Molecular Microbiology (2000) 37(1), 67±80
Temperature-regulated efflux pump/potassium
antiporter system mediates resistance to cationic
antimicrobial peptides in Yersinia
Jose Antonio Bengoechea and Mikael Skurnik*
Department of Medical Biochemistry, Institute of
Biomedicine, University of Turku, Kiinamyllynkatu 10,
FIN-20520 Turku, Finland.
Summary
Most bacterial pathogens are resistant to cationic
antimicrobial peptides (CAMPs) that are key components of the innate immunity of both vertebrates and
invertebrates. In Gram-negative bacteria, the known
CAMPs resistance mechanisms involve outer membrane (OM) modifications and specifically those in
the lipopolysaccharide (LPS) molecule. Here we
report, the characterization of a novel CAMPs resistance mechanism present in Yersinia that is dependent on an efflux pump/potassium antiporter system
formed by the RosA and RosB proteins. The RosA/
RosB system is activated by a temperature shift to
378C, but is also induced by the presence of the
CAMPs, such as polymyxin B. This is the first report
of a CAMPs resistance system that is induced by the
presence of CAMPs. It is proposed that the RosA/
RosB system protects the bacteria by both acidifying
the cytoplasm to prevent the CAMPs action and
pumping the CAMPs out of the cell.
Introduction
Cationic antimicrobial peptides (CAMPs) play an important role in the defence of plants, insects and vertebrates
against microorganisms (Boman, 1995; Nicolas and Mor,
1995). CAMPs are structurally diverse molecules that
share two features: all of them have a net positive charge
and an amphipathic character (Hancock, 1984; Vaara,
1992; Nicolas and Mor, 1995). In vertebrates, they
represent both a constitutive and inducible antimicrobial
component of the innate immune response (Boman,
1995; Nicolas and Mor, 1995). In mammals, CAMPs are
present within the granules of neutrophils, in platelets,
macrophages, mucosal or skin secretions, in the gut,
tongue, trachea, or can be obtained as the degradation
Accepted 5 April, 2000. *For correspondence. E-mail mikael.skurnik@
utu.fi; Tel. (135) 82 333 7441; Fax (135) 82 333 7229.
Q 2000 Blackwell Science Ltd
products of proteins (Boman, 1995). Antimicrobial activity
against Gram-negative bacteria is initiated by the interaction of CAMPs with the negatively charged groups at
the lipopolysaccharide (LPS) core and lipid A level, which
causes disruption of the outer membrane (OM), and the
CAMPs gain access to the cytoplasmic membrane (the so
called self-promoted pathway) (Hancock, 1984; Vaara,
1992; Nicolas and Mor, 1995). The mode of antimicrobial
action involves the formation of pores in, or solubilization
of, the cytoplasmic membrane (Hancock and Chapple,
1999). However, the cytoplasmic membrane may not be
the final target of these peptides. Because of their wide
antimicrobial activity spectrum, they are under extensive
studies for the development of new antimicrobial agents.
Indeed, preclinical, phase I and phase III trials are under
way with the most promising candidates (Hancock and
Lehrer, 1998). There is increasing evidence showing that
resistance to CAMPs is characteristic to bacterial pathogens and the resistance mechanisms are best studied in
the intracellular pathogens Salmonella and Brucella. In
both species, resistance has been attributed to the
structural peculiarities of the LPS molecules that prevent
the initial interaction of the CAMPs; mutations in CAMPs
sensitive strains were mapped to genes that modify the
structure of LPS, and specifically that of the lipid A part
(MartõÂnez de Tejada and MoriyoÂn, 1995; Freer et al., 1996;
Gunn et al., 1998; Guo et al., 1997; 1998; Sola-Landa et al.,
1998).
The genus Yersinia contains 11 species of which only
Y. pestis, Y. pseudotuberculosis and some serotypes of
Y. enterocolitica are human pathogens. Y. enterocolitica
is a common human enteropathogen that causes gastrointestinal syndromes of different severity (Bottone, 1997).
Y. enterocolitica is ingested as a contaminant of food or
water and it invades the tissues in the terminal ileum
(Hanski et al., 1989; GruÈtzkau et al., 1990). It is probable
that Y. enterocolitica will encounter CAMPs during the
infection process, and thus it is reasonable to assume that
the bacteria have developed resistance to them. In this
regard, we have shown that pathogenic and non-pathogenic
Y. enterocolitica strains grown at room temperature (RT)
are more resistant to the action of CAMPs than
enteropathogenic Escherichia coli (Bengoechea et al.,
1996), whereas when grown at 378C only the pathogenic
strains remain comparatively resistant to these peptides
68 J. A. Bengoechea and M. Skurnik
and the rosAB mutants are indicated with ±RosA2B2,
±RosA2 or ±RosB2 suffixes.
(Bengoechea et al., 1996). These data led us to
hypothesize that CAMPs resistance could be important
for the pathogenicity of Y. enterocolitica and more
generally for other extracellular pathogens.
We recently identified in Y. enterocolitica serotype O:8
two genes, named rosA and rosB, that affected temperature regulation of the O-antigen biosynthesis (Zhang,
1996). Interestingly, RosA contains motifs identified in a
large group of transport proteins belonging to the major
facilitator superfamily (Paulsen et al., 1996). RosA is
similar to members of cluster I of this family, i.e. to
proteins involved in drug resistance such as Fsr (fosmidomycin resistance protein) of E. coli (Fujisaki et al.,
1996), Bmr (multidrug-efflux transporter) of Bacillus
subtilis (Neyfakh et al., 1991) and Bcr (bicyclomycin
resistance protein) of E. coli (Bentley et al., 1993). All
these proteins pump drugs out of the cells, using the
proton-motive force as the energy supply. RosB shows
similarity to a number of proteins involved in the
glutathione-regulated potassium efflux systems such
KefB and KefC of E. coli (Munro et al., 1990) and KefX
of Haemophilus, and to the sodium antiporter (Na1/H1)
NaH of Lactococcus lactis and NapA of Enterococcus
hirae (Waser et al., 1992).
These similarities suggested that RosA and RosB
might also have a role that is not directly involved in the
O-antigen regulation. In this report, we present data
showing that the rosAB locus encodes a temperatureregulated efflux pump that is coupled to a potassium
antiporter and that the RosA/RosB system mediates
resistance to cationic antimicrobial peptides in Yersinia.
Characterization of the rosAB mutant
We tested whether the rosAB mutant could display a
phenotype consistent (i) with the putative role of RosA as
a drug efflux pump protein, i.e. sensitivity to some drugs,
and/or (ii) with the putative role of RosB as a cation
exchanger, i.e. sensitivity to sodium or potassium.
Sensitivity of the rosAB mutant to drugs and CAMPs First,
the sensitivities of YeO8c±RosA2B2 and of the parental
strain YeO8c were tested with a panel of different
antibiotics and chemicals by the disc diffusion method,
and the results are shown in Table 1. When the strains
were grown at 378C, YeO8c±RosA2B2 was more
sensitive to the antibiotics novobiocin, rifampicin and
tetracycline and to the disinfectant benzalkonium chloride
(BK) than YeO8c. No differences were found in the
sensitivity of the strains to crystal violet, ethidium bromide
(EB) and cetyltrimethylammonium bromide (CTAB).
When the strains were grown at RT, no differences in
the sensitivity to any tested agent were found (data not
shown). We also tested the sodium deoxycholate and
ethanol sensitivities but no differences were found
between the strains at both growth temperatures (data
not shown). The increased sensitivity of YeO8c±
RosA2B2 to a restricted set of drugs was in agreement
with the putative role of RosA as a drug efflux pump.
These drugs only have in common their amphipathic
character although this was not a general feature as no
differences were found in the sensitivities to other
amphipathic compounds like crystal violet, CTAB and
EB. These results prompted us to test other compounds
with amphipathic characteristics, more likely to be
encountered by Yersinia during the infection process.
The best candidates fulfilling these characteristics are
CAMPs (Hancock, 1984; Vaara, 1992; Hancock and
Chapple, 1999). To this end, we tested the sensitivity of
YeO8c±RosA2B2 to polymyxin B, taken as a model of
Results
Nomenclature of the Y. enterocolitica serotype O:8 strains
Restriction-negative derivatives of Y. enterocolitica serotype O:8 wild-type strain 8081 were used in this study and
for simplicity of presentation the virulence plasmid positive
strain is called YeO8, its plasmid cured derivative, YeO8c,
Table 1. Susceptibilities of Yersinia strains to antibiotics and chemicals. Bacteria were grown at 378C in TSB. Results are expressed in units (1 unit
equal to 10 mm).
Strains
Agent (mg disc21)
YeO8c
YeO8c±RosA2B2
YeO8c±RosA2B2/pLZ4k
YeO8c±RosA2B2/pLZ4kKpn
Novobiocin (50 mg)
Rifampicin (10 mg)
Tetracycline (10 mg)
Crystal violet (40 mg)
CTAB (100 mg)
BK (100 mg)
EB (20 mg)
15
55
62
45
0
80
0
80
70
90
40
0
100
0
0
60
65
42
0
78
0
0
60
65
48
0
82
0
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
Antimicrobial peptide resistance in Yersinia
CAMPs action (Vaara, 1992). The sensitivity was determined by an indirect assay in which bacterial suspensions
are incubated with polymyxin B in the presence of
lysozyme (a lytic enzyme acting in the periplasm).
Penetration of the probe (lysozyme) causes cell lysis,
which can be measured as a turbidity drop. Figure 1
shows that YeO8c±RosA2B2 was more sensitive to the
action of polymyxin B than YeO8c. Again this difference
manifested only when the strains were grown at 378C but
not at RT (compare Fig. 1A and B). To find out whether
this CAMPs sensitivity was also applicable to other
CAMPs that are not related structurally to polymyxin B
and that have less bactericidal activity on Yersinia (Vaara,
1992; Skurnik et al., 1999), the sensitivities of the strains
to cecropin P1 (from pig intestine) and melittin (from bee
venom) were tested using a radial diffusion assay,
keeping polymyxin B as a control. The minimal inhibitory
concentrations (MICs) of YeO8c grown at 378C to
polymyxin B, cecropin P1 and melittin were 3, 24 and
9 mg ml21, while those of YeO8c±RosA2B2 were 1, 4
and 3 mg ml21 respectively. Thus, YeO8c±RosA2B2 was
3±6 times more sensitive to these CAMPs than YeO8c.
One likely explanation for the increased sensitivity could
be that the OM architecture of the mutant was somehow
affected allowing an easier interaction of the CAMPs with
the bacterial surface. To examine this possibility, we
measured the amounts of polymyxin B bound by live
bacteria grown either at RT or 378C. No significant
differences were found between the amount of polymyxin
B adsorbed by YeO8c±RosA2B2 and YeO8c whether
grown at RT or 378C (data not shown). Noteworthy, the
amount of polymyxin B bound by the bacteria of both
strains grown at RT was significantly less than the amount
bound by the bacteria grown at 378C.
Fig. 1. The role of RosAB in polymyxin B sensitivity determined as
lysozyme-induced cell lysis.
A. Bacteria were grown at 378C in TSB.
B. Bacteria were grown at RT in TSB. Each point represents the
mean and standard deviation (covered by the symbol) of six
samples from two independently grown batches of bacteria.
Symbols: X, YeO8c; P, YeO8c±RosA2B2 ±rev; W, YeO8c±
RosA2B2; L, YeO8c±RosA2B2/pLZ4k.
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
69
Sensitivity of the rosAB mutant to sodium or potassium.
The growth properties of the two strains in minimal media
(MM) with different cation availability at two growth
temperatures (RT and 378C) were analysed. When a
medium designed to study the transport of potassium was
used (Epstein and Kim, 1971), we did not find any
significant growth differences between the strains grown
at 378C, irrespective of the potassium concentration (no
KCl [K0] or 115 mM KCl [K115]) used (Fig. 2A). Similar
results were obtained with bacteria grown at RT although
both strains reached higher final optical density (OD)
values than those obtained at 378C (data not shown).
However, when the growth in the M63 MM was tested
differences between the strains became apparent. At
378C in the absence of sodium, the growth of YeO8c±
RosA2B2 was significantly impaired when compared with
the YeO8c and this effect was more pronounced at pH 8
than at pH 7. The growth of YeO8c±RosA2B2 was
Fig. 2. Growth of Yersinia strains in MM with different cation
availability and pH.
A. YeO8c (open bar) and YeO8c±RosA2B2 (dashed bar) bacteria
were grown overnight at RT in TSB, and diluted 50-fold into: (i)
M63 medium adjusted to pH 7 or 8 with or without NaCl; and (ii)
potassium transport medium supplemented with 115 mM KCl (K115)
or 0 mM KCl (K0). The OD540 values of the cultures after 17 h
growing in an orbital shaker (250 r.p.m.) at 378C are shown.
B. Cells were grown overnight at RT in TSB and diluted fivefold into
TM (pH 7). After 6 h at RT, the culture was diluted 50-fold into TM
with different pH supplemented with 300 mM KCl. The OD540
values of the cultures after 17 h growing in an orbital shaker
(250 r.p.m.) at 378C are shown.
C. Cells were grown as described in (B), but they were diluted 50fold into TM with different pH. Symbols: X, YeO8c; P, YeO8c±
RosA2B2 ±rev; W, YeO8c±RosA2B2; L, YeO8c±RosA2B2/pLZ4k;
B YeO8c±RosA2B2/pLZ4kKpn.
70 J. A. Bengoechea and M. Skurnik
rescued by the addition of sodium (300 mM NaCl) to the
medium (Fig. 2A). Interestingly, these differences were
not seen with bacteria grown at RT. The amount of
potassium in M63 is relatively high (over 100 mM)
suggesting that the growth defect of YeO8c±RosA2B2
in this medium might be due to a lethal effect of potassium
and that this effect could be suppressed by sodium. The
apparent contradictory result of the ability of YeO8c±
RosA2B2 to grow in the presence of 115 mM potassium
(K115) in the potassium medium could be due to the
presence of enough sodium (<115 mM) in the medium.
To confirm these roles of potassium and pH, YeO8c±
RosA2B2 was grown in Tris medium (TM) with or without
300 mM KCl and with different pH (Fig. 2B and C). In
good agreement with the above results, YeO8c±
RosA2B2 could not propagate to the same extent as
YeO8c in TM supplemented with 300 mM KCl (Fig. 2B),
especially at basic pH. When the effect of pH was tested
alone (Fig. 2C), at higher pH YeO8c±RosA2B2 grew
much less than YeO8c although the OD values were
higher than in the presence of potassium in the growth
medium. When the growth experiments were repeated
with TM supplemented with 300 mM NaCl no growth
differences between the strains were seen in good
agreement with the proposed role of sodium (data not
shown). The lethal effect of both alkaline pH and
potassium on YeO8c±rosA2B2 was also measured
directly by determining the viability of YeO8c±rosA2B2
in TM adjusted to pH 7.5 or pH 8 in the presence or
absence of KCl. The number of viable YeO8c±rosA2B2
cells recovered after 1 and 2 h incubation in TM adjusted
to pH 7.5 or pH 8 was significantly lower than those of
YeO8c (at pH 8 after 2 h 60% of YeO8c±rosA2B2 and
105% of YeO8c were viable). When the same experiment
was repeated in the presence of 300 mM KCl, the number
of YeO8c±rosA2B2 viable cells was even lower than in
the absence of KCl (40% and 100% respectively). No
growth differences were found when the strains were
grown in Luria broth (LB) or tryptic soya broth (TSB) (data
not shown).
In summary, we can conclude that at 378C but not at RT
(i) YeO8c±RosA2B2 was more sensitive to CAMPs and
certain antibiotics than the wild-type strain; (ii) YeO8c±
RosA2B2 was not able to survive in a medium containing
potassium as the main cation; and (iii) it was impaired in
its ability to adapt to alkaline pH in the presence of
potassium.
Complementation studies
To confirm that the above phenotypes of YeO8c±
RosA2B2 were caused by the kanamycin cassette in
the ros locus and not by a spontaneous mutation
elsewhere, we trans complemented YeO8c±RosA2B2
using plasmids carrying the wild-type genes. pLZ6k
contains a 6 kb BamHI fragment with the gsk, rosB,
rosA genes and part of the ushA gene cloned into
pTM100, and pLZ4k contains an internal 4.4 kb EcoRI
fragment of pLZ6k cloned into a low-copy vector
pMMB207. Only the rosA and rosB genes are complete
in pLZ4k and can be induced from the tac promoter of the
vector.
pLZ4k complemented in trans the antibiotic sensitivity
of YeO8c±RosA2B2 (Table 1). To clarify whether the
complementation was due to the expression of RosA or
RosB, a KpnI fragment of pLZ4k, including most of the
rosB gene, was deleted to generate pLZ4kKpn. Because
pLZ4kKpn was able to complement YeO8c±RosA2B2
(Table 1), this suggested that RosA was responsible for
the antibiotic resistance. Interestingly, neither pLZ4k
(Fig. 1A) nor pLZ6k (not shown) complemented the
polymyxin B sensitivity.
The ability of these constructs to complement the other
phenotype of the YeO8c±RosA2B2 strain, sensitivity to
potassium at basic pH, was then analysed. pLZ4k was
able to suppress the lethal effect of potassium to some
extent, but YeO8c±RosA2B2/pLZ4k did not grow as well
as YeO8c (Fig. 2B). As expected, pLZ4kKpn (missing the
rosB gene) showed no complementation of the potassium
sensitivity of YeO8c±RosA2B2 (Fig. 2B).
Even though pLZ4k is based on a low-copy vector, we
considered that overexpression of the rosA and rosB
genes was possible. To circumvent the copy number
effects, the mutation in YeO8c±RosA2B2 was corrected
by reverse allelic exchange as described in Experimental
procedures. Indeed, the revertant strain was fully complemented with regard to both phenotypes: the potassium
and polymyxin B sensitivity (Figs 1 and 2).
In conclusion, the complementation results confirmed
that all the phenotypes described above were due to the
inactivation of the rosAB locus, and that, for proper
function, the rosA and rosB expression must be maintained at very delicate balance.
Both RosA and RosB are involved in the resistance to
CAMPs
Because RosA alone could complement the antibiotic
sensitivity phenotype, we then asked whether both RosA
and RosB were necessary for the resistance to CAMPs.
To address this question, non-polar mutations were
engineered into the rosA and rosB genes of YeO8c (see
Experimental procedures) and the mutants were tested in
the polymyxin B sensitivity assay. The results indicated
that both YeO8c±RosA2 and YeO8c±RosB2 were more
sensitive to polymyxin B than YeO8c (Fig. 3A). No
significant differences in sensitivities were seen either
between them or between them and YeO8c±RosA2B2.
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
Antimicrobial peptide resistance in Yersinia
71
Interestingly, when IPTG was used to induce the ros
locus expression in C600/pLZ4k the strain lost the CAMPs
resistance (data not shown). Thus, also in E. coli overexpression of rosAB resulted in functional problems similar
to the situation in the Yersinia genetic background (see
above). This indicated that the expression of rosA and rosB
should be under tight regulation.
Functions of RosA and RosB in the resistance to CAMPs
Fig. 3. The role of RosA and RosB in polymyxin B sensitivity in
Yersinia and E. coli.
A. Survival of bacteria (% of the colony counts of cells not exposed
to polymyxin B) after 30 min incubation with different amounts of
polymyxin B. Each point represents the mean and standard
deviation (covered by the symbol) of four samples from two
independently grown batches of bacteria. Symbols: X, YeO8c; P,
YeO8c±RosA2B2 ±rev; W, YeO8c±RosA2B2; A, YeO8c±RosA2;
K, YeO8c±RosB2.
B. E. coli C600 (open bar), E. coli C600/pLZ4kKpn (grey bar), and
E. coli C600/pLZ4k (black bar) survival (% of the colony counts of
cells not exposed to polymyxin B) after 30 min incubation with
125 ng ml21 of polymyxin B. Error bars display standard deviation
from the mean for one experiment.
These data therefore suggested that both RosA and RosB
were involved in the resistance to CAMPs, but could not
reveal whether they had a direct or an indirect role.
We speculated that if the RosA/RosB system could
transfer CAMPs resistance to a heterologous host
originally highly sensitive to CAMPs this would provide
evidence that RosA/RosB have a direct role in the
resistance to CAMPs. To this end, the CAMPs resistance
of E. coli C600, C600/pLZ4k and C600/pLZ4kKpn was
assayed using the above assay but with a fixed amount of
polymyxin B (125 ng ml21). pLZ4k but not pLZ4kKpn
significantly enhanced the survival of E. coli C600
(Fig. 3B) indicating that both RosA and RosB were
needed to confer resistance to polymyxin B in E. coli
and, taking into account our hypothesis, suggested that
RosA and RosB play a direct role in the resistance to
polymyxin B (see below). However, our results do not
exclude the possibility that in E. coli RosA and RosB are
needed to activate the expression of other factors
involved in CAMPs resistance.
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
Having established that both RosA and RosB are involved
in the resistance to CAMPs, we wanted to determine how
they mediate this resistance. Above we showed that
YeO8c±RosA2B2 was not able to survive in a medium
containing potassium as the main cation, and that it was
impaired in its ability to adapt to alkaline pH in the
presence of potassium. Therefore, RosB could be
involved in regulating the intracellular pH based on its
potassium antiporter function (K1/H1). As RosB shows
similarity to KefB and KefC that are both glutathioneregulated potassium efflux proteins of E. coli, we
hypothesized that RosB's function could resemble that
of KefB and KefC. In an elegant series of studies,
Ferguson et al. (1995) showed that KefB and KefC are
involved in the release of potassium from the cells in the
presence of electrophiles. This potassium efflux is
accompanied by an influx of protons into the cell, resulting
in acidification of the cytoplasm and this protects E. coli
cells against methylglyoxal. Furthermore, artificial lowering of the intracellular pH with a weak acid results in
complete protection of a kefB/kefC double mutant against
methylglyoxal. Based on these considerations, we
hypothesized that treatment of YeO8c±RosB2 with a
weak acid could mimic RosB's function and result in
complementation of the CAMPs resistance defect. However, this treatment should not be sufficient to protect the
YeO8c±RosA2B2 cells against the CAMPs because the
resistance was shown to be dependent on a functional
RosA.
To carry these speculations even further, the role of
RosA in the resistance to antibiotics (see above) and the
similarity of RosA with drug efflux pumps suggested that
the function of RosA could be to force the CAMPs outside
the cytoplasmic membrane using the proton-motive force
as an energy supply. Thus, inhibition of the energy supply
could interfere with the function of RosA.
To assess experimentally the above speculations,
survival of YeO8c±RosB2 in 250 ng ml21 polymyxin B
in the presence or absence of a weak acid, sodium
acetate, was tested. In line with our hypothesis, the
survival of YeO8c±RosB2 was greatly enhanced by the
presence of 15 mM sodium acetate (Fig. 4). Interestingly,
5 or 10 mM sodium acetate did not enhance the survival
(data not shown). However, the increase in the survival
72 J. A. Bengoechea and M. Skurnik
Table 2. Luciferase activity exhibited by YeO8c±RosA2B2::pRV4kluc grown in different induction conditions.
Fig. 4. Effect of sodium acetate and CCCP on the survival of
YeO8c±RosA2B2 (open bars) and YeO8c±RosB2 (black bars) to
polymyxin B. Cell suspensions were prepared in 1% tryptone (w/v)PBS (pH 6.9) and incubated with 250 ng ml21 of polymyxin B at
378C in the presence or absence of sodium acetate and/or CCCP.
Results indicate the survival of bacteria (% of the colony counts of
bacteria not exposed to polymyxin B) after 30 min incubation.
Sodium acetate or CCCP alone or in combination did not affect the
viability of the strains.
was dependent on a functional RosA because the double
mutant (YeO8c±RosA2B2) showed only a small but
significant increase in the survival (P , 0.05, two-tailed
z-test). To interfere with the function of RosA, an
uncoupler, 2-carbonyl cyanide m-chlorophenylhydrazone
(CCCP) was used. This agent has been used extensively
in studies dealing with the proton-motive force-dependent
efflux of antibiotics by bacteria (Nikaido, 1996). Typically,
the addition of CCCP to strains possessing proton-motive
force-dependent efflux pumps enhances accumulation of
the antibacterial compounds that are normally removed by
the respective efflux system (Nikaido, 1996). Accordingly,
CCCP greatly reduced the survival of YeO8c±RosB2
(Fig. 4). Noteworthy, CCCP also reduced the survival of
YeO8c to polymyxin B (data not shown).
Regulation of the ros locus
The following results show that the ros locus played a role
only when the bacteria were grown at 378C, indicating that
the expression of the ros locus could be regulated by
temperature. Moreover, its role in the resistance to
CAMPs also prompted us to study whether the expression
was modulated by the presence of these antimicrobial
agents. To address these questions, we constructed a
reporter strain, YeO8c±RosA2B2::pRV4Kluc, in which a
promoterless lucFF gene was cloned under the control of
the promoter of the ros locus (see Experimental procedures). Luciferase activity was monitored in response to
different growth conditions and agents (Table 2). Luciferase activity in the reporter bacteria grown at 378C was 16fold higher than that of bacteria grown at RT showing that
Induction conditions
RLUa
RT
RT 1 125 ng ml21 Polymyxin B
378C
378C 1 125 ng ml21 Polymyxin B
378C 1 low calcium
378C 1 300 mM KCl
378C 1 pH 5.5
5321
14 240
78 301
101 661
80 875
70 839
19 895
^
^
^
^
^
^
^
482
2455b
5559b
5745b,c
7855b
8486b
1103b,c
a. Results (mean ^ standard deviation) expressed in relative
luminescence units (RLU).
b. Significantly different from the RT result (P , 0.05, two-tailed
z-test).
c. Significantly different from the 378C result (P , 0.05, two-tailed
z-test).
the ros locus is indeed temperature regulated. Polymyxin
B significantly increased the expression of the ros locus in
bacteria grown both at 378C and RT. Potassium, on the
other hand, did not potentiate the temperature induction,
although YeO8c±RosA2B2 was sensitive to it. Restricted
calcium availability, a signal known to modulate the
expression of some virulence factors in Yersinia (Straley
and Perry, 1995; Cornelis and Wolf-Watz, 1997) and
probably to be found in vivo, did not further increase the
temperature induction of the ros locus. Interestingly, when
the reporter strain was grown in acidic pH (pH 5.5) at
378C, the expression of the ros locus was downregulated.
This prompted us to study the sensitivities of YeO8c±
RosA2B2 and YeO8c to polymyxin B when grown under
these conditions. Both strains grown at acidic pH were
more resistant to polymyxin B than the strains grown at
neutral pH, but YeO8c±RosA2B2 was still more sensitive
than YeO8c (data not shown).
The presence of rosAB in genus Yersinia
Because different Yersinia species and biogroups vary in
their susceptibility to CAMPs (Bengoechea et al., 1996;
1998), we investigated whether sequences homologous
to rosAB were present in other yersiniae. Slot-blot
hybridization analysis using a 4.4 kb EcoRI fragment of
pLZ4k as a probe gave strong positive hybridization
signals with all Yersinia strains (not shown), including Y.
pestis EV76, Y. pseudotuberculosis strains of 10 serotypes (O:1a±O:5a) and Y. enterocolitica strains of 15
serotypes (O:1, O:1,2,3, O:3, O:4,32, O:5, O:5,27,
O:6 : 30, O:8, O:9, O:10, O:13,7, O:14, O:28,50, O:35,36,
O:41,43).
The above results indicated that rosAB homologues are
present in all yersiniae and we asked whether they also
mediate CAMPs resistance. A rosAB double mutant of
Y. enterocolitica serotype O:3 strain 6471/76-c was
constructed using the same suicide vector previously
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
Antimicrobial peptide resistance in Yersinia
2
2
used for the construction of the YeO8c±RosA B mutant
(Zhang, 1996). The mutant was named YeO3c±RosA2B2
and its polymyxin B sensitivity when grown at 378C was
studied. In good agreement with the proposed function
for rosAB, YeO3c±RosA2B2 was more sensitive to
polymyxin B than the parental strain (data not shown).
Discussion
The main novel finding described in this report is that
bacteria can use an efflux pump/potassium antiporter
system as a mechanism of CAMPs resistance. We
demonstrated that in Yersinia, RosA and RosB, which
exhibit sequence similarity to drug efflux pumps and
cation antiporter proteins, make up this system. As far as
we are aware, this is the first report of such a system
mediating resistance to CAMPs.
The most important mechanisms of CAMPs resistance
in Gram-negative bacteria known so far involve OM
modifications and specifically those in the LPS molecule
(Vaara, 1992). Salmonella typhimurium regulates
mechanisms of resistance to CAMPs through the two
component signalling systems phoP/phoQ and pmrA/
pmrB. The PhoP/PhoQ system activates the transcription
of genes involved in LPS modifications at the lipid A level.
These modifications include (i) addition of aminoarabinose to the phosphate groups at lipid A; (ii) replacement
of the myristate acyl group of the lipid A by 2-OHmyristate; and (iii) the formation of heptaacylated lipid A
by the addition of palmitate (Guo et al., 1997; 1998).
Transcription of the pmrA/pmrB genes is activated by
PhoP/PhoQ but also mild acid conditions can independently activate them (Gunn and Miller, 1996; Soncini and
Groisman, 1996). The products of PmrA/PmrB activated
genes modify the LPS core and lipid A regions with
ethanolamine and add aminoarabinose to the 4 0 phosphate
of lipid A (Gunn et al., 1998). Interestingly, substitution of the
4 0 lipid A phosphate with aminoarabinose has also been
shown in Proteus mirabilis, Burkholderia (Pseudomonas)
cepacia and Chromobacterium violaceum, bacteria that
are resistant to polymyxin B (Hase and Reitschel, 1977;
Sidorczyk et al., 1983; Cox and Wilkinson, 1991; Vaara,
1992), and also in E. coli, Klebsiella pneumoniae and
Salmonella mutants resistant to polymyxin B (Helander
et al., 1994; 1996; Nummila et al., 1995). Noteworthy, an
increased heptaacylated lipid A has been reported in the
Klebsiella mutants (Helander et al., 1996). Recently, we
described another mechanism of resistance that does
not involve changes in the lipid A section of the LPS.
Y. enterocolitica O:3 mutants lacking the LPS outer core
hexasaccharide but expressing the O-antigen were more
susceptible than the wild-type bacteria to polymyxin B,
melittin, poly L-lysine and poly L-ornithine. We also
showed that these mutants bound more CAMPs than
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
73
the wild-type strain (Skurnik et al., 1999). Because the
outer core does not contain amino sugars, a change in the
overall charge of LPS is not likely. Therefore, we
postulated, based on the outer core size and its position,
that the outer core sterically hinders the access of CAMPs
to bind to the deeper parts of LPS (Skurnik et al., 1999).
With this LPS structure modification background, it was
somewhat unexpected to find out that a putative efflux
pump, RosA, was involved in the resistance to CAMPs in
Y. enterocolitica. The following evidence supports the
conclusion that RosA itself functions as a CAMPs efflux
pump: (i) RosA shows similarity to proteins that experimentally have been demonstrated to act as drug efflux
pumps; (ii) similar to efflux pump mutants the rosAB
mutant had increased sensitivity to some chemicals and
antibiotics, and the sensitive phenotype could be complemented with a plasmid containing the rosA gene alone;
(iii) both YeO8c±RosA2B2 and YeO8c±RosA2 showed
increased sensitivity to CAMPs; and (iv) YeO8c±RosB2
showed increased sensitivity to CAMPs when the proton
conductor CCCP was used to poison the cells. It is well
established that addition of CCCP to strains possessing
proton-motive force-dependent efflux pumps for antibacterial agents enhance accumulation of the agents that are
normally removed by the respective efflux system
(Nikaido, 1996). Therefore, if the CAMPs resistance of
YeO8c was dependent on the efflux pumps CCCP should
increase the sensitivity of bacteria to CAMPs, and this
was indeed the case.
The involvement of efflux pumps in the bacterial
resistance to CAMPs was, in fact, already predicted by
Nikaido (1996), who stated that these pumps are
important for bacteria in their attempts to remove toxic
compounds that interact with the cytoplasmic membrane.
Along this line, there are two recently described examples
of efflux pump-mediated resistance to CAMPs: (i) Shafer
et al. (1998) demonstrated that the susceptibility of
Neisseria gonorrhoeae to protegrins and to a-helical
human CAMPs is modulated by an energy-dependent
efflux system named mtr; and (ii) the qacA encoded
multidrug pump of Staphylococcus aureus was shown to
be involved in the resistance to the platelet microbicidal
protein, which is a small cationic peptide released from
rabbit platelets (Kupferwasser et al., 1999).
In contrast to the above two cases, in Y. enterocolitica
the efflux function of RosA alone was not sufficient to
counteract the antibacterial action of the CAMPs. The
CAMPs sensitivity of the non-polar rosB mutant indicated
that the function of RosB was also needed. Supporting
this conclusion, the E. coli C600 resistance to CAMPs
was increased only if both RosA and RosB were present.
What is the role of RosB in this scenario? We showed that
YeO8c±RosA2B2 did not grow in a medium containing
potassium as the main cation (Fig. 2A) and that it was
74 J. A. Bengoechea and M. Skurnik
Fig. 5. A Model of the mechanisms of
CAMPs resistance in Y. enterocolitica. The
left side illustrates the situation when
bacteria are grown at RT (relatively
resistant to CAMPs). The right side
illustrates the situation when bacteria are
grown at 378C (relatively sensitive to
CAMPs). The thickness of the arrows
above the genes indicates the level of gene
expression. Arrowed lines indicate
regulatory effects on gene expression
shown by circled plus or minus signs. A
dashed line indicates a hypothetical
regulatory effect on the gene expression
whereas a solid line a proven regulatory
effect. OM, outer membrane; CM,
cytoplasmic membrane.
impaired in its ability to adapt to alkaline pH in the
presence of potassium (Fig. 2B). This indicated that RosB
is involved in regulating the intracellular pH using its
potassium antiporter function (K1/H1) and that this
function may be linked to the CAMPs resistance (see
also Fig. 5). We also showed that sodium was able to
rescue the growth defect caused by the impaired
potassium antiporter function (Fig. 2A). In LB, TSB or
the medium used for the antimicrobial assays enough
sodium is present to rescue the growth defect caused by
the impaired potassium antiporter function of YeO8c±
RosB2 and therefore the only possible conclusion is that
the additional cytoplasmic acidification due to the RosB
activity must be associated with the CAMPs resistance.
Therefore, artificial lowering of the intracellular pH should
mimic the RosB activity, and furthermore this should be
reflected as CAMPs resistance in the rosB mutant. The
sensitivity experiments performed in the presence of
sodium acetate were in line with this hypothesis. The
resistance of YeO8c±RosB2 to polymyxin B reached the
same levels as that of YeO8c when the assays were
performed in the presence of 15 mM sodium acetate, but
5 or 10 mM sodium acetate were not able to rescue the
mutant (Fig. 4). The effect of weak acids on the intracellular pH depends upon the pKa of the acid, the
transmembrane pH gradient and the concentration of the
acid used (Salmond et al., 1984). Therefore, it is likely that
5 and 10 mM sodium acetate concentrations were too low
to reach a critical intracellular threshold pH. It is worth
noting that similar results were obtained when sodium
acetate was used to protect an E. coli kefB/kefC double
mutant against methylglyoxal (Ferguson et al., 1995). It is
important to point out that the survival of the YeO8c±
RosA2B2 mutant in the presence of polymyxin B was also
significantly enhanced in the presence of sodium acetate
(Fig. 4), providing further evidence to support our hypothesis that cytoplasmic acidification is involved in the
resistance to CAMPs. However, with YeO8c±RosA2B2,
in contrast to YeO8c±RosB2, artificial acidification did not
give full protection (Fig. 4) and therefore we believe that
RosA is needed to remove the CAMPs from the
cytoplasmic membrane. Because RosA probably resides
in the cytoplasmic membrane it will pump the CAMPs into
the periplasmic space where they would accumulate
without an accessory protein (Fig. 5). A likely candidate
for this is TolC, which was recently shown to enable the
transport of substrates from a cytoplasmic acceptor to the
extracellular environment without the presence of a
periplasmic protein that would bridge the cytoplasmic
and outer membranes (Koronakis et al., 1997). In line with
this, a functional TolC is required for the operation of the
AcrAB efflux system of E. coli leading to antibiotic
resistance (Ma et al., 1995; Fralick, 1996). Studies are
in progress to prove this hypothesis.
At present, we can only speculate on the mechanism of
how cytoplasmic acidification could protect against
CAMPs. This is even more difficult because it is not
completely understood how CAMPs exert their antimicrobial action (Hancock and Chapple, 1999). A variety of
mechanisms has been proposed of which the best
postulates that CAMPs interfere with the bacterial DNA
and that acidification of the cytoplasm could act against
this. Acidification could induce DNA repair enzymes or
DNA-binding proteins that physically could protect DNA
or, more directly, the low pH could prevent the interaction
of CAMPs with DNA. Alternatively, it is possible that the
low pH could act as a signal to induce other systems
required for CAMPs resistance that might even play a role
at the OM level (Fig. 5). Further studies are needed to
clarify these issues.
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
Antimicrobial peptide resistance in Yersinia
Another fundamental finding in this work was that the
ros locus is upregulated both by growth at 378C and by the
presence of the CAMPs. The upregulation by the presence
of polymyxin B happened at both growth temperatures. To
the best of our knowledge this is the first report of a
CAMPs resistance system which is induced by the
presence of CAMPs. Temperature regulates many virulence factors of Yersinia so that their optimal expression
takes place at 378C (Straley and Perry, 1995). However,
the expression of the O-antigen gene cluster of YeO8c is
downregulated at 378C (Zhang, 1996). Under these
conditions, Y. enterocolitica is more susceptible to
CAMPs than when grown at RT and this correlates with
a higher binding of the CAMPs to the OM (this work and
Bengoechea et al., 1998). Moreover, a rough YeO8c
mutant was more sensitive to CAMPs than the wild-type
strain (data not shown). Therefore, it is reasonable to
assume that the ros system becomes induced under
conditions in which the OM barrier to CAMPs is not fully
functional, which can even be due to the presence of the
CAMPs themselves (Fig. 5). To support this idea further,
we found that both the wild type and the rosAB strains,
when grown at 378C at acidic pH, were more resistant to
polymyxin B than when grown at neutral pH, and that the
ros locus expression was downregulated at acidic pH
(Table 2). Furthermore, at acidic pH both the wild type and
the rosAB bacteria expressed a typical O-antigen ladder in
contrast to bacteria grown at neutral pH (data not shown).
Based on these findings, we postulate that the ros
regulation is inversely coupled to the expression of the
O-antigen cluster. Supporting this conclusion, the ros
locus was no longer temperature regulated in YeO8
strains lacking the O-antigen (unpublished).
It is important to note that the CAMPs and potassium
sensitivity could not be fully complemented when the ros
locus was expressed even from a low-copy number
plasmid and that the E. coli C600/pLZ4k resistance to
CAMPs was abrogated if the ros locus was induced from
a plasmid tac promoter. These data indicate that the
RosAB system must not be overexpressed otherwise its
function is compromised. The reason for this is not known
at present but one can speculate that the cytoplasmic
membrane can accommodate only a certain number of
the RosAB systems and that oversaturation of the
membrane is disadvantageous. Furthermore, if the ros
locus belongs to the same network as the O-antigen
expression it would not be surprising to find that it is tightly
regulated. For example, in Shigella flexneri the balance
between the O-antigen polymerase (Wzy) and the regulator of the of O-antigen chain length (Wzz) is important
for a normal O-antigen expression (Daniels and Morona,
1999). Indeed, we have some new data indicating that in
YeO8 there is a cross-talk between Wzz and RosAB
affecting somehow the O-antigen expression.
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
75
Based on this discussion and data, as well as ideas of
other investigators, we have formulated a working model
to account for the current data, which we have summarized in Fig. 5. When bacteria are grown at RT they are
relatively resistant to the CAMPs antimicrobial action, and
the CAMPs resistance mechanisms involve OM modifications and specifically those of the LPS molecule. One
mechanism consists of reduced electrostatic interactions
between CAMPs and the negatively charged bacterial
surface, which occurs through modification of lipid A
phosphate with aminoarabinose. Another one involves the
presence of long O-antigen molecules that sterically
hinder the access of CAMPs to bind to the deeper parts
of LPS. On the contrary, when bacteria are grown at 378C,
the O-antigen molecules no longer provide a shield to
prevent CAMPs interaction. Moreover, there is also a
marked decrease in the lipid A phosphate modifications.
In these conditions, the bacterial surface is more
susceptible to the CAMPs interaction. In this situation,
the RosA/RosB system is induced and it becomes a
critical mechanism of CAMPs resistance in two ways: (i)
by pumping out the CAMPs once they reach the
cytoplasmic membrane; and (ii) by inducing the acidification of the cytoplasm. The lower intracellular pH could act
as a positive regulatory signal for the induction of other
CAMPs resistance mechanisms that might involve
changes at the OM level. Alternatively, as we have
discussed above, the low intracellular pH could play a
direct role on the resistance to CAMPs.
We have recently shown that different Yersinia species
and biogroups vary in their susceptibility to CAMPs,
specially when the strains are grown at 378C (Bengoechea
et al., 1996; 1998). We have also demonstrated that OM
differences between yersiniae, and specifically those in
the LPS molecule, could account at least in part for such
differences (Bengoechea et al., 1998). In view of the
results presented in this work, we reasoned that another
possible explanation could be the absence or presence of
the rosAB genes in these strains. However, the slot-blot
hybridization analysis showed that sequences homologous to rosAB were present in all yersiniae tested. One
might speculate that the rosAB system serves another
unknown role than CAMPs resistance in non-pathogenic
yersiniae that might have evolved in pathogenic yersiniae
to withstand CAMPs antimicrobial action.
Experimental procedures
Bacterial strains, plasmids and growth conditions
Listed in Table 3 are the bacterial strains and plasmids used
in this work. Bacteria were routinely cultured in LB, TSB or in
Luria agar (LA) plates. Other media used were the potassium
MM described by Epstein and Kim (1971), M63 medium and
TM adjusted to different pH [50 mM Tris-HCl; (NH4)SO4,
76 J. A. Bengoechea and M. Skurnik
Table 3. Bacterial strains and plasmids used in this work.
Bacterial strains and plasmids
Genotype or comments
References or sources
E. coli
C600
SY327 lpir
Sm10 lpir
thi thr leuB tonA lacY supE
l (lac pro) argE (Am) rif nalA recA56 (l pir)
thi thr leuB tonA lacY supE recA::RP4-2-Tc::Mu-Km (l pir)
Appleyard (1954)
Miller and Mekalanos (1988)
Simon et al. (1983)
R2M1 derivative of wild type strain 8081, serotype O:8, pYV1
R2M1 derivative of 8081-c the pYV-cured derivative of 8081
Serotype O:3, patient isolate, wild type
Virulence plasmid cured derivative of 6471/76
YeO8c, rosAB::Km-GenBlock, KmR
rosA1, non-polar frame shift mutation
rosB1, non-polar frame shift mutation
Reporter strain, pRV4kluc integrated into the rosAB
locus by single cross-over
Zhang et al. (1997)
Zhang and Skurnik (1994)
Skurnik (1984)
Skurnik (1984)
Zhang (1996)
This work
This work
This work
Zhang (1996)
This work
This work
This work
This work
This work
Skurnik et al. (1995)
This work
pMP200
6 kb BamHI fragment of pLZ5005 cloned into pTM100, ClmR
4 kb EcoRI fragment of pLZ6k cloned into pMMB207, ClmR
KpnI deletion derivative of pLZ4k, ClmR
4 kb EcoRI fragment of pLZ4k cloned into pUC18, AmpR
KpnI±SphI deletion derivative of pUC18, AmpR
4 kb EcoRI fragment of pLZ4k cloned into pUCD, AmpR
Suicide vector, a derivative of pJM703.1, ClmR
4 kb EcoRI fragment of pUC4k cloned into EcoRV
site of pRV1, ClmR
Derivative of pUCD4k with a non-polar frameshift
mutation in rosA, AmpR
4 kb fragment of pUCD4kAM cloned into EcoRV site of pRV1, ClmR
Derivative of pRV4k with a non-polar frameshift
mutation in rosB, ClmR
Derivative of pUC4k, NsiI fragment replaced by lucFF, AmpR
EcoRI fragment of pUC4kluc cloned into EcoRV
site of pRV1, ClmR
lucFF cloned in a pTM100 derivative, ClmR
pTM100
pMMB207
Mobilizable vector, pACYC184-oriT of RK2, ClmR
Expression vector, ClmR
Y. enterocolitica
8081-R2M1 (YeO8)
8081-res (YeO8c)
6471/76 (YeO3)
6471/76-c (YeO3c)
YeO8c±RosA2B2
YeO8c±RosA2
YeO8c±RosB2
YeO8c±RosA2B2 ±::pRV4kluc
Plasmids
pLZ6k
pLZ4k
pLZ4kKpn
pUC4k
pUCD
pUCD4k
pRV1
pRV4k
pUCD4kAM
pRV4kAM
pRV4kKpn
pUC4kluc
pRV4kluc
This work
This work
This work
This work
This work
S. Kiljunen, M. Pajunen and
M. Skurnik; unpublished data
Michiels and Cornelis (1991)
Morales et al. (1991)
2 g l21; FeSO4 7H2O, 0.5 mg l21; 1 mM MgSO4 7H2O;
0.2% glucose; 1 mg ml21 thiamine]. The influence of low pH
was tested by growing the bacteria in TSB acidified to pH 5.5
with HCl. E. coli strains were grown at 378C and Y.
enterocolitica strains at RT (22±258C) unless otherwise
indicated. When appropriate, antibiotics were added to the
growth media at the following concentrations: ampicillin
(Amp), 100 mg ml21; kanamycin (Km), 100 mg ml21 in agar
plates and 20 mg ml21 in broth; chloramphenicol (Clm),
20 mg ml21; streptomycin (Sm), 25 mg ml21; and cycloserine,
2 mg ml21.
applied to the slots as recommended by the manufacturer.
DNA fragments were labelled with digoxigenin using the High
Prime DNA labelling kit (Roche). Hybridizations were carried
out under high-stringency conditions with the buffers and
times described by the manufacturer (Roche). Detection of
hybridization was performed using the DIG luminescence
Detection kit for Nucleid Acids (Roche). Mobilization of
plasmids from E. coli strains to Y. enterocolitica strains
were performed as described earlier (Skurnik et al., 1995).
The suicide vector pRV1 derivatives were maintained in
E. coli SY327lpir or in Sm10lpir (Table 3).
Nucleic acid manipulation
Construction of YeO8c±RosA2B2 ±rev by reverse allelic
exchange
Routine techniques for plasmid isolation, restriction digestion,
ligation and transformation were used (Sambrook et al.,
1989). PCR was performed according to the supplier of the
thermostable DNA polymerase, DynaZyme II (Finnzymes).
Reaction conditions for PCR cycles were adjusted according
to the oligonucleotide primers used and the length of the
amplified fragment. Slot blotting was performed using the
minifold II Slot Blot Manifold (Schleisser and Schuell).
Denatured genomic DNA isolated by the CTAB method was
The 4.4 kb EcoRI fragment of pLZ4k that contains the rosAB
genes cluster and part of the flanking gsk and ushA genes,
was cloned into the EcoRI site of pUC18, and the resulting
plasmid was named pUC4k. This plasmid was digested with
EcoRI; the fragments were blunt-ended by Klenow treatment,
the 4.4 kb fragment was gel purified and finally cloned into
the EcoRV site of the suicide vector pRV1, giving plasmid
pRV4k. pRV4k was transformed into E. coli Sm10lpir, which
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
Antimicrobial peptide resistance in Yersinia
2
2
then mobilized the plasmid into YeO8c±RosA B . Ten
KmRClmR transconjugants, having pRV4k integrated into
the genome by homologous recombination (YeO8c±
RosA2B2::pRV4k), were pooled and subjected to cycloserine
enrichment to select for derivatives that by a second
homologous recombination event had deleted the Km GenBlock and the vector sequences from the genome. One KmS
ClmS recombinant surviving the cycloserine enrichment was
selected and named YeO8c±RosA2B2 ±rev.
Construction of non-polar mutations in the rosA and rosB
genes
rosA non-polar mutant. First, the unique BamHI site of the
polylinker region of pUC18 was eliminated by deletion of
the DNA fragment between the KpnI and SphI sites, and the
resulting plasmid was named pUCD. The 4.4 kb EcoRI
fragment of pUC4k was cloned into pUCD to obtain plasmid
pUCD4k.
To construct a deletion using the PCR-based method
described earlier (Byrappa et al., 1995) primers LZ60 (5 0 CACTGCCGATTGGCATGG-3 0 ) and JA-6 (5 0 -GGGATCCGGATGCTTATCGGTATAA-3 0 ), with a BamHI site (underlined), were phosphorylated at the 5 0 end and used without
further purification in the PCR reaction. The primers were
designed to introduce a four-nucleotide deletion in the rosA
gene and a new BamHI site for screening purposes.
Amplifications were carried out in a 50 ml reaction volume
containing 25 pmol of each phosphorylated primer pair,
400 ng of pUCD4k from minipreprations, and 0.5 ml of Vent
DNA polymerase (2 units ml21; New England BioLabs). The
PCR steps (948C 30 s, 498C 30 s and 728C 7 min) were
repeated 15 times. The appropriate sized band was gel
purified and self-ligated to obtain pUCD4kAM. This plasmid
was digested with EcoRI and blunt-ended with Klenow and
the 4.4 kb fragment was gel purified and cloned into the
EcoRV site of pRV1 to give pRV4kAM. pRV4kAM was
transformed into E. coli Sm10lpir and from there mobilized
into YeO8c±RosA2B2. A pool of KmRClmR transconjugants,
having the plasmid integrated into the genome by homologous recombination (YeO8c±RosA2B2::pRV4kAM), was
subjected to cycloserine treatment to enrich derivatives that
had deleted by a second homologous recombination event
the Km GenBlock and the vector sequences. One KmS ClmS
recombinant was selected and named YeO8c±RosA2
rosB non-polar mutant. To generate a non-polar mutation in
rosB we took advantage of the presence of the unique KpnI
site in the rosB gene and its absence in the suicide vector
pRV1. Thus, pRV4k was KpnI digested, treated with T4 DNA
polymerase to remove the 3 0 -overhang and blunt-end ligated
to give pRV4kKpn. pRV4kKpn was mobilized from E. coli
Sm10lpir into YeO8c±RosA2B2. A KmRClmR transconjugant
having pRV4kKpn integrated to the genome by homologous
recombination (YeO8c±RosA2B2::pRV4kKpn) was selected.
A derivative in which the Km GenBlock and the vector
sequences were lost by a second homologous recombination was selected after cycloserine enrichment, and called
YeO8±RosB2.
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
77
Permeability to lysozyme
Bacteria were grown in 10 ml of TSB in 50 ml flasks on an
orbital shaker (250 r.p.m.), and growth was monitored by
measuring the optical density (OD) at 540 nm (OD540). Cells
were harvested (5000 g, 15 min, 58C) in the exponential
phase of growth, and resuspended in 2 mM HEPES (pH 7.2)
at an OD500 of 0.7, and lysozyme (3 mg ml21) and polymyxin
B (9 mg ml21) were added. The suspensions were incubated
at the growth temperature, and cell lysis was monitored by
the decrease in the OD500. Lysozyme or polymyxin B alone
did not produce any OD decrease under these experimental
conditions. Results were expressed as percentages of the
optical density of controls incubated in the absence of
lysozyme and polymyxin B. All experiments were run with
triplicate samples from two independently grown cultures of
cells.
Antimicrobial assays
Radial diffusion assay. The antimicrobial activities of polymyxin B, cecropin P1 and melittin (all purchased from Sigma
Chemical) were assayed using the radial diffusion method
previously described (Lehrer et al., 1991). Briefly, bacteria
were grown as described in the above section and collected
in the exponential phase of growth. An underlay gel that
contained 1% (w/v) agarose (SeaKem LE agarose, FMC),
2 mM HEPES (pH 7.2) and 0.3 mg of TSB powder ml21 was
equilibrated at 508C, and inoculated with the different bacteria
to a final concentration of 6.1 105 cfu ml21 of molten gel.
This gel was poured into standard Petri dishes, and after
polymerization, small wells of 10 ml capacity were carved.
Aliquots of 5 ml of the different CAMPs were added and
allowed to diffuse for 2 h at 378C. After that, a 10 ml overlay
gel composed of 1% agarose and 6% TSB powder in water
was poured on top of the previous one and the plates were
incubated overnight at 378C. The next day, the diameters of
the inhibition halos were measured to the nearest 0.1 mm,
and after subtracting the diameter of the well, were
expressed in inhibition units (10 units ˆ 1 mm). MIC was
estimated by performing linear regression analysis (units
compared with log10 concentration) and determining the
x-axis intercepts. All the experiments were run in
quadruplicate in two independent occasions.
Polymyxin B sensitivity assay. Strains were grown at 378C in
10 ml of TSB in 50 ml flasks on an orbital shaker (250 r.p.m.)
and harvested (5000 g, 15 min, 58C) in the exponential phase
of growth. Then, a suspension that contained approximately
2.1 105 cfu ml21 was prepared in 1% tryptone (w/v)-PBS
(pH 7.4). Next, 10 ml of this cell suspension was mixed in
Eppendorf tubes with various concentrations of polymyxin B
in a volume of 200 ml, and incubated at the growth
temperature of the bacteria for 30 min. After that, 100 ml of
the suspensions was directly plated on LA plates and the
plates were incubated at the growth temperature of bacteria.
Colony counts were determined and results were expressed
as percentages of the colony counts of bacteria not exposed
to polymyxin B. All experiments were carried out with
duplicate samples in at least two independent occasions.
78 J. A. Bengoechea and M. Skurnik
Binding of polymyxin B by viable cells
Construction of a luciferase reporter strain
The assay described by Freer et al. (1996) was carried out
with minor modifications. Briefly, cells grown as described
above were resuspended in 2 mM HEPES pH 7.2 at
approximately 1.25 1010 cfu ml21, and 100 ml aliquots
were mixed with 12 ml containing different amounts of
polymyxin B. After 5 min of incubation at RT, the cells with
the bound polymyxin B were sedimented (12 000 g, 10 min,
RT), the supernatant centrifuged two more times under the
same conditions, and the unbound polymyxin B measured in
a bioassay. To this end, the radial diffusion assay described
above was used with E. coli ATCC352183 as the indicator
bacterium. The amount of bound polymyxin B was calculated
using polymyxin B dilutions tested in the same plate.
The firefly luciferase reporter and Pir-dependent suicide
vector was constructed as follows. The firefly luciferase gene
(lucFF) coding region was amplified by PCR using pMP200
(S. Kiljunen, M. Pajunen and M. Skurnik, unpublished
construct) as a template with phosphorylated primers
BCCP-1 (5 0 -GAGGAGAAATTAACTATGAGGGG-3 0 ) and
BCCP-2 (5 0 -TTACAATTTGGACTTTCCGCC-3 0 ). Plasmid
pUC4k was digested with NsiI and treated with T4 DNA
polymerase to generate blunt ends. The resulting 5.5 kb
fragment was gel purified and ligated with the lucFF-PCR
fragment. The obtained plasmid, pUC4kluc, contained the
internal 1.5 kb NsiI fragment of the ros cluster replaced by the
1.7 kb promoterless lucFF gene. Plasmid pUC4kluc was
XhoI±SalI digested and the ends filled in with the Klenow
fragment of DNA polymerase I. The generated 5.5 kb
fragment was gel purified and cloned into the EcoRV site of
pRV1 to create pRV4kluc. Light production was used to verify
the expression of the lucFF gene in all the constructs before
the next cloning or transformation step.
This suicide vector was mobilized into YeO8c±RosA2B2
from E. coli Sm10lpir, and a KmRClmR transconjugant in
which the suicide vector was integrated by homologous
recombination (YeO8c±RosA2B2::pRV4Kluc) was selected.
Sensitivity to antibiotics and chemicals
Sensitivities to crystal violet, rifampicin, novobiocin, tetracycline, EB, BK and CTAB (all purchased from Sigma) were
assessed on LA using the disc diffusion test. The discs
(concentration discs, é 6.5 mm, Difco laboratories) were
prepared by loading, per disc, 40 mg of crystal violet, 10 mg of
rifampicin, 50 mg of novobiocin, 10 mg of tetracycline, 20 mg
of EB, 100 mg of CTAB or 100 mg of BK dissolved in 20 ml of
distilled water. The discs were dried overnight at 378C and
kept at 48C until needed. Fresh exponentially growing
bacteria were resuspended in PBS to a final concentration
of 108 cfu ml21, and a lawn was prepared on the agar plates
with a sterile swab. The antibiotic-loaded discs were placed
on the lawns, and the plates incubated for 18 h at the growth
temperature of the cultures. The diameters of the inhibition
halos were measured to the nearest 1 mm and, after
subtracting the disc diameter, they were expressed in units
(1 unit equal to 10 mm). All experiments were performed with
duplicate samples from two independently grown cultures of
bacteria, with similar results.
Sodium deoxycholate (DOC) (Sigma) sensitivity experiments were performed as previously described (Bengoechea
et al., 1998) with minor modifications. The strains were grown
in TSB at 378C until the exponential phase of growth, pelleted
and resuspended in 2 mM HEPES (pH 7.2) to an OD450 of
0.21. Aliquots of 2 ml were transferred to two tubes, one
containing 200 ml of DOC (0.5±1% final concentration) in
2 mM HEPES, and the other, the control tube, 200 ml of
2 mM HEPES. The tubes were incubated for 60 min at
the growth temperature of the bacteria, and the OD450 of
the suspensions was measured. The DOC sensitivity of the
bacteria resulted in cellular lysis, which was seen as
decrease in OD450. The results were expressed as percent
turbidity of the DOC tube from the control tube without DOC.
All experiments were performed in triplicate with two
independently grown cultures of bacteria.
Ethanol sensitivity was assessed by following the bacterial
growth in liquids culture supplemented with the organic
solvent. Briefly, 1 ml of a 5 ml overnight culture grown at RT
was added to 50 ml of TSB and incubated on an orbital
shaker (250 r.p.m.) at 378C. At the middle of the exponential
phase, ethanol [2±4% (v/v) final concentration] was added
and growth was monitored by measuring the increase in OD
at 540 nm for a further 4 h.
Luciferase assay
The reporter strains were grown until mid-log phase, pelleted
and resuspended to an OD540 of 0.3 in 100 mM Tris-HCl
(pH 7.5). A 100 ml aliquot of the bacterial suspension was
transferred to a luminometer tube (Bio-Orbit) and mixed with
100 ml of luciferase assay reagent (Bio-Orbit) (1 mM in
100 mM citrate buffer pH 5). Luminescence was immediately
measured with a 1254 Luminova luminometer (Bio-Orbit),
and expressed as relative luminescence units (RLU). To test
the effect of polymyxin B and potassium chloride in the
expression of the luciferase fusion, they were added when
the culture reached the exponential phase and, after 1 h of
incubation, the culture was treated as described above. All
measurements were carried out in quadruplicate in at least
two separate occasions.
Acknowledgements
We thank S. Haataja, P. Ollikka and M. Pajunen for critical
reading of the manuscript, and the Skurnik laboratory
members for their help and advice. This work was supported
by grants from the Academy of Finland, Sigrid Juselius
Foundation and the Centre for International Mobility.
References
Appleyard, R.K. (1954) Segregation of new lysogenic types
during growth of doubly lysogenic strain derived from
Escherichia coli K12. Genetics 39: 440±452.
Bengoechea, J.A., DõÂaz, R., and MoriyoÂn, I. (1996) Outer
membrane differences between pathogenic and environmental Yersinia enterocolitica biogroups probed with
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
Antimicrobial peptide resistance in Yersinia
hydrophobic permeants and polycationic peptides. Infect
Immun 64: 4891±4899.
Bengoechea, J.A., Lindner, B., Seydel, U., DõÂaz, R., and
MoriyoÂn, I. (1998) Yersinia pseudotuberculosis and Yersinia pestis are more resistant to bactericidal cationic
peptides than Yersinia enterocolitica. Microbiology 144:
1509±1515.
Bentley, J., Hyatt, L.S., Ainley, K., Parish, J.H., Herbert, R.B.,
and White, G.R. (1993) Cloning and sequence analysis of
an Escherichia coli gene conferring bicyclomycin resistance. Gene 127: 117±120.
Boman, H.G. (1995) Peptide antibiotics and their role in
innate immunity. Annu Rev Immunol 13: 61±92.
Bottone, E.J. (1997) Yersinia enterocolitica: the charisma
continues. Clin Microbiol Rev 10: 257.
Byrappa, S., Gavin, D.K., and Gupta, K.C. (1995) A highly
efficient procedure for site-specific mutagenesis of fulllength plasmids using Vent DNA polymerase. Genome Res
5: 404±407.
Cornelis, G.R., and Wolf-Watz, H. (1997) The Yersinia yop
virulon: a bacterial system for subverting eukaryotic cells.
Mol Microbiol 23: 861±867.
Cox, A.D., and Wilkinson, S.G. (1991) Ionizing groups in
lipopolysaccharides of Pseudomonas cepacia in relation to
antibiotic resistance. Mol Microbiol 5: 641±646.
Daniels, C., and Morona, R. (1999) Analysis of Shigella
flexneri Wzz (Rol) function by mutagenesis and crosslinking: Wzz is able to oligomerize. Mol Microbiol 34: 181±
194.
Epstein, W., and Kim, B.S. (1971) Potassium transport loci in
Escherichia coli K-12. J Bacteriol 108: 639±644.
Ferguson, G.P., McLaggan, D., and Booth, I.R. (1995)
Potassium channel activation by glutathione-S-conjugates
in Escherichia coli: protection against methylglyoxal in
mediated by cytoplasmic acidification. Mol Microbiol 17:
1025±1033.
Fralick, J.A. (1996) Evidence that TolC is required for
functioning of the Mar/AcrAB efflux pump of Escherichia
coli. J Bacteriol 178: 5803±5805.
Freer, E., Moreno, E., MoriyoÂn, I., Pizzarro-CerdaÂ, J.,
Weintraub, A., and Gorvel, J.P. (1996) Brucella/Salmonella-lipopolysaccharide chimeras are less permeable to
hydrophobic probes and more sensitive to cationic peptides
than their native Brucella spp. counterparts. J Bacteriol
178: 5867±5876.
Fujisaki, S., Ohnuma, S., Horiuchi, T., Takahashi, I., Tsukui,
S., Nishimura, Y., et al. (1996) Cloning of a gene from
Escherichia coli that confers resistance to fosmidomycin as
a consequence of amplification. Gene 175: 83±87.
GruÈtzkau, A., Hanski, C., Hahn, H., and Riecken, E.O. (1990)
Involvement of M cells in the bacterial invasion of Peyer's
batches: a common mechanism shared by Yersinia
enterocolitica and other enteroinvasive bacteria. Gut 31:
1011±1015.
Gunn, J.S., and Miller, S.I. (1996) Pho-PhoQ activates
transcription of pmrAB, encoding a two-component regulatory system involved in Salmonella typhimurium antimicrobial peptide resistance. J Bacteriol 178: 6857±6864.
Gunn, J.S., Lim, K.B., Krueger, J., Kim, K., Guo, L., Hackett,
M., et al. (1998) PmrA-PmrB-regulated genes necessary
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80
79
for 4-aminoarabinose lipid A modification and polymyxin
resistance. Mol Microbiol 27: 1171±1182.
Guo, L., Lim, K.B., Gunn, J.S., Bainbridge, B., Darveau, R.P.,
Hackett, M., et al. (1997) Regulation of lipid A modifications
by Salmonella typhimurium virulence genes phop-phoQ.
Science 276: 250±253.
Guo, L., Lim, K.B., Poduje, C.M., Daniel, M., Gunn, J.S.,
Hackett, M., et al. (1998) Lipid A acylation and bacterial
resistance against vertebrate antimicrobial peptides. Cell
95: 189±198.
Hancock, R.E.W. (1984) Alterations in outer membrane
permeability. Annu Rev Microb 38: 237±264.
Hancock, R.E., and Lehrer, R. (1998) Cationic peptides: a
new source of antibiotics. Trends Biotechnol 16: 82±88.
Hancock, R.E.W., and Chapple, D.S. (1999) Peptide antibiotics. Antimicrob Agents Chemother 43: 1317±1323.
Hanski, C., Kutschka, U., Schmoranzer, H.P., Naumann, M.,
Stalmach, A., Hahn, H., et al. (1989) Immunohistochemical
and electron microscopic study of interaction of Yersinia
enterocolitica serotype O8 with intestinal mucosa during
experimental enteritis. Infect Immun 57: 673±678.
Hase, S., and Reitschel, E.T. (1977) The chemical structure
of the lipid A component of lipopolysaccharides from
Chromobacterium violaceum NCTC 9694. Eur J Biochem
75: 23±34.
Helander, I.M., KilpelaÈinen, I., and Vaara, M. (1994)
Increased substitution of phosphate groups in lipopolysaccharides and lipid A of the polymyxin-resistant pmrA
mutants of Salmonella typhimurium: a31 P-NMR study.
Mol Microbiol 11: 481±487.
Helander, I.M., Kato, Y., KilpelaÈinen, Y., Kostianen, R.,
Lindner, B., Nummila, K., et al. (1996) Characterization of
lipopolysaccharides of polymyxin-resistant and polymyxinsensitive Klebsiella pneumoniae O3. Eur J Biochem 237:
272±278.
Koronakis, V., Li, J., Koronakis, E., and Stauffer, K. (1997)
Structure of TolC, the outer membrane component of the
bacterial type I efflux system, derived from two-dimensional
crystals. Mol Microbiol 23: 617±626.
Kupferwasser, L.I., Skurray, R.A., Brown, M.H., Firth, N.,
Yeaman, M.R., and Bayer, A.S. (1999) Plasmid-mediated
resistance to thrombin-induced platelet microbicidal protein
in staphylococci: role of the qacA locus. Antimicrob Agents
Chemother 43: 2395±2399.
Lehrer, R.I., Rosenman, M., Harwig, S.S., Jackson, R., and
Eisenhauer, P. (1991) Ultrasensitive assays for endogenous antimicrobial polypeptides. J Immunol Methods 137:
167±173.
Ma, D., Cook, D.N., Alberti, M., Pon, N.G., Nikaido, H., and
Hearst, J.E. (1995) Genes acrA and acrB encode a stressinduced efflux system of Escherichia coli. Mol Microbiol 16:
45±55.
MartõÂnez de Tejada, G., and MoriyoÂn, I. (1995) The outer
membranes of Brucella spp. are resistant to bactericidal
cationic peptides. Infect Immun 63: 3054±3061.
Michiels, T., and Cornelis, G.R. (1991) Secretion of hybrid
proteins by the Yersinia Yop export system. J Bacteriol
173: 1677±1685.
Miller, V.L., and Mekalanos, J.J. (1988) A novel suicide
vector and its use in construction of insertion mutations:
Osmoregulation of outer membrane proteins and virulence
80 J. A. Bengoechea and M. Skurnik
determinants in Vibrio cholerae requires. Toxr J Bacteriol
170: 2575±2583.
Morales, V.M., BaÈckman, A., and Bagdasarian, M. (1991) A
series of wide-host-range low-copy-number vectors that
allow direct screening of recombinants. Gene 97: 39±47.
Munro, A.W., Ritchie, G.Y., Lamb, A.J., Douglas, R.M., and
Booth, I.R. (1990) The cloning and DNA sequence of the
gene for the glutathione-regulated potassium-efflux system. Mol Microbiol 5: 607±616.
Neyfakh, A.A., Bidnenko, V.E., and Chen, L.B. (1991) Effluxmediated multidrug resistance in Bacillus subtilis: similarities and dissimilarities with the. Proc Natl Acad Sci USA
88: 4781.
Nicolas, P., and Mor, A. (1995) Peptides as weapons against
microorganisms in the chemical defence system of
vertebrates. Annu Rev Microbiol 49: 277±304.
Nikaido, H. (1996) Multidrug efflux pumps of gram-negative
bacteria. J Bacteriol 178: 5853±5859.
Nummila, K., KilpelaÈinen, I., ZaÈhringer, U., Vaara, M., and
Helander, I.M. (1995) Lipopolysaccharides of polymyxin
B-resistant mutants of Escherichia coli are extensively
substituted by 2-aminoethyl pyrophosphate and contain
aminoarabinose in lipid A. Mol Microbiol 16: 271±278.
Paulsen, I.T., Brown, M.H., and Skurray, R.A. (1996) Protondependent multidrug efflux systems. Microb Rev 60: 575±
608.
Salmond, C.V., Kroll, R.G., and Booth, I.R. (1984) The effect
of food preservatives on pH homeostasis. J Gen Microbiol
130: 2845±2850.
Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989)
Molecular Cloning: a Laboratory Manual, 2nd edn. Cold
Spring Harbor, NY: Cold Spring Harbor Laboratory Press.
Shafer, W.M., Qu, X., Waring, A.J., and Lehrer, R.I. (1998)
Modulation of Neisseria gonorrhoeae susceptibility to
vertebrate antibacterial peptides due to a member of the
resistance/nodulation/division efflux pump family. Proc Natl
Acad Sci USA 95: 1829±1833.
Sidorczyk, Z., Zahringer, U., and Rietschel, E.T. (1983)
Chemical structure of the lipid A component of the
lipopolysaccharide from a Proteus mirabilis Re-mutant.
Eur J Biochem 137: 15±22.
Simon, R., Priefer, U., and PuÈhler, A. (1983) A broad host
range mobilization system for in vivo genetic engineering:
Transposon mutagenesis in Gram negative bacteria. Bio/
Technology 1: 784±791.
Skurnik, M. (1984) Lack of correlation between the presence
of plasmids and fimbriae in Yersinia enterocolitica and
Yersinia pseudotuberculosis. J Appl Bacteriol 56: 355±
363.
Skurnik, M., Venho, R., Toivanen, P., and Al-Hendy, A.
(1995) A novel locus of Yersinia enterocolitica serotype O:3
involved in lipopolysaccharide outer core biosynthesis. Mol
Microbiol 17: 575±594.
Skurnik, M., Venho, R., Bengoechea, J.-A., and MoriyoÂn, I.
(1999) The lipopolysaccharide outer core of Yersinia
enterocolitica serotype O:3 is required for virulence and
plays a role in outer membrane integrity. Mol Microbiol 31:
1443±1462.
Sola-Landa, A., Pizarro-Cerda, J., Grillo, M.J., Moreno, E.,
Moriyon, I., Blasco, J.M., et al. (1998) A two-component
regulatory system playing a critical role in plant pathogens
and endosymbionts is present in Brucella abortus and
controls cell invasion and virulence. Mol Microbiol 29: 125±
138.
Soncini, F.C., and Groisman, E.A. (1996) Two-component
regulatory systems can interact to process multiple
environmental signals. J Bacteriol 178: 6796±6801.
Straley, S.C., and Perry, R.D. (1995) Environmental modulation of gene expression and pathogenesis in Yersinia.
Trends Microbiol 3: 310±317.
Vaara, M. (1992) Agents that increase the permeability of the
outer membrane. Microb Rev 56: 395±411.
Waser, M., Hess Bienz, D., Davies, K., and Solioz, M. (1992)
Cloning and disruption of a putative NaH-antiporter gene of
Enterococcus hirae. J Biol Chem 267: 5396±5400.
Zhang, L. (1996) Molecular genetics of the O-antigen of
Yersinia enterocolitica lipopolysaccharide. PhD Thesis. In
Annales Universitatis Turkuensis, Ser. D. Turku, Finland:
University of Turku, pp. 118.
Zhang, L., and Skurnik, M. (1994) Isolation of an R2M1
mutant of Yersinia enterocolitica serotype 0:8 and its
application in construction of rough mutants utilizing miniTn5 derivatives and lipopolysaccharide-specific phage. J
Bacteriol 176: 1756±1760.
Zhang, L., Radziejewska-Lebrecht, J., Krajewska-Pietrasik,
D., Toivanen, P., and Skurnik, M. (1997) Molecular and
chemical characterization of the lipopolysaccharide Oantigen and its role in the virulence of Yersinia enterocolitica serotype O:8. Mol Microbiol 23: 63±76.
Q 2000 Blackwell Science Ltd, Molecular Microbiology, 37, 67±80