Screening Mycobacterium tuberculosis Secreted Proteins Identifies Mpt64 as a Eukaryotic Membrane-Binding Bacterial Effector

Advances have been made to identify secreted proteins of Mycobacterium tuberculosis during animal infections. These data, combined with transposon screens identifying genes important for M. tuberculosis virulence, have generated a vast resource of potential M. tuberculosis virulence proteins. However, the function of many of these proteins in M. tuberculosis pathogenesis remains elusive. We have integrated three cell biological screens to characterize nearly 200 M. tuberculosis secreted proteins for eukaryotic membrane binding, host subcellular localization, and interactions with host vesicular trafficking. In addition, we observed the localization of one secreted protein, Mpt64, to the endoplasmic reticulum (ER) during M. tuberculosis infection of macrophages. Interestingly, although Mpt64 is exported by the Sec pathway, its delivery into host cells was dependent upon the action of the type VII secretion system. Finally, we observed that Mpt64 impairs the ER-mediated unfolded protein response in macrophages.

Effectors are proteins that promote bacterial survival by manipulating vital cellular processes, including signal transduction and vesicular trafficking, and the cytoskeleton (49)(50)(51). Like the secretion systems themselves, the repertoire of effectors expressed by each pathogen can differ, adapted specifically for each unique life cycle. However, host membranes are a major common target for effectors. For example, SifA from Salmonella enterica serovar Typhimurium (S. Typhimurium) is prenylated inside the host cell and localizes to the Salmonella-containing vacuole (SCV) (52). SifA recruits lysosomes to maintain SCV membrane integrity, and its membrane interaction is vital to Salmonella pathogenicity (52,53). The L. pneumophila type IV secreted effector SidM is anchored to the Legionella-containing vacuole and disrupts host vesicle trafficking by sequestering and modifying Rab1 (54). Some effectors can also function by directly modifying membranes such as IpgD from Shigella flexneri that hydrolyzes phosphatidylinositol 4,5-bisphosphate [PI(4,5)P 2 ] to phosphatidylinositol 5-monophosphate (PI5P), leading to membrane blebbing and bacterial uptake into host cells (55). Thus, both host membranes themselves and membrane-dependent processes represent valuable targets for bacterial effectors (49,51,56) as we recently showed for a variety of bacterial pathogens (57). Because membrane processes are high-value targets of many bacterial effectors (57,58) and Mtb has a large repertoire of secreted proteins of unknown function (59)(60)(61)(62)(63), we hypothesized that some of the Mtb secreted proteins are membrane-binding effectors with virulence activities.
To test our hypothesis, we generated a library of 200 secreted proteins from Mtb, tested whether they individually bound yeast membranes in a life-or-death assay, and characterized their ability to alter host protein secretion in an inducible secretion assay. We also determined the subcellular localization of membrane-binding proteins using fluorescence microscopy. By combining data from the cell biological screens, we identified five Mtb secreted proteins that localized to eukaryotic membranes and disrupted the host secretory pathway in a model system. One protein, Mpt64 (Rv1980c), localized to the endoplasmic reticulum (ER) during both heterologous expression in HeLa cells and Mtb infection of macrophages. Though Mpt64 is a Sec substrate, its access to the macrophage cytoplasm was dependent on the ESX-1 secretion system. Finally, Mpt64 alone was sufficient to partially inhibit the unfolded protein response (UPR), suggesting a possible role for Mpt64 in regulating the macrophage response to infection.

RESULTS
Categorization of putative effector-like proteins from Mtb. Through the analysis of published data sets, we identified Mtb proteins that may function as secreted effectors (see Table S1 in the supplemental material). For simplicity, we define these putative effectors as mycobacterial secreted proteins (MSPs), as this encompasses proteins that may be secreted to the mycobacterial surface, into the exoproteome (i.e., the extracellular milieu), or delivered into the host cell (64)(65)(66)(67). We used the following criteria to assemble a library of MSPs: (i) Mtb proteins identified via unbiased proteomic approaches that are either in the cell wall or the exoproteome (60-63, 68, 69), (ii) Mtb proteins known to be involved in manipulation of host vesicular trafficking pathways, such as ones that induce mammalian cell entry (MCE) (70)(71)(72) or phagosome maturation arrest (13,14,73), (iii) a subset of PE/PPE proteins and proteins related to those encoded by type VII (ESX-1) loci (42,48,74), and (iv) proteins involved in virulence, ranging from defined to unknown functions (19,(75)(76)(77)(78). We then used Gateway recombination cloning to subclone MSPs from the freely available Mtb ORFome Gateway compatible library (BEI) into destination vectors for a variety of subsequent assays. The comprehensive list of MSPs is shown in Table S1.
Mtb genes encode secreted proteins that interact with eukaryotic membranes. To identify membrane-binding Mtb proteins, we used a system that leverages the signal transduction of the essential Saccharomyces cerevisiae GTPase Ras to promote growth and division (79). Ras is lipidated at its CaaX box sequence that promotes its localization to the plasma membrane, where it can be activated by Cdc25, a guanine nucleotide exchange factor (79,80). In a yeast strain with a temperaturesensitive CDC25 allele, yeast can grow only at the permissive temperature (25°C), not the restrictive temperature (37°C), because Ras activation requires interaction with Cdc25. Heterologous expression of a nonlipidated, constitutively active Ras whose activity is independent of Cdc25 (mutant Ras [Ras mut ]) can rescue yeast growth at the restrictive temperature when Ras is recruited to intracellular membranes by fusion to a membrane-binding protein (Fig. 1A). This system has been used to successfully identify membrane-binding effectors from Gram-negative pathogens (57). To identify membrane-localizing proteins from Mtb, we subcloned MSPs into a destination vector for yeast expression that generates an in-frame fusion of the MSP to Ras mut . We transformed S. cerevisiae cdc25 ts (57,79) individually with each of the 200 MSPs fused to Ras mut and incubated them at both permissive and restrictive temperatures (Fig. 1B). Shown are 20 examples of yeast growth at 25°C and 37°C when expressing individual Ras mut -MSP fusion proteins. All strains grew when incubated at 25°C. However, at the restrictive temperature, only some yeast strains, indicated in red, were rescued. For example, we observed yeast growth rescue in yeast expressing a Ras mut fusion to SapM (Rv3310), as expected due to its PI3P phosphatase activity (16,18), but not for yeast expressing Ras mut to the protein tyrosine phosphatase PtpA (Rv2234) (81,82), which did not grow at 37°C (Fig. 1B). We identified 52 Mtb proteins that rescued S. cerevisiae cdc25 ts growth at the restrictive temperature ( Fig. 1B and C).
We confirmed expression of the Ras mut -MSP fusion proteins by Western blotting (Fig. 1D). In addition, we determined the membrane localization of each MSP by fluorescence microscopy of GFP-MSP fusion proteins in yeast (Fig. 1E). It has been established that Ras can function from membranes other than the plasma membrane (83,84), and Ras mut maintains this function (57). Thus, using fluorescence microscopy, we observed green fluorescent protein (GFP)-MSP fusion proteins localizing to distinct subcellular compartments, including vacuoles, ER, and plasma membrane (Fig. 1E). Together, these results show that 25% of the MSPs tested could associate with the membranes of a variety of organelles in S. cerevisiae. Subcellular localization of membrane-localizing MSPs. While many cellular processes are conserved in eukaryotes, humans represent the primary natural host for Mtb. Therefore, to confirm that MSPs that rescued S. cerevisiae cdc25 ts growth at 37°C also bound mammalian membranes and to determine their subcellular localization in human cells, we transiently transfected HeLa cells with vectors for constitutive expression of GFP-MSP fusion proteins and then used fluorescence microscopy with colocalization markers to identify the specific membrane to which each MSP localized ( Fig. 2A). We identified GFP-MSP fusion proteins that localized to a variety of subcellular compartments, including the ER, Golgi apparatus, mitochondria, and peroxisomes ( Fig. 2A and B and Table S1). The largest proportion of the GFP-MSP fusion proteins expressed in human cells colocalized with the ER marker calreticulin (Fig. 2B). We observed in yeast a similar number of GFP-Mtb fusion proteins that localized to compact, punctate structures that could not be definitively localized. Although there was only moderate overlap in the subcellular localization identified between yeast and HeLa cells (Fig. 2B), we were able to verify that the proteins identified by the Ras rescue assay are localized to membranous organelles in human cells.
A subset of mycobacterial secreted proteins alter eukaryotic vesicular transport. Membrane-bound cargo is transported within the host cell and to the extracellular space in dynamic vesicular trafficking pathways. The membrane-bound and soluble proteins important for these processes are frequent targets of bacterial effectors (58,85). Indeed, Mtb is known to target and manipulate trafficking pathways through incompletely understood mechanisms (82,86,87). To determine whether Mtb proteins can broadly affect the host vesicular trafficking pathways as an indicator of interaction with membranes, we took advantage of a reverse dimerization system (88). In this system, a protein detectable by ELISA is sequestered in the ER by fusion to a conditional aggregation domain (CAD). Addition of a solubilization molecule that disrupts the CAD then frees the fusion protein for trafficking and release into the extracellular space. We used a fusion of human growth hormone (hGH) to the CAD domain of the ligand-reversible cross-linking protein, FKBP F36M. Thus, hGH can be quantified in cell supernatants by ELISA after the addition of the small molecule D/D Solubilizer (Fig. 3A) (88,89). Another advantage of this system is that it permits determination of the impact of expressed proteins on vesicular trafficking events independent of the bacterial effect on host innate immune responses. We transfected HeLa cells expressing hGH-CAD with each MSP individually, a negative-control protein (GFP), or an enterohemorrhagic Escherichia coli effector (EspG) that inhibits vesicular trafficking by promoting the tethering of vesicles to the Golgi apparatus (89,90). When we treated transfected cells with D/D Solubilizer, we observed increased, decreased, and normal hGH release ( Fig. 3B and Table S1). Using a cutoff of normalized hGH release below 0.25 or above 1.75, we identified 18 proteins that decreased hGH release and 11 proteins that increased hGH release compared to the GFP control (Fig. 3C). We next compared the MSPs that altered host vesicular trafficking to those that bound eukaryotic membranes and identified five proteins with overlapping activities: Rv0594, Rv1646, Rv1810, Rv1980c, and Rv2295 (Fig. 3D). During expression in HeLa cells, all but one protein localized to the ER, and all five proteins reduced hGH release (Fig. 3E).
Mpt64 N terminus is important for ER localization and inhibition of vesicular trafficking. We focused on the protein Rv1980c, also known as Mpt64, as it is a secreted protein that is highly antigenic during human tuberculosis infection (91,92). Furthermore, mpt64 is a component of the region of difference 2 (RD2) locus, one of the genomic regions deleted during attenuation of the M. bovis BCG vaccine strain (93). Loss of RD2 from Mtb attenuates its virulence, and complementation with a three-gene cluster that includes mpt64 can partially restore virulence (94).
Mpt64 is a 25-kDa protein with a predicted signal peptidase 1 cleavage site between amino acids 23 and 24, such that the mature, secreted form of the protein starts at amino acid 24 (61,62,95). While the solution structure of Mpt64 was previously solved (96), the structure does not align to a known catalytic domain but does contain a domain of unknown function (DUF3298). This domain is also present in the lysozymebinding anti-sigma factor RsiV (97). Despite structural homology between Mpt64 and RsiV (see Fig. S1 in the supplemental material) (98), there is little primary sequence homology. To determine whether Mpt64 binds lysoszyme, we purified recombinant Mpt64 from E. coli and tested binding to human or hen egg white lysozyme in an in vitro pulldown assay (97). Using this assay, we were unable to demonstrate lysozyme binding by Mpt64 (Fig. S1). We next used the solution structure to guide truncation analysis of Mpt64 in order to identify the membrane-binding sequences of Mpt64 ( Fig. 4A and B). S. cerevisiae cdc25 ts expressing a fusion of Ras mut with either full-length Mpt64, mature Mpt64 lacking its predicted signal peptide, or the N-terminal half of Mpt64 also lacking the signal peptide (Mpt64_24-143) were able to grow at 37°C, whereas S. cerevisiae cdc25 ts expressing Ras mut fused to the C-terminal half of Mpt64 (Mpt64_144-228) could not (Fig. 4C). We detected expression of Ras mut fusions of full-length Mpt64 and mature Mpt64 by Western blotting. In contrast, we could not detect Ras mut fusions of Mpt64_24-143 or Mpt64_144-228 despite the fact that the Mpt64_24-143 fusion rescued yeast growth, suggesting that expression of Mpt64_24-143 below the limit of detection by Western blotting was still sufficient to rescue yeast growth (Fig. 4D). However, we could not determine whether the C-terminal domain plays a role in membrane binding because we were unable to demonstrate stable fusion protein expression in yeast. Finally, we sought to determine whether the N-terminal portion of Mpt64 was also sufficient to inhibit hGH release using the hGH-CAD assay. We cotransfected HeLa cells with plasmids for expression of hGH-CAD and Mpt64 truncation alleles and determined their ability to inhibit hGH release in the presence of drug. Similar to the Ras rescue assay, full-length Mpt64, mature Mpt64, and Mpt64_24-143 inhibited hGH release compared to the GFP control. In contrast, cotransfection of Mpt64_144-228 with hGH-CAD had no effect on its release (Fig. 4E). These data suggest that the ability of Mpt64 to bind membranes and to inhibit host vesicular trafficking in vitro is dependent on the N terminus of the protein.
Mpt64 ER localization depends on its N terminus. As full-length Mpt64 localized to the ER in yeast and HeLa cells ( Fig. 1 and 3), we next tested the impact of Mpt64 truncations on ER localization. We first determined the phenotypic localization of Mpt64 truncations expressed as GFP fusions in yeast using fluorescence microscopy. Mpt64_1-228 and Mpt64_24-228 localized in a ring indicative of the ER (99, 100) (Fig. 5A). In contrast, Mpt64_144-228, which did not rescue yeast growth in the Ras rescue assay (Fig. 4C), was diffuse throughout the yeast cell (Fig. 5A). Interestingly, Mpt64_24-143 localized to bright puncta within the cells (Fig. 5A). To confirm the N-terminal dependence of Mpt64 localization, we transfected HeLa cells with GFP fusions to each Mpt64 truncation or GFP alone and assayed for colocalization with calreticulin through immunofluorescence microscopy. While full-length Mpt64, mature Mp64, and Mpt64_24-143 colocalized with calreticulin, GFP did not (Fig. 5B). Mpt64_144-228 localized to a bright aggregate that did not colocalize with calreticulin, suggesting that the C terminus of Mpt64 is misfolded and/or unstable when expressed on its own. These results demonstrate that Mpt64 localizes to the ER during exogenous expression in both yeast and mammalian cells and that the N-terminal 143 amino acids are sufficient to mediate subcellular localization of Mpt64 to the ER.
Mpt64 ER localization in yeast is dependent on PI3P and PI(3,5)P 2 . On the basis of the results from the PIP strips ( Fig. 6A) and to further characterize the lipid binding of Mpt64 in vivo, we took advantage of yeast strains mutated in phosphatidylinositide (PI) kinases that either lack or have reduced levels of specific phosphatidylinositol phosphates (PIPs) (Fig. 6B). Because PIPs are geographically restricted within cells and their position-specific phosphorylation patterns can function as organelle-specific markers to recruit proteins to areas with unique membrane constituents (57,101,102), inactivation of yeast PI kinases causes mislocalization of bacterial effectors that need such PIP interactions for appropriate membrane targeting (57) (Fig. 6C). We inhibited expression of each yeast PI kinase gene either by isogenic knockout of the nonessential PI kinase gene (VPS34, FAB1, and LSB6) or by doxycycline (Dox)-mediated repression of TetO 7 -promoter alleles of essential PI kinase genes (PIK1, STT4, and MSS4). We optimized repression conditions for the TetO 7 -promoter alleles by monitoring the distribution of the PI4P-specific binding protein Osh2, which shuttles between the plasma membrane and Golgi apparatus in a PI4P-dependent manner (Fig. 6D) (57, 103, 104). Thus, Dox-mediated loss of PIK1 and STT4 expression led to Osh2 redistribution to the plasma membrane (PM) or Golgi apparatus, respectively, as previously reported (57). For a further control for the isogenic deletion strains, we observed that deletion of LSB6 and VPS34 along with Dox-mediated loss of MSS4 and PIK1 caused relocalization of the S. Typhimurium effector SopA from the PM to internal puncta, consistent with its affinity for several PIP isoforms (57). Using this assay, Mpt64 relocalized from the ER to the PM and internal puncta in the absence of VPS34 and FAB1 (Fig. 6D). As these strains lack PI3P and PI(3,5)P 2 and because the VPS34 strain lacks both PIPs while the FAB1 strain lacks only PI(3,5)P 2 , we conclude that recruitment of Mpt64 to the ER in yeast is likely dependent on PI(3,5)P 2 . Though the PIP strips ( Fig. 6A) showed strong binding to PI4P, we propose that the failure to relocalize in yeast lacking various PI kinases that produce PI4P (i.e., LSB6, PIK1, and STT4) is due to their redundancy. Taken together, the  data indicate that Mpt64 interacts with PIPs in vitro and in vivo to facilitate its localization to the ER.
Secreted Mpt64 localizes to the ER during infection. Although we observed Mpt64 localization to the ER in yeast ( Fig. 1E and Fig. 5A) and HeLa cells (Fig. 5B), we wanted to determine whether endogenous, untagged Mpt64 localizes to the ER during an Mtb infection of macrophages. To that end, we infected mouse RAW267.4 macrophages with mCherry-labeled Mtb at a multiplicity of infection (MOI) of 20:1 and fixed cells at various time points after infection. We then used a rabbit polyclonal antibody developed against recombinant, mature Mpt64 protein to track Mpt64 secretion from Mtb into macrophages using immunofluorescence microscopy. Of note, this antibody was generated without complete Freund's adjuvant in order to avoid any crossreactivity against Mtb antigens generated by the use of this adjuvant. As little as 4 h after infection, endogenous Mpt64 was detected in the cytoplasm host cells in a lacy pattern indicative of ER localization (Fig. 7A, top panels). When we infected macrophages with MtbΔeccD1, a strain that is deficient in ESX-1 secretion (105, 106) and does not result in communication between the phagosome and cytoplasm (27,45,46,(107)(108)(109), Mpt64 appeared to be secreted but trapped adjacent to the bacteria (Fig. 7A, bottom panels), suggesting that it could not escape the phagosome. Importantly, Mpt64 was detected in the culture filtrate prepared from MtbΔeccD1 (Fig. S3). Thus, although Mpt64 is likely secreted from Mtb by the canonical Sec-dependent pathway, its access to the macrophage cytoplasm and other targets in the cell was dependent on the type VII secretion system.
In order to better understand the role of Mpt64 in Mtb virulence, we used mycobacteriophage (110)(111)(112) to introduce the hygromycin resistance cassette into the mpt64 gene to create an in-frame deletion (Fig. 7B). We confirmed disruption of mpt64 by PCR (Fig. S2) and loss of Mpt64 by the absence of protein on Western blots (Fig. 7C). We then complemented MtbΔmpt64 with either full-length mpt64 (MtbΔmpt64::mpt64) or Mpt64 lacking its signal peptide (MtbΔmpt64::mpt64-NS) under the control of the constitutive mycobacterial strong promoter (113). Both complemented strains expressed Mpt64, but only full-length Mpt64 (MtbΔmpt64::mpt64) could be detected in the supernatant of cultures, confirming that deletion of the signal peptide inhibits Mpt64 secretion from Mtb (Fig. 7C). Furthermore, the MtbΔmpt64::mpt64 strain had modestly higher expression of Mpt64 compared to wild-type Mtb by Western blotting, consistent with our use of a strong constitutive promoter for complementation. All four strains grew equally under axenic growth conditions (Fig. S3), and we confirmed that both Mtb and MtbΔmpt64 produced phthiocerol dimycocerosate (PDIM) by mass spectrometry (Fig. S3).
To test whether secreted Mpt64 localizes to the ER during infection, we assessed its colocalization with calreticulin in RAW267.4 cells using confocal immunofluorescence microscopy. When we infected RAW267.4 macrophages, the Mpt64 signal in Mtbinfected macrophages colocalized with calreticulin, confirming the subcellular localization of Mpt64 secreted during infection ( Fig. 7D and E and Fig. S3). However, this colocalization was lost in cells infected with MtbΔmpt64::mpt64-NS bacteria ( Fig. 7D and F and Fig. S4). For a control for antibody specificity, no Mpt64 was detected in macrophages infected with MtbΔmpt64 mutant bacteria ( Fig. 7D and Fig. S4). From these data, we can confirm that the signal peptide of Mpt64 is sufficient for the protein's secretion in vivo and is required (with concerted action of the type VII secretion system) for Mpt64 to localize to the ER during infection.
Mpt64 inhibits the unfolded protein response. ER stress and the UPR have recently been associated with Mtb infection and pathogenesis (114)(115)(116). Because we found that Mpt64 localizes to the ER during infection, we hypothesized that it might regulate the UPR. To test whether Mpt64 was sufficient on its own to impact the UPR, we stably transduced murine RAW267.4 cells with empty lentivirus or a lentivirus with Mpt64 under the control of a cytomegalovirus (CMV) promoter and induced the UPR by treating with thapsigargin, a known UPR activator (117). In the presence of thapsi-gargin, we detected robust accumulation of the UPR-activated transcription factor CCAAT enhancer-binding protein homologous protein (CHOP), a protein whose expression is low under nonstressed conditions but high in the setting of ER stress (117) (Fig. 8A). Expression of Mpt64 resulted in a 75% reduction in CHOP compared to control cells, indicating that Mpt64 alone could inhibit the UPR (Fig. 8B).
Mpt64 contributes to early Mtb growth after aerosol infection of mice. Because mpt64 is part of the Mtb RD2 locus that partially accounts for the attenuation of Mtb (94) and our data indicating that Mpt64 may function as a secreted effector that modulates the UPR, we investigated the role of Mpt64 in Mtb virulence in a murine model of infection. We infected BALB/c mice via aerosol with a low bacterial inoculum (ϳ50 to 100 CFU Mtb) and collected lungs at various time points to determine CFU and histopathology. We compared the infections of four strains: wild-type Mtb, MtbΔmpt64, MtbΔmpt64::mpt64, and MtbΔmpt64::mpt64-NS. While all mice received equal numbers of bacteria of the four strains at day 0, there were one-third fewer Mtb isolated from the lungs of mice infected with the MtbΔmpt64 mutant compared to mice infected with the wild type at 21 days (mean CFU of 2.7 ϫ 10 6 for wild-type Mtb versus 1.7 ϫ 10 6 for MtbΔmpt64; P ϭ 0.07) and 42 days (mean CFU of 5.0 ϫ 10 5 for wild-type Mtb versus 3.4 ϫ 10 5 for MtbΔmpt64) postinfection. Though these effect sizes were modest and consistent for both time points, they did not meet statistical significance at an alpha of P Ͻ 0.05. By 42 days postinfection, we observed statistically significant decreases in the CFU isolated from lungs of mice infected with the MtbΔmpt64::mpt64-NS strain (mean CFU of 5.0 ϫ 10 5 for wild-type Mtb versus 1.8 ϫ 10 5 for MtbΔmpt64::mpt64-NS; P ϭ 0.001) (Fig. 9A). At these time points, we also observed a reduction in the area of inflammation in hematoxylin-and-eosin (H&E)-stained lungs of mice infected with MtbΔmpt64::mpt64-NS compared to mice infected with the wild type ( Fig. 9B and D). Despite modest reductions in CFU in the lungs of mice infected with mutant bacteria, there was no impact on mouse survival (Fig. 9C).
Mpt64 localized to the ER in primary human macrophages but is dispensable for Mtb survival. Next, we assessed whether the localization of Mpt64 in human cells is similar to that in murine macrophages. To that end, we infected primary human monocyte-derived macrophages with mCherry-expressing wild-type (WT) Mtb or MtbΔeccD1 and stained for Mpt64. Consistent with our data in RAW267.4 cells (Fig. 7A), Mpt64 localization to extraphagosomal sites in primary human macrophages was dependent on the type VII secretion system (Fig. 10A). We then infected primary human macrophages with WT Mtb, MtbΔmpt64, MtbΔmpt64::mpt64, or MtbΔmpt64::mpt64-NS and determined the colocalization of Mpt64 with calreticulin by fluorescence microscopy ( Fig. 10B and Fig. S5). At 4 h postinfection (hpi), we detected colocalization of Mpt64 with calreticulin in cells infected with WT Mtb and MtbΔmpt64::mpt64 but not in cells infected with MtbΔmpt64 or MtbΔmpt64::mpt64-NS (Fig. 10B).
To better understand the contribution of Mpt64 in the context of human Mtb infection, we determined the growth of wild-type Mtb, MtbΔmpt64, MtbΔmpt64::mpt64, MtbΔmpt64::mpt64-NS, and MtbΔeccD1 as a control for attenuation in primary monocyte-derived human macrophages during acute infection. We recovered CFU from cells directly after infection (day 0) and 1 and 3 days postinfection. We observed significant donor-to-donor variability both in the ability to restrict intracellular Mtb replication (compare the growth of WT Mtb between representative donor 1 and donor 2) and the relative growth of MtbΔmpt64, MtbΔmpt64::mpt64, and MtbΔmpt64:: mpt64-NS in various donors. Thus, while in some donors, the CFU at day 3 postinfection of strains lacking mpt64 was modestly but not statistically significantly lower compared to WT Mtb (i.e., donor 1), in other donors, there was no impact on the presence or absence of mpt64 (i.e., donor 2). Thus, in this acute primary human macrophage infection model, the presence of Mpt64 appeared to be dispensable for Mtb survival.

DISCUSSION
Numerous efforts have been undertaken to identify Mtb secreted proteins, from lipoproteins that are incorporated into the cell wall to virulence factors that reach the extracellular environment such as ESAT-6 (59-61). However, little is known about the function of this "exoproteome" as a whole. Here we took a systematic approach toward characterizing host-dependent interactions of a collated list of putative secreted Mtb proteins. We created a library of 200 putative secreted proteins and then through a series of cell biological screens characterized these MSPs for their ability to bind eukaryotic membranes, their subcellular localization, and their ability to modulate release of a model substrate. In addition, we demonstrate that one secreted protein, Mpt64, localized to the ER during infection of mouse and human macrophages and inhibited the UPR. The cohort of 200 MSPs we generated was large but not necessarily exhaustive (see Table S1 in the supplemental material). For example, the 76 PE/PPE genes we included represent less than half (45%) of the total number of PE/PPE genes in the Mtb genome (118). In addition, a recent technology called exported in vivo technology (EXIT) identified 593 Mtb proteins secreted during intravenous infection of mice, including 38 proteins that are significantly enriched only during in vivo infection compared to growth on Middlebrook 7H10 agar, suggesting a virulence function for these proteins (67). Of the 200 MSPs we characterized, 51 overlap with those identified by EXIT, and of the 51 overlapping proteins, 25 are membrane associated in our study. This emphasizes that host membranes can be targets of Mtb secreted proteins. We found 52 Mtb proteins that associated with eukaryotic membranes, representing nearly 25% of the total screened. When the membrane association of type III and type IV effectors from several Gram-negative pathogens was explored, about 30% of effectors screened also associated with eukaryotic membranes (57). While our data are in agreement with this value, pathogens that replicate intracellularly in vacuoles had even higher numbers of membrane-associated effectors (57). This suggests that there may be additional secreted virulence proteins from Mtb that associate with the host membranes than our screen identified. Indeed, while we corroborated previously known membrane-interacting proteins such as the SecA2 secreted PI3P phosphatase SapM (17,18), the Rac1-binding protein Ndk (119), and the cholesterol-binding Mce4A (120), we failed to identify others such as LipY which hydrolyzes extracellular lipids (121) and the ESX1 substrate ESAT-6, whose ability to directly interact with the phagosomal membrane (122,123) has recently been questioned (65,124).
The vast majority of MSPs localized to the ER when expressed in HeLa cells. Whether this localization reflects the importance of modulating ER function for Mtb survival cannot be determined directly, as we did not raise protein-specific antibodies against the proteins like we did for Mpt64 in order to determine the localization of untagged, endogenous, secreted protein during infection. It is possible that some of the observed localizations represent false-positive results, as overexpression of Mtb proteins with signal sequences results in aberrant ER localization in 293T cells (125). In addition, we found some MSPs that rescued yeast growth in the Ras rescue assay but localized to the cytoplasm when expressed as GFP fusions in HeLa cells. These proteins represent either false-positive results from the Ras rescue assay or proteins that do associate with lipids whose abundance is sufficient in yeast cells but not in HeLa cells.
GFP-Mpt64 localized to the ER in both yeast cells and mammalian cells. Additionally, endogenous Mpt64 localized to the ER during Mtb infection of macrophages, suggesting that the observed localization of Mpt64 is not an artifact of heterologous overexpression. Mpt64 did not colocalize with the ER after infection with a Mtb type VII secretion system mutant, underscoring the importance of the phagosome-disrupting properties of the ESX-1 system in establishing communication with the host cell (65,126). This ESX-1-dependent mechanism of cytoplasmic access is similar to the route taken by the autotransporter-like protein tuberculosis necrotizing toxin (TNT) (127,128). Thus, our data strengthen the argument that the type VII secretion system facilitates access of non-ESX-1 substrates beyond the phagosome and into the host cell.
Mpt64 bound several PIPs in vitro with the most prominent interactions being with the monophosphatidylinositols PI3P, PI4P, and PI5P. Furthermore, the ER localization of Mpt64 in yeast changed when the PIP kinases generating PI3P and PI(3,5)P were deleted, suggesting that its function in vivo is connected to its ability to interact with membrane PIPs. While PI3P is mainly detected in endosomes and autophagosomes (129), PI3P has also been identified at specialized ER sites called omegasomes where a dynamic exchange of PI3P-positive vesicles and ER occurs, allowing for assembly of autophagy proteins and expansion of autophagosome membranes, leading to initiation of autophagosome formation (130). PI4P is thought to be enriched in the Golgi apparatus (131), but it also has an established role in mediating protein trafficking from ER exit sites (132,133). Less is known about PI5P, as its basal level is only about 1% of the basal PI4P level (134). However, PI5P is increased during bacterial infection and other stresses and can be found throughout the cell, including the ER (134) and on lipid droplets that arise from the ER (135). Similarly, PI(3,5)P 2 is present at low abundance, but its levels are elevated by stress such as hyperosmotic shock in yeast (136). Conserved functions of PI(3,5)P 2 from yeast to mammals include regulation of autophagy and retrograde trafficking, activation of some ion channels, and cargo sorting into multivesicular bodies (136). Thus, the subcellular localization of Mpt64 at the ER may result from its interaction with the monophosphatidylinositols PI3P, PI4P, and PI5P in addition to PI(3,5)P 2 , thus allowing Mpt64 to interfere with ER to Golgi trafficking and prevent release of the model substrate hGH and also inhibit the UPR. However, the relative contribution of binding to individual PIPs to the activity of Mpt64 remains unknown.
Although an enzymatic activity for Mpt64 could not be deduced from its structure, we were able to demonstrate that the N terminus of Mpt64 was sufficient to mediate membrane binding, interaction with PIPs, and inhibition of vesicular trafficking. Mature Mpt64 was also able to inhibit the UPR in the setting of thapsigargin-induced ER stress in macrophages. Since ER stress has been observed in vivo during Mtb infection and ER stress is known to activate autophagy and apoptosis (137), the ability of Mpt64 to downregulate the UPR may allow Mtb to fine-tune the host response in order to provide a long-term replicative niche. Indeed, since both autophagy (138,139) and apoptosis (140) have critical roles in the outcome of Mtb infection, engagement by Mtb of high-value targets such as the ER, Golgi apparatus, and mitochondria-organelles vital to cellular regulation of autophagy and apoptosis-with its entire secreted effector armament such as the membrane-binding MSPs identified here likely allows Mtb to tightly control the host response to facilitate successful infection.
Disruption of the RD2 locus in Mtb H37Rv leads to decreased bacterial burdens in the lungs and spleen of aerosol-infected mice at 3 weeks after infection (94). As mpt64 is within the RD2 locus, we hypothesized that the single, in-frame deletion of mpt64 might explain the attenuation phenotype of the RD2 mutant. Though we did not perform a head-to-head comparison of an MtbΔRD2 strain versus our MtbΔmpt64 strain on the Erdman genetic background, we did observe modestly decreased bacterial burdens of MtbΔmpt64 compared to WT Mtb in the lungs of mice at 3 weeks postinfection. Other genes located in the RD2 locus that were not complemented in the RD2 survival study (94) such as pe_pgrs35 (Rv1983) and cfp21 (Rv1984) may also contribute to the virulence defect observed in RD2 deletions. Furthermore, it is possible that one or more of the other Mtb secreted proteins we identified, including the 27 proteins that also localized to the ER, are able to perform a redundant function to that of Mpt64 during an animal infection. In a similar vein, whereas L. pneumophila encodes more than 300 effectors, individual L. pneumophila effector deletion mutants are not defective for growth in cells or mice (141,142). Indeed, a L. pneumophila strain in which 11 effectors that function to inhibit protein translation are deleted is still able to inhibit host protein translation, though a mutant in the type IV secretion machine itself cannot (143), indicating remarkable redundancy in effector activity. Thus, future work disrupting multiple Mtb MSPs simultaneously will help address this issue.
When we infected mice with MtbΔmpt64::mpt64-NS, a strain of Mtb that still expresses Mpt64 but cannot secrete it into the host cell, we recovered fewer CFU compared to WT from the lungs of mice infected with MtbΔmpt64::mpt64-NS. We hypothesize that this strain suffers from two detrimental consequences. First, blocking Mpt64 secretion prevents it from exerting its function in the host. Second, nonsecreted Mpt64 can still be cross-presented to the adaptive immune system (144), thus leading to a cell-mediated immune response against Mpt64. This observation is consistent with data that both human patients with active tuberculosis and their PPD-positive contacts have T-cell responses to Mpt64 (145) and T-cell-reactive Mpt64 epitopes have been mapped (146). Furthermore, Mpt64 staining is observed in granulomas of infected individuals (147,148). Thus, Mpt64 is highly immunogenic during human infection with Mtb and suggests an evolutionary trade-off between the effector function of Mpt64 and its antigenicity. When we explored the importance of Mpt64 in human disease, we observed that Mpt64 secreted from wild-type bacteria localized to the ER of infected human monocyte-derived macrophages, though a mpt64 mutant did not have a consistent defect in survival within macrophages. Future work on other mycobacterial secreted, ER-binding proteins may ultimately reveal functional redundancy with Mpt64 important for the virulence of Mtb.
Yeast strains and assays. The Saccharomyces cerevisiae strain INVSc1 (Invitrogen) was grown at 30°C in histidine dropout medium (synthetic defined [SD] base dropping out histidine [SD/ϪHIS]) or agar plates (Clontech). The construction of the cdc25 ts strain was previously described (80). The cdc25 ts strain was grown at 25°C in leucine dropout medium or agar plates (SD/ϪLEU) (Clontech). Yeast knockout strains ΔFAB1, ΔLSB6, and ΔVPS34 are on the S. cerevisiae BY4741 background and were grown in YPD (Difco) prior to transformation. The tetracycline off strains (MSS4, PIK1, and STT4; Thermo Fisher Scientific) are on the genetic background of R1158 and were grown in YPD supplemented with 300 g/ml G418 (catalog no. 345810; Millipore) prior to transformation.
Yeast strains were transformed using a lithium acetate (LiAc) protocol. The yeast strains were grown to high density overnight at the appropriate temperature with shaking. The cultures were diluted to an optical density at 600 nm (OD 600 ) of 0.2 in 50 ml YPD and allowed to reach mid-log phase. Cells were washed, resuspended in 0.1 M LiAc, and incubated 10 min at room temperature. The sample DNA was mixed with an equal volume of preboiled Yeastmaker Carrier DNA (Clontech). To the DNA was added 100 l yeast and 500 l of a solution of LiAc plus PEG (40% [wt/vol] PEG, 0.1 M LiAc). This solution was incubated 30 min at 25°C (cdc25 ts ) or 30°C with agitation every 10 min. DMSO was added, and the cells were heat shocked at 42°C for 15 min. The cells were pelleted, washed in TE (10 mM Tris [pH 7.4], 1 mM EDTA) and resuspended in 500 l TE. The transformed cells were plated on selective agar plates and incubated at the appropriate temperature for 2 to 4 days.
To perform the Ras rescue assay, three or four fresh colonies were combined in 30 l SD/ϪLEU, and 3 l was spotted onto duplicate plates that were subsequently incubated at either 25°C or at 37°C for 2 days.
INVSc1, deletion mutants and tetracycline off yeast were transformed with a galactose-inducible vector (p413GALGFP) containing GFP-Mtb fusion proteins and selected on SD/ϪHIS. To induce GFP fusion protein expression in wild-type (WT) and deletion mutant strains, yeast cells were inoculated in 5 ml galactose/raffinose (Gal/Raf) base lacking histidine (Clontech) and allowed to grow for 16 to 20 h at 30°C with shaking. Cultures were pelleted, resuspended in 3 ml Gal/Raf/ϪHIS and incubated for another 4 h at 30°C with shaking. Yeast cells were pelleted and resuspended in 30 to 50 l PBS and immobilized on an agar pad prior to visualization. For tetracycline off strains, yeast cells were inoculated into 5 ml SD/ϪHIS supplemented with 300 g/ml G418 and 50 g/ml (PIK1 and STT4) or 300 g/ml (MSS4) doxycycline (catalog no. 324385; Millipore) and incubated overnight at 30°C with shaking. Yeast cells were washed in TE buffer and resuspended in Gal/Raf/ϪHIS (to induce GFP fusion protein expression) with 300 g/ml G418 and 50 g/ml (PIK1 and STT4) or 300 g/ml (MSS4) doxycycline. After overnight incubation, yeast cells were immobilized on agar pads as described above.
Yeast lysis and Western blotting. Yeast cells (cdc25 ts ) were inoculated into 5 ml SD/ϪLeu and incubated overnight at room temperature with shaking (250 rpm). To lyse, 1.5 ml of each culture was centrifuged at 14,000 rpm for 1 min. Each pellet was resuspended in 100 l of 2.0 M LiAc and incubated on ice for 5 min. Samples were centrifuged at 14,000 rpm for 1 min to pellet, resuspended in 100 l of 0.4 M NaOH, and incubated on ice for 5 min. Samples were pelleted as before, resuspended in 75 l of 1ϫ SDS Laemmli sample buffer, and boiled at 100°C for 5 min. Lysates were centrifuged at 14,000 rpm for 1 min to remove debris, separated by SDS-polyacrylamide gel electrophoresis, and transferred to polyvinylidene difluoride membrane for Western blotting. Fusion proteins were detected by rabbit anti-Ras (1:100), and equal loading was confirmed by detection with rabbit anti-glucose-6-phosphate dehydrogenase (anti-G6PDH) (1:10,000).
To isolate primary human macrophages, 50 ml of blood from each donor was added to an equal volume of phosphate-buffered saline (PBS) and then separated by centrifugation over a Ficoll-Paque Plus (catalog no. GE17-1440-03; Sigma) gradient at 750 ϫ g for 20 min with no break. The lymphocyte/ monocyte layer was collected and incubated for 1 to 2 min with 1 ml ACK lysing buffer (catalog no. A10492-01; Gibco) to remove red blood cells. The cells were diluted to 50 ml in PBS and centrifuged 350 ϫ g for 10 min at 4°C. The supernatant was removed, and the cells were washed in 25 ml PBS and pelleted at 160 ϫ g for 15 min at 4°C. The cells were washed again in 25 ml PBS but centrifuged at 300 ϫ g for 10 min at 4°C. This final pellet was resuspended in 5 to 10 ml of RPMI 1640 supplemented with 10% human AB serum (catalog no. 35-060-CI; Corning). To differentiate into macrophages, cells were cultured in RPMI 1640 supplemented with 10% human AB serum for at least 4 h to allow for attachment. Cells were washed in PBS, which was then replaced with RPMI 1640 plus 10% human AB serum and 50 ng/ml human macrophage colony-stimulating factor (M-CSF) (catalog no. 216-MC-025; R&D Systems) for 7 days with the medium being changed every 1 or 2 days.
Molecular biology. Unless otherwise stated, all M. tuberculosis (Mtb) proteins were cloned from the BEI Resources Gateway Mtb ORF library using Gateway cloning technology (Life Technologies). The Mpt64 truncation mutants were PCR amplified (see Table S2 in the supplemental material) and cloned into pENTR (Life Technologies) prior to cloning into subsequent destination vectors.
hGH release assay and quantification. HeLa cells were plated in 24-well plates to achieve approximately 50,000 cells/well 24 h prior to transfection. Cells were cotransfected with 1 g hGH-CAD and 1 g GFP-Mtb effector or GFP alone using FuGene 6 (Promega) per the manufacturer's instructions. Cells were transfected 16 to 18 h at 37°C and 5% CO 2 . The transfection medium was then aspirated and replaced with DMEM containing 2 M D/D Solubilizer (catalog no. 635054; Clontech) and incubated for 2 h at 37°C and 5% CO 2 . The plates were centrifuged at 1,500 rpm for 5 min to pellet debris, and the culture supernatants were saved at Ϫ80°C prior to human growth hormone (hGH) quantification.
Released hGH was quantified by ELISA (catalog no. 11585878001; Roche). Briefly, samples were thawed on ice, and 20 l was transferred to each well containing 180 l sample buffer (1:10). The plate was incubated for 1 h at 37°C, washed five times in 250 l wash buffer, and incubated 1 h at 37°C with a polyclonal antibody to hGH conjugated to digoxigenin (anti-hGH-DIG). The plate was washed as described and incubated 1 h at 37°C with a polyclonal antibody to digoxigenin conjugated to peroxidase (anti-DIG-POD). The plate was washed and developed in peroxidase substrate (2,2'-azinobis [3ethylbenzothiazoline-6-sulfonic acid]-diammonium salt). The absorbance was read on a Biotek plate reader at 405 nm.
PIP strips membrane binding. 6xHIS-Mpt64_24-228 and 6xHIS-Mpt64_24-143 were purified by cobalt Talon affinity resin (Clontech). PIP strips membranes (catalog no. P23751; Invitrogen) were blocked for 1 h at room temperature in 3% fatty acid-free bovine serum albumin (BSA) (catalog no. A7030; Sigma) in TBST. Mpt64_24-228 or Mpt64_24-143 was diluted to 1.5 g/ml in 3 ml of 3% fatty acid-free BSA and incubated with the PIP strips for 3 h at room temperature with agitation. Membranes were washed three times in 3% fatty acid-free BSA prior to incubation with anti-Mpt64 or preimmune serum (1:3,000) overnight at 4°C with agitation. Membranes were washed three times in 3% fatty acid-free BSA and then incubated with HRP-conjugated donkey anti-rabbit (1:2,000) for 30 min at room temperature. Membranes were washed three times before detection of Mpt64 lipid interactions by chemiluminescence.
Transfection and colocalization of MSPs in HeLa cells. HeLa cells were transfected overnight with GFP fusion proteins using FuGene 6 transfection reagent (Roche). Cells were fixed in 4% paraformaldehyde (PFA) for 15 min, washed in PBS, and permeabilized in 0.25% Triton X-100 for 3 min at room temperature or in 100% methanol for 10 min at Ϫ20°C when using Tom20 antibody. Cells were stained with organelle-specific antibodies for 1 h at room temperature. Antibodies were visualized by secondary antibodies conjugated to Alexa Fluor 594. Cells were mounted in ProLong Gold plus DAPI (catalog no. P36931; Invitrogen), and z-stacks were collected on an AxioImager M2 microscope (Zeiss).
Detection of CHOP accumulation in macrophages. RAW267.4 cells stably expressing Mpt64ΔSP under a cytomegalovirus (CMV) promoter or control cells transduced with an empty lentivirus were seeded in 12-well plates at 5 ϫ 10 5 cells/well. To induce the unfolded protein response (UPR), the culture medium was replaced with media supplemented with 50 nM thapsigargin (catalog no. T9033; Sigma) or an equal volume of vehicle (dimethyl sulfoxide [DMSO]), and cells were incubated for 4 h. The cells were washed twice in PBS and lysed in ice-cold RIPA buffer supplemented with protease inhibitor tablets (catalog no. 11836153001; Roche). Lysates (15 to 25 g) were separated by SDS-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane for Western blotting. Accumulation of CCAAT enhancer-binding protein homologous protein (CHOP) was detected by mouse anti-CHOP (1:2,000), and band density was normalized to bands detected by the loading control mouse antiactin (1:3,000).
Infection and colocalization of Mpt64 in macrophages. Bacteria were washed repeatedly in PBS and then sonicated to create a single-cell suspension. RAW267.4 cells were infected in DMEM plus 10% horse serum (catalog no. 26050088; Invitrogen) at a multiplicity of infection (MOI) of 20:1 with mycobacteria expressing mCherry. Cells were centrifuged at 1,500 rpm for 10 min to permit bacterial attachment and then allowed to phagocytose for 1.5 h at 37°C and 5% CO 2 . Cells were fixed after 4 h postinfection in 4% PFA for 60 min. Cells were permeabilized in 0.25% Triton X-100 for 3 min at room temperature and then blocked in 5% normal donkey serum (Sigma). Mpt64 was detected with rabbit anti-Mpt64 antibody (1:500) and an HRP-conjugated goat anti-rabbit secondary antibody (1:1,000; Santa Cruz). Antibody signal was amplified by the addition of biotinylated tyramide (1:50; PerkinElmer) with detection by Alexa Fluor 488-conjugated streptavidin (1:250; Jackson Immunoresearch) or cyanine 5 tyramide (1:50; PerkinElmer). Z-stack slices were acquired with an AxioImager M2 microscope (Zeiss).
Macrophage infections for CFU. Primary human macrophages were seeded in low-evaporation 24-well plates at approximately 5 ϫ 10 5 cells/well. Bacteria were washed repeatedly in PBS and then sonicated to create a single-cell suspension. Macrophages were infected in RPMI 1640 plus 10% human AB serum at a MOI of 0.1:1. Cells were centrifuged at 1,500 rpm for 10 min to permit bacterial attachment and then allowed to phagocytose for 15 min at 37°C and 5% CO 2 . The cells were washed in PBS and then replaced with RPMI 1640 plus 10% human AB serum, and cells were washed every day between time points. The cells were lysed at time zero and subsequent time points in 500 l of0.5% Triton X-100 in PBS. Serial dilutions were plated on Middlebrook 7H11 plates, and colonies were enumerated after 2 to 3 weeks.
Construction of the Mtb mpt64 deletion mutant and complementation. An in-frame mpt64 deletion in Mtb was made using mycobacteriophage as previously described (112). Briefly, 500 bp 5= to the mpt64 start codon and 500 bp 3= to the mpt64 stop codon were amplified from Erdman genomic DNA (Table S2) and sequentially cloned into the multiple cloning sites of pMSG360HYG. This vector was linearized with AflII and DraI (catalog no. R0520 and R0129; New England Biolabs [NEB]) and transformed into E. coli EL350/phAE87 by electroporation. Phagemid DNA was isolated from pooled colonies and transformed into Mycobacterium smegmatis by electroporation. Plaques were isolated and pooled from M. smegmatis lawns, and high-titer phage was produced. Log-phase M. tuberculosis Erdman was transduced with phage at 42°C for 4 h. Mutants were selected on 7H11 plus hygromycin (100 g/ml). Wild-type mpt64 strain and a mpt64 strain lacking its secretion signal were cloned into an integrating vector containing a constitutive promoter (pMV306_MSP), conferring zeocin resistance. The MtbΔmpt64 was transformed by electroporation, and complements were selected on 7H11 plus zeocin (25 g/ml).
To confirm expression and secretion of Mpt64 complements, Mtb strains were grown to late-log phase and pelleted by centrifugation. The culture supernatants were saved and passed twice through 0.22-m filters. Bacterial pellets were boiled for 30 min in lysis buffer (50 mM Tris [pH 7.4], 150 mM NaCl) supplemented with Complete Mini protease inhibitor and then subjected to bead beating to lyse the cells. Protein content in lysates was determined by Bradford assay. Mpt64 expression in the lysates and culture supernatants was detected by Western blotting using a rabbit polyclonal antibody to Mpt64 (1:10,000). Equal loading of samples in the lysates and supernatants was confirmed by Western blotting with anti-GroEL2 (1:500) and anti-antigen 85 (1:1,000), respectively.
Mouse infections. Female BALB/c mice (The Jackson Laboratory) were infected via aerosol as described previously (149). Briefly, mid-log-phase Mtb bacteria were washed in PBS repeatedly and then sonicated to create a single-cell suspension. Bacteria were resuspended to yield an OD 600 of 0.1 in PBS. This suspension was transferred to the nebulizer of a GlassCol aerosolization chamber calibrated to infect mice with ϳ100 bacteria per animal. On the day of infection, whole lungs were collected from five mice per group, homogenized, and plated on Middlebrook 7H11 to determine the initial inoculum. At subsequent time points, the left lung, spleen, and left lobe of the liver were used to determine CFU, while the right lung was insufflated with 10% neutral buffered formalin for histopathology.
Lysozyme pulldown. E. coli lysates containing six-histidine (6xHIS)-tagged Mpt64 or an unrelated protein Cor were incubated with cobalt affinity resin (Talon; Clontech) to bind histidine-tagged proteins. After extensive washing, 1 mg/ml of either hen egg white lysozyme (catalog no. BP535; Fisher Scientific) or human lysozyme (catalog no. L1667; Sigma) was forced to flow over the immobilized beads and incubated for 5 min. The beads were washed two more times before proteins were eluted with 300 mM imidazole.
Mtb genomic DNA isolation. Late-exponential-phase Mtb was collected by centrifugation and washed once in PBS. Pellets were boiled for 20 to 30 min to sterilize. Pellets were washed once in GTE (25 mM Tris [pH 8.0], 10 mM EDTA, 50 mM glucose) and incubated overnight in lysozyme solution (10 mg/ml in GTE) at 37°C. Samples were incubated in 10% SDS and 10 mg/ml proteinase K for 40 min at 55°C, followed by incubation in NaCl and CTAB (2.4 M NaCl, 274 mM cetrimonium bromide [catalog no. H9151; Sigma]) at 60°C for 10 min. Genomic DNA was then isolated using a phenol-chloroform extraction, followed by ethanol precipitation.
Extraction of apolar lipids and PDIM analysis. Log-phase Mtb or MtbΔmpt64 were synchronized to OD 600 of 0.2 in Middlebrook 7H9 supplemented with 0.01% Tween 80 and grown for 24 h. Bacteria were collected by centrifugation at 1,600 ϫ g for 10 min, resuspended in 1 ml of 15% isopropanol, and transferred to a glass tube containing 5 ml chloroform-methanol (17:1, vol/vol), and incubated 24 h at room temperature. Samples were centrifuged at 1,600 ϫ g for 5 min, and the apolar lipids were collected from the bottom organic layer and dried. Apolar lipids were resuspended in 1.5 ml of 100% methanol. Tween 80 was removed by the addition of cobalt/thiocyanate solution and vortexed. The remaining lipids were extracted by the addition of 4 ml hexane. After centrifugation, the organic layer was saved and the aqueous layer was reextracted with 4 ml hexane. Both hexane fractions were combined, dried, and resuspended in 1 ml chloroform-methanol (2:1, vol/vol). The phthiocerol dimycocerosate (PDIM) standard was similarly resuspended. The PDIM standard or apolar lipids extracted from Mtb or MtbΔmpt64 were infused into an AbSciex TripleTOF 5600/5600ϩ mass spectrometer. Samples were analyzed in the positive mode.
Statistical analysis. Statistical analysis was performed using GraphPad Prism software. For in vitro studies, one-tailed analysis of variance (ANOVA) tests were used for experiments with multiple comparisons using Dunnett's test. For experiments with single comparisons, two-tailed unpaired Student's t test was used. For experiments containing samples with nonnormal distributions such as in vivo CFU measurements and area of lung inflammation, the Kruskal-Wallis nonparametric test was used with Dunn's correction for multiple comparison. Analysis of survival studies was performed by Kaplan-Meier test.
Ethics statement. Primary human macrophages were isolated from buffy coats from anonymous donors provided by a local blood bank (Carter Bloodcare). This study was reviewed by the University of Texas (UT) Southwestern Institutional Review Board and deemed to be exempt.
Animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Texas Southwestern (protocol 2017-102086) and followed the eighth edition of the Guide for the Care and Use of Laboratory Animals (150). The University of Texas Southwestern is accredited by the American Association for Accreditation of Laboratory Animal Care (AAALAC).