The ΔfbpA mutant derived from Mycobacterium tuberculosis H37Rv has an enhanced susceptibility to intracellular antimicrobial oxidative mechanisms, undergoes limited phagosome maturation and activates macrophages and dendritic cells

Summary

Mycobacterium tuberculosis H37Rv (Mtb) excludes phagocyte oxidase (phox) and inducible nitric oxide synthase (iNOS) while preventing lysosomal fusion in macrophages (MΦs). The antigen 85A deficient (ΔfbpA) mutant of Mtb was vaccinogenic in mice and the mechanisms of attenuation were compared with MΦs infected with H37Rv and BCG. ΔfbpA contained reduced amounts of trehalose 6, 6, dimycolate and induced minimal levels of SOCS-1 in MΦs. Blockade of oxidants enhanced the growth of ΔfbpA in MΦs that correlated with increased colocalization with phox and iNOS. Green fluorescent protein-expressing strains within MΦs or purified phagosomes were analysed for endosomal traffick with immunofluorescence and Western blot. ΔfbpA phagosomes were enriched for rab5, rab11, LAMP-1 and Hck suggesting enhanced fusion with early, recycling and late endosomes in MΦs compared with BCG or H37Rv. ΔfbpA phagosomes were thus more mature than H37Rv or BCG although, they failed to acquire rab7 and CD63 preventing lysosomal fusion. Finally, ΔfbpA infected MΦs and dendritic cells (DCs) showed an enhanced MHC-II and CD1d expression and primed immune T cells to release more IFN-γ compared with those infected with BCG and H37Rv. ΔfbpA was thus more immunogenic in MΦs and DCs because of an enhanced susceptibility to oxidants and increased maturation.

Introduction

Tuberculosis is the leading cause of mortality owing to infections in man today with at least 3 million deaths reported worldwide (Harries and Dye, 2006). The causative agent Mycobacterium tuberculosis (Mtb) has the unique ability to survive within macrophages (MΦs) using diverse strategies and become dormant, but reactivate to cause adult tuberculosis. The ability of Mtb to survive within the hostile environment of MΦs depends upon several unique features. The MΦs generate reactive oxygen species (ROS) from NADPH oxidase (phagocyte oxidase; phox) and nitric oxide (NO) via inducible nitric oxide synthase (iNOS) to kill intracellular pathogens. The phagosome–lysosome (P-L) fusion followed by enzymatic degradation appears to be another major mechanism of pathogen destruction (Nathan, 1983; Clemens, 1996; Deretic et al., 1997; Russell, 2001). Earlier studies have shown that virulent strains of Mtb and BCG strains produce abundant superoxide dismutase that inactivates superoxide, shed abundant lipids that scavenge ROS and modulate ROS in human neutrophils (Scherer et al., 1997; Perskvist et al., 2002a). Interestingly, H37Rv and BCG phagosomes also prevent colocalization of iNOS and phox with their phagosomes within MΦs (Scherer et al., 1997; Perskvist et al., 2002a; Miller et al., 2004; Daniel et al., 2006). Armstrong and Hart (1971) made the seminal observation that Mtb phagosomes do not fuse with lysosomes in MΦs that eventually led to the unraveling of a unique mechanism of pathogen survival. Thus, it seems clear that Mtb has multiple strategies to bypass MΦ antimicrobial mechanisms and become a successful pathogen.

A vaccine has been proposed to be the most effective means to prevent tuberculosis (Donald et al., 2007). However, the mechanisms through which successful vaccines can be developed remain complex. The widely used BCG vaccine has a variable efficacy (0–80%) in children, geographic variability and does not protect against adult tuberculosis (Fine, 1989; Brewer and Colditz, 1995; Fine, 2001). As a consequence, several attenuated strains of Mtb have been produced and tested in animal models as potential vaccines. The isocitrate lyase (icl) knockout (KO) mutant of Mtb had reduced virulence compared with the wild type in mice, but the organisms were still present in sufficient numbers in the organs to preclude vaccination strategies (McKinney et al., 2000). The mammalian cell entry (mce) gene mutant was attenuated but was virulent through specific routes in mice (Gioffre et al., 2005). We first reported that an attenuated mutant of H37Rv which lacked the expression of antigen 85A (ΔfbpA) offered better protection than BCG against tuberculosis in the mouse model (Copenhaver et al., 2004). Subsequently, a pantothenate auxotroph of Mtb was found attenuated in mice and also protect mice against a lethal challenge with virulent Mtb, but the protection afforded was inferior to that provided under similar circumstances with BCG (Sambandamurthy et al., 2002). More recent reports show that double gene KO strains from Mtb offer better protection than BCG in mice and safer in immunodeficient mice (Sambandamurthy et al., 2006). These studies lend an optimism that attenuated candidate vaccines can be derived from pathogenic Mtb, although, the molecular basis and rationale for the use of Mtb-derived vaccines remains confusing because of the perplexing relationships between attenuation and immunogenicity.

It is generally held that, for bacterial strains to be effective vaccines, they need to be killed and processed efficiently by antigen-presenting cells (APCs) like MΦs and dendritic cells (DCs) (Harding, 1991). This involves oxidative killing followed by lysosomal destruction, where enzymes degrade microbial proteins into peptides that are then exported to the MHC-II containing compartments (MIIC) and presented in the context of MHC-II to CD+ T cells (Harding, 1991; Levine and Chain, 1991). There is of course, a cytosol dependent processing of antigens for MHC-I presentation (Harding, 1991). Paradoxically, these mechanisms appear deficient during antigen presentation by mycobacterial vaccines. For example, virulent Mtb as well as BCG prevent P-L fusion in MΦs (Clemens, 1996; Russell, 1998; Deretic and Fratti, 1999). Mtb has been found to downregulate the expression of MHC-II in MΦs and desensitize them for the activating effects of IFN-γ (Noss et al., 2000) (Banaiee et al., 2006). Human monocytes and mouse MΦ infected with Mtb show defective antigen presentation (Pancholi et al., 1993; Gercken et al., 1994; Ramachandra et al., 2001). Modulation of MΦ function by Mtb has been shown to correlate with certain immunosuppressive antigens (Stewart et al., 2005). We recently showed that attenuated BCG and wild-type H37Rv do not efficiently present certain immunodominant antigens like antigen 85B to T cells, an event related to phagosome maturation arrest (Singh et al., 2006). Together, these studies suggest that wild-type Mtb and to a certain extent BCG vaccine appear to modulate the ability of MΦs to be immunogenic for T cells. This raised a question whether attenuated Mtb-derived candidate vaccines similarly affect MΦs. Therefore, in this study, we have compared the intracellular survival strategies of a novel ΔfbpA mutant derived from H37Rv with the BCG vaccine and wild-type H37Rv. We have analysed the differences in cell wall lipids and modulation of signal transduction and correlated them to intracellular susceptibility for oxidants, endosomal traffic and activation of APCs. We demonstrate that ΔfbpA is deficient in cell wall lipids and more susceptible to the antibacterial mechanisms of MΦs. We suggest that the immunogenicity of mycobacterial vaccines may well depend upon an efficient intracellular processing by APCs.

Results

Deletion of fbpA alters the phenotype with alterations in cell wall lipids

That fbpA, fbpB and fbpC transfer mycolic acids to trehalose to synthesize 6,6, trehalose dimycolate (TDM) is well established (Belisle et al., 1997). Our gene disruption studies indicated that ΔfbpA and ΔfbpB had different phenotypes, ΔfbpA being attenuated and ΔfbpB remaining unchanged (Armitige et al., 2000; Copenhaver et al., 2004). In order to determine whether deletion of fbpA affected lipids, we analysed the mycolic acid content of whole cells using column purification of followed by thin layer chromatography (TLC). Initial studies showed no differences in the content of alpha, keto and methoxymycolic acids between the three strains during TLC analysis (not shown). We then determined the cell wall content of TDM using radioisotope incorporation into the wild-type H37Rv, ΔfbpA and the reconstituted ΔfbpA. The strains were cultured in broth and metabolically labelled with ¹⁴C-acetate and incorporation into TDM measured using TLC analysis of cell wall extracts followed by spectroscopy. Figure 1 shows that ΔfbpA had a decreased incorporation of the label into cell wall derived from TDM compared with H37Rv while the reconstituted mutant incorporated levels more than the wild type. We suggest that overexpression of fbpA gene in the multiple copy plasmids may have enhanced ¹⁴C-acetate incorporation.

Fig. 1.

Disruption of fbpA gene leads to reduced cell wall lipids in ΔfbpA mutant compared with wild-type Mycobacterium tuberculosis (Mtb, H37Rv) and reconstituted ΔfbpA. Bacteria were grown in the presence of ¹⁴C-acetate in 7H9 broth and at hourly intervals thin layer chromatographic analysis of cell wall extracts was performed followed by radiography. The bars represent c.p.m. ± SD (not apparent) of two similar experiments. Wild-type H37Rv incorporates more label than ΔfbpA mutant and the reconstitution of the gene restores the incorporation of radiolabel into the mutant (P-values shown for strains compared; Student's t-test).

ΔfbpA modulates signal transduction in MΦs through the reduced expression of suppressor of cytokine signalling 1

BCG and Mtb as well as their product TDM strongly induce suppressor of cytokine signalling 1 (SOCS-1) in MΦs (Imai et al., 2003). SOCS-1 in turn, has been found to desensitize MΦ response to IFN-γ, affect maturation of dendritic cells (DCs) and suppress the MHC-II expression of DCs and MΦs in response to IFN-γ (O'Keefe et al., 2001; Bartz et al., 2006). We hypothesized that the differences in the cell wall content of TDM in mycobacteria may affect their ability to induce SOCS-1 and thereby, the ability to modulate MΦs and DCs. MΦs were infected with ΔfbpA, BCG and H37Rv followed by analysis of SOCS-1 levels by Western blot. MΦs showed an early and strong expression of SOCS-1 when infected with either BCG or H37Rv that continued over 4 h (Fig. 2A). In contrast, ΔfbpA induced none at 1 h post infection, while, trace levels of SOCS-1 were apparent in MΦs by 4 h. The positive control TDM by itself induced a strong signal for SOCS-1 (Fig. 2B). A nearly identical profile was found also in immature DCs purified from bone marrows of mice (not shown). As SOCS-1 desensitizes MΦs to the effects of IFN-γ, we tested the susceptibility of ΔfbpA, BCG and H37Rv infected MΦs to 400 U ml^–1 of IFN-γ followed by colony-forming unit (cfu) counts 48 h later. Figure 2C illustrates that under these conditions, IFN-γ activated MΦs were able to kill ΔfbpA better than BCG or H37RV, killing more 0.8 log₁₀ organisms in 48 h. These results suggest that ΔfbpA was more susceptible to the bactericidal mechanisms induced by IFN-γ in MΦs compared with BCG or H37Rv.

Fig. 2.

ΔfbpA affects signal transduction in MΦs through the delayed induction of suppressor of cytokine signalling 1 (SOCS-1).

A. ΔfbpA induces less strong SOCS-1 response. C57BL/6 mouse bone marrow-derived macrophages (MΦs) were infected with mycobacteria at an moi of 1:1 and incubated for 1 and 4 h. MΦ lysates were analysed using 7–12% SDS gels and electroblotted membranes were probed with an antibody to SOCS-1, followed by HRP-conjugate. Bands were detected using ECL. H37Rv and BCG induce copious expression of SOCS-1 in MΦs 1 and 4 h after infection. ΔfbpA does not induce SOCS-1 at 1 h but relatively lower levels of SOCS-1 are apparent at 4 h post infection (Mr, molecular weight marker). Lanes were loaded with equal amounts of protein as indicated by β-actin control.

B. Purified trehalose 6,6, dimycolate (TDM) induces SOCS-1. Varying amounts of TDM suspension in mineral oil were added to MΦs and lysates examined at 1 h after treatment for SOCS-1 as above. Naïve MΦs were treated with mineral oil control.

C. ΔfbpA is more susceptible to IFN-γ. MΦs were infected with mycobacteria (moi 1:1), washed to remove extracellular bacteria and 4 h later incubated with 400 U ml^–1 of IFN-γ. 48 h later, the MΦs were lysed and plated for colony-forming unit (cfu) counts on 7H11 agar. Data summarizing three separate experiments show that IFN-γ is more effective in reducing the cfu of ΔfbpA when compared with its effects on MΦs infected with either BCG or H37Rv (P-values shown for groups compared; Student's t-test).

Growth curves of ΔfbpA, BCG and wild-type H37Rv in murine MΦs and susceptibility to oxidants

We have previously described the growth profiles of ΔfbpA mutant and wild-type H37Rv in the murine J774.A1 and human THP1 MΦs (Copenhaver et al., 2004). Figure 3A shows that the mutant had a similar inability to grow in un-activated primary murine MΦs, while, H37Rv registered a significant growth. BCG was also attenuated. Prior activation of MΦs with IFN-γ reduced the cfu counts of all three strains well below that found for the same strains in naïve MΦs. This is consistent with the known ability of IFN-γ to preactivate MΦs and the reduced response of mycobacteria infected MΦs to the activation effects of IFN-γ (Gercken et al., 1994; Banaiee et al., 2006). As IFN-γ activates oxidant response in MΦs, intracellular susceptibility of mycobacteria to oxidants, was determined by incubating infected MΦs with diphenylene iodonium (DPI), a ROS inhibitor and NG-monomethyl-l-arginine (NMMA), an inhibitor of NO synthesis or their combination. Preliminary studies showed that among inhibitor-treated MΦs, between days 5–7, excessive growth of wild-type Mtb detached MΦs after infection. Thus, in order to maintain the integrity of MΦs and avoid errors in bacterial counts, cfu were determined on day 3 post treatment. Figure 3B shows that both DPI or NMMA enhanced the growth of ΔfbpA, although their combination was more effective. In contrast, the growth of H37Rv was significantly enhanced only with NMMA but not DPI. BCG showed a modest increase in growth when NO was blocked through NMMA. These results suggest that ΔfbpA was susceptible to the individual and combined action of ROS and NO within MΦs. The wild type and BCG were more susceptible to intracellular NO and less susceptible to ROS.

Fig. 3.

Growth and susceptibility of ΔfbpA mutant, H37Rv and BCG vaccine to intracellular oxidants.

A. Untreated (open symbols) or IFN-γ activated MΦs (closed symbols; 400 U ml^–1 for 24 h) were infected with the three strains of mycobacteria and incubated. On days shown, triplicate wells of MΦs per strain were lysed and plated for cfu on 7H11 agar in three separate experiments. ΔfbpA (Δ) and BCG (○) are unable to grow while the wild-type H37Rv (□) shows an increase in cfu counts over 7 days. IFN-γ activation of MΦs decreases the cfu of all three strains by more than 1.5 log₁₀ (P-values for growth in naïve MΦs versus IFN-γ activated MΦs; Student's t-test).

B. MΦs were infected as above and cfu determined on day 3 (see Results) in the presence or absence of an inhibitor of reactive oxygen species (ROS; diphenylene iodonium, DPI) or nitric oxide (NO) synthesis (NG-monomethyl-l-arginine, NMMA) or their combination. DPI and NMMA enhance the growth of ΔfbpA. DPI is ineffective against H37Rv while NMMA is effective. BCG shows a modest increase in growth with a combination of DPI and NMMA (P-values are shown above the bars versus growth of mycobacteria in naïve MΦs; Student's t-test). MΦs infected with Erdman strain show a growth profile similar to H37Rv and was included as another positive control.

Increased intracellular targeting of phox and iNOS to ΔfbpA phagosomes in MΦs

NO is produced by iNOS that assembles against the phagosomal membrane, and ROS is produced by the phox enzyme that also colocalizes against bacterial phagosomes in MΦs (Vazquez-Torres and Fang, 2001; Miller et al., 2004; Daniel et al., 2006). In order to determine whether susceptibility to oxidants (Fig. 3B) correlated with the increased intracellular targeting of iNOS and phox, MΦs were infected with gfp ΔfbpA, gfpH 37Rv or gfpBCG and immunostained for iNOS and p47^phox proteins. MΦs were stained untreated and after IFN-γ activation. Colocalization was determined using laser confocal scanning microscopy (LSM). Figure 4A illustrates that gfp ΔfbpA had an increased targeting with iNOS using a specific mab compared with gfpH 37Rv or gfpBCG phagosomes in naïve MΦs. The quantitative scoring of colocalizations derived from triplicate experiments using naïve MΦs is shown below that confirms an enhanced targeting of iNOS to gfp ΔfbpA phagosomes (Fig. 4B). A similar profile was observed for colocalization of these strains with p47^phox; gfp ΔfbpA acquired more phox than either gfpH 37Rv or gfpBCG (Fig. 4C and D). As a positive control, MΦs were activated with IFN-γ that markedly increased the colocalization of phox and iNOS to gfp ΔfbpA, gfpBCG and gfpH 37Rv phagosomes. As more than 75% of phagosomes of these three strains acquired phox and iNOS, and no significant differences in colocalization were found after IFN-γ activation, data are not shown separately. However, these positive control studies confirmed that the bactericidal mechanisms of MΦs were intact and activated by IFN-γ.

Fig. 4.

Phagosomes of ΔfbpA more frequently acquire inducible nitric oxide synthase (iNOS) and p47^phox component of phagocyte oxidase in MΦs. MΦs were infected with green fluorescent protein (gfp) expressing mycobacterial strains (gfpDfbpA, gfpBCG and gfpH 37Rv) for 4 h, washed and incubated for 24 or 72 h. MΦs were then fixed, permeabilized and stained with a mouse monoclonal antibody to iNOS or goat anti-p47^phox followed by anti-Ig-Texas red conjugates. MΦs were examined for colocalization using a Deltavision laser scanning microscope (LSM). Arrows illustrate colocalization in A and C.

A. LSM profiles illustrate that gfpΔfbpA phagosomes acquire iNOS while gfpH 37Rv and gfpBCG acquire less.

B. Per cent phagosomes positive for iNOS colocalization was tabulated from three experiments and P-values are shown for the difference between gfpDfbpA, gfpDBCG and gfpH 37Rv phagosomes (Student's t-test). At least 200 microscopic fields were scored blind per mouse MΦ preparation. Colocalizing bacteria were averaged from three independent experiments each with MΦs from three mice; each microscopic field contained about five MΦs, each of which contained about five bacilli.

C. gfpΔfbpA phagosomes acquire more p47^phox than either gfpH 37Rv or gfpBCG.

D. Phagosomes positive for colocalization are shown; data averaged from three separate experiments show increased phox colocalization for gfpΔfbpA versus gfpDBCG and gfpH 37Rv.

Increased oxidant activity in MΦs infected with ΔfbpA

As a combination of DPI and NMMA increased the growth of ΔfbpA (Fig. 3B), and oxidant-producing enzymes colocalized with ΔfbpA phagosomes within MΦs (Fig. 4), additional studies were performed to confirm the levels of intracellular ROS as well as NO released by the infected MΦs. The intracellular ROS measured using dihydrodichloro-fluorescein diacetate (H₂DCFDA) was expressed as mean fluorescence intensity (MFI) using flow cytometry. ΔfbpA induced higher and sustained levels of ROS than either H37Rv or BCG in MΦs (Fig. 5A). In additional studies, quantitative techniques were used to measure and compare the intracellular accumulation of oxidants; ROS was measured using fluorometry and NO by colorimetry. ΔfbpA infected MΦs again contained higher levels of ROS as well as the NO-derived product, nitrite (Fig. 5B and C). Data indicate that decreased growth of ΔfbpA in MΦs was likely due to an increased oxidative activity.

Fig. 5.

A. Intracellular oxidative activity within MΦs infected with ΔfbpA, BCG and H37Rv. MΦs or C57BL/6 bone marrow (BM)-derived BMA.A3 macrophage cell lines (both similar, MΦs shown) were infected with mycobacteria at an moi of 1:1, washed and incubated for 72 h.

A. At time points shown, cells were treated with 1 μg ml^–1 dihydrodichloro-fluorescein diacetate (H₂DCFDA), a probe for intracellular reactive oxygen species (ROS) for 5 min, washed and analysed in BD-Facscan using Cellquest software. Viable naïve MΦs have a basal level of fluorescence indicated by a red fill. MΦs infected with ΔfbpA show an enhanced and sustained fluorescence over time (green). MΦs infected with H37Rv show minimal ROS activity (black) while BCG infected MΦs show an intermediate level (blue). A positive control of phorbol myristyl acetate (an agonist for ROS) activated MΦs showed a burst of activity that declines rapidly within minutes (not shown). Mean fluorescent intensity (MFI, inset) values for ΔfbpA calculated from triplicate experiments (one histogram illustrated) was significant using Student's t-test (*P < 0.0092).

B and C. Fluorometry and colorimetry for oxidant activity in MΦs. MΦs were infected as above. Twenty-four and 72 h post infection, the cultures from triplicate experiments were tested for ROS using H₂DCFDA and fluorometry (B) and nitrite, a product of NO synthesis by MΦs using the Greiss reagent and colorimetry (C). ΔfbpA induces accumulation of higher levels of ROS and nitrite and P-values are shown for gfpΔfbpA versus gfpΔBCG and gfpH 37Rv (Student's t-test). Background fluorescence for ROS or nitrite in uninfected MΦs is shown as a dotted line in both B and C.

Enhanced acquisition of early and late endosomal markers by Δ fbpA phagosomes

Endosome fusion events precede P-L fusion which is the third mechanism that is known to kill intracellular bacteria and we sought to determine if ΔfbpA was attenuated because of increased P-L fusion (Clemens, 1996; Russell, 1998; Vieira et al., 2002; Vergne et al., 2004a). Virulent Mtb phagosomes are known to fuse selectively with early endosomes while excluding late endosomes as well as lysosomes (Clemens and Horwitz, 1996; Deretic et al., 1997; Russell, 2001). The early endosomes deliver nutrients and organelles to enable growth of bacteria. In order to determine the sequential endosome fusion events, MΦs were first loaded with endocytic tracers, chased and infected with gfp-mycobacteria for 4 h. Colocalization was determined using LSM. Figure 6A illustrates that gfp ΔfbpA rapidly colocalized with the acidotropic dye LTR, while gfpH 37Rv excluded LTR. The ability of H37Rv to exclude LTR is consistent with similar previous reports (Via et al., 1997). LSM studies were also performed using the early endosome marker, transferrin Texas red (TRR) and the late endosomal marker, mannosylated BSA rhodamine (MBR). The time bound colocalization of LTR, MBR and TRR by mycobacterial phagosomes within MΦs is shown in Fig. 6B. All three mycobacterial phagosomes were similar in the acquisition of TRR, an early endosomal marker. However, gfp ΔfbpA phagosomes showed an enhanced capacity to fuse with late endosomes (LTR and MBR) when compared with either gfpH 37Rv or gfpBCG.

Fig. 6.

ΔfbpA phagosomes efficiently acquire both early and late endosomal markers when compared with H37Rv and BCG within MΦs. MΦs were labelled with lysotracker red (LTR), mannosylated BSA rhodamine (MBR), or transferrin Texas red (TRR), chased with medium and infected with gfpΔfbpA, gfpBCG or gfpH 37Rv strains. MΦs were washed and examined for fluorescence colocalization.

A. LSM profile illustrates that live gfpΔfbpA has markedly enhanced colocalization with LTR while the wild-type gfpH 37Rv excludes LTR.

B. Per cent phagosomes positive for colocalization was determined by using LSM in three separate experiments. gfpΔfbpA significantly colocalizes with both LTR and MBR compared with either gfpBCG or gfpH 37Rv (Student's t-test). Colocalization with TRR, an early endosomal fusion marker, was comparable for all three strains tested. Scoring was carried out as in Fig. 4.

Phagosomes of ΔfbpA acquire phagosome maturation markers but exclude lysosomal markers

Mycobacteria growing within MΦs fuse differentially with early and late endosomes depending upon their virulence, but virulent strains such as H37Rv do not fuse with lysosomes (Via et al., 1998; Fratti et al., 2000). As P-L fusion leads to mycobacterial killing, we sought to determine if ΔfbpA was attenuated because of an altered ability to fuse with lysosomes. In order to characterize this event, two types of markers were analysed. Lysosomes of murine and human MΦs express CD63 and rab7 markers (Clemens and Horwitz, 1995; Astarie-Dequeker et al., 1999; Astarie-Dequeker et al., 2002; Fratti et al., 2003). Other lysosomal markers such as lysosome-associated membrane protein 1 (LAMP-1) and the src family kinase, Hck have been reported to occur also on late endosomes (Mommaas et al., 1995; Astarie-Dequeker et al., 2002; Perskvist et al., 2002b). For example, LAMP-1 is present on all mycobacterial phagosomes but at variable levels on virulent and avirulent strains (Ullrich et al., 1999; Fol et al., 2006). MΦs infected with gfp-expressing mycobacteria were separately labelled with fluorescent tagged antibodies to LAMP-1, Hck, CD63 and rab7 in an effort to colocalize the mycobacteria with these markers. Figure 7A and 7B illustrates that gfpΔfbpA phagosomes colocalized with Hck but not with rab7. Quantitative scoring of colocalization indicated that the mutant phagosomes stained for LAMP-1 and Hck more than BCG vaccine or H37Rv but excluded the lysosome-specific markers, CD63 and rab7 (Fig. 7C).

Fig. 7.

ΔfbpA phagosomes acquire late endosomal maturation markers but exclude lysosomal markers. MΦs were infected with gfpΔfbpA, gfpH37Rv and gfpBCG strains for 4 h, washed and incubated for 24 h. MΦs were fixed, permeabilized and stained for three markers of late endosomes (lysosome-associated membrane protein 1, LAMP-1; src family kinase, Hck) and lysosomes (CD63 and rab7) using specific antibodies and counterstained with anti-Ig-Texas red conjugates. Phagosomes were examined for colocalization (arrow and arrowhead).

A. LSM profile illustrates that gfpΔfbpA has an enhanced colocalization (arrow) with the Hck.

B. Illustration that gfpΔfbpA largely excludes rab7.

C. Per cent phagosomes positive for fluorescence colocalization with the markers was determined by using LSM. gfpΔfbpA phagosomes are more enriched for LAMP-1 and Hck but not CD63 and rab7. gfpBCG and gfpH 37Rv also acquire LAMP-1 to a lower level but exclude Hck, CD63 and rab7. Scoring from three separate experiments was averaged and significance determined using Student's t-test.

D. MΦs were infected with mycobacteria at an moi of 1:1 and phagosomes were purified on sucrose gradients at 24 h post infection. Latex beads (LB) were internalized for 4 h and phagosomes prepared as a positive control. Phagosomal proteins were then analysed using Western blot with specific antibodies followed by anti-Ig conjugates and HRP-dependent chemiluminescence. LB phagosomes acquire nearly all docking proteins along with CD63 and rab7, the specific marker of lysosomes. gfpΔfbpA phagosomes acquire multiple markers that suggest an ability to fuse with early and recycling endosomes such as rab5A, rab5B and rab11. While they acquire late endosomal markers such as Hck, they exclude lysosome-specific markers like CD63 and rab7. LAMP-1 and phosphatidyl inositol-3-kinase are also enriched on LB and gfpΔfbpA phagosomes compared with gfpBCG and gfpH 37Rv.

In additional confirmatory studies, MΦs were infected with mycobacterial strains and latex bead controls followed by sucrose gradient purification of phagosomes. The latter were analysed using Western blot with a panel of antibodies directed against early endosomes (rab5A, rab5B), recycling endosomes (rab11), late endosomes (LAMP-1, Hck) and lysosomes (CD63, rab7). An antibody against phosphatidyl inositol 3-kinase (PI-3K) was also used, as it has been reported to be one main regulator of endosomal docking process (Deretic et al., 1997; Deretic and Fratti, 1999; Deretic et al., 2004). Western analysis showed interesting results (Fig. 7D). As anticipated, the latex beads that are known to undergo rapid P-L fusion expressed multiple markers of lysosomal fusion including CD63 and rab7. ΔfbpA phagosomes were enriched for the docking proteins, ra5A and rab5B that characterize early endosomes while they excluded rab7, the marker of lysosomes as well as CD63, another marker of lysosomes (Fig. 7D). They were also enriched for rab11, a marker for recycling endosomes and for PI-3K. In addition, ΔfbpA phagosomes readily acquired the late endosomal markers such as Hck and LAMP-1 when compared with either H37Rv or BCG. The latter excluded CD63, rab7 and rab11 consistent with their previously reported behaviour (Deretic et al., 1997; Deretic and Fratti, 1999; Deretic et al., 2004). Previous studies show that wild-type H37Rv acquires the early endosomal marker rab5 but excludes the lysosomal marker rab7 in MΦs, and are thus known to be ‘immature’ phagosomes and ‘arrested’ at the level of early endosomes (Via et al., 1997; Russell, 2001). As ΔfbpA phagosomes acquired rab5, rab11 and PI-3K, we suggest that their phagosomes were more mature than either H37Rv or BCG phagosomes in MΦs, but were still not competent to fuse with the lysosomes.

MΦs and dendritic cells (DCs) infected with ΔfbpA have a more mature phenotype

Virulent H37Rv strain of Mtb downregulates the expression of MHC-II in MΦs as well as desensitizes the MΦs to the activating effects of IFN-γ (Noss et al., 2000; Banaiee et al., 2006). BCG has also been shown to subvert the phenotype of APCs like DCs affecting their maturation status (Gagliardi et al., 2005). Furthermore, both Mtb and BCG induce SOCS-1 in MΦs that can inhibit the activating effects of IFN-γ (Imai et al., 2003; Manca et al., 2005). To investigate whether ΔfbpA influenced the phenotype of APCs, MΦs and DCs were first infected with ΔfbpA, H37Rv and BCG and then tested for the surface expression of receptors using immunocytometry. Figure 8A illustrates the histograms of infected MΦs while, the footnote Table outlines the MFI values obtained from three independent experiments. MΦs infected with ΔfbpA had an enhanced expression of MHC-II, CD1d, CD86 and CD40 when compared with MΦs infected with either H37Rv or BCG. CD80 expression was comparable. Similarly, among DCs, ΔfbpA enhanced the expression of MHC-II, CD1d and CD40 but not CD80 or CD86 (Fig. 8B). These studies suggested that ΔfbpA infected MΦs and DCs have a more activated phenotype than those infected with either BCG or H37Rv.

Fig. 8.

The ΔfbpA mutant induces an enhanced expression of MHC-II and CD1d molecules in MΦs and dendritic cells (DCs).

A. MΦs or BM-derived BMA.A3 cell line MΦs (both showed similar results, MΦs shown) were infected with mycobacteria at an moi of 1:1 for 4 h, washed, incubated for 24 h and stained for surface receptor expression using specific antibodies conjugated to fluorochromes. Receptor expression was analysed using BD-Facscan and Cellquest software. MΦ histograms (one of three experiments illustrated) show the unstained or isotype stained cells (red fill); ΔfbpA (green); H37Rv (blue) and BCG (yellow) infected MΦs. Histograms illustrate that MΦs infected with ΔfbpA show an enhanced expression of CD40, CD86 and MHC-II when compared with BCG- or H37Rv-infected MΦs. Mean fluorescence intensity values (MFIs) from triplicate separate experiments are shown below the histograms (*P < 0.01; **P < 0.007; ***P < 0.008; #0.01 compared with H37Rv- or BCG-infected MΦs, Student's t-test).

B. BM-derived CD11c bead purified DCs were infected and analysed for receptor expression as above. Histograms illustrate that ΔfbpA (blue) induces a stronger expression of MHC-II in DCs than BCG (black) or H37Rv (blue). MFIs from three independent experiments are shown below the histograms (*P < 0.009; **P < 0.009 compared BCG- or H37Rv-infected DCs, Student's t-test).

Antigen presentation by APCs infected with mycobacteria

Phagosome maturation and fusion with lysosomes has been suggested to be a prelude to microbial degradation, loading of peptides to MHC-II and antigen presentation (Harding, 1991; Levine and Chain, 1991). As ΔfbpA phagosomes showed signs of enhanced maturation and activated receptor expression, experiments were performed to determine the efficacy of antigen presentation by ΔfbpA infected MΦs and DCs. We previously showed that MΦs infected with ΔfbpA triggered IFN-γ response from unpurified T cells from Mtb sensitized mice (Copenhaver et al., 2004). In this study, we compared ΔfbpA infected MΦs and DCs for their ability to prime using CD3-bead purified T cells. ΔfbpA infected MΦs and DCs induced a stronger IFN-γ response from Mtb sensitized T cells from mice (supplemental data). Figure 9 illustrates the levels of IFN-γ released by sensitized T cells over 72 h when primed from either MΦs (Fig. 9A) or DCs (Fig. 9B). The wild-type H37Rv induced a poor IFN-γ response over the 3 day incubation while BCG showed a reduced ability to induce IFN-γ compared with ΔfbpA infected APCs. Furthermore, IL-12p70 levels were significantly higher in ΔfbpA APC–T cell co-cultures than either BCG or H37Rv co-cultures (Fig. 9C and D). Increased levels of IL-12 found in latter cultures is consistent with our previous observation that ΔfbpA enhanced IL-12 and IFN-γ mRNA expression in mouse lungs during acute infection (Copenhaver et al., 2004). Together, these data suggest that ΔfbpA drives a stronger Th1 response in mice compared with BCG or H37Rv.

Fig. 9.

ΔfbpA is more immunogenic in MΦs or DCs than H37Rv or BCG. Antigen-presenting cells (APCS) such as MΦs (A, C) or DCs (B, D) were infected with mycobacteria for 4 h, washed and co-cultured with naïve or sensitized CD3-bead purified T cells from spleens of mice immunized with Mtb. After 72 h, the culture supernatants were assayed for IFN-γ (A and B) or IL-12p70 (C,D) using sandwich ELISA.

A and B. ΔfbpA induces a stronger response of IFN-γ in immune T cells (shown as bars) than BCG or H37Rv within MΦs (A) or DCs (B) (P-values determined using Student's t-test). Neither naïve T cells co-cultured with infected APCs nor uninfected APCs secrete significant levels of IFN-γ. The IFN-γ levels for these controls are represented as horizontal dotted lines.

C and D. ΔfbpA infected MΦs (C) or DCs (D) show elevated levels of IL-12p70 compared with BCG or H37Rv. Basal levels of IL-12p70 in uninfected in APCs or naïve T cell co-cultures are shown as horizontal dotted lines.

Discussion

Mtb is internalized by MΦs through various receptors that may influence the fate of organisms (Ernst, 1998; Blander and Medzhitov, 2004). We found earlier that ΔfbpA mutant and wild-type Mtb differed in growth within MΦs although both were phagocytosed by MΦs equally well (Copenhaver et al., 2004). In this study, we examined the molecular basis for the attenuation of ΔfbpA mutant that lacks the antigen 85A-mycoloyl transferase. ΔfbpA had reduced levels of TDM that led to a decreased synthesis of SOCS-1 in MΦs. SOCS-1 has emerged as a major regulator of cytokine-mediated antibacterial activity, TLR mediated signalling via IFN-γ in MΦs and maturation of DCs (O'Keefe et al., 2001; Imai et al., 2003; Bartz et al., 2006). Consistent with this observation, MΦs activated even after infection with mycobacteria were better able to kill ΔfbpA than either BCG or wild-type H37Rv. This suggesting that ΔfbpA mutant was less suppressive for MΦs through reduced induction of SOCS-1.

Additional studies focused on mechanisms through which MΦs were able to contain ΔfbpA in comparison with BCG and H37Rv. ROS and NO constitute the major anti-mycobacterial mechanisms of MΦs and the growth of ΔfbpA mutant was enhanced by the blockade of ROS and NO, respectively, by DPI and NMMA (Fig. 3C). Interestingly, blockade of ROS enhanced the growth of ΔfbpA but not that of BCG or H37Rv while all three strains were affected by the blockade of NO. An additional explanation for the enhanced susceptibility of ΔfbpA to ROS was suggested by the in situ localization studies of oxidant enzymes around phagosomes in MΦs. Phox assembles from several different components (p47^phox, p67^phox, gp91^phox and others) during the invagination and formation of the phagosome in MΦs and neutrophils (Babior, 1999). Phagosomal membrane phox then secretes copious amounts of superoxide into the lumen of the phagosomes which spontaneously dismutates into hydrogen peroxide (H₂O₂) (Suh et al., 2006). The latter can in turn yield more bactericidal species such as hydroxyl radical and hypochlorous acid depending upon the iron and chloride ions available within the phagosomes. Our studies showed that ΔfbpA colocalized more with phox enzyme and therefore, we suggest that ΔfbpA is more susceptible to ROS (Figs 3 and 4). In support of this, virulent H37Rv phagosomes were found to exclude phox in MΦs (Fig. 4) (Daniel et al., 2006), and superoxide dismutase defective strains of H37Rv were found more susceptible to oxidants (Edwards et al., 2001). Similar considerations appear to apply to the increased tendency of ΔfbpA phagosomes to colocalize with iNOS as NO is a more powerful antimyco-bacterial agent (Fig. 4). In summary, ΔfbpA appears more susceptible to intracellular oxidants.

Recent studies show that cell wall lipids make a difference in the ability of mycobacteria to traffic within MΦs (Chua and Deretic, 2004; Chua et al., 2004). Lipoarabiomannan has been found to interfere with phagosome maturation while TDM coated liposomes were reported to repulse each other in vitro alluding that it may affect endosome fusion within MΦs (Spargo et al., 1991). Interestingly, a cord forming trahalose lipid defective mutant of M. marium was also reported to be defective in phago-some maturation (Robinson et al., 2007). As ΔfbpA was TDM deficient, these considerations prompted us to examine its intracellular trafficking pattern. Traditionally, phagocytosed bacteria in MΦs are internalized into phagosomes that fuse with various types of early and late endosomes that deliver various cellular cargo and ultimately the phagosome fuses with lysosomes where bacterial degradation takes place owing to proteases and lipases (Gabay et al., 1986) (Levine and Chain, 1991). Armstrong and Hart made the historical observation of that wild-type Mtb avoids destruction by avoiding P-L fusion that has been confirmed by others using wild-type H37Rv as well as BCG (Armstrong and Hart, 1971; McDonough et al., 1993; Clemens and Horwitz, 1995; Russell, 1995; Deretic and Fratti, 1999; Pieters, 2001; Kusner, 2005). However, Mtb and BCG phagosomes retain the selective ability to fuse with early endosomes acquiring molecules such as transferrin, transferrin receptor, organelles as well as MHC-II molecules (Clemens and Horwitz, 1996; Sturgill-Koszycki et al., 1996; Russell, 2003). In addition, they acquire glycosphingolipids from MΦ plasma membrane through recycling endosomes. Finally, proteolytic enzymes such as Cathepsin-D (Cat-D) are delivered to their phagosomes although, owing to the exclusion of vacuolar protonATPase (vATPase) by Mtb, BCG and M. avium phagosomes, Cat-D remains inactive (Ullrich et al., 1999).

These observations indicate that Mtb and BCG phagosomes fuse selectively with early endosomes to derive nutrients and organelles for growth but do not fuse with lysosomes to evade killing through degradation. Thus the ‘maturation arrest’ of Mtb and BCG facilitates their survival within MΦs. Using fluorescent endocytic tracers (LTR, TRR and MBR) we initially found that ΔfbpA phagosomes fused with early endosomes acquiring transferrin. However, unlike H37Rv or BCG, their phagosomes also showed an enhanced capacity to fuse with late endosomes containing either LTR or MBR probes. This suggested that the mutant was more fusion competent. We next investigated the molecular basis for enhanced endosomal fusion of ΔfbpA phagosomes and characterized whether P-L fusion occurred as a consequence.

Studies show that several rab GTPases regulate endosomal trafficking and early endosomes carry the rab5 GTPase, recycling endosomes label for rab11 while, lysosomes label for rab7 and CD63. A number of downstream effector proteins such as early embryonic antigen 1 (EEA1), a protein that binds rab5 and triggers the recruitment of phosphatidyl inositol-3 kinase (PI-3K), and Hepatocyte growth factor-regulated tyrosine kinase substrate (Hrs) are also involved in endosome fusion. Interestingly, Mtb and BCG phagosomes acquire rab5, an early endosomal docking protein while excluding rab7, a marker of lysosomes (Deretic et al., 1997; Via et al., 1997). More recent studies show that rab5 to rab7 conversion is defective on Mtb and BCG phagosomes preventing lysosomal fusion (Rink et al., 2005). The defect in maturation also extends to the exclusion of downstream docking and sorting proteins. Thus, EEA1 is absent and levels of phosphatidyl inositol phosphate are reduced on Mtb phagosomes, and Hrs is defectively recruited by mycobacteria. In contrast, these markers are recruited actively and are abundant on inert latex bead phagosomes (Vergne et al., 2004b; Vieira et al., 2004; Deretic et al., 2007).

The differential presence of rab and other docking proteins on early and recycling endososomes and lysosomes prompted us to analyse ΔfbpA, H37Rv and BCG phagosomes in MΦs. ΔfbpA phagosomes acquired rab5A, rab5B as well as rab11 that appeared to explain their enhanced ability to fuse with ‘early’ and ‘recycling’ endosomes respectively (Vieira et al., 2002). They also acquired the protein tyrosine kinase (Hck) a marker of late endosomes that seems to explain their ability to fuse with late endosomes (Astarie-Dequeker et al., 2002). Paradoxically, ΔfbpA phagosomes consistently excluded rab7 and CD63 markers of lysosomes. Thus, ΔfbpA phagosomes differed from H37Rv and BCG in being able to fuse with early, recycling as well as late endosomes but resembled them by avoiding lysosomal fusion. Together, these studies show that ΔfbpA phagosomes are more fusion competent than either BCG or H37Rv, have more maturation markers and are more susceptible to intracellular oxidants, attributing them a unique phenotype.

As ΔfbpA mutant was effective as a candidate vaccine, we investigated whether these features translated into an increased efficiency of antigen presentation (Copenhaver et al., 2004). Antigen presentation consists of an intracellular component where bacteria are degraded by proteases such as cathepsins in an acidic environment and the membrane expression of the MHC-II bound peptides in conjunction with costimulatory and adhesion molecules (Levine and Chain, 1991). The MHC-II forms an immunological synapse along with CD80, CD86 and CD40 to optimally present peptides to T cells (Harding, 1991; Levine and Chain, 1991). ΔfbpA enhanced the expression of MHC-II and costimulatory molecules on both MΦs and DCs. As SOCS-1 has been found to downregulate the MHC-II expression of MΦs, we suggest that the ability of ΔfbpA to enhance MHC-II expression is related to its reduced induction of SOCS-1 (Fig. 2). Finally, ΔfbpA infected MΦs and DCs were more efficient to induce a stronger recall IFN-γ response in the immune T cells of mice. We suggest that this was due to an increased expression of MHC-II as well as costimulatory molecules such as CD40 and CD86.

The unique behaviour of ΔfbpA in MΦs raised interesting questions on the essentiality and function of the genes encoding Ag85A, Ag85B and Ag85C. Following our initial report that fbpA disruption resulted in an attenuated phenotype, others showed that disruption of the gene encoding Ag85C (fbpC) also resulted in changes in cell wall permeability and fatty acid composition (Puech et al., 2002). Subsequently, Ronning et al. (2004) reported that despite similarity in function, the amino acid residues near the substrate binding sites of Ag85 complex are different suggesting that all three were required for the optimal synthesis of TDM. We suggest that the deletion of fbpA led to a deficiency in TDM levels perhaps because its function could not be complemented by either fbpB or fbpC and this, in turn, led to its attenuation. That ΔfbpA is more essential for the survival of Mtb is also supported by the observation that antisense oligonucleotides against the fbpA mRNA inhibited Mtb growth in broth to a greater extent than oligonucleotides against mRNA from fbpB or fbpC (Harth et al., 2002). Together, these studies indicate that fbpA is more essential for the virulence of Mtb despite being an important antigen inducing T cell responses (Ramachandra et al., 2001). Furthermore, it is important to note that the fbpB-encoded Ag85B is a more significant immunodominant antigen of Mtb (Launois et al., 1994). Initially identified to contain the immunodominant Th1 epitope ‘peptide 25’, Ag85B is an integral component of multiple successful DNA vaccines including those under clinical trials (Skeiky and Sadoff, 2006). Thus, we suggest that deletion of fbpA appears to have led not only to attenuation but also enhanced immunogenicity because of an altered intracellular behaviour in APCs and intact Ag85B. In conclusion, our studies suggest that the mechanisms through which a vaccine is processed in APCs may ultimately determine their efficacy and candidate antituberculosis vaccines perhaps need to be optimized for efficient processing by APCs.

Experimental procedures

Bacteria

The wild-type M. tuberculosis H37Rv and Mycobacterium bovis BCG (#35734) were from ATCC repository (MD, USA). The ΔfbpA and the reconstituted ΔfbpA mutant were prepared as described before (Armitige et al., 2000). The green fluorescent protein (gfp)-expressing strains of ΔfbpA, H37Rv and BCG were prepared as per previously described techniques (Dhandayuthapani et al., 1995). They were grown once in bulk in 7H9 broth without additives (H37Rv and BCG), Middlebrook 7H9 broth with 25 μg ml^–1 kanamycin (Dfbp) and 7H9 broth with 25 μg ml^–1 kanamycin and 50 μg ml^–1 hygromycin (gfpΔfbp). After 7–10 days, bacteria were harvested, washed, sonically dispersed and stored in aliquots at –70°C. Infections were performed with freshly re-grown, washed, log phase organisms (day 7–10) generated from these single use seed aliquots. For fluorescence microscopy, gently sonicated single cell suspensions were used at an multiplicity of infection (moi) of 1:1. It should be noted that phagocytosis of multiple bacteria into MΦs did not affect their colocalization pattern unless they were large clumps.

Radiolabelled acetate incorporation assay for determination of cell wall TDM

Metabolic radiolabelling of Mtb and mutants

ΔfbpA mutant, the reconstituted ΔfbpA mutant and H37Rv strains were grown with continuous shaking for 7 days at 37°C (to late log phase) in 7H9 broth (20 ml) supplemented withADC enrichment containing 0.2% glucose and 1% glycerol. An aliquot of 5 ml was removed, turbidity adjusted to McFarland #1 standard and 40 μCi of sodium ¹⁴C-acetate (57 mCi mmol^-1) (Amersham Pharmacia) added to tubes which were incubated for 3, 6, 12 and 24 h. Plating of McFarland #1 adjusted culture showed that the start inoculum contained 1 × 10⁸ cfu ml^–1 for the wild-type, ΔfbpA and reconstituted ΔfbpA. After radiolabelling, the cultures were harvested by centrifugation (5000 r.p.m. for 10 min) and the bacterial pellets were washed in cold PBS (three washes by centrifugation) and ethanol.

Extraction and analysis of radiolabelled TDM

The bacterial pellets were then extracted four times with 1 ml portions of chloroform-methanol (CM) 2:1. After centrifugation, the upper liquid phases (containing cell wall-bound mycolic acids) were collected. Extracts of the upper phases were pooled and dried in glass vials. These ¹⁴C-acetate incorporated glycolipids were dissolved in 200 μl of CM (2:1), radioactivity determined by scintillation counting, and analysed for presence of radiolabelled TDM by TLC, autoradiography and phosporimage analysis.

TLC of TDM bands

The TDM mixtures were separated by TLC on silica gel 60 plates in solvent C. For radiolabelled glycolipid samples, about 40 000 c.p.m. per lane were applied. TLC standards of TDM (20 μg, Sigma Chemical, MO) were cochromatographed with the radiolabelled glycolipid samples. After separation, the TLC plates were allowed to dry, exposed to a scanning Storm phosphorimager, and radioactivity quantified by the use of ImageQuant software. For autoradiography, the TLC plates were exposed to photographic films for 24–48 h.

MΦs and DCs

C57BL/6-derived MΦs were cultured in McCoy's medium with 10% FBS, penicillin and gentamicin (medium) supplemented with 10 ng ml^–1 recombinant mouse GM-CSF (Cell Sciences). MΦs were rested in GM-CSF-free medium for 2 days and used for mycobacterial infections. C57BL/6 bone marrow-derived macrophage-like cell line BMA.A3 was kindly provided by Dr Kenneth L. Rock (University of Massachusetts Medical School). This cell line maintained in the same medium, has been well characterized to study phagocytosis and antigen presentation by MTB (Teitelbaum et al., 1999). Primary and cell line MΦs were maintained as monolayers in 24-well plates for growth curve determinations or oxidant assays, eight-well slide chambers for fluorescence microscopy or in 75 ml flasks for preparation of phagosomes during phagosome maturation studies. DCs were cultured from mouse bone marrow cells grown in GM-CSF and IL-4 containing medium for 7 days and CD11c+ cells purified using anti-CD11c-coated magnetic beads (Miltenyi, USA) as described by the manufacturer. The cells were more than 97% pure and contained a mix of CD8α+ and CD8α– DCs as described (Moulton et al., 2007).

SOCS-1 signalling in MΦs after infection

MΦs were infected with mycobacteria at an moi of 1:1 and washed after 1 and 4 h. Lysates were estimated for protein content, mixed with an antiprotease mix and equal volumes loaded per lane and analysed by Western blot against a SOCS-1-specific antibody (Santa Cruz Biotechnology, USA). Bands were visualized using chemiluminiscence. MΦs were also activated with TDM suspension in mineral oil suspension or vehicle alone to determine SOCS-1 induction. Blots were probed with an antibody to β-actin to ensure equal protein load in lanes. MΦs were separately stained using acid fast stain to confirm that they contained approximately equal numbers of intracellular mycobacteria (not shown).

Growth curves for Mtb or BCG strains in MΦs

These were carried out as described before (Copenhaver et al., 2004; Daniel et al., 2006) using organisms to infect naïve MΦs or BMA.3 cells (both yield similar results, data not shown) in 24-well format at an moi 1:1 and plating cfu on days 0, 3 and 7 post infection. To study the effect of MΦ activation on growth curves, in addition to naïve MΦs, replicates of MΦ cultures were activated with 400 U ml^–1 of IFN-γ for 24 h prior to infection and cfu counts. MΦs were fully viable (> 90%) until day 7 as tested by alamar blue conversion method (Benghuzzi, 1995). The blockade of ROS and NO was achieved by incubating infected MΦs with 10 μM DPI a non-scavenging NADPH oxidase inhibitor or 50 mM NMMA an inhibitor of NO synthesis, or their combination (Lee et al., 2005; Wang et al., 2006). These doses were based on initial dose–response titrations to determine inhibition of ROS and NO respectively.

Localization of phox proteins and iNOS on phagosomes within MΦs using immunofluorescence

MΦs were infected with live gfpΔfbpA, gfpH 37Rv or gfpBCG using eight-well chambers and stained for colocalization with goat antibodies to mouse p47^phox (Santa Cruz Biotechnology, CA) or a mouse monoclonal antibody to mouse iNOS (Sigma Chemical, MO) using the protocol described earlier (Daniel et al., 2006). Cells were mounted in Elvanol mountant and examined using a Deltavision LSM to confirm colocalization patterns and analysis of the staining patterns of anti-p47^phox or iNOS antibodies. A positive control of phorbol myristyl acetate activated uninfected MΦs and a negative control consisting of uninfected, untreated and rested MΦs were stained with anti-p47^phox antibodies for each type of MΦ and time point analysed. In addition, antibody isotype controls were included using gfpMTB-infected MΦs stained with normal goat or mouse IgG followed by Texas redconjugated antispecies IgG. Unactivated or uninfected MΦs presented a uniformly low background staining with this procedure. phox-specific staining on phagosomes was evident as dense tubular or spherical staining, strongly colocalizing with phagosomes (Daniel et al., 2006). INOS-specific staining with the mab was more granular and localized in cytoplasm or around phagosomes. Per cent colocalization of gfp-mycobacteria with either phox or iNOS was determined by counting at least 100 fields of a quadruplicate slide chamber per mouse bone marrow-derived preparation and averaging the numbers. All scoring was blinded. The standard deviations were calculated from three independent experiments using MΦs from three mice in each experiment and P-value determined for colocalization scores using the t-test. It should be noted that after gentle sonication we obtained mostly a single cell suspension. However, occasional clumps of bacteria in MΦs were unavoidable and we eliminated such clumps from being scored for colocalization purposes. Furthermore, MΦ monolayers were washed four times with media to ensure that extracellular bacteria were not a problem. In our hands, when moi was below 1:5 such bacteria were rare.

Fluorescent immunostains for early, recycling and late endosome-lysosome markers on phagosomes

The immunostaining of MΦs infected with gfp-mycobacteria was carried out similar to the procedure described for phox proteins. The antibodies used were rab5A, rab5B, rab11, rab7, CD63, Hck and PI-3 kinase (Santa Cruz Biotechnology, CA). LAMP-1 (ID4B clone) was from the Developmental Hybrodoma Bank, IA).

Measurement of oxidant activity

Intracellular ROS using flow cytometry

BMA.A3 MΦs were infected with mycobacteria at an moi of 1:1 and washed after 4 h. The cells were suspended in 5 μM H₂DCFDA for 5 min, and analysed for intracellular fluorescence using a Cellquest software and BD-Facscan instrument. The non-fluorescent H₂DCFDA probe is cleaved within living cells by esterases and the resultant DCF is oxidized to DCFH by ROS produced within MΦs yielding a green fluorescent compound, which in turn, is trapped within cells by hydrophobic binding to cytoplasmic proteins. Histograms were analysed for MFIs in three independent experiments and P-value calculated using Student's t-test.

Quantitation of ROS using fluorometry

Superoxide generated by phox dismutates to yield H₂O₂ which can oxidize H₂DCFDA to a fluorescent product which can then be measured by fluorometry (Daniel et al., 2006). BMA.A3 cells in 24 wells (10⁶ cells per well; three wells were used for each strain) were infected with myco-bacteria as above for 4 h and washed. They were added with 100 μM H₂DCFDA and fluorescence in average fluorescence units (AFUs) quantified using an Ascent fluoroscan. AFUs from infected as well as control MΦs in three independent experiments were plotted against time.

Quantitation of NO using colorimetry

Nitrite derived from NO in medium of BMA.A3 MΦs infected as per ROS detection was quantified using Griess reagent and spectrophotometry as described before (Jagannath et al., 1998).

Phagosomal acquisition of late endosomal markers

LTR labelling of late endosomes

LTR (DND99; Molecular probes Inc, Eugene OR) localizes in LAMP-1 positive late endosomes which do not fuse with live MTB containing phagosomes (Via et al., 1998). Colocalization of green Mtb with red coloured LTR thus suggests phagosome-late endosome fusion. MΦs or BMA.A3 MΦs (both gave similar results) were preloaded with 1/1000 dilution of LTR for 4 h, washed and infected with gfp-mycobacterial strains for another 4 h. After another step of washing cells were incubated for and fixed cells were then examined at 24 h for phagosomal labelling of LTR using LSM.

MBR labelling of late endosomes

The late endosomes were created by endocytosing mannosylated BSA labelled with rhodamine (MBR; Sigma Chemical, MO). MBR enters through the MΦ mannose or lectin type receptors and matures from early to late endosomes in about 2 h (Astarie-Dequeker et al., 1999). The latter have been reported to fuse with MTB phagosomes after activation with IFN-γ although live virulent Mtb has been reported to exclude this marker in unactivated MΦs (Kang et al., 2005). MΦs were labelled with MBR for 2 h, washed twice with warm medium and infected with gfp-mycobacterial strains for 4 h before being washed again three times and re-incubated for another 24 h. Colocalization with MBR was also determined using LSM.

Labelling of MΦs with TRR

BMA.A3 MΦs were washed with serum-free McCoy medium and incubated with TRR at 5 μg ml^–1 for 90 min. Cells were then washed and chased with complete medium three times for 120 min. MΦs were then infected with gfp-mycobacteria for 4 h, washed and incubated for another 24 h. MΦs were examined as above for colocalization. Per cent colocalization of LTR, MBR and TRR against gfp-mycobacteria was determined as described for phox and iNOS using similar numbers of mice.

Detection of phagosomal membrane proteins by immunoblot analysis

BMA.A3 MΦs were grown as monolayers in 75 ml flasks and infected at moi of 1:1–1:5 with mycobacteria for 4 h, washed and incubated for another 24 h. Mycobacterial phagosomes were fractionated as per the procedures described by Ullrich et al. at the end of these time points (Ullrich et al., 1999; Ullrich et al., 2000). Briefly, 2 h before fractionation, MΦ monolayers were washed once with warm serum-free medium, gently scraped, washed three times in a fractionation buffer (PFB) with 10 mM Hepes, 5 mM EDTA, 5 mM EGTA, pH 7.0 and suspended in PFB with an antiprotease mix consisting of 1 μg ml^–1 leupeptin, 1 μg ml^–1 pepstatin and 1 mM phenylmethyl sulfonylfluoride. Pellets were then homogenized in a glass tissue homogenizer 10 times and passed 10 times through a 28-gauge needle. Lysates were centrifuged at 500 g for 5 min to sediment nuclei and the post nuclear supernatant was layered on a step gradient of 50% and 12% sucrose in PFB. After centrifugation at 1000 g for 60 min, the interphase of phagosome fraction was collected and further purified by passing through two successive cushions of 70 and 400 kDa ficoll in PFB as described before. The final purified phagosomal pellet was collected by centrifugation at 10 000 g for 15 min and suspended in SDS sample buffer and subjected to ultrasonic disruption for 5 min. Latex beads (IgG coated, 2 μM diameter, Sigma chemical, MO) were added into MΦs for 60 min, washed and phagosomes fractionated 4 h after phagocytosis as they mature earlier. Phago-some pellets were heated at 98°C for 5 min were then electro-phoresed using a 4–20% gradient SDS gel, electroblotted and membranes were probed with antibodies as used above for immunostains. Bands were visualized using a chemiluminiscence kit from Amersham. The post-nuclear supernatants obtained from MΦ lysates served as control cytosol preparations and were used to determine the basal level of endosome markers after estimation of protein (Ullrich et al., 1999; Ullrich et al., 2000). All the markers tested for phagosome analysis were found in cytosol of all MΦs infected with mycobacteria, but were present differentially on purified phagosome pellets, suggesting that phagosomal translocation was dependent on the phagocytosed material.

Antigen presentation assay

The methods have been previously described (Copenhaver et al., 2004; Connelly et al., 2007; Moulton et al., 2007). Sensitized and naïve T cells were derived from C57BL/6 mice. One group of mice was left un-immunized (naïve) and another group was immunized via subcutaneous route with one dose of H37Rv given at the base of tail 2 weeks apart. Mice were then treated with an oral dose of INH (25 mg kg^–1) for 10 days and mice were sacrificed 3 days after the last drug dose. Spleen cells were gently teased apart, suspended in ACK buffer to lyse red cells, washed and incubated in RPMI-1640 medium with 10% FBS (medium) for 24 h for monocytes to adhere. CD3+ bead purified T cells were fractionated using magnetic columns (Miltenyi, USA) and layered on APCs. APCs infected with ΔfbpA, H37Rv or BCG were separately overlaid with H37Rv sensitized T cells and the supernatants collected to determine the cytokine levels. To generate APCs, bone marrow-derived adherent MΦs cultured in GM-CSF containing medium for 7 days were infected with H37Rv bacteria or left uninfected as monolayers in 24-well plates. The DCs were obtained by culturing bone marrow cells for 7 days in medium with GM-CSF and IL-4 (10 ng ml^–1 each) and purifying CD11c+ cells through a magnetic column. The cells were routinely 98% pure and were used as such as immature DCs without additional activation (Moulton et al., 2007). The DCs were also infected with mycobacteria at an moi of 1:1. After a 4 h infection, monolayers of MΦs or DCs were washed extensively to remove non-phagocytosed bacteria and overlaid with T cells. For both MΦ and DC co-cultures, T cells were overlaid at a ratio of 1:20. For each type of APCs on the monolayer, triplicate wells were overlaid with T cells from three H37Rv immunized mice and naive splenocytes from three unimmunized mice. Twenty-four, 48 and 72 h after the addition of T cells, 200 μl aliquots of supernatants were collected from each well of the APCs along with control wells leaving behind 400 μl at the end of 72 h when supernatant collections were terminated. Medium was not replenished as no loss of viability was detected among T cells in pilot experiments and viability was measured of the APC/T-cell overlay using Alamar blue viability indicator. IFN-γ or IL-12p70 in the medium was determined using sandwich ELISA kits from R&D (USA).