INFECTION AND IMMUNITY, May 1978, p. 559-566 0019-9567/78/0020-0559$02.00/0 Copyright © 1978 American Society for Microbiology
Vol. 20, No. 2
Printed in U.S.A.
Characterization of Plasmids That Encode for the K88 Colonization Antigen PATRICIA L. SHIPLEY, CARLTON L. GYLES,t AND STANLEY FALKOW* Department of Microbiology and Immunology, University of Washington School of Medicine, Seattle, Washington 98195 Received for publication 5 December 1977
K88 antigen, an important virulence factor in porcine enteropathogenic Escherichia coli (EEC), can be transferred along with the ability to ferment the trisaccharide raffinose (Raf). The plasmids from a number of EEC strains that encode these two properties were isolated and characterized. In most strains the K88 and Raf genes were found on a single nonconjugative plasmid approximately 50 x 106 daltons in size. This plasmid core was conserved with only slight variation among the strains tested. In some transconjugants, larger conjugative plasmids were observed that were apparently recombinants between the Raf/K88 plasmid and a transfer factor. Occasionally plasmids carrying only the raffinose fermentation genes arose by deletion of a deoxyribonucleic acid segment of about 20 x 106 daltons that included the K88 antigen gene(s).
Diarrheal disease caused by the common intestinal bacterium Escherichia coli is a major cause of death among young domestic animals such as pigs, calves, and lambs. One of the salient features of the E. coli strains able to induce diarrheal disease is the ability to proliferate in the upper small intestine of the infected animal (16). Many enteropathogenic strains isolated from young pigs possess, in addition to their normal surface antigen complement, a proteinaceous antigen designated K88 (17, 24). The adhesive properties of this antigen allow K88positive bacteria to adhere to the intestinal mucosa and thereby avoid the normal clearing mechanisms of the gut. Subsequently K88+ cells may multiply and reach high population densities (9, 20). The structural gene for the K88 antigen, like those for the E. coli enterotoxin, is located on a transmissible plasmid (18). The introduction of both the K88 antigen and enterotoxin plasmids into an E. coli strain from the normal flora of a pig converts the recipient strain to pathogenicity, but neither plasmid alone will do so (20). Although a number of studies have addressed themselves to the nature and role of the K88 antigen (9-11, 19, 20, 23, 24), little is known about the genetics of K88 and the properties of the plasmid or plasmids on which these genes reside. The 1966 study of 0rskov and 0rskov (18) suggested that the transmissible K88 genes might be part of a class I transfer system (1) in which a nonconjugative plasmid carrying the
K88 structural gene is mobilized by an fi+ transfer factor plasmid. Electron microscopic studies by Bak and co-workers in 1972 (2) were in agreement with this interpretation and further suggested that the K88 and transfer genes might reside on a 50-megadalton (Mdal) composite plasmid that can dissociate into a 10-Mdal plasmid carrying the K88 structural gene and a 40Mdal transfer plasmid. However, since these plasmid entities were not observed singly in separate cells, it was not possible to make a definite assignment of phenotype to each molecular species. More recently, Smith and Parsell (21) reported that transmissible raffinose-utilizing ability (Raf) was a feature of porcine enteropathogenic strains possessing the K88 antigen. Furthermore, K88 was always transferred together with Raf from these strains. Since raffinose fermentation also appeared to dissociate from the K88 antigen under certain conditions, it was not clear whether these genes were on the same or different plasmids. The aim of this study was to further characterize the plasmids that encode the K88 antigen in porcine enteropathogenic E. coli and to determine what relationship there is between the K88, transfer, and raffinose fermentation properties. MATERIALS AND METHODS Bacterial strains. E. coli K-12 derivatives 711 (F-
lac-28 his-51 tryp-30proC23 Phe- Nalr), C600 (F- Athr-1 leu-6 thi-1 supE44 lacYl tonA21), and SF185 t Present address: Department of Veterinary Microbiology (F- Nalr derivative of W1485) were used as recipients and Immunology, University of Guelph, Ontario, Canada. in conjugation and transformation experiments. The 559
560
SHIPLEY, GYLES, AND FALKOW
strains used as a source of the plasmids in this study, with the exception of D520, were isolated from diseased pigs in Ontario, Canada. All are prototrophic and capable of fermenting raffinose. Their serogroup and antibiotic resistance patterns are shown in Table 1. Strain D520 is one of the strains used in the original studies of Stirm et al. (24) and contains the K88 plasmid from the same source as that in the studies of Bak et al. (2). It was included in this study to facilitate comparison of our results with those in the literature. Media and reagents. Bacterial cultures were grown in brain heart infusion broth (Difco) or L broth (14). Raffinose fermentation properties were determined in phenol red broth with 1% raffinose or on MacConkey agar base with 1% raffinose. Bacteria were grown on either nutrient agar (Difco) or L agar for K88 testing. Purified K88ab or K88ac antigen (24) was used to prepare rabbit anti-K88 antiserum. Selection of transconjugants or transformants capable of utilizing raffinose as a sole carbon source was made on M9 minimal salts medium (13) containing 0.5% raffinose and the appropriate amino acid and antibiotic supplements for selection of the recipient bacteria. Methods of plasmid transfer. (i) Conjugation. Donor and recipient strains were grown without aeration to a density of approximately 108 cells/ml. Equal volumes of donor and recipient cultures were mixed and incubated at 37°C. Transfer frequencies were determined after 4 h of mixed incubation. In cases where the transfer frequencies were low, transconjugants were obtained from 16-h matings. (ii) Transformation. Purified plasmid deoxyribonucleic acid (DNA) was used to transform E. coli C600 by the method of Cohen et al. (4). Preparation of plasmid DNA. Cleared lysates of bacterial cells were prepared as described by Elwell et al. (6). Plasmid DNA was purified by isopycnic centrifugation in CsCl-ethidium bromide gradients. The density was adjusted to approximately 1.625 g/cm3, and the solution was centrifuged for 40 to 48 h at 15°C and 35,000 rpm in a Beckman type 40 fixed-angle rotor. The dense plasmid DNA band was removed by dripping from the bottom of the tube, and the ethidium bromide was extracted with CsCl-saturated isopropanol. The nick translation procedure reported by Chilton et al. (3) was used to prepare labeled plasmid DNA. TABLE 1. Bacterial strains Strain Resistance pattern' Serogroup CG1121 Sm, Tc, Am 0149:K91, 88ac CG1193 Sm, Tc, Nm 0157:"V17", K88ac CG2047 Sm 0157:"V17", K88ac Sm CG82 0157:"V17", K88ac CG116 Sm, Tc 0147:K89, 88ac Sm CG139 045:K?, 88ab Nm CG140 045:K?, 88ab CG145 Tc, Sm 0149:K91, 88ac CG146 Tc, Sm, Am 0149:K91, 88ac CG151 Sm, Cm 0149:K91, 88ac CG152 Cm, Nm 0149:K91, 88ac CG157 Tc, Sm, Cm 0149:K91, 88ac D520 08:K27(A)-, 88ab a Sm, Streptomycin; Tc, tetracycline; Am, ampicillin; Nm, neomycin; Cm, chloramphenicol.
INFECT. IMMUN.
Agarose gel electrophoresis. Agarose gels (0.7%) were prepared and run as described by Meyers et al. (15). For survey purposes, the sodium dodecyl sulfate-salt precipitation technique (7) was used to prepare cleared lysates from 30-ml bacterial cultures. These lysates were partially purified by ribonuclease digestion followed by phenol extraction. Salt was added to 0.3 M, and the DNA was precipitated in ethanol at -20°C. The DNA precipitate was suspended in 0.1 ml of distilled water, and samples of 4 to 25 pl were run on the gel. Plasmids of known molecular weights or EcoRI-cleaved fragments of F DNA were used as standards for molecular-weight determinations of whole plasmids or endonuclease-cut fragments, respectively. Extraction of unlabeled DNA for hybridization. Unlabeled whole-cell DNA from plasmid-containing strains was prepared by the method of So et al. (22). DNA-DNA duplex studies. Labeled and unlabeled DNAs were sheared by sonic treatment in a Branson model 140D sonic oscillator for 5 min at 4°C and dialyzed against 0.42 M NaCl. Hybridization and the Sl-nuclease reaction to determine the degree of duplex formation were carried out according to the procedure of Crosa et al. (5). To determine the percentage of homology among plasmid species, the degree of duplex formation was calculated, the chromosomal control was subtracted, and each heterologous reaction was divided by the homologous reaction. DNA contour length. The contour length of plasmid DNA molecules was measured by spreading the DNA according to the Kleinschmidt technique (12). Negatives of photographs of isolated open circular molecules were projected through a photographic enlarger and traced by hand. The contour lengths were measured with an Electronic Graphics Calculator (Numonics Corp.). The molecular weight was determined by comparison with a plasmid of known molecular weight spread on the same grid and photographed at the same magnification. Restriction endonuclease digestion. Endonuclease digestion reactions were carried out in a volume of 30 ,ul with 0.1 to 0.5 ,ug of purified plasmid DNA. EcoRI reactions contained 0.1 M tris(hydroxymethyl)aminomethane (Tris), pH 7.4, 0.05 M NaCl, and 0.01 M MgCl2. HindIII reactions contained 0.05 M NaCl, 0.005 M MgCl2, 50 ,ug of bovine serum albumin per ml, and 0.006 M Tris, pH 7.4.
RESULTS Coincidence of raffinose fermentation and K88 antigen. Initially, 31 enterotoxigenic strains isolated from diseased pigs in Ontario, Canada, were screened for the production of the K88 antigen and the ability to ferment raffinose. Twenty-two of these strains were K88 positive. All of the K88-positive strains fermented raffinose with the production of acid and gas. Among the K88-negative strains, two utilized raffinose with the production of acid only, whereas the remaining seven were raffinose negative. Eighteen Raf+ K88+ strains were tested for their ability to transfer the raffinose fermentation prop-
VOL. 20, 1978
PLASMIDS CODING FOR K88 ANTIGEN
561
erty to an E. coli K-12 recipient, and the 12 duce enterotoxin was transferred by three strains that did so at a detectable frequency strains. In several instances where the frequency of Raf transfer was low, all transconjugants were selected for further study. Transfer of Raf and K88. Matings were tested had acquired either enterotoxin or resistcarried out between the Raf+ K88+ strains and ance genes. Not enough testing was done to E. coli K-12 recipients, using chromosomal nal- determine the frequency of cotransfer of Ent or idixic acid resistance to select for the recipient antibiotic resistance with Raf; however, it was cells after mating. Since selection can be made clear that although cotransfer was frequent from for the ability to utilize raffinose but not for K88 some donors, the overall incidence of cotransfer antigen production, frequencies were deter- did not approach that seen with the K88 antigen. mined for Raf transfer, and the raffinose-ferPlasmid content of Raf+ K88+ transconmenting transconjugants were tested by slide jugants. The high incidence of cotransfer of the agglutination for K88. The 13 donor strains K88 antigen and raffinose fermentation sugtransferred Raf at frequencies ranging from gested that these two properties might be lo10-2 per recipient cell (Table 2). In cated on a single plasmid. To investigate this each case, Raf transfer was accompanied by the possibility, the plasmid content of the K-12 K88+ acquisition of the K88 antigen. For most of the Raf+ transconjugant strains was surveyed by matings there was 100% cotransfer of the raffi- agarose gel electrophoresis of partially purified nose fermentation and K88 antigen properties; lysates. At least one transconjugant from each however, single isolates that had received the Raf+ K88+ donor was included. Where possible, Raf character alone were obtained from several strains were chosen that had received only the matings (Table 2). Two of these strains will be raffinose and K88 markers. A representative gel discussed in detail later in this report. is shown in Fig. 1. A summary of the results of With the exception of the D520 strain, the a number of gels is shown in Table 3. Moleculardonors in these matings had several character- weight values above 65 Mdal in this table were istics in addition to raffinose fermentation and confirmed by electron microscopy since molecK88 antigen production that are typical of por- ular-weight differences in this range are difficult cine enteropathogenic E. coli strains and may to determine under the electrophoresis condibe carried on plasmids. These include produc- tions used. Several Raf+ K88+ strains contained tion of enterotoxins and hemolysins and anti- only one plasmid, confirming the linkage bebiotic resistance. Therefore, many of the raffi- tween Raf and K88 (Table 3). Although the nose-fermenting transconjugants were tested to plasmid content varied even among strains that had apparently acquired only raffinose and K88, see whether they had also acquired one or more of these characteristics (Table 2). No Raf+ trans- it is apparent that every strain carried either a conjugants were found to have received the abil- plasmid of approximately 50 Mdal or one of ity to produce hemolysin; however, seven donors about 90 Mdal. Since each of these was found as transferred antibiotic resistance along with raf- the only plasmid in at least one strain, they finose fermentation, whereas the ability to pro- probably represent two forms of Raf/K88 plasTABLE 2. Transfer of raffinose fermentation and other markers to E. coli K-12 Donor
D520 CG82 CG116 CG139 CG140 CG145 CG146 CG151 CG152 CG157 CG1121 CG1193 CG2047
Cotransfer of other markers: frequency~~ K88 cotransfer (no. Raf transfer frequency" h K88+/no. tested).
3.0 x 10-2 1.5 x 10-4 4.3 x 10-5 2.3 x 10-3 4.8 x 10-:3 3.2 x 10-5 7.8 x 105.7 x 10-6
Vol. 20, No. 2
Printed in U.S.A.
Characterization of Plasmids That Encode for the K88 Colonization Antigen PATRICIA L. SHIPLEY, CARLTON L. GYLES,t AND STANLEY FALKOW* Department of Microbiology and Immunology, University of Washington School of Medicine, Seattle, Washington 98195 Received for publication 5 December 1977
K88 antigen, an important virulence factor in porcine enteropathogenic Escherichia coli (EEC), can be transferred along with the ability to ferment the trisaccharide raffinose (Raf). The plasmids from a number of EEC strains that encode these two properties were isolated and characterized. In most strains the K88 and Raf genes were found on a single nonconjugative plasmid approximately 50 x 106 daltons in size. This plasmid core was conserved with only slight variation among the strains tested. In some transconjugants, larger conjugative plasmids were observed that were apparently recombinants between the Raf/K88 plasmid and a transfer factor. Occasionally plasmids carrying only the raffinose fermentation genes arose by deletion of a deoxyribonucleic acid segment of about 20 x 106 daltons that included the K88 antigen gene(s).
Diarrheal disease caused by the common intestinal bacterium Escherichia coli is a major cause of death among young domestic animals such as pigs, calves, and lambs. One of the salient features of the E. coli strains able to induce diarrheal disease is the ability to proliferate in the upper small intestine of the infected animal (16). Many enteropathogenic strains isolated from young pigs possess, in addition to their normal surface antigen complement, a proteinaceous antigen designated K88 (17, 24). The adhesive properties of this antigen allow K88positive bacteria to adhere to the intestinal mucosa and thereby avoid the normal clearing mechanisms of the gut. Subsequently K88+ cells may multiply and reach high population densities (9, 20). The structural gene for the K88 antigen, like those for the E. coli enterotoxin, is located on a transmissible plasmid (18). The introduction of both the K88 antigen and enterotoxin plasmids into an E. coli strain from the normal flora of a pig converts the recipient strain to pathogenicity, but neither plasmid alone will do so (20). Although a number of studies have addressed themselves to the nature and role of the K88 antigen (9-11, 19, 20, 23, 24), little is known about the genetics of K88 and the properties of the plasmid or plasmids on which these genes reside. The 1966 study of 0rskov and 0rskov (18) suggested that the transmissible K88 genes might be part of a class I transfer system (1) in which a nonconjugative plasmid carrying the
K88 structural gene is mobilized by an fi+ transfer factor plasmid. Electron microscopic studies by Bak and co-workers in 1972 (2) were in agreement with this interpretation and further suggested that the K88 and transfer genes might reside on a 50-megadalton (Mdal) composite plasmid that can dissociate into a 10-Mdal plasmid carrying the K88 structural gene and a 40Mdal transfer plasmid. However, since these plasmid entities were not observed singly in separate cells, it was not possible to make a definite assignment of phenotype to each molecular species. More recently, Smith and Parsell (21) reported that transmissible raffinose-utilizing ability (Raf) was a feature of porcine enteropathogenic strains possessing the K88 antigen. Furthermore, K88 was always transferred together with Raf from these strains. Since raffinose fermentation also appeared to dissociate from the K88 antigen under certain conditions, it was not clear whether these genes were on the same or different plasmids. The aim of this study was to further characterize the plasmids that encode the K88 antigen in porcine enteropathogenic E. coli and to determine what relationship there is between the K88, transfer, and raffinose fermentation properties. MATERIALS AND METHODS Bacterial strains. E. coli K-12 derivatives 711 (F-
lac-28 his-51 tryp-30proC23 Phe- Nalr), C600 (F- Athr-1 leu-6 thi-1 supE44 lacYl tonA21), and SF185 t Present address: Department of Veterinary Microbiology (F- Nalr derivative of W1485) were used as recipients and Immunology, University of Guelph, Ontario, Canada. in conjugation and transformation experiments. The 559
560
SHIPLEY, GYLES, AND FALKOW
strains used as a source of the plasmids in this study, with the exception of D520, were isolated from diseased pigs in Ontario, Canada. All are prototrophic and capable of fermenting raffinose. Their serogroup and antibiotic resistance patterns are shown in Table 1. Strain D520 is one of the strains used in the original studies of Stirm et al. (24) and contains the K88 plasmid from the same source as that in the studies of Bak et al. (2). It was included in this study to facilitate comparison of our results with those in the literature. Media and reagents. Bacterial cultures were grown in brain heart infusion broth (Difco) or L broth (14). Raffinose fermentation properties were determined in phenol red broth with 1% raffinose or on MacConkey agar base with 1% raffinose. Bacteria were grown on either nutrient agar (Difco) or L agar for K88 testing. Purified K88ab or K88ac antigen (24) was used to prepare rabbit anti-K88 antiserum. Selection of transconjugants or transformants capable of utilizing raffinose as a sole carbon source was made on M9 minimal salts medium (13) containing 0.5% raffinose and the appropriate amino acid and antibiotic supplements for selection of the recipient bacteria. Methods of plasmid transfer. (i) Conjugation. Donor and recipient strains were grown without aeration to a density of approximately 108 cells/ml. Equal volumes of donor and recipient cultures were mixed and incubated at 37°C. Transfer frequencies were determined after 4 h of mixed incubation. In cases where the transfer frequencies were low, transconjugants were obtained from 16-h matings. (ii) Transformation. Purified plasmid deoxyribonucleic acid (DNA) was used to transform E. coli C600 by the method of Cohen et al. (4). Preparation of plasmid DNA. Cleared lysates of bacterial cells were prepared as described by Elwell et al. (6). Plasmid DNA was purified by isopycnic centrifugation in CsCl-ethidium bromide gradients. The density was adjusted to approximately 1.625 g/cm3, and the solution was centrifuged for 40 to 48 h at 15°C and 35,000 rpm in a Beckman type 40 fixed-angle rotor. The dense plasmid DNA band was removed by dripping from the bottom of the tube, and the ethidium bromide was extracted with CsCl-saturated isopropanol. The nick translation procedure reported by Chilton et al. (3) was used to prepare labeled plasmid DNA. TABLE 1. Bacterial strains Strain Resistance pattern' Serogroup CG1121 Sm, Tc, Am 0149:K91, 88ac CG1193 Sm, Tc, Nm 0157:"V17", K88ac CG2047 Sm 0157:"V17", K88ac Sm CG82 0157:"V17", K88ac CG116 Sm, Tc 0147:K89, 88ac Sm CG139 045:K?, 88ab Nm CG140 045:K?, 88ab CG145 Tc, Sm 0149:K91, 88ac CG146 Tc, Sm, Am 0149:K91, 88ac CG151 Sm, Cm 0149:K91, 88ac CG152 Cm, Nm 0149:K91, 88ac CG157 Tc, Sm, Cm 0149:K91, 88ac D520 08:K27(A)-, 88ab a Sm, Streptomycin; Tc, tetracycline; Am, ampicillin; Nm, neomycin; Cm, chloramphenicol.
INFECT. IMMUN.
Agarose gel electrophoresis. Agarose gels (0.7%) were prepared and run as described by Meyers et al. (15). For survey purposes, the sodium dodecyl sulfate-salt precipitation technique (7) was used to prepare cleared lysates from 30-ml bacterial cultures. These lysates were partially purified by ribonuclease digestion followed by phenol extraction. Salt was added to 0.3 M, and the DNA was precipitated in ethanol at -20°C. The DNA precipitate was suspended in 0.1 ml of distilled water, and samples of 4 to 25 pl were run on the gel. Plasmids of known molecular weights or EcoRI-cleaved fragments of F DNA were used as standards for molecular-weight determinations of whole plasmids or endonuclease-cut fragments, respectively. Extraction of unlabeled DNA for hybridization. Unlabeled whole-cell DNA from plasmid-containing strains was prepared by the method of So et al. (22). DNA-DNA duplex studies. Labeled and unlabeled DNAs were sheared by sonic treatment in a Branson model 140D sonic oscillator for 5 min at 4°C and dialyzed against 0.42 M NaCl. Hybridization and the Sl-nuclease reaction to determine the degree of duplex formation were carried out according to the procedure of Crosa et al. (5). To determine the percentage of homology among plasmid species, the degree of duplex formation was calculated, the chromosomal control was subtracted, and each heterologous reaction was divided by the homologous reaction. DNA contour length. The contour length of plasmid DNA molecules was measured by spreading the DNA according to the Kleinschmidt technique (12). Negatives of photographs of isolated open circular molecules were projected through a photographic enlarger and traced by hand. The contour lengths were measured with an Electronic Graphics Calculator (Numonics Corp.). The molecular weight was determined by comparison with a plasmid of known molecular weight spread on the same grid and photographed at the same magnification. Restriction endonuclease digestion. Endonuclease digestion reactions were carried out in a volume of 30 ,ul with 0.1 to 0.5 ,ug of purified plasmid DNA. EcoRI reactions contained 0.1 M tris(hydroxymethyl)aminomethane (Tris), pH 7.4, 0.05 M NaCl, and 0.01 M MgCl2. HindIII reactions contained 0.05 M NaCl, 0.005 M MgCl2, 50 ,ug of bovine serum albumin per ml, and 0.006 M Tris, pH 7.4.
RESULTS Coincidence of raffinose fermentation and K88 antigen. Initially, 31 enterotoxigenic strains isolated from diseased pigs in Ontario, Canada, were screened for the production of the K88 antigen and the ability to ferment raffinose. Twenty-two of these strains were K88 positive. All of the K88-positive strains fermented raffinose with the production of acid and gas. Among the K88-negative strains, two utilized raffinose with the production of acid only, whereas the remaining seven were raffinose negative. Eighteen Raf+ K88+ strains were tested for their ability to transfer the raffinose fermentation prop-
VOL. 20, 1978
PLASMIDS CODING FOR K88 ANTIGEN
561
erty to an E. coli K-12 recipient, and the 12 duce enterotoxin was transferred by three strains that did so at a detectable frequency strains. In several instances where the frequency of Raf transfer was low, all transconjugants were selected for further study. Transfer of Raf and K88. Matings were tested had acquired either enterotoxin or resistcarried out between the Raf+ K88+ strains and ance genes. Not enough testing was done to E. coli K-12 recipients, using chromosomal nal- determine the frequency of cotransfer of Ent or idixic acid resistance to select for the recipient antibiotic resistance with Raf; however, it was cells after mating. Since selection can be made clear that although cotransfer was frequent from for the ability to utilize raffinose but not for K88 some donors, the overall incidence of cotransfer antigen production, frequencies were deter- did not approach that seen with the K88 antigen. mined for Raf transfer, and the raffinose-ferPlasmid content of Raf+ K88+ transconmenting transconjugants were tested by slide jugants. The high incidence of cotransfer of the agglutination for K88. The 13 donor strains K88 antigen and raffinose fermentation sugtransferred Raf at frequencies ranging from gested that these two properties might be lo10-2 per recipient cell (Table 2). In cated on a single plasmid. To investigate this each case, Raf transfer was accompanied by the possibility, the plasmid content of the K-12 K88+ acquisition of the K88 antigen. For most of the Raf+ transconjugant strains was surveyed by matings there was 100% cotransfer of the raffi- agarose gel electrophoresis of partially purified nose fermentation and K88 antigen properties; lysates. At least one transconjugant from each however, single isolates that had received the Raf+ K88+ donor was included. Where possible, Raf character alone were obtained from several strains were chosen that had received only the matings (Table 2). Two of these strains will be raffinose and K88 markers. A representative gel discussed in detail later in this report. is shown in Fig. 1. A summary of the results of With the exception of the D520 strain, the a number of gels is shown in Table 3. Moleculardonors in these matings had several character- weight values above 65 Mdal in this table were istics in addition to raffinose fermentation and confirmed by electron microscopy since molecK88 antigen production that are typical of por- ular-weight differences in this range are difficult cine enteropathogenic E. coli strains and may to determine under the electrophoresis condibe carried on plasmids. These include produc- tions used. Several Raf+ K88+ strains contained tion of enterotoxins and hemolysins and anti- only one plasmid, confirming the linkage bebiotic resistance. Therefore, many of the raffi- tween Raf and K88 (Table 3). Although the nose-fermenting transconjugants were tested to plasmid content varied even among strains that had apparently acquired only raffinose and K88, see whether they had also acquired one or more of these characteristics (Table 2). No Raf+ trans- it is apparent that every strain carried either a conjugants were found to have received the abil- plasmid of approximately 50 Mdal or one of ity to produce hemolysin; however, seven donors about 90 Mdal. Since each of these was found as transferred antibiotic resistance along with raf- the only plasmid in at least one strain, they finose fermentation, whereas the ability to pro- probably represent two forms of Raf/K88 plasTABLE 2. Transfer of raffinose fermentation and other markers to E. coli K-12 Donor
D520 CG82 CG116 CG139 CG140 CG145 CG146 CG151 CG152 CG157 CG1121 CG1193 CG2047
Cotransfer of other markers: frequency~~ K88 cotransfer (no. Raf transfer frequency" h K88+/no. tested).
3.0 x 10-2 1.5 x 10-4 4.3 x 10-5 2.3 x 10-3 4.8 x 10-:3 3.2 x 10-5 7.8 x 105.7 x 10-6
