Abstract

An unexpected observation led us to examine the relationship between β-lactam exposure and synthesis of colonic acid capsular polysaccharide in Escherichia coli. Strains containing a cps-lacZ transcriptional fusion were challenged with antibiotics having various modes of action, and gene expression was detected by a disk-diffusion assay and in broth cultures. The cps genes were induced by a subset of β-lactams but not by agents inhibiting protein synthesis or DNA replication, indicating that cps expression was specific and not due to stresses accompanying cell death or by a general inhibition of peptidoglycan synthesis. A narrow concentration just below the MIC triggered cps expression in liquid culture, suggesting the response may be triggered by near-lethal levels of antibiotic. Because colanic acid is important for maturation of biofilm architecture, antibiotics that increase its synthesis might exacerbate the formation or persistence of biofilms.

Introduction

Members of the Enterobacteriaceae constitutively produce small amounts of the exopolysaccharide colanic acid, but when the cells are traumatized by osmotic shock, dehydration, or destabilization of the outer membrane, production of this capsular material increases markedly [1,2]. Because colanic acid is produced when bacteria encounter a hostile environment, its synthesis appears to be a survival mechanism for damaged cells.

Colanic acid is produced by enzymes encoded by the cps (wca) genes [3,4], which are regulated by rcsA, rcsB, and rcsC[2,5], by rcsF[6], and by lon, which degrades RcsA [3]. Induction is initiated by a two-component signaling pathway in which a sensor kinase, RcsC, phosphorylates RcsB, which subsequently forms a heterodimer with RcsA and activates cps transcription [4,5,7]. Although the steps in the rcs-cps pathway are fairly well defined, the signal sensed by RcsC remains unknown [7,8]. What is clear is that the resulting polysaccharide is an important component in biofilm formation and architecture [9,10]. Biofilms function as a dense protective matrix for imbedded microbes and contribute to nosocomial infections. The host immune response is often ineffective against cells protected within biofilms, and antibiotics, disinfectants and UV light have difficulty penetrating these masses [11].

While constructing Escherichia coli mutants from which the penicillin binding proteins (PBPs) were deleted in many combinations [12,13], we noted the appearance of a mucoid capsule due to the presence of colanic acid in strains lacking PBP 1b and in strains lacking PBP 1a plus the proteins encoded by the yrfEFGHI loci (unpublished observations, and reference [14]). Since it appeared that alteration of peptidoglycan synthesis might trigger capsule production, we tested the effect of antibiotics on cps induction, because any stressor, including antibiotic therapy, which triggers capsule synthesis, could potentially exacerbate biofilm problems.

Materials and methods

The following E. coli strains were tested for antibiotic susceptibility and cps induction: SG20781 (MC4100 lon+cpsB10::lacZ Mu-immλ) (supplied by S. Gottesman) and FSSG02-1K (SG20781 ΔampC). FSSG02-1K was constructed by moving the ΔampC::Kan allele from E. coli CS14-2K [12] into SG20781 via P1 transduction [15]. Strains were grown in Luria–Bertani (LB) broth or agar [15].

Antibiotic susceptibility testing was performed by disk diffusion. An overnight culture of bacteria was diluted to a density equal to a 0.5 McFarland standard [16] and spread onto LB agar containing 0.1 mg ml−1 5-bromo-4-chloro-3-indolyl-β-d-galactoside (X-gal) [17]. Standard antibiotic disks were obtained from Becton Dickinson (Franklin Lake, NJ, USA), and Fischer Scientific (Pittsburgh, PA, USA). Disks containing cephaloridine (Sigma Chemical Co., St. Louis, MO, USA) and ticarcillin (Research Products International Corp., Mount Prospect, IL, USA) were prepared by saturating filter paper with solutions of 300 and 150 µg ml−1 of cephaloridine and ticarcillin, respectively, and air-drying. Disks were placed onto the plates, incubated for 24 h at 37°C, and the diameters of zones of inhibition were measured. Induction of the cps genes was indicated by a blue halo of β-galactosidase activity that appeared at the periphery of the zones of inhibition. Antibiotics showing weak or negative cps induction were checked again at 48 h and at 1 week, but no changes were observed.

For quantitation of induction in broth cultures, equal numbers of cells were inoculated into LB containing dilutions of antibiotic. Cells in tubes at and below the minimum inhibitory concentration (MIC) were collected and tested for β-galactosidase by measuring the enzymatic activity towards the substrate ONPG (o-nitrophenyl-β-d-galactoside), as described [15].

Results and discussion

While constructing E. coli mutants deficient in different PBPs, we observed that colonies of two strains became mucoid after overnight incubation on plates. This led us to ask if inhibiting PBPs directly with antibiotics would induce expression of the colanic acid capsular genes. E. coli FSSG02-1K and SG20781 contain a cpsB10::lacZ transcriptional fusion, so that induction of the colanic acid structural genes can be monitored by measuring β-galactosidase levels. The strains were exposed to various antimicrobial agents in a disk diffusion assay, and we monitored the presence and relative levels of β-galactosidase induction by observing the formation of the blue precipitate created by action of the enzyme on the substrate, X-gal, included in the agar medium.

When grown in the presence of a disk containing cephaloridine, a β-lactam with high specificity for PBP 1a, an intensely blue ring appeared at the edge of the zone of inhibition (Fig. 1A), indicating that a subinhibitory concentration of this antibiotic induced expression of genes involved in colanic acid synthesis. To determine if cps expression depended on inhibition of a particular component or function (cell wall, membrane, protein or nucleic acid synthesis), we tested a range of antibiotics with different modes of action. Photographs of typical results are reproduced in Fig. 1, and the data are summarized in Table 1. (Results using the ampC+ strain, E. coli SG20781, gave equivalent results except that the zones of inhibition for some β-lactams were smaller because of intrinsic β-lactamase activity.)

1

Examples of cps-lacZ induction by antibiotics. E. coli FSSG02-1K was plated and tested for antibiotic sensitivity and expression of β-galactosidase by disk diffusion assay. Representative reactions are shown. β-galactosidase acting on the substrate X-gal in the medium produced blue rings at the periphery of zones of inhibition. Results for all antibiotics tested are summarized in Table 1. A: Cephaloridine. B: Colistin. C: Ampicillin. D: Chloramphenicol. E: Nalidixic acid.

1

Examples of cps-lacZ induction by antibiotics. E. coli FSSG02-1K was plated and tested for antibiotic sensitivity and expression of β-galactosidase by disk diffusion assay. Representative reactions are shown. β-galactosidase acting on the substrate X-gal in the medium produced blue rings at the periphery of zones of inhibition. Results for all antibiotics tested are summarized in Table 1. A: Cephaloridine. B: Colistin. C: Ampicillin. D: Chloramphenicol. E: Nalidixic acid.

1

Antibiotic induction of cps-lacZ in E. coli FSSG02-1K

Antibiotic (µg/disk) Zone (mm) cps induction 
β-Lactams (PBP inhibitors) 
ampicillin 10 23 − 
ampicillin/sulbactam 20 23 − 
carbenicillin 100 26 ++ 
cefotetan 30 32 ++ 
ceftazidime 30 31 − 
ceftriaxone 30 33 − 
cephaloridine 300 19 ++++ 
cephalothin 30 20 ++ 
oxacillin − 
penicillin G 10 − 
ticarcillin 150 36 ++ 
ticarcillin/clavulinate 85/85 26 
Other cell wall inhibitor 
vancomycin 30 
Cell membrane disruption 
colistin (polymyxin E) 10 15 ++++ 
Protein synthesis inhibitors 
chloramphenicol 30 18 − 
erythromycin 15 − 
gentamicin 10 24 − 
kanamycin 30 17 − 
neomycin 30 11 − 
streptomycin 10 − 
tetracycline 30 19 − 
DNA synthesis inhibitors 
ciprofloxacin 40 − 
nalidixic acid 30 21 − 
sulfisoxazole 250 16 − 
novobiocin − 
Multiple modes of action 
nitrofurantoin 300 22 − 
Antibiotic (µg/disk) Zone (mm) cps induction 
β-Lactams (PBP inhibitors) 
ampicillin 10 23 − 
ampicillin/sulbactam 20 23 − 
carbenicillin 100 26 ++ 
cefotetan 30 32 ++ 
ceftazidime 30 31 − 
ceftriaxone 30 33 − 
cephaloridine 300 19 ++++ 
cephalothin 30 20 ++ 
oxacillin − 
penicillin G 10 − 
ticarcillin 150 36 ++ 
ticarcillin/clavulinate 85/85 26 
Other cell wall inhibitor 
vancomycin 30 
Cell membrane disruption 
colistin (polymyxin E) 10 15 ++++ 
Protein synthesis inhibitors 
chloramphenicol 30 18 − 
erythromycin 15 − 
gentamicin 10 24 − 
kanamycin 30 17 − 
neomycin 30 11 − 
streptomycin 10 − 
tetracycline 30 19 − 
DNA synthesis inhibitors 
ciprofloxacin 40 − 
nalidixic acid 30 21 − 
sulfisoxazole 250 16 − 
novobiocin − 
Multiple modes of action 
nitrofurantoin 300 22 − 

Diameter of zones of inhibition, measured across disks of 7 mm diameter. Antibiotics that did not inhibit growth of the bacterial lawn were assigned a value of 0 mm.

β-Galactosidase production: (−) none; (+) weak (a pale blue ring at the periphery of the zone of inhibition); (++) intermediate; (++++) high (a deep blue ring at the periphery of the zone of inhibition).

Concentration of solution used to saturate paper disks.

Increased resistance to neomycin and kanamycin was due to the neo gene inserted into the ampC gene in strain FSSG02-1K.

Most antibiotics did not induce capsule synthesis, indicating that cps induction was not simply the result of a generalized stress response accompanying antibiotic-induced cell death or slow growth. In particular, agents inhibiting protein or DNA synthesis were ineffective (Table 1). It was no surprise that colistin initiated cps expression (Fig. 1B) because this antibiotic disrupts the cytoplasmic membrane [18], and membrane perturbations are known to activate colanic acid synthesis [19]. The only other agents to induce cps expression were β-lactams, which inhibit the final stages of peptidoglycan synthesis (Table 1). Surprisingly, some β-lactams were effective inducers (carbenicillin, cefotetan, cephaloridine, cephalothin and ticarcillin), while others did not induce any cps expression (ampicillin, ceftazidime, ceftriaxone, oxacillin and penicillin G) (Fig. 1 and Table 1). Thus, colanic acid expression was not a simple, generalized response to inhibition of peptidoglycan synthesis.

The qualitative results from the disk diffusion assay were confirmed by quantitative assays of cps gene induction in broth cultures (Table 2). Cephaloridine was by far the most effective inducer, the extremely high level of β-galactosidase corresponding to the intense blue halo observed in the disk diffusion assay. Carbenicillin and cephalothin triggered much less expression, mirroring the light blue rings exhibited around those antibiotics in the disk assay. Of the other antibiotics tested, none induced a significant level of gene expression (Table 2). Strikingly, the cps genes were induced only at a single antibiotic concentration just below the MIC for each antibiotic.

2

Antibiotic induction of cps-lacZ in E. coli SG20781

β-Lactam (µg ml−1(Miller Units) 
Cephaloridine 0.90 1600.0 
 0.45 1.2 
 0.22 
Carbenicillin 7.5 83.0 
 3.75 
Cephalothin 3.75 27.0 
Cloxacillin 125.0 5.7 
 62.5 1.1 
 31.2 
Phenethillin 125.0 5.8 
 62.5 3.4 
 31.2 
Penicillin G 15.0 6.4 
 7.5 
 3.75 
Ticarcillin 3.75 1.4 
Ampicillin 3.75 
Azlocillin 3.75 
β-Lactam (µg ml−1(Miller Units) 
Cephaloridine 0.90 1600.0 
 0.45 1.2 
 0.22 
Carbenicillin 7.5 83.0 
 3.75 
Cephalothin 3.75 27.0 
Cloxacillin 125.0 5.7 
 62.5 1.1 
 31.2 
Phenethillin 125.0 5.8 
 62.5 3.4 
 31.2 
Penicillin G 15.0 6.4 
 7.5 
 3.75 
Ticarcillin 3.75 1.4 
Ampicillin 3.75 
Azlocillin 3.75 

E. coli SG20781 was incubated in the presence of two-fold dilutions of the indicated antibiotics. Only those concentrations at or just below the MIC are reported in the table. The β-galactosidase activity in all other concentrations was negative. The results for E. coli FSSG02-1K (SG20781 ΔampC) were qualitatively the same, though the amounts of β-galactosidase produced after exposure to cephaloridine (350 units) and cephalothin (14 units) were slightly lower than observed for strain SG20781, due to the higher effective antibiotic concentration in the absence of the AmpC β-lactamase protein.

β-Galactosidase activities are reported as Miller Units (two significant figures only) [17].

Among the β-lactams, even small structural differences influenced expression of the cps genes (Fig. 2). Although carbenicillin (Fig. 2C) induced cps expression, two closely related β-lactams, penicillin G (Fig. 2A) and ampicillin (Fig. 2B), did not. These three compounds differ at a single position, suggesting that the carboxyl substitution in carbenicillin may be important (Fig. 2C). In addition, ticarcillin, which contains an identically placed carboxyl group, also induced the colanic acid genes, although a sulfur-containing five-member ring replaces the benzyl ring of penicillin G (Fig. 2D). However, a carboxyl moiety at this position is not absolutely required because both cephaloridine and cephalothin induced gene expression (Table 1), even though both have hydrogen in place of a carboxyl group at the same position (Fig. 2E).

2

Side groups of selected antibiotics and effects on cps-lacZ induction. A single asterisk (*) indicates that the side group is attached to a penicillin β-lactam core; a double asterisk (**) indicates that the side group is attached to a cephalosporin β-lactam core. Dashed circles denote differences from penicillin G. R-groups attached to the other side of the β-lactam core are not shown for cephaloridine, cephalothin or ticarcillin. A: Penicillin G. B: Ampicillin. C: Carbenicillin. D: Ticarcillin. E: Cephaloridine and cephalothin.

2

Side groups of selected antibiotics and effects on cps-lacZ induction. A single asterisk (*) indicates that the side group is attached to a penicillin β-lactam core; a double asterisk (**) indicates that the side group is attached to a cephalosporin β-lactam core. Dashed circles denote differences from penicillin G. R-groups attached to the other side of the β-lactam core are not shown for cephaloridine, cephalothin or ticarcillin. A: Penicillin G. B: Ampicillin. C: Carbenicillin. D: Ticarcillin. E: Cephaloridine and cephalothin.

Although some of the physiological triggers for cps induction are known (osmotic shock, dehydration and destabilization of the outer membrane), the signal recognized by the sensor kinase RcsC remains unidentified. The results reported here suggest that the action of particular β-lactams creates just such a signal. Because β-lactams inhibit PBP transpeptidase activity, one possibility is that the signaling compound may be some unusual fragment of peptidoglycan. If so, the difference between inducing and non-inducing β-lactams must be in their targets or effects. Interestingly, two antibiotics with high specificities for PBP 1a (cephaloridine and cephalothin) [20] induced the cps genes, suggesting that the signal might be generated by inactivating this enzyme. However, deleting PBP 1a did not result in expression of the cps-lacZ fusion (data not shown), indicating that the absence of PBP 1a alone is not responsible for inducing colanic acid synthesis. A second possibility is that the inducing β-lactams might damage the outer membrane directly or indirectly as a consequence of inhibiting peptidoglycan crosslinking. If so, then membrane damage would constitute the common signal for cps induction. Finally, a third possibility is that specific β-lactams might induce colanic acid gene expression by a mechanism not connected with their interactions with PBPs. For example, particular β-lactams could bind a separate receptor, which, in turn, might generate the signal for cps induction.

The discovery that several β-lactams induce expression of capsular polysaccharide genes in E. coli is similar to the observation that specific antibiotics induce expression of the polysaccharide adhesin Ica proteins, which mediate biofilm formation in Staphylococcus epidermidis[21]. In this organism, tetracycline and a semisynthetic streptogramin induce expression of the ica operon but several other antibiotics do not, including two β-lactams and vancomycin. The effect in S. epidermidis occurs at two or three antibiotic concentrations just below their MICs, a slightly larger window of activating concentrations than we report here. Most important, though, is that S. epidermidis biofilm formation is enhanced by subinhibitory concentrations of these antibiotics [21], providing a practical link between gene induction and bacterial behavior.

This induction of exopolysaccharides by specific classes of antibiotics bears close attention. In particular, administering agents that increase the levels of colanic acid in E. coli could exacerbate the maturation of biofilms prevalent in catheters, endotracheal tubes, implants and other medical devices [9,22–24]. It would be unfortunate if the antimicrobials used to treat infections also facilitated biofilm formation, thus protecting microbes from these or other agents.

Acknowledgments

This work was supported by Grant GM61019 from the National Institutes of Health. We thank Susan Gottesman for providing E. coli SG20781 and David Bradley for photographic assistance.

References

[1]
Sledjeski
D.
Gottesman
S.
(
1996
)
Osmotic shock induction of capsule synthesis in Escherichia coli K-12
.
J. Bacteriol.
 
178
,
1204
1206
.
[2]
Gottesman
S.
Stout
V.
(
1991
)
Regulation of capsular polysaccharide synthesis in Escherichia coli K12
.
Mol. Microbiol.
 
5
,
1599
1606
.
[3]
Ebel
W.
Trempy
J.
(
1999
)
Escherichia coli RcsA, a positive activator of colanic acid capsular polysaccharide synthesis, functions to activate its own expression
.
J. Bacteriol.
 
181
,
577
584
.
[4]
Stout
V.
(
1996
)
Identification of the promoter region for the colanic acid polysaccharide biosynthetic genes in Escherichia coli K-12
.
J. Bacteriol.
 
178
,
4273
4280
.
[5]
Stout
V.
Gottesman
S.
(
1990
)
RcsB and RcsC: a two-component regulator of capsule synthesis in Escherichia coli
.
J. Bacteriol.
 
172
,
659
669
.
[6]
Gervais
F.
Drapeau
G.
(
1992
)
Identification, cloning, and characterization of rcsF, a new regulator gene for exopolysaccharide synthesis that suppresses the division mutation ftsZ84 in Escherichia coli K-12
.
J. Bacteriol.
 
174
,
8016
8022
.
[7]
Gottesman
S.
(
1995
)
Regulation of capsule synthesis: modification of the two-component paradigm by an accessory unstable regulator
. In:
Two-Component Signal Transduction
  (
Hoch
J.A.
Silhavy
T.J.
, Eds.).
American Society for Microbiology
,
Washington, DC
.
[8]
Whitfield
C.
Roberts
I.S.
(
1999
)
Structure, assembly, and regulation of expression of capsules in Escherichia coli
.
Mol. Microbiol.
 
31
,
1307
1319
.
[9]
Prigent-Combaret
C.
Prensier
G.
Le Thi
T.T.
Vidal
O.
Lejeune
P.
Dorel
C.
(
2000
)
Developmental pathway for biofilm formation in curli-producing Escherichia coli strains; role of flagella, curli and colanic acid
.
Environ. Microbiol.
 
4
,
450
464
.
[10]
Danese
P.N.
Pratt
L.A.
Kolter
R.
(
2000
)
Exopolysaccharide production is required for development of Escherichia coli K-12 biofilm architecture
.
J. Bacteriol.
 
182
,
3593
3596
.
[11]
Watnick
P.
Kolter
R.
(
2000
)
Biofilm, city of microbes
.
J. Bacteriol.
 
182
,
2675
2679
.
[12]
Denome
S.A.
Elf
P.K.
Henderson
T.A.
Nelson
D.E.
Young
K.D.
(
1999
)
Escherichia coli mutants lacking all possible combinations of eight penicillin binding proteins: viability, characteristics, and implications for peptidoglycan synthesis
.
J. Bacteriol.
 
181
,
3981
3993
.
[13]
Meberg
B.M.
Sailer
F.C.
Nelson
D.E.
Young
K.D.
(
2001
)
Reconstruction of Escherichia coli mrcA (PBP 1a) mutants lacking multiple combinations of penicillin binding proteins
.
J. Bacteriol.
 
183
,
6148
6149
.
[14]
Young
K.D.
(
2001
)
Approaching the physiological functions of penicillin-binding proteins in Escherichia coli
.
Biochimie
 
83
,
99
102
.
[15]
Miller
J.H.
(
1972
).
Experiments in Molecular Genetics
 .
Cold Spring Harbor Laboratory
,
Cold Spring Harbor, NY
.
[16]
Baron
E.
Finegold
S.
(
1990
)
Methods for testing antimicrobial effectiveness
. In:
Bailey and Scott's Diagnostic Microbiology
 ,
8
th edn., pp.
172
173
.
C.V. Mosby
,
St. Louis, MO
.
[17]
Miller
J.H.
(
1992
).
A Short Course in Bacterial Genetics: a Laboratory Manual and Handbook for Escherichia coli and Related Bacteria
 .
Cold Spring Harbor Laboratory Press
,
Plainview, NY
.
[18]
Storm
D.R.
Rosenthal
K.S.
Swanson
P.E.
(
1977
)
Polymyxin and related peptide antibiotics
.
Annu. Rev. Biochem.
 
46
,
723
763
.
[19]
Ebel
W.
Vaughn
G.J.
Peters
H.K.
III
Trempy
J.E.
(
1997
)
Inactivation of mdoH leads to increased expression of colanic acid capsular polysaccharide in Escherichia coli
.
J. Bacteriol.
 
179
,
6858
6861
.
[20]
Tomasz
A.
(
1983
)
Mode of action of β-lactam antibiotics — a microbiologist's view
. In:
Antibiotics Containing the beta-Lactam Structure
  (
Demain
A.L.
Solomon
N.A.
, Eds.), vol.
I
, p.
15
96
.
Springer
,
Berlin
.
[21]
Rachid
S.
Ohlsen
K.
Witte
W.
Hacker
J.
Ziebuhr
W.
(
2000
)
Effect of subinhibitory antibiotic concentrations on polysaccharide intercellular adhesin expression in biofilm-forming Staphylococcus epidermidis
.
Antimicrob. Agents Chemother.
 
44
,
3357
3363
.
[22]
Tsukamoto
T.
Matsukawa
M.
Sano
M.
Takahashi
S.
Hotta
H.
Itoh
N.
Hirose
T.
Kumamoto
Y.
(
1999
)
Biofilm in complicated urinary tract infection
.
Int. J. Antimicrob. Agents
 
11
,
233
236
.
[23]
Feldman
C.
Kassel
M.
Cantrell
J.
Kaka
S.
Morar
R.
Goolam-Mohamed
A.
Philips
J.
(
1999
)
The presence and sequence of endotracheal colonization in patients undergoing mechanical ventilation
.
Eur. Respir. J.
 
13
,
546
551
.
[24]
Tunney
M.M.
Patrick
S.
Curran
M.D.
Ramage
G.
Hanna
D.
Nixon
J.R.
Gorman
S.P.
Davis
R.I.
Anderson
N.
(
1999
)
Detection of prosthetic hip infection at revision arthroplastry by immunofluorescence microscopy and PCR amplification of the bacterial 16S RNA gene
.
J. Clin. Microbiol.
 
10
,
3281
3290
.