The Mycobacterium bovis BCG Cyclic AMP Receptor-Like Protein Is a Functional DNA Binding Protein In Vitro and In Vivo, but Its Activity Differs from That of Its M. tuberculosis Ortholog, Rv3676

Abstract

Mycobacterium tuberculosis Rv3676 encodes a cyclic AMP (cAMP) receptor-like protein (CRP_Mt) that has been implicated in global gene regulation and may play an important role during tuberculosis infection. The CRP_Mt ortholog in Mycobacterium bovis BCG, CRP_BCG, is dysfunctional in an Escherichia coli CRP competition assay and has been proposed as a potential source of M. bovis BCG's attenuation. We compared CRP_BCG and CRP_Mt in vitro and in vivo, in M. bovis BCG and M. tuberculosis, to evaluate CRP_BCG's potential function in a mycobacterial system. Both proteins formed dimers in mycobacterial lysates, bound to the same target DNA sequences, and were similarly affected by the presence of cAMP in DNA binding assays. However, CRP_Mt and CRP_BCG differed in their relative affinities for specific DNA target sequences and in their susceptibilities to protease digestion. Surprisingly, CRP_BCG DNA binding activity was stronger than that of CRP_Mt both in vitro and in vivo, as measured by electrophoretic mobility shift and chromatin immunoprecipitation assays. Nutrient starvation-associated regulation of several CRP_Mt regulon members also differed between M. bovis BCG and M. tuberculosis. We conclude that CRP_BCG is a functional cAMP-responsive DNA binding protein with an in vivo DNA binding profile in M. bovis BCG similar to that of CRP_Mt in M. tuberculosis. However, biologically significant functional differences may exist between CRP_BCG and CRP_Mt with respect to gene regulation, and this issue warrants further study.

Tuberculosis (TB) remains a global epidemic, with one-third of the world's population infected, 9 million active cases, and over 2 million deaths annually (3). Increased drug resistance and a lethal synergism with human immunodeficiency virus exacerbate the morbidity and mortality already associated with Mycobacterium tuberculosis infection (7, 9). The molecular pathogenesis of M. tuberculosis is poorly understood, and better characterization of the mechanisms by which the tubercle bacillus responds to its host environment is needed if we are to control this deadly infection.

Cyclic AMP (cAMP), generated by adenylate cyclase, is an important signaling molecule in many bacterial species and eukaryotic cells. cAMP-mediated gene regulation in Escherichia coli, a process which requires interaction with the cAMP receptor protein (CRP), has been well defined (16). Additional roles for cAMP in metabolism have been reported for many pathogens (2, 6, 8, 13, 18, 19, 23, 30, 36), and recent studies increasingly point to an important role for cAMP signaling in microbial pathogenesis. Vfr, a CRP homolog in Pseudomonas aeruginosa, coordinates the expression of a variety of virulence-associated factors including type IV pili and the type III secretion system (35). Mutation of a cyclase-encoding gene (cya) in Vibrio vulnificus resulted in attenuation of the bacterium, with a 100-fold increase in the 50% lethal dose for mice (15). Pathogenic fungi also use the highly conserved cAMP signal transduction pathway to regulate cellular differentiation and pathogenesis (20, 21, 31).

The potential significance of cAMP signaling in M. tuberculosis pathogenesis was first explored in an early study that reported elevated levels of cAMP in the cytoplasms of macrophages infected with M. microti (24). Those authors correlated such increased cAMP levels with impaired phagosome-lysosome fusion. However, little more work was done on this topic until an in silico study identified genes for 15 adenylate cyclases and 10 cNMP-binding proteins in the M. tuberculosis genome (27). At least 10 of these adenylate cyclases are functional, as demonstrated by in vitro biochemical assays (33). It was recently shown that cAMP signaling is likely to be an important global regulator of gene regulation in M. tuberculosis and also that Rv3676 (CRP_Mt) is a CRP-like cAMP-responsive transcription factor with a putative regulon of more than 100 genes (4, 32, 34). Deletion of Rv3676 caused impaired growth in laboratory medium, in bone marrow-derived macrophages, and in a mouse model of TB (32).

CRP_Mt and its ortholog in M. bovis are identical. However, the Mycobacterium bovis BCG Pasteur ortholog, which we refer to as CRP_BCG, contains two amino acid differences, L47P and E178K, relative to the CRP_Mt sequence. It was recently reported that the L47P change resulted in loss of cAMP binding and that the E178K substitution caused a DNA binding defect in an E. coli model system (34). The authors proposed a role for CRP_BCG in M. bovis BCG's attenuation based on these results. However, potential differences between CRP_Mt and CRP_BCG function have not been examined biochemically or in TB complex mycobacteria. Such studies are a critical first step in the assignment of a role for this CRP-like protein in the attenuation of M. bovis BCG.

We compared various functional aspects of CRP_Mt and CRP_BCG in vitro and in vivo in M. tuberculosis and M. bovis BCG by use of members of CRP_Mt's putative regulon (4). We report here that CRP_BCG is a fully functional DNA binding protein with enhanced activity relative to CRP_Mt, both in vitro and in vivo. However, differences were found between CRP_Mt's and CRP_BCG's overall structural conformations, and the starvation-associated regulation of several CRP_Mt regulon members differed between M. tuberculosis and M. bovis BCG. The implications of these findings for CRP_BCG's potential as an attenuating factor in M. bovis BCG are discussed.

MATERIALS AND METHODS

Bacterial strains and culture.

M. tuberculosis H37Rv and recombinant M. bovis BCG (Pasteur strain; Trudeau Institute) were grown in Middlebrook 7H9 medium supplemented with 0.5% glycerol, 10% oleic acid-albumin-dextrose-catalase, 0.05% Tween 80 as previously described (10) or on Middlebrook 7H10 agar (Difco) supplemented with 10% oleic acid-albumin-dextrose-catalase and 0.01% cycloheximide. Fresh cultures were inoculated from frozen stocks for every experiment. Bacteria were typically used in late log phase, at 7 days of growth. Cultures were grown under shaking conditions and under ambient air. E. coli strains were grown in Luria-Bertani broth or on Luria-Bertani agar plates. Kanamycin at 25 μg/ml was added for recombinant strains. All cultures were grown at 37°C.

Nutrient starvation conditions.

For some experiments, bacteria were subjected to nutrient starvation conditions, as previously described by others (5). Briefly, M. tuberculosis or M. bovis BCG was grown in 7H9 broth with supplements for 6 days with shaking in tissue culture flasks. The culture was split into two aliquots and harvested by centrifugation. One aliquot was resuspended into phosphate-buffered saline (PBS), and the other aliquot was resuspended in the same volume of fresh 7H9 broth with supplements. After agitation for another 24 h, bacteria were harvested, and RNA was prepared for reverse transcriptase PCR (RT-PCR) to assess the expression of selected genes. Primers for these experiments are shown in Table 1.

TABLE 1.

Primers used in this study

Assay type and gene namea	Primer	Oligonucleotide sequenceb
EMSA and ChIP
Rv3857c	KM1218	F: GGATCCTAGCCAACCTGATACGTGCC
	KM1219	R: GGATCCGAGGGCACAGTTCATC
serC-Rv0885	KM1229	F: GGATCCTATGCGCGTCCTGTCCATTC
	KM1230	R: GGATCCTGGTCGGCCATGCCATCAG
frdA	KM1260	F: GGATCCGTAAACACAATCGCAGAACTGG
	KM1261	R: GGATCCATAACCACGATGTTGTGTTGGG
Rv0019c	KM1262	F: GGATCCTCGTCCGCATGCACTGAACC
	KM1263	R: GGATCCGTCAGTTGCAGTACCAACC
PE15	KM1266	F: GGATCCTATCTCGCTGCCGTGTAGC
	KM1267	R: GGATCCGGAACGACTCGCAACGTCAC
Rv0145	KM1270	F: GGATCCACCTGACCGAGACCTAAC
	KM1271	R: GGATCCTCGCGAGGATGGTAACGAGG
Rv0950c-sucC	KM1601	F: GGATCCAACTCCTTGGCTTGATACTC
	KM1602	R: GGATCCGAGGTGTGCGAATCGCTG
Rv1158c-Rv1159	KM1641	F: GGATCCTAGCGATCGCCGAGCGCACC
	KM1642	R: GGATCCGCGGACAAACGTCCAGATGG
sigA internal fragment	KM1309	F: GATGACCGAGCTTAGCGAGC
	KM1310	R: CGTAGGTGGAGAACTTGTACC
RT-PCR
Rv3676	KM859	F: ACGGCCGAGAAAACCTGTTAACC
	KM860	R: GACCCGCAATGCGACCTTCC
16S rRNA	KM886	F: GCGAACGGGTGAGTAACACG
	KM887	R: TGAAAGAGGTTTACAACCCG
Rv0145	KM1272	F: ATAGCTGCCGGATTCGATGG
	KM1273	R: GGCTCGTGGTAGGTCAACG
frdA	KM1274	F: GCGATAGCCGAAACCAATCC
	KM1275	R: GTGAAATCCCGTCTTGTCGG
Rv0019c	KM1276	F: TCATCTGGTCCGTGCTACGG
	KM1277	R: ATCCTAGATCTTCGACGTACC
PE15	KM1280	F: CTGCCATCGAAGCAGTGACC
	KM1281	R: CCACCGCTGAGATACGACGC
Rv0885	KM1282	F: CGAGTACGTGGAAATGCTGG
	KM1283	R: GAACGTGTAGTTCATCAGGC
serC	KM1286	F: GGCTATGAGGTGATACTGGG
	KM1287	R: GGCGTCGATGACGACCAAGG
sucC	KM1288	F: TTCGCCAAGCACAACGTGCC
	KM1289	R: TATCGCTAGCCTCAGCGACC
sigA	KM1309	F: GATGACCGAGCTTAGCGAGC
	KM1310	R: CGTAGGTGGAGAACTTGTACC
Rv1158c	KM1643	F: GGTATCGCTCACGCAGATCC
	KM1644	R: CGTCAGTCCTGGAAAGGTGG
Rv0950c	KM1675	F: GTGCTGACCGCTCATGCATC
	KM1676	R: GCGTTAGCCAGGTCGATACC
Rv1159	KM1677	F: CATTTTCGGTGGTCGTGTTCC
	KM1678	R: GCAGCAGGCCGAACAACTCG
Rv3857c	KM1679	F: CTCGGATTTGATACGAAACC
	KM1680	R: CAGTTGCCCGAAGACGACCC

^aThe target sequence for each gene is the intergenic region for EMSA and the ChIP assay and an internal region for RT-PCR, according to the complete sequence of M. tuberculosis H37Rv on the TubercuList World Wide Web server (http://genolist.pasteur.fr/TubercuList/).

^bOligonucleotide sequences are from 5′ to 3′. F, forward primer; R, reverse primer.

Expression and purification of CRP_Mt and CRP_BCG.

Expression and purification of CRP_Mt have been reported previously (4). Expression of CRP_BCG was by the same procedure, except that M. bovis BCG DNA was used as template for PCR amplification of the encoding DNA. The DNA was cloned into pET28a(+) vector (Novagen) between EcoRI and HindIII to generate pMBC570. The sequence was verified and maintained in E. coli BL21(DE3). The expression of CRP_Mt and CRP_BCG was induced with 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for 3 h at 37°C. The His-tagged CRP_Mt and CRP_BCG were purified using a HisTrap column (Amersham Biosciences) as previously reported (4). Protein concentration was measured with a NanoOrange protein quantitation kit (Molecular Probes) and diluted to 2 mg/ml before storage in aliquots at −70°C.

Preparation of polyclonal antibody and Western blot analysis.

Two New Zealand White rabbits were immunized subcutaneously with 1 mg purified His-CRP_Mt emulsified 1:1 with Titermax Gold (Sigma) and boosted initially and again 2 weeks later with the same amount of protein and adjuvant. The specificity of serum was confirmed by Western blotting with His-CRP_Mt and extract of M. bovis BCG as the antigen. Extracts of M. tuberculosis and M. bovis BCG were mixed with sample buffer and subjected to 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were blotted onto polyvinylidene difluoride membranes and sequentially probed with anti-CRP_Mt serum and with a peroxidase-conjugated goat anti-rabbit immunoglobulin G secondary antibody (Jackson ImmunoResearch Laboratories). Peroxidase detection was carried out with the ECL Western blotting detection reagent and analysis system (Amersham Biosciences).

Cross-linking of CRP_Mt and CRP_BCG by glutaraldehyde.

Cross-linking of purified His-tagged CRP_Mt and CRP_BCG was modified from a previously reported method (22). CRP_Mt and CRP_BCG were diluted in cross-linking buffer (50 mM sodium phosphate, pH 7.4, 20% glycerol, 5 mM MgCl₂) to 310 nM and were then incubated with glutaraldehyde at a final concentration of 7 mM for 1 h at room temperature. The reactions were quenched by the addition of SDS-PAGE sample buffer, and a portion of each was separated on a 12% SDS-PAGE gel. The proteins were transferred onto a polyvinylidene difluoride membrane and visualized with anti-CRP_Mt antibody.

Proteolysis of CRP_Mt and CRP_BCG with trypsin.

The effects of 100 μM cAMP on CRP_Mt's and CRP_BCG's conformation were compared by incubation of 5 μg of CRP_Mt and CRP_BCG with 0.2, 0.1, or 0.02 μg trypsin (Sigma) in a 20-μl reaction mixture for 10 min at 37°C, as described by others (17). One-half of the digested protein was separated on a 15% SDS-PAGE gel. AMP (100 μM) was used as a control for cAMP.

Electrophoretic mobility shift assays (EMSA).

PCR forward primer was labeled with [γ-³³P]ATP (MP Biomedicals) by use of T4 DNA polynucleotide kinase (New England Biolabs). DNA fragments were then labeled by PCR, and about 0.05 pmol DNA probe was used in each 10-μl binding reaction mixture, as described previously (4). Samples were loaded on an 8% nondenaturing polyacrylamide gel and run for 2 to 3 h at 14 V/cm. Gels were vacuum dried, exposed on a phosphor screen, scanned with a Storm 860 PhosporImager (Molecular Dynamics), and analyzed with ImageQuant software (Molecular Dynamics).

DNase I footprinting.

A 156-bp DNA fragment containing the serC-Rv0885 intergenic region was PCR amplified with primers KM1299 (GCTGGTACCGGGAATTATCC) and KM1300 (TGCCATCAGGGTAGTGAGGG). Either KM1299 or KM1300 was end labeled with T4 polynucleotide kinase (New England Biolabs) and [γ-³²P]ATP (MP Biomedicals) to generate a single end-labeled probe. Binding reactions were performed in a final volume of 100 μl containing approximately 0.25 pmol DNA probe, 100 nM of either CRP_Mt or CRP_BCG in binding buffer (10 mM Tris-HCl [pH 8.0], 50 mM KCl, 1 mM EDTA, 50 μg/ml bovine serum albumin, 1 mM dithiothreitol, 0.05% nonionic P-40 detergent, 100 μM cAMP) and 20 μg/ml poly(dI-dC) as a nonspecific competitor. Binding reactions were performed for 30 min at room temperature and were then subjected to DNase I digestion in 10 mM MgCl₂, 5 mM CaCl₂ for 1 min. Digestion was terminated by the addition of 200 μl stop solution (1% SDS, 200 mM NaCl, 20 mM EDTA, and 20 μg/ml salmon sperm DNA). Samples were extracted once with phenol-chloroform, and DNA was precipitated with ethanol, dried, and resuspended in formamide loading dye. Samples were loaded on a 7 M urea-6% polyacrylamide sequencing gel. A dideoxynucleotide sequencing ladder was generated using a Thermo Sequenase cycle sequencing kit (USB) with the same end-labeled primer as was used for the footprinting.

ChIP.

Chromatin immunoprecipitation (ChIP) experiments were carried out by a method modified from the literature (12, 29). M. tuberculosis and M. bovis BCG were grown until late log phase. Paraformaldehyde was added to a final concentration of 4% for M. tuberculosis, and formaldehyde was added to a final concentration of 1% for M. bovis BCG, for 30 min at room temperature. Cells were harvested, washed twice in PBS containing 1% Tween 80, and then resuspended in 1 ml lysis buffer containing 0.5% SDS, 50 mM Tris-HCl (pH 8.0), 10 mM EDTA (pH 8.0), 50 μl protease inhibitor cocktail (Sigma), and 10 mg lysozyme. After incubation for 20 min at 37°C, bacterial DNA was sheared by sonication as previously reported, except that the sonication time was increased to 10 min at setting 8 (28). Cell debris was removed by centrifugation, and the supernatant was diluted 1:5 in IP buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% Triton X-100) for use as the input sample. A 1-ml aliquot of the input sample was incubated overnight at 4°C on a nutator with 10 μl of antiserum or the preimmune serum from the same rabbit. Protein A-agarose beads (50 μl; Boehringer Mannheim) was added and incubated for 1 h at room temperature on a nutator, followed by four washes with IP buffer and two elutions in 150 μl of elution buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, pH 8.0, and 1% SDS). The eluate was incubated overnight at 65°C, followed by the addition of proteinase K to 100 μg/ml and incubation for 2 h at 37°C. Immunoprecipitated DNA was recovered by ethanol precipitation after phenol-chloroform extraction and used for PCR amplification of selected intergenic regions (30 cycles). An internal fragment of sigA amplified using KM1309 and KM1310 (Table 1) was used to normalize the amount of DNA between CRP-immune and preimmune IP samples.

RNA extraction and RT-PCR.

Total RNA extraction from cultures was carried out as described by others (25). After isopropanol precipitation, RNA was treated twice with RNase-free DNase I (RNase Free DNase set; QIAGEN) on an RNeasy mini kit column (QIAGEN). The RNA concentration was determined spectrophotometrically with a biophotometer (Eppendorf) at 260 nm. cDNA was prepared and amplified as previously described (11). Control reactions were performed against 16S RNA (1) by use of cDNA dilutions of 10⁻⁴ or 10⁻⁵ and against sigA (26) by use of cDNA diluted 10⁻². PCR of 16S RNA and of sigA was also performed using total RNA without reverse transcription to ensure the absence of DNA contamination. PCR was run for 30 cycles. Gels were scanned with a FluorImager595 (Molecular Dynamics) and analyzed with ImageQuant software.

Statistical analysis.

A two-tailed t test was performed using GraphPad InStat version 3.0b for Macintosh (GraphPad Software, CA) to determine the significance of binding and expression differences between genes in M. tuberculosis and M. bovis BCG. P values of <0.05 were considered significant.

RESULTS

Dimerization of CRP_Mt and CRP_BCG.

Many transcriptional regulators form oligomers when bound with their DNA ligands. An example is E. coli CRP, which binds specific DNA targets as a dimer (16). CRP_Mt recognizes a CRP-like palindromic core sequence, suggesting that it also functions as a dimer in M. tuberculosis.

His-tagged CRP_Mt and CRP_BCG were overexpressed in E. coli and then purified for in vitro analyses. Dimer formation was examined by glutaraldehyde cross-linking and Western blot analysis using anti-CRP_Mt serum (Fig. 1). Untreated proteins migrated predominantly as monomers during SDS-PAGE, but bands corresponding in mobility to dimers were present for both CRP_Mt and CRP_BCG following glutaraldehyde treatment (Fig. 1B). At high protein concentrations, CRP_BCG, but not CRP_Mt, also produced a faint higher-molecular-weight band and large aggregates that failed to enter the gel (not shown). Bands corresponding in size to both dimers and monomers were also observed for native bacterial protein extracts from both M. tuberculosis and M. bovis BCG (Fig. 1C). These results indicate that native expression and dimerization of these CRP-like proteins are similar in M. tuberculosis and M. bovis BCG.

FIG. 1.

Expression and dimerization of CRP_Mt and CRP_BCG. (A) CRP_Mt and CRP_BCG were overexpressed in E. coli and purified with a HisTrap column (Amersham Biosciences). Lanes: M, broad-range protein molecular weight markers (Promega); 1 to 3, CRP_Mt; 4 to 6, CRP_BCG. Lanes 1 and 4 are lysates of uninduced bacteria; lanes 2 and 5 are lysates of bacteria induced with IPTG; lanes 3 and 6 are purified proteins. (B) Glutaraldehyde cross-linking of purified CRP_Mt and CRP_BCG analyzed by Western blotting. Lanes: M, MagicMark XP Western protein standards (Invitrogen); 1 and 2, CRP_Mt; 3 and 4, CRP_BCG. Proteins in lanes 2 and 4 were cross-linked with glutaraldehyde. (C) Dimerization in bacterial lysates of M. tuberculosis (lane 1) and M. bovis BCG (lane 2) analyzed by Western blotting. Lane M, MagicMark XP Western protein standards. Note that the overexpressed proteins in panel B contain a His tag of about 4.4 kDa. Therefore, the antiserum-recognized bands in panel B and panel C differ in size.

Conformational differences between CRP_Mt and CRP_BCG.

We previously reported that CRP_Mt changed its conformation upon binding with cAMP (4). Here we evaluated both CRP_Mt and CRP_BCG by limited proteolysis with trypsin to compare their interactions with cAMP. CRP_BCG was more sensitive than CRP_Mt to trypsin digestion, whether or not cAMP was present (Fig. 2). CRP_Mt showed partial protection from trypsin digestion by cAMP when 5 μg of protein was treated with 0.2 μg trypsin (Fig. 2A), similar to our previous result (4). However, CRP_BCG was completely digested and could not be evaluated for cAMP binding in these conditions (Fig. 2A). CRP_BCG still showed substantial cleavage when the quantity of trypsin was reduced to 0.02 μg, but CRP_Mt exhibited no obvious proteolysis under these conditions (Fig. 2B). CRP_BCG showed a banding pattern in the presence of cAMP only slightly different from what was seen with trypsin only or AMP controls (Fig. 2B), which differed from the results obtained with CRP_Mt (Fig. 2A and C). These results indicate that CRP_Mt and CRP_BCG differ in their overall sensitivities to trypsin and possibly in their interactions with cAMP.

FIG. 2.

Evidence of allosteric alteration of CRP_Mt and CRP_BCG by cAMP. CRP_Mt and CRP_BCG were digested with 0.2 μg (A), 0.02 μg (B), or 0.1 μg (C) trypsin. Lanes: M, broad-range molecular markers (Promega); 1 to 4, CRP_Mt; 5 to 8, CRP_BCG. Lanes 4 and 8 are undigested controls. Lanes 1 to 3 and 5 to 7 were digested with trypsin; 100 μM cAMP was supplemented in lanes 2 and 6; 100 μM AMP was supplemented in lanes 3 and 7. Figures are representative of at least two independent experiments.

CRP_Mt and CRP_BCG have similar DNA binding specificities.

DNA binding of CRP_Mt and CRP_BCG was tested with the serC-Rv0885 intergenic DNA used as a model sequence. Our previous results indicated that CRP_Mt binds to this sequence specifically, and with enhancement in the presence of 100 μM of cAMP (4). In the present study, CRP_BCG showed higher DNA binding affinity than did CRP_Mt in either the absence (Fig. 3A) or presence (Fig. 3B) of 100 μM cAMP. Addition of cAMP enhanced the DNA binding of CRP_BCG to the same extent as seen for CRP_Mt (1.5- to 2-fold) (Fig. 3C). However, CRP_BCG was also able to form a higher-molecular-mass protein-DNA complex that was most prominent in the presence of cAMP (Fig. 3B and C). These results indicate that CRP_BCG is fully competent for DNA binding, although its binding properties are not identical to those of CRP_Mt.

FIG. 3.

Comparison of DNA binding between CRP_Mt and CRP_BCG, and the effects of cAMP on CRP_BCG binding. (A) Binding of CRP_Mt and CRP_BCG with the serC-Rv0885 intergenic region in the absence of cAMP. Lanes: 1, free probe; 2 to 6, CRP_Mt; 7 to 11, CRP_BCG. Lanes 2 to 6 and 7 to 11 are in the presence of 0.35 nM, 0.7 nM, 1.7 nM, 3.5 nM, and 7 nM purified protein, respectively. (B) Comparison of CRP_Mt's and CRP_BCG's binding with the serC-Rv0885 intergenic region in the presence of 100 μM cAMP. Probe and protein dilutions are the same as the lanes in panel A. (C) CRP_BCG's binding with the serC-Rv0885 intergenic region in the absence (lanes 1 to 6) or in the presence (lanes 7 to 12) of cAMP. Lanes 1 and 7 are free probes. Lanes 2 to 6 and 8 to 12 are in the presence of 0.35 nM, 0.7 nM, 1.7 nM, 3.5 nM and 7 nM purified protein, respectively. DNA binding data are plotted to the right of each panel. For each panel, quantitation of band intensity was done using ImageQuant software, and the percent DNA bound was calculated as the intensity of each shifted band divided by the intensity of the unshifted band in the probe-only sample (lane 1). The apparent dissociation constant (k_d) (37) in nM is also indicated.

CRP_Mt's and CRP_BCG's DNA binding specificities were further explored by DNase I footprinting of the serC-Rv0885 intergenic region. The same sequence was protected from digestion with DNase I, in the presence of either CRP_Mt or CRP_BCG (Fig. 4). This protected region is consistent with the binding motif that we identified previously (4), but it includes asymmetric flanking sequences as well (Fig. 4).

FIG. 4.

DNase I footprinting of CRP_Mt's and CRP_BCG's binding sites in the serC-Rv0885 intergenic region. The −155/+1 DNA region upstream of the serC translational start site was labeled at the −155 end (A) or the +1 end (B) and then digested with DNase I in the absence of protein (−) or in the presence of 100 nM of either CRP_Mt (H37Rv) or CRP_BCG (BCG). A dideoxynucleotide sequencing ladder, run in parallel with the footprinting reactions (data not shown), was used to determine the sequence of the protected region. The black bars indicate the previously reported CRP_TB binding motif (TGTGAN₆TCACA).

Together, EMSA, DNase I footprinting, and limited proteolysis results indicate that CRP_BCG retains the ability to interact with both DNA and cAMP at a functional level, despite its conformational differences with CRP_Mt.

CRP_Mt and CRP_BCG bind similar target sequences with differing affinities.

CRP_Mt's and CRP_BCG's DNA binding interactions were further examined, with a broader range of DNA sequences, in vitro and in vivo. In vitro binding was measured by EMSA, while in vivo binding was assessed by ChIP assays that used the same intergenic DNA sequences. Intergenic regions of serC-Rv0885, Rv0950c-sucC (Rv0951), and Rv1158c-Rv1159, as well as regions upstream of PE15 (Rv1386), frdA (Rv1552), Rv3857c, and Rv0019c, were selected from the CRP_Mt's list of putative binding sequences (4). Another sequence upstream of Rv0145 which binds with CRP_Mt but was not included in that previous study was also tested. All of these intergenic DNA sequences are identical in M. bovis BCG and M. tuberculosis H37Rv, except that Rv3857c has one base difference 164.5 bp downstream from the binding motif axis, and frdA has two nucleotide substitutions located 136.5 and 175.5 bp upstream of the binding motif axis. The H37Rv sequences were used for all experiments.

All putative binding sequences tested using EMSA, except Rv0019c, showed clear binding with both CRP_Mt and CRP_BCG (Fig. 5). The Rv0019c intergenic region bound CRP_BCG with very low affinity and failed to bind at all with CRP_Mt at molarities up to 35 nM. Neither CRP_Mt nor CRP_BCG bound to a mycobacterial sigA internal fragment, which was used as a negative control. The affinity of CRP_BCG for DNA was slightly higher than that of CRP_Mt, and in some cases, (e.g., serC-Rv0885, frdA, Rv0145, Rv1158c-Rv1159), the mobility of CRP_BCG-DNA complex bands was supershifted in the presence of 3.5 nM or more protein (Fig. 5). This phenomenon was not observed with CRP_Mt (Fig. 5), even at protein concentrations up to 3.5 μM (data not shown).

FIG. 5.

In vitro and in vivo binding of CRP_Mt and CRP_BCG with selected intergenic regions. Lanes 1 to 3, EMSA of CRP_Mt (H37Rv) and CRP_BCG (BCG) with selected intergenic probes amplified from M. tuberculosis H37Rv. No cAMP was supplemented in the binding buffer. Lane 1 is free probe. Protein-DNA interaction is shown with 3.5 nM (lane 2) and 35 nM (lane 3) protein. The intergenic regions of genes tested are labeled to the left. serC, sucC, and Rv1158c are divergent from Rv0885, Rv0951, and Rv1159, respectively. Note that the size of the protein-DNA complex varies according to the probe size of different genes. Lanes 4 to 6, ChIP in M. tuberculosis and M. bovis BCG. Lane 4 shows PCR from immunoprecipitation with anti-CRP_Mt serum, and lane 5 is mock precipitation with preimmune serum prepared from the same rabbit. Lane 6 shows PCR control results from input DNA. The intergenic regions of genes tested (except the sigA control, which is in the open reading frame [ORF]) are listed to the left. Note that the results are pooled from different gels, so that bands of different genes are not quantitatively comparable. The ChIP results represent three biological repeats.

Bacterial protein-DNA complexes were cross-linked in culture suspension and analyzed by ChIP to evaluate DNA binding in vivo. A parallel sample using preimmune serum from the same rabbit was run as a background reference control, and an internal sigA DNA sequence was used for PCR normalization. All DNA sequences that showed in vitro binding by EMSA were also enriched by ChIP in both M. tuberculosis and M. bovis BCG lysates (Fig. 5). Rv0019c, which was negative by EMSA, was not enriched by ChIP. Two sequences, frdA and Rv0950c-sucC, reproducibly showed stronger binding with CRP_BCG than with CRP_Mt (Fig. 5; Table 2). These results indicate that although CRP_Mt and CRP_BCG bind similar regulatory sequences in vivo, they differ in their relative affinities for certain DNA sequences.

TABLE 2.

Quantitation of ChIP assay results^a

DNA amplifiedb	Immune/preimmune ratio (mean ± SEM) forc:		P value (H37Rv vs BCG)d
	H37Rv	BCG
serC-Rv0885	2.6 ± 0.6	2.8 ± 0.2	0.76
frdA	1.6 ± 0.1	4.5 ± 0.1	0.001
Rv0950c-sucC	1.7 ± 0.1	4.1 ± 0.1	0.0009
Rv0145	2.3 ± 0.1	2.0 ± 0.7	0.65
PE15	2.3 ± 0.4	2.1 ± 0.2	0.65
Rv1158c-Rv1159	1.5 ± 0.2	1.9 ± 0.2	0.30
Rv3857c	1.7 ± 0.1	2.1 ± 0.4	0.38
Rv0019c	0.9 ± 0.1	1.1 ± 0	-

^aPCR bands were scanned and analyzed with ImageQuant software.

^bDNA amplified from the intergenic regions of divergent genes or upstream of others. sigA was amplified from within its ORF.

^cRatio of bands amplified from immunoprecipitated immune serum to bands amplified from the preimmune serum control. The results are presented as means ± standard errors of the means of three biological repeats.

^dP values from H37Rv versus BCG comparisons were compared using a two-tailed t test. Comparisons with P values of <0.05 are shown in bold. It was not possible to calculate a P value for Rv0019c (indicated by the dash), due to the 0 value for the standard error of the means.

Gene regulation of CRP_Mt regulon members under nutrient starvation conditions.

The differing affinities of CRP_Mt and CRP_BCG for some regulatory sequences suggested that members of the CRP regulon might be regulated to different extents in M. tuberculosis and M. bovis BCG. We previously observed that a high percentage of the putative CRP_Mt regulon candidates were associated with starvation in another published study (4, 5). Both anaerobic and starved cultures have been used as models to study the molecular basis of dormancy (14), and the CRP_Mt regulon has been implicated in the persistence pathway (32).

We compared expression levels of putative regulon gene members by semiquantitative RT-PCR in M. tuberculosis and M. bovis BCG subjected to starvation versus nutrient-rich conditions (Fig. 6; Table 3). Rv0885, frdA, Rv0950c, Rv0145, PE15, and Rv1158c were up-regulated at least 1.5-fold in response to starvation in at least one of the bacteria. frdA was similarly up-regulated, while Rv0950c was down-regulated, in both species. Three genes, Rv0145, PE15, and Rv1158c, showed statistically significant differences in their regulatory patterns between M. tuberculosis and M. bovis BCG. Rv0885 was generally more strongly regulated in M. tuberculosis than M. bovis BCG, but the difference was not statistically significant. These results indicate that the gene regulatory responses of M. tuberculosis and M. bovis BCG to starvation are not identical. Further studies are needed to determine the extent of these regulatory differences and whether they are mediated by the CRP-like proteins in TB complex bacteria.

FIG. 6.

Transcription of selected genes after starvation of M. tuberculosis and M. bovis BCG for 24 h in PBS. RT-PCR with RNA prepared from M. tuberculosis H37Rv and M. bovis BCG after 24 h of starvation in PBS. Lanes: 1, 7H9 broth control; 2, starvation samples. The genes tested are listed to the left. sigA is a control for normalizing the cDNA level. Note that the results are pooled from PCRs with different concentrations of cDNA. Therefore, the bands are not comparable between different genes. The figure is a representative of three independent experimental repeats.

TABLE 3.

Quantitation of RT-PCR results^a

ORF	Gene name	Fold (mean ± SEM) induction by starvationb		P value (H37Rv vs BCG)c
		H37Rv	BCG
Rv3676		1.03 ± 0.07	0.93 ± 0.03	0.25
Rv0884c	serC	1.00 ± 0.20	0.67 ± 0.07	0.19
Rv0885		2.30 ± 0.60	1.27 ± 0.20	0.18
Rv1552	frdA	2.70 ± 0.42	2.30 ± 0.25	0.46
Rv0950c		0.40 ± 0.10	0.30 ± 0.06	0.44
Rv0951	sucC	0.70 ± 0.10	0.97 ± 0.03	0.06
Rv0145		0.53 ± 0.12	0.90 ± 0.06	0.05
Rv1386	PE15	0.53 ± 0.12	0.97 ± 0.03	0.03
Rv1158c		1.30 ± 0.20	0.53 ± 0.07	0.02
Rv1159		0.83 ± 0.09	1.13 ± 0.20	0.24
Rv3857c		0.97 ± 0.32	1.17 ± 0.07	0.57
Rv0019c		0.97 ± 0.17	1.07 ± 0.20	0.72

^aGels were scanned and analyzed with ImageQuant software.

^bResults were expressed as means ± standard errors of the means of three biological repeats and normalized by the induction (n-fold) of sigA.

^cP values for H37Rv versus BCG comparisons were compared using a two-tailed t test. P values relating to significant species-related differences in response to starvation are shown in bold.

DISCUSSION

Allosteric and functional comparison of CRP_Mt and CRP_BCG.

CRP_Mt is an established CRP-like gene regulator in M. tuberculosis (4, 32). However, a recent report indicated that two amino acid changes alter CRP_BCG's function relative to CRP_Mt, making CRP_BCG unable to compete with CRP for DNA and/or cAMP binding in an E. coli functional model (34). This finding raised the possibility that this cAMP-responsive transcription regulator is inactive or defective in M. bovis BCG, possibly contributing to BCG's attenuation.

We found that CRP_BCG is a fully functional DNA binding protein in M. bovis BCG, but its pattern of DNA binding is not identical to that of CRP_Mt in M. tuberculosis. Both CRP_Mt and CRP_BCG were detected as dimers in vitro and in vivo, but the trypsin digestion experiments indicated that the amino acid differences in CRP_BCG affected its conformation. The L47P change in CRP_BCG deletes one trypsin recognition site, but the E178K change generates another, as determined by PeptideCutter (http://au.expasy.org/tools/peptidecutter/). Therefore, the total number of trypsin recognition sites is the same in each protein, although their positions vary.

It is not clear whether CRP_BCG also differs significantly from CRP_Mt in its functional interactions with cAMP (Fig. 3). The presence of cAMP had very little effect on CRP_BCG's sensitivity to trypsin digestion compared with the protection it afforded to CRP_Mt (Fig. 2). In fact, CRP_BCG was slightly more sensitive to trypsin digestion in the presence of cAMP, which is more similar to what is observed with CRP in E. coli than with CRP_Mt (17). However, the cAMP-associated DNA binding effects observed by EMSA suggest that CRP_BCG retains a functional association with cAMP that is not readily detected in the trypsin experiments, due to CRP_BCG's increased lability in the presence of protease (Fig. 3). Additional functional studies will be needed to more completely resolve this issue.

In contrast to our expectations based on a previous report (34), we found that CRP_BCG shows higher affinity than CRP_Mt for most target DNA sequences in vitro (Fig. 5), indicating that the conformational changes associated with the L47P and/or E178K substitutions are actually beneficial for binding with DNA. This effect may have biological significance, because in vivo binding by CRP_BCG was also significantly enhanced relative to CRP_Mt in two cases (Fig. 5).

Comparison of in vitro and in vivo DNA binding of CRP_Mt and CRP_BCG.

Our in vitro (EMSA) and in vivo (ChIP) DNA binding results showed good agreement across all eight sequences tested, indicating that EMSA provides a useful predictor for in vivo binding of CRP_Mt regulon promoters. Seven intergenic sequences from the putative CRP_Mt regulon showed clear binding by both approaches, thereby supporting our previous in silico prediction of the regulon (4). Nonetheless, these regulon predictions need further optimization. Rv0145 binds CRP_Mt and CRP_BCG in vitro and in vivo, making it a likely regulon member, despite its not being detected in the in silico analysis (4). Rv0145 also regulates in response to starvation in M. tuberculosis (but not M. bovis BCG) but was not previously identified as a member of the starvation regulon (5). In contrast, the regulatory sequence for another putative regulon member predicted in silico, Rv0019c, was not detected in vivo, and it showed only very weak in vitro binding with CRP_BCG and no binding with CRP_Mt (Fig. 5). Rv0019c is also the only previously identified starvation regulon member that failed to be regulated in response to starvation in the present study (5).

Rv0019c and Rv0145 may aid us in better defining the criteria that are most important for binding and in optimizing the CRP_Mt binding motif. Specifically, alignment of the CRP_Mt motifs for all DNA sequences used in this study shows that only Rv0019c and Rv0145 have mismatches in both 5-bp halves of the predicted core sequence. Rv0019c has three mismatches, compared with two for Rv0145. None of the remaining sequences had more than one mismatched nucleotide (Fig. 7). Future studies will determine whether it was the total number of mismatches, or the specific nucleotide changes, that caused Rv0019c's lack of activity in this assay.

FIG. 7.

Alignment of CRP_Mt and CRP_BCG binding regions of selected intergenic sequences. Forty-six nucleotides centered on the motif core axis from each intergenic sequence were aligned. Blocks of similar nucleotides are indicated with yellow shading. Identical nucleotides are shown in red, and conserved nucleotides are shown in green. The binding arms in the consensus sequence are indicated with black bars. The protection region from DNase I footprinting in the serC-Rv0885 intergenic sequence is underlined. The sequences for which the experimental results are inconsistent with the previous prediction (4) are indicated with asterisks.

EMSA was a reliable predictor of in vivo DNA binding activity, but the relative affinities of CRP_Mt and CRP_BCG for various DNA sequences in vitro and in vivo did not always coincide (Fig. 5; Table 2). Only two DNA sequences, frdA and Rv0950c-sucC, were bound with more affinity by CRP_BCG than CRP_Mt in vivo, whereas CRP_BCG's DNA binding activity was higher for all sequences in vitro (Fig. 5). This is probably because many different factors that can affect binding are included in the complex biological environment but are not present in the defined and “minimalist” in vitro binding conditions used for the EMSA experiments. In this respect, the ChIP approach is likely to be more reliable than EMSA for interpreting differences between CRP_Mt's and CRP_BCG's interactions with different regulatory sequences. Nonetheless, the differing in vivo DNA binding affinities measured for these DNA sequences did not correlate with regulatory differences found in the starvation assay, even though both frdA and Rv0950c were regulated in response to starvation. It is possible that factors other than CRP_Mt/CRP_BCG mediate the starvation response for these genes or that DNA binding affinity is not the primary determinant of regulatory function.

Responses of M. tuberculosis and M. bovis BCG to nutrient starvation.

Twenty-four members of the putative CRP_Mt regulon (∼21%) (4) were previously identified as being starvation regulated in M. tuberculosis (5). Five of these genes (Rv0885, frdA, Rv0019c, PE15, and Rv1158c) were examined in the present study, and results for all except Rv0019c were consistent with those of a previous microarray analysis (5). In addition, we observed down-regulation of Rv0950c and Rv0145 expression in response to starvation, a finding which has not been reported previously. The reason for these differences in regulation is not clear, but they likely reflect subtle differences in experimental conditions or assay sensitivities between laboratories.

PE15, Rv1158c, and Rv0145 showed differing expression patterns between M. bovis BCG and M. tuberculosis (Table 3; Fig. 6). PE15 and Rv0145 expression was not regulated in M. bovis BCG, while Rv1158c expression was regulated only in M. bovis BCG, indicating differences in the responses of these bacteria to starvation. However, all three genes code for hypothetical proteins of unknown function, making it difficult to assess the possible importance of their dysregulation to M. bovis BCG's attenuation.

In summary, M. tuberculosis and M. bovis BCG contain functional CRP orthologs that behave similarly, but not identically, with respect to DNA binding in vitro and in vivo. Conformational differences may be responsible for these differences in DNA affinity and could also affect the interactions of these CRP-like proteins with RNA polymerase and other gene regulatory factors. Starvation-associated gene regulatory differences between M. bovis BCG and M. tuberculosis may have implications for pathogenesis and warrant further investigation. Further studies are also needed to assess the roles of these CRP orthologs in the gene regulatory responses of M. bovis BCG and M. tuberculosis to starvation and other host-associated environmental conditions, as well as the environmental signals to which these CRP-like global regulators respond.