Multitier regulation of the E. coli extreme acid stress response by CsrA

ABSTRACT CsrA is an RNA-binding protein that regulates processes critical for growth and survival, including central carbon metabolism, motility, biofilm formation, stress responses, and expression of virulence factors in pathogens. Transcriptomics studies in Escherichia coli suggested that CsrA repressed genes involved in surviving extremely acidic conditions. Here, we examine the effects of disrupting CsrA-dependent regulation on the expression of genes and circuitry for acid stress survival and demonstrate CsrA-mediated repression at multiple levels. We show that this repression is critical for managing the trade-off between growth and survival; overexpression of acid stress genes caused by csrA disruption enhances survival under extreme acidity but is detrimental for growth under mildly acidic conditions. In vitro studies confirmed that CsrA binds specifically to mRNAs of structural and regulatory genes for acid stress survival, causing translational repression. We also found that translation of the top-tier acid stress regulator, evgA, is coupled to that of a small leader peptide, evgL, which is repressed by CsrA. Unlike dedicated acid stress response genes, csrA and its sRNA antagonists, csrB and csrC, did not exhibit a substantial response to acid shock. Furthermore, disruption of CsrA regulation of acid stress genes impacted host-microbe interactions in Caenorhabditis elegans, alleviating GABA deficiencies. This study expands the known regulon of CsrA to genes of the extreme acid stress response of E. coli and highlights a new facet of the global role played by CsrA in balancing the opposing physiological demands of stress resistance with the capacity for growth and modulating host interactions. IMPORTANCE To colonize/infect the mammalian intestinal tract, bacteria must survive exposure to the extreme acidity of the stomach. E. coli does this by expressing proteins that neutralize cytoplasmic acidity and cope with molecular damage caused by low pH. Because of the metabolic cost of these processes, genes for surviving acid stress are tightly regulated. Here, we show that CsrA negatively regulates the cascade of expression responsible for the acid stress response. Increased expression of acid response genes due to csrA disruption improved survival at extremely low pH but inhibited growth under mildly acidic conditions. Our findings define a new layer of regulation in the acid stress response of E. coli and a novel physiological function for CsrA.

The central component of the Csr system is CsrA, a homodimeric protein that recognizes and binds to specific RNA sequences (1,7).CsrA generally represses stress respon ses and systems associated with the stationary phase of growth while activating the expression of genes associated with exponential growth (7,8).CsrA-mediated regulation involves binding to sites containing a critical GGA motif, which is often found in the single-stranded loop of an RNA hairpin (9,10).These binding sites are typically located in the 5′ leader or early mRNA coding regions.CsrA binding can regulate translation initiation, RNA stability, riboswitch activity, or transcription elongation (5,7,(11)(12)(13)(14)(15)(16)(17)(18).
The expression of csrA is tightly regulated, both transcriptionally and posttranscrip tionally (19).In addition, CsrA activity is extensively regulated by other components of the Csr system.In E. coli, two small RNA (sRNA) antagonists, CsrB and CsrC, contain multiple high-affinity CsrA binding sites that act to sequester CsrA from other regulatory targets (20,21).csrB/C transcription is activated in response to environmental stresses, the accumulation of metabolic end products such as acetate, and quorum sensing (22,23).CsrB/C levels are also regulated by CsrD via RNase E-dependent turnover, which is activated by glucose in E. coli (24)(25)(26)(27).These findings imply that elevated CsrB/C levels cause decreased CsrA activity under environmental stresses and upon metabolic end product accumulation but increased activity in the presence of preferred carbon sources.
Recent high-throughput studies identified CsrA as a likely regulator of acid stress resistance genes in E. coli (Fig. 1) (3,9,(28)(29)(30).Despite being characterized as a neutro phile, E. coli can survive extremely acidic conditions for extended periods of time (31).Tolerance for extremely acidic environments is thought to protect against exposure to gastric acidity (32) and contribute to the low oral infectious dose of certain E. coli pathogens (33,34).
FIG 1 Hypothetical targets of CsrA-dependent regulation in the EvgA-YdeO-GadE circuit.RNA-seq studies include Log 2 fold changes > 1 (3,24,28).Pulldown refers to CsrA-bound RNA targets (29).In silico predictions refer to the computational identification of putative CsrA-binding sites (30).CLIP-seq (3) and HITS-CLIP-seq (28) refer to CsrA-binding targets identified by these methods.Blue squares indicate that the gene was identified as a possible regulatory target of CsrA in the respective study.
Transcriptional regulation of GDAR and ydeP is complex and involves several circuits such as EvgA-YdeO-GadE that activate the expression of gadA, gadBC, and ydeP during the exponential phase (Fig. 1) (39,50).The EvgA-YdeO-GadE circuit responds to mildly acidic pH when the sensor kinase EvgS phosphorylates the response regulator EvgA (Fig. 1) (51,52).EvgA in turn activates the transcription of ydeP and overexpression of evgA leads to increased expression of gadA and gadB (53).EvgA also activates the transcrip tion of ydeO and gadE.YdeO is a transcription factor that activates gadE transcription, represses its expression, and indirectly represses ydeP expression (39,50,54).GadE is a transcription factor that is critical for the expression of the GDAR system and for activating its expression.GadE is under complex transcriptional and posttranscriptional regulation involving over a dozen regulators (55)(56)(57).
The goal of our study was to investigate CsrA-dependent regulation of the acid stress response in E. coli.We demonstrate that CsrA represses acid stress circuitry at multiple levels.Disruption of CsrA causes overexpression of this circuitry, conferring a survival advantage under extremely acidic conditions but a growth defect under mildly acidic conditions.Our findings highlight a new and important role played by CsrA in managing the trade-off between bacterial growth and stress survival.

Disruption of csrA causes a pH-dependent growth defect
Recent studies identified several likely mRNA targets of CsrA-dependent regulation involved in growth and survival in acidic conditions including evgA and gadA (Fig. 1; Table S1) (3,9,(28)(29)(30).To assess the physiological implications of CsrA-dependent regulation of acid stress systems, wild-type (WT) E. coli strain MG1655 and its isogenic csrA::kan mutant were grown under different pH conditions ranging from 7.5 to 5 (Fig. 2A; Table S1).At pH 6 or below, the csrA mutant displayed a strong growth defect that intensified as the pH decreased further.While the growth rate of the WT strain also decreased under acidic conditions, the effect was substantially weaker.Complementing the csrA mutation with a plasmid-borne wild-type csrA gene restored WT growth at pH 5.5, confirming the role of CsrA in this phenotype (Fig. S2).These results suggest that CsrA represses genes detrimental for the growth under acidic conditions, activates genes beneficial for growth under acidic conditions, or both.Interestingly, pH 6 is the upper limit of activation for the EvgS sensor kinase and increased activity and expression of the EvgAS TCS are associated with a pH-dependent growth defect, suggesting that CsrA may negatively regulate EvgAS function or downstream regulatory targets of EvgAS (47,49).

CsrA regulates survival under extremely acidic conditions
When E. coli cells were grown in neutral media (pH 7) and then shocked by exposure to extremely acidic conditions (pH 2.5) in the absence of exogenous glutamate, little to no survival was observed for WT or csrA-mutant strains (Fig. 2C).However, when shocked under extremely acidic conditions in the presence of glutamate, mean survival of the csrA mutant was ~12 fold higher than WT, suggesting that the GDAR system is more highly expressed or more active in the csrA mutant (Fig. 2D; Table S1).Complementation of the csrA mutant with plasmid-borne csrA reversed this survival phenotype, confirming the role of CsrA in the acid stress response.
When cells were grown in mildly acidic media (pH 5.5), preadapting them before challenge with extreme acidity, the csrA mutant exhibited a much higher survival rate than WT in all conditions (Fig. 2B, E, and F).Without added glutamate, the csrA mutant survived ~100-fold better than WT, and with added glutamate ~5-fold better than WT (Fig. 2E and F).Complementation of the csrA mutation reversed these phenotypes (Fig. 2).
The effect of CsrA on survival also extended to a condition designed to simulate fasted human stomach contents after drinking a glass of water (Biorelevant Media).After a 10-min challenge, the csrA mutant survived ~55-fold better than WT.Complementation of csrA abolished this survival phenotype (Fig. 2B).These results suggest that CsrA likely impacts survival in stomach acidity during host colonization.Survival under extremely acidic conditions in the absence of external amino acids has been associated with evgA and ydeP overexpression (49), offering a possible explanation for this phenotype.

Acid stress phenotypes of the csrA mutant depend on the EvgA-YdeO-GadE circuitry
We used epistasis analysis to identify genes responsible for the pH-dependent physiol ogy of the csrA mutant (Fig. 1; Table S1).Deletion of evgS, gadB, or gadC did not suppress the growth defect of the csrA mutant (Fig. 3A, G, and H).However, deletion of several other genes in the EvgA-YdeO-GadE circuit suppressed the pH-dependent growth defect of the csrA mutant grown under mildly acidic conditions, including evgA, ydeO, ydeP, gadE, and gadA (Fig. 3B through F; Fig. S3).Complementing these genes in mutant strains restored the growth defect of the csrA mutant.Previous studies indicated that overex pression of evgA and ydeP impaired the growth of an otherwise WT strain (49), similar to the growth defective phenotype observed in the csrA mutant.Our results suggest that the EvgA-YdeO-GadE circuit participates in the pH-dependent growth defect of the csrA mutant and imply that CsrA may regulate the EvgA-YdeO-GadE circuit.Deletion of evgS, gadB, or gadC did not suppress the growth defect of the csrA mutant (Fig. 3A, G, and H).
We next tested whether the deletion of genes in the EvgA-YdeO-GadE circuit affected the survival of WT and csrA mutant strains that were preadapted by growth under mildly acidic conditions (pH 5.5) and then subjected to extremely acidic conditions (pH 2.5).Deleting evgS, evgA, ydeO, ydeP, or gadA abolished the survival of WT and csrA mutant strains in the absence of exogenous glutamate (Fig. 4A and C).Deleting gadE reduced the survival of WT and the csrA mutant under extremely acidic conditions with or without added glutamate (Fig. 4C and D).Interestingly, deleting gadB did not affect survival with or without glutamate, whereas deleting gadA resulted in a survival defect when grown without glutamate but not in its presence (Fig. 4C and D).
Deleting evgS in a csrA mutant abolished the survival phenotype but it did not alter the growth defect of this strain (Fig. 3A and 4A).This paradox may be due to the influence of two factors: in the absence of EvgS, EvgA can be phosphorylated by acetylphosphate and EvgS is thought to be necessary for dephosphorylation of EvgA (58)(59)(60).stomach acid (pH 1.6) for 10 min.(C) Cells grown exponentially under neutral conditions and then challenged by extremely acidic conditions without glutamate (0 mM glutamate) for 2 h.No survival of any strain was seen under these conditions.(D) Cells grown exponentially under neutral conditions and then challenged by extremely acidic conditions with 1.5 mM glutamate for 2 h.(E) Cells grown exponentially under mildly acidic conditions and then challenged by extremely acidic conditions without glutamate for 2 h.(F) Cells grown exponentially under mildly acidic conditions and then challenged by extremely acidic conditions with 1.5 mM glutamate for 2 h.Error bars represent standard deviation (sd) from at least three independent experiments.Statistical significance was determined using unpaired t-tests and is denoted as follows: *P < 0.05; **P < 0.01; ***P < 0.001.Perhaps, sufficient phosphorylated EvgA accumulates in an evgS knockout to cause impaired growth of the csrA mutant but not enough to increase survival of the csrA mutant.
Glutamate supplementation eliminated survival defects of the evgS, evgA, ydeO, ydeP, and gadA deletions in WT and csrA mutant backgrounds, consistent with similar findings from prior studies (Fig. 4B and D); however, gadA was not tested during exponential growth in those studies (39,48,49).Deleting either gadA or gadB when supplemen ted with glutamate had little to no effect on the survival of the WT or csrA mutant strains.However, deleting both gadA and gadB in the WT and csrA mutant backgrounds abolished survival in the presence or absence of glutamate (Fig. 4C and D), highlighting the redundant nature of these genes.Deleting gadE impaired the survival of the WT and csrA mutant strains when supplemented with glutamate (Fig. 4D), similar to previous observations (61).

CsrA binds to 5′ leaders of mRNAs in the EvgA-YdeO-GadE circuit
While previous CLIP-Seq and RNA pulldown studies suggested that CsrA binds to evgA, gadE, gadA, and gadB mRNAs, these in vivo studies did not determine if CsrA binds independently of other factors or assess the binding affinity and specificity (3,28,29).Hence, gel shift assays were utilized to explore direct CsrA binding to the 5′ leader of transcripts in the EvgA-YdeO-GadE circuit independently of other factors (Fig. 5A).
Previous studies demonstrated that CsrA-mediated regulation of complex genetic circuitry can occur via the top-tier regulator, as is the case for the regulation of motility via flhDC (4) or biofilm regulation via nhaR (62).EvgA appears to be the top-tier regulator of the acid stress response and is the first gene of the evgA-evgS operon.CsrA bound with high affinity to the 5′ leader of the evgA transcript (Fig. 5B), with an apparent K D of 7 nM.As the CsrA concentration was increased, two shifted species were observed, suggesting that more than one CsrA dimer can bind to the evgA 5′ leader transcript.This RNA contains 4 GGA motifs that might be components of CsrA-binding sites (Fig. 5A).Competition assays with unlabeled specific (evgA) and nonspecific (phoB) RNA estab lished the specificity of this interaction.
CsrA also bound to the 5′ leaders of gadA, gadB, gadE, and ydeO with apparent K D values of 73 nM, 49 nM, 138 nM, and 84 nM, respectively (Fig. 5B).Two distinct shifts were observed for the gadA and gadB transcripts, suggesting that more than one CsrA dimer may bind to these transcripts.Given that the gadA RNA contains 3 GGAs and gadB contains 4 GGAs, this explanation is plausible.Competition assays indicated that CsrA binding to both gadA and gadB RNA was specific (Fig. 5B).On a native gel, the ydeO transcript adopted two forms (F1 and F2), and CsrA bound to both forms.Because this transcript produced a single band on a denaturing gel (data not shown), these forms represent alternative RNA conformations.The potential regulatory implications of these two RNA conformations are not known.CsrA did not bind to the 5′ leader of ydeP, suggesting that ydeP is not a direct target of CsrA-dependent regulation.
Together, these findings reveal that CsrA binds directly to the transcripts of the EvgA-YdeO-GadE circuit in the absence of other factors.Interestingly, CsrA bound most tightly to mRNA of the top-tier regulator EvgA and with weaker affinity to mRNAs of down stream genes within the circuit.

CsrA regulates evgA translation via translational coupling to a small leader peptide
Having identified the regulatory circuitry of interest for CsrA effects on acid stress resistance and the genes that CsrA likely regulates directly, we next assessed the effects of CsrA on gene expression using lacZ reporter fusions.First, evgA expression was examined under neutral and acidic conditions throughout the growth cycle in WT and csrA mutant backgrounds.Expression of an evgA'-'lacZ translational fusion was higher in the csrA mutant under neutral and acidic conditions (Fig. 6A and B).In addition, a P lacUV5 -evgA′-′lacZ leader fusion was tested in which the evgA promoter was replaced with the lacUV5 promoter.This promoter is unaffected by the csrA::kan mutation either directly or indirectly and was used to assess posttranscriptional regulation.The leader fusion was expressed at a higher level in the csrA mutant (Fig. 6C and D).Both the translational and leader fusions for evgA showed a similar level of higher expression (~2 fold) in the csrA mutant, suggesting that the difference of expression of evgA'-'lacZ in the csrA mutant is  likely a result of disrupted posttranscriptional regulation rather than transcriptional regulation.To determine whether CsrA directly represses evgA translation, a P T7 -evgA'-'lacZ translational fusion expressed from a T7 promoter was used in the PURExpress system.The decrease in expression caused by increasing concentrations of CsrA indicated that CsrA directly represses evgA translation (Fig. 7B).Recent studies identified a leader peptide, EvgL, in the evgA leader mRNA such that the evgL stop codon overlaps with the start codon of evgA, suggestive of translational coupling (9,63).We investigated the possibility that evgA translation is coupled to that of evgL by determining if evgL is translated and then by examining the impact of evgL translation on evgA translation.Hence, we constructed an evgL'-'lacZ translational fusion and found that expression was regulated by CsrA under both neutral and acidic conditions in a pattern similar to the expression of evgA (Fig. 6E and F).To assess the effects of evgL translation on evgA translation, we changed the start codon of evgL to a stop codon in the context of an evgA'-'lacZ translational fusion.When evgL trans lation was disrupted by the stop codon, evgA'-'lacZ expression was eliminated, which is indicative of translational coupling (Fig. 6G).Similar CsrA-mediated regulation via translational coupling of iraD to a short leader peptide was observed previously (64).
RNA footprint and toeprint analyses of CsrA binding suggested that CsrA-mediated regulation of evgA expression involves a more complex mechanism.Footprinting with RNase T1 showed CsrA-dependent protection of GGA4, which overlaps the evgA SD sequence.Footprints were not observed for the other GGA motifs (Fig. 7E).However, toeprinting assays, which identify the 3′ boundary of a bound protein or stable RNA secondary structure (65), suggested that CsrA has four binding sites in the evgA 5′ leader.The use of Avian Myeloblastosis Virus reverse transcriptase (AMV) indicated that CsrA interacts with the two GGAs in the 5′ leader that overlap the evgL SD sequence (GGA2) and just upstream of its start codon (GGA3), as well as the GGA overlapping the evgA SD sequence (GGA4), while toeprinting with SuperScript III reverse transcriptase (SSIII) suggested that CsrA interacts with GGA1 in the 5′ leader (Fig. 7C and D).These results imply that CsrA regulates the translation of evgA directly and via translational coupling with evgL.

CsrA regulates the expression of the EvgA-GadE-YdeO circuit at multiple levels
We next assessed the effects of CsrA on the expression of other acid stress resistance genes using lacZ translational fusions for mid-level regulators, ydeO and gadE, and for genes under the control of EvgA (ydeP) or GadE (gadA and gadB).The ydeO'-'lacZ fusion was expressed at a higher level in the csrA mutant strain under neutral (pH 7) and mildly acidic conditions (pH 5.5) (Fig. 8A and B).Expression of the gadE'-'lacZ translational fusion was similar in WT and csrA mutant strains under neutral conditions during exponential growth, but the expression in the csrA mutant was higher in the stationary phase (Fig. 8C).By contrast, under mildly acidic conditions, gadE'-'lacZ expression was higher in the csrA mutant strain throughout growth (Fig. 8D).
Expression of both gadA'-'lacZ and gadB'-'lacZ translational fusions exhibited the effects of CsrA (Fig. 8E through H), the details of which helped to explain their differential effects on physiology, where GadA played an important role in the growth defect of the csrA mutant but GadB did not (Fig. 3).Under mildly acidic conditions, both fusions were expressed at higher levels in the csrA mutant during the exponential phase of growth.At maximal expression in the csrA mutant (~4 h), the gadA'-'lacZ fusion was expressed at a level that was ~2.5-fold higher than the gadB'-'lacZ fusion.In addition, complementing the gadA knockout with gadB on a plasmid reintroduced the growth defect of the csrA mutant, suggesting that the level of expression, and not the particular isozyme, is responsible for the effects on growth (Fig. S4).Under neutral conditions, gadB'-'lacZ expression showed little to no effect of CsrA during the exponential growth but expression was higher in the csrA mutant in the stationary phase, while expression of the gadA'-'lacZ fusion was comparable in the WT and csrA mutant strains (Fig. 8E and G).
Interestingly, the ydeP'-'lacZ translational fusion was expressed at a higher level throughout growth in the csrA mutant strain under both neutral and mildly acidic conditions (Fig. 8I and J).Since CsrA did not bind to the ydeP 5′ leader (Fig. 5B), these results imply that CsrA-dependent regulation is mediated indirectly, likely via direct effects of CsrA on evgA expression.
The translational fusion assays showed that the disruption of CsrA-dependent regulation results in increased expression of genes in the EvgA-YdeO-GadE circuit under acidic conditions, highlighting the importance of CsrA-dependent regulation of the acid stress response.As described above, we observed increased expression of evgA in the csrA mutant (Fig. 6A and B).While evgA did not respond to acidic conditions in the WT background, this was not the case in the csrA mutant in which expression increased earlier during growth at ~3 h (Fig. 6A and B).This lack of an evgA response to mildly acidic conditions was observed in previous studies using transcriptional fusions, despite the fact that EvgA is reported to be autoregulatory (50,52,66).To investigate whether CsrA regulated the circuitry downstream of GadA primarily via evgA, we examined the effects of knocking out evgA on the expression of a gadE'-'lacZ translational fusion.Deletion of evgA did not eliminate the effects of CsrA on gadE expression (Fig. S5).This result is consistent with the gel shift data indicating that CsrA binds directly to the gadE transcript (Fig. 5B) and confirms the multi-tier nature of CsrA-dependent regulation of this circuit.However, gadE is subject to several transcription activation and repression mechanisms and is itself positively autoregulatory, which may complicate the interpreta tion of CsrA effects on gadE expression in an evgA mutant (50,67).

CsrA directly regulates gadA and gadB
We next examined the molecular mechanism of CsrA-dependent regulation of gadA and gadB expression.To determine whether CsrA repressed translation of gadA and gadB, we constructed plasmids with P T7 -gadA'-'lacZ and P T7 -gadB'-'lacZ translational fusions in which the 5′ leaders were fused to a T7 promoter and assessed their expression in the PURExpress system.In both cases, CsrA repressed gadA and gadB, indicative of direct translational repression (Fig. 9B).
We also performed RNA toeprinting and footprinting experiments to further investi gate the regulatory mechanisms by which CsrA represses gadA and gadB translation.CsrA-dependent toeprints were observed just downstream of two GGA motifs in the gadA transcript.One GGA overlaps the gadA SD sequence (GGA1) and the second is in the early coding sequence (GGA3).A toeprint was not observed corresponding to GGA2 (Fig. 9A and C).In the case of gadB, CsrA-dependent toeprints were observed in regions that overlap the SD sequence (GGA1), and two in the early coding region (GGA3 and GGA4).A toeprint was not observed corresponding to GGA2 (Fig. 9A and D).To further examine the sites of CsrA binding, RNase T1 footprinting was performed.In both cases, this analysis suggested that the two most important CsrA interaction sites corresponded to GGA1, which overlaps the respective SD sequences, and GGA3, which is in the early coding sequences (Fig. 9E and F).These binding patterns are somewhat similar to CsrA-sdiA interaction, although CsrA binds exclusively within the coding region of sdiA (5).

Acid shock does not trigger the expression of csr genes
Previous studies on the Csr system demonstrated that it is often subject to feedback loops in which CsrA represses the expression of genes involved in activating transcrip tion of CsrB and CsrC sRNAs (2,5,7).Furthermore, while acid stress does not appear to affect CsrA transcription (50), the protonated forms of formate or acetate stimulate transcription of csrB and csrC by binding to the sensor-kinase BarA of the BarA-UvrY TCS which phosphorylates the transcription factor UvrY, which, in turn, activates csrB/C transcription (22).We therefore explored the effects of a pH shift from 7 to 5.5 on the expression of a csrA'-'lacZ translational fusion, as well as on the expression of csrB-lacZ and csrC-lacZ transcriptional fusions in exponentially growing cells.A ydeP'-'lacZ translational fusion served as a positive control, as ydeP expression increases in response to a decrease in pH (50,51).The ydeP'-'lacZ fusion showed the expected response (Fig. S6A).Expression of the csrA'-'lacZ fusion was unaffected by the decrease in pH, consistent with previous observations on its transcription (Fig. S6B) (50).Expression of the csrC-lacZ fusion was not affected by the decrease in pH, whereas the csrB-lacZ showed a small increase in expression, starting after 25 min of exposure to acidic conditions (Fig. S6C and  D).These results reveal that the key genes of the Csr system do not respond or respond weakly to acid stress.

The csrA mutant alleviates GABA deficit in Caenorhabditis elegans
The csrA mutant overexpresses genes involved in the production of GABA, a neurotrans mitter that regulates neuronal excitability in certain organisms including C. elegans.We investigated whether the csrA mutation would increase GABA production sufficiently to restore a GABA deficiency in C. elegans by assessing its effect on a GABA-dependent biological response.Nematodes harboring loss-of-function mutations in the unc-25 gene are deficient in endogenous GABA synthesis and exhibit pronounced convulsions upon exposure to pentylenetetrazole (PTZ), a competitive GABA A receptor antagonist that is used to chemically induce seizures to model epilepsy and other seizure disorders (68)(69)(70)(71).Animals deficient in GABA lack the inhibitory signaling necessary to prevent the excitatory effects of PTZ on cholinergic signaling, thereby leading to PTZ-induced convulsions (71).Prior research demonstrated that exogenous GABA supplementation effectively restores normal phenotypes in unc-25 mutants (72), and dietary utilization of a GABA-producing E. coli strain has neuroprotective effects in nematodes (73).
After 15 min of PTZ exposure, the number of convulsing unc-25 nematodes fed the WT E. coli strains OP50 or MG1655 was approximately fivefold higher than nematodes fed the csrA mutant derivative of MG1655 or those exogenously supplemented with GABA (Fig. 10).We used epistasis analysis to identify the genes responsible for the restorative effects of the csrA mutant on the GABA-deficient phenotype and discovered that deletion of either of the GABA producing glutamate decarboxylases, gadA or gadB, dampened the effects of the csrA mutant on convulsions (Fig. 10).Interestingly, the gene encoding the glutamate-GABA antiporter gadC appeared to have limited involvement in the restorative phenotype afforded by the csrA mutant, perhaps due to the activity of additional GABA transporters, such as the GABA permease YhiM (74).Together, these findings underscore the potential therapeutic relevance of csrA and bacterial GABA production in host disorders where GABA plays a critical biological role.

DISCUSSION
Despite being a neutrophile, E. coli is capable of surviving extremely acidic conditions for extended periods of time through its expression of several acid resistance systems (31,38).These systems are critical for survival in extremely acidic conditions but are detrimental to growth under mildly acidic conditions, presumably due to the metabolic drain that they exert (39,48,50,75,76).As a result, the acid stress systems are subject to complex and fine-tuned transcriptional regulatory circuits that are still not fully understood despite having been extensively studied (50,55,61,77).In compar ison, posttranscriptional regulation of the acid stress response has not been exten sively studied, despite evidence suggesting that posttranscriptional regulation plays an important role in modulating the expression of many genes, which is the basis of CsrA-mediated regulation (3,28,36,57,78,79).
We demonstrate that CsrA represses the EvgA-YdeO-GadE circuit of E. coli at multiple levels, from the top-tier regulator evgA to the structural genes gadA and gadB.Expres sion and activity of this circuitry are triggered by mildly acidic (pH 5-6) conditions, in anticipation of more extreme conditions (pH 1-3).CsrA is critical in preventing overexpression of the EvgA-YdeO-GadE circuit, and, in turn, preventing growth defects under mild acidity.CsrA deficiency results in severe growth defects under mildly acidic conditions and increased acid resistance in extremely acidic conditions, highlighting the critical role played by CsrA in balancing the trade-off between growth and survival.
Epistasis analysis showed that CsrA-dependent regulation of the EvgA-YdeO-GadE circuit was multitier in nature, and deleting evgA did not completely abolish the overexpression of gadE'-'lacZ in a csrA mutant background (Fig. S5).Given that CsrA binds directly to gadE mRNA, the latter results were not surprising and demonstrate that CsrA represses gadE expression at more than one level, and not simply via effects on EvgA (Fig. 5B; Table S5).
High-affinity binding of CsrA to the evgA, gadE, ydeO, gadB, and gadA transcripts and in vivo lacZ reporter results also indicate that CsrA is involved in binding to and repressing multiple genes of the EvgA-YdeO-GadE circuit (Fig. 5, 7 and 9).The binding data are consistent with in vitro expression results, which demonstrated that CsrA directly repressed not only the top-tier regulator evgA but also the structural genes gadA and gadB.In the case of gadA and gadB, toeprinting showed that CsrA interacted with CsrA binding sites overlapping the SD sequence and the early coding sequences, which was supported by footprinting assays (Fig. 7 and 9).In the case of evgA, toeprinting and footprinting showed that CsrA interacted at multiple sites raising the possibility of a complex mechanism.This hypothesis is consistent with the results of lacZ reporter assays showing that translation of the leader peptide evgL is required for translation of evgA, implying that CsrA-dependent regulation of evgA involves translational coupling of the two genes, which is similar to CsrA-dependent regulation of iraD expression (64).However, unlike iraD where CsrA binds entirely upstream of the leader peptide coding sequence, CsrA binds both upstream and within the coding sequence of evgL.CsrA-dependent regulation of evgLA may require binding of a CsrA dimer to one site and then bridging to a second site in the segment preceding the evgL coding sequence as demonstrated previously (80).Doing so may enable another CsrA dimer to bind to a segment preceding evgL and to the site overlapping the evgA SD sequence, which is also within the coding region of evgL.This binding pattern would be consistent with gel shift results showing that CsrA binding to the 5′ leader of evgA results in at least two different shifts (Fig. 6).However, further analysis is required to fully elucidate the regulatory mechanism.
Interestingly, gadA but not gadB played an important role in the physiology of the csrA mutant under acidic conditions, affecting both growth and survival.These two isozymes were considered as interchangeable due to their high sequence similarity and similar phenotypes in the stationary phase (81).Complementing a gadA knockout with gadB restored the pH-dependent growth defect of the csrA mutant, suggesting that the combined level of GadA/B activity is more important for this growth phenotype than the particular isozyme (Fig. S4).This interpretation is consistent with the lacZ reporter results under acidic conditions in a csrA mutant background, where gadA'-'lacZ was more highly expressed than gadB'-'lacZ (Fig. 8).GadA also appears to play a role in survival under extremely acidic conditions without glutamate supplementation in exponentially growing cells.Previous work quantifying the amount of GadA and GadB produced in E. coli ATCC 11246, which is closely related to E. coli MG1655, showed that 80% of the isolated protein was GadA, explaining the disproportionate role of GadA in surviving extreme acidity (82,83).While mutating gadA and gadB had differing effects on the physiology in a csrA mutant background, this pattern did not extend to the C. elegans studies.However, it is important to note that the latter studies were conducted on agar plates, where the cells were presumably in a stationary phase.Previous studies examining acid stress survival in the stationary phase noted no difference in the effects of gadA and gadB, suggesting that their distinct physiology may be limited to the exponential phase of growth (38,84).
It is also interesting that reporter fusions for csrA, csrB, and csrC did not respond in a substantial way to acid stress induction, suggesting that expression of the Csr system itself is not strongly responsive to drops in pH.Despite this finding, we cannot definitively conclude that the Csr system is unresponsive to acid stress as the influence of CsrB/C RNA turnover was not examined in this study.
As CsrA is important for growth in E. coli and other species, it has become an appealing target for drug design (85)(86)(87).However, inhibiting CsrA causes increased biofilm formation and enhances the expression of virulence factors in some species, both of which are counterproductive for therapeutics (6,88).The present study adds to the growing list of reasons why targeting CsrA for drug design may be counterproduc tive for antibiotic-based therapy of E. coli infections.Specifically, impairing CsrA-depend ent regulation would greatly increase the expression of acid tolerance regulators of E. coli.This increase is relevant as acid tolerance has already been proposed to be involved in the low infectious dose of certain pathogenic E. coli (33).However, disrupting CsrA-dependent regulation of gadA/B may prove to be an appealing strategy for GABA production.GABA is a neurotransmitter that plays a key role in human mental health and microbiome research suggests that GABA-producing bacteria are inversely correlated with depression (44).Previous work explored engineering E. coli for batch production of GABA or even as a probiotic to deliver GABA directly; however, no previous studies have explored disrupting CsrA repression of gadA/B (89,90).Our findings suggest that loss of repression of GDAR genes may be useful for increasing GABA levels in the host.

Bacterial strains, plasmids, bacteriophage, culture conditions and oligonu cleotides
All E. coli strains, plasmids, and bacteriophages used in this study are listed in Tables S2 to S4, respectively.Bacterial strains were grown and maintained in LB medium (0.5% yeast extract, 1% tryptone, and 1% NaCl, pH 7.4).Overnight cultures were inoculated in LB medium from frozen glycerol stocks of bacterial strains.These cultures were grown at 37°C or 30°C with shaking (250 rpm).Gene deletions were introduced by P1 vir transduc tion from E. coli donor strains from the Keio library (91).To remove antibiotic resistance cassettes, pCP20 encoding the Flp recombinase was used (92).Plasmids to complement knockouts were sourced from the ASKA collection (93).
M9 supplemented medium (1 × M9 salts supplemented with 2 mM MgSO 4 , 0.1 mM CaCl 2 , 0.2% casamino acids, and 0.4% glucose) was used for assessing growth and gene expression.For media used in survival assays, M9 was only supplemented with 0.4% glucose, and the pH was adjusted to pH 2.5 using HCl.When studying GDAR activity, growth media were supplemented with 1.5 mM glutamate.The pH of the medium was adjusted to 2.5 using HCl.For simulated stomach acid (pH 1.6), Fasted State Simulated Gastric Fluid (FaSSGF) Biorelevant medium was used.
Deletion of csrA results in severe growth defects and genetic instability (94).To avoid this problem, experiments were performed with strains carrying a csrA truncation (csrA::kan) that retains ~10% of its RNA binding activity (95).The csrA::kan allele was moved into various strains via P1 transduction (1).

Construction of lacZ reporter fusions
Translational fusions to 'lacZ were constructed in plasmid vector pLFT (29).Posttranscrip tional fusions to 'lacZ were constructed in plasmid vector placUV5 (29).All fusions were integrated into the chromosome using the CRIM system (96)(97)(98).Translational fusions were constructed as follows.About 500 nt of DNA upstream of the promoter region, the promoter region, and one or more codons downstream of the translation start site were amplified by PCR using the relevant primers (Table S3).Since gadE transcrip tional regulation involves an extremely large upstream region, the translational fusion started ~750 nt upstream of the first promoter (67).The PCR products and pLFT were digested with PstI and BamHI and then ligated together.The products were transformed into DH5α λpir cells.The plasmids were isolated and fusion sequences were sequenced and verified before being integrated into the λatt site of strain MG1655 ΔlacZ using the helper plasmid pFINT as previously described (29).Single integrates were confirmed via PCR (96).Refer to Table S3 for the primer sequences.
The evgA leader fusion was constructed by PCR amplifying the 5′ leader of evgA (−114 to +7 relative to the first nucleotide of the start codon as +1) with primers that introduced restriction sites flanking the leader sequence.The resulting PCR product and pLacUV5 were digested with EcoRI and BamHI and then ligated together.To avoid the accumulation of spontaneous mutations that occurred when pLacUV5 plasmids were maintained in DH5α λpir cells, the ligated plasmids were integrated directly into the λatt site of MG1655 ΔlacZ using the Int λ expressing helper plasmid pFINT (29).The resulting single integrates were confirmed by PCR and sequencing.

β-galactosidase assay
Bacterial cultures containing lacZ fusions were grown in LB at 37°C to exponential phase (OD 600 = 0.5), diluted to an optical density at 600 nm (OD 600 ) of 0.02 in fresh M9 media supplemented with 0.4% glucose and 0.2% casamino acids.Cells were harves ted at appropriate time points throughout growth during experiments.Acid induction experiments were based on a previously published protocol where a predetermined volume (1.6 mL) of 0.75 M HCl or water was added to actively growing cultures, HCl was added to decrease the pH from 7 to 5.5 and water was used as a negative control (50).β-Galactosidase activity was determined as described previously (29).Total cellular protein was measured following precipitation with 10% trichloroacetic acid, using the bicinchoninic acid (BCA) assay (Pierce Biotechnology) with bovine serum albumin as the protein standard.

Growth curves and kinetics
Bacterial cultures were grown in LB at 37°C overnight and diluted to an OD 600 of 0.01 in fresh M9 media.OD 600 was measured every h for 8 h and at 24 h.The growth rate constant (μ) was calculated from the exponential phase of growth: μ = 2.303(logOD 2 − logOD 1 )/(t 2 -t 1 ).

Survival assays
Bacterial cultures were grown in LB at 37°C overnight and diluted to OD 600 of 0.01 in fresh M9 media.Cells were grown to an OD 600 of 0.5, pelleted by centrifugation, and then washed with a M9 medium containing 0.4% glucose.The wash medium was the same pH as the growth media and contained no added amino acids.Cells were then diluted 1:50 into pH 2.5 M9 medium containing 0.4% glucose with or without 1.5 mM glutamate and incubated for 2 h without shaking at 37°C.Aliquots were serially diluted in triplicate and were plated onto LB agar plates.Colonies were counted after being grown overnight at 37°C.Percent survival was calculated as follows: ((CFU per mL at 2 h) / (CFU per mL at 0 h)) * 100.

Gel shift assays
Binding of CsrA to the 5′ leader of gadA (60 nt; −27 to +33 relative to the first nucleotide of the start codon as +1), gadB (73 nt; −27 to +43 relative to the start codon), gadE (28 nt; −21 to +7 relative to the start codon), evgA (132 nt; −124 to +8 relative to the start codon), ydeO (84 nt; −74 to +10 relative to the start codon), and ydeP (188 nt; −178 to +10 relative to the start codon) was monitored using a gel shift assay.The transcripts were synthesized in vitro using the MEGAshortscript Kit (Invitrogen).Templates for in vitro transcription reactions were generated by PCR and subjected to PCR cleanup using Monarch PCR & DNA Clean up Kit (New England BioLabs, NEB) before using for in vitro transcription reactions.In vitro synthesized transcripts were gel purified on denaturing urea polyacrylamide gels, eluted overnight, extracted with phenol-chloroform, ethanol precipitated, resuspended in TE buffer, and quantified with a spectrophotometer.Twenty pmol of the RNA was dephosphorylated with Antarctic phosphatase (NEB) and 5′-end-labeled with γ-32 P ATP and T4 polynucleotide kinase (NEB), gel purified, eluted overnight, phenol-chloroform extracted, ethanol precipitated, and resuspended in TE.The concentration of labeled RNA was determined using a standard curve constructed with γ-32 P ATP.Binding reactions contained 0.08-0.2nM end-labeled RNA, 10 mM MgCl 2 , 100 mM KCl, 32.5 ng SUPER-ase In (Ambion) with various concentrations of CsrA-H6 as indicated in the figures and incubated at 37°C for 30 min (99).Reaction mixtures were separated on a native polyacrylamide gel and imaged using a phosphorimager.The radioactive signals of free and shifted/bound RNA-protein complexes were quantified with Quantity One software and used for determining the apparent equilibrium dissociation constants (K D ).

Coupled transcription-translation assays
Coupled transcription-translation assays were conducted in vitro using the PURExpress kit (NEB) following a published protocol (12).Plasmid pGadA-T7 contains a T7 promoter to drive transcription of the gadA translational fusion (nt +1 to +54 relative to the transcriptional start site).Plasmid pGadB-T7 contains a T7 promoter to drive transcription of the gadB translational fusion (nt +1 to +78 relative to the transcriptional start site).Plasmid pEvgA-T7 contains a T7 promoter to drive transcription of the evgA translational fusion (nt +1 to +122 relative to the transcriptional start site).The transcription start site of the σ S promoter was chosen over the sigma σ 70 promoter due to previous results suggesting it may be more active (66).A similar P T7 -pnp'-'lacZ translational fusion plasmid was used as a negative control (65).These plasmids were used as templates for coupled transcription-translation reactions according to the manufacturer's instructions.Each 6.7 µL reaction contained 250 ng of plasmid DNA and various concentrations of purified CsrA-His6 with 1 U of RNase inhibitor (Promega) and 2.5 mM dithiothreitol (DTT), 2.7 µL of solution A and 2 µL of solution B. The mixtures were incubated for 2 h at 37°C and β-galactosidase activity was determined.OD 420 data without CsrA were normalized to 100%.

Toeprint assays
CsrA-RNA toeprint assays followed a previously published procedure (19).gadA RNA (from nt −27 to +26 relative to the gadA translational start), gadB RNA (nt −27 to +50 relative to the gadB translational start), and evgA RNA (nt −114 to +8 relative to the evgA translational start) were synthesized with the RNAMaxx kit (Agilent technologies) using PCR-generated DNA templates.Each gel-purified RNA (150 nM) in TE buffer was hybridized to a 5′ end-labeled DNA oligonucleotide complementary to the 3′ end of the vector-derived 3′ extension by heating for 3 min at 80°C followed by slow cooling for 10 min at room temperature.Toeprint reaction mixtures (10 µL) contained 2 µL of the hybridization mixture (30 nM final concentration), 1 µM CsrA-H6, 375 µM each dNTP, 10 mM DTT and Superscript III (SSIII), or AMV reverse transcriptase buffer.Mixtures were incubated for 30 min at 37°C to allow CsrA-RNA complex formation.After the addition of 0.125-0.25 U SSIII (Invitrogen) or 0.5-2 U AMV reverse transcriptase (Sigma Aldrich) incubation was continued for 15 min at 37°C.Reactions were terminated by the addition of 10 µL of gel loading buffer (95% formamide, 0.025% SDS, 20 mM EDTA, 0.025% bromophenol blue, and 0.025% xylene cyanol).Samples were heated to 90°C for 5 min and fractionated through standard 6% polyacrylamide-8 M urea sequencing gels.Toeprint patterns were visualized with a phosphorimager.

Footprint assays
CsrA footprints were performed according to the published procedure (95).The gadA, gadB, and evgA RNAs described for the toeprint assay were used for footprinting.Gel-purified RNA was dephosphorylated and then 5′ end-labeled using T4 polynucleo tide kinase (NEB) and [γ−32P]ATP (7,000 Ci/mmol).Labeled RNAs were renatured by heating for 1 min at 90°C followed by slow cooling for 10 min at room temperature.Binding reaction mixtures (10 µL) contained 2 nM labeled RNA, 10 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 100 mM KCl, 40 ng of yeast RNA, 7.5% glycerol, 0.1 mg/mL xylene cyanol, and various concentrations of purified CsrA-H6.After a 30-min incubation at 37°C to allow for CsrA-RNA complex formation, RNase T1 (0.016 U) was added, and the incubation was continued for 15 min at 37°C.The reactions were stopped by adding 10 µL of gel loading buffer.Samples were heated for 5 min at 90°C and fractionated through standard 6% polyacrylamide-8 M urea sequencing gels.Cleaved patterns were examined using a phosphorimager.

C. elegans studies
Nematodes were maintained as previously described (100,101).The C. elegans strains used in this study, CB156 (unc-25) and N2 (Bristol), were obtained from the Caenorhab ditis Genetics Center.Bacterial strains used in all C. elegans experiments were grown overnight in Lennox Broth (LB) at 37°C with shaking, seeded onto nematode growth media (NGM), and dried in a biosafety cabinet for 2-4 h.Worms were age-synchronized using the standard sodium hypochlorite method.Age-synchronized worms were plated onto NGM seeded with appropriate bacteria and were kept at 23°C for 2 days.Positive control plates were bathed in 30 mM GABA solution (102) for 3 h prior to starting the experiments.Worms were picked and placed on NGM plates supplemented with PTZ (7 mg/mL) for 15 min (68).Following the incubation, the number of worms displaying a convulsing (head bobbing) phenotype was counted.As a control, PTZ-treated N2 worms did not display any convulsions, consistent with previous reports (68).

Datamining methods
Hypothetical targets of CsrA-dependent regulation were selected from previously published studies.Evidence for CsrA affecting transcript levels was determined by examining RNA-seq data (3,24,28).Genes were considered hypothetical targets if the log 2 fold change was >1 or < −1 and was statistically significant.Potential RNA targets derived from pulldown results (29), in silico prediction (30), CLIP-seq (3), and HITS-CLIPseq (28) were obtained from previously published studies.The data were compiled and are represented in Table S1; Fig.

FIG 2
FIG 2 The csrA mutant exhibits pH-dependent growth and survival phenotypes.pBR322 is a control plasmid and p2VR112 overexpresses csrA.Stippling indicates that the strain carries a plasmid.(A) Relative growth rates of E. coli MG1655 (WT) and its isogenic csrA::kan mutant are shown.Error bars represent the standard deviation (sd) from four independent experiments.(B) Cells grown exponentially under mildly acidic conditions and then challenged by simulated (Continued on next page)

Full 5 FIG 3
FIG 3 Suppressors of the pH-dependent growth defect of the csrA mutant.(A through H) Growth rates (μ) of exponential phase E. coli strains in mildly acidic (pH 5.5) M9 media normalized to WT. Stippling indicates complemented strains.Gene products of possible suppressors are indicated in the heading of each panel.Error bars represent the standard deviation (sd) from six independent experiments.Statistical significance was determined using an unpaired t-test and is denoted as follows: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.The absence of a comparison bar indicates a comparison to WT.

FIG 4
FIG 4 Suppression of the csrA mutant survival phenotype in preadapted cells.Exponentially growing cells (OD = 0.5) were preadapted at pH 5.5 and then challenged for 2 h at pH 2.5.(A) Knockouts of genes relevant to the EvgA-YdeO regulon were challenged under extremely acidic conditions without glutamate or (B) extremely acidic conditions with 1.5 mM glutamate.(C) Knockouts of genes relevant to the GDAR system were challenged under extremely acidic conditions without glutamate or (D) extremely acidic conditions with 1.5 mM glutamate.Error bars represent the standard deviation (sd) from six independent experiments.Statistical significance was determined using unpaired t-tests and is denoted as follows: *P < 0.05; **P < 0.01; ***P < 0.001.The absence of a comparison bar indicates a comparison to WT.

FIG 5
FIG 5 CsrA binding interactions with evgA, gadA, gadB, gadE, ydeO, and ydeP mRNA.(A) Sequences of probes used for gel shift assays.GGA motifs that may be components of CsrA binding sites are shown in red, start codons are underlined, and the Shine-Dalgarno (SD) sequence is shaded.(B) 5′-end-labeled transcripts were incubated with CsrA at the indicated concentrations.Competition reactions were performed in the presence of unlabeled specific (self ) or unlabeled nonspecific (phoB) competitor RNAs at the concentrations shown.The positions of free (F) and bound (B) RNA are marked.

9 FIG 6
FIG 6 Effects of CsrA on evgA expression in vivo.Translation fusion expression of evgA (A) at pH 7 and (B) at pH 5.5.Leader fusion expression of evgA (C) at pH 7 and (D) at pH 5.5.Translational fusions of evgL (E) at pH 7 and (F) at pH 5.5.(G) Translational fusion expression of evgA with the wild type evgL start codon mutated to a stop codon (A89T:T90A).Growth curves (OD 600 ) are shown in the panel insets.Error bars represent the standard deviation (sd) from three independent experiments.

FIG 7
FIG 7 CsrA binding represses translation of evgA in vitro.(A) evgA leader sequence.GGA motifs are shown in red, the start codon of evgL in teal, the start codon of evgA in blue, and toeprint positions from AMV are in yellow and SSIII is in brown.The evgL and evgA SD sequences and the evgL stop codon are underlined.(B) Effects of CsrA on in vitro translation of evgA'-'lacZ and pnp'-'lacZ (control) translational fusions driven by a T7 RNAP promoter.Relative β-galactosidase activity depicts the mean and standard deviation of activity relative to reaction mixtures without CsrA.(C) CsrA-dependent toeprints on evgA RNA using AMV reverse transcriptase.Nucleotides in red indicate a GGA motif, blue indicates the evgA start codon, and teal the evgL start codon.Arrows (+) indicate bands upon the addition of CsrA and (-) indicate the loss of a band upon the addition of CsrA.(D) Toeprinting of the evgA transcript using SSIII reverse transcriptase.Color coding and arrows are the same as in (C).(E) CsrA-evgA RNA footprint.5′-end-labeled evgA RNA was treated with RNase T1 ±CsrA, as shown.RNA exposed to partial alkaline hydrolysis (OH), RNase T1 digestion of denatured RNA (T1), and untreated control RNA (C) are also shown.Red vertical lines correspond to CsrA binding sites that contain the GGA motifs.

FIG 8 (
FIG 8 (Continued) at pH 7, (H) gadB at pH 5.5, (I) ydeP at pH 7, and (J) ydeP at pH 5.5.Expressions in the WT and csrA mutant strains are shown in black and red, respectively.Growth is shown in panel insets.Error bars represent the standard deviation (sd) from three independent experiments.

FIG 9
FIG 9 CsrA binding represses translation of gadA and gadB in vitro.(A) gadA and gadB leader sequences.The GGA motifs are shown in red and the start codons are in blue.Toeprint positions are indicated in yellow.(B) Effects of CsrA on in vitro translation of gadA'-'lacZ, gadB'-'lacZ and pnp'-'lacZ (control) translational fusions driven by a T7 RNAP promoter.Relative β-galactosidase activity depicts the mean and standard deviation relative to reaction mixtures lacking CsrA.(C) CsrA-gadA RNA toeprint using AMV reverse transcriptase.Toeprint positions are indicated on the right.(D) CsrA-gadB RNA toeprint using AMV reverse transcriptase.Toeprint positions are indicated on the right.(E) CsrA-gadA RNA footprint and (F) CsrA-gadB RNA footprint.5′-end-labeled RNA was treated with RNase T1 ±CsrA as shown.Partial alkaline hydrolysis (OH) and RNase T1 digestion of denatured RNA (T1), and untreated control RNA are also shown.

FIG 10
FIG 10 The effect of the csrA mutation on GABA-deficient C. elegans.Data are represented as the average percent of worms convulsing normalized to that in worms fed WT E. coli (MG1655) after 15 min of exposure to PTZ.Each data point is the average of a minimum of three independent experiments for a minimum of 60 worms.Error bars represent standard deviation (sd).Statistical significance was determined using unpaired t-tests and is denoted as follows: *P < 0.05; **P < 0.01; ***P < 0.001 ****P < 0.0001.The absence of a comparison bar indicates a comparison to WT.