Amelioration of the Fitness Costs of Antibiotic Resistance
Due To Reduced Outer Membrane Permeability by
Upregulation of Alternative Porins
Michael Knopp1 and Dan I. Andersson*,1
1
Department of Medical Biochemistry and Microbiology, Uppsala University, Uppsala, Sweden
*Corresponding author: E-mail: [email protected].
Associate editor: Miriam Barlow
Abstract
The fitness cost of antibiotic resistance is a key parameter in determining the evolutionary success of resistant bacteria.
Studies of the effect of antibiotic resistance on bacterial fitness are heavily biased toward target alterations. Here we
investigated how the costs in the form of a severely impaired growth rate associated with resistance due to absence of two
major outer membrane porins can be genetically compensated. We performed an evolution experiment with 16 lineages
of a double mutant of Escherichia coli with the ompCF genes deleted, and reduced fitness and increased resistance to
different classes of antibiotics, including the carbapenems ertapenem and meropenem. After serial passage for only 250
generations, the relative growth rate increased from 0.85 to 0.99 (susceptible wild type set to 1.0). Compensation of the
costs followed two different adaptive pathways where upregulation of expression of alternative porins bypassed the need
for functional OmpCF porins. The first compensatory mechanism involved mutations in the phoR and pstS genes, causing
constitutive high-level expression of the PhoE porin. The second mechanism involved mutations in the hfq and chiX genes
that disrupted Hfq-dependent small RNA regulation, causing overexpression of the ChiP porin. Although susceptibility
was restored in compensated mutants with PhoE overexpression, evolved mutants with high ChiP expression maintained
the resistance phenotype. Our findings may explain why porin composition is often altered in resistant clinical isolates
and provide new insights into how bypass mechanisms may allow genetic adaptation to a common multidrug resistance
mechanism.
Key words: antibiotic resistance, fitness cost, compensatory evolution, Escherichia coli, porin.
Introduction
Article
The evolution and dissemination of antibiotic-resistant bacteria represent a major public health threat jeopardizing important medical advances. To rationally combat antibiotic
resistance, we need to understand which factors influence
the emergence and persistence of resistant clones. One key
factor is the fitness cost of resistance, that is, the impact the
resistance mechanism has on bacterial growth and survival. In
the presence of high levels of antibiotics, the acquisition of an
antibiotic resistance mechanism provides a fitness advantage,
in comparison with susceptible competitors, whereas in the
absence of antibiotics resistance mechanisms generally confer
deleterious effects, typically observed as an increased generation time and reduced survival inside a host (Andersson and
Levin 1999; Andersson and Hughes 2010). These observations
suggest that resistant bacteria should generally be outcompeted by susceptible, higher fitness populations after removal
of the selective pressure. However, several factors can cause
the resistance to be maintained: 1) coselection of resistance
genes with other functions that confer a fitness advantage
(Enne et al. 2001; Baker-Austin et al. 2006; Yates et al. 2006), 2)
presence of resistances that impose a very low or no cost
(Ramadhan and Hegedus 2005; Criswell et al. 2006), 3) selection for resistant strains by the presence of very low amounts
of antibiotics and heavy metals present as environmental
pollutants (Gullberg et al. 2011, 2014), and 4) compensatory
evolution that reduces the fitness cost, often without loss of
the resistance (Lenski 1998; Schulz zur Wiesch et al. 2010;
Andersson and Hughes 2011). The latter entails the acquisition of second site mutations (intra- or extragenic), generating higher fitness-resistant strains that are competitive even
in the absence of antibiotics. Understanding compensatory
evolution is of high importance to allow prediction of the
persistence and dissemination of antibiotic resistance, but to
date studies of fitness compensation have focused almost
exclusively on resistance caused by target alterations. In
these studies, evolution toward higher fitness was generally
achieved by mutations in the same target gene, which restore
functionality, or by mutations in direct interaction partners of
the target gene (Levin et al. 2000; Andersson and Hughes
2010; Maclean et al. 2010). Compensation of a multidrug
resistance mechanism, such as the reduction in membrane
permeability, has to our knowledge not yet been studied.
The outer membrane of Gram-negative bacteria forms the
most exterior barrier between the environment and the bacterium. It is composed of phospholipids and lipopolysaccharides, which form a hydrophobic coat impermeable to
hydrophilic compounds. Although this barrier prevents the
ß The Author 2015. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please
e-mail: [email protected]
3252
Mol. Biol. Evol. 32(12):3252–3263 doi:10.1093/molbev/msv195 Advance Access publication September 9, 2015
MBE
Fitness Costs of Antibiotic Resistance . doi:10.1093/molbev/msv195
import of many toxic substances, it also hinders the export of
such. Furthermore, external nutrients are not able to freely
diffuse through this hydrophobic layer. To allow an efficient
exchange with the environment, the outer membrane contains outer membrane proteins (OMPs) that form hydrophilic
channels (pores) facilitating nutrient uptake and export of
toxic substances. Escherichia coli encodes three major
porins, OmpC, OmpF, and PhoE. These proteins fold into
ß-barrels and assemble into trimers in the outer membrane
forming aqueous pores (Gehring and Nikaido 1989; Cowan
et al. 1992, 1995; Basle et al. 2006). The pores allow the passive
diffusion of molecules smaller than 500–600 Da, with OmpC
having a slightly smaller exclusion size than OmpF and PhoE
(Nikaido and Rosenberg 1983). OmpR, the response regulator
of the EnvZ–OmpR two-component system, regulates expression of OmpC and OmpF (Sarma and Reeves 1977; Hall
and Silhavy 1981). Depending on the osmolarity of the surrounding medium, OmpR enhances transcription of either
ompC or ompF. Even though both these porins are expressed
constitutively to some extent, OmpR is crucial for high-level
expression (Mizuno and Mizushima 1987). At the posttranscriptional level, ompC and ompF, as well as the majority
of other OMPs, are also regulated by Hfq-dependent small
RNAs (sRNAs). Hfq acts as an RNA chaperone and mediates
the pairing of sRNAs with their trans-encoded targets, which
enables or represses translation or degradation of the target
mRNA, depending on the regulatory function of the sRNA.
Examples of regulatory sRNAs targeting the expression of
OMPs are MicA, MicC, MicF, OmrAB, or ChiX negatively
regulating the expression of ompA, ompC, ompF, ompT, and
chiP, respectively (Guillier et al. 2006; Vogel and Papenfort
2006; Valentin-Hansen et al. 2007). The third major porin,
PhoE, is a member of the Pho regulon. This regulon contains
over 47 genes (VanBogelen et al. 1996; Han et al. 1999;
Suziedeliene et al. 1999; Baek and Lee 2006) whose expression
is under the control of the PhoRB two-component regulatory
system and is induced upon phosphate starvation
(Vershinina and Znamenskaia 2002; Lamarche et al. 2008).
Under phosphate-limiting conditions, the response regulator
PhoB is phosphorylated by the sensor kinase PhoR. After
phosphorylation, PhoB can bind to Pho boxes and enable
transcription of Pho-regulated genes. In the presence of
high levels of phosphate, PhoR enters an alternative state,
which disturbs PhoB phosphorylation. The detailed mechanism of this inhibition is still unclear, but PhoU is essential for
this regulatory process. PhoU is located in an operon together
with the Pst system (phosphate-specific transporter). These
genes, encoding PstS, PstC, PstA, and PstB, form a highaffinity/low-velocity phosphate uptake system in the inner
membrane and are crucial for PhoR/PhoB regulation.
Besides its transport function, the Pst system acts as a
sensor of external Pi concentrations and transmits the
signal to PhoR, inducing an activation or inhibition complex.
The Pst proteins play a crucial role in sensing periplasmic
phosphate levels and activating expression of Pho-regulated
genes (Hsieh and Wanner 2010) and it was previously shown
that loss of function mutations in any members of the
pstSCAB–phoU operon lead to constitutive activation of
the Pho regulon (for a review see Lamarche et al. 2008).
Besides the three major porins OmpC, OmpF, and PhoE, E.
coli encodes an additional set of pore-forming OMPs. Among
these are unspecific porins like OmpN and OmpG, which
allow passive diffusion of small hydrophilic molecules, as
well as specialized porins, which only facilitate the uptake of
specific substrates. Examples of specific porins are ChiP and
LamB, which enable the uptake of chitobiose and maltodextrin, respectively (Wang et al. 1997; Plumbridge et al. 2014).
OMPs, especially the abundant major porins OmpC and
OmpF, play an important role in the development of antibiotic resistance. Many antibiotics, for example, carbapenems,
quinolones, aminoglycosides, and tetracyclines, cannot easily
diffuse through the outer membrane (Delcour 2009). Instead,
these antibiotics utilize porins to overcome the hydrophobic
barrier and translocate into the periplasm. The molecular
mechanisms of antibiotic uptake by porins have been studied
intensively (Ceccarelli et al. 2004; Danelon et al. 2006; Hajjar
et al. 2010; Ziervogel and Roux 2013). Considering the importance of porins for antibiotic entry into the periplasm, it is not
surprising that loss of porins can increase resistance toward
multiple antibiotics (Delcour 2009). Often, single point mutations alone that slightly alter the pore properties are sufficient to significantly decrease antibiotic uptake (Simonet et al.
2000; Bredin et al. 2002), and mutations that lead to loss or
alterations of porins are frequent among resistant clinical
isolates: A study analyzing porin abundance in 45 ß-lactamresistant clinical isolates of E. aerogenes showed that over 40%
lacked porins (Charrel et al. 1996). Similarly, sequencing of the
two major porins OmpK36 and OmpK35, which are homologs of the E. coli OmpC and OmpF, in five pan-resistant
clinical isolates of Klebsiella pneumoniae showed that all
strains carried mutations in both the major porins (Shi
et al. 2013).
Antibiotic resistance due to reduced porin expression and
membrane permeability comes at a significant cost, because
simultaneous with the exclusion of antibiotics important nutrients are concomitantly excluded from the periplasm
(Ferenci 2005). During antibiotic treatment, a decreased nutrient uptake might be an acceptable trade-off; however, in
the absence of antibiotics an impermeable outer membrane
will be deleterious. We present a detailed study on how E. coli
copes with porin loss in the absence of antibiotics and what
evolutionary pathways are available to ameliorate the fitness
cost due to loss of the two major porins OmpC and OmpF.
Results
Evolution of Porin-Deficient Mutants and
Identification of Compensatory Mutations
Removal of the two major porins OmpC and OmpF in E. coli
increased the resistance to carbapenems by 4- (meropenem)
to 60-fold (ertapenem) with an associated reduction in exponential growth rate of 15% in the absence of drugs (fig. 1). To
investigate if and how this cost can be genetically reduced, we
conducted an evolution experiment starting with 16 parallel
lineages of a parental strain lacking OmpC and OmpF
(DA31944). All lineages were serially passaged daily by
3253
Knopp and Andersson . doi:10.1093/molbev/msv195
MBE
Fig. 1. Relative growth rate and resistance profile of the ompCF double knockout and eight evolved growth-compensated mutants. The exponential
growth rates are presented as relative values (filled circles) compared with the wild-type growth rate on the primary y-axis. The dotted line indicates the
relative wild-type growth rate and the corresponding genotype of each strain is indicated above the graph. Each data point represents the average of
three biological and three technical replicates; the error bars represent the standard deviation. The MIC values of all investigated strains are shown as
bars on the secondary y-axis.
1:1,000 dilutions in 1 ml Lysogeny broth (LB) medium,
resulting in approximately 10 generations of growth per
day. After 250 generations, we isolated individual clones
from each population and assessed the exponential growth
rate and the resistance phenotype (supplementary fig. S1,
Supplementary Material online). Fourteen of the 16 clones
showed a significant increase in growth rate compared with
the unevolved parental strain, with relative growth rates ranging from 0.89 to 0.99 (as compared with 0.85 in the parental
ompCF double mutant). Eight clones that showed the highest
increase in fitness were chosen for subsequent analysis (fig. 1).
Notably, in some instances the increase in growth rate was
accompanied by a complete loss of the resistance phenotype,
while other mutants maintained partial or full resistance to
ertapenem and meropenem.
Eight clones were whole-genome sequenced to investigate
the genetic basis of adaptation (table 1). All the mutants
contained 1–2 SNPs and/or 1–2 insertion sequence (IS) element insertions. No duplications, deletions, or inversions were
observed. Genetic changes were found mainly in five different
genes (hfq, chiX, phoR, pstS, and rpoS) in different combinations. Among these, hfq and phoR were found to be mutated
in multiple strains, whereas the two mutants that carried the
wild-type alleles of phoR and hfq (DA32616 and DA32628)
carried mutations in genes directly associated with Hfq or
PhoR function: DA32616 contained an insertion in the chiX
gene, encoding an sRNA which regulates the porin ChiP in an
Hfq-dependent fashion (Rasmussen et al. 2009), while
DA32628 carried an insertion in pstS, a known suppressor
of the Pho regulon (Ruiz and Silhavy 2003). Four mutants
carried a mutation inside or upstream rpoS. This gene
3254
encodes the alternative sigma factor RpoS which is involved
in the expression of genes related to stress responses and
secondary metabolism (Battesti et al. 2011). The order in
which the mutations occurred during the evolution experiment was determined by Sanger sequencing the relevant
genes from frozen populations isolated at different time
points. The two insertions in chiX and rpoS present in
DA32616 could be detected at the earliest time point.
Isolation and sequencing of DNA from individual clones
from populations saved after 100 and 150 generations
showed either wild-type alleles of chiX and rpoS or double
mutants carrying insertions in both genes, indicating that
both insertions occurred contemporaneously early during
the evolution experiment (supplementary tables S4 and S5,
Supplementary Material online). In general, mutations in
phoR, hfq, and pstS occurred very early during the evolution
experiment, while mutations in rpoS, gltP, and sapC could
only be detected after 200 or 250 generations.
Analysis of Growth-Compensated Mutants Showed
that phoR and pstS Mutations Upregulate PhoE,
Whereas hfq and chiX Mutations Upregulate ChiP
We hypothesized that the compensatory mutations caused
an upregulation of alternative porins that could compensate
for the lack of OmpC and OmpF. To investigate the role of
the phoR and pstS mutations in compensation with porin
loss, we first measured the activity of the alkaline phosphatase
(AP) PhoA in the growth-compensated mutants in stationary
and exponential growth phase (fig. 2). This enzyme is a
member of the Pho regulon and can be used as an indicator
Fitness Costs of Antibiotic Resistance . doi:10.1093/molbev/msv195
Table 1. Mutations in growth-compensated mutants.
Strain
DA32613
DA32616
DA32619
DA32620
DA32621
DA32622
DA32627
DA32628
SNP
phoR M189Ra
rpoS I128Nb
—
hfq R16S
hfq P10La
(sapC S69P)b
hfq P10L
phoR T151del
phoR L146Pa
rpoS T-21Gb
(gltP A-115T)c
Insertion
—
chiX:insAa
rpoS-71insBa
—
—
—
—
pstS:insAa
rpoS -528insHb
Note.—Within parenthesis are mutations that are unlikely to contribute to the
observed phenotypes.
a–c
The order in which the mutations appeared in the populations.
of the activation level of this regulon. In the stationary phase,
AP activity was increased 200- to 500-fold in the growthcompensated mutants that carried a mutation in either
phoR or pstS, indicating that these mutations result in an
activation of the Pho regulon. In contrast, strains carrying
mutations in hfq or chiX showed the same AP activity as
the wild type, demonstrating that these mutations do not
influence the Pho regulon. Interestingly, the unevolved porindeficient mutant exhibited high AP concentrations in exponential phase. This activation was maintained for a phoR
mutant (DA32613), while wild-type levels were restored in
an hfq mutant (DA32621).
To further examine whether the compensatory mutations
cause upregulation of alternative porins, we used qPCR to
determine transcription levels of several different porins.
The Pho regulon contains the third major porin in E. coli,
PhoE. Under normal conditions the PhoR and PstS proteins
repress phoE expression. To test whether the expression of
phoE was altered in the compensated mutants, we determined its transcription level in a selected set of evolved mutants (fig. 3). As predicted, coinciding with increased PhoA
activity, phoE expression was strongly upregulated in strains
carrying a mutation in phoR or pstS, whereas mutations in hfq
and chiX had no or smaller effects on PhoE expression.
Besides PhoE, the E. coli genome encodes several OMPs,
many of which constitute pore-forming channels. To test
whether upregulation of other porins contribute to the
growth compensation in evolved OmpCF-deficient double
mutants, we analyzed the expression levels of four other
genes encoding putative outer membrane pores: ompG,
ompN, ompW, and chiP. No difference in expression levels
was observed for ompG, ompN, or ompW in any of the tested
strains. In contrast, the expression of chiP, a specific porin
facilitating the uptake of chitooligosaccharide (Plumbridge
et al. 2014), was increased in DA32616 (IS insertions in chiX
and rpoS) and DA32621 (hfq P10L) by 76- and 61-fold, respectively. Expression of the OMP ChiP is tightly repressed by
MBE
the sRNA ChiX that binds to the 50 UTR of the chiP mRNA
and sequesters the ribosome binding site (Rasmussen et al.
2009). Furthermore, Hfq is essential for the regulatory function of ChiX (Zhang et al. 2003; Mandin and Gottesman
2009), and mutations in hfq can disrupt the binding affinity
of Hfq to its substrates, thereby disrupting the regulation by
sRNAs. Thus, the mutations found in chiX (DA32616) and
hfq (DA32621) result in the same downstream effect:
Activation of chiP expression, most likely by impairing the
repressing function of the ChiX sRNA. Unexpectedly, the
transcript level of the second regulatory target of ChiX, the
sensor kinase DpiB of the two-component system DpiAB,
was not increased in any of the analyzed strains (fig. 3).
Outer Membrane Profiling Shows that PhoE Is
Upregulated in phoR and pstS Mutants
The outer membrane of Gram-negative bacteria can be selectively purified and OMPs extracted. In wild-type E. coli K12
(strain MG1655), the porins PhoE and ChiP cannot be easily
detected under normal conditions. To examine porin expression in growth-compensated mutants, we analyzed the protein composition of the outer membrane of the wild type
(DA4201), the unevolved OmpCF-deficient mutant
(DA31944), and four selected growth-compensated mutants
(fig. 4). Consistent with the transcriptional analysis, PhoE
levels were increased in strains carrying a mutation in phoR
(DA32622) or pstS (DA32628), whereas in the growthcompensated strains with mutations in hfq (DA32621) or
chiX (DA32616), PhoE levels remained low. We were not
able to detect ChiP in any of the samples. Despite the strongly
increased transcription of chiP, we suspect that the protein
level is still too low to be detected with this method. We tried
to detect ChiP by introducing a C-terminal FLAG tag to the
wild-type and mutant strains (for strain numbers see supplementary table S1, Supplementary Material online), but even
with this tag attached no ChiP protein was detectable in a
western blot (data not shown), possibly because the tag was
not accessible to the antibody or it prevented proper folding
into the membrane and therefore led to degradation of ChiP.
Expression of chiP and phoE Is Both Necessary and
Sufficient for Compensation of the Growth Defects
Caused by ompCF Deletion
We performed two different tests to unambiguously demonstrate the need for ChiP and PhoE upregulation for growth
compensation in the evolved mutants. If ChiP and PhoE are
necessary and sufficient to confer compensation, it is expected that deletion of the corresponding genes should prevent compensation and conversely that overexpression
should confer compensation. First, we constructed chiPand phoE-knockout mutations in the wild type, in the
unevolved porin-deficient mutant, and in a subset of the
compensated strains. Although removal of chiP and phoE
had no effect on exponential growth rate in the wild-type
(DA4201) and the unevolved porin-deficient mutant
(DA31944), the effects in the evolved strains were as expected
(fig. 5). Thus, deletion of chiP from mutants carrying
3255
Knopp and Andersson . doi:10.1093/molbev/msv195
MBE
Fig. 2. Relative AP activities in exponential and stationary phase. AP activities from two OD units of stationary phase culture (dark grey) and
exponential phase cultures (light grey) were measured from three biological replicates. All values are relative to the AP activity in the wild-type
sample (DA4201) in stationary phase (set to 1.0). The error bars represent the standard deviation. The relevant mutated genes of each strain are
indicated above the chart.
Fig. 3. mRNA expression analysis of selected genes. Transcript levels of ompG, ompN, ompW, dpiB, phoE, and chiP were measured using qRT-PCR. RNA
was harvested in exponential phase (OD600 = 0.15) from three biological replicates. Error bars represent the standard deviation. Values for each gene are
relative to the corresponding wild-type value (set to 1.0). The relevant mutated genes of each strain are indicated above the chart.
compensatory mutations in chiX (DA32616) and hfq
(DA32621) reduced the exponential growth rate by 6% and
4%, respectively, whereas deletion of phoE from these genetic
backgrounds caused no reduction in exponential growth rate.
Conversely, removal of the phoE gene from mutants harboring mutations in phoR (DA32622) or pstS (DA32628) decreased exponential growth rate by 13% and 21%,
respectively, whereas deletion of chiP in these mutants did
3256
not change the growth rate. These results demonstrate that
the compensatory effect of the chiX and hfq mutations requires a functional chiP gene and that the compensatory
effect of the phoR and pstS mutations requires a functional
phoE gene.
As a second test, we investigated if overexpression of alternative porins could abrogate the growth defects of the
unevolved porin-deficient mutant. We used six different
Fitness Costs of Antibiotic Resistance . doi:10.1093/molbev/msv195
Fig. 4. Outer membrane profiling. OMPs were extracted in exponential
phase and separated on an SDS-Gel containing 10% acrylamide and 6 M
urea. Protein bands were visualized using colloidal coomassie staining.
Shown is a fraction of the gel containing OmpC, OmpA, and PhoE. The
proteins were cut out and identified using mass spectrometry analysis.
The relevant mutated genes of each strain are indicated above the
picture.
plasmids from the ASKA library encoding chiP, phoE, ompN,
ompG, ompW, and an empty vector control to study the
influence of these porins on exponential growth rates.
Because strong induction of porin expression with the high
copy ASKA-vector pCA24N had a deleterious effect on
growth even in a wild-type control (supplementary fig. S2,
Supplementary Material online), we used a relatively low
(15 mM) concentration of the inducer Isopropyl ß-D-1-thiogalactopyranoside (IPTG). As expected, overexpression of either
ChiP or PhoE could completely restore growth in an OmpCFdeficient strain, whereas overexpression of OmpN or OmpW
did not increase the exponential growth rate (fig. 6).
Surprisingly, OmpG was also able to compensate for the
loss of OmpC and OmpF. These results show that overexpression of either ChiP, PhoE, or OmpG is sufficient to compensate for the fitness reduction caused by lack of OmpCF.
Discussion
To better understand how antibiotic resistance evolves and
persists in bacterial populations, we need to understand how
bacteria can overcome resistance-associated fitness costs by
compensatory mutations. So far, studies of fitness compensation have been strongly biased toward resistances based on
target alterations (Andersson and Hughes 2010). Here, we
present a detailed genetic analysis of the evolutionary pathways that can ameliorate the growth defects conferred by a
general resistance mechanism (see supplementary table S3,
Supplementary Material online, for resistance profile to different antibiotic classes) that reduces permeability of the
outer membrane. The loss of outer membrane porins as a
result of antibiotic pressure is commonly observed among
clinical isolates (Fernandez and Hancock 2012). We found
that the reduction in fitness caused by porin loss can be
rapidly compensated. Thus, 250 generations of growth were
sufficient for the evolution of high-fitness mutants in almost
all tested lineages. Based on our results, we propose a model
where the exponential growth rate in an OmpCF-deficient
strain of E. coli K12 MG1655 is restored by upregulation of
either of the porins PhoE or ChiP. This upregulation was
achieved by mutations in genes involved in two different
regulatory pathways: 1) by activation of the Pho regulon
due to mutations in phoR and pstS or 2) by activation of
MBE
chiP expression by perturbation of Hfq-dependent sRNA regulation due to mutations in chiX and hfq (fig. 7). We demonstrated by quantification real-time PCR (qRT-PCR) (fig. 3),
AP activity assays (fig. 2), and outer membrane profiling
(fig. 4) that the dysregulation of these pathways leads to an
increased expression of 1) PhoE and 2) ChiP. Furthermore, by
either removing or overexpressing PhoE and ChiP (figs. 5
and 6) we showed that these porins are needed and sufficient
to confer full or partial growth compensation in the ompCF
double mutant.
Mutations acquired in phoR and pstS in evolved ompCFdeficient mutants resulted in a constitute activation of the
Pho regulon and restored the growth rate by overproduction
of PhoE. We hypothesize that PhoE can replace the physiological function of OmpC and OmpF allowing uptake of compounds that are growth limiting in the unevolved ompCF
mutant. Associated with the restoration of growth PhoE overexpression also restored susceptibility toward ertapenem and
meropenem, suggesting that PhoE allows uptake of these
antibiotics similarly to OmpC and OmpF. Three of the four
mutants that carried a mutation in phoR or pstS also contained a mutation in or upstream of rpoS, a gene encoding
the alternative sigma factor RpoS. The mutant allele found in
DA32613 (rpoS I128N) was reported previously to have a
strongly decreased activity behaving like an RpoS null
mutant (Spira et al. 2011). The nature of the other mutations
found in rpoS, an SNP in the promoter, and two insertions
disrupting transcriptional and translational start sites are
likely to result in decreased expression and therefore decreased RpoS activity in these mutants as well. It was
shown previously that overexpression of RpoS increases the
expression of pstS and reduces the expression of the other
Pho-regulated genes. Vice versa, a mutant lacking rpoS shows
a strong increased activity of Pho-regulated genes (Taschner
et al. 2004). It is therefore likely that the mutations in rpoS
observed in some evolved mutants contribute to the growth
compensation by increasing the expression of phoE. However,
mutations in rpoS were always accompanied by additional
mutations, suggesting only a minor influence on growth compensation in exponential phase. This notion is supported by
the observation that a mutation in phoR (DA32622) increased
the growth rate to the same extent as double mutants carrying a mutant allele of phoR and rpoS (DA32613 and
DA32627) or pstS and rpoS (DA32628). We therefore suggest
that rpoS mutations are not the primary cause of phoE upregulation, but possibly they provide additional fitness benefits,
for example, by increasing growth yield in stationary phase
(Paulander et al. 2009). Surprisingly, Pho-regulated genes were
already upregulated (to a lower extent) in the unevolved
ompCF-deficient mutant in exponential phase. This could
indicate a stress-response mechanism due to nutrient limitation in an import-deficient strain of E. coli. Deletion of phoE in
the unevolved mutant did not further reduce the growth rate,
suggesting that the observed increase in phoE expression was
not high enough to affect the growth rate (fig. 5). This is
supported by the fact that we did not see a high abundance
of PhoE in the outer membrane profile of the unevolved
mutant (fig. 4), even though the mRNA levels of phoE were
3257
Knopp and Andersson . doi:10.1093/molbev/msv195
MBE
Fig. 5. Effects of removal of chiP and phoE on exponential growth rates. The exponential growth rates are presented as relative values compared with
the wild-type growth rate (set to 1.0). White bars represent the growth rate of the unmodified strains, dark gray bars represent the growth rate after
deletion of chiP, and light grey bars represent the growth rates after deletion of phoE in the corresponding strains. The difference between bars marked
with *** are considered statistically significant with P values below 0.0001. All values are based on three biological and three technical replicates. The
relevant mutated genes of each strain are indicated above the chart.
Fig. 6. Overexpression of OMPs. Plasmids for expression of selected porins were transformed into the unevolved porin-deficient mutant (DA31944).
The plasmid was selected with 15 mg/l chloramphenicol and expression was induced with 15 mM IPTG. All values are relative to a wild type (DA4201)
carrying the corresponding plasmid. All values are based on three biological and three technical replicates. The insert encoded in the ASKA-plasmid
pCA24N carried by each strain is indicated above the chart.
strongly elevated even without further compensatory mutations (fig. 3). The increase in phoE expression was not observed in stationary phase, possibly due to RpoS-mediated
repression of the Pho regulon.
Besides upregulation of PhoE, we discovered an alternative
independent pathway to growth compensation mediated by
ChiP. Unlike OmpC, OmpF, and PhoE, ChiP does not function
as an unspecific pore, but facilitates the specialized uptake of
chitin-derived oligosaccharides (Figueroa-Bossi et al. 2009).
Under normal conditions, the sRNA ChiX represses the expression of chiP. This Hfq-dependent sRNA is constitutively
expressed and inhibits the chiP mRNA by sequestering its
3258
ribosomal binding site (Rasmussen et al. 2009). In the presence of chitosugars, the expression of the chitobiose operon
chbBCARFG is induced, which contains an exceptional trapping mechanism for the sRNA ChiX. Pairing of the sRNA with
the trap triggers its degradation, which ultimately leads to an
increased translation of the chiP transcript (Figueroa-Bossi
et al. 2009; Overgaard et al. 2009). It is reported that ChiP is
constitutively expressed in chiX deletion mutants (Rasmussen
et al. 2009). This coincides with the high expression of chiP in
DA32616, which carries an insertion in chiX disrupting its
regulatory function. In addition to ChiX inactivation, single
point mutations in hfq (DA32621) could derepress expression
Fitness Costs of Antibiotic Resistance . doi:10.1093/molbev/msv195
Fig. 7. Model for compensation of ompCF deficiency in Escherichia coli.
Antibiotic pressure can select for porin loss resulting in resistant mutants with reduced membrane permeability and fitness. Evolutionary
adaptation to reduce the fitness costs selects for mutations in multiple
negative regulators that cause either 1) overactivation of the Pho regulon and a resulting upregulation of phoE or 2) perturbation of sRNA
regulation resulting in chiP upregulation. Genes marked with * acquired
loss of function mutations, while mutations in genes marked with **
resulted in a reduced expression/functionality.
of the chiP transcript, presumably by altering the binding to
ChiX. Hfq-dependent regulation of OMPs by sRNAs is
common and the estimates suggest that about one-third of
known sRNAs are involved in regulation of outer membrane
components (Vogel and Papenfort 2006; Valentin-Hansen
et al. 2007). One of our growth-compensated mutants acquired a mutation in the arginine residue at position 16. This
residue, positioned on the rim of Hfq, is highly conserved and
essential for full functionality of Hfq-dependent regulation
and mutations in this specific position disturb the binding
of Hfq to its sRNA substrates (Panja et al. 2013). In another
study comparing the regulatory capabilities of 14 different
Hfq mutants for several substrates, it was shown that mutations in Hfq could completely diminish some sRNA-regulatory pathways while leaving others unaltered (Zhang et al.
2013). Mutations in position 16 in particular were shown to
render ChiX nonfunctional. Complete loss of Hfq was shown
MBE
previously to cause increased susceptibility of E. coli to toxic
compounds, including the ß-lactam antibiotics oxacillin and
cefamandole (Yamada et al. 2010). Because hfq mutants selected in our study largely maintained their resistance phenotype, it further demonstrates that the functionality of Hfq is
likely to be only partly impaired rather than completely lost.
The P10L mutation that we observed in this study (DA32620
and DA32621) has not been described previously, but our
qRT-PCR showed that ChiX-dependent chiP sequestration
is severely impaired in these mutants, suggesting a similar
mechanism of action. Surprisingly, the second target of
ChiX, the dpiB transcript, was not affected by either the insertion in chiX or the mutant hfq alleles P10L and R16S, indicating secondary regulatory mechanisms for dpiB expression.
Considering that the deletion of chiP from growth-compensated mutants with mutations in hfq (DA32621) and chiX
(DA32616) did not abolish the growth compensation completely suggests that other factors might also contribute to
the increased growth rate in these strains. Interestingly, compensation via chiP upregulation restored PhoA activity in exponential phase to wild-type levels and it is possible that the
potential stress signal causing activation of the Pho regulon in
the unevolved mutant is abrogated by elevated ChiP
expression.
A major difference between compensation via PhoE and
ChiP is the effect on the resistance phenotype: While PhoEmediated compensation is accompanied by a complete loss
of resistance, mutants that evolved high-level expression of
ChiP were able to maintain the resistance at least partly. An
explanation for this effect could be that ChiP does not allow
passage of ertapenem and meropenem to the same extent as
OmpC, OmpF, or PhoE. Overproduction of ChiP would therefore lead to high-fitness strains that selectively exclude carbapenems, an unwanted outcome in a clinical setting. In fact, a
recent study by Tsai et al. (2015) demonstrated that overproduction of ImpR, the ChiP homolog in Proteus mirabilis,
induces decreased carbapenem susceptibility. Similar to our
results, ImpR was upregulated in an hfq mutant, increasing its
resistance toward ertapenem and meropenem compared
with a wild type, demonstrating the importance of ChiP for
carbapenems resistance across species.
Although the compensatory capability of PhoE overexpression is expected due to the high similarity to OmpC and
OmpF, compensation by ChiP overexpression was surprising.
Interestingly, overexpression of ChiP restored the activity of
Pho-regulated genes in exponential phase to wild-type levels.
This finding shows that the presumed stress signal, leading to
Pho activation in exponential phase in an ompCF mutant, is
removed by overproduction of ChiP. Overexpression of multiple porins showed a high specificity, with PhoE and ChiP
being able to restore the growth rate to wild-type levels. In
spite of being highly homologous to the three major porins
OmpC, OmpF, and PhoE (Prilipov et al. 1998), OmpN was not
able to functionally replace them, demonstrating the high
specificity of porin function and potential for compensation.
However, growth rates could be compensated by a third
porin, OmpG, which forms large pores allowing unspecific
transport of compounds up to 600 Da (Fajardo et al. 1998;
3259
MBE
Knopp and Andersson . doi:10.1093/molbev/msv195
Subbarao and van den Berg 2006). Expression of this protein
in E. coli has been observed only after rare chromosomal
rearrangements placing it under a new promoter. In a wildtype strain it is believed that ompG lacks a functional promotor (Misra and Benson 1989). In contrast, expression of the
phoE and chiP genes is under extensive negative regulation
that can be disrupted at multiple levels. Thus, the high
number of mutational targets that can lead to PhoE and
ChiP overexpression, as compared with OmpG overexpression, presumably explains why OmpG upregulation was not
observed in our evolution experiment.
We focused on exponential growth rate as an output for
fitness in our experimental set-up, and future studies are
needed to examine the influence of PhoE and ChiP upregulation on other aspects of bacterial fitness. Even though ompCF
double mutants are attenuated in vivo when given orally
(Chatfield et al. 1991), the frequent occurrences of these mutants among clinical isolates document their clinical relevance. Furthermore, it is known that inactivation of the Pst
system in E. coli can reduce virulence, leading to increased
serum sensitivity, impaired capsular antigen display, as well as
a reduced resistance toward acids and antimicrobial peptides
(Harel et al. 1992; Ngeleka et al. 1992; Daigle et al. 1995;
Lamarche et al. 2005). Similarly, mutations in the PhoR–
PhoB two-component system that induce constitutive activation of Pho-regulated genes have been reported to reduce
virulence in Agrobacterium tumefaciens (Mantis and Winans
1993) and Corynebacterium glutamicum (Kocan et al. 2006).
Despite the negative effects on virulence, alterations in PhoE
expression were observed in a pair of K. pneumoniae isolated
from the same patient (Kaczmarek et al. 2006). One of the
strains displayed high phoE expression and low resistance,
while the other one displayed low phoE expression while
maintaining high-level resistance. This provides strong evidence that compensatory evolution of uptake deficiencies
does occur in clinical settings, and that the adaptive pathways
presented in our study might be significant for the evolution
of carbapenem-resistant, high-fitness strains in vivo.
Materials and Methods
wild-type strain (DA4201) and the resistance cassette was
removed by FLP recombination using the plasmid pCP20.
In a consecutive step, ompF was removed likewise, using E.
coli MG1655ompC as a recipient. Other deletion mutants
described in this study were constructed by P1 transduction
from a corresponding clone from the Keio collection (Baba
et al. 2006) and subsequent removal of the kanamycin cassette by introduction of pCP20 (Datsenko and Wanner 2000).
Evolution Experiments
Sixteen independent lineages of E. coli MG1655
ompCompF were grown overnight in 1 ml LB to full density.
After 24 h, 1 ml (about 5 106 cells) of the culture was transferred into 1 ml fresh medium, allowing the growth of approximately 10 generations per serial passage. This procedure
was repeated for 250 generations. Every 50 generations, a
fraction of the overnight culture was saved at 80 C. At
the end of the evolution experiment, each population was
plated on LA to obtain individual clones and one clone of
each lineage was saved at 80 C and characterized with
respect to growth rate and resistance phenotype. Eight selected clones were subjected to genomic DNA isolation using
the Genomic Tip 100 G DNA kit (Qiagen). Whole-genome
sequencing of these clones was performed at Beijing Genome
Institute, China, using Illumina sequencing technology.
Determination of Exponential Growth Rate
Three independent overnight cultures were diluted 1:1,000 in
LB, corresponding to ~5 106 CFU/ml. Aliquots of three
times 300 ml per dilution were transferred into BioscreenC
plates (Oy Growth Curves Ab, Ltd). The samples were
grown at 37 C with shaking for 18 h in a BioscreenC
Analyzer and OD600 values were monitored every 4 min.
The calculation of the maximum growth rates was based
on OD600 values between 0.024 and 0.09, where growth was
observed to be exponential. Relative growth rates were calculated by dividing the generation time of the sample by the
generation time of the parental strain, usually DA4201, which
was included in each individual experiment.
Bacterial Strains and Growth Conditions
Minimal Inhibitory Concentration Determination
All strains used in this study were derived from E. coli MG1655
(DA4201) and are listed in supplementary table S1,
Supplementary Material online. Unless otherwise indicated,
bacteria were cultivated in LB for liquid growth and Lysogeny
broth agar (LA) for growth on solid media at 37 C. When
appropriate, liquid and solid media were supplemented with
12.5 mg/l chloramphenicol, 50 mg/l kanamycin, or 15 mM
IPTG.
The ompCF double knockout was constructed using
lambda Red-mediated recombineering (Datsenko and
Wanner 2000). For this, a chloramphenicol cassette flanked
by FLP recombinase target sequences was PCR amplified from
plasmid pKD3 (GenBank accession number AY048742.1). The
primers carried overhangs of 35 nt homologous to the adjacent regions of ompC. After introduction of the chloramphenicol cassette, it was moved by P1 transduction into a
The minimal inhibitory concentrations were determined
using Etest strips (BioMerieux) according to the manufacturer’s recommendations. In brief, bacteria were grown overnight in Mueller–Hinton (MH) broth and diluted 1,000-fold
before spreading on MH-agar plates. The Etest strips were
placed onto the plate and the result was read after 20 h of
incubation at 37 C.
3260
AP Assay
The AP activity in the samples was measured using the colorimetric SensoLyte pNPP Alkaline Phosphatase Assay Kit
(AnaSpec) according to the manufacturer’s recommendations. In brief, cells were grown overnight in 1-ml LB and
washed twice with 500 ml 1 assay buffer. The cells were
pelleted and dissolved in 150 ml 1 assay buffer with 3 ml
Triton-X-100. After 10-min incubation at room temperature,
MBE
Fitness Costs of Antibiotic Resistance . doi:10.1093/molbev/msv195
the debris was pelleted and 50-ml supernatant was transferred
to a 96-well plate. After addition of 50-ml pNPP, the samples
were incubated for 60 min at room temperature and the absorbance at 405 nm was measured. An AP standard of known
concentration was included to calculate the total activity of
AP in the samples. For detection of AP activity in exponential
growing cells, an overnight culture was diluted 1:500 in 20-ml
prewarmed LB and grown until OD600 = 0.15. Two OD600
units (13.3 ml) were pelleted and used for subsequent AP
detection. Values acquired from exponential phase cultures
were multiplied by four to compensate for the difference in
cell number compared with stationary phase cultures. The AP
activity is presented as relative values compared with wild
type (DA4201) in stationary phase.
Isolation of Total RNA and Relative qRT-PCR
Overnight cultures were diluted 1:500 in 20-ml prewarmed LB
and incubated at 37 C until OD600 = 0.15. Three milliliters of
the culture were added to 6-ml RNAprotect Bacteria Reagent
(Qiagen). Total RNA was extracted using the RNeasy Mini Kit
(Qiagen) according to the manufacturer’s recommendation.
The total RNA extract was subjected to gel electrophoresis for
quality control. Chromosomal DNA was removed using the
DNase Turbo DNA-free kit (Ambion) according to the manufacturer’s recommendation. A total of 500 ng of DNA-free
RNA was subjected to reverse transcription using the High
Capacity Reverse Transcription kit (Applied Biosystems) according to the manufacturer’s instruction, in a total reaction
volume of 50 ml. Relative cDNA levels were measured using
PerfeCTa SYBR Green SuperMix (Quanta Biosciences) in an
Eco Real-Time PCR System (Illumina). Each primer pair (supplementary table S2, Supplementary Material online) was
tested for efficiency in a set of six 10-fold dilutions. The
genes hcaT and cysG were used as reference genes in the
calculations for relative expression. All measurements were
repeated for three biological and three technical replicates in
the form of undiluted 1:10 and 1:100 dilutions of the template.
Outer Membrane Profiling
Overnight cultures were diluted 1:500 into 20-ml prewarmed
LB and incubated at 37 C until OD600 = 0.15. Two OD600
units (13.3 ml) were centrifuged for 15 min at 4,500 rpm.
The pellet was suspended in 1 ml, 100-mM Tris–HCl, pH
8.0, 20% sucrose, and incubated on ice for 10 min. After centrifugation (10 min at 5,000 rpm) the pellet was suspended in
1 ml, 100-mM Tris–HCl, pH 8.0, 20% sucrose, 10-mM EDTA.
Lysozyme was added to a final concentration of 100 mg/ml
and incubated on ice for 10 min. MgSO4, DNaseI, and RNaseA
were added to final concentrations of 20 mM, 5 mg/ml, and
5 mg/ml. The spheroplasts were disrupted by 7 freeze/thaw
cycles in dry ice/ethanol and a room temperature water bath.
The last thawing step was performed for 2 h on ice until the
sample was completely fluid. The membranes were pelleted
for 25 min at 15,000 rpm, washed in 500 ml, 20-mM NaPO4,
pH 7, pelleted again for 15 min at 15,000 rpm and resuspended in 1 ml, 20-mM NaPO4, pH 7, 0.5% sarcosyl. After
30-min incubation at room temperature, the pellets were
washed twice in 1 ml, 20-mM NaPO4, pH 7 and finally
taken up in 60-ml Laemmli sample buffer. The samples were
boiled 5 min at 95 C and subjected to sodium dodecyl sulfate
polyacrylamide gel electrophoresis (SDS-PAGE) (10% polyacrylamide/6 M urea). Protein bands were stained using colloidal coomassie and bands of interest were cut out for
subsequent mass spectrometry (MS) analysis.
In-Gel Digestion and Liquid Chromatography-MS
Cut-out bands were subjected to in-gel tryptic digestion as
described (Shevchenko et al. 2000). Briefly, the bands were
divided into smaller fragments and after destaining they were
subjected to reduction and alkylation using dithiothreitol and
iodoactamide before trypsin (sequence grade, modified
Promega, Madison, WI) was added at a concentration of
12.5 ng/ml. Digestion was performed over night at 37 C.
Peptides were extracted from the gel using a solution composed of 60% acetonitrile (ACN), 5% formic acid (FA), and
35% MilliQ water and dried using a SpeedVac. Peptides
were subsequently dissolved in 15 mL, 0.1% FA. The LC separation was carried out using a Thermo Easy-nano Liquid
Chromatography instrument (ThermoFisher Scientific,
Bremen, Germany). The sample injection volume was 5 mL
and the column was a 10 cm 75 mm, C18-A2 column (Easy
column; ThermoFisher Scientific) with a particle size of 3 mm
and an H2O:ACN:FA solvent system (H2O, 0.1% FA mobile
phase [A]; ACN, 0.1% FA mobile phase [B]) was used. Peptides
were eluted using a gradient of mobile phases A and B up to
80% B during 35 min. An LTQ-Orbitrap Velos Pro Mass spectrometer (ThermoFisher Scientific) was used to acquire mass
data. The MS was operating in positive ion mode and a survey
spectrum of high resolution was used to detect the peptide
mass. Thereafter consecutive collisionally induced dissociation spectra of the ten most abundant ions were recorded
in the ion trap. The acquired data (.RAW files) were processed
in the Proteome Discoverer 1.4 (Thermo Fisher Scientific)
software toward proteins from E. coli in the SwissProt database. Common search parameters included the following:
Maximum 10 ppm and 0.8 Da error tolerance for the
survey scan and MS/MS analysis, respectively; enzyme specificity was trypsin; maximum two missed cleavage sites allowed; cysteine carbamidomethylation was set as static
modification; oxidation (M) and deamidation (N,Q) were
set as variable modifications. The search criteria for protein
identification were set to at least two matching peptides of
95% confidence level per protein.
Complementation Assays
For growth complementation experiments, the wild-type and
selected mutants were transformed with plasmids isolated
from the ASKA library (Kitagawa et al. 2005) containing
phoE, chiP, ompG, ompN, ompW, or an empty control plasmid
pCA24N. After introduction of the plasmids into the strains,
continuous selection with chloramphenicol ensured maintenance of the plasmid. Induction of expression was achieved
by adding 15-mM IPTG. Higher amounts of IPTG had deleterious effects on growth rate.
3261
Knopp and Andersson . doi:10.1093/molbev/msv195
Supplementary Material
Supplementary tables S1–S5 and figures S1 and S2 are available at Molecular Biology and Evolution online (http://www.
mbe.oxfordjournals.org/).
Acknowledgments
This work was supported by grants from the Swedish
Research Council and European Commission (project
EvoTAR). The Science for Life Laboratory Mass
Spectrometry Based Proteomics Facility in Uppsala performed
the protein analysis and data storage was obtained and supported by Bioinformatics Infrastructure for Life Sciences.
References
Andersson DI, Hughes D. 2010. Antibiotic resistance and its cost: is it
possible to reverse resistance? Nat Rev Microbiol 8:260–271.
Andersson DI, Hughes D. 2011. Persistence of antibiotic resistance in
bacterial populations. FEMS Microbiol Rev. 35:901–911.
Andersson DI, Levin BR. 1999. The biological cost of antibiotic resistance.
Curr Opin Microbiol. 2:489–493.
Baba T, Ara T, Hasegawa M, Takai Y, Okumura Y, Baba M, Datsenko KA,
Tomita M, Wanner BL, Mori H. 2006. Construction of Escherichia
coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Mol Syst Biol. 2:2006.0008.
Baek JH, Lee SY. 2006. Novel gene members in the Pho regulon of
Escherichia coli. FEMS Microbiol Lett. 264:104–109.
Baker-Austin C, Wright MS, Stepanauskas R, McArthur JV. 2006. Coselection of antibiotic and metal resistance. Trends Microbiol.
14:176–182.
Basle A, Rummel G, Storici P, Rosenbusch JP, Schirmer T. 2006. Crystal
structure of osmoporin OmpC from E. coli at 2.0 A. J Mol Biol.
362:933–942.
Battesti A, Majdalani N, Gottesman S. 2011. The RpoS-mediated general
stress response in Escherichia coli. Annu Rev Microbiol. 65:189–213.
Bredin J, Saint N, Mallea M, De E, Molle G, Pages JM, Simonet V. 2002.
Alteration of pore properties of Escherichia coli OmpF induced
by mutation of key residues in anti-loop 3 region. Biochem J.
363:521–528.
Ceccarelli M, Danelon C, Laio A, Parrinello M. 2004. Microscopic
mechanism of antibiotics translocation through a porin. Biophys J.
87:58–64.
Charrel RN, Pages JM, De Micco P, Mallea M. 1996. Prevalence of
outer membrane porin alteration in beta-lactam-antibioticresistant Enterobacter aerogenes. Antimicrob Agents Chemother.
40:2854–2858.
Chatfield SN, Dorman CJ, Hayward C, Dougan G. 1991. Role of ompRdependent genes in Salmonella typhimurium virulence: mutants
deficient in both ompC and ompF are attenuated in vivo. Infect
Immun 59:449–452.
Cowan SW, Garavito RM, Jansonius JN, Jenkins JA, Karlsson R, Konig N,
Pai EF, Pauptit RA, Rizkallah PJ, Rosenbusch JP, et al. 1995. The
structure of OmpF porin in a tetragonal crystal form. Structure
3:1041–1050.
Cowan SW, Schirmer T, Rummel G, Steiert M, Ghosh R, Pauptit RA,
Jansonius JN, Rosenbusch JP. 1992. Crystal structures explain functional properties of two E. coli porins. Nature 358:727–733.
Criswell D, Tobiason VL, Lodmell JS, Samuels DS. 2006. Mutations conferring aminoglycoside and spectinomycin resistance in Borrelia
burgdorferi. Antimicrob Agents Chemother. 50:445–452.
Daigle F, Fairbrother JM, Harel J. 1995. Identification of a mutation in the
pst-phoU operon that reduces pathogenicity of an Escherichia coli
strain causing septicemia in pigs. Infect Immun. 63:4924-4927.
Danelon C, Nestorovich EM, Winterhalter M, Ceccarelli M, Bezrukov
SM. 2006. Interaction of zwitterionic penicillins with the OmpF
channel facilitates their translocation. Biophys J. 90:1617–1627.
3262
MBE
Datsenko KA, Wanner BL. 2000. One-step inactivation of chromosomal
genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci
U S A. 97:6640–6645.
Delcour AH. 2009. Outer membrane permeability and antibiotic resistance. Biochim Biophys Acta. 1794:808–816.
Enne VI, Livermore DM, Stephens P, Hall LM. 2001. Persistence of sulphonamide resistance in Escherichia coli in the UK despite national
prescribing restriction. Lancet 357:1325–1328.
Fajardo DA, Cheung J, Ito C, Sugawara E, Nikaido H, Misra R. 1998.
Biochemistry and regulation of a novel Escherichia coli K-12 porin
protein, OmpG, which produces unusually large channels.
J Bacteriol. 180:4452–4459.
Ferenci T. 2005. Maintaining a healthy SPANC balance through regulatory and mutational adaptation. Mol Microbiol. 57:1–8.
Fernandez L, Hancock RE. 2012. Adaptive and mutational resistance: role
of porins and efflux pumps in drug resistance. Clin Microbiol Rev.
25:661–681.
Figueroa-Bossi N, Valentini M, Malleret L, Fiorini F, Bossi L. 2009. Caught
at its own game: regulatory small RNA inactivated by an inducible
transcript mimicking its target. Genes Dev. 23:2004–2015.
Gehring KB, Nikaido H. 1989. Existence and purification of porin heterotrimers of Escherichia coli K12 OmpC, OmpF, and PhoE proteins.
J Biol Chem. 264:2810–2815.
Guillier M, Gottesman S, Storz G. 2006. Modulating the outer membrane with small RNAs. Genes Dev. 20:2338–2348.
Gullberg E, Albrecht LM, Karlsson C, Sandegren L, Andersson DI. 2014.
Selection of a multidrug resistance plasmid by sublethal levels of
antibiotics and heavy metals. MBio 5:e01918-01914.
Gullberg E, Cao S, Berg OG, Ilback C, Sandegren L, Hughes D, Andersson
DI. 2011. Selection of resistant bacteria at very low antibiotic concentrations. PLoS Pathog. 7:e1002158.
Hajjar E, Mahendran KR, Kumar A, Bessonov A, Petrescu M, Weingart H,
Ruggerone P, Winterhalter M, Ceccarelli M. 2010. Bridging
timescales and length scales: from macroscopic flux to the molecular mechanism of antibiotic diffusion through porins. Biophys J.
98:569–575.
Hall MN, Silhavy TJ. 1981. Genetic analysis of the major outer membrane
proteins of Escherichia coli. Annu Rev Genet. 15:91–142.
Han JS, Park JY, Lee YS, Thony B, Hwang DS. 1999. PhoBdependent transcriptional activation of the iciA gene during
starvation for phosphate in Escherichia coli. Mol Gen Genet.
262:448–452.
Harel J, Forget C, Ngeleka M, Jacques M, Fairbrother JM. 1992. Isolation
and characterization of adhesin-defective TnphoA mutants of septicaemic porcine Escherichia coli of serotype O115:K-:F165. J Gen
Microbiol. 138:2337–2345.
Hsieh YJ, Wanner BL. 2010. Global regulation by the seven-component
Pi signaling system. Curr Opin Microbiol. 13:198–203.
Kaczmarek FM, Dib-Hajj F, Shang W, Gootz TD. 2006. High-level carbapenem resistance in a Klebsiella pneumoniae clinical isolate is due to
the combination of bla(ACT-1) beta-lactamase production, porin
OmpK35/36 insertional inactivation, and down-regulation of the
phosphate transport porin phoe. Antimicrob Agents Chemother.
50:3396–3406.
Kitagawa M, Ara T, Arifuzzaman M, Ioka-Nakamichi T, Inamoto E,
Toyonaga H, Mori H. 2005. Complete set of ORF clones of
Escherichia coli ASKA library (a complete set of E. coli K-12 ORF
archive): unique resources for biological research. DNA Res.
12:291–299.
Kocan M, Schaffer S, Ishige T, Sorger-Herrmann U, Wendisch VF, Bott M.
2006. Two-component systems of Corynebacterium glutamicum:
deletion analysis and involvement of the PhoS-PhoR system in
the phosphate starvation response. J Bacteriol. 188:724–732.
Lamarche MG, Dozois CM, Daigle F, Caza M, Curtiss R 3rd, Dubreuil JD,
Harel J. 2005. Inactivation of the pst system reduces the virulence of
an avian pathogenic Escherichia coli O78 strain. Infect Immun.
73:4138–4145.
Lamarche MG, Wanner BL, Crepin S, Harel J. 2008. The phosphate
regulon and bacterial virulence: a regulatory network connecting
Fitness Costs of Antibiotic Resistance . doi:10.1093/molbev/msv195
phosphate homeostasis and pathogenesis. FEMS Microbiol Rev.
32:461–473.
Lenski RE. 1998. Bacterial evolution and the cost of antibiotic resistance.
Int Microbiol. 1:265–270.
Levin BR, Perrot V, Walker N. 2000. Compensatory mutations, antibiotic
resistance and the population genetics of adaptive evolution in
bacteria. Genetics 154:985–997.
Maclean RC, Hall AR, Perron GG, Buckling A. 2010. The evolution of
antibiotic resistance: insight into the roles of molecular mechanisms
of resistance and treatment context. Discov Med. 10:112–118.
Mandin P, Gottesman S. 2009. A genetic approach for finding small
RNAs regulators of genes of interest identifies RybC as regulating the
DpiA/DpiB two-component system. Mol Microbiol. 72:551–565.
Mantis NJ, Winans SC. 1993. The chromosomal response regulatory
gene chvI of Agrobacterium tumefaciens complements an
Escherichia coli phoB mutation and is required for virulence.
J Bacteriol. 175:6626–6636.
Misra R, Benson SA. 1989. A novel mutation, cog, which results in
production of a new porin protein (OmpG) of Escherichia coli
K-12. J Bacteriol. 171:4105–4111.
Mizuno T, Mizushima S. 1987. Isolation and characterization of deletion
mutants of ompR and envZ, regulatory genes for expression of the
outer membrane proteins OmpC and OmpF in Escherichia coli.
J Biochem. 101:387–396.
Ngeleka M, Harel J, Jacques M, Fairbrother JM. 1992. Characterization of
a polysaccharide capsular antigen of septicemic Escherichia coli
O115:K “V165” :F165 and evaluation of its role in pathogenicity.
Infect Immun. 60:5048–5056.
Nikaido H, Rosenberg EY. 1983. Porin channels in Escherichia coli: studies
with liposomes reconstituted from purified proteins. J Bacteriol.
153:241–252.
Overgaard M, Johansen J, Moller-Jensen J, Valentin-Hansen P. 2009.
Switching off small RNA regulation with trap-mRNA. Mol
Microbiol. 73:790–800.
Panja S, Schu DJ, Woodson SA. 2013. Conserved arginines on the rim of
Hfq catalyze base pair formation and exchange. Nucleic Acids Res.
41:7536–7546.
Paulander W, Maisnier-Patin S, Andersson DI. 2009. The fitness cost of
streptomycin resistance depends on rpsL mutation, carbon source
and RpoS (sigmaS). Genetics 183:539–546, 531SI–532SI.
Plumbridge J, Bossi L, Oberto J, Wade JT, Figueroa-Bossi N. 2014.
Interplay of transcriptional and small RNA-dependent control
mechanisms regulates chitosugar uptake in Escherichia coli and
Salmonella. Mol Microbiol. 92:648–658.
Prilipov A, Phale PS, Koebnik R, Widmer C, Rosenbusch JP. 1998.
Identification and characterization of two quiescent porin genes,
nmpC and ompN, in Escherichia coli BE. J Bacteriol. 180:3388–3392.
Ramadhan AA, Hegedus E. 2005. Survivability of vancomycin resistant
enterococci and fitness cost of vancomycin resistance acquisition.
J Clin Pathol. 58:744–746.
Rasmussen AA, Johansen J, Nielsen JS, Overgaard M, Kallipolitis B,
Valentin-Hansen P. 2009. A conserved small RNA promotes silencing of the outer membrane protein YbfM. Mol Microbiol.
72:566–577.
Ruiz N, Silhavy TJ. 2003. Constitutive activation of the Escherichia coli
Pho regulon upregulates rpoS translation in an Hfq-dependent fashion. J Bacteriol. 185:5984–5992.
MBE
Sarma V, Reeves P. 1977. Genetic locus (ompB) affecting a major outermembrane protein in Escherichia coli K-12. J Bacteriol. 132:23–27.
Schulz zur Wiesch P, Engelstadter J, Bonhoeffer S. 2010. Compensation
of fitness costs and reversibility of antibiotic resistance mutations.
Antimicrob Agents Chemother. 54:2085–2095.
Shevchenko A, Chernushevich I, Wilm M, Mann M. 2000. De Novo
peptide sequencing by nanoelectrospray tandem mass spectrometry using triple quadrupole and quadrupole/time-of-flight instruments. Methods Mol Biol. 146:1–16.
Shi W, Li K, Ji Y, Jiang Q, Wang Y, Shi M, Mi Z. 2013. Carbapenem and
cefoxitin resistance of Klebsiella pneumoniae strains associated with
porin OmpK36 loss and DHA-1 beta-lactamase production. Braz J
Microbiol. 44:435–442.
Simonet V, Mallea M, Pages JM. 2000. Substitutions in the eyelet region
disrupt cefepime diffusion through the Escherichia coli OmpF channel. Antimicrob Agents Chemother. 44:311–315.
Spira B, de Almeida Toledo R, Maharjan RP, Ferenci T. 2011. The uncertain consequences of transferring bacterial strains between laboratories—rpoS instability as an example. BMC Microbiol. 11:248.
Subbarao GV, van den Berg B. 2006. Crystal structure of the monomeric
porin OmpG. J Mol Biol. 360:750–759.
Suziedeliene E, Suziedelis K, Garbenciute V, Normark S. 1999. The acidinducible asr gene in Escherichia coli: transcriptional control by the
PhoBR operon. J Bacteriol. 181:2084–2093.
Taschner NP, Yagil E, Spira B. 2004. A differential effect of sigmaS on the
expression of the PHO regulon genes of Escherichia coli. Microbiology
150:2985–2992.
Tsai YL, Wang MC, Hsueh PR, Liu MC, Hu RM, Wu YJ, Liaw SJ. 2015.
Overexpression of an outer membrane protein associated with decreased susceptibility to carbapenems in Proteus mirabilis. PLoS One
10:e0120395.
Valentin-Hansen P, Johansen J, Rasmussen AA. 2007. Small RNAs controlling outer membrane porins. Curr Opin Microbiol. 10:152–155.
VanBogelen RA, Olson ER, Wanner BL, Neidhardt FC. 1996. Global analysis of proteins synthesized during phosphorus restriction in
Escherichia coli. J Bacteriol. 178:4344–4366.
Vershinina OA, Znamenskaia LV. 2002. The Pho regulons of bacteria.
Mikrobiologiia 71:581–595.
Vogel J, Papenfort K. 2006. Small non-coding RNAs and the bacterial
outer membrane. Curr Opin Microbiol. 9:605–611.
Wang YF, Dutzler R, Rizkallah PJ, Rosenbusch JP, Schirmer T. 1997.
Channel specificity: structural basis for sugar discrimination and
differential flux rates in maltoporin. J Mol Biol. 272:56–63.
Yamada J, Yamasaki S, Hirakawa H, Hayashi-Nishino M, Yamaguchi A,
Nishino K. 2010. Impact of the RNA chaperone Hfq on multidrug
resistance in Escherichia coli. J Antimicrob Chemother. 65:853–858.
Yates CM, Shaw DJ, Roe AJ, Woolhouse ME, Amyes SG. 2006.
Enhancement of bacterial competitive fitness by apramycin resistance plasmids from non-pathogenic Escherichia coli. Biol Lett.
2:463–465.
Zhang A, Schu DJ, Tjaden BC, Storz G, Gottesman S. 2013. Mutations in
interaction surfaces differentially impact E. coli Hfq association with
small RNAs and their mRNA targets. J Mol Biol. 425:3678–3697.
Zhang A, Wassarman KM, Rosenow C, Tjaden BC, Storz G, Gottesman S.
2003. Global analysis of small RNA and mRNA targets of Hfq. Mol
Microbiol. 50:1111–1124.
Ziervogel BK, Roux B. 2013. The binding of antibiotics in OmpF porin.
Structure 21:76–87.
3263