The Production of F41 Fimbriae by Piglet Strains of Enterotoxigenic Escherichia coli that Lack K88, K99 and 987P Fimbriae

Transcription

1 Journal of General Microbiology (1 983), 129, Printed in Great Britain 2753 The Production of F41 Fimbriae by Piglet Strains of Enterotoxigenic Escherichia coli that Lack K88, K99 and 987P Fimbriae By J. A. MORRIS,* C. J. THORNS, G. A. H. WELLS, A. C. SCOTT AND W. J. SOJKA Central Veterinary Laboratory, New Haw, Weybridge, Surrey KT1.5 3NB, U.K. (Received 9 February 1983; revised 2.5 April 1983) The enterotoxigenic Escherichia coli strains 1676,1706,175 1 and KEC96a, which do not produce fimbrial adhesive antigens of the K88, K99 or 987P antigen type reacted both in vitro and in vivo with antiserum to F41 fimbriae in an indirect immunofluorescent antibody technique. Antiserum used to demonstrate material B, an adhesive antigen thought to mediate the adhesive and mannose-resistant (MR) haemagglutinating properties of E. coli strains 1676, 1706 and 1751, reacted in vitro with an F41 strain. The antiserum also inhibited the MR haemagglutinating activity of F41 antigen and gave an anionic precipitation line in immunoelectrophoresis experiments with an extract containing F41 antigen. The MR haemagglutinating properties of an antigen extract containing material B from E. coli strain 1706 was neutralized by antiserum to F41 fimbriae and by OK antisera to E. coli strains that produce both F41 and K99 fimbriae. These sera also gave an anionic precipitation line with the MR haemagglutinin from E. coli strain 1706 and the MR haemagglutinin gave a line of identity with F41 in gel diffusion experiments with antiserum to F41 fimbriae. OK antisera to K99 F41- bacteria and OK antisera to K88 bacteria and 987P bacteria did not react with this haemagglutinin. Transmission electron microscopy on the ileum of newborn gnotobiotic piglets infected with E. coli strain 1706 showed irregular, poorly defined filamentous material surrounding some, though not all, bacteria but regular fimbrial structures were not visible. Fine thread-like material was visible surrounding most of the bacteria in ultra thin sections from a piglet infected with an E. coli strain that produces both F41 and K99. Staining with ruthenium red suggested that this material was associated with polysaccharide. INTRODUCTION The K88, K99 and 987P fimbriae produced by enterotoxigenic strains of Escherichia coli are involved in the adhesion to the microvillous borders of the villous enterocytes in the small intestine of newborn piglets (Moon et al., 1979). However, some strains of E. coli have been isolated that do not produce any of these three types of fimbriae and yet colonize the small intestine and cause diarrhoea in the suckling pig (Moon et al., 1980). It was suggested that these E. coli, which Moon referred to as 3P- strains, might produce fimbriae of unknown antigen type that facilitate attachment of these organisms to the ileum. Two surface antigens were isolated : material A, which contained fimbriae, caused mannose-sensitive haemagglutination and did not appear to mediate adhesion of the E. coli strains to the pig intestine; and material B, which did not appear to contain fimbriae, caused mannose-resistant haemagglutination and was thought to be responsible for the adhesive properties of the bacteria (Awad-Masalmeh et al., 1982). Escherichia coli strain B41, the K99 reference strain, and certain other K99 organisms produce, in addition to K99, a second adhesive antigen which also exhibits mannose-resistant haemagglutination (Morris et al., 1980). This activity was associated with the presence of /83/ Crown Downloaded copyright from by

2 2154 J. A. MORRIS AND OTHERS fimbriae (Morris et al., 1982, De Graaf & Roorda, 1982) which were provisionally referred to as F41 (Morris et al., 1982). In the present study it is demonstrated that the 3P- E. coli strains described by Moon et al. (1980) and a further enterotoxigenic E. coli strain (KEC 96a) that did not produce K88, K99 or 987P fimbriae all produce F41 antigen in vitro and in vivo. METHODS Bacteria'The strains of Escherichia coliexamined are listed in Table 1 and Table 4 which also give details of their isolation. Bacteria were grown routinely on sheep blood agar for 18 h at 37 "C unless stated otherwise but for the original detection of F41 antigen, bacteria were grown in Minca broth (Guinee et al., 1976) supplemented with 0.1 % yeast extract. Antisera. OK antisera were produced in rabbits by intravenous injection of bacteria (Sojka, 1965). Details of antisera to the different fimbriae are given in Table 1, Table 2 and Table 4. indirect immunofuorescent antibody technique. The first antibody was rabbit antiserum to the relevant fimbriae; as controls (a) normal rabbit serum, and (b) OK antiserum to the homologous organism was used. The fluoresceinlabelled IgG fraction of goat anti-rabbit immunoglobulin (Nordic Immunobiological Laboratories, Maidenhead, Berks, U.K.) was used as the second antibody. Isolation of mannose resistant (MR) haemagglutinins. The original technique for the isolation of K99 was used (Morris et al., 1977). Briefly, the bacteria were grown on sheep blood-agar overnight and the haemagglutinins were extracted by shaking at 60 "C. The haemagglutinins were precipitated from the claiified supernate at ph 4.0 using acetic acid, partially purified by five cycles of re-precipitation and adjusted to a protein concentration of 8 mg ml-' in 100 mm-phosphate buffer ph 7.5. Haemagglutination. Direct haemagglutination tests using 3% (v/v) suspensions of sheep erythrocytes were performed in microtitre trays at 4 "C with D-mannose present at a final concentration of 0.5% (Morris et al., 1977). The effect of various antisera on the activity of the haemagglutinins was investigated by incubating equal volumes of sample and antiserum at 37 "C overnight. Immunoelectrophoresis and gel dirkion. The method of Scheidegger as described previously (Morris & Hussaini, 1,974) was used for immunoelectrophoresis. Ouchterlony double diffusion tests were performed in agar gels (Morris et al., 1978). Diffusions proceeded overnight at 4 "C. Bacterial adhesion in vitro to enterocytes. Enterocytes were isolated from the small intestine of healthy calves (Sellwood et al., 1975). Enterocyte suspensions (100 pl) were incubated at 37 "C for 30 min with an equal volume of bacteria. In competition experiments, 50 pl enterocyte suspension (lo5 cells) was incubated with an equal volume Table 1. Indirect immunofluorescent antibody staining of E. coli strains with antisera to K99 and F41 Jimbriae E. coli strain B4 1 B41M K EC96a Serogroup 0101 :K99 F :F41 O'K 12': K :K :K :K :K28 History Isolated from calf with diarrhoea Laboratory derived K99- form of strain B41M K 12 recombinant containing K99 plasmid from strain B41 Isolated from piglets with diarrhoea1 Isolated from piglets with diarrhoea :K28 K99 E. coli in same herdt * Supplied by F. K. De Graaf. t Supplied by P. A. M. Guinee. 1 Supplied by H. W. Moon. Indirect immunofluorescence with antisera to purified fimbrial antigen type I 7 F41* K99t - zk

3 F41 Jimbriae on E. coli lacking K88, K99 and 987P 2755 of competitor (2 mg ml-l) or saline for 30 min before the addition of lo8 bacteria in 50 p1. All experiments were performed in the presence of 0.5% (w/v) D-mannose and each experiment was performed on at least three separate occasions. The mean number of adherent bacteria per enterocyte was calculated from at least 60 enterocytes. Experimental animals. Full term, hysterotomy derived, colostrum deprived piglets were assigned in groups of two or three to plastic isolators where they were inoculated orally with 5 x lo8 to 5 x lo9 E. coli and maintained under gnotobiotic conditions (Morris et al., 1982). Between 16 and 20 h post inoculation the clinical status of piglets was noted. Segments from the ileum were obtained by laparotomy under halothane anaesthesia. One portion was fixed in glutaraldehyde for examination by electron microscopy, and a second was quenched in liquid nitrogen for cryostat section. The latter was cut at 10 pm, fixed in acetone and examined by the indirect immunofluorescent antibody technique. A third portion was fixed in 10% (v/v) neutral buffered formalin for conventional histological examination in selected experimental groups. Electron microscopy. For scanning electron microscopy glutaraldehyde-fixed samples were post fixed in osmium tetroxide and coated with gold; and for transmission electron microscopy the fixed material was embedded in Araldite and ultra thin sections stained either with methanolic uranyl acetate and lead citrate, or with 0.01 % ruthenium red in cacodylate buffer as described in detail (Morris et al., 1982). RESULTS In vitro demonstration of F41 antigen The indirect fluorescent anti body test using antiserum to purified F41 fimbriae demonstrated F41 antigen on each of the 3P- E. coli strains 1676, 1706 and 1751 (Table 1). Strain KEC96a fluoresced weakly with the antiserum but F41 antiserum failed to react with the K99 E. coli strain This strain, together with strain B44, fluoresced strongly with K99 antiserum which did not react with any of the other E. coli. Thus, strain KEC96a and each of the 3P- E. coli produced F41 in vitro. B antiserum reacted with E. coli strain B41 thereby indicating that this strain produces material B. The other bacteria were not examined with this antiserum. Immunological properties of the MR haemagglutinin(s) from E. coli strain I706 MR haemagglutinin prepared from E. coli strain 1706 agglutinated sheep erythrocytes in the presence of 0.5% D-mannose to a reciprocal titre of 64. B antiserum completely inhibited this reaction (Table 2). The activity was also completely inhibited by antiserum to F41 prepared (a) directly from the purified antigen, and (b) by absorption of OK antiserum to strain B41 M. It was also inhibited by OK antisera to K99-positive E. coli of serogroups 09 or 0101 (Table 2). Antiserum to K99 positive E. coli of serogroups 08 or 064 had no effect on the haemagglutinating activity of the 1706 haemagglutinin and antibodies to K88 or 987P fimbriae were also without effect. The 1706 haemagglutinin gave a single anionic precipitation line on immunoelectrophoresis against B antiserum and with the two F41 antisera. All three antisera inhibited its haemagglutinating activity (Table 2). The haemagglutinating property of F41 antigen prepared from E. coli strain B41M was completely inhibited by absorption with B antiserum which also gave a single anionic precipitation line with F41 antigen. Surface antigen extracts from E. colistrain B41 (0101 : K99, F41) contained both K99 and F41 adhesins (Morris et al., 1982). Consequently immunoelectrophoresis of these extracts against OK antiserum to E. coli strain B41 gave both an anionic precipitation line, characteristic of F41, and a cationic precipitation line, characteristic of K99. Only the F41 line developed on diffusion of the adhesin mixture against F41 (abs) antiserum or B antiserum, whereas only the K99 line was visible on diffusion against OK antiserum to E. coli K12 K99 (Fig. 1). In gel diffusion experiments the E. coli strain 1706 haemagglutinin gave a line of identity when diffused against B antiserum and F41 (abs) antiserum. The F41 (abs) antiserum and OK antiserum to E. coli strain B44 (09 : K30, K99, F41) gave a line of identity on diffusion against the 1706 haemagglutinin but there was no precipitation with OK antiserum to E. coli K12 K99 or with OK antiserum to E. coli strain B117 (08 :K85, K99). Diffusion of F41 adhesin against F41 antiserum and B antiserum gave a line of identity.

4 2756 J. A. MORRIS AND OTHERS Table 2. Effect of dfferent antisera on the mannose resistant haemagglutinating properties, and the immunoelectrophoresis pattern of the mannose resistant haemagglutinin from E. coli strain 1706 Immunoelectrophoresis precipitin arc & Serum Antibodies to : HA inhibition Anionic Cationic B* F41 (abs)t F41$ OK B85 OK B44 OK B42 OK B41 OK RVC 1616 OK B79 OK 431 OK 613 OK B117 OK S13 OK 637 OK 1475 OK 987 OK G7 OK Abbotstown OK X91 '3P-' F4 1 F : K99 F41 09 :K30 K99 F41 09:K35 K99 F :K99 F :K30 K99 F :K32 K99 F :K30 K99 F :K30 K99 F41 08 : K85 K99 08 :K85 K :K'V142' K99 OK12' : K99 09 : K P 08:K87 K88ab 0149:K91 K88 ac 08 :K87 K88ad * OK antiserum to E. coli strain 1706 absorbed with boiled culture of strain 1706 and then with strain 1706 cultured at 18 "C supplied by H. W. Moon. OK antiserum to E. coli strain B41M absorbed with strain B41M cultured at 18 "C. 2 Prepared against purified fimbriae. Supplied by F. K. De Graaf. - Fig. 1. Immunoelectrophoresis of surface antigen extract of Escherichia coli strain B41 (0101 :K99 F41). Each well contains antigen extract; trough (a) contains OK antiserum to E. coli strain B41, trough (b) contains F41 (abs) antiserum, trough (c) contains B antiserum, trough (d) contains OK antiserum to E. coli K12 K99. In vitro bacterial adhesion to caw enterocytes Escherichia coli strains 1676, 1706 and 1751 each adhered in vitro to the brush borders of calf enterocytes in the presence of 0.5% D-mannose (Table 3). The attachment of E. coli strain B41M was comparable whereas E. coli strain KEC96a did not attach at all. This was consistent with the immunofluorescence studies that indicated that strain KEC96a produced less F41 in vitro than the other F41 E. coli strains. Preparations of E. coli strain 1706 MR haemagglutinin and F41 antigen each reduced the attachment of all four strains.

5 F41 Jimbriae on E. coli lacking K88, K99 and 981P Animal studies The 3P- strains 1676, 1706, 1751 and strain KEC96a each colonized the small intestine of newborn piglets (Table 4). Adherence of these strains to villi was demonstrated by immunofluorescence with B antiserum and F41 antiserum but not with antisera to K99, K88 or 987P. With the exception of strain KEC96a, each of the strains were capable of producing diarrhoea. The control strain B44 (producing K99 and F41 fimbriae) gave positive immunofluorescence with both F41 and K99 antiserum while the other control strain X177/81 (possessing 987P fimbriae only) gave positive results only with antiserum to 987P. (The failure to observe adherence of this strain by scanning electron microscopy was probably due to artifacts produced by the preparation of the specimen.) Except in experiments with E. coli strain 1751, these results were confirmed on two other occasions using groups of three piglets in each experiment. In a second experiment with strain 1751 colonization was minimal in two of three piglets and in a third experiment two of the piglets died overnight and were not examined, but colonization was readily demonstrated in the survivor. Whenever strain did colonize the small intestine, F4 1 antigen was always demonstrated. When immunofluorescence occurred it was confined to the luminal border of villous enterocytes forming a continuous layer around most villi. Examination by transmission electron microscopy of ultra thin sections of infected ileum showed that bacteria of strain 1706 were located in close association with each other and with the tips of the microvilli. Irregular, poorly defined electron dense thread-like material was seen surrounding some of the bacteria but no Table 3. Efect of F41 antigen and E. coli strain 1106 mannose-resistant haemagglutinin on the in vitro attachment of E. coli strains to calfenterocytes Mean no. of bacteria per enterocyte in the presence of:, E. coli strain Saline F MR haemagglutinin B41M 13.7 (k6.5) 2.5 ( f 2.6) 2.6 ( f 3.4) (210.0) 5.4 (k 6.3) 1.4 (f 2-0) ( f 9.0) 4.2 (f4.1) 1-2 (f 1.8) ( f 9.4) 3-8 (k4.0) 1.2 ( 1.9) KEC96a <I NT NT NT, Not tested. Table 4. Diarrhoea, colonization of the small intestine and the production of Jimbriae in piglets orally inoculated with diflerent strains of E. coli Diarrhoea Immunofluorescence of villus-associated (no. affected/ bacteria using antisera to : E. coli no. surviving colonization of r A \ strain to termination) small intestine* B F4l(abs) F41 K99 K88t 987P$ (1 died) - B@ KEC96a 012 X177/81 I1 212 Mil k/con trol NT NT - NT NT, Not tested. * Colonization demonstrated by scanning electron microscopy. OK antisera to E. coli strain D214 (0100:K88ab). #, OK antiserum to E. coli strain 987(09:K103, 987P) absorbed with boiled culture of strain Isolated from calf with diarrhoea. 11 Isolated from piglet with diarrhoea.

6 2758 J. A. MORRIS AND OTHERS structures identifiable as fimbriae were visible. Sections stained with ruthenium red revealed a definite electron dense layer of material surrounding the bacteria and microvilli. In contrast, ultra thin sections of the ileum from a piglet infected with the K99-positive E. coli strain B44 showed bacteria apparently surrounded by an electron lucent area which separated them from each other and from the microvilli. Fine, irregular thread-like electron dense material radiated from the surface of these bacteria. Similar structures were also observed in sections stained with ruthenium red and on several occasions these structures made contact with the microvilli. DISCUSSION The enteropathogenic Escherichia coli strains 1676, 1706 and do not produce fimbriae of the antigen types K88, K99 or 987P and yet they colonize the small intestine of neonatal piglets (Moon et al., 1980). It was postulated that these organisms produce fimbriae of unknown antigen type(s) which mediate adhesion and colonization of the small intestine. The present study demonstrated that each of these E. coli strains produce F41 fimbriae both in vitro and in vivo. Furthermore, F41 bacteria reacted with the B antiserum used by Awad-Masalmeh et al. (1982) to demonstrate material B, the antigen thought to be responsible for the MR haemagglutination activity and adhesive properties of E. coli strains 1676, 1706 and Escherichia coli strain KEC96a (0101 : K28) another enterotoxigenic strain which does not produce K88, K99 or 987P, also produces F41 fimbriae but this antigen was demonstrated clearly only when grown in viuo. This organism was isolated from piglets in a herd from which E. coli of the serogroup 0101 : K28, K99 had also been isolated and it seems likely that E. coli strain KEC96a is a naturally occurring K99- variant. Surface antigen with MR haemagglutinating activity was isolated from E. coli strain 1706 and haemagglutination inhibition experiments with B antiserum suggested that this activity was due to material B. However, the purity of material B was not recorded by Awad-Masalmeh et al. (1982) and the purity of the 1706 haemagglutinin was not assessed in the present study. It is possible, therefore, that the serum inhibition of haemagglutination may not be specific. Nevertheless, F41 fimbriae have only been found on K99 E. coli from the 09 and 0101 serogroups and the MR haemagglutinin isolated from E. coli strain 1706 reacted only with OK antisera to K99 E. coli from serogroups 09 and 0101 in haemagglutination inhibition experiments and immunoelectrophoresis experiments. Furthermore neutralization of (a) the E. coli strain 1706 MR haemagglutinin by F41 antiserum, and (b) the F41 MR haemagglutinating activity by B antiserum, suggests that material B contains F41 antigen. Moreover, the inhibition of adhesion to calf enterocytes of E. coli strains 1706, and 1676 by F41, and F41' E. coli strain B41 M by 1706 haemagglutinin, is further evidence that the adhesive property of material B is due, at least in part, to F41 antigen. Experiments with gnotobiotic piglets demonstrate that F41 is produced in viuo and confirm earlier reports that E. coli strain 1676, 1706 and each colonize the small intestine and do not produce fimbrial antigens of the K88, K99 or 987P types (Moon et al., 1980). Transmission electron microscopy studies confirm that E. coli strain 1706 comes into close contact with microvilli in the small intestine and that fimbriae cannot be demonstrated readily in vivo. This contrasts with the finding with E. coli strain B44 where an electron lucent zone traversed by filamentous structures is observed between the bacteria and the microvilli. The staining of these filaments with ruthenium red suggests that they may be associated with polysaccharide and further supports a role for capsular polysaccharide in the colonization of the small intestine (Sojka et al., 1978; Hadad & Gyles, 1982). The fimbrial antigens involved in the colonization of the piglet small intestine by enterotoxigenic strains of E. cozi have hitherto included K88, K99 and 987P but it seems probable that F41 is yet another fimbrial antigen involved in this process. If F41 does play a role in uiuo, then its incorporation into a multivalent vaccine to stimulate colostral antibodies for the passive protection of suckling piglets would warrant consideration and its detection would be of value in the diagnosis and epidemiology of E. coli neonatal diarrhoea.

7 F41 firnbriae on E. coli lacking K88, K99 and 987P 2759 We are grateful to Drs M. Awad-Masalmeh, F. K. De Graaf, P. A. M. Guinee and H. W. Moon for their interest and generosity in supplying bacterial cultures and antisera. REFERENCES AWAD-MASALMEH, M., MOON, H. W., RUNNELS, P. L. agglutination test for bovine leptospirosis. Journal of & SCHNEIDER, R. A. (1982). Pilus production, Hygiene 73, hemagglutination, and adhesion by porcine strains MORRIS, J. A., STEVENS, A. E. & SOJKA, W. J. (1977). of enterotoxigenic Escherichia coli lacking K88, K99 Preliminary characterization of cell-free K99 antiand 987P antigens. Infection and Immunity 35, 305- gen isolated from Escherichia coli B41. Journal of 313. General Microbiology 99, DE GRAAF, F. K. & ROORDA, I. (1982). Production, purification and characterization of the fimbrial adhesive antigen F41 isolated from the calf enteropathogenic Escherichia coli strain B41M. Infection and Immunity 36, GUINEE, P. A. M., JANSEN, W. H. &AGTERBERG, C. M. (1976). Detection of the K99 antigen by means of agglutination and immunoelectrophoresis in Escherichia coli isolates from calves and its correlation with enterotoxigenicity. Infection and Immunity 13, HADAD, J. J. B. & GYLES, C. L. (1982). Scanning and transmission electron microscopic study of the small intestine of colostrum-fed calves infected with selected strains of Escherichia coli. American Journal of Veterinary Research 43, MOON, H. W., ISAACSON, R. E. & POHLENZ, J. (1979). Mechanisms of association of enteropathogenic Escherichia coli with intestinal epithelium. American Journal of Clinical Nutrition 32, MOON, H. W., KOHLER, E. M., SCHNEIDER, R. A. & WHIPP, S. C. (1980). Prevalence of pilus antigens, enterotoxin types and enteropathogenicity among K88-negative enteroxigenic Escherichia coli from neonatal pigs. Infection and Immunity 27, MORRIS, J. A. & HUSSAINI, S. N. (1974). Characterization of the antibodies detected by the microscopic MORRIS, J. A., STEVENS, A. E. & SOJKA, W. J. (1978). Anionic and cationic components of the K99 surface antigen from Escherichia coli B41. Journal of General Microbiology 107, MORRIS, J. A., THORNS, C. J., Scon, A. C., SOJKA, W. J. & WELLS, G. A. H. (1982). Adhesion in uitro and in uiuo associated with an adhesive antigen (F41) produced by a K99- mutant of the reference strain Escherichia coli B41. Infection and Immunity 36, MORRIS, J. A. THORNS, C. J. & SOJKA, W. J. (1980). Evidence for two adhesive antigens on the K99 reference strain Escherichia coli B41. Journal of General Microbiology 118, SELLWOOD, R., GIBBONS, R. A., JONES, G. W. & RUTTER, J. M. (1975). Adhesion of enteropathogenic Escherichia coli to pig intestinal brush border. The existence of two pig phenotypes. Journal oj hfedical Microbiology 8, SOJKA, W. J. (1965). Escherichia coli in Domestic Animals and Poultry. Farnhamn Royal : Commonwealth Agricultural Bureaux. SOJKA, W. J, WRAY, C. & MORRIS, J. A. (1978). Passive protection of lambs against experimental enteric colibacillosis by colostral transfer of antibodies from K99-vaccinated ewes. Journal of Medical Microbiology 11,