PsaF Is a Membrane-Localized pH Sensor That Regulates psaA Expression in Yersinia pestis

ABSTRACT The Yersinia pestis pH 6 antigen (PsaA) forms fimbria-like structures and is required for full virulence during bubonic plague. High temperature and low pH regulate PsaA production, and while recent work has uncovered the molecular aspects of temperature control, the mechanisms underlying this unusual regulation by pH are poorly understood. Using defined growth conditions, we recently showed that high levels of PsaE and PsaF (two regulatory proteins required for expression of psaA) are present at mildly acidic pH, but these levels are greatly reduced at neutral pH, resulting in low psaA expression. In prior work, the use of translational reporters suggested that pH had no impact on translation of psaE and psaF, but rather affected protein stability of PsaE and/or PsaF. Here, we investigated the pH-dependent posttranslational mechanisms predicted to regulate PsaE and PsaF stability. Using antibodies that recognize the endogenous proteins, we showed that the amount of PsaE and PsaF is defined by a distinct pH threshold. Analysis of histidine residues in the periplasmic domain of PsaF suggested that it functions as a pH sensor and indicated that the presence of PsaF is important for PsaE stability. At neutral pH, when PsaF is absent, PsaE appears to be targeted for proteolytic degradation by regulated intramembrane proteolysis. Together, our work shows that Y. pestis utilizes PsaF as a pH sensor to control psaA expression by enhancing the stability of PsaE, an essential psaA regulatory protein. IMPORTANCE Yersinia pestis is a bacterial pathogen that causes bubonic plague in humans. As Y. pestis cycles between fleas and mammals, it senses the environment within each host to appropriately control gene expression. PsaA is a protein that forms fimbria-like structures and is required for virulence. High temperature and low pH together stimulate psaA transcription by increasing the levels of two essential integral membrane regulators, PsaE and PsaF. Histidine residues in the PsaF periplasmic domain enable it to function as a pH sensor. In the absence of PsaF, PsaE (a DNA-binding protein) appears to be targeted for proteolytic degradation, thus preventing expression of psaA. This work offers insight into the mechanisms that bacteria use to sense pH and control virulence gene expression.


RESULTS
A distinct pH threshold controls the levels of PsaE and PsaF. We previously reported that high levels of PsaE and PsaF are detected at 37°C and pH 6.3 and that both proteins are greatly reduced at 37°C and pH 7.3 (18). Translation of psaE and psaF is not impacted by pH (18), and thus it is likely that posttranslational mechanisms are involved. To begin to address how PsaE and PsaF are affected by pH, Y. pestis strain CO92 cured of the virulence plasmid pCD1 (YP6; here referred to as the wild type [WT]) was grown at 37°C in brain heart infusion (BHI) buffered to pH 6.3, 6.5, 6.7, 7.0, or 7.3. Consistent with our previous findings (18), high levels of PsaE and PsaF were detected at pH 6.3, and levels were much lower at pH 7.3 (Fig. 1A). While both proteins were readily detected at pH 6.5 and pH 6.7, PsaF was undetectable at a pH of .6.7. PsaE levels were noticeably reduced at pH 6.7 and were very low at pH 7 and 7.3. These data confirm that pH impacts PsaE and PsaF levels and reveal a distinct threshold between pH 6.7 and 7 that defines the levels of these proteins.
Because pH does not appear to affect translation of psaF (18), we hypothesized that pH affects the stability of PsaF. To test this, the WT strain was first grown at 37°C pH 6.3 to the late exponential phase to allow production of PsaE and PsaF. Bacterial cells were then pelleted and suspended in phosphate-buffered saline (PBS) (a nutrient-limiting condition) at pH 6.3 or pH 7.3, and PsaF levels were monitored over time (Fig. 1B). At pH 7.3 in PBS, PsaF levels were reduced within 2 h, whereas a similar change in PsaF levels was not observed at pH 6.3 in PBS. A comparable number of bacteria were recovered after 2 h from both conditions, and this was comparable to the number of bacteria present when they were initially suspended in PBS (data not shown), indicating that cell viability did not contribute to the changes in PsaF levels at pH 7.3. In contrast, once PsaE was produced during growth at pH 6.3, its stability was minimally affected by the shift to pH 7.3 in PBS (Fig. 1B). These data suggest that PsaF is less stable at pH 7.3 than at pH 6.3, but PsaE is relatively stable at both pH levels under these conditions. The relative stability of PsaE at pH 7.3 could be explained by low levels of PsaF being sufficient to stabilize PsaE, or it could indicate that, once produced, PsaE is stable under these conditions.
DegP and Tsp influence levels of PsaE. As with PsaF, translation of psaE is not affected by pH (18), and we thus predicted that during bacterial growth, pH may also control the stability of PsaE. Regulated intramembrane proteolysis (RIP) is a two-step degradation of membrane proteins that occurs in most domains of life, and recent work shows that bacteria utilize RIP to control virulence gene expression (40,41). During RIP, a target protein is sequentially cleaved, first by a periplasmic protease (site 1 protease) and then by an integral membrane protease (site 2 protease), resulting in complete degradation of the target protein. The PsaE-like proteins, TcpP and ToxR, are degraded via RIP (30,38,39), and ToxS and TcpH (PsaF-like proteins) inhibit degradation of ToxR and TcpP, respectively (29,31). Therefore, we wanted to determine if PsaE also could The wild type (WT; YP6) was grown at 37°C in brain heart infusion (BHI) buffered to pH 6.3, 6.5, 6.7, 7.0, and 7.3, and PsaE and PsaF were analyzed via Western blot as described in Materials and Methods. As a control, whole-cell lysates of the DpsaEF mutant (YPA18) grown at 37°C in BHI pH 6.3 were also analyzed. (B) WT was grown in BHI pH 6.3 for 8 h, cells were pelleted, washed and suspended in phosphate-buffered saline (PBS) adjusted to pH 6.3 or pH 7.3 as indicated in Materials and Methods, and PsaE and PsaF were analyzed over time. LC, loading control from Ponceau S-stained membrane.
be subjected to proteolytic degradation via RIP. Tsp and YaeL (RseP) are the site 1 and site 2 proteases, respectively, that degrade TcpP in Vibrio cholerae, and since deletion of tsp and/or yaeL restores the ability of TcpP to activate transcription of toxT (38, 39), we speculated that preventing degradation of PsaE at pH 7.3 would restore transcription of psaA. To test this, genes predicted to encode periplasmic (site 1; tsp, degS, degQ, and degP) and membrane (site 2; rseP, ypfJ, ypo0398, and ftsH) proteases were mutated and psaA expression was analyzed after growth at 37°C and pH 6.3 or pH 7.3. The insertion mutations in degS, degQ, ypfJ, ypo0398, and ftsH did not impact psaA expression, suggesting that the proteases encoded by these genes are not required for degradation of PsaE. However, we cannot exclude the possibility that these insertions did not completely disrupt gene function.
Of the eight mutants tested, only the tsp and degP insertion mutants exhibited increased psaA expression at pH 7.3 relative to that of the WT strain ( Fig. 2A). Mutations in the tsp and degP genes largely attenuated Y. pestis growth, indicating these proteases play a critical role in Y. pestis physiology. Despite the lower growth rate of these mutants, these data suggest that DegP and Tsp individually contribute to low psaA expression under this noninducing condition. The rseP insertion mutant had  reduced psaA expression at pH 6.3 and was severely attenuated for growth on agar plates and in broth medium. We were unable to generate an DrseP mutant, which is consistent with previous reports that rseP is an essential gene in Y. pestis (42,43). Thus, it is difficult to determine if RseP plays a role in PsaE degradation or if the effect on psaA expression of the rseP insertion mutant is a consequence of the growth defect. The increase in psaA expression in the degP and tsp mutants at 37°C and pH 7.3 suggested that PsaE levels and/or activity may be partially restored by loss of these site 1 proteases. As insertion disruption mutations can have polar effects on downstream genes, we constructed individual degP and tsp deletion mutants (DdegP and Dtsp).
These deletion mutants were then tested to determine if the altered psaA expression at pH 7.3 was associated with changes in PsaE levels. The WT and mutant strains were grown at pH 6.3 and pH 7.3, and PsaE was analyzed via Western blot. The amount of PsaE present at pH 7.3 was elevated in both the DdegP and Dtsp mutants relative to that in WT, suggesting that increases in psaA transcription were due to an increase in PsaE levels ( Fig. 2B and C). Yet, PsaF remained undetectable in both mutants at pH 7.3, indicating that Tsp and DegP are not individually responsible for low levels of PsaF under this condition (see Fig. S1 in the supplemental material).
To validate the impact of the predicted site 1 proteases on psaA expression, strains with degP or tsp restored were generated by reintroducing degP (YPA476) or tsp (YPA438) at the native site of the DdegP and Dtsp mutants, respectively. As seen with the insertion mutants, expression of psaA in both the Dtsp and DdegP mutants was higher relative to that in the WT at pH 7.3, suggesting that the absence of these proteases is sufficient to partially restore PsaE activity at pH 7.3 ( Fig. 2D and E). Reintroduction of tsp and degP into the Dtsp and DdegP mutants resulted in psaA expression comparable to that of the WT at 37°C pH 7.3, indicating that the observed effect of the Dtsp and DdegP mutations was due to the loss of Tsp and DegP, respectively, rather than due to secondary site mutations. While expression of psaA increased in the DdegP mutant (4-fold) and in the Dtsp mutant (5-fold) relative to that in the WT at pH 7.3, expression levels remained low relative to those at pH 6.3. It is possible that this is due to the potential redundancy in Tsp and DegP activity. Our attempt to generate a DdegP Dtsp double mutant to examine this was unsuccessful, likely due to the significant impacts on growth observed with the individual mutants. Under the growth conditions used for these assays, the final optical density at 600 nm (OD 600 ) for the degP and tsp mutants were 17% and 24% that of the WT at pH 7.3, respectively. Together, these data suggest DegP and Tsp individually influence PsaE levels, but it is difficult to determine if these proteases have redundant roles at 37°C and pH 7.3.
Cysteine residues in PsaE are required for PsaE and PsaF stability. While PsaE is thought to directly activate psaA transcription via its DNA-binding domain (37), the role of the periplasmic domain is unknown. Cysteines in the periplasmic domain affect the stability of TcpP and are predicted to impact the function of ToxR (33,(44)(45)(46)(47). Like ToxR and TcpP, PsaE contains two cysteine residues in its periplasmic domain. PsaE variants with individual cysteine-to-serine substitutions (PsaE C206S and PsaE C211S ) and a double cysteine-to-serine substitution (PsaE C206S/C211S ) were constructed. Expression of psaA was similar to that in the DpsaEF mutant in all three PsaE cysteine-to-serine mutants, indicating the importance of these cysteine residues (Fig. 3A).
In V. cholerae, substitution of either cysteine in the TcpP periplasmic domain greatly reduces the stability of both TcpP and TcpH (33). To determine if the cysteine substitutions in PsaE impacted the stability of PsaF, the PsaE cysteine-to-serine mutants were grown at pH 6.3, and PsaF was analyzed via Western blot (Fig. 3B). Strikingly, PsaF was undetectable in all three PsaE mutants. Because the psaE open reading frame overlaps with that of psaF, it is possible that the nucleotide alterations encoding PsaE C206S (25 nucleotides upstream of psaF) or PsaE C211S (10 nucleotides upstream of psaF) disrupted translation of psaF. As PsaF is required for PsaE stability (18), a disruption of psaF translation could lead to an absence of PsaE in strains encoding PsaE cysteine-to-serine mutations. To test this, psaF was expressed in trans (pPsaF) and introduced in the psaE cysteine mutants. Introduction of pPsaF into a DpsaF mutant (psaE 1 ) restores PsaE levels (18); however, the presence of pPsaF was not sufficient to restore detection of PsaE C206S , PsaE C211S , or PsaE C206S/C211S , indicating that reduced translation of psaF in these mutants likely was not responsible for the lack of detectable PsaE (Fig. 3C). Together, these data indicate that both cysteines in the C-terminal periplasmic domain of PsaE impact the stability of both PsaE and PsaF.
PsaF is a pH sensor. The large periplasmic domain of PsaF seems ideal for sensing environmental cues such as extracellular pH. This region contains nine histidine residues. The pK a of histidine imidazole is ;6, and histidine protonation at mildly acidic pH levels can impart a "pH-sensing" role shown to modulate the conformation and/or activity of multiple bacterial two-component sensor kinases and viral membrane proteins (48)(49)(50)(51)(52)(53)(54). The Yersinia enterocolitica PsaF homologue, MyfF, also contains a histidine-rich periplasmic domain, and six of these histidines are conserved in PsaF (Fig. 4A). While the effect of pH on MyfF has not been reported, the expression of myfA (psaA homologue) is elevated at acidic pH (55,56). Thus, we wanted to determine if these histidines drive pH-dependent function and/or stability of PsaF. To test this, psaF mutants encoding individual histidine-to-alanine substitutions were expressed with a wild-type psaE allele on the chromosome at the native site. Individual substitutions were made for the six histidines shared between PsaF and MyfF. Western blot analysis of lysates from the six individual PsaF mutants grown at 37°C and pH 6.3 indicated that only PsaF H40A was readily detected at levels similar to those of WT PsaF (Fig. 4B). Corresponding with the presence of PsaF, psaA was highly expressed only in the WT strain and the mutant PsaF H40A strain (Fig. 4C). A triple His-to-Ala substitution was also made for the three histidines unique to PsaF (PsaF H38A , PsaF H116A , and PsaF H159A ). Of note, this triple His-to-Ala substitution mutant also produced PsaA (see Fig. S2 in the supplemental material), suggesting that H40 and the three histidine residues unique to PsaF do not significantly impact PsaF function under this condition. The other five PsaF mutants (H54A, H87A, H141A, H153A, and H155A) were undetectable at pH 6.3, suggesting these residues are essential for PsaF stability under this condition. Consistent with the absence of PsaF, very low levels of psaA expression were detected in these mutants (Fig. 4C).
Since histidine protonation can influence protein conformation (57), we wanted to determine if the stability of the PsaF histidine substitution mutants was impacted by pH. The same strains described above were grown at 37°C and pH 6.5 or 6.0, and PsaE and PsaF were analyzed by Western blotting. The Y. enterocolitica homologue of psaA, myfA, is only expressed at an even lower pH, pH 5 (56); therefore, we also grew these strains at 37°C and pH 5.5 for Western blot analysis of PsaE and PsaF. None of the PsaF mutants, including PsaF H40A , were detected at pH 6.5, a condition under which WT PsaF is abundant (Fig. 5A)  However, in samples from three of the six mutants (PsaF H40A , PsaF H87A , and PsaF H153A ), PsaF was detected when the pH was lowered to 6.0. Of these mutant proteins, only PsaF H40A was present at pH 6.3, suggesting that both PsaF H87A and PsaF H153A are not stable at pH of .6.0.
When the pH was further acidified to 5.5, all six PsaF mutant proteins were detected (PsaF H54A was detected at a low level), indicating that the stability of PsaF histidine mutants can be rescued by decreasing the pH. The presence of near-WT levels of PsaE correlated with the presence of each PsaF histidine mutant. One exception to this was that at pH 6.0 or 6.5, PsaE appeared to be present at levels comparable to those in the WT in the absence of PsaF H155A . To determine if these PsaF mutant proteins influence psaA expression in response to pH, the WT strain and these psaF histidine mutants were then assayed for psaA expression after growth at 37°C in BHI at pH 6.5, 6.0, and 5.5 (Fig. 5B). Expression of psaA in the WT increased when the pH was more acidic, and this corresponded with an apparent increase in levels of PsaE and PsaF (Fig. 5). Expression of psaA in the PsaF His-to-Ala mutants was roughly comparable to that in the WT when the stability of PsaF was restored by growth at lower pH. Collectively, these data suggest that these histidine residues of PsaF are involved in sensing pH and impact stability of both PsaF and PsaE.

DISCUSSION
PsaA is a bubonic plague virulence factor and is produced when Y. pestis is grown in vitro under high temperature and low pH conditions (9,10,(17)(18)(19). Despite the strong impact of temperature and pH on psaA expression, the underlying mechanisms have remained largely unknown. Using defined growth conditions, we recently demonstrated that high temperature and low pH activate psaA transcription in Y. pestis, and our work suggests that posttranscriptional regulation of two key psaA regulatory proteins, PsaE and PsaF, is primarily responsible for this unusual expression pattern (18). From our previous work, we predicted that posttranslational mechanisms also control the observed pH-dependent effect on PsaE and PsaF levels. Our earlier study examining the absence of PsaE or PsaF in DpsaF or DpsaE mutants, respectively, indicated that PsaE and PsaF exhibit codependent stability (18). Here, we provide evidence that residues in the periplasmic domains of each protein impact their stability. Building upon our previous work, we propose a model by which PsaF uses its histidine-rich periplasmic domain to sense pH (Fig. 6)  transcription of psaA, we predict that Y. pestis utilizes PsaF to stabilize PsaE and control psaA transcription in response to pH.
Acidic pH is known to influence gene expression in multiple bacteria (58)(59)(60)(61)(62). Yet, knowledge of the proteins and mechanisms underlying these pH-sensing pathways is limited. The data presented in this study into how PsaE and PsaF control expression of psaA offer insight into the mechanisms bacteria use to sense pH and activate gene expression. We show that a distinct pH threshold (pH . 6.7) determines whether or not PsaF is present. Because translation of psaF occurs at both pH 6.3 and pH 7.3 (18), it appears that pH controls protein stability rather than production. After production at pH 6.3, PsaF levels decrease following a shift to neutral pH, whereas levels remain high when left at mildly acidic pH; thus, PsaF functions as a membrane-localized pH-sensor, with its stability linked to extracellular pH. As with other proteins that have been implicated as pH sensors (48-50, 54, 63), PsaF has a periplasmic domain that is ideally positioned for such environmental sensing. This domain contains nine histidine residues, and we found that five were essential for PsaF stability at pH 6.3. The stability of PsaF in these five mutants is rescued by further reducing the pH, indicating that these mutants can sense and respond to this signal. While additional studies are needed to determine how histidines enhance PsaF stability, the involvement of pH suggests that their ionization state plays a role such that their protonation promotes stability. Stability of individual PsaF histidine mutants is rescued by growth at lower pH. One possible explanation for this is that individual residues become protonated within the folded protein at different pH levels. While the pK a of histidine imidazole is ;6.0, protein structure can influence the pK a of individual residues within a protein (64,65).
We previously reported that the expression of psaF from a plasmid restores PsaE to high levels in a DpsaF mutant, suggesting that one key role of PsaF is to enhance PsaE stability (18). The ability of PsaF His-to-Ala mutants to rescue PsaE stability and activate psaA transcription at lower pH levels, when the stability of these PsaF mutants themselves is restored, further supports this role. We predict that low pH alters the structural conformation of PsaF and that this affects its ability to interact with and enhance the stability of PsaE. PsaE is present at very low levels at pH of .6.7, and the absence of PsaF above this pH threshold is likely responsible for the observed loss of PsaE. However, we cannot exclude the possibility that PsaE also senses pH.
Recent work on PsaE-like proteins (ToxR and TcpP) identified proteases involved in targeted protein degradation that indicated these proteins are degraded by RIP (30,31,38,39). Here, we present evidence that DegP and Tsp, two proteases that contribute to ToxR and TcpP degradation, respectively (30,39,66), also contribute to the low levels of PsaE observed at pH 7.3. Tsp and DegP are both periplasmic proteases and may directly recognize and target the PsaE periplasmic domain. As PsaF is absent at pH 7.3, it seems likely that PsaF serves to block access of these proteases to PsaE and thus plays a similar role to that of ToxS and TcpH, which inhibit RIP of ToxR and TcpP, respectively (29,31).
The cytoplasmic N-terminal domain of PsaE contains an OmpR-like winged helixturn-helix DNA-binding domain that is thought to bind the psaA promoter to directly activate psaA expression (19,37). While PsaE-like proteins possess a conserved topology, the primary sequences of their periplasmic domains do not resemble each other, and a full understanding of the function of this domain is lacking. As was observed with TcpP (33), PsaE contains two cysteines in the periplasmic domain and our data indicate that both influence protein stability. Conversely, loss of either cysteine does not as readily impact the overall stability of ToxR but may alter its activity (44)(45)(46)(47)66). In both TcpP and ToxR, the cysteines influence protein conformation (33, 44-46, 66, 67). Notably, mutation of the cysteines in PsaE also negatively impacts PsaF stability, supporting our previous work indicating that the stability of PsaE and PsaF is mutually dependent (18). A similar effect of cysteine mutations in TcpP on TcpH stability has been observed (33). A recent study using the purified periplasmic domains of ToxR and ToxS suggested that intradomain disulfide bond formation of ToxR promotes binding to ToxS (68). This finding raises the possibility that intradomain disulfide bond formation in the periplasmic domain of PsaE is necessary for interaction with PsaF. In this scenario, if PsaF does not fold properly due to pH levels above 6.7 and cannot bind PsaE, then PsaE is susceptible to proteolysis via Tsp and DegP. When unable to interact with PsaE, PsaF itself appears to be unstable. Unlike that of PsaE, loss of Tsp or DegP alone is not sufficient to restore stability of PsaF at pH 7.3. These data are consistent with the observation that cysteine-to-serine mutations of PsaE result in loss of stability of both PsaE and PsaF.
While the topology of PsaE seems to be ideal for both environmental sensing and gene activation, it alone cannot activate psaA transcription. In a previous study, the absence of PsaF was associated with low psaA expression even when PsaE was produced at low temperatures (18). At pH levels of .6.7, where only low levels of PsaE and no PsaF are detected, there is negligible psaA transcription and PsaA cannot be detected. In most cases, PsaE levels were significantly reduced when PsaF was undetectable. The analysis of the PsaF histidine mutants, and their rescue by lowering the pH, indicated that in all growth conditions when both PsaE and PsaF were detectable, psaA expression was comparable to that in the WT. Together, these data indicate that PsaF may serve to do more than solely inhibit PsaE degradation and may also influence PsaE activity.
Our data resemble that of a recent report that suggested a direct interaction between the ToxR and ToxS periplasmic domains; the strength of this interaction was negatively impacted at alkaline pH, a growth condition under which ToxR is degraded (32). It is tempting to speculate that an analogous phenomenon occurs with PsaE and PsaF. In a similar manner, pH influences the stability of PsaE. The amount of PsaE is slightly lower at pH 6.7 relative to that at pH 6.3, yet PsaF levels are not altered at this pH range. This may indicate that PsaF and PsaE do not interact as effectively at intermediate pH (e.g., pH 6.7) resulting in reduced amounts of PsaE. In the absence of PsaF at pH of .6.7, levels of PsaE are even further reduced.
Bacteria control protein stability as a means of rapidly modulating gene expression in response to changes in environmental cues via RIP (69), and our data suggest that regulation of PsaE and PsaF stability drives pH-dependent psaA transcription. PsaE and PsaF belong to an unusual family of paired transcriptional regulators that are localized to the inner membrane (19,21,23), and most studies have focused on the function and regulation of the "PsaE-like" effector (i.e., ToxR and TcpP), likely because of their predicted role as a direct transcriptional regulator. PsaF clearly plays a vital role in enhancing PsaE stability (18), as has been proposed for the PsaF-like proteins TcpH and ToxS (29,31). Yet despite this role, little attention has been focused on PsaF-like proteins. This work provides new information on the role of PsaF by defining it as the pH sensor contributing to the pH dependent expression of psaA. As with PsaE and PsaF, TcpP and TcpH exhibit codependent stability; however, it is unknown if TcpP or TcpH levels are regulated in response to specific environmental signals. It is also not known if the PsaF-like proteins TcpH and ToxS directly function as sensors of specific environmental signals. Our work supports the current paradigm while offering new insights into mechanisms that influence the codependent stability of these protein pairs by revealing critical regulation of PsaF in response to an environmental signal, pH. The observation that PsaE is stable following a shift from pH 6.3 to 7.3 but that PsaF is not further suggests that PsaF serves as the direct pH sensor. Thus, we propose the temperature-and pH-dependent regulation of psaA transcription can be explained by temperature-dependent translation of psaE and psaF, an effect of pH on folding of PsaF and its interaction with PsaE, which in turn impacts the susceptibility of PsaE to proteolysis.

MATERIALS AND METHODS
Bacterial strains and growth conditions. Bacterial strains and plasmids used in this study are listed in Table 1. Y. pestis strains were cultivated on brain heart infusion (BHI) agar (BD Biosciences, Bedford, MA) at 26°C for 48 h and in BHI broth with aeration at 26°C or at 37°C. Escherichia coli strains were grown in LB (BD Biosciences) at 37°C. When indicated, bacteria were grown in BHI broth that was buffered to the indicated pH and filter sterilized as described previously (18). When necessary, antibiotics were added at the following concentrations: kanamycin (Kan), 50 mg/ml; carbenicillin (Carb), 100 mg/ml; and Irgasan (Irg), 2 mg/ml. Plasmid and strain construction. Primers used in this study are listed in Table 2. All in-frame deletion mutants and mutants in which alleles were introduced at the native site were constructed via allelic exchange using the pSR47S suicide vector (70) as previously described (18). All plasmids were constructed via Gibson assembly (NEB) and were confirmed by sequencing.
(i) In-frame deletions. The plasmids for in-frame deletions of degP and tsp were constructed by amplifying ;500-bp DNA fragments upstream and downstream of the target gene. These fragments were cloned into pSR47S, and the resulting plasmids, pJQ047 (DdegP) and pJQ018 (Dtsp), were introduced into YP6 (WT) or YPA18 (DpsaEF) via conjugation as previously described (18). Briefly, transconjugants were selected on BHI plates with Kan 50 and Irg 2 . The second recombination event was selected for by streaking Kan r (Kan-resistant)/Irg r colonies onto BHI agar plates containing 5% sucrose. Deleted genes were confirmed by PCR. (ii) Reintroduction of degP and tsp in DdegP and Dtsp mutants. The plasmids used to reintroduce degP and tsp at the native site were constructed as follows. The target gene coding sequence, including 500-bp upstream and downstream flanking regions, was amplified and cloned into pSR47S, and the resulting plasmids, pJQ054 and pJQ050, were introduced into YPA425 (DdegP) and YPA350 (Dtsp), respectively, via conjugation and integration of the plasmids, and were identified by selection on BHI plates with Kan 50 and Irg 2 .
gfp transcriptional reporter assay. To analyze psaA promoter activity, a gfp transcriptional reporter plasmid (pEW102) was electroporated into the indicated strains and assayed as previously described (18). Saturated cultures grown in unbuffered BHI were subcultured to an OD 600 of 0.2 in buffered BHI and grown for 8 h with aeration at 37°C. Relative fluorescent units (RFU) were measured using a Synergy HT microplate reader (BioTek Instruments, Winooski, VT) and normalized to OD 600 . Data are represented as RFU/OD 600 6 standard deviation.
PsaF stability assay. Saturated cultures of YP6 (WT) were subcultured to an OD 600 of 0.2 in BHI buffered to pH 6.3 and grown for 8 h at 37°C. Cells (OD = 10) were pelleted, washed once with PBS adjusted to pH 6.3, split into equal volumes, pelleted, and suspended in PBS adjusted to either pH 6.3 or 7.3, and incubated at 37°C for 4 h. At 0 h, 2 h, and 4 h, cells (OD = 2) from each condition were collected and prepared for Western blot analysis.
Western blot analysis. Saturated cultures grown in unbuffered BHI were subcultured to an OD 600 of 0.2 in BHI buffered to the indicated pH and grown for 8 h at 37°C. PsaE, PsaF, and PsaA were analyzed as previously described (18). Briefly, cell lysates were resuspended in Laemmli buffer, boiled, separated via SDS-PAGE, and transferred to a polyvinylidene difluoride (PVDF) membrane. Loading of protein samples was qualitatively assessed via Ponceau S staining. Anti-PsaE, anti-PsaF, and anti-PsaA sera were used as previously described (18). Anti-IgG horseradish peroxidase (HRP)-conjugated secondary antibodies were used at a titer of 1:20,000.
Statistical analysis and reproducibility. Analyses were performed using Prism 8 (GraphPad Software, San Diego, CA). All experiments were performed at least three times, with biological triplicates in each assay. Unless otherwise noted, representative assays are shown.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.1 MB.