Two component regulatory systems (2CRS) comprise a sensor histidine kinase (HK) and a response regulator (RR), which sense and respond to various environmental conditions, thus promoting bacterial survival [1]. Upon detection of a specific stimulus, the HK autophosphorylates and transfers a phosphoryl group to the cognate RR, which promotes DNA binding of the latter to initiate transcription of the appropriate adaptive response [2],[3]. However, under certain circumstances, bacterial RR may also be phosphorylated by a noncognate HK or small molecule acetyl phosphate, merging different stimuli or diversifying the response to a single signal [4], or may function without being phosphorylated [5]-[8].

Twelve complete pairs of 2CRS and a few orphan HK and RR genes have been found in the Mycobacterium tuberculosis (Mtb) genome. SenX3-RegX3 was among the first 2CRS to be identified in Mtb [9],[10]. The senX3-regX3 operon is required for Mtb survival during murine infection [11]-[13]. However, conflicting results have been reported for the virulence of strains deficient in senX3 or regX3 in the lungs and spleens of BALB/c mice [13]. Unlike the case in other 2CRS, expression of senX3 and regX3 is monocistronic as well as polycistronic [14],[15], raising the possibility that each may function independently of the other under clinically relevant conditions and during host infection. In M. smegmatis, co-transcription of senX3 and regX3 is required for the expression of phosphate-dependent genes and for optimal growth under inorganic phosphate (P_i)-limiting conditions [16],[17]. Previously, we have shown that the senX3-regX3 operon is involved in the Mtb phosphate starvation response (PSR) [12], however the individual role of each gene in Mtb survival during P_i depletion has not been characterized. Moreover, RegX3, when overexpressed in M. smegmatis, can be phosphorylated in the absence of SenX3 in P_irich medium [17], and the SenX3-RegX3 homolog PhoBR in E. coli also responds to nutrient starvation in addition to P_i depletion [18]. These findings suggest that Mtb SenX3-RegX3 may have a broader role in Mtb virulence beyond the PSR and that these two factors may function independently of each other.

In order to further understand the role of each gene of the senX3-regX3 2CRS in Mtb virulence, and in particular the senX3-independent contribution of regX3, two different mutants deficient in senX3 or regX3 were studied under multiple conditions and their phenotypes were compared to the isogenic wild-type parent strain Mtb CDC1551, as well as their respective complemented strains. The expression of senX3 and regX3, as well as that of the phosphate specific transport gene pstC2, were evaluated in all the tested strains under physiologically relevant stress conditions. In addition, the growth and survival of each of these strains were assessed during axenic growth in nutrient-rich broth, P_i depletion, nutrient starvation, and after aerosol infection in the lungs of BALB/c mice. Our data support the hypothesis that RegX3 may have a SenX3-independent role in Mtb virulence.

SenX3-RegX3 is among the best characterized 2CRS in Mtb [9],[10], yet many questions remain about its role in virulence. Homologs of senX3 and regX3 are found in the genomes of various mycobacterial species, indicating that this evolutionarily conserved 2CRS may play a fundamental regulatory role in mycobacterial physiology [19]. Unlike other 2CRS, in which expression of the operon is bicistronic and auto-regulatory, senX3 and regX3 can be transcribed independently of each other [15]. In particular, the intergenic region between the two genes contains several mycobacterium interspersed repetitive units (MIRU), which exist only in M. leprae and members of the Mtb complex [9],[14],[17],[20],[21], and which can drive senX3-independent expression of regX3[15]. In the current study, we sought to investigate the senX3- independent contribution of regX3 to Mtb gene regulation, growth, and survival under various physiologically relevant growth-limiting conditions in vitro and in vivo.

We found that the genes senX3 and regX3, while dispensable for Mtb growth in nutrient-rich broth, are equally required for optimal growth and morphological response during P_i depletion. In addition, as evidenced by the statistically greater survival of senX3::Tn relative to regX3::Tn, we found that regX3 contributes to Mtb survival independent of senX3 during prolonged nutrient starvation and chronic infection in mouse lungs.

Although SenX3-RegX3 has been implicated in the mycobacterial PSR [12],[16],[17], the functional role of each gene in this operon has not been characterized previously during P_i depletion. P_i depletion is believed to be a physiologically relevant microenvironment encountered by Mtb within the arrested macrophage phagolysosome [22]. Mtb senX3-regX3 is upregulated in response to P_i depletion and expression of the phosphate-specific transport operon pstS3-pstC2-pstA1 is RegX3-dependent [12]. Previously, we have shown monocistronic upregulation of senX3 and regX3 during P_i depletion [15], which differs from M. smegmatis, in which the senX3-regX3 operon is co-transcribed. In the current study, we found decreased expression of regX3 in the senX3-deficient mutant relative to the wild type during P_i depletion, which could be restored by reintroduction of senX3. Decreased regX3 expression in P_i-depleted senX3::Tn may be due to decreased co-transcription with senX3 or down-regulation of independently expressed regX3. Although the senX3 promoter is autoregulatory and activated by RegX3, which is phosphorylated by SenX3 [23], it is unknown whether the intergenic region driving independent regX3 expression is also RegX3-dependent, and, hence, SenX3-dependent.

We found that Mtb mutants deficient in senX3 or regX3 exhibited similar growth defects during P_i depletion. Despite expression of regX3 in the senX3-deficient mutant, it appears that RegX3 was not functional, as expression of the RegX3-dependent, P_i-specific transport gene pstC2 was dramatically downregulated in this strain. Our results are consistent with monocistronic, P_i-dependent expression of each gene and the putative role of SenX3 as a P_i sensor [23]. Therefore, although monocistronic expression of regX3 is preserved in senX3::Tn [15], the cognate HK SenX3 may be required to phosphorylate and activate RegX3 during P_i depletion, thereby triggering the PSR. In vitro studies in many bacteria have provided evidence in favor of the specificity of 2CRS, in which the HK show a remarkable kinetic preference for their cognate RR primarily at the level of molecular reorganization [24]-[31]. For example, in B. subtilis the HK KinA can phosphorylate either Spo0F or Spo0A, but with more than 50,000-fold preference for Spo0F during sporulation [24],[25].

Interestingly, despite similar growth phenotypes of senX3::Tn and regX3::Tn during P_i limitation, the senX3-deficient mutant showed markedly higher survival relative to the regX3-deficient mutant upon prolonged nutrient starvation and in the lungs of mice, while both mutants were attenuated compared with the wild-type. Our gene expression results demonstrate reduced but detectable expression of regX3 in the senX3-deficient mutant. This discordance in phenotypes suggests the possibility that independently transcribed regX3 may be translated and phosphorylated in the absence of SenX3, thereby initiating transcription of the RegX3 regulon [15],[17],[23]. In M. smegmatis, the essential gene regX3 can be activated in the absence of the nonessential gene senX3 in P_i-replete broth, suggesting the existence of alternative mechanisms of RegX3 phosphorylation [17]. Similarly, in E. coli the RegX3 homologue PhoB can be phosphorylated by the HK EnvZ [32], CreC (PhoM), or acetyl phosphate in the absence of PhoR in P_i-independent pathways [33]. Precedents for phosphorylation of RR by noncognate HK are also present in Mtb. Thus, the Mtb dormancy response regulator DosR can be phosphorylated by its cognate HK DosS or by the orphan HK DosT [34],[35]. Although 2CRS usually maintain specificity to prevent unwanted cross-talk [24]-[31], significant sequence and structure similarity in HK and RR, as well as the existence of small molecule phosphodonor acetyl-phosphate make cross-regulation possible [36]-[38]. In addition, cross-regulation could occur at the transcriptional level [39],[40]. Under certain circumstances many bacteria appear to employ cross-regulation in order to integrate multiple signals or diversify the response to a single input [4],[41]. Therefore, based on our initial findings, we speculate that although Mtb senX3-regX3 2CRS maintains specificity during P_i depletion, cross-regulation of RegX3 may facilitate a rapid response of Mtb to a changing environment in vitro and in vivo under other stresses, in which the identity of alternative HK and/or other activating factors remains to be determined. Alternatively, independently transcribed RegX3 may be functional in the absence of phosphorylation. For example, the Pseudomonas aeruginosa RR AlgB and AlgR do not require phosphorylation by the cognate HK KinB for alginate production in vivo, indicating that these RR may mediate activation of gene expression through alternative mechanisms [5]. We speculate that the SenX3-RegX3 2CRS may provide a common means by which to transduce the signals of P_i depletion and nutrient starvation, leading to induction of the stringent response, which is critical for Mtb persistence in the host [42],[43]. Our findings are consistent with the function of the homologous PhoBR 2CRS in E. coli, which regulates the transcriptional response to P_i depletion and nutrient starvation [18],[44]. The mechanism of regX3 regulation during nutrient starvation and in vivo infection remains to be identified.

Higher survival of senX3::Tn relative to regX3::Tn during chronic infection in mice is consistent with previous data showing higher survival of a senX3 deletion mutant relative to a regX3 deletion mutant in mouse spleens [13]. However, our data conflict with those of Rickman et al. in that they showed greater attenuation of the senX3 deletion strain in mouse lungs relative to the regX3 deletion strain. These discrepant findings may be due to methodological differences between our study and that by Rickman et al. The latter study used H37Rv, in which the senX3-regX3 IR comprises three identical 77-bp MIRU rather than two 77-bp MIRU and one 53-bp MIRU, as in the CDC1551 strain used in our study [13],[14]. In addition, Rickman et al. used intravenous infection rather than aerosol infection, although both studies used BALB/c mice. Nevertheless, these two studies corroborate our recent finding that senX3 and regX3 do not simply function as a bicistronic operon, as previously thought [14], since, in such a case, disruption of senX3 would be expected to yield mutant phenotypes identical to those of a regX3-deficient strain. In the current study, both complemented strains restored wild-type growth and survival during P_i depletion, and, in the case of regX3::Tn Comp, during nutrient starvation and in the lungs of mice. However, unlike in a previous study [13], our complemented strains failed to restore the wild-type phenotype during mouse infection. There are several potential explanations for these discrepant findings. In comparison to the cloning strategy by Rickman et al., our senX3-regX3 complementation construct lacked a 211-bp sequence upstream of the senX3 gene, which may have altered regulation of this gene during the in vitro stress conditions studied and mouse infection. Another difference between the two complementation approaches is that in the earlier study the senX3 and regX3 genes were fused together by deleting the intergenic region [13]. On the other hand, each of our complemented strains included this intergenic region, since we recently described the potential importance of its promoter activity in driving senX3- independent expression of regX3[15]. The precise role of this intergenic region in regulating expression of each of these genes requires further study. In addition, the presence of a second copy of the regX3 gene may have had a deleterious effect on in vitro and in vivo survival. As in the case of other defective complemented strains, the chromosomal location at the recombination site may have affected proper expression of senX3 and regX3. Alternatively, the development of other occult mutations during genetic construction of these strains may have contributed to their reduced virulence relative to the wild-type strain. However, deficiency of senX3 or regX3 was reported previously to be associated with reduced Mtb virulence in mice, and the attenuated mutant phenotypes were fully complemented in the study by Rickman et al. [13]. Therefore, we favor the explanation that our complementation strategy, as described above, contributed to the incomplete restoration of the wild-type phenotypes in the complemented strains. The mechanisms by which senX3 and regX3 are regulated under different stress conditions require further characterization.

Our data raise the interesting possibility that differential regulation of senX3 and regX3 may have evolved to allow for simultaneous coordination of the Mtb PSR [12], while providing a common means by which bacilli may respond to a variety of different physiologically relevant stress conditions. In particular, the mechanisms underlying the SenX3-independent contribution of RegX3 to Mtb virulence deserves further investigation.

The experimental design was developed by DR and PCK. The experiments were performed by DR. DB was responsible for histological analysis. All authors contributed to writing of the manuscript. All authors read and approved the final manuscript.