|Home | About | Journals | Submit | Contact Us | Français|
Mycobacterium tuberculosis CRPMt, encoded by Rv3676 (crp), is a CRP-like transcription factor that binds with the serC – Rv0885 intergenic region. In the present study, we evaluated CRPMt’s regulation of serC and Rv0885 in M. tuberculosis and M. bovis BCG, using site-specific mutagenesis, promoter fusions and RT-PCR. The CRPMt binding site was required for full expression of serC and Rv0885, and expression of both genes was reduced in M. tuberculosis and M. bovis BCG crp mutants. These data show that CRPMt binding directly activates both serC and Rv0885 expression. M. tuberculosis serC restored the ability of an Escherichia coli serC mutant to grow in serine-dropout medium, demonstrating that M. tuberculosis serC encodes a phosphoserine aminotransferase. Serine supplementation, or overexpression of serC, accelerated the growth of M. tuberculosis and M. bovis BCG crp mutants in mycomedium, but not within macrophages. These results establish a role for CRPMt in the regulation of amino acid biosynthesis, and show that reduced serine production contributes to the slow-growth phenotype of M. tuberculosis and M. bovis BCG crp mutants in vitro. Restoration of serine biosynthesis by serC expression will facilitate identification of additional CRPMt-regulated factors required by M. tuberculosis during macrophage and host infection.
Tuberculosis (TB) remains a global epidemic, with nine million new cases and two million deaths annually, and one-third of the world population latently infected. Drug resistance and a lethal synergy between the causative agent Mycobacterium tuberculosis and human immunodeficiency virus (HIV) are major causes of this high morbidity and mortality (http://www.who.int/tb/publications/global_report/2009/update/en/index.html). Gene regulation is critical for M. tuberculosis’ successful response to the diverse environments encountered within the host during infection. However, little is known about gene regulatory mechanisms in M. tuberculosis.
Adenosine 3′, 5′-cyclic monophosphate (cAMP) is an important signaling molecule in diverse organisms, including prokaryotic and eukaryotic microbes as well as their mammalian hosts. The classical prokaryotic cAMP signaling pathway established in Escherichia coli for catabolite repression requires interaction of cAMP with the cAMP receptor protein (CRP) (Kolb et al., 1993). However, cAMP’s broader roles in regulating metabolism and pathogenesis of many pathogens are being increasingly recognized (Alspaugh et al., 2002, Caler et al., 2000, D’Souza & Heitman, 2001, Gross et al., 2003, Petersen & Young, 2002, Wolfgang et al., 2003). CRP homologs in Pseudomonas aeruginosa, Yersinia pestis and Salmonella enterica serovar Typhimurium coordinate expression of a range of virulence factors, including type IV pili and the type III secretion system (TTSS) (Whitchurch et al., 2005, Miao et al., 1999, Petersen & Young, 2002). Mutation of the adenylyl cyclase-encoding gene (cya) in Vibrio vulnificus caused attenuation in a murine infection model (Kim et al., 2005), and a cAMP- dependent Cl− secretory pathway is associated with the diarrhea induced by Aeromonas sobria (Tanoue et al., 2005). Pathogenic fungi use the highly conserved cAMP signal transduction pathway to regulate cellular differentiation as well as their virulence potential (Pukkila-Worley & Alspaugh, 2004), and the cAMP-PKA signal transduction pathway is required for Aspergillus fumigatus pathogenicity (Liebmann et al., 2003, Liebmann et al., 2004).
Recent work shows that M. tuberculosis increases cAMP production upon phagocytosis by macrophages (Agarwal et al., 2009, Bai et al., 2009), suggesting that cAMP signaling is important for M. tuberculosis pathogenesis. This finding is consistent with results of an early study showing that infection of macrophages with M. microti elevated cAMP levels and impaired phagosome-lysosome fusion in the infected cells (Lowrie et al., 1975). Genes have been identified for at least 15 adenylyl cyclases in M. tuberculosis, and activity has been confirmed for the products of at least 10 of these genes (Castro et al., 2005, Guo et al., 2001, Linder et al., 2002, Reddy et al., 2001, Shenoy & Visweswariah, 2006a, Sinha et al., 2005, Tews et al., 2005, Abdel Motaal et al., 2006). The further presence of 10 genes encoding putative cNMP-binding proteins suggests a wide range of functional roles for cAMP in M. tuberculosis biology (McCue et al., 2000, Shenoy & Visweswariah, 2006b).
CRPMt (Rv3676) is one of two cAMP-associated transcription factors identified in M. tuberculosis (McCue et al., 2000, Gazdik et al., 2009, Gazdik & McDonough, 2005). CRPMt is a CRP-like protein that binds a DNA motif resembling E. coli CRP’s DNA binding site (Bai et al., 2005, Rickman et al., 2005, Spreadbury et al., 2005). Two amino acid differences in M. bovis BCG’s CRPMt ortholog (CRPBCG) increase its DNA binding affinity and have some effects on gene expression, but no significant functional or virulence-related differences between the two proteins have been found (Bai et al., 2007, Hunt et al., 2008, Stapleton et al., 2010).
We previously identified a putative regulon for CRPMt comprising 114 genes (Bai et al., 2005), and additional regulon members were proposed in a recent study (Krawczyk et al., 2009). However, CRPMt’s direct role in regulation has not been established for most of the above genes. The putative CRPMt regulon includes serC, which has been identified as an essential gene in M. tuberculosis (Sassetti et al., 2003). In E. coli, serC encodes a phosphoserine aminotransferase that is essential for growth in the absence of serine supplementation (Clarke et al., 1973). An M. tuberculosis crp deletion strain showed impaired growth in bone marrow-derived macrophages and in a mouse model of TB (Rickman et al., 2005). Growth was also reduced in vitro, but the bases of these growth defects are not known. We reasoned that decreased expression of serC could contribute to the reduced growth rate of the M. tuberculosis crp mutant.
In this study, we used serC and its divergently transcribed gene, Rv0885, to study CRPMt’s role in regulation of gene expression. We found that expression of both genes is directly activated by CRPMt’s binding to its predicted binding site. We also identified the gene product of serC in M. tuberculosis (and in M. bovis BCG) as a phosphoserine aminotransferase involved in the synthesis of serine. In addition, we showed that serine supplementation increases the crp mutants’ growth in vitro, but not within macrophages. These studies establish a role for CRPMt in the regulation of amino acid biosynthesis in M. tuberculosis and provide mechanistic insights into that regulation.
The upstream intergenic region and open reading frame (orf) sequences of serC and Rv0885 are identical in M. bovis BCG and M. tuberculosis, with the exception of a single base change that results in a T234I difference in the M. bovis BCG Rv0885 ortholog, relative to that of M. tuberculosis. The transcription start sites (TSSs) of serC and Rv0885 were determined by primer extension, using total RNA isolated from exponentially growing M. bovis BCG. The TSS of Rv0885 mapped to a G, which is 84 nt from the translation start codon (Fig. 1A). A potential Pribnow (−10) box (AAGGGT) with 3-of-6 bp match to the consensus sequence TAYGAT (where Y is either a C or a T) (Mulder et al., 1997) is centered at a suboptimal 7 nt upstream of the Rv0885’s TSS, while the CRP-binding motif is centered at −55.5 nt from this TSS.
Several possible TSSs for serC were detected with oligonucleotides KM1608 and KM1609, as we have observed for some other mycobacterial genes (Vasudeva-Rao & McDonough, 2008). Use of KM1609 mapped one potential TSS to a G located 103 nt upstream from the start codon (Fig. 1B). The CRP-binding motif is centered at +21.5 nt from this serC TSS, while a potential −10 region (TATGAC) with a 5-of-6-bp match with the consensus sequence (TAYGAT) is centered at −13 nt. A second possible TSS (Fig. 1C) was also mapped using oligonucleotide KM1608, which is 70 nt from the start codon and 13.5 nt downstream of the CRP binding motif axis. An additional faint band of greater size was also observed with each primer. The TSSs associated with these faint bands are separated by 6 nt and flank the divergent Rv0885 TSS. RT-PCR was performed using a series of primers (Table 2) stepping progressively upstream through this region to confirm the presence of the longer extension products. Product was obtained with all primers except one that binds a region beyond the upstreammost site identified by primer extension (Fig. S1). While these results validate the upstream TSSs, differences in the efficiencies of the different primers prevent us from drawing conclusions regarding the actual relative use of each possible TSS.
We previously identified a CRPMt binding motif in the serC-Rv0885 intergenic region (Bai et al., 2007, Bai et al., 2005) (Fig. 1D). In the present study, we constructed promoter:lacZ reporter fusions for serC and Rv0885 using their intergenic DNA region to evaluate the role of CRPMt in regulation of these two genes. We measured the activities of both promoters in M. bovis BCG, given that the serC-Rv0885 intergenic DNA binds CRPMt and CRPBCG similarly (Bai et al., 2007).
Truncated reporter fusions of serC and Rv0885 promoters that excluded DNA sequences upstream of the CRPMt-binding motif (‘long’ fragments) had 50% (serC) to 33% (Rv0885) of the activity of their respective full length promoters in M. bovis BCG (Fig. 2). The decreased promoter activities indicate that DNA sequences upstream of the CRPMt binding site contribute to expression of both genes. However, truncation beyond the CRPMt-binding motif (‘short’ fragments) completely abolished the expression of both promoters (Fig. 2), demonstrating the importance of the CRPMt-binding site for the expression from either promoter.
We replaced 16 bp of the motif core (TGTGANNNNNNTCACA) with a 16 bp random sequence (AGGCTTTCGTTCCGCT) to more specifically test the role of the CRPMt binding site. Constructs in which the same DNA sequence was used to replace regions 18 bp upstream or downstream of the motif were included as controls (Fig. 3A). The activities of the mutant promoters were compared with WT, in M. bovis BCG. CRPMt motif replacement alone reduced the expression of each promoter ~5-fold (Fig. 3B), consistent with the promoter truncation results. As expected, substitution of DNA sequences in the ‘Ctl 2’ region (Fig. 1D, ,3A)3A) had no effect on either promoter’s activity (Fig. 3B). However, the ‘Ctl 1’ DNA replacement (Fig. 1D, ,3A)3A) increased expression ~5-fold for serC, and ~3-fold for Rv0885, compared with the WT promoters.
The increased expression associated with the ‘Ctl 1’ replacement suggested that the modified DNA sequence either enhanced the binding with CRPMt, or reduced binding of an unknown repressor for both serC and Rv0885. To determine whether the CRPMt binding was altered by this sequence substitution, we labeled the WT and mutant promoter regions as probes, and used EMSA to assess CRPMt’s binding with each DNA sequence. Replacement of the binding site totally abolished the binding with CRPMt (Fig. 3C), while both substitution control fragments retained their binding with CRPMt at levels similar to those of the WT promoter sequences (Fig. 3C). The same results were observed using CRPBCG (data not shown). We conclude that the ‘Ctl 1’ region contains a binding site for an unidentified repressor (Fig. 3A).
The role of the CRPMt binding site in serC and Rv0885 regulation was further defined with point mutations. We previously showed that nt G4 and C17 in the CRPMt motif model NNTGTGANNNNNNTCACANN are highly conserved, and are crucial for CRPMt-binding to the serC-Rv0885 intergenic region (Bai et al., 2005). CRPBCG produced the same result (Fig. 4A), so we generated serC and Rv0885 promoter reporters with G4 and C17 point mutations (Fig. 4B), and measured their expression levels in M. bovis BCG. Expression of these mutated promoters was reduced ~2-fold for serC and ~3-fold for Rv0885, compared with the WT levels (Fig. 4C). These results indicate that the altered nucleotides are required for full activity of the promoters, further suggesting that CRPMt positively regulates expression of both serC and Rv0885.
M. tuberculosis crp and M. bovis BCG mutants were generated by homologous recombination (Fig. 5A) to establish the role of CRPMt (or CRPBCG) itself in expression of serC and Rv0885. The replacement of crp with a hygromycin resistance marker in each bacterium was confirmed by PCR (Fig. S2A, B). The mutation was complemented with a single copy of crp expressed from its own promoter and integrated into the chromosome at the L5 attachment site (Lee et al., 1991). A similar M. tuberculosis crp mutant was also generously provided by Dr. R. Buxton, and used in these studies until our new mutant was available. The growth of both M. tuberculosis crp and M. bovis BCG crp mutants was significantly slower than that of their respective WT or complemented strains grown in mycomedium (data not shown), consistent with the phenotype of the M. tuberculosis crp mutant described previously (Rickman et al., 2005).
The serC and Rv0885 promoter fusion plasmids were transformed into the M. bovis BCG crp mutant, and their expression was compared in the mutant and WT strains. Deletion of crp significantly reduced the expression levels of both serC and Rv0885 (Fig. 5B), consistent with the promoter mutation data. CRPMt’s role in serC and Rv0885 expression in M. tuberculosis was further confirmed by semi-quantitative RT-PCR using RNA from WT M. tuberculosis H37Rv, the crp mutant and its complemented strain. Several additional CRPMt regulon members were included for comparison. Deletion of crp reduced the expression of both serC and Rv0885 (Fig. 5C). In contrast, expression of another putative CRPMt regulon member, lipQ (Rv2485c), was enhanced in the crp mutant. This lipQ result is consistent with the finding of a previous report (Rickman et al., 2005). As expected, expression of Rv0019c, which does not bind with CRPMt (Bai et al., 2007), was unaffected by crp deletion and served as a negative control. Taken together, these data show that CRPMt directly regulates expression of both serC and Rv0885, and that full activation of these promoters is mediated through binding of the previously identified CRPMt binding site.
M. tuberculosis Rv0885 is a conserved hypothetical protein of unknown function, while the predicted product of the divergently expressed serC gene is annotated as a phosphoserine aminotransferase (http://genolist.pasteur.fr/Tuberculist). However, M. tuberculosis SerC has only 18% identity with E. coli SerC at the amino acid level, and the function of M. tuberculosis SerC has not been confirmed. We cloned the serC orfs of M. tuberculosis and E. coli separately into pMBC664, and expressed each in E. coli under control of the E. coli crp promoter (Fig. 6A, B). Either construct was able to complement a serC mutation in E. coli (KL285) for growth on minimal medium lacking serine (Fig. 6C), while the KL285(pMBC664) vector-only control strain was unable to grow on serine-free plates (Fig. 6C). This result indicates that the biological function of M. tuberculosis serC’s product is similar to that of E. coli SerC, so we conclude that M. tuberculosis SerC is also a phosphoserine aminotransferase.
In E. coli, SerC is an essential enzyme for the serine and pyridoxine biosynthesis pathways (Fig. 7A) (Man et al., 1997). Our data showed that serC expression was significantly decreased in M. tuberculosis and M. bovis BCG crp mutants, so we reasoned that reduced production of serine and/or pyridoxine could be a factor in the slow-growth phenotype of the crp mutants. We supplemented mycomedium with 5 mM serine or 5 mM pyridoxine to test the effects of these components on the growth of the M. bovis BCG and M. tuberculosis crp mutants. Addition of serine accelerated the growth of the mutants ~2 fold during log phase, but addition of pyridoxine was toxic to the mutants. Neither serine nor pyridoxine affected the growth of WT or the complemented mutant strains (Fig. 7B, 7C, S3). These results suggested that the slow-growth of the crp mutants was related to the serine biosynthesis pathway.
E. coli SerC catalyzes the interconversion of 3-phospho-hydroxypyruvate and glutamate to 3-phosphoserine and α-ketoglutarate (Man et al., 1997)(Fig. 7A). The 3-phosphoserine is then converted into serine (Man et al., 1997)(Fig. 7A). We tested the effects of these compounds on both crp mutants, and found that their growth was significantly increased at mid- to late-log phase by addition of 5 mM phosphoserine, but was not by 5 mM glutamate or α-ketoglutarate (Fig. 7B, C, S3). Because serine is a substrate for biosynthesis of tryptophan, and interconverts with glycine and cysteine (Fig. 7A), we also tested the effects of each of these amino acids on the mutants’ growth. Addition of 5 mM glycine or cysteine restored the mutants’ growth to a level similar to that with serine present. In contrast, addition of 5 mM tryptophan did not improve growth (Fig. 7B, C, S3).
The serine supplementation data suggested that dysregulation of serC by crp deletion contributed to the mutants’ slow growth in vitro. We further tested this possibility using plasmid pMBC1091, which expresses M. tuberculosis serC under control of the constitutive M. tuberculosis tuf promoter, such that its expression is independent of CRPMt. Colonies of both the M. bovis BCG and M. tuberculosis crp mutants harboring pMBC1091 were significantly larger than those harboring the pMBC283 vector-only control (Fig. 8A,B). Constitutive expression of serC also restored growth of the recombinant M. bovis BCG crp mutant to WT levels by late log phase (Fig. 8C). While constitutive expression of serC also improved growth of M. tuberculosis crp strain in liquid medium relative to that of the mutant carrying only the vector plasmid (Fig. 8D), it was not to the same extent as was observed on solid media (Fig. 8B), or with direct serine supplementation (Fig. 7B, C). Carriage of the pMBC283 plasmid itself also slowed the mutant’s growth, for reasons that are not clear.
The ability of constitutive serC overexpression to restore growth of M. tuberculosis crp within macrophages was also tested, as the M. tuberculosis crp mutant was previously shown to be defective for growth in this environment (Rickman et al., 2005). Macrophages were infected with WT or M. tuberculosis crp mutant bacteria that carried either empty vector (pMBC283) or the serC-expression vector (pMBC1091). Bacterial growth within macrophages was monitored over a nine-day period by microscopy and plating for colony forming units CFUs (Figure 9). Microscopy was used in parallel with CFUs as an alternative read-out to ensure that the poor plating efficiency of the crp mutants at low density did not result in under-counting of mutant bacteria.
As expected, M. tuberculosis crp showed little to no growth within macrophages when assayed by either method, while the mutant’s growth was fully restored by expression of crp in trans. In contrast, constitutive expression of serC had no effect on growth M. tuberculosis crp within macrophages. Infection levels of the crp mutant remained steady throughout a 9 day period, with or without the addition of the constitutive serC gene construct. This suggests that bacterial growth, rather than survival is most affected by crp deletion. In contrast, the number of WT and crp-complemented mutant M. tuberculosis bacteria increased significantly over time within macrophages. Thus, the crp mutant’s intramacrophage defect is not due solely to lack of serC expression and its source likely differs from that of its in vitro growth deficiency.
Appropriate gene regulation in response to different environmental conditions is critical for M. tuberculosis’s pathogenicity during host infection. CRPMt is a global transcription factor with a putative regulon that includes more than 100 genes (Bai et al., 2005, Krawczyk et al., 2009). We previously showed that the serC-Rv0885 intergenic region contains a CRPMt binding motif that specifically interacts with CRPMt and CRPBCG in vitro and in vivo (Bai et al., 2007, Bai et al., 2005). However, neither serC nor Rv0885 were identified in a microarray study that compared gene expression between an M. tuberculosis crp mutant and WT M. tuberculosis, raising questions about the role of CRPMt in their regulation (Rickman et al., 2005). The present study clearly shows that expression of both serC and Rv0885 is directly activated by CRPMt through association with the previously defined CRPMt binding site (Bai et al., 2005). The absence of these genes from the previous report (Rickman et al., 2005) is most likely due to the sensitivity of the microarray assay or differences in growth conditions between laboratories. Future studies are needed to confirm CRPMt-mediated regulation of other putative regulon members (Bai et al., 2005).
In E. coli, cAMP is a second messenger within a global regulatory network that includes multiple overlapping regulons that respond to signals such as starvation and anaerobiosis (Kolb et al., 1993). The environmental signals to which CRPMt responds have not yet been defined. M. tuberculosis is also exposed to restricted nutrient conditions and hypoxic environments in the host, and previous studies showed that Rv0885 is up-regulated in both conditions (Bai et al., 2007, Betts et al., 2002, Sherman et al., 2001). However, a direct effect of cAMP levels on Rv0885’s regulation has not yet been demonstrated, despite evidence that CRPMt (and CRPBCG) allosterically binds this second messenger (Bai et al., 2005, Reddy et al., 2009, Stapleton et al., 2010).
CRPMt is a homodimer, and cAMP binding reorients the helix-turn-helix domains to maximize specific DNA binding. However, reports differ as to the extent of its asymmetry in the absence of cAMP (Gallagher et al., 2009, Kumar et al., 2010). In contrast with E. coli CRP, apoCRPMt shifts from an open to a closed structure in the presence of cAMP, with the closed structure being the more active one (Reddy et al., 2009). This structural change is consistent with results of an earlier study showing that cAMP binding reduces CRPMt’s and CRPBCG’s sensitivity to trypsin digestion, while protecting and enhancing their DNA binding ability (Bai et al., 2005, Bai et al., 2007).
Nonetheless, CRPMt’s cAMP binding affinity and the allosteric effects associated with cAMP binding are relatively low compared with those of E. coli CRP (Stapleton et al., 2010), and cAMP is not actually required for at least some of CRPMt’s DNA binding interactions (Bai et al., 2005, Stapleton et al., 2010). While cAMP levels can be high in mycobacteria, intracytoplasmic levels of cAMP in mycobacteria are similar to those of E. coli in many conditions (Bai et al., 2011, Bai et al., 2009, Botsford, 1981, Padh & Venkitasubramanian, 1976, Biville & Guiso, 1985). Thus, the mechanisms by which these mycobacterial CRP orthologs regulate gene expression in response to cAMP may differ from the classical E. coli CRP paradigm, particularly with respect to sensitivity thresholds. It is also likely that secondary regulatory factors, such as the repressor predicted from these studies, play important roles as co-regulators, as has been reported for Lrp co-regulation of several CRP-regulated genes in E. coli, including serC and serA (Bilge et al., 1993, Lange et al., 1993, Mathew et al., 1996, Man et al., 1997, Yang et al., 2002). Future studies are clearly needed to address the roles of cAMP levels and co-regulation of CRPMt-regulated genes in mycobacteria.
Our results indicate that CRPMt serves as both an activator (serC and Rv0885) and, possibly, a repressor (lipQ). This study establishes that CRPMt directly regulates expression of serC and Rv0885, but the nature of CRPMt’s role in lipQ regulation has not been determined. cAMP also regulates serC expression in E. coli, but in this case cAMP-CRP is thought to mediate its regulatory effects indirectly (Lim et al., 1994, Man et al., 1997). In E. coli, the position of CRP-binding with respect to the TSS largely determines whether the resulting regulation is positive or negative (Kolb et al., 1993, Botsford & Harman, 1992). CRP-activated promoters tend to have binding sites centered at −41, −61, or −70, placing CRP and the polymerase on the same face of a DNA helix with a periodicity of 10.5 (Kolb et al., 1993). In contrast, CRP binding at other sites (e.g., −60 to +20) typically represses transcription, due to negative interactions with RNA-polymerase (Kolb et al., 1993). The CRPMt binding site is centered at −55.5 bp from the Rv0885 TSS, so CRPMt’s activation of Rv0885 expression is consistent with the positive regulation predicted by the E. coli CRP model. However, binding of CRPMt to the site centered at either + 21.5 bp or −12.5 bp from two possible dominant serC TSSs also resulted in positive regulation, a result that fits less well with the E. coli CRP regulatory paradigm. Nonetheless, the positioning of sites relative to the TSS corresponds well with the 10.5 bp periodicity associated with positive regulation, suggesting that this periodic spacing is a dominant regulatory determinant. Additional minor TSSs are separated by half a helix turn and located even further upstream, at−51.5 or−57.5 relative to the CRPMt motif. While the relative use of these possible start sites in different conditions remains to be determined, these data suggest that CRPMt’s binding is not an impediment to RNA polymerase’s progression once transcription has been initiated. Detailed regulatory studies are also needed to establish the means by which CRPMt is able to exert a positive regulatory effect from this unusual position. CRP in Vibrio cholerae regulates expression of CqsA at the level of mRNA stability (Liang et al., 2008), and such additional levels of CRPMt-mediated control cannot be ruled out in the case of M. tuberculosis serC.
DNA sequence substitution of the ‘Ctl 1’ region, which contains the 16 bp replacement located 18 bp upstream of the CRPMt motif core for serC (Fig. 1D) significantly increased the expression of both serC and Rv0885, relative to the expression levels of the WT promoters (Fig. 3). This DNA region is between the TSSs of serC and Rv0885, proximal to the −10 sequences of both genes. We propose that an unidentified repressor binds within the ‘Ctl 1’ region, interfering with RNA polymerase’s access to both −10 sequences. This putative repressor is unlikely to be CRPMt, because the CRPMt- DNA binding is not changed by the sequence replacement, nor does CRPMt bind the replacement sequence. Data from another recent study also indicate a high level of complexity in CRPMt’s regulatory mechanisms (Stapleton et al., 2010). CRPMt-mediated regulation of whiB1 expression can be either positive or negative, depending on which of two upstream CRPMt binding sites is occupied (Agarwal et al., 2006, Stapleton et al., 2010). Additional studies are clearly needed to better define the scope CRPMt’s regulatory mechanisms for serC and other genes.
Our data indicate that CRPMt regulates expression of serC, which encodes a phosphoserine aminotransferase with activity similar to that of E. coli SerC. The growth of M. tuberculosis and M. bovis BCG crp mutants is slower than that of their respective WT strains, consistent with CRPMt -mediated regulation of mycobacterial metabolism. The reported requirement of serC for M. tuberculosis growth in vitro (Sassetti et al., 2003) led us to test whether decreased levels of serC expression in M. tuberculosis and M. bovis BCG crp mutants caused the mutants’ slow-growth phenotypes. The improved growth associated with either addition of 5 mM serine or constitutive expression of serC from the tuf promoter indicates that serine starvation did contribute to the mutants’ growth defects. However, serC expression restored growth less efficiently than did direct serine supplementation in the case of the M. tuberculosis crp mutant. It is possible that higher levels and/or proper regulation of serC is required for complete phenotypic complementation, or that serC must be co-expressed with other genes in M. tuberculosis for the enzyme to be fully effective. In contrast, the ability of serC expression alone to fully restore WT levels of growth in the M. bovis BCG crp mutant indicates that there are metabolic differences between M. bovis BCG and M. tuberculosis that are at least partly regulated by CRP. Such differences clearly warrant further investigation, and are consistent with previous reports of CRP-associated gene regulatory differences in M. tuberculosis and M. bovis BCG (Bai et al., 2007, Hunt et al., 2008, Spreadbury et al., 2005).
Serine can be used for biosynthesis of glycine, cysteine, and tryptophan in many organisms (Kitabatake et al., 2000, Largen & Belser, 1975, Pizer, 1965, Ulane & Ogur, 1972). In the present study, addition of glycine and cysteine, but not tryptophan, enhanced the growth of the crp mutants at levels similar to those observed with addition of serine. This result supports the proposed biosynthesis pathway (Fig. 7A) in which glycine and cysteine, but not tryptophan, are able to interconvert with serine. The biosynthesis pathways of these three amino acids have not been reported in mycobacteria. However, all of the enzymes necessary to convert serine into these amino acids have been predicted (Table 1). It is likely that a serine bypass exists for tryptophan biosynthesis, or that tryptophan synthesis is sufficient even with low serine levels caused by dysregulation of serC in the crp mutants.
It is clear that the dysregulation of serine biosynthesis contributes to the growth defect of crp mutants in vitro, but restoration of serine biosynthesis by constitutive serC expression did not improve growth of the M. tuberculosis crp mutant within macrophages (Fig. 9). Therefore CRPMt-regulated genes other than serC are also critical for M. tuberculosis replication within macrophages, and likely in vivo during animal infection. The missing factor(s) could either provide protection against inhibitory components or facilitate metabolic adaptations for growth within the specialized macrophage environment. For an example, we found that the presence of 5 mM pyridoxine (a component of vitamin B6, which exists in vivo) is toxic to crp mutants, but not to the WT strains (Fig. 7, S3), and expression of serC in the crp mutant of M. bovis BCG did not reduce this toxicity (Data not shown).
The defective growth of the crp mutants with impaired serine biosynthesis suggests that CRPMt and SerC are potential drug targets for TB therapy. Restoration of in vitro growth by serC complementation will facilitate the identification of CRPMt regulon members that specifically contribute to M. tuberculosis’s fitness in vivo, leading to a better understanding of CRPMt’s role in TB pathogenesis.
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 (OADC), 0.05% Tween-80 (defined as mycomedium), as previously described (Florczyk et al., 2003) or on Middlebrook 7H10 agar (Difco) supplemented with 10% OADC, and 0.01% cycloheximide. Fresh cultures were inoculated from frozen stocks for every experiment. Bacteria were typically used in late log phase after 7 days of growth. Cultures were grown with gentle rocking in ambient air conditions except as specified. Escherichia coli strains were grown in Luria broth or Luria agar plates. Kanamycin at 25 μg/ml, or hygromycin at 50 μg/ml, was used where specified. All cultures were grown at 37°C. Addition of cAMP or amino acids is indicated in the Results section for specific experiments.
J774.16 murine macrophage cells were maintained by twice-weekly passages in antibiotic-free tissue culture medium (Dulbecco’s modified Eagle’s medium (Gibco) supplemented with 5% NCTC (Gibco), 1% nonessential amino acids, 20% fetal bovine serum, and 1% glutamine) as described previously (McDonough & Kress, 1995).
Plasmid p0004S (kindly provided by Dr. William Jacobs, Albert Einstein College of Medicine) (Vilcheze et al., 2008) was modified by digestion with NotI, blunted and religated to generate pMBC819. The PacI-MscI fragment of pMBC819 was further replaced with PacI and MscI digested PCR fragment amplified with KM2271and KM2257 from pMBC819 to introduce a unique NotI site after PacI and to remove the two PflMI (isochizomer of Van91I) sites within the original PacI-MscI fragment. This construct was designated as pMBC851 and was used to sequentially clone an upstream crp-flanking DNA sequence into the newly introduced unique NotI site, and to replace the remaining PflMI fragment with a DNA fragment from downstream of crp, to generate pMBC879. We amplified the M. tuberculosis crp upstream region with KM772 and KM773 (with NotI linkers), and a downstream fragment with KM2061 and KM2062 (with PflMI ends). The final pMBC879 construct was sequence verified before being used to generate crp knockouts in M. tuberculosis and M. bovis BCG, according to published methods (Bardarov et al., 2002, Tufariello et al., 2004). Briefly, pMBC879 was linearized with PacI and ligated with PacI-digested shuttle phasmid phAE159 (kindly provided by Dr. William Jacobs) (Vilcheze et al., 2008). The ligation mixture was packaged using MaxPlax lambda packaging extracts (Epicentre Biotechnoloies) and transduced into E. coli HB101. Phasmid DNA was prepared from hygromycin-resistant colonies and transformed into M. smegmatis mc2155 to produce the knockout phage stocks. Mutants were screened by PCR with primer pairs KM844 + KM2585 and KM2085 + KM845, which were designed to amplify the junction regions between the insertion marker and flanking chromosomal DNA beyond that which was included in the knockout construct. The disruptions were further confirmed by PCR with primers KM859 + KM860, which amplify internal M. tuberculosis crp sequences. The M. bovis BCG mutant was used throughout these studies, while the corresponding M. tuberculosis mutant was used for the macrophage infection experiments. Other M. tuberculosis experiments were performed with the crp mutant generously provided by Dr. R. Buxton prior to completion of the new M. tuberculosis mutant. The deletion endpoints of our M. tuberculosis crp mutant are identical to those in our M. bovis BCG mutant, but differ slightly from those of the mutant generated by the Buxton laboratory. However, the phenotypes of both M. tuberculosis crp mutants were similar in all assays (not shown).
M. tuberculosis crp was amplified with KM1327 + KM963 and was cloned into the integrating expression vector pMBC409 at HindIII, to generate pMBC587. The M. bovis BCG crp homolog was amplified with KM1327 + KM963 and cloned into pMBC409 at HindIII site, to generate pMBC1029. Plasmids pMBC587 and pMBC1029 were transformed into crp mutants of M. tuberculosis and M. bovis BCG, respectively, to generate single-copy complemented strains.
serC was expressed in the crp mutants using the expression plasmid pMBC283, which contains oriM and the M. tuberculosis tuf promoter, along with both kanamycin and hygromycin resistance markers (from pCRII and pEZ::TNpMOD, respectively). The serC orf was amplified using primers KM2713 and 2714 (Table 2), and ligated into pMBC283 at HindIII to generate pMBC1091 for expression of serC under control of the tuf promoter.
Mycomedium was modified by addition of 5 mM of each supplement as specified, and was adjusted to pH 6.6 – 6.8. The growth of WT and crp mutants in both M. tuberculosis and M. bovis BCG was compared in the differentially supplemented media within 25 cm2 flasks under gentle shaking conditions, at 37°C with ambient air. Samples of 100 μl were withdrawn and diluted with 100 μl of mycomedium in 96-well plates for each time point. Bacterial concentrations were determined by measurement of the optical densities at 620 nm (OD620) using a spectrophotometric microplate reader (Tecan).
J774.16 macrophages were seeded at 1×105 cells/ml in 6 well plates approximately 24 h prior to infection, and incubated at 37°C with 5% CO2. Macrophages were then infected with WT or mutant M. tuberculosis strains at a multiplicity of infection of approximately one bacterium per macrophage, as described previously (McDonough & Kress, 1995). Briefly, after 4 h infection, monolayers were washed twice with phosphate-buffered saline (PBS) solution at pH 7.4 and left in fresh tissue culture medium, which was replaced every 2–3 days throughout the infection. At the specifed timepoints, macrophages were lysed with 0.5% Triton X-100 and 0.5% sodium deoxycholate for 3 min. Dilutions of cell lysate were then plated onto Middlebrook 7H10 plates and CFUs were enumerated after three weeks for WT and six weeks for mutant strains. For microscopy, infected macrophages were fixed with 4% paraformaldehyde in PBS, stained using a Kinyoun acid-fast protocol ((Kinyoun, 1915), and mounted onto slides for microscopy, as described previously (McDonough & Kress, 1995).
CRPMt and CRPBCG were expressed and purified as previously described (Bai et al., 2007, Bai et al., 2005). PCR primers were labeled as probes with γ- 33P-ATP using T4 DNA polynucleotide kinase (New England Biolabs). DNA fragments were amplified by PCR, and approximately 0.05 pmol DNA probe was used in each 10 μl of binding reaction, as described previously (Bai et al., 2005). A 500-fold excess of unlabeled DNA was used for competition experiments. Samples were loaded on a non-denaturing 8% polyacrylamide gel and run for 2 – 3 h at 14V/cm. Gels were vacuum dried, exposed on a phosphor screen, scanned with a Storm 860 PhosphorImager (Molecular Dynamics), and analyzed with ImageQuant software.
Total RNA was extracted using method of Mangan and co-workers (Mangan et al., 1997). After isopropanol precipitation, RNA was treated twice with RNase-free DNase I on a spin column using the RNeasy Mini Kit and RNase-Free DNase Set (Qiagen). The RNA concentration was determined spectrophotometrically with a Biophotometer (Eppendorf) at 260 nm.
The primer extension experiments were performed as described previously (Vasudeva-Rao & McDonough, 2008). [γ-32P]-ATP labeled primers that were complementary to regions downstream of serC (KM1608 and KM1609) and Rv0885 (KM1335) promoters, respectively, were hybridized separately to 20 μg total RNA isolated from exponentially growing M. bovis BCG. Hybridization was performed at 100°C for 1 min, followed by annealing for 20 min at 60°C or 65°C for serC, or at 52°C for Rv0885. Extension reactions were performed with SSRIII reverse transcriptase (Invitrogen) at 42°C for 1 h. Samples were separated on a 6% polacrylamide sequencing gel containing 8 M urea. In each case, the adjacent lane contained a DNA sequence ladder generated by dideoxy sequencing (Sanger et al., 1977) of plasmid pMBC648, which contains the serC-Rv0885 intergenic sequence, using the relevant oligonucleotide as primer. Gels were vacuum dried, exposed on phosphor screens, and scanned with a Storm 860 PhosphorImager (Molecular Dynamics).
The intergenic region between serC and Rv0885 was PCR-amplified using primers KM1229 and KM1230 (Table 2), and cloned into the single-copy integrative vector, pLACint, upstream of a promoterless lacZ gene, as previous reported (Purkayastha et al., 2002). Point mutations and whole motif core substitutions were generated by strand overlap extension (SOEing)-PCR (Horton, 1997, Vasudeva-Rao & McDonough, 2008). KM1229, KM1230, KM1448 and KM1449 were used to generate the G4-to-C mutation. KM1229, KM1230, KM1450 and KM1451 were used to generate the C17-to-G mutation. The motif core sequence (TGTGAGCTGTTCACA) was substituted with a 16 bp random sequence (ATATCTACTGAGTTAC) by using primers KM1229, KM1230, KM1521 and KM1522. A 16 bp sequence (GACACTATGACCCAGC), located 18 bp upstream of the motif core, was substituted with the same random sequence using primers KM1229, KM1230, KM1563 and KM1564 to generate the control-1 DNA fragment. A 16 bp sequence (AGGCTTTCGTTCCGCT), which is located 18 bp downstream of the motif core, was replaced with the same random sequence by using primer KM1229, KM1230, KM1565 and KM1566 to generate the control-2 DNA fragment. The truncated promoter fragments were amplified using primers KM1463 and KM1230, or KM1464 and Rv1230 for serC, and KM1465 and KM1229, or KM1466 and KM1229 for Rv0885. All promoter fragments were cloned into pLACint at the BamHI site and sequence-verified before being transformed into mycobacteria, to generate the reporter strains.
Bacteria were grown to exponential phase by shaking at 37°C in mycomedium. A 300 μl portion of each sample was transferred to a fresh tube and sonicated, as previously described (McDonough et al., 2000). The β-galactosidase activity of the recombinant cultures was measured as described previously (Vasudeva-Rao & McDonough, 2008) by using 5-acetylamino-fluorescein di-β-D-galactopyranoside (C2FDG) (Molecular Probes), which is cleaved by beta-galactosidase to produce fluorescein. Fluorescein was measured in a CytoFluor Multi-Well Plate Reader (PerSpective Biosystems), and activity readings in arbitrary units were normalized by the optical density of the cultures at 650 nm, using a THERMOmax microplate reader (Molecular Devices). Results were expressed as relative β-galactosidase activity/OD.
M. tuberculosis WT, crp mutant and the complemented strain (kindly provided by Dr. Roger Buxton) (Rickman et al., 2005), were grown to late log phase in mycomedium. cDNA was prepared and amplified as previously described (Gazdik & McDonough, 2005). All samples were standardized against 16S RNA (Alland et al., 1998) using cDNA diluted 10−4 or 10−5, and against sigA (Manganelli et al., 1999) using cDNA diluted 10−2. PCR of 16S RNA and sigA was also performed prior to cDNA generation, using total RNA without reverse transcription, to ensure the absence of DNA contamination. Gels were scanned with a FluorImager595 (Molecular Dynamics) and analyzed with ImageQuant software (Molecular Dynamics).
The E. coli crp promoter was amplified with primers KM1491 and KM1472 (Table 2), and cloned into pBR322 to generate pMBC664, while the M. tuberculosis serC orf was amplified with primers KM2360 and KM2361, and inserted into pMBC664 to generate pMBC886. E. coli serC was amplified with primers KM2362 and KM2363, and was cloned into pMBC664 to generate pMBC887. The plasmids pMBC664, pMBC886 and pMBC887 were individually transformed into an E. coli serC mutant strain KL285 (CGSC4310) for further analyses (Clarke et al., 1973). Growth among the recombinant strains was compared as described in a previous report (Helgadottir et al., 2007). Briefly, they were grown on M9 minimal medium plates containing 0.2% (w/v) D-glucose and 19 amino acids (excluding serine) at 10 μg/ml each. Serine was added at 50 μg/ml for plates with serine. All plates contained 100 μg/ml ampicillin, and were grown for 48 h at 30°C.
A two-tailed t test was used to determine the significance of differences between sample results using Prism 4 software. P values <0.05 were considered significant.
We thank Dr. Bill Jacobs for providing p0004S and phAE159; Dr. Roger Buxton for providing the M. tuberculosis crp mutant; and the E. coli Genetic Center for providing the KL285 strain. We also thank Dr. Gwen Knapp, Dr. Hema Vasudeva-Rao and Ms. Tachanda Byrant for valuable discussions and technical assistance. DNA sequencing was done through the Wadsworth Center Molecular Genetics Core. This work was supported in part by National Institutes of Health grant AI063499.