Vol. 60, No. 7
INFECTION AND IMMUNITY, JUlY 1992, p. 2648-2656 0019-9567/92/072648-09$02.00/0 Copyright ©) 1992, American Society for Microbiology
and pil-Related DNA Sequences and Other Virulence Determinants Associated with Escherichia coli Isolated from Septicemic Chickens and Turkeys
pap-
CHARLES M. DOZOIS, JOHN M. FAIRBROTHER,* JOSEE HAREL, AND MARC BOSSE
Departement de Pathologie et de Microbiologie, Faculte de Medecine Veterinaire, Universite de Montreal, C.P. 5000, Saint-Hyacinthe, Quebec, Canada J2S 7C6 Received 16 September 1991/Accepted 7 April 1992
Escherichia coli isolates from septicemic or healthy chickens and turkeys from Quebec were serotyped, examined genotypically by using DNA probes specific for the pil and pap fimbrial systems and the aerobactin siderophore system, and examined phenotypically for lethality in day-old chicks, hemagglutination, serum resistance, and aerobactin production. Serogroups 078 and 01 were most common in septicemic chickens and turkeys. pap' isolates from chickens were associated with septicemia, and pap' isolates from turkeys were associated with lethality in day-old chicks. Four of nine pap' isolates from septicemic turkeys expressed P adhesin, whereas all pap' isolates from septicemic chickens were negative for P adhesin. The pil+ genotype was associated with septicemia in chickens and with serum resistance in isolates from turkeys. Mannose-sensitive hemagglutination of guinea pig erythrocytes was associated with septicemia in chickens and turkeys, although this phenotype was not associated with pil+ isolates from turkeys. Serum resistance was associated with isolates from septicemic turkeys and with lethality in isolates from chickens. The aerobactin system was associated with isolates from septicemic chickens and turkeys. Overall, results indicated that (i) genotypic examination may reveal virulence-associated traits which differ from the typically expected phenotype and/or are not readily expressed in vitro, and (ii) certain phenotypic and genotypic traits associated with E. coli causing extraintestinal disease in humans and animals are also associated with E. coli causing avian septicemia.
of strains from serogroups 01 and 02 (14, 17, 55). Most strains expressing either FlA or FlA-like fimbriae have been shown to adhere to chicken respiratory epithelial cells in vitro (17, 25). In addition, mannose-resistant hemagglutinating (MRHA) P or P-related fimbriae may be expressed by some E. coli of avian origin (1, 46, 56). To date, the characterization of virulence factors of E. coli from poultry has consisted mainly of phenotypic studies. Whereas phenotypic studies of virulence factors require in vitro expression, which may vary during infection of the host (4, 48), DNA hybridization techniques allow the examination of virulence factors without relying on phenotypic expression (27, 33, 43). The purpose of this study was to characterize E. coli isolated from either healthy or septicemic chickens and turkeys by examination of both phenotypic and genotypic characteristics. Virulence, the presence of related sequences for the pil (FlA), pap (P [F13]), sfa, and afa operons, MSHA and MRHA, serum resistance, and the presence and expression of aerobactin and hemolysin genes were investigated in order to establish the relationship between genotype and phenotype and to compare the frequency of these factors in pathogenic versus commensal E. coli, thus elucidating the role of these factors in the pathogenesis of colisepticemia in chickens and turkeys. Results of the study demonstrated that (i) certain phenotypes and/or genotypes were associated with septicemia and/or virulence in chickens and/or turkeys, (ii) differences were observed between isolates from septicemic chickens and turkeys, and (iii) a good correlation between genotype and phenotype was observed for the aerobactin system, but not for the pap and pil fimbrial systems.
Colibacillosis is one of the principal causes of morbidity and mortality in chickens and turkeys and causes significant economic losses to the poultry industry. Of the various forms of Eschenichia coli infection occurring in poultry, the most common syndrome involves airsacculitis, frequently followed by a generalized infection. The air sacs are the first organs affected, and extension of the infection may result in pericarditis, fibrinous perihepatitis, and an often fatal septicemia (23, 40, 54). Infection is often enhanced or initiated by predisposing agents, which include environmental and biological factors (e.g., viruses and Mycoplasma spp.) (12). Pathogenic strains belong mainly to serogroups 01, 02, and 078 (15, 28, 52). The aerobactin iron-sequestering system is frequently associated with E. coli causing extraintestinal infections in humans (13) and other animals (19, 35) and is correlated with virulence of E. coli isolates from chickens (34). In addition, extraintestinal E. coli is often significantly resistant to the bactericidal effects of normal serum. For example, a correlation between virulence and resistance to normal turkey serum has been observed in E. coli isolated from turkeys (18). The characterization of fimbriae of E. coli from poultry has centered on the examination of strains from select serogroups (01, 02, and 078) isolated from diseased birds. E. coli isolated from septicemic chickens has been shown to possess mainly 17-kDa mannose-sensitive hemagglutinating (MSHA) FlA (type 1A) fimbriae, in the case of strains from serogroup 078, and MSHA fimbriae which have been classified as FlA-like on the basis of differences in the molecular weight and immunological reactivity, in the case
*
Corresponding author. 2648
VIRULENCE DETERMINANTS OF E. COLI FROM POULTRY
VOL. 60, 1992
MATERIALS AND METHODS Bacterial strains. One hundred and seventy-five E. coli isolates were included in this study. Eighty-three isolates were from lesions of visceral organs of 2- to 52-day-old chickens with colisepticemia. Twenty-nine isolates were similarly obtained from 3- to 12-week-old turkeys with colisepticemia. Isolates of disease origin were from birds submitted to the Laboratoire de Pathologie Animale, SaintHyacinthe, Quebec, from 1987 to 1990. In addition, 29 and 34 isolates were obtained from the feces and litter of healthy chickens and turkeys, respectively; in all cases, they were from farms with no abnormal incidence of colisepticemia. 0 serotyping was done by using standard slide and tube agglutination techniques as described previously (45) with 30 antiserum samples of the most common serogroups of E. coli isolated from birds with colisepticemia. Isolates were stored on Dorset agar slants at 4°C. Unless otherwise stated, bacteria were grown for 18 h in tryptic soy broth (Difco Laboratories, Detroit, Mich.) prior to examination. E. coli reference strains included 4787 (F165+, aerobactin positive), 6894 (aerobactin positive, Hly+), and BAM (rough, F1A+). In addition, strains MC1061(pPILL1005), MC1061 (pPILL1006), and MC1061(pPILL1007) (2), carrying internal fragments of the afa, pap, and sfa operons, respectively, were used as controls and for the preparation of gene probes. E. coli K-12 HB101 was used as a negative control. Lethality test. The virulence of 101 isolates randomly selected from groups of different origins was evaluated by a lethality test in 1-day-old chicks by using the technique of Dho and Lafont (16) with some modifications. Briefly, for each isolate, groups of five 1-day-old chicks were inoculated subcutaneously with 0.5 ml of 10-, 100-, or 1,000-fold-diluted volumes of overnight culture. Bacterial counts were estimated by using a spread plate method on brain heart infusion agar (Difco). Chicks were observed for 3 days, mortalities were recorded, and the 50% lethal dose (LD50) was calculated. Isolates were grouped into three lethality classes: 1, LD50, c5 x 106 CFU ml-'; 2, LD50, 5 x 106 to 5 x 108 CFU ml-1; and 3, LD50 .5 x 108 CFU ml-'. Hemagglutination. MSHA and MRHA tests were performed essentially as described by Fairbrother et al. (19). For MRHA, isolates were grown overnight at 37°C on minimal Davis agar (Difco) plus Casamino Acids (MD-1 agar) and were tested with chicken, human 0P1, and human A1P1 erythrocytes. Isolates were also grown overnight at 37°C on tryptic soy agar-5% (vol/vol) bovine blood (bovine blood agar) and tested with chicken, bovine, ovine, human OP1, and human A1P1 erythrocytes. Hemagglutination (HA) of chicken erythrocytes was also examined with isolates grown in MD-1 broth. MRHA-positive isolates were examined for recognition of the aL-D-galactosyl-(1,4)-,-galactose (Gal-Gal) receptor, as observed for P fimbriae, by using the P-latex test (2). Certain isolates were reexamined for MRHA and P-latex agglutination after 10 passages on bovine blood agar. For MSHA, isolates were subjected to four consecutive passages for 48 h each in static tryptic soy broth at 37°C and tested with guinea pig erythrocytes in the presence or absence of 5% D-mannose. Preparation of DNA probes. DNA probes for the fimbrial operons afa, pap, and sfa were generated from the recombinant plasmids pPILL1005, pPILL1006, and pPILL1007, respectively (kindly provided by A. Labigne, Institut Pasteur, Paris, France) and were digested by the restriction endonuclease PstI as described previously (2). The Afa probe is a 1.1-kb DNA fragment internal to afaC, the Pap
2649
probe is a 0.3-kb fragment internal topapC (60), and the Sfa probe is a 0.8-kb fragment previously designated F, which incorporates part of the sfa biogenesis and structural genes (47). The FlA probe for the pil fimbrial gene cluster was a 2.2-kb BamHI-PvuII fragment of the pilC gene generated from pSH2 (kindly provided by P. E. Orndorff, North Carolina State University, Raleigh) (39). The aerobactin receptor gene probe was a 2.2-kb PvuII fragment of pABN1 (7), and the aerobactin synthesis gene probe was a 1.8-kb AvaI fragment of pABN5 (7) (both pABN1 and pABN5 were kindly provided by J. B. Neilands, University of California, Berkeley). The hemolysin probe was a 1.8-kb PvuII fragment generated from pSF4000 (61) (kindly provided by R. Levesque, Laval University, Quebec, Canada). Plasmid DNA was extracted and purified by ultracentrifugation in a cesium chloride gradient (38). After restriction enzyme digestion, the resulting fragments were separated by agarose gel electrophoresis or polyacrylamide gel electrophoresis. Appropriate fragments were cut from the gel, concentrated by ethanol precipitation, purified, and radiolabeled with [a_-32P]dCTP by using a multiprimer DNA labeling kit (Pharmacia LKB Biotechnology Inc., Baie d'Urfe, Quebec, Canada), according to the instructions of the manufacturer.
Colony hybridization. Isolates were spotted onto Luria agar (Luria broth containing 15 g of agar per liter) and incubated at 37°C for 18 h. Colonies were then transferred to Whatman 541 filter paper (Whatman, Inc., Clifton, N.J.), and the filters were processed, hybridized, and revealed by autoradiography as described previously (10). Aerobactin production. A bioassay to determine the phenotypic expression of aerobactin was performed by using a cross-feeding method (34) with slight modifications (19). Serum bactericidal assay. The bacterial survival in serum of 64 isolates randomly selected from the groups of different origins and previously tested for lethality in day-old chicks was examined by using a bactericidal assay adapted from Taylor and Kroll (59). Briefly, a 10-fold dilution of overnight culture was resuspended in tryptic soy broth and incubated under agitation for 1.5 h. A 10-,ul volume of this culture was resuspended in 990 ,u of gelatin-Veronal-buffered saline plus magnesium and calcium ions (pH 7.35) to obtain approximately 107 CFU ml-1. A 50-,ul volume of the suspension was added and mixed with 450 pul of normal serum pooled from three birds (either 15-day-old chickens or turkeys, according to the species of origin of the isolate). After mixing, the serum-bacteria suspensions were incubated at 37°C for 3 h. Viable counts were taken at hourly intervals. The demarcation between the serum resistance and the serum sensitivity of bacteria was defined as a 2-log decrease in bacterial population after 3 h of incubation. Strains were examined prior to the serum resistance assay for the production of FlA and P fimbriae by MSHA of guinea pig erythrocytes and the P-latex test, respectively. Statistical analysis. Establishment of statistical differences between and among groups was evaluated using the G test of independence (53).
RESULTS O serotyping. Among 112 isolates from cases of colisepticemia, serogroup 078 was most frequent in chickens and turkeys, followed by 01 and 055 in turkeys and 01 and 018 in chickens (Table 1). With the exception of serogroups 055, 02, and 011, the most common serogroups from septicemic birds were rarely found in the environment or feces of
2650
INFECT. IMMUN.
DOZOIS ET AL. TABLE 1. 0 serogroups of E. coli isolated from colisepticemic and healthy chickens and turkeys No. of isolates (%) in:
Serogroup
Chickens
078 01 055 018 011 02 035 045 Rough
Diseased
Healthy
(52) (6.2) (2.4) (3.6) (1.2)
1 (3.4) 1 (3.4) 4 (14)
(2.4) (1.2) (2.4) (2.4) (8.4) 15 (18)
1 (3.4) 1 (3.4) 4 (17) 15 (52)
83
29
43 5 2 3 1 2 1 2 2 7
Othera Nontypeable Total
Chickens & Turkeys
Turkeys Diseased 7 3 2 1 1
2 (6.9)
(24) (10) (6.5) (3.4) (3.4)
Healthy
Diseased
2 (5.9)
50 8 4 4 2
Healthy
(45) (7.1) (3.6) (3.6) (1.8)
1 (1.6) 1 (1.6) 6 (9.5)
1 (2.9)
2 (1.8) 2 (1.8) 2 (1.8)
2 (3.2)
2 (6.9) 1 (3.4) 11 (38)
4 (12) 7 (21) 19 (56)
4 (3.6) 8 (7.1) 26 (23)
5 (7.9) 11 (17.5) 32 (54)
29
34
1 (2.9)
1 (3.4)
3 (4.8)
112
63
a Isolates from cases of colisepticemia: turkey, 0115; chicken, 015, 022, 053, 071, 083, 0131, and 0138. Isolates from healthy birds (number of isolates): turkey, 06 (2), 08 (2), 054 (2), and 021; chicken, 015, 021/053, 071, and 083.
healthy birds. Sixty-eight (81%) of 83 isolates from septicemic chickens, in contrast to only 18 (62%) of 29 isolates from septicemic turkeys, were typeable with the antisera utilized. The frequency of nontypeable isolates from septicemic turkeys was closer to that observed in isolates from the environment of healthy birds than to that observed in isolates from septicemic chickens. Relationship between origin, lethality, HA, and the presence of fimbrial genes. Isolates were examined by colony hybridization for the presence of the afa, pil, pap, and sfa gene systems by using DNA probes specific for each system. Thirty-six (44%) of 81 isolates and 9 (31%) of 29 isolates from septicemic chickens and turkeys, respectively, were positive for the pap probe (pap' isolates) (Table 2). The pap' isolates were significantly more frequent in septicemic chickens (P < 0.02) than in healthy chickens. In contrast, the pap' isolates from septicemic chickens did not appear to be directly associated with lethality in day-old chicks (Table 3). Isolates from septicemic turkeys were also more frequently pap' than isolates from healthy turkeys (Table 2), but the difference was not statistically significant (P < 0.1). Nevertheless, in isolates from septicemic turkeys, a significant difference in the frequency of pap' isolates between lethality classes 1 and 3 (P < 0.001) and among all three lethality classes (P < 0.05) (Table 4) was observed. The HA patterns of the pap' isolates were not typical of strains expressing P adhesin (Tables 5 and 6). Only four pap' isolates, from septicemic turkeys, showed MRHA of
human erythrocytes and specific Gal-Gal binding typical of P adhesins. One pap' sfa+ isolate from a diseased chicken was MRHA positive for human type OP1 and type A1Pj erythrocytes but was negative for the P-latex test. None of the other 41 pap' isolates, from either chickens or turkeys, expressed the P adhesin. Five of these pap' isolates were further passaged 10 times and still did not express the P adhesin. The four MRHA-positive isolates from septicemic turkeys and the MRHA-positive isolate from a septicemic chicken were in lethality class 1. Three of the turkey isolates belonged to serogroup 01, and one was nontypeable. The MRHA-positive isolate from a chicken was from serogroup 018. All isolates were negative for HA of ovine erythrocytes which contain the Forssman antigen specific for the F adhesin expressed on fimbriae encoded by the prs paprelated system. Sixty-four (79%) of 81 isolates and 19 (65%) of 29 isolates from septicemic chickens and turkeys, respectively, were positive for the pil probe (pil+ isolates). The pil+ isolates TABLE 3. Relationship between the presence of aerobactin and fimbrial genes, HA, origin, and lethality of chicken E. coli isolates Isolate source and
lethality
Total
classa
No. of isolates With genotype' Showing HA'
aero+ pap+ pil+ HuMRHA ChMSHA GpMSHA
Septicemic birds
TABLE 2. Presence of pap and pil sequences in E. coli isolated from chickens and turkeys No. of isolates from the following
Genotypea
Chicken
pap+ pap
pil+ pil Based
origin':
on
b Isolates were from
21 20 4
21 19 4
11 12 2
16 14 3
3 7 3
3 6
1 4
2 2 1
1
6 5 2
12 14 3
1 3 1
3 7 2
Healthy birds
Turkey
Sept
Env
Sept
Env
36 45 64 17
5 24 14 15
9 20 19 10
4 29 21 12
positive hybridization with pil and/or pap DNA probes. colisepticemic birds (Sept) or from the environment of healthy birds (Env). a
1 2 3
1 2 3
2
.5
LD_50 classes: 1,
INFECTION AND IMMUNITY, JUlY 1992, p. 2648-2656 0019-9567/92/072648-09$02.00/0 Copyright ©) 1992, American Society for Microbiology
and pil-Related DNA Sequences and Other Virulence Determinants Associated with Escherichia coli Isolated from Septicemic Chickens and Turkeys
pap-
CHARLES M. DOZOIS, JOHN M. FAIRBROTHER,* JOSEE HAREL, AND MARC BOSSE
Departement de Pathologie et de Microbiologie, Faculte de Medecine Veterinaire, Universite de Montreal, C.P. 5000, Saint-Hyacinthe, Quebec, Canada J2S 7C6 Received 16 September 1991/Accepted 7 April 1992
Escherichia coli isolates from septicemic or healthy chickens and turkeys from Quebec were serotyped, examined genotypically by using DNA probes specific for the pil and pap fimbrial systems and the aerobactin siderophore system, and examined phenotypically for lethality in day-old chicks, hemagglutination, serum resistance, and aerobactin production. Serogroups 078 and 01 were most common in septicemic chickens and turkeys. pap' isolates from chickens were associated with septicemia, and pap' isolates from turkeys were associated with lethality in day-old chicks. Four of nine pap' isolates from septicemic turkeys expressed P adhesin, whereas all pap' isolates from septicemic chickens were negative for P adhesin. The pil+ genotype was associated with septicemia in chickens and with serum resistance in isolates from turkeys. Mannose-sensitive hemagglutination of guinea pig erythrocytes was associated with septicemia in chickens and turkeys, although this phenotype was not associated with pil+ isolates from turkeys. Serum resistance was associated with isolates from septicemic turkeys and with lethality in isolates from chickens. The aerobactin system was associated with isolates from septicemic chickens and turkeys. Overall, results indicated that (i) genotypic examination may reveal virulence-associated traits which differ from the typically expected phenotype and/or are not readily expressed in vitro, and (ii) certain phenotypic and genotypic traits associated with E. coli causing extraintestinal disease in humans and animals are also associated with E. coli causing avian septicemia.
of strains from serogroups 01 and 02 (14, 17, 55). Most strains expressing either FlA or FlA-like fimbriae have been shown to adhere to chicken respiratory epithelial cells in vitro (17, 25). In addition, mannose-resistant hemagglutinating (MRHA) P or P-related fimbriae may be expressed by some E. coli of avian origin (1, 46, 56). To date, the characterization of virulence factors of E. coli from poultry has consisted mainly of phenotypic studies. Whereas phenotypic studies of virulence factors require in vitro expression, which may vary during infection of the host (4, 48), DNA hybridization techniques allow the examination of virulence factors without relying on phenotypic expression (27, 33, 43). The purpose of this study was to characterize E. coli isolated from either healthy or septicemic chickens and turkeys by examination of both phenotypic and genotypic characteristics. Virulence, the presence of related sequences for the pil (FlA), pap (P [F13]), sfa, and afa operons, MSHA and MRHA, serum resistance, and the presence and expression of aerobactin and hemolysin genes were investigated in order to establish the relationship between genotype and phenotype and to compare the frequency of these factors in pathogenic versus commensal E. coli, thus elucidating the role of these factors in the pathogenesis of colisepticemia in chickens and turkeys. Results of the study demonstrated that (i) certain phenotypes and/or genotypes were associated with septicemia and/or virulence in chickens and/or turkeys, (ii) differences were observed between isolates from septicemic chickens and turkeys, and (iii) a good correlation between genotype and phenotype was observed for the aerobactin system, but not for the pap and pil fimbrial systems.
Colibacillosis is one of the principal causes of morbidity and mortality in chickens and turkeys and causes significant economic losses to the poultry industry. Of the various forms of Eschenichia coli infection occurring in poultry, the most common syndrome involves airsacculitis, frequently followed by a generalized infection. The air sacs are the first organs affected, and extension of the infection may result in pericarditis, fibrinous perihepatitis, and an often fatal septicemia (23, 40, 54). Infection is often enhanced or initiated by predisposing agents, which include environmental and biological factors (e.g., viruses and Mycoplasma spp.) (12). Pathogenic strains belong mainly to serogroups 01, 02, and 078 (15, 28, 52). The aerobactin iron-sequestering system is frequently associated with E. coli causing extraintestinal infections in humans (13) and other animals (19, 35) and is correlated with virulence of E. coli isolates from chickens (34). In addition, extraintestinal E. coli is often significantly resistant to the bactericidal effects of normal serum. For example, a correlation between virulence and resistance to normal turkey serum has been observed in E. coli isolated from turkeys (18). The characterization of fimbriae of E. coli from poultry has centered on the examination of strains from select serogroups (01, 02, and 078) isolated from diseased birds. E. coli isolated from septicemic chickens has been shown to possess mainly 17-kDa mannose-sensitive hemagglutinating (MSHA) FlA (type 1A) fimbriae, in the case of strains from serogroup 078, and MSHA fimbriae which have been classified as FlA-like on the basis of differences in the molecular weight and immunological reactivity, in the case
*
Corresponding author. 2648
VIRULENCE DETERMINANTS OF E. COLI FROM POULTRY
VOL. 60, 1992
MATERIALS AND METHODS Bacterial strains. One hundred and seventy-five E. coli isolates were included in this study. Eighty-three isolates were from lesions of visceral organs of 2- to 52-day-old chickens with colisepticemia. Twenty-nine isolates were similarly obtained from 3- to 12-week-old turkeys with colisepticemia. Isolates of disease origin were from birds submitted to the Laboratoire de Pathologie Animale, SaintHyacinthe, Quebec, from 1987 to 1990. In addition, 29 and 34 isolates were obtained from the feces and litter of healthy chickens and turkeys, respectively; in all cases, they were from farms with no abnormal incidence of colisepticemia. 0 serotyping was done by using standard slide and tube agglutination techniques as described previously (45) with 30 antiserum samples of the most common serogroups of E. coli isolated from birds with colisepticemia. Isolates were stored on Dorset agar slants at 4°C. Unless otherwise stated, bacteria were grown for 18 h in tryptic soy broth (Difco Laboratories, Detroit, Mich.) prior to examination. E. coli reference strains included 4787 (F165+, aerobactin positive), 6894 (aerobactin positive, Hly+), and BAM (rough, F1A+). In addition, strains MC1061(pPILL1005), MC1061 (pPILL1006), and MC1061(pPILL1007) (2), carrying internal fragments of the afa, pap, and sfa operons, respectively, were used as controls and for the preparation of gene probes. E. coli K-12 HB101 was used as a negative control. Lethality test. The virulence of 101 isolates randomly selected from groups of different origins was evaluated by a lethality test in 1-day-old chicks by using the technique of Dho and Lafont (16) with some modifications. Briefly, for each isolate, groups of five 1-day-old chicks were inoculated subcutaneously with 0.5 ml of 10-, 100-, or 1,000-fold-diluted volumes of overnight culture. Bacterial counts were estimated by using a spread plate method on brain heart infusion agar (Difco). Chicks were observed for 3 days, mortalities were recorded, and the 50% lethal dose (LD50) was calculated. Isolates were grouped into three lethality classes: 1, LD50, c5 x 106 CFU ml-'; 2, LD50, 5 x 106 to 5 x 108 CFU ml-1; and 3, LD50 .5 x 108 CFU ml-'. Hemagglutination. MSHA and MRHA tests were performed essentially as described by Fairbrother et al. (19). For MRHA, isolates were grown overnight at 37°C on minimal Davis agar (Difco) plus Casamino Acids (MD-1 agar) and were tested with chicken, human 0P1, and human A1P1 erythrocytes. Isolates were also grown overnight at 37°C on tryptic soy agar-5% (vol/vol) bovine blood (bovine blood agar) and tested with chicken, bovine, ovine, human OP1, and human A1P1 erythrocytes. Hemagglutination (HA) of chicken erythrocytes was also examined with isolates grown in MD-1 broth. MRHA-positive isolates were examined for recognition of the aL-D-galactosyl-(1,4)-,-galactose (Gal-Gal) receptor, as observed for P fimbriae, by using the P-latex test (2). Certain isolates were reexamined for MRHA and P-latex agglutination after 10 passages on bovine blood agar. For MSHA, isolates were subjected to four consecutive passages for 48 h each in static tryptic soy broth at 37°C and tested with guinea pig erythrocytes in the presence or absence of 5% D-mannose. Preparation of DNA probes. DNA probes for the fimbrial operons afa, pap, and sfa were generated from the recombinant plasmids pPILL1005, pPILL1006, and pPILL1007, respectively (kindly provided by A. Labigne, Institut Pasteur, Paris, France) and were digested by the restriction endonuclease PstI as described previously (2). The Afa probe is a 1.1-kb DNA fragment internal to afaC, the Pap
2649
probe is a 0.3-kb fragment internal topapC (60), and the Sfa probe is a 0.8-kb fragment previously designated F, which incorporates part of the sfa biogenesis and structural genes (47). The FlA probe for the pil fimbrial gene cluster was a 2.2-kb BamHI-PvuII fragment of the pilC gene generated from pSH2 (kindly provided by P. E. Orndorff, North Carolina State University, Raleigh) (39). The aerobactin receptor gene probe was a 2.2-kb PvuII fragment of pABN1 (7), and the aerobactin synthesis gene probe was a 1.8-kb AvaI fragment of pABN5 (7) (both pABN1 and pABN5 were kindly provided by J. B. Neilands, University of California, Berkeley). The hemolysin probe was a 1.8-kb PvuII fragment generated from pSF4000 (61) (kindly provided by R. Levesque, Laval University, Quebec, Canada). Plasmid DNA was extracted and purified by ultracentrifugation in a cesium chloride gradient (38). After restriction enzyme digestion, the resulting fragments were separated by agarose gel electrophoresis or polyacrylamide gel electrophoresis. Appropriate fragments were cut from the gel, concentrated by ethanol precipitation, purified, and radiolabeled with [a_-32P]dCTP by using a multiprimer DNA labeling kit (Pharmacia LKB Biotechnology Inc., Baie d'Urfe, Quebec, Canada), according to the instructions of the manufacturer.
Colony hybridization. Isolates were spotted onto Luria agar (Luria broth containing 15 g of agar per liter) and incubated at 37°C for 18 h. Colonies were then transferred to Whatman 541 filter paper (Whatman, Inc., Clifton, N.J.), and the filters were processed, hybridized, and revealed by autoradiography as described previously (10). Aerobactin production. A bioassay to determine the phenotypic expression of aerobactin was performed by using a cross-feeding method (34) with slight modifications (19). Serum bactericidal assay. The bacterial survival in serum of 64 isolates randomly selected from the groups of different origins and previously tested for lethality in day-old chicks was examined by using a bactericidal assay adapted from Taylor and Kroll (59). Briefly, a 10-fold dilution of overnight culture was resuspended in tryptic soy broth and incubated under agitation for 1.5 h. A 10-,ul volume of this culture was resuspended in 990 ,u of gelatin-Veronal-buffered saline plus magnesium and calcium ions (pH 7.35) to obtain approximately 107 CFU ml-1. A 50-,ul volume of the suspension was added and mixed with 450 pul of normal serum pooled from three birds (either 15-day-old chickens or turkeys, according to the species of origin of the isolate). After mixing, the serum-bacteria suspensions were incubated at 37°C for 3 h. Viable counts were taken at hourly intervals. The demarcation between the serum resistance and the serum sensitivity of bacteria was defined as a 2-log decrease in bacterial population after 3 h of incubation. Strains were examined prior to the serum resistance assay for the production of FlA and P fimbriae by MSHA of guinea pig erythrocytes and the P-latex test, respectively. Statistical analysis. Establishment of statistical differences between and among groups was evaluated using the G test of independence (53).
RESULTS O serotyping. Among 112 isolates from cases of colisepticemia, serogroup 078 was most frequent in chickens and turkeys, followed by 01 and 055 in turkeys and 01 and 018 in chickens (Table 1). With the exception of serogroups 055, 02, and 011, the most common serogroups from septicemic birds were rarely found in the environment or feces of
2650
INFECT. IMMUN.
DOZOIS ET AL. TABLE 1. 0 serogroups of E. coli isolated from colisepticemic and healthy chickens and turkeys No. of isolates (%) in:
Serogroup
Chickens
078 01 055 018 011 02 035 045 Rough
Diseased
Healthy
(52) (6.2) (2.4) (3.6) (1.2)
1 (3.4) 1 (3.4) 4 (14)
(2.4) (1.2) (2.4) (2.4) (8.4) 15 (18)
1 (3.4) 1 (3.4) 4 (17) 15 (52)
83
29
43 5 2 3 1 2 1 2 2 7
Othera Nontypeable Total
Chickens & Turkeys
Turkeys Diseased 7 3 2 1 1
2 (6.9)
(24) (10) (6.5) (3.4) (3.4)
Healthy
Diseased
2 (5.9)
50 8 4 4 2
Healthy
(45) (7.1) (3.6) (3.6) (1.8)
1 (1.6) 1 (1.6) 6 (9.5)
1 (2.9)
2 (1.8) 2 (1.8) 2 (1.8)
2 (3.2)
2 (6.9) 1 (3.4) 11 (38)
4 (12) 7 (21) 19 (56)
4 (3.6) 8 (7.1) 26 (23)
5 (7.9) 11 (17.5) 32 (54)
29
34
1 (2.9)
1 (3.4)
3 (4.8)
112
63
a Isolates from cases of colisepticemia: turkey, 0115; chicken, 015, 022, 053, 071, 083, 0131, and 0138. Isolates from healthy birds (number of isolates): turkey, 06 (2), 08 (2), 054 (2), and 021; chicken, 015, 021/053, 071, and 083.
healthy birds. Sixty-eight (81%) of 83 isolates from septicemic chickens, in contrast to only 18 (62%) of 29 isolates from septicemic turkeys, were typeable with the antisera utilized. The frequency of nontypeable isolates from septicemic turkeys was closer to that observed in isolates from the environment of healthy birds than to that observed in isolates from septicemic chickens. Relationship between origin, lethality, HA, and the presence of fimbrial genes. Isolates were examined by colony hybridization for the presence of the afa, pil, pap, and sfa gene systems by using DNA probes specific for each system. Thirty-six (44%) of 81 isolates and 9 (31%) of 29 isolates from septicemic chickens and turkeys, respectively, were positive for the pap probe (pap' isolates) (Table 2). The pap' isolates were significantly more frequent in septicemic chickens (P < 0.02) than in healthy chickens. In contrast, the pap' isolates from septicemic chickens did not appear to be directly associated with lethality in day-old chicks (Table 3). Isolates from septicemic turkeys were also more frequently pap' than isolates from healthy turkeys (Table 2), but the difference was not statistically significant (P < 0.1). Nevertheless, in isolates from septicemic turkeys, a significant difference in the frequency of pap' isolates between lethality classes 1 and 3 (P < 0.001) and among all three lethality classes (P < 0.05) (Table 4) was observed. The HA patterns of the pap' isolates were not typical of strains expressing P adhesin (Tables 5 and 6). Only four pap' isolates, from septicemic turkeys, showed MRHA of
human erythrocytes and specific Gal-Gal binding typical of P adhesins. One pap' sfa+ isolate from a diseased chicken was MRHA positive for human type OP1 and type A1Pj erythrocytes but was negative for the P-latex test. None of the other 41 pap' isolates, from either chickens or turkeys, expressed the P adhesin. Five of these pap' isolates were further passaged 10 times and still did not express the P adhesin. The four MRHA-positive isolates from septicemic turkeys and the MRHA-positive isolate from a septicemic chicken were in lethality class 1. Three of the turkey isolates belonged to serogroup 01, and one was nontypeable. The MRHA-positive isolate from a chicken was from serogroup 018. All isolates were negative for HA of ovine erythrocytes which contain the Forssman antigen specific for the F adhesin expressed on fimbriae encoded by the prs paprelated system. Sixty-four (79%) of 81 isolates and 19 (65%) of 29 isolates from septicemic chickens and turkeys, respectively, were positive for the pil probe (pil+ isolates). The pil+ isolates TABLE 3. Relationship between the presence of aerobactin and fimbrial genes, HA, origin, and lethality of chicken E. coli isolates Isolate source and
lethality
Total
classa
No. of isolates With genotype' Showing HA'
aero+ pap+ pil+ HuMRHA ChMSHA GpMSHA
Septicemic birds
TABLE 2. Presence of pap and pil sequences in E. coli isolated from chickens and turkeys No. of isolates from the following
Genotypea
Chicken
pap+ pap
pil+ pil Based
origin':
on
b Isolates were from
21 20 4
21 19 4
11 12 2
16 14 3
3 7 3
3 6
1 4
2 2 1
1
6 5 2
12 14 3
1 3 1
3 7 2
Healthy birds
Turkey
Sept
Env
Sept
Env
36 45 64 17
5 24 14 15
9 20 19 10
4 29 21 12
positive hybridization with pil and/or pap DNA probes. colisepticemic birds (Sept) or from the environment of healthy birds (Env). a
1 2 3
1 2 3
2
.5
LD_50 classes: 1,
