Microbial Pathogenesis 1996; 20: 325–333

The role of the O-antigen lipopolysaccharide on thecolonization in vivo of the germfree chicken gut byAeromonas hydrophila serogroup O:34

Susana Merino,1 Xavier Rubires,1 Alicia Aguillar,1 Jean F. Guillot2

and Juan M. Tomas1∗

1Departamento de Microbiologıa. Facultad de Biologıa. Universidad de Barcelona.Diagonal 645, 08071 Barcelona, Spain, 2Station de Pathologie Aviare, Centre deRecherches de Tours, INRA, Nouzilly, 37380 Monnaie, France

(Received November 13, 1995; accepted in revised form January 16, 1996)

Merino, S. (Departamento de Microbiologıa. Facultad de Biologıa, Universidad de Barce-lona. Diagonal 645, 08071 Barcelona, Spain), X. Rubires, A. Aguillar, J. F. Guillot and J. M.Tomas. The role of the O-antigen lipopolysaccharide on the colonization in vivo of thegermfree chicken gut by Aeromonas hydrophila serogroup O:34. Microbial Pathogenesis1996; 20: 325–333.

We compared the ability of different Aeromonas hydrophila strains from serogroupO:34 grown at different temperatures to colonize in vivo the germfree chicken gut. Wefound a good colonization when the strains were grown at 20°C but not when they weregrown at 37°C. We previously described that these strains were able to form the O-antigenlipopolysaccharide (LPS) when they grow at low temperature but not at high temperature.We also obtained by transposon mutagenesis mutants only devoid of the O-antigen LPS(rfb mutants), and showed that they were unable to colonize the germfree chicken gut. Allthese results prompted us to conclude that the O-antigen LPS, in these strains, is a mainfactor for colonization in this animal model system. 1996 Academic Press Limited

Key words: Aeromonas hydrophila serogroup O:34; colonization in vivo; germfree chickengut.

Introduction

Motile Aeromonas species are ubiquitous inhabitants of the aquatic environmentand are also considered to be part of the normal microflora of the intestinal tractof different animals.1 Aeromonas hydrophila is both an opportunistic and a primarypathogen of a variety of aquatic and terrestrial animals, including humans;the clinical manifestations range from gastroenteritis to soft-tissue infections,including septicemia, and meningitis.2 Aeromonas strains have been serotypedon the basis of the O-antigen LPS,3 the polysaccharide chains in the smooth LPS,also known as the somatic antigen.

Serogroup O:34 strains of A. hydrophila have been recovered from moribundfish4,5 or from clinical specimens,6 being the single most common Aeromonasserogroup accounting for 26.4% of all clinical infections (mainly gastroenteritisand septicemia). Previous investigations have documented O:34 strains as animportant cause of infections in humans.7,8

∗Author to whom correspondence should be addressed.

0882–4010/96/060325+09 $18.00/0 1996 Academic Press Limited

S. Merino et al.326

1 2 3 4 5 6 7

Fig. 1. SDS-PAGE of LPS from A. hydrophila strains serogroup O:34. Lanes: 1, AH-405 grown at20°C; 2, AH-405 grown at 37°C; 3, AH-457 grown at 0°C; 4, AH-405 (LPS mutant with incomplete LPS-core12) grown at 20°C; 5, AH-406 grown at 20°C; 6, AH-406 grown at 37°C; 7, AH-459 grown at 20°C.

The ecosystem that we selected to study the role of the O-antigen LPS oncolonization in vivo was the germfree chicken gut, which is a well-controlledsystem that has been well studied previously.9 Furthermore, A. hydrophila is aregular inhabitant of the intestinal ecosystem of the chicken, whereas some strainsare able to cause acute diarrhoea in these birds, as well as to extraintestinalinfections on these animals.10 Then, chicken seems to be a possible Aeromonasfood contamination source for humans.

We describe here the isolation of LPS mutants by transposon mutagenesis, andshow that they were only devoid of the O:34-antigen LPS without changes in theLPS-core region. With these isogenic mutants we study the role of O-antigen LPSon the colonization of the germfree chicken gut.

Results

Spontaneous isogenic rifampicin resistant mutants from A. hydrophila strainsfrom serogroup O:34 occurred at a frequency of 1 to 3×10−7. These mutants (AH-405 and AH-406) showed identical surface properties as their wild-type strains.No changes in bacteriophage sensitivity were detected [both were sensitive tobacteriophages PM1 (a bacteriophage-specific for the O:34-antigen LPS11) or PM2(a bacteriophage-specific for the LPS-core12)]. No changes in the outer-membrane(OM) protein profile or LPS with cells grown at different temperatures wereobserved when we compared the wild-type strains with their isogenic rifampicinresistant mutants. As we previously described,13 these strains showed a smoothLPS (with O:34-antigen LPS) when they were grown at 20°C and a rough LPS(without O:34-antigen LPS) when grown at 37°C (Fig. 1).

In order to demonstrate the role of the O:34-antigen LPS, we isolated mutantsdevoid of the O:34-antigen LPS from strains AH-405 and AH-406 by Tn5 transposonmutagenesis and immunoscreening as described in Materials and methods.

O-antigen lipopolysaccharide colonization of germfree chicken gut 327

Table 1 Strains and plasmids used, their relevant characteristics and origin

Strain Relevant properties Origin

A. hydrophila strains from serogroup O:34AH-3 Smooth when grow at 20°C, but rough at 37°C. (13)Ba5 Smooth when grow at 20°C, but rough at 37°C. (13)AH-405 Rifampicin resistant mutant from AH-3 This studyAH-406 Rifampicin resistant mutant from Ba5 This studyAH-22 Isogenic rough mutant with complete LPS-core from AH,3 (11)AH-35 Isogenic rough mutant with incomplete LPS-core from AH-3 (12)AH-457 O:34− with complete LPS-core; Tn5 mutant derived from AH-405 This studyAH-459 O:34− with complete LPS-core; Tn5 mutant derived from AH-406 This study

E. coli strainsHB101 pro,leu,thi,lacY,strr,endoI−, recA− (19)S17-1 pro,thi,HsdR−,HsdM+, recA− (18)

PlasmidspRK2073 Self-transmissible plasmid with resistance to (19)

spectinomycin, streptomycinpSUP2021 Mobilizable transposon carrier plasmid Tn5 inserted with the (18)

mob genes from RP4

Table 2 Main chemical composition of purified LPS from A. hydrophila O:34strains

lmol/mg of LPS

Strain KDOa L-Heptoseb Glucoseb Hexosaminesb

AH-3 (20°C)c 0.018 0.31 1.57 0.41AH-3 (37°C) 0.029 0.99 1.67 0AH-457 (20°C) 0.040 0.98 1.69 0

Ba5 (20°C) 0.019 0.31 1.55 0.43Ba5 (37°C) 0.042 0.96 1.68 0AH-459 (37°C) 0.041 0.97 1.66 0

AH-22 (20°C) 0.038 0.96 1.64 0AH-35 (20°C) 0.051 1.25 1.32 0

aAssayed by colorimetric method as previously described by us.13

bAssayed by gas-liquid chromatography as described in Materials and methods.cGrowth temperature.Note: LPS from wild-type strains AH-3 and Ba5 is smooth when they grow at 20°C and rough when

they grow at 37°C. LPS from strain AH-457, AH-459 and AH-22 is rough, and LPS from strain AH-35is deep rough at both growth temperatures.

We obtained different mutant colonies, selecting strains AH-457 and AH-459(respectively from AH-405 and AH-406) as representatives of them. These mutantsshowed a rough LPS either when they were grown at 20°C or 37°C (Table 1). Theywere resistant to bacteriophage PM1 and sensitive to bacteriophage PM2, as ithappens with rough mutants with complete LPS-core.11 These Tn5 mutants alsoshowed identical LPS chemical composition as their wild-type strains grown at37°C (rough LPS) or LPS mutants with a complete LPS-core (Table 2). Furthermore,in LPS gels with Tricine buffer (in order to compare LPS-core oligosaccharides),

S. Merino et al.328

1 2 3 4 5 6 7

Fig. 2. Gel analysis of LPS-core oligosaccharides from A. hydrophila strains serogroup O:34. Lanes:1, AH-3 grown at 37°C; 2 and 6, AH-22 grown at 20°C; 3, AH-457 grown at 20°C; 4, AH-35 (LPS mutantand incomplete LPS-core12) grown at 20°C; 5, Ba5 grown at 37°C; 7, AH-459 grown at 20°C.

the Tn5 mutants showed the same relative electrophoretic mobility as their wild-type strains grown at 37°C (rough LPS) or LPS mutants with a complete LPS-core(Fig. 2).

All the strains grown at 20°C (with smooth or rough LPS) were agglutinated byan specific antibody for the fimbriae (antiserum kindly provided by M. Sato14).

In vivo studies

The experimental birds were inoculated on day 12 after hatching with 108–9 cellsof different A. hydrophila O:34 strains grown at different temperatures and theirisogenic Tn5 LPS mutants, and the number of A. hydrophila cells recovered fromthe faeces was counted. The birds inoculated with the wild-type strains grown at20°C had approximately equal bacterial densities in their faeces (108–9 cfu/g offaeces) throughout the experiment (days 12 to 30). However, when the birds wereinoculated with the same strains grown at 37°C, we found a great decrease ofthe number of cfu/g of faeces and were under our detection limit (102/g of faeces)in 2–3 days (Fig. 3). Identical results were obtained if we used the spontaneousrifampicin resistant mutants (AH-405 and AH-406) from the wild-type strains (AH-3 and Ba5, respectively) (Fig. 3).

In order to prove that these results were not a phenomenon temperaturedependent, we used the Tn5 LPS mutants (AH-457 and AH-459) grown at 20°C toinoculate the birds (always rough LPS, with a complete LPS-core). The resultsobtained are showed in Figure 4. The faeces of the inoculated birds with strainsAH-457 and AH-459 showed a great decrease of the number of cfu/g, and wereagain under the detection limit in two days.

Finally, at the end of the experiments, we remove the intestine of the birds. Theintestine of the colonize birds showed a large quantity of A. hydrophila (average

O-antigen lipopolysaccharide colonization of germfree chicken gut 329

302

Days

Log

cfu

/g o

f fa

eces

16

10

8

6

4

10 12 14 18 20 22

B

24 26 28

302

16

10

8

6

4

10 12 14 18 20 22 24 26 28

A

Fig. 3. Colonization in vivo of the germfree chicken gut by A. hydrophila strain AH-3 (Χ—Χ, A) andits isogenic rifampicin resistant mutant AH-405 (Ο—Ο, B) grown at 20°C, and the same strains AH-3(Ε—Ε, A) and AH-405 ( — , B) grown at 37°C. Similar results were obtained with strain Ba5 andits isogenic rifampicin resistant mutant AH-406. The arrows indicate the inoculation day (day 12) aftereach group of 10 axenic chickens had been placed in an axenic-controlled germfree isolator. Eachvalue is the average of 10 independent animals and the vertical bars at each point show the range ofvariation between birds.

of 107/g), while no A. hydrophila could be detected in the intestine of theuncolonized birds.

Discussion

We decided to study the role of the LPS of A. hydrophila strains O:34 on thecolonization in vivo of the germfree chicken gut, because in preliminaryexperiments we found a good colonization when the wild type strains were grown

S. Merino et al.330

302

Days

Log

cfu

/g o

f fa

eces

16

10

8

6

4

10 12 14 18 20 22

B

24 26 28

302

16

10

8

6

4

10 12 14 18 20 22 24 26 28

A

Fig. 4. Colonization in vivo of the germfree chicken gut by A. hydrophila strains grown at 20°C. (A)strain AH-495 (Χ—Χ) and its isogenic Tn5 rfb mutant AH-457 (Ε—Ε). (B), strain AH-406 (Ο—Ο) andits isogenic Tn5 rfb mutant AH-459 ( — ). The arrows indicate the inoculation day (day 12) and theconditions are the same as in Fig. 3.

at 20°C but not when they were grown at 37°, and we previously described thatthese strains showed a smooth LPS when they were grown at 20°C but roughwhen they were grown at 37°C.13 It is important to point out that the change onLPS is clearly a phenotypic change, because 37°C-grown cells produce smoothdescendants if cultured at 20°C. These facts, together with the results reportedin colonization of non-axenic animals by LPS-deficient mutants of Salmonellatyphimurium,15,16 prompted us to study the role of the O:34-antigen LPS in thiscolonization process.

Initially, we obtained transposon mutants devoid of the O:34-antigen LPS butwith a complete LPS-core, because in order to study the role of the O:34-antigen LPS, the LPS-core should be complete. The combination of transposon

O-antigen lipopolysaccharide colonization of germfree chicken gut 331

mutagenesis and immunoscreening with specific O:34 antiserum allowed us toselect mutants only devoid of the O:34-antigen LPS. This fact could be observedby different methodologies: bacteriophage sensitivity to specific phages for theO:34-antigen LPS and LPS-core, by LPS gels (Fig. 1), by specific gels for the LPS-core oligosaccharides (Fig. 2), and by the chemical composition of LPS (Table 3).From all these results we can conclude that mutants AH-457 and AH-459 areclearly rfb mutants, i.e. the Tn5 is inserted in the DNA region coding for the genesresponsible of the O:34-antigen biosynthesis.

The wild type strains or their derivative rifampicin resistant mutants grown at20° (smooth LPS) showed a high density of colonization of the germfree chickengut that remained uniform through the days of the experiment. By contrast, thesame strains grown at 37°C (rough LPS; LPS without the O:34-antigen LPS) orthe rfb mutants obtained by transposon mutagenesis (AH-457 and AH-459) grownat 20°C were unable to colonize the germfree chicken gut under the sameconditions in which the strains with smooth LPS were able to do it. Furthermore,all these strains grown at 20°C showed the same ability in a qualitative test tofimbriate, then all of them were able to produce fimbriae (an important colonizationfactor17).

This study has demonstated for the first time an important role for the O:34-antigen LPS in in vivo colonization of axenic animals by A. hydrophila O:34 strains.

Materials and methods

Bacteria, bacteriophages and media. The A. hydrophila strains from serogroup O:34used are listed in Table 1. Bacteriophages PM1 and PM2 have been characterized previouslyby us; they were propogated on A. hydrophila AH-3.11–12 The basal medium used forbacterial growth and phage propagation was Tryptone Soya Broth (TSB) or TSB with 1.5%agar (TSA). To prepare soft agar we added 0.6% agar to TSB. Bacteriophage sensitivity orresistance was assayed by spot test.

Spontaneous rifampicin resistant mutants were selected by plating 109 cells on TSA plus100 lg/ml of rifampicin. They were reisolated by restreaking on the same plates.

Transposon Tn5-mutants were obtained after a triparental mating using as donors E.coli S17-1 carrying the plasmid pSUP202 (a derivative of plasmid pBR322 unable to replicateon Aeromonas carrying the mob genes and transposon Tn5)18 and a helper conjugativeplasmid pRK2073 (spectinomycin resistant) in E. coli HB101,19 and using as a recipientstrain the rifampicin resistant mutants from the A. hydrophila wild type strains. Thetriparental matings were performed at 30°C and the selection was in TSA plates with100 lg/ml of rifampicin and 25 lg/ml of kanamycin. The transposon Tn5-mutants obtainedwere finally selected by immunoscreening after restreaking on the same selective plates.

Immunoscreening of O:34-antigen LPS mutants. Mutants grown on selective agar plates(plus 100 lg/ml of rifampicin and 25 lg/ml of kanamycin) were overlayed with precutnitrocellulose filters (Millipore Corp., Bedford, MA). Dried filters were then blocked,immunostained with specific O:34-antigen rabbit antiserum (1:1000) and stained with goatanti-rabbit immunoglobulin G alkaline phosphatase (Boehringer). The colonies that gavea negative signals were purified and rescreened again by the same procedure.

LPS isolation and analyses. LPS was purified by the method of Westphal and Jann,20

and as modified by Osborn21 for mutants devoid of the O-antigen LPS. Purified LPS wasanalysed by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE)and silver stained by the method of Tsai and Frasch.22 For analysis of the LPS-coreoligosaccharides, proteinase K digests of outer membrane fractions were separated on18% polyacrylamide gels using a SDS-tricine buffer and visualized by silver staining asdescribed.23 2-Keto-3-doxyoctulosonic acid (KDO) was measured by the thiobarbituric acidmethod as previously described by us.13 Monosaccharides were also analysed to their

S. Merino et al.332

alditol acetate derivatives by gas-liquid chromatography on a 3% SP-3840 column (Supelco)as previously described by us.11

Antiserum. Polyclonal antiserum against purified LPS of strain AH-3 (O:34) grown at20°C was raised in adult New Zealand white rabbits as previously described by us, andalso rendered specific for O:34 antigen LPS by extensive adsorption using LPS mutant AH-22 (only devoid of O-antigen LPS but with a complete LPS-core) as previously describedby us.13

In vivo studies.(i) Animals. White Leghorn chickens, strain PA12, were used as previously.24

(ii) Inoculation of animals. A 1 ml amount of late exponential phase cultures of the differentA. hydrophila strains (108–9/ml) grown at different conditions was successively force-fed to1-week-old-axenic chickens, using an esophageal catheter mounted on a syringe. A groupof 10 birds were used for each strain tested.(iii) Faeces sampling. The faeces were collected from the cloaca into separate sterile tubes,weighed, and their contents were diluted in sterile saline and shaken vigorously for severalminutes to homogenize the faeces. Appropriate dilutions were plated on TSA or Aeromonasselective medium25 with or without antibiotic (25 lg/ml of kanamycin or 100 lg/ml ofrifampicin) depending on the A. hydrophila strain inoculated. The detection limit was 102

cfu/g of faeces.

This work has been supported by a research grant from DGICYT (PB94-0906). We thankM. Sato for antiserum against fimbriae and Maite Polo for her technical assistance. X.R.and A.A. are a recipient of a fellowship from Ministerio de Educacion y Ciencia.

References

1. Trust TJ, Sparrow RAH. The bacterial flora in the alimentary tract of fresh-water salmonid fish.Can J Microbiol 1984; 20: 1219–28.

2. Freij BJ. Aeromonas: biology of the organism and diseases in children. Pediatr Infect Dis 1984; 3:164–75.

3. Sakazaki R, Shimada T. O-serogrouping scheme for mesophilic Aeromonas strains. Jpn J Med SciBiol 1984; 37: 247–55.

4. Lallier R, Bernard F, Lalonde G. Difference in the extracellular products of two strains of Aeromonashydrophila virulent and weakly virulent for fish. Can J Microbiol 1984; 30: 900–4.

5. Merino S, Benedı VJ, Tomas JM. Aeromonas hydrophila strains with moderate virulence. Microbios1989; 59: 165–73.

6. Misra SK, Shimada T, Bhadra RK, Pal SC, Nair GB. Serogroups of Aeromonas species from clinicaland environmental sources in Calcutta, India. J Diarrhoeal Dis Res 1989; 7: 8–12.

7. Kokka RP, Janda JM, Oshiro LS, Altwegg M, Shimada T, Sakazaki R, Brenner DJ. Biochemical andgenetic characterization of autoagglutinating phenotypes of Aeromonas species associated withinvasive and noninvasive disease. J Infect Dis 1991; 163: 890–4.

8. Janda JM, Guthertz LS, Kokka RP, Shimada T. Aeromonas species in septicemia: laboratorycharacteristics and clinical observations. Clin Infect Dis 1994; 19: 77–83.

9. Savage DC. Microbial ecology of the gastrointestinal tract. Ann Rev Microbiol 1977; 31: 107–33.10. Kirov S. The public health significance of Aeromonas spp. in foods. Int J Food Microbiol 1993; 20:

179–98.11. Merino S, Camprubı S, Tomas JM. Characterization of an O-antigen bacteriophage from Aeromonas

hydrophila. Can J Microbiol 1991; 38: 235–40.12. Merino S, Camprubı S, Tomas JM. Isolation and characterization of bacteriophage PM2 from

Aeromonas hydrophila. FEMS Microbiol Lett 1990; 68: 239–44.13. Merino S, Camprubı S, Tomas JM. Effect of the growth temperature on outer membrane

components and virulence of Aeromonas hydrophila strains of serotype O:34. Infect Immun 1992;60: 43434–9.

14. Sato M, Arita M, Honda T, Miwatami T. Characterization of a pilus produced by Aeromonashydrophila. FEMS Microbiol Lett 1989; 59: 325–30.

15. Nevola JJ, Stocker BAD, Laux DC, Cohen PS. Colonization of the mouse intestine by an avirulentSalmonella typhimurium strain and its lipopolysaccharide-defective mutant. Infect Immun 1985;50: 152–9.

16. Nevola JJ, Laux DC, Cohen PS. In vivo colonization of the mouse large intestine and in vivopenetration of intestinal mucus by an avirulent strain of Salmonella typhimurium and itslipopolysaccharide deficient mutant. Infect Immun 1987; 55: 2884–990.

O-antigen lipopolysaccharide colonization of germfree chicken gut 333

17. Salyers AA, Whitt DD. Virulence factors that promote colonization. In: Bacterial pathogenesis, amolecular approach. Washington, DC: ASM Press 1994: 30–46.

18. Simon R, Priefer U, Puhler A. A broad host range mobilization system for in vivo geneticengineering: transposon mutagenesis in Gram negative bacteria. Biotechnology 1983; 1: 784–91.

19. Ditta G. Tn5 mapping of Rhizobium nitrogen fixation genes. Methods Enzymol 1985; 118: 519–28.20. Westphal O, Jann K. Bacterial lipopolysaccharides: extraction with phenol-water and further

applications of the procedure. Methods Carbohydr Chem 1965; 5: 83–91.21. Osborn MJ. Preparation of lipopolysaccharide from mutant strains of Salmonella. Methods

Enzymol 1966; 8: 161–4.22. Tsai CM, Frasch CE. A sensitive silver stain for detecting lipopolysaccharide in polyacrylamide

gels. Anal Biochem 1982; 119: 115–9.23. Hitchcock PJ, Brown TM. Morphological heterogeneity among Salmonella lipopolysaccharide

chemotypes in silver-stained polyacrylamide gels. J Bacteriol 1983; 154: 269–77.24. Guillot JF, Chaslus-Dancia E, Lafont JP. Spontaneous implantation of antibiotic-resistant

Enterobacteriaceae in the digestive tract of chickens in the abscence of selective pressure.Antimicrob Agents Chemother 1977; 12: 697–702.

25. Palumbo SA, Maxino F, Williams AC, Buchanan RL, Thayer DW. Starch-ampicillin agar for thequantitative detection of Aeromonas hydrophila. Appl Environ Microbiol 1985; 50: 1027–30.