Temperature Control of psaA Expression by PsaE and PsaF in Yersinia pestis

Y. pestis is a Gram-negative bacterial pathogen that causes bubonic plague. As a vector-borne pathogen, Y. pestis fluctuates between an arthropod vector (flea) and mammalian host. As such, Y. pestis must recognize environmental signals encountered within each host environment and respond by appropriately regulating gene expression. PsaA is a key Y. pestis mammalian virulence determinant that forms fimbriae. Our work provides evidence that Y. pestis utilizes multiple posttranscriptional mechanisms to regulate the levels of two PsaA regulatory proteins in response to both temperature and pH. This study offers insight into mechanisms that bacteria utilize to sense environmental cues and regulate the expression of determinants required for mammalian disease.

blood meal. The transmission of Y. pestis to a flea from an infected mammal also occurs during a blood meal and requires high levels of bacteria in mammalian blood (ϳ10 8 CFU/ml); thus, both survival within a flea and systemic dissemination within a mammalian host are essential for the biphasic life cycle of Y. pestis (10). Transcriptome analyses reveal distinct Y. pestis expression profiles within a flea and mammalian host (11)(12)(13), suggesting that Y. pestis regulates gene expression in response to differential environmental signals encountered within each host. While the signals that distinguish the mammalian and flea microenvironments are not well defined, the temperature difference between a flea (ϳ26°C) and mammalian host (37°C) is thought to serve as a key environmental signal, as the expression of many Y. pestis virulence genes increases following an upshift in temperature from 26°C to 37°C (13,14).
One such temperature-regulated virulence factor of Y. pestis is the "pH 6 antigen" (PsaA). PsaA forms fimbria-like structures on the cell surface (15,16) and is required for the full virulence of Y. pestis in multiple mouse models of disease (15,(17)(18)(19)(20). While the functional role of PsaA during mammalian infection has not been defined, in vitro studies using various cell lines suggest that PsaA functions to both inhibit phagocytosis and promote host cell adherence (21,22). Intriguingly, PsaA production requires a combination of high temperature (Ͼ35°C) and acidic pH (pH Ͻ6.7) (23), and, since the detection of psaA transcripts corresponds with the detection of PsaA (24), it is predicted that Y. pestis utilizes temperature and pH to regulate the transcription of psaA. Despite this unusual expression pattern, the regulation of psaA transcription has not been examined in Y. pestis, and the mechanisms by which temperature and pH contribute to psaA transcription are not known.
The psa locus consists of five genes responsible for the production, translocation, and assembly of PsaA subunits into fimbria-like structures; psaE and psaF encode regulators, psaA encodes the fimbrial subunit, and psaB and psaC encode proteins that resemble the PapD and PapC families of chaperone and usher proteins, respectively (16). In the absence of psaBC, PsaA accumulates in the cell, indicating that the products of these genes contribute to PsaA export (15,16). Encoded upstream of psaA, PsaE and PsaF are predicted to be transcriptional activators that coregulate psaA transcription (15,24). However, the precise role(s) of PsaE and PsaF in Y. pestis is not thoroughly understood, and their predicted function(s) is extrapolated in part from studies in Yersinia pseudotuberculosis (psa) and of homologues in the closely related species Yersinia enterocolitica (myf) (25)(26)(27). In Y. pseudotuberculosis, psaE and psaF are both required to detect psaA transcripts (27). PsaE is predicted to have a DNA-binding domain, and there is evidence that PsaE contributes to psaA expression in Y. pestis (15,24); however, PsaF has no conserved domains, and its role remains elusive. Analysis of fusions of alkaline phosphatase to PsaE and PsaF in Y. pseudotuberculosis suggests that both proteins are integral membrane proteins, each with a single transmembrane domain (27). While pairs of regulatory proteins with similar topology to PsaE and PsaF have been identified in other bacteria (28)(29)(30)(31), PsaE and PsaF exhibit little primary sequence similarity with known proteins. The transcription of psaE and psaF in Y. pseudotuberculosis is not affected by temperature or pH, and thus, the transcription of these regulatory genes is not sufficient to activate psaA transcription (27). This has led to speculation that the production and/or activity of PsaE/F is subject to posttranscriptional regulation (27). However, it has yet to be determined whether PsaE and/or PsaF is influenced by temperature or pH and how PsaE and PsaF influence the expression of psaA.
We set out to define the mechanisms that contribute to the regulation of PsaA production in Y. pestis by PsaE and PsaF. In addition to monitoring psaEF expression, we generated antibodies against both PsaE and PsaF to monitor the levels of these proteins. Analysis of psaE and psaF transcription and PsaE and PsaF protein levels revealed that PsaE and PsaF are impacted by temperature and pH via posttranscriptional mechanisms. Our data suggest that temperature affects synthesis and that pH influences the stability of these key regulators of psaA expression.

RESULTS
Temperature and pH provide discrete signals that activate psaA transcription in Y. pestis. Temperature and pH are well-established environmental signals that affect the expression of psaA in Y. pestis (15,23,24). Despite this, the mechanisms utilized by Y. pestis to regulate the transcription of psaA in response to high temperature (Ͼ35°C) and mildly acidic pH (Ͻ6.7) are not known. To investigate how temperature and pH affected psaA expression and PsaA production in Y. pestis, a psaA-gfp transcriptional reporter (pEW102) was introduced into Y. pestis strain CO92 cured of the virulence plasmid pCD1 (YP6; here referred to as the wild-type strain [WT]). Prior studies on psaA expression used cultures grown under conditions in which the starting pH of the growth medium was adjusted but was not buffered to maintain the pH during growth (15,24,27). Thus, we first examined the expression of psaA-gfp in WT grown in unbuffered brain heart infusion (BHI) broth at 26°C and 37°C; both expression and the pH of the medium were monitored over time (Fig. 1A). Consistent with previous studies, psaA expression was observed only at 37°C. During growth at 37°C, psaA-gfp activity was initially detected at 6 h (late log phase), coinciding with acidification of the growth medium to just below pH 6.3. Thus, it appeared that the combination of temperature and pH was responsible for activating psaA transcription. However, the pH of the growth medium at 26°C never got below 6.5, so we could not rule out the possibility that low pH, rather than temperature, was the activating signal. Furthermore, as the pH of the growth medium at 37°C did not drop below 6.3 until 6 h, it was also possible that growth phase played a role in the activation of psaA transcription. To test these possibil- ities, cultures of WT containing the psaA-gfp plasmid were grown in BHI broth buffered at pH 7.3, 6.7, or 6.3 (Fig. 1C). During growth at 37°C and pH 6.3 (37°C/pH 6.3), psaA-gfp activity was detected within 2 h, suggesting that low pH, rather than growth phase, induced the expression of psaA. The expression of psaA remained minimal throughout growth at 37°C/pH 7.3, but an intermediate level of expression was detected at 37°C/pH 6.7. Minimal psaA-gfp expression was detected after growth at 26°C/pH 6.3 (Fig. 1D). Importantly, levels of bacterial growth were not significantly different between the conditions tested (data not shown).
To determine if PsaA production corresponded with psaA-gfp expression, the WT was grown at 37°C in unbuffered BHI broth and in BHI broth buffered to pH 6.3 and pH 6.7, and PsaA was analyzed via Western blotting using an anti-PsaA antibody (Fig. 1E). In unbuffered BHI broth, PsaA was detectable shortly after psaA-gfp expression was detected (8 h) in late log phase of growth. In contrast, PsaA was detected within 2 h of growth at 37°C/pH 6.3. Thus, the production of PsaA exhibited a similar pattern of regulation in response to pH as psaA transcription. Based on these results, 37°C/pH 6.3 was defined as an inducing growth condition for psaA transcription (and PsaA production), whereas 37°C/pH 7.3 and 26°C/pH 6.3 were defined as noninducing growth conditions. Thus, high temperature and low pH provide discrete signals that are both required to activate psaA transcription.
To characterize the role of PsaE and PsaF in the expression of psaA in Y. pestis, the psaA-gfp plasmid was introduced into the ΔpsaEF mutant (YPA18), and expression was compared between the ΔpsaEF mutant and WT strains ( Fig. 2A). Consistent with previous data in Y. pseudotuberculosis (27), expression of psaA-gfp was not detected in the ΔpsaEF mutant even at 37°C/pH 6.3 (inducing condition), indicating that PsaE and/or PsaF is required for psaA transcription in Y. pestis. The psaE and psaF genes were introduced into the ΔpsaEF mutant at the native site of the chromosome (YPA260). Complementation with both psaE and psaF restored the expression of psaA to WT levels ( Fig. 2A), indicating that the loss of psaA expression in the ΔpsaEF mutant was due to the deletion of psaEF. The complementation of either psaE (YPA265) or psaF (YPA279) alone was not sufficient to restore psaA transcription ( Fig. 2A). Similarly, both psaE and psaF were required for the production of PsaA, as PsaA was undetectable when psaE or psaF alone was introduced into the ΔpsaEF mutant (Fig. 2B)  both psaE and psaF are required for psaA transcription, and thus PsaA production, in Y. pestis.
Temperature and pH impact PsaE and PsaF levels through separate posttranscriptional mechanisms. psaE and psaF coding sequences overlap by 4 bp and are predicted to be cotranscribed from a promoter upstream of psaE. To test this, RNA was isolated from the WT strain grown at 37°C/pH 6.3, and reverse transcription-PCR (RT-PCR) was used to analyze the psaE-psaF junction. Using a single set of primers internal to psaE and psaF, a product was obtained from a template that was subjected to reverse transcription, indicating that these genes are indeed cotranscribed (Fig. 3A). In Y. pseudotuberculosis, the expression of psaE and psaF is not significantly affected by temperature or pH (27). To determine if this is also true in Y. pestis, a psaEF-gfp transcriptional reporter plasmid (pJC126) was introduced into the WT, and expression was measured after growth at 26°C and 37°C in BHI broth buffered at pH 6.3 or 7.3. The expression of psaEF-gfp was detected under all four growth conditions (Fig. 3B). While expression at 37°C was moderately higher than that at 26°C (3-fold), there was no significant difference between expression levels at pH 6.3 and pH 7.3 at either temperature. Notably, psaEF-gfp expression was much higher than in the WT containing the vector control under the four conditions, suggesting that the expression of psaEF occurred under all four growth conditions and, unlike the expression of psaA, was not largely impacted by pH. Thus, the temperature-and pH-dependent regulation of psaA expression is not dictated by transcriptional regulation of psaEF in Y. pestis.
Based on experiments in Y. pseudotuberculosis, Yang and Isberg speculated that PsaE and PsaF are subject to posttranscriptional regulation in response to temperature and WT containing a psaEF transcriptional reporter plasmid (pJC126, psaEF-gfp) was grown at 37°C and 26°C in buffered BHI broth, and the RFU/OD 600 was determined under each condition, as described in Materials and Methods. The number depicted over each bar indicates fold change in psaEF-gfp expression in the WT compared to the expression from the vector control under the given condition. Each bar represents the mean RFU/OD 600 , and error bars represent standard deviations. Each sample was assayed in biological triplicates, and at least three independent experiments were performed. ***, P Ͻ 0.0001; ns, not significant by one-way ANOVA and Tukey's multiple-comparison test. (C) Whole-cell lysates of WT Y. pestis grown at 37°C and 26°C in buffered BHI broth were prepared, and PsaE, PsaF, and PsaA were analyzed via Western blotting, as indicated in Materials and Methods. As a control, whole-cell lysates of the ΔpsaEF mutant grown at 37°C in BHI broth buffered to pH 6.3 were also analyzed. Prior to probing with antibody, a PVDF membrane was stained with Ponceau S to assess loading (LC, loading control). At least three independent experiments were performed. The data presented are from a representative experiment.
pH (27). While transcription of psaE and psaF in Y. pseudotuberculosis occurred under all growth conditions tested, PsaE and/or PsaF protein levels were not determined (27). Therefore, to investigate if the levels of the PsaE and PsaF proteins were impacted by temperature and/or pH, antibodies were generated against PsaE and PsaF, and wholecell lysates of the WT grown under inducing and noninducing conditions were analyzed by Western blotting (Fig. 3C). PsaE and PsaF were detected only in samples grown at 37°C/pH 6.3 and corresponded with the detection of PsaA. Although the expression of psaEF-gfp at 37°C/pH 6.3 was similar to that at pH 7.3, PsaE levels were consistently lower at 37°C/pH 7.3, and PsaF was never detected at 37°C/pH 7.3. Furthermore, neither PsaE nor PsaF was detected at 26°C at either pH. Thus, the production of PsaE and PsaF appears to be regulated by posttranscriptional mechanisms, and the mechanisms appear to impact PsaE and PsaF differently in response to temperature and pH. Since the detection of PsaA requires PsaE and PsaF, these mechanisms indirectly impact downstream PsaA production. Temperature-dependent translation of psaE is mediated by the psaE 5= UTR. The absence of PsaE and PsaF at 26°C, despite the high level of transcription, suggested that temperature may regulate the translation of psaE and psaF. To investigate this, we generated a psaE native -lacZ translational reporter (pJQ021) containing lacZ fused to the native psaEF promoter, 5= untranslated region (UTR), and psaE start codon, including the first 18 nucleotides of the psaE coding sequence, such that the translation of lacZ was directly dependent on the psaE 5= UTR (Fig. 4A, "native"). This reporter was introduced into the ΔlacZ mutant (YPA87) at the Tn7 site, and this reporter strain (YPA355) was grown under the four conditions tested above. ␤-Galactosidase activity from psaE native -lacZ was significantly higher at 37°C than at 26°C (6-fold), suggesting that  translation initiation occurred more readily at 37°C (Fig. 4B). At both temperatures, ␤-galactosidase activity was only moderately lower at pH 7.3 than at pH 6.3, so translation initiation did not appear to be affected by pH. These results suggest that the psaE 5= UTR mediates temperature-dependent translation. Recent analysis of the RNA "structurome" in Y. pseudotuberculosis suggests the existence of numerous RNA thermometers mediating temperature-dependent translation (32). The transcription start site of psaEF mRNA has been mapped (33), and within the psaEF 5= UTR, there is a string of uridine residues (ϩ7 to ϩ11) that resemble a fourU RNA thermometer (34). FourU thermometers regulate the translation of downstream genes by modulating a temperature-responsive mRNA structure formed by imperfect base pairing of uridine residues to residues in the ribosome binding site (RBS) (35), and nucleotide substitutions that disrupt this base pairing can impact the translation of downstream genes (36,37). Thus, we hypothesized that the string of uridine residues within the psaE 5= UTR functioned as an RNA thermometer and that mutating these residues would disrupt the temperature-dependent translation mediated by the psaE 5= UTR. To test this, we generated two additional translational reporters in which the string of uridine residues in the psaE 5= UTR was altered. In one construct, the nucleotides were all changed to guanine (5G), which would potentially prevent base pairing with guanine residues in the RBS (pJQ028) (Fig. 4A and 5G). In the second construct, the nucleotides were all changed to cytosine (5C), which would potentially strengthen base pairing with guanine residues in the RBS (pJQ027) (Fig. 4A and 5C). These reporters were integrated into the ΔlacZ mutant at the Tn7 site to generate psaE 5G -lacZ (YPA359) and psaE 5C -lacZ (YPA357) reporter strains, respectively. These strains were grown under the four growth conditions as described before, and ␤-galactosidase activity was measured (Fig. 4B). The activity of psaE 5G -lacZ was similar at both 37°C and 26°C, and expression of psaE 5G -lacZ was significantly higher than psaE native -lacZ at 26°C. There was only a modest reduction in the expression of psaE 5G -lacZ at pH 7.3 compared to pH 6.3, suggesting that translation of psaE 5G -lacZ occurred under all four growth conditions. Conversely, psaE 5C -lacZ had low ␤-galactosidase activity even at 37°C. These data suggest that the 5= UTR of psaE mediates temperature-dependent translation and that uridine residues within the 5= UTR contribute to this regulatory mechanism.
As the lacZ translational reporters demonstrated that temperature-dependent translation in the psaE 5= UTR sequence is influenced by the uridine residue motif, we wanted to determine if PsaE and PsaF are produced at 26°C when psaE and psaF were encoded downstream of the psaE 5G 5= UTR. To test this, the psaE and psaF genes were fused to the psaE 5G 5= UTR and native promoter and introduced into the ΔpsaEF mutant at the native site on the chromosome, effectively replacing the native psaE 5= UTR, to generate the psaE 5G strain (YPA361). PsaE, PsaF, and PsaA were analyzed by Western blotting following growth of the WT, psaE 5G mutant, and psaE 5C mutant under the four growth conditions (Fig. 4C). At 26°C/pH 6.3, where PsaE is not normally detected in WT, low levels of PsaE were detected in the psaE 5G mutant, indicating that PsaE can be produced at 26°C. However, PsaF and PsaA could still be detected only in cultures of the psaE 5G mutant grown at 37°C/pH 6.3, indicating that synthesis of PsaE at 26°C was not sufficient for production of PsaF or PsaA at 26°C. Notably, PsaE levels were relatively low under all conditions in the psaE 5G mutant, indicating that the stability of PsaE may be altered, even at 37°C/pH 6.3. Curiously, the levels of PsaF (and PsaA) appeared to be moderately low in the psaE 5G mutant (relative to psaE native ) at 37°C/pH 6.3, corresponding with the slightly lower level of PsaE. No PsaE, PsaF, or PsaA was detected under any of the same four growth conditions in the psaE 5C mutant, indicating that PsaE, PsaF, and PsaA were not produced under any growth conditions in this mutant. Taken together, these data suggest that the 5= UTR of psaE regulates the translation of psaE in response to temperature, but PsaF is influenced by temperature and pH through additional mechanisms.
Translation of psaF is temperature dependent. To determine if the production of PsaF was affected by temperature or pH independently of the psaE 5= UTR, the psaEF coding sequences were cloned into an inducible expression construct such that the expression of psaEF could be induced by the addition of anhydrotetracycline (ATc). Importantly, the psaE 5= UTR was not present upstream of psaE in this construct and thus would not influence PsaE or PsaF production. This plasmid (pPsaEF) was introduced into the ΔpsaEF mutant; this strain was grown under all four growth conditions in the presence and absence of ATc, and cell lysates were analyzed by Western blotting (Fig. 5A). PsaE was detected under all four growth conditions when ATc was added, even at low temperature. Yet, despite the presence of PsaE under all four conditions, PsaF was detectable only in samples grown at 37°C/pH 6.3, as seen in the psaE 5G mutant. These data indicate that the regulation of PsaF production by temperature and pH occurs through a mechanism distinct from the psaE 5= UTR.
While it was possible that the translation of psaF was regulated by temperature and/or pH, it was also possible that the stability of PsaF was impacted by these signals.
To determine whether translation initiation of psaF was affected by temperature and/or pH, we made a psaF up -lacZ translational reporter (pJQ043) containing lacZ fused to sequence upstream of psaF (psaE coding sequence), the psaF start codon including the first 30 nucleotides of the psaF coding sequence, and the native psaEF promoter (up to the psaEF transcription start site) (Fig. 5B). Importantly, this expression construct lacks the native psaE 5= UTR (including the psaE RBS) and psaE start codon; therefore, the translation of lacZ is directly dependent upon sequences upstream of the psaF start codon. Since the psaEF promoter was active under all growth conditions tested, we reasoned that any significant differences in ␤-galactosidase activity would be dependent on the translation of psaF. This reporter was introduced into the ΔlacZ mutant at the Tn7 site to generate the psaF up -lacZ strain (YPA424). This strain was grown under the four growth conditions as described above, and ␤-galactosidase activity was measured (Fig. 5C). ␤-Galactosidase activity was higher (8-fold) at 37°C than at 26°C, and while there was a slight difference in expression between pH 6.3 and pH 7.3, translation did not appear to be largely impacted by pH. Thus, the translation of psaF requires high temperature and this regulation is mediated by the sequence upstream of psaF. However, control of PsaF production in response to temperature did not account for observed differences in PsaF levels in response to pH.
PsaF influences psaA transcription by affecting PsaE levels. While PsaE is predicted to be the direct transcriptional activator of psaA transcription, the role of PsaF is less clear. While psaF is required for psaA transcription ( Fig. 2A), in the psaE 5G mutant (where psaF is present but PsaF is not produced at 26°C/pH 6.3), PsaA was not detected in samples grown at 26°C/pH 6.3 despite low levels of PsaE. These data suggest that the production of PsaF, in addition to PsaE, is required for psaA expression. To further dissect this phenomenon and investigate the role of PsaF in the regulation of psaA transcription, the psaA-gfp plasmid was introduced in the psaE native (YPA260), psaE 5G (YPA361), and psaE 5C (YPA360) strains, and expression was analyzed after growth at 37°C and 26°C in BHI broth buffered to pH 6.3 (Fig. 6A). High levels of psaA-gfp expression occurred in the psaE native and psaE 5G strains at 37°C/pH 6.3, while no expression occurred in the psaE 5C mutant. These data correspond with the absence of PsaE, PsaF, and PsaA in the psaE 5C mutant and the presence of these proteins in the psaE native and psaE 5G strains under this growth condition. The expression of psaA-gfp was slightly lower in the psaE 5G strain than in the psaE native strain, corresponding with the slightly lower levels of PsaA previously noted in the psaE 5G strain (Fig. 4C). Conversely, minimal psaA-gfp expression occurred in the psaE native and psaE 5G strains at 26°C/pH 6.3, while expression was not detected in the psaE 5C strain. Despite psaA-gfp expression being low, expression in the psaE 5G strain was 2-fold higher than in the psaE native strain.  This moderate increase in psaA-gfp expression corresponded with the presence of low levels of PsaE in the psaE 5G strain and the absence of PsaE in the psaE native strain under these growth conditions (Fig. 4C). These data further support the hypothesis that PsaE acts as a direct transcriptional activator of psaA transcription; however, the presence of PsaE alone is not sufficient for high-level expression of psaA, indicating that PsaF is required for maximal activation of psaA transcription. The topologies of PsaE and PsaF resemble those of the ToxR-ToxS and TcpP-TcpH regulatory protein pairs in Vibrio cholerae (27,29,30,38). ToxS and TcpH enhance the stability of ToxR and TcpP, respectively (39,40), and since PsaE levels are reduced in the absence of PsaF, we wondered if PsaF influenced PsaE stability. To test this, we constructed a plasmid expressing psaF under the control of the psaEF promoter and the psaF 5= UTR (pPsaF; same sequence driving the psaF up -lacZ reporter in Fig. 5B). This plasmid and a vector control plasmid were introduced into the strain expressing only psaE at the native site (YPA265), and PsaE levels were analyzed in this strain grown at 37°C/pH 6.3 (Fig. 6B). As previously noted, when both psaE and psaF are present, PsaE was detected at high levels. However, in the absence of psaF, PsaE was detected only at low levels, suggesting that PsaF plays a role in stabilizing PsaE. When pPsaF was introduced into YPA265, generating a strain that contains psaE on the chromosome and psaF on a plasmid, higher levels of PsaE were detected than in the vector control strain lacking pPsaF. Curiously, lower levels of PsaE were detected when psaF was expressed in trans than with native chromosomal expression, thus suggesting that cotranscription of psaE and psaF may be required for maximum levels of PsaE. Despite the slight reduction from trans complementation of psaF (compared to WT), these data indicate that PsaF increases levels of PsaE, and this likely occurs through enhanced stability.

DISCUSSION
As a vector-borne pathogen, Y. pestis moves between a flea vector and mammalian host and must use environmental signals as cues to regulate the expression of key virulence determinants that aid in immune evasion and survival within the host tissue. The Y. pestis pH 6 antigen (PsaA) is a virulence factor with an unusual expression pattern that requires high temperature and low pH (15,(17)(18)(19)23). It is well established that the combination of high temperature and acidic pH is required for psaA transcription and PsaA production in Y. pestis, but the underlying mechanisms have remained elusive (23,24). Here, we show that both psaA transcription and PsaA production occur rapidly when Y. pestis is grown at high temperature in medium buffered to low pH (37°C/pH 6.3). By utilizing buffered medium for growth, we ruled out the possibility that expression of psaA is impacted by the growth phase of the bacteria rather than by low pH. We also investigated the roles of PsaE and PsaF, key transcriptional regulators of psaA, to understand how temperature and pH influence psaA expression. We propose a model in which temperature and pH impact PsaE and PsaF levels and, thus, the expression of psaA (Fig. 7).
Consistent with findings in Y. pseudotuberculosis (27), we show that both psaE and psaF are required for psaA transcription in Y. pestis and that the transcription of psaE and psaF, unlike psaA, is not affected by pH and only moderately impacted by temperature. These results support previous speculation that the function and/or levels of PsaE and PsaF are subject to posttranscriptional regulation in Y. pseudotuberculosis (27) and led us to investigate how temperature and pH influence PsaE and PsaF in Y. pestis. Prior to this study, the direct detection of PsaE and PsaF had not been reported. By generating antibodies that recognize endogenous PsaE and PsaF protein, we showed that the levels of both PsaE and PsaF are influenced by temperature and pH. Neither PsaE nor PsaF was detected after growth at low temperature (26°C), corresponding with the absence of psaA transcription and, thus, PsaA. Intriguingly, PsaE and PsaF appear to have different sensitivities to pH; PsaE is detectable at low levels at 37°C/pH 7.3, whereas we are unable to detect PsaF at 37°C/pH 7.3. Reduced levels of PsaE (and absence of PsaF) at 37°C/pH 7.3 also correspond with the absence of both the PsaA protein and psaA promoter activity. Since the presence of PsaA at 37°C/pH 6.3 corre-sponds with the presence of both PsaE and PsaF, understanding how temperature and pH impact PsaE and PsaF levels became the focus of this study.
Our data suggest that the translation of both psaE and psaF is temperature dependent and regulated by sequences upstream of each gene. Our finding that the 5= UTR of psaE regulates translation initiation of psaE in response to a temperature upshift resembles the recently described mechanism regulating temperature-dependent synthesis of LcrF, a major virulence regulator of the type III secretion system (T3SS) in Yersinia spp. (37,41). Similar to our findings with PsaE/PsaF, the synthesis of LcrF in Y. pestis is significantly increased at high temperature and corresponds with the expression of type III secretion system (T3SS) genes regulated by LcrF (42,43). In Y. pseudotuberculosis, lcrF translation was shown to be regulated by a fourU RNA thermometer located in the mRNA upstream of lcrF (37). Initially characterized for the regulation of bacterial heat shock genes in response to temperature (44), recently, RNA thermometers have been identified as a mechanism to regulate the expression of virulence genes in pathogenic bacteria (36,37,45,46). Additionally, a recent study of the structurome in Y. pseudotuberculosis suggests that these elements may be extensively utilized by Yersinia spp. (32). By mutating the uridine residues within the psaE 5= UTR, we were able to overcome the thermal regulation of psaE translation to inhibit the detection of PsaE at 37°C (psaE 5C ) and detect low levels of PsaE at 26°C (psaE 5G ). Yet, despite the presence of PsaE at 26°C in the psaE 5G mutant, the detection of PsaF and PsaA remains limited to 37°C/pH 6.3. When psaEF was expressed using an inducible expression construct (pPsaEF; YPA366), in which psaE is expressed without the native psaE 5= UTR, PsaE was present at high levels at 37°C and 26°C, but PsaF was present only at 37°C/pH 6.3. Thus, the native psaE 5= UTR imparts thermal regulation on psaE translation, which appears to explain why PsaE is absent at 26°C in the WT.
Differences in the activity of the psaF up -lacZ translational reporter at 37°C and 26°C indicated that the sequence upstream of psaF (psaE coding sequence) imparts temperature-dependent regulation on psaF translation. This additional layer of thermal regulation likely explains why PsaF remains absent in the psaE 5G mutant at 26°C, despite the ability to induce the synthesis of PsaE. The discrete regulation of both PsaE and PsaF synthesis by high temperature reveals that Y. pestis incorporates multiple mechanisms to ensure that the production of PsaA is specific for high temperature. Similarly, when thermal regulation of lcrF translation was overcome by disrupting the  RNA thermometer upstream of lcrF in Y. pseudotuberculosis, downstream Yop production remained specific for high temperature, despite detectable LcrF at low temperature (37), thus suggesting the presence of multiple layers of regulation. While it is possible that the uridine residues (ϩ7 to ϩ11) in the psaE mRNA 5= UTR function as a fourU RNA thermometer, additional studies will be needed to determine if the psaE 5= UTR influences folding of psaEF mRNA in response to temperature in a manner similar to characterized fourU RNA thermometers (34). The topologies of PsaE and PsaF are similar to that of the ToxR/ToxS-like family of transcriptional regulatory protein pairs in V. cholerae (29,38). Members of this protein family serve as key regulators of virulence gene expression (28-31, 39, 40, 47-51), and our data indicate that PsaE and PsaF are key regulators of psaA transcription in Y. pestis. ToxR-like proteins are predicted to sense environmental signals and influence downstream gene expression, as both sensing (periplasmic) and DNA-binding (cytoplasmic) domains are contained within a single protein. However, the mechanisms by which ToxR-like proteins integrate signal sensing and downstream gene activation are not well understood. The membrane spanning topology of ToxR-like proteins offers an unusual potential to function as a one-component signal transduction system, as seen with the acid-sensing regulator CadC (52). Intriguingly, the stability and/or function of other ToxR-like family proteins, such as TcpP, and ToxR itself, are affected by a second effector protein (TcpH and ToxS, respectively) (31,39,40,49,53,54). Characteristic of ToxR-like proteins, the N-terminal cytoplasmic domain of PsaE contains a winged helix-turn-helix DNA-binding motif resembling the OmpR family of response regulators (16). While an interaction of PsaE with the psaA promoter has not been demonstrated, it is predicted that PsaE directly activates psaA transcription. However, as PsaA is not detected unless both PsaE and PsaF are present, PsaE alone is not sufficient to activate psaA transcription. Thus, like ToxR and TcpP, PsaE requires a partner protein to regulate gene expression.
The N terminus of PsaF is thought to be anchored to the inner membrane (27), and therefore, PsaF is likely not directly involved in binding to psaA promoter DNA. We found that PsaE levels are influenced by PsaF; thus, it is plausible that the primary role of PsaF is to enhance PsaE stability to allow for psaA transcription. In support of this, high levels of PsaE and PsaA correspond with the presence of PsaF. However, there is evidence to suggest that PsaF also influences PsaE function. When the WT is grown at 37°C/pH 7.3, low levels of PsaE are detected, but PsaF is not detected. Under these conditions, the level of psaA expression is just as low as in the ΔpsaEF mutant, further indicating that PsaE alone cannot activate psaA transcription. These data clearly indicate that both PsaE and PsaF are required for the expression of psaA and suggest that the role of PsaF may be to promote the stability and/or function of PsaE.
The role of pH as an environmental signal for the expression of psaA remains unclear, as it is not known when Y. pestis encounters an acidic pH during mammalian infection. While it has been speculated that a macrophage phagosome may provide the low-pH environment that is necessary for psaA transcription (16), the expression of psaA in host cells during a mammalian infection has not been demonstrated. Strikingly, little is known about mechanisms that bacteria utilize to sense pH and influence gene expression. Our work suggests that Y. pestis may utilize PsaE and PsaF to sense changes in pH to regulate the expression of psaA. Despite thermal regulation of psaE and psaF translation, the translation of psaE and psaF is not influenced by changes in pH. However, the levels of both PsaE and PsaF do show a significant change in response to pH, suggesting that there are additional posttranslational pH-dependent mechanisms regulating PsaE and PsaF. Since PsaE levels are affected by PsaF, it is tempting to speculate that PsaF contributes to pH-dependent stability of both proteins. To address this, we are currently investigating the mechanisms by which pH regulates PsaE and PsaF. Together, our data suggest that Y. pestis utilizes temperature and pH to influence levels of PsaE and PsaF, two key transcriptional regulatory proteins of psaA in Y. pestis. The temperature-dependent regulation of psaE and psaF translation, in addition to the regulation of PsaE and PsaF by pH, allows Y. pestis to precisely control the expression of psaA through multiple environmental signals.

MATERIALS AND METHODS
Bacterial strains and growth conditions. All bacterial strains and plasmids used in this study are listed in Table 1. Y. pestis CO92/pCD1 Ϫ (YP6) was cultivated on brain heart infusion (BHI) agar (BD Biosciences, Bedford, MA) at 26°C for 48 h and in BHI broth cultures grown with aeration at 26°C or 37°C. E. coli strains were cultivated on Luria-Bertani (LB) agar (BD Biosciences) at 37°C overnight and in liquid cultures with aeration at 37°C or 26°C. When indicated, bacteria were grown in BHI broth that was adjusted and buffered to the appropriate pH. BHI broth was buffered with 100 mM MES [2-(Nmorpholino)ethanesulfonic acid; Sigma] and then adjusted to pH 6.3 or 6.7, or it was buffered with 100 mM MOPS [3-(N-morpholino)propanesulfonic acid; Fisher Scientific] and then adjusted to pH 7.3 and filter sterilized. When necessary, antibiotics were added to the growth medium at the following concentrations: kanamycin (Kan), 50 g/ml; carbenicillin (Carb), 100 g/ml; and irgasan (Irg), 2 g/ml. For the expression of genes cloned into pMWO-005, 50 ng/ml anhydrous tetracycline (ATc) was added to the liquid medium when strains were subcultured.
Plasmid and strain construction. All primers used in this study are listed in Table 2. The ΔpsaEF mutant and all mutants in which psaE and/or psaF were introduced at the native site in the ΔpsaEF mutant were constructed via allelic exchange using the pSR47S suicide vector (55). All plasmids were constructed via Gibson Assembly (NEB), unless otherwise described, and were confirmed by sequencing.
(i) psaEF deletion and complementation. The plasmid for generating an in-frame deletion of psaEF was constructed by amplifying ϳ500-bp DNA fragments upstream and downstream of psaE and psaF, respectively. These fragments were digested and cloned into pSR47S to generate pJC306. This plasmid was introduced into YP6 via conjugation, essentially as described previously (17). Briefly, transconjugants were selected on BHI plates with 50 g/ml Kan (Kan 50 ) and 2 g/ml Irg (Irg 2 ). The second recombination event was selected for by streaking colonies resistant to Kan 50 and Irg 2 onto BHI agar plates containing psaEF promoter fused to psaF upstream sequence into pJQ003 This work 5% sucrose. PCR was performed on candidate colonies to identify those with the deletion of psaEF. A single clone was selected for experimentation and named YPA18. The plasmid used to introduce psaE and psaF at the native site on the chromosome was constructed as follows. The psaEF coding sequence including 500-bp upstream and downstream flanking sequences was amplified and cloned into pSR47S to generate pEW104. The plasmids expressing only psaE (pEW105) or only psaF (pEW106) were similarly constructed. For pEW105, the psaEF promoter region and psaE coding sequence were amplified. For pEW106, the psaF coding sequence and the psaEF promoter sequence were amplified separately and cloned into pSR47S. These plasmids were introduced into YPA18 (ΔpsaEF) via conjugation and subjected to the same selection procedure used to generate deletion strains. The resulting strains are YPA260 (pEW104, psaEF ϩ ), YPA265 (pEW105, psaE ϩ only), and YPA279 (pEW106, psaF ϩ only).
Plasmids containing the psaE translational fusions to lacZ were each made by amplifying the native psaEF promoter and the native or mutant psaE 5= UTR (5G or 5C) and cloned into pJQ003, such that the native promoter drives the expression of the 5= UTR and gene fusion. The psaEF promoter and native psaE 5= UTR (no substitutions) were amplified as a single fragment that was cloned into pJQ003, generating pJQ021 (psaE native -lacZ). The 5C or 5G nucleotide mutations were introduced into the psaE 5= UTR by first amplifying the UTR with primers containing the substituted nucleotides. These products, along with the native psaEF promoter fragment (up to the transcriptional start site), were then cloned a Restriction sites are in bold. Sequence overlaps for Gibson Assembly cloning are underlined. b How the primer was used for cloning. The primer was used to construct the indicated plasmid or construct. F, forward primer; R, reverse primer.
into pJQ003 to generate pJQ027 (psaE 5C -lacZ) and pJQ028 (psaE 5G -lacZ), respectively. The psaF translational reporter plasmid was constructed similarly. A fragment spanning ϳ800 bp upstream of the psaF start codon (within the psaE gene) was amplified and cloned along with the same native psaEF promoter fragment used for the psaE fusions into pJQ003 to generate pJQ043 (psaF up -lacZ). All lacZ reporter plasmids were electroporated into E. coli S17-1 pir and then introduced into YPA87 (ΔlacZ) via conjugation using a triparental mating with E. coli containing the pTNS-2 helper plasmid to mediate integration at the Tn7 site on the chromosome (58). Transconjugates were selected on BHI plates with Kan 50 /Irg 2 and were then patched onto BHI Kan 50 and BHI Carb 100 plates to identify clones that no longer contained pTNS-2.
(iv) psaE 5= UTR mutants at the native site. A similar set of plasmids with altered psaE 5= UTR sequences, as described above, was constructed for chromosomal complementation. These plasmids contain the native psaEF promoter and either the 5C and 5G psaE 5= UTR and were amplified as described above with primers designed for Gibson Assembly into pSR47S. The resulting plasmids, pJQ029 and pJQ030, respectively, were introduced into YPA18 via conjugation and selected for growth on BHI plates with Kan 50 and Irg 2 . The resulting complemented strains are YPA360 (psaE 5C ) and YPA361 (psaE 5G ).
(v) trans complementation. The plasmid for expressing psaEF from the tet promoter was constructed by amplifying the psaEF coding sequence and cloning into pMWO-005 (60). pMWO-005 contains the tet operator/promoter and RBS for inducible transcription and efficient translation. The resulting plasmid, pPsaEF, was introduced into YPA18 via electroporation.
The plasmid for complementation of psaF in trans was constructed as follows. The psaEF promoter (up to the ϩ1 site) and a DNA fragment extending ϳ800 bp upstream of psaF and ϳ500 bp downstream were amplified, digested with SalI and BamHI, and cloned into pWKS30 (61) via Gibson Assembly. The resulting plasmid, pPsaF, was introduced into YPA265 via electroporation.
gfp transcriptional reporter assay. To analyze promoter activity, gfp transcriptional reporter plasmids were introduced in Y. pestis CO92 strains via electroporation. Saturated cultures were subcultured to an optical density at 600 nm (OD 600 ) of 0.2 in unbuffered BHI broth or BHI broth buffered to pH 6.3, 6.7, or 7.3 and grown for 8 h with aeration at 26°C or 37°C. Relative fluorescent units (RFU) from each sample were measured using a Synergy HT microplate reader (BioTek Instruments, Winooski, VT) and normalized to the OD 600 to determine the RFU per OD 600 (RFU/OD 600 ). For the gfp time course in Fig. 1A and B, strains were subcultured as described above into 30 ml unbuffered BHI broth and grown at 37°C or 26°C for 10 h with aeration. Every 2 h, a 2-ml aliquot was removed for OD 600 and RFU measurements, and then the cells were removed by centrifugation to determine the pH of the supernatant.
␤-Galactosidase assays. Saturated cultures of Y. pestis strains containing translational reporters grown in unbuffered BHI broth at 26°C were subcultured to an OD 600 of 0.2 in BHI broth buffered to pH 6.3 or 7.3 and grown for 8 h at 26°C or 37°C. Assays were performed as previously described (62).
RNA isolation and RT-PCR. Saturated cultures of YP6 were subcultured to an OD 600 of 0.2 in BHI broth buffered to pH 6.3 or pH 7.3 and grown at 37°C or 26°C. After 8 h, 10 OD 600 of cells were pelleted and resuspended in 1 ml TRIzol reagent (Sigma). RNA was extracted and treated with DNase I according to the manufacturer's (Sigma) instructions. Using 2 g of RNA for the template, cDNA synthesis was performed with SuperScript III (Life Technologies), following the manufacturer's protocol. RT-PCR was performed using cDNA as the template and primers for psaE and psaF. DNase I-treated RNA that was not treated with reverse transcriptase served as a control for DNA contamination, and YP6 genomic DNA (gDNA) was used as a positive PCR control. RT-PCR products were separated on a 1% agarose gel and visualized by staining with GelRed nucleic acid stain (Biotium).
Peptide synthesis and antibody production. To generate antibodies that recognize PsaE and PsaF, two PsaE and PsaF peptide fragments containing keyhole limpet hemocyanin (KLH) conjugations were synthesized by LifeTein (Somerset, NJ). The synthesized peptides were sent to Cocalico Biologicals, Inc. (Stevens, PA), and both peptide fragments were used for immunization of New Zealand White rabbits, following standard protocols. To generate an antibody against PsaA, a His-PsaA fusion protein was expressed in E. coli, purified, and sent to Covance Research Products, Inc. (Denver, PA) for immunization of a New Zealand White rabbit, following standard protocols.
Western blot analysis. Saturated cultures of indicated strains were subcultured to an OD 600 of 0.2 in BHI broth buffered to pH 6.3 or 7.3 and grown for 8 h at 26°C or 37°C. Whole-cell lysates were prepared from 1.5 ml of cells that were pelleted, washed once with ice-cold phosphate-buffered saline (PBS), and resuspended in Laemmli buffer containing 5% ␤-mercaptoethanol. Samples were boiled for 10 min, and 0.2 OD 600 separated via SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membrane for Western blot analysis. Loading was qualitatively assessed by Ponceau S staining of the PVDF membrane. Anti-PsaE, anti-PsaF, and anti-PsaA sera were used to probe for PsaE, PsaF, and PsaA, respectively. Prior to use, the anti-PsaE serum was adsorbed against Y. pestis ΔpsaEF mutant lysates and was used at a titer of 1:100. Anti-PsaF serum was used at a titer of 1:1,000. Anti-PsaA serum was absorbed against E. coli lysates and used at a titer of 1:2,500. Anti-IgG horseradish peroxidase (HRP)-conjugated secondary antibodies were used at a titer of 1:20,000.

SUPPLEMENTAL MATERIAL
Supplemental material for this article may be found at https://doi.org/10.1128/JB .00217-19.