The Cell Wall Lipid PDIM Contributes to Phagosomal Escape and Host Cell Exit of Mycobacterium tuberculosis

TFRs are among the most abundant family of transcriptional regulators in bacteria and as such have been extensively studied (26, 27). Rv3167c is annotated as a TFR since 89% of its protein sequence can be modeled by the Phyre2 software with 99.9% confidence using the highest-scoring template TFR (SCO0332) of Streptomyces coelicolor. The predicted structure depicts the N terminus containing 3 α-helices that form a typical helix-turn-helix DNA binding domain common to TFRs and the additional 6 α-helices that constitute the C-terminal ligand binding and dimerization domains (Fig. 1a). TFRs typically function as homodimers (26, 27). In order to test whether Rv3167c can dimerize, we made use of the mycobacterial protein fragment complementation (M-PFC) assay (28), where bacterial growth is seen on agar containing trimethoprim when the proteins of interest interact. As shown in Fig. 1b, growth was observed when Rv3167c was dually expressed, indicating that Rv3167c interacts with itself and likely forms a dimer. Depicted are the Rv3167c interaction, the positive control (GCN4, a yeast transcriptional regulator known to dimerize), and the negative control (empty plasmids). The full-length image, including all negative controls, can be found in Fig. S6 in the supplemental material. Additionally, we demonstrated that purified Rv3167c (molecular mass of 24 kDa) has a retention volume identical to that of ovalbumin (molecular mass of 43 kDa), indicating again that Rv3167c forms a homodimer (data not shown). TFRs are typically encoded in the genome in close proximity and in opposing orientation to the genes they regulate (29). Figure 1c depicts the genomic organization of the Rv3167c locus with the sequence of the intergenic region between Rv3167c and the adjacent operon, Rv3168-Rv3169. Based on transcriptome sequencing (RNA-seq) data, Rv3168 and Rv3169 are two of the most highly upregulated genes in the M. tuberculosis ΔRv3167c mutant (see Fig. S1 in the supplemental material), suggesting direct regulation by Rv3167c.

TFRs are frequently autoregulatory and bind to palindromic DNA sequences (26, 27). Interestingly, a palindrome (palRv3167c) exists within the Rv3167c-Rv3168 intergenic region (sequence highlighted in red in Fig. 1c). Complementary oligonucleotides corresponding to both strands of palRv3167c were annealed to form a double-stranded product, which was incubated with purified Rv3167c and analyzed by electrophoresis mobility shift assay (EMSA). As shown in Fig. 1d, incubation of increasing concentrations of Rv3167c with palRv3167c resulted in an increased mobility shift indicating binding of Rv3167c to the sequence. Importantly, when Rv3167c was incubated with randomized double-stranded DNA of similar size, no shift was observed, demonstrating that Rv3167c’s interaction with palRv3167c is specific. We further characterized Rv3167c interaction with palRv3167c using isothermal titration calorimetry (ITC). Figure 1e is a representative example of three independent ITC experiments. Rv3167c bound palRv3167c in a 1:1 protein-DNA stoichiometry and had a dissociation constant of 12.3 ± 3.2 μM, which is similar to other TFR interactions with DNA. Importantly, Rv3167c demonstrated no significant interaction with randomized control DNA (see Fig. S2 in the supplemental material). Taken together, these data indicate that Rv3167c functions as a typical TFR that represses its own promoter and also negatively regulates the downstream gene(s)—in this case Rv3168 and Rv3169.

In order to characterize the regulon of Rv3167c, we performed RNA-seq on the wild-type M. tuberculosis, M. tuberculosis ΔRv3167c deletion mutant, and M. tuberculosis ΔRv3167c complement (ΔRv3167::Comp) strains grown in standard growth media and harvested during logarithmic growth. Our RNA-seq data were highly reproducible, as evidenced by the high degree of clustering between biological replicates, and indicated that we are analyzing distinct transcriptional states (Fig. 2a). The inclusion of the M. tuberculosis ΔRv3167::Comp strain in our analysis provided a valuable internal control for determining the Rv3167c regulon. To do this, we determined genes differentially expressed in the M. tuberculosis ΔRv3167c mutant compared to M. tuberculosis as well as genes differentially expressed in the M. tuberculosis ΔRv3167c strain compared to the ΔRv3167c::Comp strain. We compared both gene sets (Fig. 2b) and only considered genes differentially expressed in both contrasts as this gave us more confidence Rv3167c was involved in their regulation. In total, 442 genes were considered to be regulated by Rv3167c (see Table S1b in the supplemental material). The differential expression of such a number of genes indicates that Rv3167c has a relatively broad transcriptional impact. Gene Ontology (GO) enrichment analysis identified three GO terms (Table 1). The GO term “DIM/DIP cell wall layer assembly,” which encompasses genes associated with the synthesis and export of PDIM, was enriched in our gene set. Indeed, the entire PDIM operon was upregulated in the M. tuberculosis ΔRv3167c mutant compared to wild-type M. tuberculosis, while being mostly unchanged or even downregulated in the M. tuberculosis ΔRv3167::Comp strain compared to the wild type, indicating these genes do complement back to wild-type levels (Fig. 2c). Using thin-layer chromatography (TLC), we showed that the transcriptional upregulation manifested in an increase in PDIM lipid in the M. tuberculosis ΔRv3167c mutant (Fig. 2d).

PDIM is synthesized in the cytosol and transported to the outer membrane of M. tuberculosis by the membrane lipid transporter mmpl7. Deletion of mmpl7 results in the failure of proper PDIM localization and the accumulation of the lipid in the cytosol (12). To assess the role of PDIM in the necrosis induced by the M. tuberculosis ΔRv3167c mutant, we deleted mmpl7 in the M. tuberculosis ΔRv3167c background, resulting in the ΔRv3167c Δmmpl7 double deletion mutant, which was confirmed by PCR (see Fig. S3 in the supplemental material). Infection of THP-1 macrophages with the M. tuberculosis ΔRv3167c Δmmpl7 mutant resulted in a significant decrease in cell death compared to the ΔRv3167c strain (Fig. 3a). As described above, RNA-seq analysis revealed more than 400 genes differentially regulated in the M. tuberculosis ΔRv3167c strain. In an effort to minimize concerns about the contribution of other genes to the cell death phenotype and to assess whether PDIM alone was sufficient to modulate host cell death, we analyzed an mmpl7 transposon insertion mutation (mmpl7::TN) in the M. tuberculosis CDC1551 strain obtained from BEI Resources (30). The deletion of mmpl7 resulted in significantly less cell death than wild-type M. tuberculosis (Fig. 3c; see Fig. S9a in the supplemental material). Complementation of mmpl7 back into the M. tuberculosis mmpl7::TN mutant reverted the phenotype to levels similar to those of the wild type (Fig. 3c). A clean deletion mutant generated through a recombination strategy (the M. tuberculosis Δmmpl7 mutant) confirmed this finding (see Fig. S7a in the supplemental material). Thin-layer chromatography confirmed that these mutants produce similar levels of PDIM to the wild type (see Fig. S8 in the supplemental material), confirming these mutants are deficient in export of PDIM, not synthesis. Previous work has implicated PDIM in macrophage invasion (15) and in resistance to killing by an early innate immune response (16, 18, 19). However, we saw similar numbers of bacteria both initially after infection and at 24 h postinfection in the M. tuberculosis ΔRv3167cΔmmpl7 (Fig. 3b) and mmpl7::TN (Fig. 3d) mutants compared to wild-type M. tuberculosis. These data thus confirm that the bacterial PDIM levels directly impact the capacity of M. tuberculosis to induce host cell necrosis and do not just reflect a lower quantity of intracellular bacteria.

Macroautophagy (autophagy) is an evolutionarily conserved catabolic process that involves sequestration of cytosolic contents into de novo-generated vesicles termed autophagosomes, which then get delivered to the lysosome for degradation. Intracellular pathogens can be targeted and killed by autophagic machinery in a process termed xenophagy (31). Previous work in our lab demonstrated that in addition to increased necrosis, infection of cells with the M. tuberculosis ΔRv3167c mutant resulted in increased induction of autophagy but not xenophagy (25). In order to determine whether PDIM played a role in M. tuberculosis-induced autophagy in addition to the induction of necrosis, we monitored the conversion of cytosolic LC3I to autophagosome-bound LC3II, a hallmark of autophagy, using LC3-green fluorescent protein (GFP)-expressing THP-1 cells and flow cytometry (32). THP-1 LC3-GFP-expressing cells were infected for 24 h and partially lysed with saponin, and the remaining LC3-GFP was measured using flow cytometry (Fig. S9b for gating and raw data). A significant decrease in autophagy was seen in cells infected with the M. tuberculosis mmpl7::TN mutant compared to wild-type M. tuberculosis (Fig. 4a), a finding confirmed by analysis of the M. tuberculosis Δmmpl7 mutant (Fig. S7b) indicating PDIM plays a role in induction of autophagy in M. tuberculosis.

The M. tuberculosis ΔRv3167c mutant accesses the cytosol earlier and to a greater extent than wild-type M. tuberculosis (25). We hypothesized that PDIM contributes to phagosomal escape and that loss of PDIM would result in decreased access to the cytosol. To test this, we made use of a fluorescence resonance energy transfer (FRET)-based assay described previously (21, 22). Briefly, THP-1 cells were loaded with CCF4-AM. Intact CCF4-AM emits green fluorescence (535 nm) due to FRET between the fluorescent moieties. Cleavage of CCF4-AM by β-lactamase expressed by cytosolic bacteria leads to FRET loss and a shift in the emission wavelength to 450 nm, which can be monitored by flow cytometry (Fig. S9c). Using this assay, we demonstrate that the M. tuberculosis mmpl7::TN mutant accesses the cytosol to a lesser extent than wild-type M. tuberculosis and the mmpl7::TN complemented strain (Fig. 4b). Previous work has shown that phagosomal escape of M. tuberculosis is dependent on the ESX-1-secreted substrate EsxA (33). We observed no difference in the secretion of EsxA in the M. tuberculosis mmpl7::TN mutant compared to wild-type M. tuberculosis (see Fig. S5 in the supplemental material). This represents the first evidence, to our knowledge, that PDIM influences the phagosomal escape of M. tuberculosis.

All M. tuberculosis cultures were grown at 37°C in 7H9 broth or on 7H11 agar medium with 1× ADC enrichment (albumin-dextrose-catalase), 0.5% glycerol, and 0.05% Tween 80 (broth). Escherichia coli cultures were grown in LB medium. Kanamycin, hygromycin, and zeocin were used at 40, 50, and 100 μg/ml, respectively, for M. tuberculosis cultures and 40, 150, and 100 μg/ml, respectively, for E. coli.

THP-1 cells from ATCC were maintained at 37°C and 5% CO₂ in RPMI (ATCC) supplemented with 10% heat-inactivated fetal calf serum (FCS) (growth medium). THP-1 LC3-GFP-expressing cells were grown in growth medium plus 100 μg/ml G418 sulfate (Cellgro). For infections, THP-1 cells were treated with 20 ng/ml phorbol 12-myristate 13-acetate (PMA) (Sigma) for 24 h to differentiate. Cells were washed twice in phosphate-buffered saline (PBS) and infected at a multiplicity of infection (MOI) of 3:1 in growth medium plus 5% human serum AB (Sigma). The infected cells were incubated at 37°C and 5% CO₂ for 4 h followed by two more PBS washes and chased with growth medium plus 100 μg/ml gentamicin (Gibco). The time point indicated as “0 h” refers to immediately after 4 h of infection and washes, while “24 h” refers to the 24-h chase period after 4 h of infection.

DNA recombinant techniques were carried out following standard procedures. All restriction and modifying enzymes were purchased from Fermentas. M. tuberculosis mutants were constructed using a recombineering approach and specialized phage transduction as described previously (47). Knockouts were confirmed by PCR. Sequences of all primers used in this study can be found in Table S1. The following reagent was obtained through BEI Resources, NIAID, NIH: Mycobacterium tuberculosis strain CDC1551 (NR-13649), transposon mutant 1224 (MT3012/Rv2942/mmpl7: open reading frame [ORF] size, 2,763; point of insertion, 354; NR-14743). Complementation of the mmpl7 transposon and deletion mutants was carried out by expression of full-length mmpl7 from the episomal plasmid pMAN-1, kindly provided by Jeff Cox (48). Primers used for complementation can be found in Table S1a.

Rv3167c was cloned into pET28c (Novagen) using the BamHI and NdeI restriction sites to generate an N-terminal 6-histidine tag. The cloned construct was transformed into E. coli BL21-DE3. BL31-DE3::pET28c-Rv3167c was grown in LB plus kanamycin to an optical density at 600 nm (OD₆₀₀) of 0.5, after which expression of Rv3167c was induced with the addition of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for 5 h. The culture was harvested, resuspended in PBS plus 200 mM NaCl, and disrupted by sonication. Cellular debris was removed by spinning at 15,000 × g for 30 min. The supernatant was applied to columns packed with His-Pur cobalt Superflow resin (Thermofisher) equilibrated with PBS plus 200 mM NaCl plus 10 mM imidazole and incubated at 4°C for 1 h with rocking. The column was washed with 10 volumes of PBS plus 200 mM NaCl plus 30 mM imidazole. Rv3167c was eluted from the column with PBS plus 200 mM NaCl plus 250 mM imidazole. Ten percent glycerol was added, and the purified protein was stored at 4°C until fast protein liquid chromatography (FPLC) purification.

All FPLC experiments were carried out on a GE AKTA Explorer chromatography system. Gel filtration chromatography was carried out on a Superdex 75HR 10/30 column (GE Healthcare Life Sciences). Eluent was then further purified with ion-exchange chromatography on a MonoQ 5/50 Gl column (GE Healthcare Life Sciences). Purified protein was confirmed to be Rv3167c by mass spectrometry. An LMW gel filtration calibration kit (Amersham Biosciences) was used as the standard for gel filtration chromatography determination of Rv3167c dimerization.

Isothermal titration calorimetry (ITC) was performed on a MicroCal iTC200 titration microcalorimeter with a starting protein concentration of 0.130 mM and a starting double-stranded DNA (dsDNA) concentration of 2.10 mM. Immediately before the experiments were conducted, the protein was dialyzed against 10 mM phosphate (pH 7.2), 136 mM NaCl, and 4 mM KCl to remove glycerol. DNA oligonucleotides were annealed in the same buffer as the protein. Each titration experiment consisted of 2-μl injections of dsDNA into protein at 25°C with a mixing speed of 1000 rpm. Data acquisition and analysis were performed with the software package ORIGIN according to a single-site binding model.

Two oligonucleotides per tested binding site were purchased from Operon, corresponding to the plus and minus strands of each site. To anneal, corresponding oligonucleotides were mixed at a 1:1 molar ratio and heated to 95°C for 10 min in a heat block, after which the block was turned off and allowed to slowly cool to room temperature. Annealed oligonucleotides were mixed with purified Rv3167c at the indicated concentrations in a buffer containing PBS plus 100 mM NaCl plus 5 mM MgCl₂ and incubated on ice for 15 min. Samples were then separated on a 5% Tris-borate-EDTA (TBE) PAGE gel (Bio-Rad) at 150 V for 1 h at 4°C. The gel was then stained with SYBR green (Thermo, Fisher) and imaged. Randomized oligonucleotides served as the negative control. All primer sequences are listed in Table S1a.

For propidium iodide (PI) staining, at the indicated time points, cells were harvested and washed once in PBS and then resuspended in PBS plus 5% FCS and 1 μg/ml PI and incubated for 10 min. The percentage of PI-positive cells was determined by flow cytometry (BD Accuri C6). Analysis of LC3-GFP-expressing THP-1 cells was performed as described previously (21). At the indicated time points, cells were harvested and washed once with PBS. Cells were then permeabilized with 0.05% saponin for 5 min, after which they were washed once in PBS and resuspended in PBS plus 5% FCS. The percentage of autophagic cells was determined by flow cytometry (BD Accuri C6). A representative example of the flow cytometry gating strategy to determine PI and LC3-GFP positivity can be found in Fig. S9a and b.

Mycobacterial protein fragment complementation (M-PFC) was conducted as described in reference 28. Rv3167c was cloned into both pUAB300 and PUAB400. Constructs were then cotransformed into Mycobacterium smegmatis. Transformants were then spotted with appropriate controls on 7H11 medium supplemented with 50 μg/ml trimethoprim (Fisher) and grown at 37°C for 3 days.

To detect mycobacterial escape from the phagosome, the CCF4 FRET assay was performed as described previously (21). Briefly, cells were stained with 8 μM CCF4 (Invitrogen) in EM buffer (120 mM NaCl, 7 mM KCl, 1.8 mM CaCl₂, 0.8 mM MgCl₂–5 mM glucose, 25 mM HEPES [pH 7.3]) containing 2.5 μM probenecid (Sigma-Aldrich) for 1.5 h at room temperature. Live populations were distinguished from dead ones by addition of LIVE/DEAD fixable red stain (Invitrogen) for 30 min at room temperature. Following staining, cells were fixed with 4% paraformaldehyde (PFA) overnight and analyzed by flow cytometry (BD FACSCanto). Representative examples of the flow cytometry gating strategy to determine loss of FRET in infected cells can be found in Fig. S9c.

M. tuberculosis cultures were grown in 7H9 until they reached an OD₆₀₀ of 0.6 to 0.8. Cultures were pelleted and resuspended in 1 ml TRIzol (Ambion). RNA was extracted with chloroform, precipitated with 100% isopropanol, and washed with 70% ethanol. Purified RNA was treated with Turbo DNase (Ambion) for 1 h. RNA was used right away or stored at −80°C.

The Maxima first strand cDNA synthesis kit (Thermo Scientific) was used to synthesize cDNA as per the manufacturer’s instructions.

Ribosomal RNA was removed from samples using the Epicentre Gram-positive Ribo-Zero rRNA magnetic removal kit. RNA-seq libraries were prepared using the Illumina ScriptSeq v2 library preparation kit according to the manufacturer’s protocol. Library quality was assayed by Bio-analyzer (Agilent) and quantified by quantitative PCR (qPCR) (Kapa Biosystems). Libraries were run on an Illumina HiSeq1500, generating 100-bp paired end reads. RNA-seq read quality was verified using FastQC, and low-quality base pairs were trimmed using Trimmomatic. The M. tuberculosis H37Rv reference genome was downloaded from TB Database, and reads were mapped to the genome using TopHat2. Count tables were generated using HTSeq along with a custom GFF, which includes features from both TB Database and Tuberculist. Count tables were loaded into R/Bioconductor, quantile normalized, and corrected for variance bias using Voom. Limma was used to find genes that were differentially expressed between each of the pairs of sample conditions. A total of 442 genes were found to be both differentially expressed in the wild-type–mutant contrast and the mutant-complement contrast and thus are potentially affected by Rv3167c regulation. This list of candidate deregulated genes was checked for functional enrichment using GOseq with multiple-testing correction using the Benjamini and Hochberg method (49).

Cultures were grown in 7H9 broth in a volume of 50 ml to an OD₆₀₀ of 0.8. Cultures were pelleted and resuspended in 2:1 methanol-chloroform and left at 4°C overnight. Samples were spun down, and supernatant was transferred to a glass beaker. The pellet was resuspended in 1:1 methanol-chloroform and left at room temperature for an hour. Samples were spun down, and the supernatant was removed. The process was repeated again with 1:2 methanol-chloroform. All three fractions were pooled, and the methanol and chloroform were allowed to evaporate overnight in a fume hood. The total lipid pellet was weighed, and 300 μg of total lipid was spotted onto a TLC plate. The plate was resolved in a mobile phase of 9:1 petroleum ether-diethyl ether. Lipid spots were revealed by spraying the plate with a solution of 3% cupric acetate in 8% phosphoric acid followed by baking at 140°C.

Bacteria were pelleted and resuspended in 25 mM Tris HCl, followed by bead beating to lyse cells. Lysate was cleared by spinning at 12,000 rpm for 5 min. The Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Scientific) was used to measure protein concentrations. Antibody against EsxA was purchased from Santa Cruz Biotechnology, Inc., and used at a 1:200 dilution. Antibodies against Fap and GroEL were obtained from BEI Resources and used at 1:7,500 and 1:50 dilutions, respectively.

Statistical analysis was performed using GraphPad Prism version 6.0 software. Data are presented as means ± standard errors of the means (SEM) from three independent experiments, and one-way analysis of variance (ANOVA) with Tukey’s posttest was used unless mentioned otherwise in the figure legends. P value significance is indicated on the figures as follows: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; and ****, P ≤ 0.0001.