Antibiotic stress induces a large amount of outer membraneprotein in Pseudomonas aeruginosa

Akira Nakajima, Mie Hoshikawa, Taiji Nakae *Department of Molecular Life Science, Tokai University School of Medicine, Isehara 259-1193, Japan

Received 17 June 1998; accepted 29 June 1998

Abstract

To investigate the bacterial response to antibiotic stress, we analyzed the outer membrane proteins of Pseudomonasaeruginosa grown in the presence of a sub-minimum inhibitory concentration of antibiotics. Among the antibiotics tested,fluoroquinolones and streptonigrin induced a large amount of outer membrane protein with a molecular mass of 43 kDa. Thisprotein is most likely the stress-responsive protein, since the quinolone-resistant mutants with a higher minimum inhibitoryconcentration of antibiotic than the wild-type strain produced a large amount of 43-kDa protein only in the presence of sub-minimum inhibitory concentration of the mutants itself, but not that of the antibiotic-susceptible wild-type strain. Thesequence of N-terminal 15 amino acids of the 43-kDa protein was identical to that of pyocin R1. However, purified pyocin R1failed to accumulate in the outer membrane. Thus, we concluded that the 43-kDa protein (pyocin R1) is the antibiotic-stress-induced outer membrane protein. z 1998 Federation of European Microbiological Societies. Published by Elsevier ScienceB.V. All rights reserved.

Keywords: Stress; Antibiotic ; Membrane protein; Pyocin

1. Introduction

Pseudomonas aeruginosa often infects patientswhose immune activity is lowered and therefore theorganism is a problematic pathogen in nosocomialinfections [1]. A major problem in P. aeruginosa in-fection may be that this pathogen exhibits a highdegree of resistance to a broad spectrum of chemo-therapeutic agents [1]. When the bacteria were ex-posed to antibiotics, they responded to the antibioticstresses by inducing antibiotic-modifying enzyme(s)[2], lowering the antibiotic sensitivity of the targets

[3], lowering the penetration rate of antibiotics bythe loss of the antibiotic-permeable porin [4,5] andpumping out antibiotics by active extrusion systems[6]. These cellular responses to antibiotics stressesoften accompanied mutational changes due to strongselective pressure. However, the inducible responseto antibiotic stress is not well understood exceptfor the induction of L-lactamase in the presence ofL-lactams [2]. We are interested in the bacterial re-sponse to the action of a sub-minimum inhibitoryconcentration (MIC) of antibiotics without accom-panying mutation(s). We analyzed membrane pro-teins of cells grown under antibiotic stress in an anal-ogy to the OprM, OprJ and OprN expression in thenalB, nfxB and nfxC type antibiotic-resistant e¥ux

0378-1097 / 98 / $19.00 ß 1998 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved.PII: S 0 3 7 8 - 1 0 9 7 ( 9 8 ) 0 0 2 8 7 - 0

FEMSLE 8293 12-8-98

* Corresponding author. Tel. : +81 (463) 93-5436;Fax: +81 (463) 93-5437; E-mail: nakae@is.icc.u-tokai.ac.jp

FEMS Microbiology Letters 165 (1998) 261^265

mutants [6,7]. We found that P. aeruginosa producesa large amount of pyocin-like outer membrane pro-tein in the presence of sub-MICs of £uoroquino-lones.

2. Materials and methods

2.1. Bacterial strains and the culture conditions

PAO1 [7] is the wild-type strain of P. aeruginosa.OCR1 [7] and Mutant10 [3] are £uoroquinolone-re-sistant strains with impaired nalB and gyrA genes,respectively. Bacterial cells were grown in 10 ml ofL-broth overnight at 37³C with shaking and the cul-ture medium was diluted with 190 ml of prewarmedfresh medium, then incubated at 37³C. Antibiotic atan appropriate concentration was added whenneeded. Cells were harvested by centrifugation at7500 rpm for 10 min and washed with 10 ml of 10mM Tris-HCl bu¡er, pH 8.0, by centrifugation.

2.2. Preparation of the inner and outer membranefractions

Cells were suspended in 10 ml of Tris-HCl bu¡er,pH 8.0 and passed through a French pressure cell at12 000 kg cm32. The cell lysate was centrifuged at45 000 rpm for 60 min in a Beckman L-70 centrifugeusing a 55.2Ti rotor after removing unbroken cellsby centrifugation at 10 000 rpm for 10 min. Pelletswere washed once with 10 mM Tris-HCl, pH 8.0,containing 0.3 M NaCl. The centrifuged pelletswere suspended in 2 ml of 1% N-laurylsarcosine,left to stand at 30³C for 30 min, then centrifugedat 45 000 rpm for 60 min [8]. The pellets were sus-pended in a small amount of deionized water. Solu-ble and insoluble fractions were used as the innerand outer membrane fractions, respectively. Themembrane fractions were also obtained by treatmentof the cells by EDTA and lysozyme as reported ear-lier [9].

2.3. Puri¢cation of the 43-kDa outer membraneprotein (Opr43)

The laurylsarcosine-insoluble fraction obtainedfrom cells grown in the presence of a 1/2 MIC of

o£oxacin for 5 h was successively washed with 1%of octyl-polyoxyethylene in 10 mM Tris-HCl, pH 8.0and 1% of sodium dodecylsulfate (SDS) containing0.5 M NaCl in 10 mM Tris-HCl, pH 8.0. The in-soluble materials were extracted with Laemm-li's sample bu¡er [10]. Opr43 to be used for immu-nization was further puri¢ed by preparative SDS-polyacrylamide gel electrophoresis. A proteinband corresponding to Opr43 was electro-elut-ed in the electrode bu¡er. SDS was removed bydialysis.

2.4. Puri¢cation of pyocin

Culture supernatant of strain PAO1 in the pres-ence a 1/2 MIC of o£oxacin was treated by 70%saturation of ammonium sulfate at 4³C overnightand centrifuged at 12 000 rpm for 15 min. After ex-tensive dialysis, the protein was subjected to DEAE-HPLC chromatography in 10 mM Tris-HCl, andeluted with a 0^1 M NaCl linear gradient.

2.5. Other methods

Antibody against Opr43 was raised by injectinghighly puri¢ed Opr43 (250 mg) in Freund adjuvantinto 2.5-kg Japanese white rabbits once a week forfour consecutive weeks. Acrylamide gel electropho-resis was carried out by the method of Laemmli [10]or that of Lugtenberg et al. [11]. The amino-terminalamino acid sequence was determined with anABI473A protein sequencer. Protein was quanti¢edby the method of Lowry et al. [12].

3. Results

3.1. Induction of a new outer membrane protein in thepresence of antibiotics

The laboratory strain of P. aeruginosa PAO1 wasgrown in the presence of a 1/2 MIC of levo£oxacin,o£oxacin, tetracycline, chloramphenicol, gentamicin,carbenicillin, piperacillin or rifampicin. As seen inFig. 1, a large amount of the outer membrane pro-tein with a molecular mass of 43 kDa was expressedin the presence of £uoroquinolone antibiotics. Theouter membrane protein pro¢les in the cells grown

FEMSLE 8293 12-8-98

A. Nakajima et al. / FEMS Microbiology Letters 165 (1998) 261^265262

in the presence of each of the antibiotics tested werecomparable with that of cells not exposed to anti-biotics. Since all the £uoroquinolones tested andstreptonigrin (data not shown) caused overexpres-sion of Opr43, it is likely that the induction of thisprotein is a stress response to the disturbed DNAmetabolism.

To test the kinetics of Opr43 induction, freshlydiluted cells were grown for 0, 0.5, 1, 2, and 4 h inthe presence of a 1/2 MIC of o£oxacin and the out-er membrane protein was analyzed. As seen in Fig. 2,a large amount of Opr43 protein was inducedat 120 min and more at 240 min, suggesting thatthe protein was inserted into the membrane as thecell grew.

3.2. Experiments to distinguish whether or not the cellproduces Opr43 sensing antibiotic concentrationor the stress exerted by the antibiotics

The wild-type strain was exposed to a 1/20, 1/10,1/5, 1/3, or 1/2 MIC of o£oxacin and the expressionof Opr43 was visualized by the Western blottingmethod. Cells that were unexposed to o£oxacin pro-duced a small amount of Opr43 and those exposedto a 1/5 MIC to 1/2 MIC of o£oxacin producedincreasing amounts of the protein (Fig. 3). This ex-periment, however, did not distinguish whether thecell senses the antibiotic concentration or the stressexerted by o£oxacin. To ascertain this issue, wetested the production of Opr43 in the strains exhib-iting di¡erent susceptibility to £uoroquinolones.Strain OCR1 and Mutant10 are nalB and gyrA mu-tants, respectively, exhibiting MICs of levo£oxacinof 1.56 Wg ml31 and 12.5 Wg ml31, respectively,whereas that of levo£oxacin in their respective par-ent strains was 0.39 Wg ml31. OCR1 and Mutant10were grown in both media containing 0.78 Wg ml31

and 6.25 Wg ml31 of levo£oxacin, respectively. If thebacteria sense the antibiotic concentration and re-spond to it, Opr43 should be produced in all thestrains at 0.2 Wg ml31. On the other hand, if thebacteria sense the antibiotic stress, levo£oxacin-re-sistant strains should express a large amount ofOpr43 only at a 1/2 MIC of levo£oxacin in thestrain, but less at a 1/2 MIC in the wild-type strain.As seen in Fig. 4, both nalB and gyrA mutants ex-pressed large amounts of Opr43 in the presence of0.78 and 6.25 Wg ml31 of levo£oxacin, respectively,

FEMSLE 8293 12-8-98

Fig. 3. E¡ect of o£oxacin concentration on the induction ofOpr43. An overnight culture of strain PAO1 was diluted with a9-fold volume of fresh medium containing di¡erent concentra-tions of o£oxacin then shaken at 37³C for 5 h. The laurylsarco-sine-insoluble fraction was subjected to electrophoresis byLaemmli's bu¡er system. Proteins were stained by the Westernblotting method as above. Lane 1, without antibiotics ; 2, 0.039Wg ml31 ; 3, 0.078 Wg ml31 ; 4, 0.156 Wg ml31 ; 5, 0.26 Wg ml31 ;6, 0.39 Wg ml31. The MIC of o£oxacin in PAO1 was 0.78 Wgml31.

Fig. 2. Time course of Opr43 expression. The overnight-grownculture of strain PAO1 was diluted with 9 volumes of a freshmedium containing 0.39 Wg ml31 of o£oxacin. Cells were har-vested at the desired time and the outer membrane fraction sub-jected to electrophoresis. Proteins were electroblotted to PVDFmembrane and stained with rabbit antibody raised against Opr43and anti-rabbit IgG-conjugated secondary antibody. Lane 1, 0min; 2, 30 min; 3, 60 min; 4, 120 min; 5, 240 min.

Fig. 1. Electrophoretogram of the outer membrane fraction of P.aeruginosa PAO1 grown in the presence of a 1/2 MIC of antibi-otics. The laurylsarcosine-insoluble fraction was suspended insample bu¡er of Laemmli's electrophoresis system, heated at95³C then 40 mg of protein was applied. Lane 1, without antibi-otics ; 2, levo£oxacin; 3, o£oxacin; 4, tetracycline; 5, chloram-phenicol ; 6, molecular mass marker; 7, without antibiotic ; 8,gentamicin; 9, carbenicillin; 10, piperacillin; 11, rifampicin.

A. Nakajima et al. / FEMS Microbiology Letters 165 (1998) 261^265 263

but not at 0.2 Wg ml31. These results clearly demon-strated that Opr43 is a stress-responsive protein.

To rule out the possibility that the induction ofOpr43 is due to a mutation(s), we tested a reset ofOpr43 expression. Opr43 was induced in the pres-ence of a 1/2 MIC of levo£oxacin for 5 h, dilutedwith a 20-fold volume of fresh medium without anti-biotics and incubated for an additional 5-h period.Opr43 was induced at a 1/2 MIC of levo£oxacin andthe amount of protein in the cells diluted with a 20-fold volume of medium without levo£oxacin pro-duced only a wild-type level of Opr43 (data notshown). Thus, the production of Opr43 appears tobe inducible.

3.3. Characterization of Opr43

Opr43 traveled to the position where OprE1, E2and E3 moved electrophoretically by Laemmli's sys-tem. To determine whether or not Opr43 is identicalto these outer membrane proteins, we carried outelectrophoresis by Lugtenberg's bu¡er system, whichcan separate these proteins. The outer membranefraction was prepared from the mutant lacking

OprE1 [13] grown in the presence of a 1/2 MIC ofo£oxacin then subjected to electrophoresis. Wefound the Opr43 band at a position distinguishablefrom the location of OprE2 and OprE3 (data notshown). Thus, Opr43 is a stress-induced novel outermembrane protein. However, this result does notrule out the possibility that Opr43 is a truncatedprotein derived from OprE1, E2 or E3. To clarifythis ambiguity, we analyzed the N-terminal aminoacid sequence of the highly puri¢ed 43-kDa protein.The sequence was S-F-F-H-G-V-T-N-V-D-I-G-A,which is di¡erent from the sequence in OprE1, E2or E3. A computer-aided amino acid similaritysearch of the GenBank database revealed that thesequence was identical to that of pyocin R1 [14].

3.4. Opr43 was not the rebound form of themedium-secreted pyocin R1

Since the N-terminal amino acid sequence ofOpr43 matched that of pyocin R1, one may arguethat Opr43 was the medium-secreted pyocin R1bound to the outer membrane. To investigate thispossibility, the wild-type strain PAO1 was culturedin fresh medium containing puri¢ed pyocin R1 andthe outer membrane fraction was prepared by theEDTA-lysozyme method. We did not ¢nd signi¢cantlevels of Opr43 bound to the outer membrane. Thus,it is unlikely that Opr43 is membrane-rebound pyo-cin R1.

4. Discussion

To investigate the bacterial response to antibioticstress, we analyzed outer membrane proteins in P.aeruginosa, and found that a large amount of Opr43was induced in the presence of a 1/5 to 1/2 MIC of£uoroquinolones. The localization of Opr43 in theouter membrane was con¢rmed both by determiningthe association of the protein with the highly puri¢edouter membrane fraction and by the well-establishedsurfactant extraction method. To characterize thisnovel outer membrane protein, we determined theN-terminal amino acid sequence and, to our surprise,found that 15 amino acid residues were identicalwith that of pyocin R1. It was reported earlier thatUV irradiation or mitomycin treatment of P. aerugi-

FEMSLE 8293 12-8-98

Fig. 4. Electrophoretogram of the outer membrane fractions ofthe wild-type strain and antibiotic-resistant mutants. Overnightculture medium of P. aeruginosa was diluted with a 20-fold vol-ume of fresh medium into two or three separate £asks containinga 1/2 MIC equivalent of levo£oxacin for the wild-type strain orthat for the antibiotic-resistant strains, then shaken at 37³C for5 additional hours. The outer membrane fraction was preparedby the laurylsarcosine extraction method and was subjected tothe electrophoresis. Lane 1, molecular mass marker; 2, PAO1without levo£oxacin; 3, PAO1 with 0.2 Wg ml31 of levo£oxacin;4, Mutant10 without levo£oxacin; 5, Mutant10 with 0.2 Wg ml31

of levo£oxacin; 6, Mutant10 with 6.25 Wg ml31 of levo£oxacin;7, OCR1 without levo£oxacin; 8, OCR1 with 0.2 Wg ml31 oflevo£oxacin; 9, OCR1 with 0.78 Wg ml31 of levo£oxacin.

A. Nakajima et al. / FEMS Microbiology Letters 165 (1998) 261^265264

nosa induced pyocin R1 in the culture medium [15].However, induction of a large amount of pyocin R1in the presence of £uoroquinolones in the outermembrane has not been reported. If we assumethat Opr43 was the same molecule as pyocin R1 inthe culture medium, how and why a large amount ofthe water-soluble molecule is located in the outermembrane remain unclear. The following possibil-ities are conceivable. (i) Opr43 is a secretion inter-mediate of pyocin R1. (ii) Pyocin R1 was adsorbedfrom the culture medium. (iii) Opr43 has an un-known physiological function. The ¢rst possibilityis unlikely, since the outer membrane-associatedOpr43 cannot be chased out in the antibiotic-freemedium (data not shown). Accumulation of Opr43seems to require cell growth. The second possibilityis less likely, since the pyocin R1 puri¢ed from theculture medium failed to readsorb to the cell. Thethird possibility seems most likely. Lines of evidenceto support this possibility are as follows. (i) Opr43was induced by antibiotic stress, but not in the pres-ence of a less harmful concentration of £uoroquino-lones. Removal of the antibiotic from the mediumterminated the Opr43 production. (ii) A smallamount of Opr43 was expressed in the cell grownin the antibiotic-free medium. Since the P. aeruginosachromosome retains the gene encoding Opr43, it islikely that the physiological role of Opr43 is yet to beelucidated.

This is the ¢rst report to our knowledge that anti-biotic stress induces a large amount of outer mem-brane protein. Since we observed simultaneousinduction of the 20-kDa and 37-kDa inner mem-brane proteins with the induction of Opr43 in thepresence of sub-MIC of £uoroquinolones, it is ofinterest to link the functions of these proteins.

Acknowledgments

This study was supported in part by grants fromthe Ministry of Education of Japan, Culture, Sportand Science, the Ministry of Health and Welfare andthe Japan Society for Promotion of Science. A.N. isthe recipient of a Tokai University Scholarship. Theauthors thank S. Kuramitsu and K. Sato for proteinsequence analysis and for the gift of strains, respec-tively.

References

[1] Brown, M.R.W. (1975) In: Resistance of Pseudomonas aeru-ginosa (Brown, M.R.W., Ed.), pp. 71^107. John Wiley andSons, London.

[2] Livermore, D.M. and Young, Y.J. (1987) L-Lactamase abilityand inducer power of newer L-lactam antibiotics in relation totheir activity against L-lactamase-inducibility mutant of Pseu-domonas aeruginosa. J. Infect. Dis. 155, 755^782.

[3] Inoue, Y., Sato, K., Fujii, T., Hirai, K., Inoue, M., Iyobe, S.and Mitsuhasi, S. (1987) Same properties of subunits of DNAgyrase from Pseudomonas aeruginosa PAO and its nalidixicacid resistant mutant. J. Bacteriol. 169, 2322^2325.

[4] Quinn, J.P., Dudek, E.J., diVincenzo, C.A., Lucks, D.A. andLerner, S.A. (1986) Emergence of resistance to imipenem dur-ing therapy for Pseudomonas aeruginosa infection. J. Infect.Dis. 154, 289^294.

[5] Bucher, K.H., Cullmann, W., Dick, W. and Opterkuch, W.(1987) Imipenem resistance in Pseudomonas aeruginosa result-ing from diminished expression of an outer membrane pro-tein. Antimicrob. Agents Chemother. 31, 703^708.

[6] Lei, Y., Sato, K. and Nakae, T. (1991) O£oxacin resistantPseudomonas aeruginosa mutants with elevated drug extrusionacross the inner membrane. Biochem. Biophys. Res. Commun.178, 1043^1048.

[7] Masuda, N., Sakagawa, E. and Oya, S. (1995) Outermembrane protein for multiple drug resistance in Pseudo-monas aeruginosa. Antimicrob. Agents Chemother. 39, 645^649.

[8] Komatsu, Y., Murakami, K. and Nishikawa, T. (1981) Pene-tration of moxalactam into its target proteins in Escherichiacoli K-12: comparison of a highly moxalactam-resistant mu-tant with its parent strain. Antimicrob. Agents Chemother. 20,613^619.

[9] Mizuno, T. and Kageyama, M. (1978) Separation and char-acterization of the outer membrane of Pseudomonas aerugino-sa. J. Biochem. 84, 179^191.

[10] Laemmli, U.K. (1970) Cleavage of structural proteins duringthe assembly of the head of bacteriophage T4. Nature 227,680^685.

[11] Lugtenberg, B., Meijers, J., Peters, R., Hoek, P. and Alphen,L. (1975) Electrophoretic resolution of the `major outer mem-brane protein' of Escherichia coli K-12 into four bands. FEBSLett. 58.

[12] Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J.(1951) Protein measurement with the Folin phenol reagent.J. Biol. Chem. 193, 265^275.

[13] Yamano, Y. (1994) Studies on the Outer Membrane Perme-ability of Pseudomonas aeruginosa. Ph.D. Thesis, Kyoto Uni-versity, Kyoto.

[14] Sasaki, T., Shinomiya, T., Kumazaki, T., Mohri, N., Ishii,S.-i. and Arisaka, F. (1997) Nucleotide sequence of the con-tractile tail sheath and tube genes of bacteriophage PS17 andamino acid sequences of their products. Res. Commun. Bio-chem. Cell Mol. Biol. 1, 93^107

[15] Kageyama, M. (1964) Studies of pyocin. I. Physical and chem-ical properties. J. Biochem. 55, 49^53.

FEMSLE 8293 12-8-98

A. Nakajima et al. / FEMS Microbiology Letters 165 (1998) 261^265 265