in vivo. We envision that the panel of reporters available will continue to expand in the coming years, bringing with them the ability to interrogate even more aspects of Mtb biology during infection. Further, the possibilities for continued innovation with this technology are vast, including combination of different reporter types, and exploitation of reporters in conjunction with varied other methods such as tissue optical clearing and MALDI-IMS. The compatibility of this new generation of tools with established animal models also enables their direct use in drug and vaccination studies, with analyses now possible at the single bacterium level, in the spatial context of tissue structure. We anticipate that the use of such fluorescent reporters will continue to open novel avenues of research, with the insight gained contributing importantly to our ability to more effectively combat tuberculosis.

FUNDING

This work was supported by research grants to ST from the National Institutes of Health (AI114952), and from the American Lung Association (RG-507805).

Conflict of interest. None declared.