ABSTRACT
Experimental evolution of Escherichia coli K-12 W3110 by serial dilutions for 2,200 generations at high pH extended the range of sustained growth from pH 9.0 to pH 9.3. pH 9.3-adapted isolates showed mutations in DNA-binding regulators and envelope proteins. One population showed an IS1 knockout of phoB (encoding the positive regulator of the phosphate regulon). A phoB::kanR knockout increased growth at high pH. phoB mutants are known to increase production of fermentation acids, which could enhance fitness at high pH. Mutations in pcnB [poly(A) polymerase] also increased growth at high pH. Three out of four populations showed deletions of torI, an inhibitor of TorR, which activates expression of torCAD (trimethylamine N-oxide respiration) at high pH. All populations showed point mutations affecting the stationary-phase sigma factor RpoS, either in the coding gene or in genes for regulators of RpoS expression. RpoS is required for survival at extremely high pH. In our microplate assay, rpoS deletion slightly decreased growth at pH 9.1. RpoS protein accumulated faster at pH 9 than at pH 7. The RpoS accumulation at high pH required the presence of one or more antiadaptors that block degradation (IraM, IraD, and IraP). Other genes with mutations after high-pH evolution encode regulators, such as those encoded by yobG (mgrB) (PhoPQ regulator), rpoN (nitrogen starvation sigma factor), malI, and purR, as well as envelope proteins, such as those encoded by ompT and yahO. Overall, E. coli evolution at high pH selects for mutations in key transcriptional regulators, including phoB and the stationary-phase sigma factor RpoS.
IMPORTANCE Escherichia coli in its native habitat encounters high-pH stress such as that of pancreatic secretions. Experimental evolution over 2,000 generations showed selection for mutations in regulatory factors, such as deletion of the phosphate regulator PhoB and mutations that alter the function of the global stress regulator RpoS. RpoS is induced at high pH via multiple mechanisms.
INTRODUCTION
Escherichia coli is a neutralophilic bacterium with the ability to grow under a wide range of pH. Before colonizing the colon, the bacteria survive exposure to acidic levels as low as pH 2.0 in the stomach and alkaline levels exceeding pH 10 at the pancreatic duct (1, 2). During growth, E. coli normally maintains a cytoplasmic pH within the range of pH 7.4 to 7.9 (3, 4), while growing over the range of external pH 5.0 to 9.0 (5). At high external pH, the cell downregulates expression of acid stress response genes and positively regulates expression of base resistance genes (6, 7). The acid-inducible phosphate regulator PhoB limits production of fermentation acids; phoB deletion thus increases acid production, which could enhance cell growth at high pH (8, 9). A different response involves the NhaA Na+/H+ antiporter, which imports protons in exchange for sodium ion at high external pH. The nhaA gene is upregulated at high pH (10, 11).
Another way that E. coli acclimates to pH stress is by inducing the global stress sigma factor rpoS, which regulates ∼500 stress-associated genes and 1,044 genes in total (12, 13). RpoS is expressed at high levels under a variety of stress conditions, such as nutrient starvation associated with stationary phase (14), as well as during exponential phase if the cells face other external stresses such as temperature shock, osmotic stress, or acid stress (12). RpoS function is required at extremely high pH (pH 9.8), where cells cannot grow but remain viable (15).
The mechanism of the RpoS contribution to the high-pH stress response is unknown. RpoS expression is regulated via transcription, translation, and proteolysis. Transcriptional upregulation or mRNA stabilization may be stimulated by the alarmone (p)ppGpp and is negatively regulated by the cAMP-cAMP receptor protein (cAMP-CRP) complex (16). Translation of rpoS is regulated by small regulatory RNAs (sRNAs) (12, 17). Inhibition of proteolysis is an additional means by which RpoS levels are elevated when cells encounter stress (16). When cells are growing well, adaptor protein RssB takes RpoS to the ClpXP protease for degradation (18). However, during stress, antiadaptors IraM, IraP, and IraD sequester RssB, thereby stabilizing RpoS; each antiadaptor is induced under a different stress condition (19).
At high pH in the gut environment, enteric bacteria face anaerobic conditions under which ATP synthesis is less than optimal. One system that enables bacteria to compete under both alkaline and anaerobic conditions is torCAD (20). The torCAD respiratory operon is induced in the presence of trimethylamine N-oxide (TMAO); the resulting proteins reduce TMAO to trimethylamine (TMA) under regulation by the periplasmic TorT protein and the two-component TorS histidine kinase and TorR response regulator (21). Production of trimethylamine alkalizes the extracellular medium (20). Another target of TorR, the tryptophan deaminase TnaA, may help the cell to anticipate and deal with alkaline stress (5, 6, 20). The torCAD operon is negatively regulated by TorI, a transcriptional regulator that originated as the excisionase of defective prophage KplE1 (22).
We devised an evolution experiment to select for mutations that enhance growth and/or stationary-phase persistence at high pH. We cultured E. coli W3110 at the upper limits of pH base stress (pH 9.0 to pH 9.3) for 2,200 generations in a 96-well microtiter plate and then sequenced (Illumina MiSeq 2 × 250) the genomes of two clones from each of four individual population wells (B11, F7, D9, and G9). Mutations were predicted by reference-based alignment using the computational pipeline breseq 0.30.0 (23). Previous laboratory evolution experiments have adapted E. coli to grow optimally in various stressful environments and have identified novel mutations that confer higher fitness in external acid, permeant acid, and minimal media (24–26). A common mechanism of evolution involves mutation of regulators that shift expression levels of many genes in complex ways, such as the Gad regulon at low pH (27).
In this paper, we present a genetic basis for adaptation under prolonged exposure to alkaline stress under semiaerobic conditions. Our studies reveal high-pH fitness advantages for mutations in regulatory genes such as phoB (activator of phosphate regulon) (8, 9) and in components affecting expression of rpoS.
RESULTS
The fitness of populations (B11, F7, D9, and G9) increases at pH 9.25.At pH levels above approximately pH 9.0, our E. coli strain W3110 grows for approximately four doublings, and then the culture declines without achieving sustained stationary phase. We began subculturing 24 populations of W3110 E. coli at pH 9.0, near the upper limit for sustained growth. The evolving cultures reached progressively higher endpoint densities (Fig. 1). As the endpoint density increased, the pH of the medium was increased to pH 9.2 (470 generations), then to pH 9.25 (1,200 generations), and finally to pH 9.3 (1,900 generations). After 2,200 generations, the stationary-phase culture density at pH 9.25 was more than 3-fold higher than that of the ancestral strain. This steady increase in tolerance of the stress condition is consistent with the pattern seen for evolved populations subcultured at low pH (24) or in the presence of benzoic acid (25).
Endpoint growth of evolved populations from generations 0, 500, 1,000, 1,500, and 2,200 at pH 9.25. Frozen plates of evolved populations were thawed and diluted 1:50 into fresh LBK (with 150 mM AMPSO; pH 9.25) and then incubated in a microplate reader at 37°C for 22 h with intermediate shaking and subsequent OD450 readings every 15 min. Optical density was measured at 16 h. Mean values are shown for 8 representative evolving populations. Error bars indicate standard error of the mean (SEM) (n = 8).
Genomes of high-pH-evolved clones show mutations.After 2,200 generations, two clones from each of four populations (B11, F7, D9, and G9) were isolated for genome sequencing (Tables 1 and 2). The sequences were aligned with the sequence determined from NCBI sequence NC_007779.1 for strain W3110; mutations found in our lab stock culture of strain W3110 were excluded from analysis. Populations F7 and D9 show several mutations in common, which might have arisen from inadvertent cross-transfer during an early time of serial culture.
Strains generated by experimental evolution, by P1 construction, or by recombineering
Mutations found in selected populations after 2,200 generations at pH 9.1 to 9.3a
The base-evolved sequences showed approximately 50 mutations in total from the eight sequenced strains compared to the ancestral W3110. These included 23 missense mutations but only one silent mutation, a ratio indicating a high likelihood of selection pressure. Insertion sequences (IS) caused 10 knockout mutations, as well as a large insertion-mediated deletion (insH-[rfbA]). The prevalence of IS-mediated knockouts is consistent with that reported for evolution experiments with acid stress (25, 27).
Three of the four populations (F7, D9, and G9) showed a deletion of prophage KplE1 (genes intS-torI), whose excisionase TorI functions also as a transcriptional regulator (28, 29). All four populations showed mutations related to rpoS, a known regulator of high-pH survival (15). Populations F7 and D9 shared a missense mutation, population B11 had a mutation upstream of rpoS, and population G9 showed a mutation in sspA, encoding an acid resistance protein that decreases H-NS expression and upregulates rpoS (30). Strain F7-1 showed a knockout insertion in phoB, the positive regulator of the phosphate regulon (8, 9). Expression of phoB is induced at low pH, and phoB deletion increases production of fermentation acids. Other mutations found in the high-pH-evolved strains affected rpoN (encoding a nitrogen starvation sigma factor), yobG (mgrB) (PhoPQ regulator), pcnB [poly(A) polymerase], ydcI (transcriptional regulator), periplasmic and envelope protein genes (imp, yahO, and ompT); and murein maintenance enzyme genes (mpl and slt).
High-pH-evolved clones show a fitness increase at high pH but not at low pH.When cultured at pH 9.25, all the evolved clones (B11-1, F7-2, D9-1, and G9-1) grew to higher endpoint densities than did the ancestor W3110 (Fig. 2A). Early growth rates showed no significant differences, but the ancestor began to decrease in culture density after about 5 h in the alkaline medium, whereas the four evolved clones persisted in stationary phase. After 16 h of growth in pH 9.25 medium, evolved clones F7-2, D9-1, and G9-1 grew to a higher density than B11-1 (Tukey test, P < 0.05). For each strain, the curve shown is one of eight replicates showing the median culture density after 16 h of incubation. Figure 2B shows all eight replicates for two of the evolved strains, G9-1 and B11-1. Overall, our high-pH-adapted strains showed a growth advantage in alkaline medium.
High-pH-evolved clones outgrow the ancestor W3110 at pH 9.25. Strains were cultured overnight at pH 8.5 as described in Materials and Methods. Overnight cultures were diluted 200-fold into microplate wells with LBK–150 mM AMPSO at pH 9.25; the final pH after 22 h of incubation was 9.15. (A) The curves shown have median OD600 values at 16 h. The 16-h culture densities of strains D9-1, F7-2, G9-1, and B11-1 showed a significant difference from that of the ancestral strain W3110 (Tukey test, P < 0.05). (B) Full set of 8 replicate growth curves for strains G9-1, B11-1, and W3110 from the same experiment as for panel A.
We sought to determine whether the fitness advantage acquired by the evolved strains was pH dependent. The growth of high-pH-evolved isolates was tested at low pH, near the end of the pH range allowing growth of E. coli (Fig. 3). The growth rates and culture densities were tested for isolates B11-1, F7-2, D9-1, and G9-1 cultured in potassium-modified Luria broth (LBK) medium buffered with homo-PIPES [homopiperazine-1,4-bis(2-ethanesulfonic acid)] (pH 4.6). All isolates reached lower 16-h culture densities than the ancestor at pH 4.6 (Tukey test, P < 0.05). Thus, the mutations selected in the high-pH-evolved populations slightly decreased the relative fitness at low pH. Growth at pH 7.0 (Fig. 3C) showed mixed results; strains D9-1 and B11-1 showed slightly higher growth than W3110, but the other strains showed no difference.
Culture densities for base-evolved strains at pH 4.6 and at pH 7.0. Strains were cultured overnight in LBK medium buffered with MES at pH 5.5 and then diluted 200-fold in medium buffered with homo-PIPES, pH 4.6. All curves shown have median OD600 values at 16 h. (A) Strains G9-1, D9-1, G7-2, and B11-1 showed 16-h culture densities significantly higher than that of W3110 (Tukey test, P < 0.05). (B) All eight replicates are shown for each strain, G9-1, B11-1, and W3110. (C) Culture densities reached at pH 7.0 in LBK buffered with 100 mM MOPS.
Deletion of phoB enhances growth at high pH.Several mutations found in the high-pH-evolved strains suggest the possibility that knockout or altered activity of these genes might enhance fitness at high pH. To test this possibility, deletions of these genes from the Keio collection were transduced into the ancestor W3110 and their growth rates tested at pH 9.1 (Fig. 4). Each deletion strain was tested in five trials of eight replicates each.
W3110 deletion strains cultured at pH 9.10 or at pH 9.25. Strains were cultured overnight at pH 8.5 as described in Materials and Methods. Overnight cultures were diluted 200-fold into microplate wells with LBK–150 mM AMPSO at pH 9.10; the final pH after 22 h of incubation was 9.1. The curves shown have median OD600 values at 16 h. (A) The strains shown are all derivatives of W3110 transduced with Keio deletion ΔphoB::kanR, ΔpcnB::kanR, ΔydcI::kanR, ΔtorI::kanR, or ΔrpoS::kanR. Of the deletions tested, the ΔphoB::kanR and ΔpcnB::kanR strains showed significantly higher growth at 16 h than strain W3110, whereas the ΔrpoS::kanR strain showed lower growth (Tukey test, P < 0.05). (B) Strain G9-1 with a recombineered ΔtorCAD deletion showed no significant difference in 16-h culture density compared to that of the parent strain G9-1. Both grew significantly more than W3110. (C) Strain W3110 with a deletion of regulator ΔtorR or of ΔtorS showed no significant difference in growth compared to that of W3110. All showed lower growth than that of the strains evolved at high pH: B11-1, F7-2, D9-1, and G9-1 (Tukey test, P < 0.05).
High-pH-evolved strain F7-1 showed an IS1 insertion knockout of phoB (Table 2). Strain W3110 phoB::kanR showed consistently higher growth than did the ancestor W3110. Thus, phoB deletion, with downregulation of the phosphate regulon (8), could be contributing to the high-pH fitness of strain F7-1. The fitness effect of ΔphoB, however, was smaller than that seen for F7-2 (Fig. 2A) and was less consistent when tested at pH 9.25. Thus, F7-1 and F7-2 must possess additional mutations contributing to high-pH fitness.
Testing of other high-pH-selected gene deletions.Deletion of other genes showed smaller contributions to high-pH fitness (Fig. 4). The ΔpcnB::kanR knockout strain outgrew W3110 in 4 out of 6 sets of eight replicates, and the ΔydcI::kanR strain outgrew the ancestor in 3 of 6 sets. No fitness effect was seen for deletions ΔyahO::kanR, Δmpl::kanR, ΔompT::kanR, Δslt::kanR, and ΔyfcD::kanR. All these genes are likely to have increased fitness during the serial culture of experimental evolution, but revealing their contributions may require extended direct-competition assays (24).
Since three out of four populations showed a deletion including torI, we tested whether the loss of torI contributes to fitness increase at high pH (Fig. 4A). Strain W3110 ΔtorI::kanR did not grow significantly better than the ancestor. The KplE1 excisionase TorI functions also as a transcriptional regulator of torCAD, encoding proteins that reduce trimethylamine N-oxide (TMAO), a reaction that may increase pH (20, 22). Since TorI inhibits torCAD expression, we hypothesized that deletion of torI increases expression of torCAD and that torCAD expression somehow enhances fitness at high pH. To test this, we recombineered a deletion of torCAD in the evolved isolate G9-1. The G9-1 ΔtorCAD deletion strain was cultured at pH 9.25. W3110 ΔtorCAD and G9-1 grew almost identically in the pH 9.25 LBK, and both grew significantly better than the ancestor (Fig. 4B) (Tukey test, P < 0.05). Thus, torI regulation of torCAD does not show a fitness effect detectable under the conditions tested. We also tested for a possible fitness contribution of the torS-torR phosphorelay system, which mediates TorI inhibition of torCAD (20). When cultured at pH 9.25, W3110 ΔtorS::kanR and W3110 ΔtorR::kanR grew as poorly as the W3110 ancestor (Fig. 4C). Similarly, deletion of torR from the evolved strain G9-1 showed no significant difference in growth compared to that of the parental strain G9-1 (data not shown).
Mutations in rpoS.Mutations affecting RpoS were of interest as they appeared in several high-pH isolates, although none appear to knock out function (Table 2). The rpoS-associated mutations found in our evolved strains are point mutations that may have more complex effects on phenotype. In fact, all four populations showed a mutation associated with RpoS, which is necessary for survival at extremely high pH beyond pH 9.8. Both clones in populations F7 and D9 acquired a missense mutation to rpoS (E247K) within the open reading frame (ORF). This mutation occurred in region 3 of the RpoS protein, which is responsible for extended −10 promoter recognition (31). Additionally, strain B11 acquired an intergenic mutation (A→G) upstream of rpoS in the leader responsible for translational regulation (12). For comparison, a strain containing ΔrpoS::kanR reached a lower maximum optical density than did the parent W3110 in 4 out of 6 sets of replicates. This result suggests that the mutations found in our experimental evolution do not eliminate RpoS function.
RpoS protein levels and activities vary in the high-pH-evolved strains.We examined the status of RpoS and RpoS activity in the evolved strains, monitoring protein level (Fig. 5A and B) and monitoring RpoS activity by a qualitative measure of catalase activity; the RpoS-dependent katE gene encodes catalase HPII (32) (Table 3). RpoS protein levels were examined under two conditions: growth to early exponential phase, when the W3110 parent does not express RpoS (Fig. 5A), and growth to stationary phase, at either pH 7 or pH 9, when W3110 has accumulated RpoS (Fig. 5B). Strains D9 and G9 had significant levels of RpoS in exponential phase, strain B11 had no visible RpoS under either growth condition, and strain F7 had a pattern similar to that of the W3110 parent. However, all strains had catalase activity less than that of the parent but significantly above that of a ΔrpoS derivative of the parent (Table 3). Thus, each of the evolved strains has lowered but not abolished RpoS activity, by different mechanisms.
RpoS expression in high-pH-evolved strains. Ancestral strain W3110, evolved isolates (B11-2, F7-1, D9-2, and G9-1), and INH7 (W3110 ΔrpoS::kan) were grown overnight in LBK buffered with 100 mM MOPS (pH 7.0) at 37°C. Cultures were diluted 1:1,000 into fresh medium (LBK with 100 mM MOPS, pH 7.0) and grown to exponential phase (OD600 of 0.1). At an OD600 of 0.1, cells were split via filtration into LBK–100 mM MOPS (pH 7.0) or LBK–100 mM TAPS (pH 9.0). (A) Immediately upon the split into pH 7 and pH 9 media, samples were taken and precipitated with 5% trichloroacetic acid (TCA). Samples were analyzed by Western blotting using fluorescent anti-RpoS serum and anti-EF-Tu serum as a loading control. Blue asterisks correspond to samples resuspended in pH 7.0 medium, and red asterisks correspond to samples resuspended in pH 9.0 medium. Note that these strains experience the filtration treatment but have no time to adjust to the changed pH, and thus the two samples can be considered duplicates. (B) Cultures were grown at either pH 7.0 or pH 9.0 for an additional 150 min after the split; these cells were then in stationary phase, where the level of RpoS protein is expected to be high, as seen for W3110. RpoS levels in all evolved isolates are high independent of pH, except for B11, which has barely detectable levels of RpoS at stationary phase.
Evolved clones exhibit lower RpoS-dependent catalase activity
In strain B11, the leader mutation may decrease translation, consistent with loss of RpoS protein (Fig. 5A and B). In strains F7 and D9, the E247K mutation may decrease RpoS activity. The basis for constitutive expression of RpoS in D9 and G9 was not determined, but possibly the sspA mutation in G9-1 contributes to this (Table 2). SspA is a starvation protein that blocks the H-NS repressor, and H-NS in turn represses expression of iraD and iraM antiadaptors that stabilize RpoS (30, 33). If the sspA allele had a gain of function, repressing H-NS more effectively, the resulting expression of the antiadaptors would be expected to increase RpoS accumulation. Given the decreased activity of RpoS in G9-1, we would propose that sspA has a loss of function or that one of the other mutations in this lineage is responsible for the decreased activity.
RpoS transcription and translation increase at pH 9.0 compared to pH 7.0.All our high-pH-evolved strains showed decreased but not absent RpoS activity (Table 3), and rpoS deletion strains showed pH-sensitive growth (Fig. 4A). Thus, we suspected that high-pH stress, like so many other conditions, might induce RpoS. However, the long-term high RpoS expression might not be optimal for the regrowth after dilution during the evolution experiments. Thus, it could be that these high-pH-evolved populations might show selection for mutations that partly decrease RpoS. We sought to determine if and how RpoS is upregulated during growth at high pH.
RpoS is regulated at multiple levels (12). In addition to transcriptional regulation, translation is inhibited by a hairpin within the 5′ untranslated region (UTR). Under specific stress conditions, small noncoding RNAs anneal to the 5′ UTR, opening the hairpin and thus activating translation. RpoS protein is degraded, dependent upon the adaptor protein RssB and the ClpXP protease; antiadaptors, induced in response to specific stresses, protect RpoS from degradation.
We first measured RpoS protein accumulation (Fig. 6A) and β-galactosidase activity (Fig. 6B) at pH 9.0 compared to pH 7.0. We used strain SG30013, a derivative of MG1655 that expresses rpoS at the wild-type locus in the genome and carries the rpoS promoters, leader, and protein fused to a lacZ translational reporter gene. This reporter measures all levels of regulation (SG30013) (Table 1). Western blot analysis of RpoS protein accumulation and β-galactosidase activity showed significant upregulation of RpoS at pH 9.0 compared to pH 7.0 within 45 min of exposure at pH 9.0 (Tukey test, P < 0.05). At both pH values the cultures showed sustained high-pH induction of RpoS during exponential phase, with a 20-fold increase (pH 9.0 versus pH 7.0) by 60 min (Fig. 6B). In contrast, at pH 7.0 there was no RpoS accumulation until 150 min of growth, when cells had entered stationary phase (Fig. 6B).
RpoS is induced during growth at pH 9.0. Strain SG30013 expresses wild-type genomic RpoS and also has a lac translational fusion to rpoS. (A) Bacteria were cultured overnight in LBK–100 mM MOPS (pH 7.0) and then diluted (1:1,000) into fresh medium. Upon growing to an OD600 of 0.1, cultures were split via filtration into LBK–100 mM MPOS (pH 7.0) and LBK–100 mM TAPS (pH 9.0). Samples were taken at the time of the initial split and subsequently as indicated. Protein samples were precipitated with 5% TCA and analyzed via Western blotting using an anti-RpoS antiserum. EF-Tu was used as a loading control. (B) Bacteria were cultured as described above. Samples for β-galactosidase activity were prepared and assayed as described by Miller (51) (n = 3; error bars = SEM). Blue bars correspond to cultures grown at pH 7.0 and red bars correspond to cultures grown at pH 9.0. Numbers above the bars indicate fold induction at pH 9.0 compared to pH 7.0. Asterisks indicate significant rpoS induction at pH 9.0 over pH 7.0 compared to time zero (Tukey test, P < 0.05). MU, Miller units.
Does the high-pH induction of RpoS involve transcriptional regulation? The gene rpoS is transcribed primarily from PrpoS, initiating within the upstream nlpD gene; two nlpD promoters also contribute to its expression. To determine if upregulation of transcription contributes to high-pH induction of RpoS, we assayed a transcriptional fusion (NM376) which contains all rpoS promoters (PrpoS, PnlpD1, and PnlpD2) and 15 nucleotides of the leader fused to lacZ. We detected small induction of RpoS within 15 min of exposure to pH 9.0 (Fig. 7A); this contrasts with the 20-fold induction seen after 60 min in Fig. 6. Therefore, only a portion of the response to pH 9.0 is via transcriptional induction. Next, we sought to determine which of the three promoters contribute to the transcriptional induction of RpoS by using an isogenic transcriptional fusion (NM378) containing only one promoter (PrpoS). With exposure to an external high pH during exponential phase, we observed a marginal 2-fold induction of RpoS. The positive transcriptional regulation (pH 9.0 versus pH 7.0) at the promoter level comes primarily from PrpoS (Fig. 7B, compared to Fig. 7A). Taking the results together, we estimate a 2-fold transcriptional contribution to RpoS induction at pH 9.0 during growth.
rpoS transcription contributes a modest induction at pH 9.0 during exponential growth. (A) Strain NM376 was prepared and assayed for β-galactosidase activity as for Fig. 1. NM376 is a transcriptional fusion that contains all rpoS promoters (Prpo, PnlpD1, and PnlpD2) and 15 nucleotides of the leader. (B) Strain NM378 was prepared and assayed for β-galactosidase activity as described for Fig. 1. NM378 possesses only the major rpoS promoter (PrpoS) and 15 nucleotides of the leader. Blue bars correspond to cultures at pH 7.0 and red bars correspond to cultures at pH 9.0. Asterisks indicate significant rpoS induction at pH 9.0 (Tukey test, P < 0.05). MU, Miller units.
We tested the role of regulation at the level of rpoS translation. RpoS translation by sRNAs depends on an inhibitory hairpin with the leader. To determine if RpoS translation is upregulated during growth at pH 9.0, we assayed two different translational fusions using the methodology described above (Fig. 8). The first fusion (BA938) contained a constitutive Cp17 prompter, full-length leader, and truncated 477-nucleotide (nt) ORF that lacks the sequence required for RssB-mediated RpoS degradation. The other translational fusion (BA943) was isogenic except for a point mutation (C125T) within the leader. This mutation prevents the formation of the mRNA hairpin loop, thereby leaving the ribosome-binding site (RBS) exposed for RNA polymerase to initiate translation (open hairpin). Both the translational fusions with the wild-type hairpin (BA938) and with the open hairpin (BA943) showed a similar 2-fold induction of RpoS beginning 45 min after exposure to pH 9.0 (Fig. 8A and B). As we saw for transcription, translation of RpoS is modestly increased at pH 9.0. There is no clear evidence of dependence on sRNAs with the hairpin loop (34).
RpoS translation increases modestly beginning after 30 min in an mRNA hairpin-independent manner. (A) Strain BA938 was prepared and assayed for β-galactosidase activity as described for Fig. 1. BA938 is an rpoS-lacZ translational fusion constitutively expressed by a Cp17 promoter. Additionally, BA938 has a truncated rpoS ORF (477 nt) which lacks the region necessary for RssB-dependent degradation. (B) Strain BA943 was prepared and assayed for β-galactosidase activity as for Fig. 1. BA943 is a derivative of BA938 with a C125T mutation, which prevents the formation of the rpoS mRNA hairpin loop, thereby preventing occlusion of the RBS. Asterisks indicate significant rpoS induction at pH 9.0 (Tukey test, P < 0.05). MU, Miller units.
RpoS induction is mediated by antiadaptors IraM, IraP, and IraD.RpoS is rapidly degraded under nonstress conditions, dependent upon adaptor protein RssB, and is stabilized upon expression of three known antiadaptors (IraM, IraP, and IraD) (35). To examine the contribution of protein stabilization to induction at high pH, we first utilized strain SG30018, a derivative of SG30013 carrying a deletion of rssB (Table 1). The rssB deletion stabilizes RpoS by inhibiting RssB-mediated delivery to the ClpXP protease, thereby mimicking the function of antiadaptor proteins; thus, RpoS should be fully stable in this strain. During growth at pH 9.0, we observe a 4-fold increase of RpoS-LacZ compared to that during growth at pH 7.0 (Fig. 9A). This increase is less than the 20-fold induction seen after 60 min in an rssB+ strain (Fig. 6B). This result implies that one or a combination of the known antiadaptors (IraM, IraP, and IraD) or not-yet-identified antiadaptors interacts with adaptor protein RssB to stabilize RpoS at pH 9.0. A derivative of SG30013 that was mutant for all three antiadaptors, iraM, iraP, and iraD (INH5), was grown and assayed as before in buffered alkaline medium (pH 9.0) (Fig. 9B). This strain had a lower level of expression of the fusion protein, consistent with loss of antiadaptor-mediated stabilization, and showed no RpoS induction at pH 9.0 compared to pH 7.0.
RpoS stability contributes a 4-fold induction of RpoS expression during growth at pH 9.0, likely through known antiadaptors IraD, IraM, and IraP. (A) A ΔrssB mutant (SG30018, with stabilized RpoS) was prepared and assayed for β-galactosidase activity as for Fig. 1. (B) A ΔiraM ΔiraP ΔiraD mutant (INH5) was grown and assayed for β-galactosidase activity as described for Fig. 1 (n = 3; error bars = SEM). Asterisks indicate significant rpoS promoter induction at pH 9.0 (Tukey test, P < 0.05). MU, Miller units.
Overall, our results demonstrate significant induction of RpoS upon exposure to pH 9.0. Stabilization of the protein, at least in part due to one or more of the known antiadaptors, contributes around 5-fold to induction; transcription and translation each contribute about 2-fold, for an overall induction, after 60 min, of 20-fold. Thus, high-pH treatment has multiple effects on the RpoS sensing pathway.
To confirm that pH 9.0 did not affect the constitutive Cp17 promoter or lacZ fusion, we assayed a control strain that had a Cp17::lacZ fusion in an SG30013 background at pH 7.0 and pH 9.0. When grown at pH 7.0 or pH 9.0, RpoS was induced equally with little variation (Fig. 10). This result indicates that pH 9.0 does not affect the expression of Cp17 or lacZ.
pH 9.0 does not affect the stability or activity of the Cp17 constitutive promoter or the lacZ gene. Strain BA926 is a derivative of MG1655 with a Cp17-lacZ transcriptional fusion. MU, Miller units.
DISCUSSION
Experimental evolution of E. coli at pH 9.0 to 9.3 yielded strains that grow at pH values significantly higher than that for the ancestral strain. Genomic analysis of eight isolates revealed approximately 50 mutations that may be contributing to fitness at pH 9.0. As found in our previous pH stress evolution experiments, many of the mutations involved knockout of regulators (25, 27). The knockout of acid-inducible phoB shows a clear contribution to high-pH fitness, which could be associated with the increased production of fermentation acids.
Three out of four populations (F7, D9, and G9) acquired a deletion of the KPlE1 prophage, which carries torI, a transcriptional activator of the torCAD operon. Induction of torCAD downregulates known acid resistance genes gadA, gadBC, hdeAB, hdeD, yhiE, and yhiM and positively regulates base resistance genes tnaLAB, possibly in anticipation of alkalization of the medium caused by TMAO reduction to trimethylamine (20). Deletion of torCAD did not appear to decrease the fitness of strain G9-1 under the conditions tested (one daily dilution cycle), but given the appearance of the deletion in multiple populations, it is likely that torCAD expression has a fitness contribution that could be demonstrated by more sensitive means such as a multicycle direct competition. For the future, our lab is establishing a sensitive fluorescence-activated cell sorting (FACS) system for competition analysis (36).
The high-pH-evolved isolates showed a striking pattern of point mutations in the RpoS stress sigma factor, whose fitness contributions remain unclear. RpoS is required for survival at extremely high pH (pH 9.8) (15), and it mediates regulation of base resistance proteins (10, 13). This connection may be related to the fact that the pH rises during culture growth to stationary phase in unbuffered tryptone-yeast extract medium (37, 38). Our subculturing cycle subjected bacteria to repeated periods of stationary phase, conditions under which rpoS mutations accumulate as part of the “growth advantage in stationary phase” (GASP) response (39, 40). Multiple rounds of stationary phase, with concomitant high pH, also lead to mutations in the rpoS activator sspA (41), which showed a mutant allele in one of our base-evolved populations (population G9).
One possible explanation for the selection of rpoS variant alleles in our populations is that the short-term benefits of RpoS at high pH have an associated cost under long-term exposure. A comparable short-term and long-term trade-off is seen for the acid-inducible amino acid decarboxylases at low pH (27). Consistent with this role, we found that RpoS is induced during growth at pH 9.0 but not at pH 7.0. This upregulation of RpoS allows expression of proteins needed for nongrowth survival at extremely high pH (pH >9.8). Analysis of rpoS-lacZ fusions revealed a complex pathway for induction. Transcriptional fusions showed a 2-fold contribution to RpoS induction primarily from PrpoS (Fig. 7). Similarly, translational fusions revealed a 2-fold induction. The regulatory details we observed must be qualified by the fact that different media and strains were used for the molecular experiments and the evolution experiments.
Under certain stresses, RpoS translation is positively regulated with the help of chaperone protein Hfq via direct binding of small regulatory RNAs (sRNAs) to the RpoS mRNA hairpin (42). When an sRNA binds to an mRNA hairpin, the hairpin is opened and the ribosome may bind to the RBS to initiate translation (17). However, we saw no clear evidence that RpoS induction is affected by the opening of the sRNA hairpin loop (Fig. 8). It is unclear what may be causing transcriptional or translational RpoS induction.
In our experiments, the greatest contribution to RpoS induction came from protein stabilization, via inhibition of RssB-mediated ClpXP proteolysis of RpoS (Fig. 9). At least one of the three characterized antiadaptors (iraM, iraP, and iraD) clearly contributes to this stabilization, but it may not account for all of it (Fig. 9B). It would be interesting to determine which of the three antiadaptors may be stabilizing RpoS and how high pH induces this response. IraP and IraD are known to be induced in response to the alarmone ppGpp (43, 44); if high pH increases ppGpp (44), we would predict that these antiadaptors will be critical for high-pH induction. Elevated ppGpp levels might also explain the minor transcriptional induction.
Our investigation of the expression of RpoS protein levels and RpoS activity in the evolved strains demonstrates that they have all taken different pathways to the same final phenotype, that is, RpoS activity levels that are lower than those of the wild-type strain but significantly above those of a strain devoid of RpoS. This proposed mechanism is consistent with the combination of a need for RpoS to cope with the perpetual high-pH stress coupled with compensation for some fitness cost of the high RpoS levels resulting from high pH. The evolution to reduced but not absent RpoS is strikingly similar to the case for rpoS alleles isolated as advantageous for long-term stationary-phase growth (45), reinforcing the importance of having enough but not too much RpoS.
MATERIALS AND METHODS
Strains and culture media.The strains used in this study are listed in Table 1. All high-pH-evolved isolates were derived from a W3110 K-12 ancestor (lab stock D-8; originally from F. Neidhardt) (24). For RpoS induction experiments, strains were constructed starting with SG30013. SG30013 is a derivative of MG1655 containing a translational fusion of lacZ fused to the promoters, leader, and first 250 amino acids of the rpoS gene (46).
Bacteria were cultured at 37°C in Luria broth supplemented with potassium (LBK) (10 g/liter tryptone, 5 g/liter yeast extract, 7.45 g/liter KCl). The medium was buffered with either 100 mM 2-(N-morpholino) ethanesulfonic acid (MES) (pKa = 5.96), 100 mM 3-(N-morpholino)propane-1-sulfonic acid (MOPS) (pKa = 7.20), [(2-hydroxy-1,1-bis(hydroxymethyl)ethyl)amino]-1-propanesulfonic acid (TAPS) (pKa = 8.4), or N-(1,1-dimethyl-2-hydroxyethyl)-3-amino-2-hydroxypropanesulfonic acid (AMPSO); pKa = 9.0). The pH was adjusted accordingly using 5 M KOH. All cultures were incubated at 37°C unless stated otherwise.
High-pH (pH 9.25) experimental evolution.High-pH experimental evolution was modeled after our previous laboratory evolution experiments conducted in the presence of external acid and permeant acid (24, 25). Twenty-four high-pH-evolved populations originated from the same K-12 W3110 (lab stock, D-8) ancestor. The ancestral strain was cultured overnight in LBK with 100 mM TAPS (pH 8.5) and then diluted (1:200) into 24 wells containing 200 μl LBK with 150 mM TAPS (pH 9.0). The plate was then incubated in a SpectraMax Plus384 MicroPlate reader (Molecular Devices) at 37°C for 22 h while optical density at 45 nm (OD450) readings were recorded every 15 min after shaking for 3 s. After 22 h, the cultures were diluted (1:100) into fresh medium in a new 96-well plate and grown identically. The old plates were stored at −80°C with addition of 100 μl 50% glycerol–100 mM MOPS (pH 7.5) to each well population. Diluting the evolving cultures 1:100 resulted in approximately 6.6 generations per day. If a culture did not grow, the experiment was restarted by taking a 1:50 dilution from the most recent frozen plate into fresh medium. All populations underwent the same number of generations.
The cells were initially subcultured in LBK with 150 mM TAPS (pH 9.0) and serially subcultured for an estimated 470 generations, at which point the buffer was changed to 150 mM AMPSO to better accommodate an increase in the pH of the medium to 9.2. Cells were subcultured at pH 9.2 for an additional 730 generations; then, at generation 1,200, the pH was increased to 9.25, and at generation 1,900 the pH was increased to 9.3. The experiment was terminated after 2,200 generations, and isolates were obtained for genome sequencing.
Whole-genome sequencing.The genomes of eight evolved strains and of the ancestral strain W3110 were resequenced (Table 1). The DNA from each strain was extracted using a DNeasy kit (Qiagen). Sequencing was performed by the Research and Technology Support Facility at Michigan State University. Libraries were prepared using the Illumina TruSeq Nano DNA library preparation kit. Illumina MiSeq 2 × 250 base calling was performed by Illumina Real Time Analysis (RTA) v1.18.54, and the output of RTA was demultiplexed and converted to Fastq format with Illumina Bcl2fastq v1.8.4. The number of reads for each of 8 genomes ranged from 900,000 to 1,900,000. Average quality scores ranged from Q30 to Q36.
Genome sequences were aligned to the reference E. coli K-12 W3110 (GenBank accession number NC_007779.1), and mutations were predicted using the breseq 0.30.0 computational pipeline (23, 47). Unassigned alignment variants were inspected for deletions and insertions.
Strain construction.P1 phage transduction was conducted to construct deletion strains with kanR insertions (24) using kanR replacement strains obtained from the Keio collection (Coli Genetic Stock Center, Yale University) (48).
A ΔtorCAD deletion was constructed by recombineering (49, 50). The counterselectable cat-sacB cassette was PCR amplified from colonies of E. coli strain XLT241 using primers that have 5′ homology to the flanking regions of torCAD (5′-TGTTCATATTTGCTCATTAAGATCGCTTCACTAAACCATAATTCTATGTGACGGAAGATCACTTCGCA-3′ and 5′-ATGCCTGATGCGACGCTAAACGCGTCTTTTCAGGCGCACAAACATCAAAGGGAAAACTGTCCATATGCA-3′). High-pH-evolved isolate G9-1 transformed with the heat-inducible pSIM6::ampR plasmid was cultured to mid-log phase at 32°C and then transferred to 42°C to induce recombineering proteins. Cells were made electrocompetent and electroporated with the PCR product of cat-sacB containing flanking homology to torCAD. The transformed cells were outgrown at 32°C for 3 to 5 h and plated on LB containing 10 μg/ml chloramphenicol. Successful colonies were screened for resistance to chloramphenicol and sensitivity to sucrose. To remove the cat-sacB cassette, the successful colonies were made electrocompetent, electroporated with a DNA oligonucleotide with flanking homology to torCAD (5′-TGCTCATTAAGATCGCTTCACTAAACCATAATTCTAGTTTGTGCGCCTGAAAAGACGCGTTTAGCGTCGCA-3′), and then plated on LB agar plates lacking NaCl but containing 6% sucrose. Successful colonies were further screened for resistance to sucrose and sensitivity to chloramphenicol. The torCAD deletion was then confirmed by PCR sequencing (with 5′-CTGTTCATATTCTGCCGTAAGC-3′ and 5′-GTGTTTGCCGGATGCGATGA-3′).
Strains NM376 and 378 were constructed by recombineering a 3-way PCR product into strain NM1100 (35). In a first step, PCR was used to amplify a zeocin cassette with a 5′-end 40-nt homology to the end of lacI, using oligonucleotides lacI-Zeo_inv.F1 (GCTGGCACGACAGGTTTCCCGACTGGAAAGCGGGCAGTGAAAAAAAAGCCCGCTCATTAG) and zeo-rpoS_inv.R1 (GCAAAATAACCAGGAAAAATCCAGGTATTTCCTCACGTTTGTTGACAATTAATCATCGGC). In parallel, a second PCR product was created carrying the rpoS fragment that has 40 nt of homology at its 5′ end to the 3′ end of the zeocin cassette and 40 nt of homology to the +1 of lacZ at the 3′ end, using oligonucleotides Zeo-rpoS_inv.F2 (GCCGATGATTAATTGTCAACAAACGTGAGGAAATACCTGG) and either rpoS-lacZ.R2a for NM376 (ATAGCTGTTTCCTGTGTGAAATTGTTATCCGCTCACAATTACTCTGTTCACCCGAAGACT) or rpoS-lacZ.R2b for NM378 (ATAGCTGTTTCCTGTGTGAAATTGTTATCCGCTCACAATTGCTCGGTTGTGCTTGCTGTT). The two PCR primers were then mixed together into a third PCR mixture, using flanking primers (lacI-Zeo inv.F1 with either rpoS-lacZ.R2a or rpoS-lacZ.R2b) to create one fragment carrying both the zeocin cassette and appropriate rpoS fragment. This final PCR product was then used in a recombineering reaction into NM1100 as described elsewhere (35), selecting for clones growing on LB-zeocin plates. All constructs were checked by sequencing.
Measurement of high-pH growth.The eight evolved strains (B11-1, B11-2, D9-1, D9-2, F7-1, F7-2, G9-1, and G9-2) (Table 1), the ancestral strain (W3110), and the transduced strains were cultured in LBK and buffered with 100 mM TAPS at a pH of 8.5 for 16 to 18 h at 37°C with shaking. Initial overnight growth with a milder stress increases the consistency of quality of the overnight culture, which varies depending on the size of the colony picked. The cultures were diluted into LBK buffered with 150 mM AMPSO at pH 9.25 at a 1:200 concentration in a 96-well plate, with eight replicate wells for each strain. Plates were incubated in the SpectraMax Plus384 MicroPlate reader (Molecular Devices) for 22 h at 37°C. The OD600 was measured every 15 min with shaking for 3 s before reads, and data were compiled using Softmax Pro version 6.4.2. Endpoint density was measured at 600 nm after 16 h. The pH of the culture medium was measured before and after growth. The final pH was measured by combining the solutions of the eight replicate wells for each strain. pH measurements were performed at room temperature. The change in pH during growth was <0.2 unit.
The rate of doublings per hour (k) was calculated during logarithmic growth phase. For each strain, eight replicates were tested from independent overnight cultures. For statistical analysis, analysis of variance (ANOVA) and Tukey honestly significant difference (HSD) tests were conducted on the log-phase growth rates and on the log2 values of the 16-h culture densities.
β-Galactosidase assay for RpoS induction.Overnight cultures were incubated for 16 to 18 h in LBK with 100 mM MOPS (pH 7.0) and then diluted (1:1,000) into fresh medium to minimize residual levels of stationary-phase RpoS. Upon reaching an OD600 of 0.1, each culture was split via filtration and suspended into equal 30-ml volumes of LBK–100 mM MOPS (pH 7.0) and LBK–100 mM TAPS (pH 9.0). Samples were taken at the time of initial exposure and subsequently at timed intervals during aerated growth in a rotating water bath set to 37°C. Samples were assayed, and enzyme activity was measured (51). For statistical analysis, ANOVA and Tukey HSD tests were conducted on the ratios of enzyme activity (pH 9.0 versus pH 7.0) from three independent trials for the assay times presented.
Western blotting for RpoS induction.Strains were cultured under conditions identical to those for the β-galactosidase assay for RpoS induction, and samples for both assays were taken simultaneously (see above). Cells were precipitated with trichloroacetic acid (5% total volume) and washed with 80% acetone. Samples were suspended in SDS loading buffer (Invitrogen) and normalized according to the OD600. Samples were isolated using a 12% Bis-Tris gel (Invitrogen) and transferred onto a nitrocellulose membrane (Invitrogen). Membranes were probed with a 1:5,000 dilution of anti-RpoS antibody, and loading was normalized to an EF-Tu control. Images were obtained with a ChemiDoc MP imaging system (Bio-Rad).
Catalase assay of RpoS activity.Strains were streaked on LB plates and incubated overnight at 37°C, and a colony was picked from a plate and submerged into 1 ml of 3% H2O2 without mixing. Bubbles (O2 formation) were immediately evaluated on a scale of 0 (none) to 5 (wild-type strain) by an observer who was not told the identity of the strains; three independent trials were done, using different colonies and different observers for each, and their ratings were averaged. The method was as described at http://learn.chm.msu.edu/vibl/content/catalase.html.
Accession number(s).Sequence data have been deposited in the NCBI Sequence Read Archive (SRA) under accession number SRP074502.
ACKNOWLEDGMENTS
This work was funded by National Science Foundation grant MCB1613278 and in part by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.
We thank Lynn Thomason and Jeremy Moore for the gift of strains and instruction in recombineering. We thank Francois De Mets and Thomas Delerue for providing the evaluation of the levels of catalase in the evolved strains.
FOOTNOTES
- Received 2 March 2018.
- Accepted 14 May 2018.
- Accepted manuscript posted online 25 May 2018.
- Copyright © 2018 American Society for Microbiology.
