Mycobacterium tuberculosis universal stress protein Rv2623 interacts with the putative ATP binding cassette (ABC) transporter Rv1747 to regulate mycobacterial growth

Abstract

We have previously shown that the Mycobacterium tuberculosis universal stress protein Rv2623 regulates mycobacterial growth and may be required for the establishment of tuberculous persistence. Here, yeast two-hybrid and affinity chromatography experiments have demonstrated that Rv2623 interacts with one of the two forkhead-associated domains (FHA I) of Rv1747, a putative ATP-binding cassette transporter annotated to export lipooligosaccharides. FHA domains are signaling protein modules that mediate protein-protein interactions to modulate a wide variety of biological processes via binding to conserved phosphorylated threonine (pT)-containing oligopeptides of the interactors. Biochemical, immunochemical and mass spectrometric studies have shown that Rv2623 harbors pT and specifically identified threonine 237 as a phosphorylated residue. Relative to wild-type Rv2623 (Rv2623WT), a mutant protein in which T237 has been replaced with a non-phosphorylatable alanine (Rv2623T237A) exhibits decreased interaction with the Rv1747 FHA I domain and diminished growth-regulatory capacity. Interestingly, compared to WT bacilli, an M. tuberculosis Rv2623 null mutant (ΔRv2623) displays enhanced expression of phosphatidyl-myo-inositol mannosides (PIMs), while the ΔRv1747 mutant expresses decreased levels of PIMs. Animal studies have previously shown that ΔRv2623 is hypervirulent, while ΔRv1747 is growth-attenuated. Collectively, these data have provided evidence that Rv2623 interacts with Rv1747 to regulate mycobacterial growth; and this interaction is mediated via the recognition of the conserved Rv2623 pT237-containing FHA-binding motif by the Rv1747 FHA I domain. The divergent aberrant PIM profiles and the opposing in vivo growth phenotypes of ΔRv2623 and ΔRv1747, together with the annotated lipooligosaccharide exporter function of Rv1747, suggest that Rv2623 interacts with Rv1747 to modulate mycobacterial growth by negatively regulating the activity of Rv1747; and that Rv1747 might function as a transporter of PIMs. Because these glycolipids are major mycobacterial cell envelope components that can impact on the immune response, our findings raise the possibility that Rv2623 may regulate bacterial growth, virulence, and entry into persistence, at least in part, by modulating the levels of bacillary PIM expression, perhaps through negatively regulating the Rv1747-dependent export of the immunomodulatory PIMs to alter host-pathogen interaction, thereby influencing the fate of M. tuberculosis in vivo.

Fig 1. M. tuberculosis Rv2623 interacts with Rv1747.

(A) The primary structure of Rv1747 with 2 FHA domains (Orange: FHA I & FHA II); elements typical of ABC transporters: NBD (nucleoside-binding domain; 1–559 amino acids; Walker A&B (Yellow; the ATP-binding domain), and transmembrane domain (Blue bars: transmembrane helices). (B) The GAL4-based Matchmaker Gold Yeast Two-Hybrid system was used to identify interacting partners of Rv2623, which was cloned into the pGBKT7 vector as a fusion to the GAL4 DNA-binding domain (pGBKT7::Rv2623). A DNA library of M. tuberculosis Erdman prey proteins were expressed as fusions to the Gal4 activation domain using pGADT7AD. The screen revealed that Rv2623 interacts with the N-terminal FHA I domain of Rv1747. Analysis of a re-cloned full-length Rv1747 FHA I domain validated the interaction (B, bottom panel: pGBKT7::Rv2623/pGADT7AD::FHA I). Rv2623 dimerization was exploited to serve as positive control (B, top panel; pGBKT7::Rv2623/pGADT7AD::Rv2623). pGBKT7::Rv2623/pGADT7AD and pGBKT7/pGADT7AD::FHA I represent negative controls. The interaction was further confirmed by affinity chromatography study (C). Purified histidine (His₆)-tagged Rv2623 (Rv2623) and FLAG-tagged FHA I (FHA I: first 100 amino acids of Rv1747) were expressed in M. smegmatis mc²155. Purified FLAG-tagged FHA I was passed over columns with or without Rv2623 immobilized onto the Nickel (Ni)-NTA resin. Western analyses of the appropriate elution fractions using anti-Rv2623 and anti-FLAG antibodies revealed that Rv2623 and Rv1747 FHA I co-eluted—upper and lower panels of lane 3 represent the results of probing eluents from column containing both (Ni)-NTA resin-immobilized (His₆)-tagged Rv2623 and FLAG-tagged Rv1747 FHA I with anti-Rv2623 and anti-FLAG antibody, respectively—thus demonstrating interaction of these two mycobacterial components. Lane 1: upper panel and lower panel represent results of reacting eluents from column with only Rv2623 with the appropriate antibody. The upper and lower panels of Lane 2 depict reactivity of eluents from column harboring only FHA I with the appropriate antibody. Lane 4 of upper and lower panels represent recombinant Rv2623 (+ve Rv) and FLAG-FHA I (+FHA); respectively, loaded as positive controls. α-Rv2623 and α-FLAG: anti-Rv2623 and anti-FLAG antibodies; respectively. Arrows indicated the molecular weight of His-tagged Rv2623 (~32.31 kDa)) and FLAG-tagged Rv1747 FHA I (~12 kDa; expressed as the first 100 amino acids of Rv1747).

Fig 2. Superposition of S. cerevisiae Rad53 FHA domain with Rv1747 FHA I.

(A) PyMol (www.pymol.org) depiction of the surface of the S. cerevisiae Rad53 FHA based on solved structures: R70, S85, and N107, the three residues in the conserved FHA domain region that have been shown to play significant roles in mediating interaction with the phosphothreonine (pT)-containing peptide motif (pTEAD) [24,32,33], are located on the surface of the yeast Rad53 FHA domain. (B) Ribbon diagram (www.pymol.org) of Rad53 FHA domain demonstrating interaction between R70, S85, and N107 of Rad53 with the pTEAD residues of its interacting partner’s conserved FHA-binding motif [24,32,33]. (A&B) H88 and G69 residues of the Rad53 FHA domain, which play a role in stabilizing the interaction between the Rad53 FHA domain and its interacting partner [24,32,33], are non-surface located. (C) The homology model of Rv1747 FHA I domain was generated via the M4T server ver 3.0 [34] based on comparative modeling using a combination of 2 templates (PDB codes 2LC1 and 1UHT). The homology model of the Rv1747 FHA I domain was then superimposed onto the Rad53 FHA domain [24,32,33] using Pymol (www.pymol.org): Note the superposition of conversed amino acids shown in Rad53 (N107, H88, S85, G69, R70) to play important roles in recognizing the pT-containing motif (pTEAD) [24,32,33] with the corresponding residues of Rv1747 FHA I domain (N69, H50, S47, G32, R33). (D) Alignment of various FHA domains with the prototypic Rad53 FHA1 and FHA2 has revealed near complete match of spacing between the conserved residues of Rad53 FHA1, known to participate in the interaction with the phosphorylated FHA domain-binding motif of its interacting partner (G69, R70, S85, H88, N107), with those of the M. tuberculosis Rv1747 FHA I domain (G32, R33, S47, H50. N69), except for one amino acid difference between Rad53 R70/S85 and the Rv1747 R33/S47 spacing. KAPP: kinase associated protein phosphatase of Arabidopsis thaliana [72]; KIAA: also known as KIAA0710/NFBD1 (nuclear factor BRCT domain 1) [73,74]. Note the highly conserved N, S, R (green asterisks: surface location) and H and G (Grey asterisks: non-surface location) residues.

Fig 3. The Rv2623 threonine residues.

(A) The solvent accessible surface (SAS) of threonine residues and the corresponding OH group in each monomer of the dimeric Rv2623 was calculated based on the crystal structure of the USP (PDB ID 3CIS) [6] using AREAIMOL program from CCP4 suite [35,36]. Among the 9 Thr residues, 5 have solvent accessible hydroxyl oxygen atoms (Shaded: T90, T103, T212, T237, T280). T90, T103, T212, and T237 also have a previously reported phosphorylation motif concerning the pT+3 residues (bracketed): pTXX(S) for T90, pTXX(D) for T103, pTXX(M) for T212, and pTXX(V) for T237 [24,32,33]. KAPP: kinase associated protein phosphatase of Arabidopsis thaliana [72]; KIAA: also known as KIAA0710/NFBD1 (nuclear factor BRCT domain 1) [73,74]; CDS-1: Checkpoint DNA synthesis protein kinase [75]; Y127_MYCTU: Cy1A11.16C [32], M. tuberculosis GarA [tuberculist.epfl.ch]. (B) The surface of the Rv2623 protein is displayed using PYMOL (www.pymol.org). The subunits A and B are colored in blue cyan and green respectively. The solvent accessible Threonine residues (T90, T103, T212, T237) are colored in yellow. (C) PYMOL display of a ribbon representation of an Rv2623 monomer based on previously solved structure [6] depicting the position of the four solvent accessible threonine residues with the corresponding pT+3 residues that have been shown to promote interaction with FHA domains [24,32,33]. The threonine and the pT+3 residues are labeled in yellow and cyan, respectively.

Fig 4. Rv2623 is post-translationally modified and phosphorylated at T237.

(A) Immunoblot of 2D-gel electrophoretically-resolved BCG lysates with anti-Rv2623 monoclonal antibody revealed three isoforms with differing isoelectric pH values, thus providing evidence for post-translational modification of the USP; (B) Dot-blot analysis of affinity-purified Rv2623 from M. smegmatis mc²155 (Top panel); M. tuberculosis Erdman (Mid panel), and M. bovis BCG (Bottom panel) demonstrating immunoreactivity with an anti-pT antibody. “+” indicate positive pT control; soy bean trypsin inhibitor serves as negative controls (S9 Fig). (C) Mass spectrometry-based phosphomapping of Rv2623. The sequence coverage of the protein was ~90%. Manual examination of the appropriate MS/MS spectra was conducted to verify the phosphopeptides identified via software programs. The graph shown represents MS/MS spectra of m/z 507.24(2+) corresponding to peptide spanning amino acid residues 231–238 of Rv2623. Labelled are b- and y- fragment ions produced from Collisional Induced Dissociation (CID) in an LTQ-Orbitrap Elite LC MS/MS instrument. Phosphorylation on T237 of Rv2623 was determined based on the MS/MS fragmentation patterns; in particular are the observed loss of phosphoric acid (indicated as "y(n)-98Da") for fragment ions y(3), y(4), y(5), y(6), y(7), and b(7) under CID conditions [76]. The "⁺⁺" sign indicates that the assigned fragment ion is doubly charged. The graph shown represents results derived from analysis of in vitro phosphorylated M smegmatis-expressed recombinant Rv2623. Analysis of vitro phosphorylated E. coli-expressed recombinant Rv2623 yielded the same results. (D) Molecular docking was employed to depict the interaction of the pTRVV-containing FHA domain-binding motif of Rv2623 with Rv1747 FHA I [38] (https://life.bsc.es/servlet/pydock/home/). This was performed with the [amino acids 152 to 294] domain of Rv2623 (with T237 phosphorylated) on the [amino acids 1 to 90] domain of Rv1747 FHA I, using PyDockWeb, with the d(N64-V240) constrained. The phosphate was built into T237 by the PyTMs plugin in Pymol prior to the docking. The PDB code of Rv2623 is 3cis. The structure of the FHA domain of Rv1747 is taken from Fig 2 (B and C). The Threonine residue and the pT+3 residues in the binding motifs of Rv2623 are labeled in yellow and cyan, respectively.

Fig 5. T237 of Rv2623 plays an important role in the interaction between the USP and Rv1747 FHA I to regulate mycobacterial growth.

(A) Various T→A mutants of M. tuberculosis Rv2623 were overexpressed in M. smegmatis, and growth of the recipient cells monitored in the BACTEC 9000MB system in triplicates [6]. The time to detection reflects the rate of bacterial growth. pMV261: M. smegmatis harboring the pMV261 vector containing no Rv2623 constructs; WT: M. smegmatis overexpressing WT Rv2623 protein. The results are representative of three independent experiments. **p<0.01 (Student's t-test; WT vs. T103A); ***p<0.001 (Student's t-test; WT vs. T237A). (B) The melting temperature of Rv2623_WT (Orange) and Rv2623_T237A (Purple) are comparable. (C) These affinity chromatography studies, set up as described in Fig 2, used purified His₆-tagged Rv2623_WT (Rv2623) and His₆-tagged Rv2623_T237A (Rv2623_T237A) derived from the pQE80L system; and cMyc-tagged FHA I (designated FHA I: first 120 amino acids of Rv1747) expressed via LIC vector pMCSG7. Purified FHA I was passed over columns with or without Rv2623_WT or Rv2623_T237A immobilized onto the Nickel (Ni)-NTA resin. Western blot analyses of the appropriate elution fractions were conducted using anti-His and anti-cMyc antibodies. Lane 1: negative controls–column without any Rv2623 proteins. Lane 2 and Land 3: the columns harbored Rv2623_WT and Rv2623_T237A; respectively, as well as FHA I—eluents probed with the appropriate antibody revealed that relative to WT Rv2623 protein, the capacity of Rv2623_T237A to bind Rv1747 FHA I is diminished. Lane 4 represents positive controls for His₆-tagged Rv2623 (Rv +ve), and cMyc-tagged FHA I (FHA +ve); respectively. The Western blot shown is representative of three independent experiments. The blue bar graph depicts densitometric analysis of the relative binding or Rv2623 proteins to Rv1747 FHA I. Error bar: standard deviation; ***p < 0.0005. Arrows indicate the molecular weight of His-tagged Rv2623_WT and Rv2623_T237A proteins (~32.31 kDa) and cMyc-tagged Rv1747 FHA I domain (~14.3 kDa; expressed as the first 120 amino acids of Rv1747). (D) The ability of Rv2623_WT to retard growth in recipient bacterial cells upon overexpression (via the multi-copy pMV261 vector) in M. tuberculosis Erdman is reversible with a T237A mutation of Rv2623, as assessed by the BACTEC 9000MB. “–ve”: untransformed M. tuberculosis Erdman; “pMV261”: Erdman transformed with pMV261 alone. ***p<0.001 (Student's t-test; comparing T237A with WT). (E) The growth-regulatory attribute of Rv2623_WT and the various T→A mutants in recipient cells upon overexpression in virulent M. tuberculosis Erdman was assessed by monitoring OD_{600 nm} in supplemented 7H9 Middlebrook medium. The results presented above are representative of three independent experiments. Error bars: Standard deviations.

Fig 6. PIM expression by M. tuberculosis ΔRv2623 and ΔRv1747.

(A) M. tuberculosis WT and ΔRv2623 in stationary phase cultures in supplemented Middlebrook 7H9 medium were let to stand for 10 min (Left tube: M. tuberculosis; Right tube ΔRv2623). The deletion mutant displays a sedimentation phenotype distinct than that of WT. (B,C) M. tuberculosis ΔRv2623, plated on 7H10 agar medium supplemented with OADC, displays a colony morphology different than that of WT. (E) Analysis of M. tuberculosis lipid extracts by 2-dimensional thin layer chromatography (TLC) revealed that compared to WT, ΔRv2623 expresses higher level of PIMs; results of densitometric quantification is depicted in (F). In a separate set of experiments designed to evaluate whether the PIM phenotype of ΔRv2623 is Rv2623-specific, bacteria were grown in Middlebrook 7H11 + OADC-supplemented agar at 37°C, 5% CO₂ and harvested 12 days post-plating, and lipid extraction and 2D-TLC were carried out as previously published [77]. Complementation of (B-D) the colony morphology phenotype and (G) PIM phenotype of ΔRv2623 via the integrating vector pMV306. The PIM results are representative of 2–3 biological samples, each analyzed in three independent experiments. (H) ΔRv1747 expresses PIMs at a level lower than that of WT. Error bars: Standard deviation. *p<0.05; **p<0.01; ***p<0.001.

Fig 7. Schematic of the regulation of Rv1747 putative PIM transport by Rv2623.

In response to certain signals encountered in the host (1), the threonine residue of Rv2623 of M. tuberculosis (a universal stress protein) at position 237 (red “T” in purple sphere) is phosphorylated (2). This results in the formation of a conserved phosphothreonine (pT237)-containing motif that enables the engagement of Rv2623 with the FHA I domain of Rv1747 (3). This interaction negatively modulates the function of the putative transporter Rv1747, turning it off (4). In the absence of the signals operative in step (1), or in the presence of additional signals (5), dephosphorylation of the phosphorylated Rv2623 occurs (6), leading to disengagement of Rv2623 from Rv1747 FHA I (7). This disengagement releases the inhibitory effect of the phosphorylated Rv2623, allowing Rv1747 to transport the putative substrates PIMs (8). Whether PIMs are the substrates for Rv1747 remains to be proven. The signals that induce the phosphorylation of Rv2623 are presently unclear–potential candidates include hypoxia and nitrosative stress, as well as nutritional restriction. The nature of the kinase that phosphorylates Rv2623 in vivo is also unknown. The Rv1747 is depicted in its monomeric form except for its transmembrane domain (TMD) for clarity. Rv2623 tethered to an orange circle with a red P represents the phosphorylated form. Red “T” in purple sphere: T237. Small red spheres represent the substrates transported by the Rv1747 transporter.