The small noncoding DsrA RNA is an acid resistance regulator in Escherichia coli.

Abstract

DsrA RNA is a small (87-nucleotide) regulatory RNA of Escherichia coli that acts by RNA-RNA interactions to control translation and turnover of specific mRNAs. Two targets of DsrA regulation are RpoS, the stationary-phase and stress response sigma factor (sigmas), and H-NS, a histone-like nucleoid protein and global transcription repressor. Genes regulated globally by RpoS and H-NS include stress response proteins and virulence factors for pathogenic E. coli. Here, by using transcription profiling via DNA arrays, we have identified genes induced by DsrA. Steady-state levels of mRNAs from many genes increased with DsrA overproduction, including multiple acid resistance genes of E. coli. Quantitative primer extension analysis verified the induction of individual acid resistance genes in the hdeAB, gadAX, and gadBC operons. E. coli K-12 strains, as well as pathogenic E. coli O157:H7, exhibited compromised acid resistance in dsrA mutants. Conversely, overproduction of DsrA from a plasmid rendered the acid-sensitive dsrA mutant extremely acid resistant. Thus, DsrA RNA plays a regulatory role in acid resistance. Whether DsrA targets acid resistance genes directly by base pairing or indirectly via perturbation of RpoS and/or H-NS is not known, but in either event, our results suggest that DsrA RNA may enhance the virulence of pathogenic E. coli.

Full text

Page 1

JOURNAL OF BACTERIOLOGY, Sept. 2004, p. 6179–6185
0021-9193/04/$08.00?0DOI: 10.1128/JB.186.18.6179–6185.2004
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Vol. 186, No. 18
The Small Noncoding DsrA RNA Is an Acid Resistance
Regulator in Escherichia coli†
Richard A. Lease,1,2* Dorie Smith,1Kathleen McDonough,1and Marlene Belfort1
Wadsworth Center, New York State Department of Health, Center for Medical Sciences, Albany, New York,1
and Department of Biophysics, The Johns Hopkins University, Baltimore, Maryland2
Received 21 April 2004/Accepted 18 June 2004
DsrA RNA is a small (87-nucleotide) regulatory RNA of Escherichia coli that acts by RNA-RNA interactions
to control translation and turnover of specific mRNAs. Two targets of DsrA regulation are RpoS, the station-
ary-phase and stress response sigma factor (?s), and H-NS, a histone-like nucleoid protein and global
transcription repressor. Genes regulated globally by RpoS and H-NS include stress response proteins and
virulence factors for pathogenic E. coli. Here, by using transcription profiling via DNA arrays, we have
identified genes induced by DsrA. Steady-state levels of mRNAs from many genes increased with DsrA
overproduction, including multiple acid resistance genes of E. coli. Quantitative primer extension analysis
verified the induction of individual acid resistance genes in the hdeAB, gadAX, and gadBC operons. E. coli K-12
strains, as well as pathogenic E. coli O157:H7, exhibited compromised acid resistance in dsrA mutants.
Conversely, overproduction of DsrA from a plasmid rendered the acid-sensitive dsrA mutant extremely acid
resistant. Thus, DsrA RNA plays a regulatory role in acid resistance. Whether DsrA targets acid resistance
genes directly by base pairing or indirectly via perturbation of RpoS and/or H-NS is not known, but in either
event, our results suggest that DsrA RNA may enhance the virulence of pathogenic E. coli.
Regulation by RNA, termed riboregulation, plays a substan-
tial role in modulating gene expression in bacteria (reviewed in
references 14, 17, 26, 33, 42, 43, and 44). In addition to their
classically studied role in plasmid maintenance (reviewed in
reference 42), Escherichia coli small RNAs act to change the
conformation of target mRNAs (DsrA and RprA), block
mRNA translation by occlusion of Shine-Dalgarno sequences
(MicF, OxyS, Spf, and RyhB), degrade target mRNAs (DsrA,
Oop, and RyhB), and titrate specific protein factors (OxyS and
CsrB RNAs, 6S RNA), sometimes in combination.
Several E. coli small RNAs coordinate stress responses or
virulence factors (reviewed in reference 14). A principal ad-
vantage of small RNAs as regulators is that they are not trans-
lated and therefore cost less energy to produce than do pro-
teins. Also, many bacterial small RNAs are relatively stable
and can persist to target transcripts with high specificity by
antisense interactions (reviewed in references 17 and 43).
Some small RNAs are degraded together with their target
mRNAs (22).
One such RNA, DsrA RNA, is a small (87 nucleotides),
multifunctional genetic regulator of E. coli. DsrA RNA mod-
ulates the levels of two global transcription regulators, RpoS
(?s, the product of the rpoS gene) and H-NS (a nucleoid
protein and transcription silencer in bacteria, produced from
the hns gene). DsrA acts by sequence-specific RNA-RNA in-
teractions to enhance translation of rpoS RNA and to stabilize
rpoS message (18–20, 35). In addition to its role at rpoS, DsrA
also binds hns mRNA by specific base-pairing interactions and
blocks H-NS translation as it sharply increases hns mRNA
turnover (18). The first stem-loop region of DsrA melts out to
contact rpoS mRNA, whereas the second stem-loop region of
DsrA base pairs with hns mRNA. This conformational change
within DsrA acts to switch the translation state of two different
mRNA targets (reviewed in references 1 and 17). Like that of
many small, noncoding regulatory RNAs, direct DsrA activity
on mRNAs requires Hfq, an Sm domain RNA-binding protein
and putative RNA chaperone (reviewed in reference 43).
DsrA perturbation of H-NS and RpoS results in increased
transcription of downstream genes repressed by H-NS or ac-
tivated by RpoS (19, 35). H-NS and RpoS also act in concert to
permit the transcription of a number of stress response and
virulence factor proteins (reviewed in references 3 and 15).
Many genes require additional regulatory proteins, in conjunc-
tion with H-NS and RpoS, to tailor specific responses to par-
ticular environmental stresses (reviewed in reference 15). The
downstream effect of DsrA is therefore predicted to be the
induction of a pleiotropic stress response.
Despite the study of key components of the DsrA regulatory
network, the phenotype of DsrA activity in the cell has re-
mained elusive. Here we used a genomics approach to define
the downstream effects of DsrA in E. coli. DNA array-based
transcriptome analysis suggests that DsrA stimulates acid re-
sistance, which is known to enhance the virulence of patho-
genic E. coli strains (9 and references therein). Both the hdeAB
and glutamate-dependent (gad) acid resistance systems were
induced, although the arginine-dependent (adi) genes were
apparently not induced by DsrA. Furthermore, both nonviru-
lent (K-12) and pathogenic (O157:H7) strains of E. coli had
compromised acid resistance in dsrA-null mutants. DsrA there-
fore plays a role in cellular acid resistance, an important fea-
ture for the survival of enteric bacteria in low-pH environ-
ments and for the virulence of pathogenic E. coli.
* Corresponding author. Mailing address: Department of Biophys-
ics, Jenkins Hall, Homewood Campus, The Johns Hopkins University,
3400 N. Charles St., Baltimore, MD 21218. Phone: (410) 516-6536.
Fax: (410) 516-4118. E-mail: ral@jhu.edu.
† Supplemental material for this article may be found at http://jb
.asm.org.
6179

Page 2

MATERIALS AND METHODS
Media, strains, and plasmids. Cells were grown in Luria-Bertani (LB) me-
dium (28). Where appropriate, antibiotics were used at the following concentra-
tions: ampicillin, 100 ?g/ml; chloramphenicol, 25 ?g/ml; tetracycline, 10 ?g/ml;
kanamycin, 50 ?g/ml; streptomycin, 50 ?g/ml. E. coli strain M182 is described
elsewhere (8). E. coli O157:H7 strain 1957 is a clinical isolate from stool or fecal
material from 2001, provided by the Wadsworth Center Bacteriology Laboratory.
Strain M182 dsrA::cat was made by P1 transduction of the cat gene from C600
dsrA::cat, which was provided by Susan Gottesman (National Institutes of
Health, Bethesda, Md.). E. coli strain SM10 tra?Kanr(?pir) (32) was provided
by Kelynne Reed (Austin College, Sherman, Tex.).
Strain O157 dsrA::cat was constructed as follows. The dsrA::cat allele, along
with ca. 900 bp of the flanking chromosomal DNA sequence, was amplified by
PCR with primers W1748 (GCT CTA GAA AGA GAC AAC GAT AAC CTC
G) and W1749 (GCT CTA GAG CGT AAT CCA TTA CCT CCA G) and was
cloned by blunt-end ligation into ?pir-dependent gene replacement vector
pCVD442 (ori-R6K mob?AmprsacB) (11) at a filled-in XbaI restriction site.
The recombinant plasmid DNA was used to transform DH5? (?pir) and plated
on LB medium containing chloramphenicol and ampicillin. The resulting plas-
mid, pCVD442 dsrA::cat, was screened by restriction analysis and then used to
transform E. coli SM10. Separately, a streptomycin-resistant variant of O157:H7
was selected by plating on LB medium containing streptomycin. This O157 (Strr)
strain was mated with SM10 Kanr/pCVD442dsrA::cat in liquid culture for 2 h at
37°C. E. coli O157 Strr/pCVD-dsrA::cat was selected on plates (LB medium
containing streptomycin and chloramphenicol). Selected clones were checked for
kanamycin sensitivity and ampicillin resistance by patching of colonies onto
plates. To prepare dsrA::cat chromosomal integrants, gene replacement was
performed by growing cultures of O157 Strr/pCVDdsrA::cat and plating on LB
medium containing 5% sucrose and chloramphenicol. Most sucrose-resistant
colonies were Amps, indicating loss of the pir-dependent gene replacement
vector. The exchange of dsrA for the dsrA::cat chromosomal allele was verified by
chloramphenicol resistance and by PCR with primers W1748 and W1749.
Plasmids pBRdsrA and pBRdsrA*Hwere constructed by cloning the BamHI
fragment of pDDS164 (34) or pDsrA*H(19) into the BamHI site within the Tetr
gene of pBR322. Potential clones were selected for ampicillin resistance and
then screened for insertional inactivation of the Tetrgene (Tets) on LB medium
containing tetracycline. Clones were confirmed by restriction analysis and DNA
sequencing.
Transcription profiling. Freshly streaked cells of strain M182 containing either
pBR322 or pBRdsrA were grown overnight in LB broth plus ampicillin at 37°C.
Cells were then diluted 1:100 into 30 ml of LB broth plus ampicillin for growth
at 30°C to induce DsrA (35). Cells were grown with vigorous shaking to an
optical density at 600 nm of 0.3 to 0.4. Total cellular RNA was prepared and used
to make labeled cDNA by reverse transcription in the presence of [?-33P]dATP
(NEN). An oligonucleotide mixture complementary to every E. coli mRNA 3?
end (Sigma-Genosys) was used to prime cDNA synthesis. The33P-labeled total
cellular cDNA was used to probe filter-based DNA arrays (Sigma-Genosys) as
previously described (39). Filters were then exposed to phosphorimager screens
(Molecular Dynamics, Inc.) and scanned. Images were quantified with Arrayvi-
sion software (Imaging Research, St. Catharines, Ontario, Canada). The exper-
iment was performed twice. Hybridization signals were normalized to total
genomic DNA standards present on each filter. Differential expression for each
set of experiments was determined for individual genes by dividing the signal
from the DsrA-overproducing strain by that of the plasmid control. For images
of arrays and spot data files, see the supplemental material.
Primer extension analysis. RNA was extracted and primer extension was
performed as previously described (5, 18). Primer sequences are as follows: adiA
? 74, CTT GAT GGA GAA ACT CGC TTT CAA C; adiY ? 83, AGT TCT
CGC TAA AGC AAA GCG ATA C; gadA ? 48, CGT TAA CAG CTT CTG
GTC CAT TTC G; gadB ? 64, GTT CCG ACC TTA AAT CCG TTA CTT G;
gadX ? 89, GGT GAG AAT ATA TTT ATG TCT TGC; nfnB ? 37, TAT CCA
TAA AGA CTC CAT GTG AAA G; W538dsrA, GAA ACT TGC TTA AGC
AAG AAG C. The hns- and stpA-specific primers (48) and the hdeA-specific
primer (2) are as previously described. All DNA oligonucleotides were purified
via elution from polyacrylamide gels. The DNA oligonucleotides were end la-
beled with [?-32P]ATP (Perkin-Elmer/NEN) and T4 polynucleotide kinase (New
England Biolabs) as previously described (28), extracted with 1 volume each of
phenol and then with chloroform-isoamyl alcohol (24:1), and purified by TE-10
spin column chromatography (Clontech).
Acid resistance assays. Cultures of strains M182 and O157 and their respective
dsrA mutants and merodiploid strains were grown overnight at 37°C and tested
by dilution into LB medium at pH 2.0 as previously described (13), except that
samples were taken each hour for plating for up to 6 h of acid treatment at 37°C.
Cells were diluted in 10 mM Tris-HCl (pH 7.5)–1 mM magnesium chloride prior
to plating. Percent survival is given as the titer of the CFU of acid-tested cells
compared to that of a zero-time, untreated control sample.
RESULTS
Global mRNA analysis. To determine the downstream ef-
fects of DsrA, we induced DsrA expression from a plasmid.
Changes in the expression of total cellular RNA from cultures
with and without DsrA overproduction at 30°C were compared
by using genomic DNA filter arrays. Whereas transcript levels
from many genes increased or decreased severalfold (Table 1),
a number of genes related to acid resistance were strongly
induced by DsrA overproduction (Table 1, top). The hdeAB
operon, which encodes the H-NS-repressed acid resistance
proteins HdeA and HdeB (12, 46, 47), produced 10- to 26-fold
higher hdeA transcript levels and 12- to 43-fold higher hdeB
transcript levels when DsrA was overproduced (Table 1, lines
1 and 2). Other H-NS- and RpoS-regulated acid resistance
operon elements were also induced. For example, the gadAX
operon, which encodes glutamate decarboxylase, GadA, and
its positive regulator, GadX (40), was induced as much as
threefold (Table 1, lines 4 and 5). Similarly, a gene that en-
codes a positive regulator of arginine-dependent acid resis-
tance, adiY, was induced ca. fivefold (Table 1, line 7) (38).
Primer extension analysis. To verify and quantitate induc-
tion of the specific genes mentioned above, as well as to check
for the overexpression of several genes expected to be induced
that were not seen by array analysis, we measured the levels of
key transcripts by gene-specific primer extension analysis. Cells
containing plasmid vector were compared to cells overproduc-
ing either DsrA or a mutant DsrA variant from a plasmid.
Identical growth conditions were used for the array and primer
extension analyses. The mRNA levels of the acid resistance
genes hdeA, gadA, gadX, gadB, adiY, and adiA were analyzed
(Fig. 1; quantitation in Table 1, column five; data for adiY and
adiA not shown). As a positive control, we determined the
transcript levels of stpA. StpA is an H-NS paralog that is
induced in hns mutants (48) and that registered a 1.2- to
2.3-fold increase in the array analysis (Table 1, line 18). As a
negative control, levels of the unrelated nfnB nitroreductase
mRNA were also checked (4).
The data confirm the induction of several types of acid
resistance genes by DsrA, and the lengths of all cDNAs are
consistent with transcription starting at the major promoter of
each gene. The levels of hdeA transcript were increased 26-
fold, corresponding exactly to those determined by the array
analysis (Fig. 1, panel 5, and Table 1, line 1). The levels of gadA
mRNA were also increased, by a factor of 3.5, again providing
correspondence between the two methods (Fig. 1, panel 2, and
Table 1, line 4), and stpA mRNA was increased 3.1-fold (Fig.
1, panel 4). A modest 2.6-fold increase observed for gadX
mRNA was again corroborative (Fig. 1, panel 1, and Table 1,
line 5). Interestingly, we also found a 3.5-fold increase in gadB
mRNA, whereas such induction had been undetectable by
array analysis (Fig. 1, panel 3, and Table 1, line 6). We at-
tribute the differences between the two methods to the fact
that oligonucleotide cDNA primers for the array analysis were
optimized for cloning of the PCR products displayed on the
6180 LEASE ET AL.J. BACTERIOL.

Page 3

arrays (25) and not necessarily for annealing to the specific
mRNAs. Conversely, although the adiY transcript was 3.7- to
5.7-fold elevated in the array analysis (Table 1, line 7), we were
unable to detect either adiY or adiA transcripts by primer
extension analysis (data not shown). Regardless, since gadB
and gadC form an operon and since gadC is required for
glutamate decarboxylase-based acid resistance (9), we can con-
clude that both the hdeAB and glutamate-dependent acid re-
sistance (gad) genes are induced by DsrA.
To confirm that changes in transcript levels resulted from
DsrA function, we measured acid resistance gene transcripts
after induction of a DsrA variant, DsrA*H, which is compro-
TABLE 1. DsrA functional genomics in E. coli K-12
Category and
genes
Location
(min)a
Function
Fold change
Filter arrayb
Primer extensionc
Acid resistance
1 hdeA
2 hdeB
3 hdeD
4 gadA
5 gadX (alias yhiX)
6 gadB
7 adiY
8 adiA
78.76
78.75
78.78
78.98
78.95
33.81
93.44
93.46
Acid resistance
Acid resistance
Acid resistance
Acid resistance
Acid resistance-virulence regulator
Acid resistance
Acid resistance regulator
Acid resistance
10–26
12–43
2.2–3.4
2.3–3.0
2.3–3.2
1
3.7–5.7
1
26
—
—
3.5
2.6
3.5
NS
NS
Membrane or transport
9 proX
10 ompX
11 ompA
12 ompF
13 ompT
14 rbsB
60.47
18.32
21.95
21.23
12.59
84.80
Osmotic shock
Outer membrane protein
Outer membrane protein
Outer membrane protein
Outer membrane protein
Periplasmic ribose-binding proteind
3.3–3.9
2.3–4.5
2–2.6
2–2.5
1.9–3.1
?2.6–3.4
—
—
—
—
—
—
Regulatory
15 malT
16 rpoE
17 hns
18 stpA
19 rcsA
20 dps
76.54
58.36
27.84
60.27
43.58
18.27
Activator-maltose regulon
Activator-sigma factor
Repressor-nucleoid
Repressor-nucleoid-RNA chaperone
Activator-capsule synthesis
DNA protection
?2.3–2.5
?2–2.5
1
1.2–2.3
1.3–1.4
1.9–3.4
—
—
1.3e
3.1
?10f
—
Motility
21 fimA
22 fliA
23 fliC
97.88
43.09
43.11
Fimbriae
Flagella
Flagella
2.2–3.5
?2.1
?2.6–7.3
—
—
—
Other
24 ftn
25 xylA
26 sucA
27 groS (alias mopB)
28 garR (alias yhaE)
29 sgbU (alias yiaR)
30 hchA (alias yedU)
31 rihC (alias yaaF)
32 nfnBg
42.83
80.34
16.34
94.16
70.48
80.77
43.84
0.59
13.02
Ferritin (iron storage)
D-Xylose isomerase
2-Ketoglutarate dehydrogenase
Chaperonin
Tartronate semialdehyde reductase
Hexulose 6-phosphate isomerase (putative)
Heat-inducible chaperone
Ribonucleoside hydrolase
Nitroreductase (control)g
2.8–3.6
2.5–2.9
?2–2.3
?2.6–2.9
?2.3–3.2
?2.4–3.4
4.7–5.9
?2–2.2
1
—
—
—
—
—
—
—
—
1.2g
Hypothetical protein
33 yagU
34 yncC
35 yeaK
36 yeaP
37 yeaQ
38 yedW
39 ybfE
6.51
32.73
40.34
40.43
40.46
43.89
15.32
Alias b0287
Alias b1450
Alias b1787
Alias b1794
Alias b1795
Alias b1969
Alias b2253
2.6–2.8
2.1–4.4
2–2.9
2–2.2
2.9–3.3
2.3–4
2.1–3.2
—
—
—
—
—
—
—
aGene location data were obtained from the EcoGene database (6, 27).
bSigma-Genosys nylon membrane gene array. Fold changes for each experiment (n ? 2) are given as ranges. All values that decrease are so indicated by minus signs.
cAverage fold effect (n ? 3 or 4). NS, no extension product seen. —, primer extension not performed.
dDsrA also contains an antisense sequence complementary to rbsD in the rbsDACBK operon (19).
eThe hns mRNA levels do not change, because of H-NS autoregulation, while H-NS protein levels decrease. Values shown here are comparable to those seen
previously for primer extension quantitation (19). RpoS mRNA is stabilized, and levels slightly increased (18).
fThe rcsA transcript levels were determined previously (34). Visual estimate of primer extension was ?10-fold. Activity of a reporter gene fusion was elevated 13-fold.
gThe nfnB gene was unaffected by DsrA overproduction and was used as a control.
VOL. 186, 2004 DsrA RNA AS ACID RESISTANCE REGULATOR IN E. COLI6181

Page 4

mised in its ability to regulate hns or rpoS (Fig. 1, all panels,
lane M) (19). Variant plasmid pBRdsrA*Hgenerated levels of
DsrA*Hmutant RNA comparable to those generated by the
wild-type DsrA plasmid (Fig. 1, panel 6, compare lanes D and
M). The plasmids that produce DsrA and DsrA*Hare isogenic
except for five point mutations within the dsrA gene. DsrA*His
unable to base pair with hns mRNA and cannot significantly
activate the translation of rpoS mRNA, although DsrA*Hcan
pair with an altered hns allele (19). These data indicate that
DsrA and not other plasmid products or sequences induced
these acid resistance genes.
Acid resistance phenotype test. The pattern of gene induc-
tion described above suggests a role for DsrA in acid resis-
tance. Consistent with these observations, in preliminary ex-
periments E. coli K-12 strain M182, overexpressing DsrA from
one of several plasmids, displayed increased acid tolerance at
pH 3.8 by a factor of 12- to 5,000-fold relative to that of control
M182 cells (data not shown). To confirm a physiological role
for DsrA in acid resistance, we compared E. coli M182 to an
otherwise isogenic dsrA-null mutant (M182 dsrA::cat) for the
ability to survive immersion in pH 2 medium. Growth and acid
treatment were at 37°C. The K-12 strains with dsrA deleted
were killed more readily at pH 2 than was the wild type (Fig.
2A and B, compare filled and open circles). The downward
trend indicates a 102- to 103-fold disadvantage in survival at
low pH between 1 and 5 h for the dsrA mutant, supporting a
physiological role for DsrA.
To verify that DsrA is responsible for the acid resistance
seen, we complemented the dsrA-null strain with wild-type or
mutant DsrA produced in trans from a plasmid (Fig. 2A, filled
and open triangles). Wild-type DsrA overproduction in the
dsrA-null mutant rendered M182 strongly acid resistant even
up to 6 h of treatment at pH 2. The strain overproducing DsrA
was up to 106-fold more acid resistant than the dsrA mutant.
Curiously, we saw a partial restoration of acid resistance in
the dsrA-null mutant when we overproduced the altered
DsrA*Hvariant in trans from a plasmid (Fig. 2A, compare
open circles and open triangles). The DsrA*H-producing plas-
mid restored a level of acid resistance comparable to that of
the wild-type strain (Fig. 2A, cf. filled circles and open trian-
gles). However, wild-type DsrA was considerably more effec-
tive than DsrA*Hfor promoting acid resistance survival when
overproduced (103-fold difference at 6 h) (Fig. 2A, cf. open
and filled triangles). As DsrA*Hcannot act directly on hns or
rpoS RNA, an independent mechanism of acid resistance is
implied, possibly via direct DsrA base pairing with other
RNAs, such as, for example, putative mRNA targets argR, ilvI,
and rbsD (19). Nevertheless, the strong and persistent acid
resistance phenotype of the DsrA-overproducing strain, cou-
pled with compromised acid resistance in dsrA-null mutants
relative to the wild type, substantiates the physiological rele-
vance of DsrA in acid resistance.
Acid resistance in a pathogen. The ability to survive at a low
pH in the stomach is considered to be a factor that permits
FIG. 1. Quantitative primer extension analysis of acid resistance
genes. RNA was extracted from cells with a vector plasmid (lanes v), a
DsrA-overproducing plasmid (lanes D), or a plasmid overproducing
inactive mutant DsrA (lanes M). A minus sign indicates a no-RNA
control lane. The labeled cDNA products of primer extension were
analyzed by polyacrylamide gel electrophoresis; representative gel data
are shown. The size of the major cDNA product is given in nucleotides
below the panel. The name of each mRNA (top) corresponds to
transcripts originating from the major promoter of each gene. The
nfnB gene was tested as an unregulated control. Carets indicate rele-
vant primer extension products. A dideoxy-GTP sequencing ladder of
an unrelated RNA (Tetrahymena thermophila L-21 IVS) was used as a
size marker. The values on the left are sizes in nucleotides. nt, nucle-
otides.
FIG. 2. Plating assay for acid resistance. The percent survival of
E. coli and isogenic dsrA mutants and overexpressers is plotted against
time spent at pH 2.0. (A) Comparison of K-12 strain M182 with
different levels of DsrA and DsrA*H. Filled circles represent wild-type
M182, and open circles represent dsrA::cat, a null variant of M182.
Filled triangles represent dsrA::cat complemented with wild-type DsrA
from a plasmid; open triangles represent dsrA::cat complemented with
the plasmid-encoded dsrA*Hvariant. Results are the average of three
trials. The average standard deviation is ?11% and never greater than
23%. (B) Comparison of K-12 strain M182 and EHEC strain O157:H7.
Circles represent nonvirulent K-12 strain M182; squares represent
pathogenic E. coli O157:H7. Filled symbols represent dsrA?strains;
open symbols represent dsrA mutant strains. Representative data are
shown from at least three independent trials; all trials gave similar
trends, but with sufficient variability that the data were not superim-
posable.
6182LEASE ET AL.J. BACTERIOL.

Page 5

pathogenic bacteria to establish an infection (9 and references
therein). We theorized that the enterohemorrhagic pathogen
E. coli O157:H7 might use DsrA to induce acid resistance.
Accordingly, we assayed E. coli O157:H7 and its dsrA null
mutant for acid resistance. An E. coli O157 clinical isolate
displayed patterns of acid resistance different from those of
K-12 laboratory strain M182 (Fig. 2B, cf. circles and squares),
with the pathogenic strain E. coli O157 being considerably
more acid resistant than K-12. Again we found a clear trend of
compromised acid resistance in the dsrA::cat mutant (Fig. 2B,
cf. filled and open squares), with a 10- to 75-fold reduction in
acid resistance of E. coli O157 in the absence of DsrA.
DISCUSSION
DsrA global effects. The mRNA levels of many genes in-
crease, while levels of several mRNAs decrease in response to
DsrA overproduction (Table 1). The affected genes are dis-
tributed around the chromosome and perform multiple func-
tions (Table 1 and Fig. 3A). Most notably, acid resistance
genes are induced by DsrA overexpression (Table 1 and
Fig. 1). Apart from acid resistance, no single unifying theme
emerged from our analysis of these data, although regulated
genes include membrane proteins and sugar metabolism oper-
ons (Fig. 3A). These array data, in combination with studies of
specific genes, serve as a departure point for generation of
testable hypotheses. The role of condition-specific regulatory
factors may predominate; for example, only two of the flagellar
regulatory and synthesis genes were induced (Table 1, lines 22
and 23), and not a multigene cascade. Thus, hypotheses gen-
erated from these data could be individually tested under ap-
propriate conditions and strains, as was done for acid resis-
tance (Fig. 2). Not all genes were detected by arrays (e.g., stpA,
Table 1, line 18), possibly because of suboptimal annealing of
cDNA primers, mRNA secondary structures, or degradation at
the 3? ends.
DsrA enhances acid resistance. In comparisons of wild-type
and dsrA-null mutants, both E. coli K-12 and pathogenic O157
strains are compromised in the ability to survive a low-pH
challenge if dsrA is knocked out (Fig. 2). Also, dsrA mutants
were rendered strongly acid resistant by overproduction of
DsrA in trans. Taken together, these results imply a physiolog-
ical role for DsrA in acid resistance. Multiple acid resistance
responses (gadAX, gadBC, and hdeAB operons) are depicted as
part of a DsrA regulatory circuit (Fig. 3B). DsrA presumably
acts indirectly, via H-NS and RpoS. However, in some cases
DsrA may act directly (Fig. 3B, gray dotted arrow), by RNA-
RNA interactions with specific target mRNAs, as it does with
hns and rpoS. One such example is the periplasmic ribose-
binding protein operon (rbs) transcript, which was identified as
a potential DsrA target by virtue of complementarity (19), and
which is depressed about threefold when DsrA is induced
(Table 1, line 14).
The greater effectiveness of wild-type DsrA than DsrA*Hin
complementing the dsrA-null mutant for survival at low pH
should be considered (Fig. 2A). Clearly, DsrA*Hcannot induce
hde or gad acid resistance genes (Fig. 1). It is feasible that
either DsrA*Hor plasmid sequences titrate out a regulatory
molecule such as Hfq or LeuO, respectively (16, 36). LeuO and
Hfq have both been shown to complement an hns mutant for
repression of certain acid-inducible genes (29, 30). Another
possibility is that an additional, H-NS-independent pathway is
responsible for DsrA*Hpartially complementing the DsrA null
mutant. DsrA interacts with hns and rpoS mRNAs by different
base pairing via discrete portions of the DsrA molecule (17).
The hypothesis that DsrA might interact with another target
via different nucleotides than those mutated in DsrA*His
FIG. 3. Control by DsrA. (A) Map of DsrA-responsive loci. Trian-
gles indicate the direction of gene transcription. Black, acid resistance
genes; light grey, osmotic-shock genes; dark grey, regulatory genes;
white, other genes. Genes known to be H-NS/RpoS regulated are
underlined (see the supplemental material for references). (B) DsrA
regulatory circuits. ?, activation; ?, repression. Different environmen-
tal signals lead to increased DsrA, which binds hns mRNA to block
translation and binds rpoS mRNA to increase translation (left) (17). A
decrease in H-NS protein (2) relieves the repression of genes, as an
increase in RpoS protein (1) coordinately activates transcription of
genes, resulting in increased acid resistance and virulence. A grey
dashed line to the right of DsrA shows putative direct DsrA binding to
other target mRNAs. A secondary GadX circuit (solid gray lines)
maintains acid resistance and blocks the production of strain-specific
virulence factors, such as Per and LEE, as described in the text.
VOL. 186, 2004 DsrA RNA AS ACID RESISTANCE REGULATOR IN E. COLI6183