Cellubrevin Alterations and Mycobacterium tuberculosis Phagosome Maturation Arrest*

The intracellular trafficking processes controlling phagosomal maturation remain to be fully delineated.Mycobacterium tuberculosis var. bovis BCG, an organism that causes phagosomal maturation arrest, has emerged as a tool for dissection of critical phagosome biogenesis events. In this work, we report that cellubrevin, a v-SNARE functioning in endosomal recycling and implicated in endosomal interactions with post-Golgi compartments, plays a role in phagosomal maturation and that it is altered on mycobacterial phagosomes. Both mycobacterial phagosomes, which undergo maturation arrest, and model phagosomes containing latex beads, which follow the normal pathway of maturation into phagolysosomes, acquired cellubrevin. However, the mycobacterial and model phagosomes differed, as a discrete proteolytic degradation of this SNARE was detected on mycobacterial phagosomes. The observed cellubrevin alteration on mycobacterial phagosomes was not a passive event secondary to a maturation arrest at another checkpoint of the phagosome maturation pathway, since pharmacological inhibitors of phagosomal/endosomal pathways blocking phagosomal maturation did not cause cellubrevin degradation on model phagosomes. Cellubrevin status on phagosomes had consequences on phagosomal membrane and lumenal content trafficking, involving plasma membrane marker recycling and delivery of lysosomal enzymes. These results suggest that cellubrevin plays a role in phagosomal maturation and that it is a target for modification by mycobacteria or by infection-induced processes in the host cell.

The mechanisms of phagosomal biogenesis and maturation into the phagolysosome represent a fundamental biological process reflecting regulatory aspects of membrane trafficking and organelle biogenesis in eukaryotic cells (1)(2)(3)(4)(5)(6)(7)(8)(9). These processes are sometimes targeted by a class of microbes capable of affecting phagosome integrity or maturation (8, 10 -19). Interference with the normal pathway of phagosomal maturation into the phagolysosome enables such intracellular pathogens to avoid microbial growth control or killing by the host phagocytic cells or to avert efficient antigen presentation to the effectors of adaptive immune response in the host (1,3,7,20,21). Consequently, investigations of microbial interference with the trafficking processes in the host cells may provide a means for dissecting the phagosomal maturation pathway. In this context, we have been using Mycobacterium tuberculosis, a highly adapted pathogen that parasitizes macrophages, as a tool to study fundamental phagosome maturation processes (reviewed in Ref. 22). It has been established that M. tuberculosis, M. tuberculosis var. bovis BCG, and Mycobacterium avium reside in specialized phagosomes that afford protection from acquisition of the characteristics of the terminal lysosomal organelles (23)(24)(25)(26), a phenomenon termed in classical microbiological literature as the inhibition of phagosome-lysosome fusion (27). The unique distribution of markers on M. tuberculosis phagosomal compartments (MPC) 1 suggests that there may be discrete alterations in MPC interactions with other intracellular organelles (22). Mycobacterial phagosomes display reduced clearance of early phagosomal proteins such as TACO (28), better known as Coronin (22,28), and several plasma membrane markers (24); they manifest impaired acidification due to the paucity of H ϩ ATPase (29); and they show limited acquisition of late endosomal/lysosomal proteins (23,30,31).
Membrane trafficking, organelle biogenesis, and maintenance of compartmental integrity are highly regulated processes and depend on a multicomponent vesicle docking and fusion machinery. The basal apparatus needed in the final stages of membrane fusion is composed of SNARE proteins forming a four-helix trans-SNARE bundle contributed by the donor and acceptor membranes (32). Prior to a fusion event, the pre-existing cis-SNARE complexes are disassembled and primed by the action of the ATPase NSF and ␣ϪSNAP, preparing the fusion proteins for the formation of trans-SNARE pairing (33,34). Purified, fusion competent SNAREs (32,(35)(36)(37)(38), in combination with other tethering and regulatory factors such as Rab GTPases and their effectors (39 -43), define the permissive fusion events in vivo.
In this work, we extended our analyses of the phagosomal organelles to the v-SNARE cellubrevin. Cellubrevin (also known as VAMP3) is a SNARE molecule involved in the recycling of plasma membrane markers from the endosome (44) and has been implicated in trafficking from the TGN to the endocytic pathway through interactions with the TGN SNARE Syntaxin 6 (45). Here we report the degradation of cellubrevin on the M. tuberculosis phagosome, as a mechanism contributing to altered trafficking from and to the mycobacterial phagosome.

EXPERIMENTAL PROCEDURES
Cell and Bacterial Culture Conditions-The murine macrophage-like cell line, J774, was maintained in Dulbecco's modified Eagle's medium (BioWhittaker) supplemented with 4 mM L-glutamine and 5% fetal bovine serum (HyClone). M. tuberculosis var. bovis BCG Pasteur (BCG) harboring phsp60-gfp was grown in Middlebrook 7H9 broth (46). Single cell suspensions were generated using a Tenbroek tissue grinder, followed by a 5-min pulse in a water bath sonicator. Remaining bacterial aggregates were removed by low speed centrifugation.
Phagosome Purification-For experiments examining brief infection periods (Ͻ1 h), infections were synchronized as described previously (8). Synchronization of latex beads and BCG phagocytosis was achieved by allowing latex beads or BCG to attach to macrophages cooled to 4°C without allowing uptake. Samples were shifted to 37°C for various periods of time and phagosomes isolated as described previously. For longer phagosome chase experiments, BCG or latex beads were taken up by macrophages for 1 h, followed by removal of unphagocytosed particles and fresh media added for 1 to 24 h (19). Phagosomes were isolated and characterized for purity as described previously (19). Briefly, cells were mechanically lysed in 8.5% sucrose (w/w in homogenization buffer: 1 mM HEPES, pH 7.0, 0.5 mM EGTA, 1 mM EDTA, 1 mg/ml bovine gelatin) supplemented with protease inhibitors. After cellular debris and nuclei were removed by low speed centrifugation to generate post-nuclear supernatants (PNS), PNS from different samples were sedimented by centrifugation through a sucrose step gradient of 8.5, 15, and 50% sucrose (w/w in homogenization buffer). The sediment collected from the 15/50% interface was loaded on a linear 32-53% sucrose gradient (w/w in homogenization buffer). After isopycnic separation of organelles, fractions were collected from the top of the gradients, and membranes were pelleted. The fraction containing purified MPC were characterized for purity as described previously (19). For latex bead phagosome purification, J774 cells were infected with latex beads and incubated as indicated above. Latex bead phagosomal compartments (LBC) were isolated from PNS by flotation and characterized for purity as described previously (47).

Modifications of the SNARE Cellubrevin on Mycobacterial
Phagosomes-Purified phagosomes containing M. bovis BCG (BCG) or latex beads were prepared and characterized as described previously (8,19) and examined for the presence and dynamics of endosomal SNAREs. The most striking observation was made with the v-SNARE cellubrevin (VAMP3). Purified LBC were found to contain and accumulate cellubrevin over time (Fig. 1). Surprisingly, purified MPC, while containing cellubrevin (14 kDa), also displayed a lower molecular mass band (12.2 kDa) ( Fig. 1A) that reacted with the cellubrevin antibody directed against the cytoplasmic N terminus of cellubrevin (44). The identity of the lower band as a VAMP was confirmed (Fig. 1B) by using the antibody anti-VAMP c10.1, which recognizes the cytoplasmic ␣-helix of VAMP3 (cellubrevin), as well as VAMP1 and VAMP 2 (51). We conclude that the lower molecular mass polypeptide corresponds to a removal of a 1.8-kDa fragment from the C terminus of cellubrevin. The truncated polypeptide, referred to as ⌬cellubrevin, was not present on LBC.
A decrease in the amount of full-length cellubrevin was seen in PNS from BCG-infected cells with the effect depending on the multiplicity of infection (Fig. 1, C and D). Cellubrevin levels were significantly reduced in macrophages infected at a multiplicity of infections (m.o.i.) of 50 or 100 bacilli per cell ( Fig. 1, C and D) (p Ͻ 0.005 for m.o.i. 50 ; p ϭ 0.0001 for m.o.i. 100 , relative to maximum cellubrevin levels; ANOVA). Actin levels were unaffected at all m.o.i. values (Fig. 1D). ⌬Cellubrevin in PNS observed from BCG-infected cells could only be detected when membranes were pelleted from PNS and overloaded on SDS-PAGE gels (data not shown).
⌬Cellubrevin Generation Is Limited to Mycobacterial Phagosomes-To test whether BCG infections can affect cellubrevin in trans, e.g. on latex bead phagosomes in the same macrophages, cells were infected with BCG for 24 h prior to superinfecting with latex beads for an additional 24 h after which LBC were purified as described under "Experimental Procedures." We found that conditioning cells with BCG at levels that did not reduce overall cellular cellubrevin levels in infected cells (Fig. 2, PNS), did not significantly alter cellubrevin levels on LBC (Fig. 2, LBC). Furthermore, ⌬cellubrevin was absent from LBC isolated from BGC-conditioned macrophages (Fig. 2). This suggests that the effects of BCG on cellubrevin are not exerted globally within BCG-infected macrophages, but represent a phenomenon specific for the mycobacterial phagosome.
Pharmacologically Induced Stagnation of Phagosomal Maturation Does Not Affect Cellubrevin Levels on Phagosomes-We next examined whether the generation of ⌬cellubrevin on MPC is only a secondary phenomenon caused by phagosome maturation arrest or could actively contribute to the arrest. This was addressed by inducing a phagosome maturation block using pharmacological inhibitors of membrane trafficking. We found that treatment with bafilomycin A and nocodozole, two known inhibitors of cellular functions essential in endosome maturation (acidification and microtubule based vectorial transport, respectively) (52, 53), did not cause reduction in cellubrevin levels on LBCs (Fig. 3A). Thus, induction of a maturation block alone, by causing stagnation of latex bead phagosomes, did not suffice to cause the appearance of ⌬Cellubrevin on LBC (Fig.  3A) or reduction of cellubrevin levels in PNS.
Brefeldin A Reduces Cellubrevin Levels on Phagosomes but Does Not Affect Cellubrevin Integrity-In contrast to bafilomycin A and nocodozole, treatment with brefeldin A (BFA), an inhibitor of ARF GTPase-based vesicular trafficking within the secretory pathway, including the trafficking from the TGN, decreased levels of cellubrevin on LBC (Fig. 3, A and B). The effect was specific, as levels of SNAP 23 on LBC were not affected in BFA-treated cells (Fig. 3A). Cellubrevin levels were reduced on LBC by 75% in cells treated with BFA (p ϭ 0.0001; ANOVA) (Fig. 3B). However, a cellubrevin degradation product was not detected on latex bead phagosomes in BFA-treated cells ( Fig. 3A; ⌬cellubrevin on MPC is shown for reference). Although cellubrevin levels were significantly reduced on LBCs in BFA-treated macrophages, overall cellular cellubrevin levels were not affected ( Fig. 3A; PNS). It is possible that BFA effects on phagosome-associated cellubrevin were due to either its decreased delivery or increased removal from phagosomes. We favor the former possibility based on the following considerations: (i) cellubrevin has been implicated in TGN-to-endosome trafficking (45); (ii) it has been shown that the TGN t-SNARE syntaxin 6 can interact with cellubrevin (45); (iii) we have also found that syntaxin 6 is excluded from MPC, a phenomenon linked to the block in delivery of lysosomal effectors (e.g. H ϩ ATPase and immature cathepsins) to phagosomes. 2 As BFA has been shown to cause the tubulation and fusion of endosomes and TGN (48), this opens the theoretical possibility that BFA-mediated alterations in the protein profiles of LBC may be linked to mixing of compartments. However, since we observed a reduction rather than increase in phagosomal acquisition of cellubrevin, we can exclude the possibility that BFA in our experiments caused indiscriminant fusion of phagosomes with endosomes or TGN. Instead, the observed reduction of cellubrevin acquisition by LBC in BFA-treated cells is most likely associated with the disruption of phagosomal interactions with the endosomal and biosynthetic compartments.
Alterations in Cellubrevin Levels Correlate with Altered Recycling from Phagosomes-We next tested whether alterations in cellubrevin levels affected recycling of markers from the phagosome. The effect of BFA (which lowers cellubrevin on phagosomes; Fig. 3) on recycling of transferrin receptor (TfR) from model phagosomes was examined. LBC were isolated from BFA-treated macrophages at 30 min post-phagocytosis. LBC from BFA-treated cells retained significantly more TfR relative to LBC from untreated control macrophages (p Ͻ 0.05; ANOVA) (Fig. 4, A and B). BFA treatment also caused increased levels of AP3 (p Ͻ 0.05; ANOVA), an adapter protein involved in TGN and endosomal trafficking (54) (Fig. 4, A and  B). Similarly, levels of coronin, an actin-binding, general phagocytosis protein (also termed TACO, proposed to accumulate on mycobacterial phagosomes (28)), was moderately increased on LBC upon BFA treatment (p Ͻ 0.01; ANOVA) (Fig. 4, A and B). In contrast, levels of the early endosomal SNARE syntaxin 13 were not affected by BFA, suggesting that the effects seen with TfR, AP3, and coronin (Fig. 4, A and B) were not a result of indiscriminate membrane mixing.
MPC Retain Increased Levels of the ATPase NSF Involved in SNARE Priming-After a round of membrane fusion is completed, SNARE proteins remain tightly associated in a cis-SNARE complex. SNARE molecules stored in cis complexes are made available for subsequent fusion events only upon a priming cycle, which involves an ATP-dependent disruption of the 2 R. A. Fratti and V. Deretic, unpublished results. cis-SNARE aggregate. The priming process is dependent on the ATPase NSF association with SNARE complexes, via interactions that include another factor, ␣-SNAP (55). Upon ATP hydrolysis, the cis-SNARE complex is dissociated, and the primed free SNARE proteins become available for new docking and fusion events. Isolated phagosomes were probed for the presence of NSF and ␣-SNAP. Both MPC and LBC displayed membrane bound NSF and ␣-SNAP throughout the experiment with some evidence of reduced amounts as the maturation progressed (Fig. 5A). When NSF was examined at 24 and 72 h post-infection, we found that MPC retained significant levels of NSF, while most of it was lost from LBC (Fig. 5B).
LBC isolated from BFA-treated cells showed an increase in NSF levels (200%) relative to those isolated from untreated cells (p Ͻ 0.05; ANOVA) (Fig. 6, A and B). Inhibiting H ϩ ATPase activity with bafilomycin A had a similar effect (p Ͻ 0.05; ANOVA) (Fig. 6, A and B). Nocodozole did not have an effect on NSF increase, and, if anything, it caused a slight reduction in NSF levels. Accumulation of NSF on LBC isolated from BFA-treated macrophages (Fig. 6) correlates with a marked reduction in cellubrevin levels (Fig. 3), while retention of NSF on MPC (Fig. 5) correlates with cellubrevin degradation. We interpret these observations as an impeded release of NSF from MPC due to disfunctional SNARE priming associated with either reduced levels of cellubrevin (e.g. as on BFA-treated LBC) or with a compromised physical integrity of this SNARE, as seen on mycobacterial phagosomes. DISCUSSION The arrest of M. tuberculosis phagosome maturation has been associated with accumulation of Rab5 and exclusion of a number of regulatory factors: the Rab5 effector EEA1 (8) and the late endosomal GTPase Rab7 (19). These alterations are believed to lead to the paucity of the late endosomal lipid lysobisphosphatidic acid (8) and late endosomal/lysosomal enzymes, H ϩ ATPase (29), cathepsin D (30,31), and mannose 6-phosphate receptor (23). Here we have shown that mycobacteria affect the level and integrity of the v-SNARE Cellubrevin with repercussions for recycling of markers from the newly formed phagosomes and possibly of significance for interactions between the TGN and phagosomes. The generation of ⌬cellubrevin on MPC can be attributed to either production of a protease by the mycobacteria targeting this host cell SNARE or can be ascribed to activation in infected cells of host homeostatic mechanisms degrading SNARE molecules (56).
The changes in cellubrevin levels and integrity are not the FIG. 3. Cellubrevin levels on phagosomes with pharmacologically induced maturation arrest. Western blot analysis of purified LBC from macrophages treated with brefeldin A, bafilomycin A, or nocodozole as described under "Experimental Procedures." A, J774 cells phagocytosed latex beads for 30 min in a synchronized infection, after which LBC were isolated from each sample. Equivalent amounts (protein) of phagosomes were separated by SDS-PAGE, transferred to Immobilon-P, and probed with antibodies against cellubrevin and SNAP 23. B, quantitative analysis of cellubrevin and SNAP 23 by Western blots. Cellubrevin levels on LBC purified from macrophages treated with solvent, BFA, bafilomycin, or nocodozole as described under "Experimental Procedures" were quantified as described in the legend to Fig. 1 (n ϭ 3; *, p ϭ 0.0001; ANOVA).

FIG. 4.
Clearance of membrane-associated phagosomal markers during pharmacologically induced maturation block. J774 cells were treated with brefeldin A as described under "Experimental Procedures." Macrophages were allowed to phagocytose latex beads for 30 min in a synchronized infection after which LBC were isolated from each sample. Equivalent amounts (protein) of phagosomes were separated by SDS-PAGE, transferred to Immobilon-P, and probed with antibodies against transferrin receptor (TfR), adapter protein 3 (AP3, 3), syntaxin 13, and coronin (also known as TACO). A, Western blot analysis. B, quantitative analysis of TfR, AP3, coronin, and syntaxin 13 by Western blots. Protein levels on LBC purified from macrophages treated with solvent, or BFA as described under "Experimental Procedures," were quantified as described in the legend to Fig. 1 (n ϭ 3; *, p Ͻ 0.05; ANOVA). only alterations on mycobacterial phagosomes. Recently, we have reported the exclusion from MPC of the tethering molecule EEA1, a factor necessary for phagosome maturation (8). EEA1 is a major Rab5 effector implicated in vesicle tethering and fusion (57), including homotypic early endosomal fusion. The absence of EEA1 from MPC may appear counterintuitive, since it has been reported that MPC accumulate Rab5 (19), and Rab5 participates in the recruitment of EEA1 to endosomal membranes (58). However, EEA1 association with membranes depends on its FYVE domain binding to phosphatidylinositol 3-phosphate (59, 60) on the target organelles. The exclusion of EEA1 from MPC has been interpreted as interference with either generation or turnover of phosphatidylinositol 3-phosphate or blocking of EEA1 binding to the membrane of mycobacterial phagosomes (8,61).
The alterations of cellubrevin on MPC most likely affect trafficking processes determining the maturation status of these phagosomes. While a role in interactions between phagosomes and endosomes cannot be excluded at present, we believe that cellubrevin may function on phagosomes in ways different from trafficking between endosomal organelles and phagosomes, a concept reinforced by the dispensability of cellubrevin in the context of early endosomal fusion (62). Cellubrevin and syntaxin 6 are the proposed cognate SNAREs (45), potentially involved in TGN-to-phagosome transport. The accumulation of cellubrevin over time on phagosomes, extending to the 24-h time point, is in keeping with cellubrevin playing a role in TGN-to-phagosome trafficking. In this context, the interaction of EEA1 with the t-SNARE syntaxin 6 is likely to be important in TGN-to-endosome trafficking (63). We interpret the effects of mycobacteria on cellubrevin presented here, EEA1 shown elsewhere (8), and on syntaxin 6, 2 as a manifestation of convergent checkpoints involved in mycobacterial phagosome maturation arrest. Interference with the integrity or function of cellubrevin may thus represent a critical event in mycobacterial inhibition of phagosomal acquisition of lysosomal characteristics. In this model, the cleavage of cellubrevin, coupled with the exclusion of EEA1 from MPC, may inhibit interactions with syntaxin 6-containing TGN vesicles, thus deflecting the delivery of lysosomal enzymes by this route.
Cellubrevin may also play a role in the previously noted (24) partially defective recycling of plasma membrane markers associated with mycobacterial phagosomes. This phenomenon represents at present a less understood feature of mycobacterial phagosomes in the context of significance for intracellular survival. Previous studies have shown that mycobacterial phagosomes are accessible to transferrin, which can be chased out from these compartments (30,64), but only at lower rates relative to model phagosomes containing beads (30,64). This suggests that TfR recycling on MPC is affected, albeit not completely abrogated. The partial defect in TfR recycling corresponds with our observation that, although a substantial portion of cellubrevin is truncated on mycobacterial phagosomes, some amount of full-length protein remains on MPC at all times. In this context, it is worth noting that two toxininsensitive VAMPs, such as VAMP7 and endobrevin (VAMP8), overlapping with TfR-containing recycling vesicles, are also present on endosomes and phagosomes (65,66). Thus, cellubre- vin function in recycling may not be completely abrogated on MPC, but instead, its lower amounts, with or without participation of other VAMPs, may function at subopitmal levels, which in turn could be important in maintaining the specialized mycobacterial phagosome.
An alternative interpretation for the effect BFA on the phagosome remodeling could involve the delivery of cellubrevin to the plasma membrane. Cellubrevin has been shown to localize to nascent phagocytic cups (4) and has been suggested to facilitate the delivery of membrane at the site of phagocytosis. Hackam et al. (67) have reported that treating cells with BFA inhibits the fusion of cellubrevin containing endosomes with the plasma membrane. Thus, it is conceivable that the effects caused by BFA in our hands could reflect a reduced delivery of cellubrevin containing membranes to the site of phagosome formation. However, BFA concentrations used by Hackam et al. (67) were much higher (6-fold) than those applied in our work and exceed the BFA concentration required for TGN disruption (68). Thus, it is likely that the effects of BFA on cellubrevin levels on LBC seen in our work are linked to disruption of phagosomal interactions with other compartments, rather than effects on cellubrevin redistribution with indirect consequences for phagosomes.
We envision two different mechanisms to explain the appearance of ⌬cellubrevin on mycobacterial phagosomes: (i) proteolytic attack by a mycobacterial product or (ii) recruitment of host cytosolic enzymes degrading cellubrevin. Since M. tuberculosis produces at least two putative metalloproteases and contains other secreted hydrolases (69), it is possible that M. tuberculosis may directly target and proteolytically degrade Cellubrevin, similar to the action of clostridial metalloproteases, the tetanus and botulinum toxins (70). Alternatively, mycobacteria may recruit or activate host cell SNARE proteolytic machinery (56). Although ⌬cellubrevin on MPC most likely lacks the transmembrane domain, it is probably retained by association with other SNAREs in a binary complex or in the ternary cis-SNARE complex that can no longer be primed, as evidenced by accumulation of NSF on MPC. The role of NSF in phagosome maturation has been suggested by Funato et al. (71), and alterations in its amounts on phagosomes of other pathogens have been reported (72). Palade and colleagues (73) have recently described the presence of a large protein and polar lipid complex, which contains cellubrevin as well as Rab5, NSF, ␣-SNAP, syntaxin, dynamin, caveolin, and cholesterol. Mycobacterial phagosomes also contain caveolin and dynamin. 2 We predict the presence of a similar complex on phagosomes, which is further modified on MPC by cellubrevin proteolysis. The observed truncation of cellubrevin on MPC is most likely involved in the long term maintenance of the specialized mycobacterial phagosomal compartment in macrophages.