APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Sept. 1993, p. 2830-28360099-2240/93/092830-07$02.00/0Copyright © 1993, American Society for Microbiology

Plasmid and Serological Differences betweenEdwardsiella ictaluri Strains

CRAIG J. LOBB,* SEYED H. GHAFFARI, J. RUSSELL HAYMAN,AND DEXTER T. THOMPSON

Department of Microbiology, University of Mississippi Medical Center,Jackson, Mississippi 39216-4505

Received 25 January 1993/Accepted 11 June 1993

Several studies have shown that isolates ofEdwardsiella ictaluri obtained from infected channel catfish in thesoutheastern United States harbor two cryptic plasmids, designated pCL1 (5.7 kb) and pCL2 (4.9 kb). Theseisolates appear to be serologically homogeneous. To extend these studies, we focused our analyses on twoisolates of nonictalurid origin. Plasmid analyses of a danio isolate showed that it harbored plasmids which weresimilar if not identical to pCL1 and pCL2. This strain was also serologically indistinguishable from thoseisolated from channel catfish. In contrast, a green knife fish (GNF) isolate harbored four plasmids with relativemobilities of 6.0, 5.7, 4.1, and 3.1 kb. Southern blot analyses indicated that only the 5.7- and 4.1-kb plasmidsstrongly hybridized under high-stringency conditions to probes specific for pCL1 and pCL2, respectively. TheGNF isolate showed minimal reactivity when reacted with polyclonal antiserum prepared against a channelcatfish isolate. However, polyclonal antiserum to the GNF isolate strongly reacted with the GNF isolate in bothsurface fluorescence and agglutination reactions. Sodium dodecyl sulfate-polyacrylamide gel electrophoresisanalyses of cell lysates showed that the protein banding patterns of the strains compared were similar. However,Western blots of proteinase K-digested cell extracts showed that 0 antigen of the GNF isolate was antigenicallydistinct from the 0 antigen of the other isolates. These studies indicate that there are different serotypes of E.ictaluri and suggest that plasmid and serological analyses of future isolates ofE. ictaluri can be used to determinewhether structurally distinct strains are emerging in major channel catfish aquaculture areas.

Edwardsiella ictaluri was first identified and characterizedin 1981 (4) and can be readily distinguished from the morecommon member of the genus, E. tarda, by biochemicalanalyses (15). E. ictalun is the bacterial pathogen responsi-ble for enteric septicemia of catfish (ESC) (3). ESC has beenviewed as an acute septicemia that can progress rapidly inapparently healthy, fast-growing catfish and can result inextensive mortality. ESC has become a significant problemin the aquaculture of channel catfish in the southeasternUnited States.The incidence of disease-associated mortality occurs

within the optimum temperature range for in vitro growth ofthe bacteria (22 to 28°C). When the bacterium is isolatedfrom kidney, brain, blood, etc., it generally requires 48 h at30°C to form typical colonies approximately 2 mm in diam-eter. This relatively slow rate of growth initiated our earlierstudies to determine the plasmid profile of different isolatesof E. ictaluri obtained from infected channel catfish raised inthe southeastern United States (7). The analyses showedthat two plasmids, designated pCL1 (5.7 kb) and pCL2 (4.9kb), were both present in each of the E. ictaluri strainsexamined. The two plasmids were isolated and mapped withvarious restriction enzymes. These analyses indicated thatthe restriction maps of the two plasmids were distinct; thisapparent lack of homology was further supported by theinability to visualize possible cointegrate structures whenplasmid pools were analyzed by agarose electrophoresis.These analyses strongly suggest that plasmid analysis shouldbe an effective method for rapidly identifying isolates of E.ictalun from channel catfish cultured in the southeastern

United States. Since this report other investigators haveconfirmed these results (8, 10).

Strains of E. ictaluri isolated from infected catfish havebeen shown to be homogeneous when analyzed in a batteryof biochemical tests (14). In addition, serological analysesusing monoclonal antibodies have indicated that differentisolates of E. ictaluri appear to be serologically homoge-neous (9). This general conclusion has led to studies directedtoward vaccination approaches to control ESC in culturedchannel catfish. However, there have now been severalreports of E. ictaluri infecting other fish besides channelcatfish. We focused our analyses on two of these isolates.The first isolate was from an infected danio (Danio devario)which was cultured in the southeastern United States (14).The second isolate was from an infected South Americangreen knife fish (GNF) (Eigemannia virescens) that wasmaintained in laboratories in La Jolla, Calif. (4a). This strainwas recovered in pure culture from the kidneys of fish dyingfrom an acute gram-negative bacteremia. Earlier studies hadindicated that the plasmid profile of the GNF isolate wasapparently distinct, bearing plasmids of different relativemobilities. The plasmid profile of the danio isolate, however,was similar to the profiles of isolates recovered from infectedchannel catfish (8, 11). This apparent difference in plasmidprofiles suggested that additional structural differences be-tween E. ictalun strains might exist. These studies have ledto this report, which reaffirms the value of plasmid analysisin characterizing E. ictaluri isolates and, equally important,establishes that there are different serotypes of E. ictaluri.

MATERIALS AND METHODS

Bacterial strains and culture conditions. The isolates of E.ictaluri recovered from infected channel catfish (Ictalurus* Corresponding author.

2830

Vol. 59, No. 9

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 21

Feb

ruar

y 20

22 b

y 22

2.11

3.99

.190

.

PLASMID AND SEROLOGICAL DIFFERENCES IN E. ICTALURI 2831

punctatus) were identified and characterized earlier (7). Thestrains are designated S85-1377, S85-1381, AL 84-142, andAL 84-187. The type strain ATCC 33202, originally isolatedin Georgia from an infected channel catfish, was obtainedfrom the American Type Culture Collection, Rockville, Md.The danio and GNF isolates of E. ictaluri were obtainedfrom James Bertolini and were designated 10.15 and 10.16,respectively, in an earlier study (1). Biochemical character-istics of the danio and GNF strains were reported (1), andseparate analyses confirmed the identification of thesestrains as E. ictaluri. Characteristic negative biochemicalreactions included indole, motility at 37°C, H2S productionon triple sugar iron medium, malonate utilization, Simmonscitrate, Voges-Proskauer, urease, lactose, adonitol, dulcitol,sorbitol, arabinose, and arginine dihydrolase. Prior to anal-ysis, the cells were recovered from frozen stocks maintainedat -70°C and cultured at 30°C in brain heart infusion medium(Difco Laboratories, Detroit, Mich.).

Plasmid and hybridization analyses. The methods used toobtain plasmid pools from the E. ictaluri isolates used in thisreport, as well as the agarose gel electrophoresis conditionsused, were identical to those reported earlier (7). For hybrid-ization analyses, pCL1 and pCL2 were first resolved byagarose gel electrophoresis, and the supercoiled plasmidswere separately recovered from individual gel slices byelectroelution. To ensure purity, the plasmids were sub-jected to a second round of purification, in which thesupercoiled plasmids were linearized with EcoRI, separatedby electrophoresis, and subsequently electroeluted fromagarose gels. The purified plasmids were radiolabeled to aspecific activity of about 1 x 109 cpm/,ug by using a randomhexamer commercial kit (Amersham, Arlington Heights,Ill.). The methods used for Southern blot analysis and thehigh-stringency hybridization conditions used followed pro-cedures described in detail elsewhere (2).

Antigenic analyses and immunoassays. Murine polyclonalantisera were prepared to isolate S85-1377 as well as to theGNF isolate of E. ictaluri by the following procedure.Bacteria were inoculated from frozen stocks into 5 ml ofbrain heart infusion medium and grown for 15 h at about30°C. The pelleted cells were washed three times in sterile0.135 M NaCl at 4°C, and the cell pellet was resuspended in10 ml of the saline solution. Six BALB/c mice, obtained froma colony maintained at the Department of Microbiology ofthe University of Mississippi Medical Center, were eachinjected subcutaneously with a total of 0.2 ml of the cellsuspension. The mice were boosted three times with freshlycultured cells prepared in the same manner at weeklyintervals. Concomitant with the last injection the animalswere injected with 0.5 ml of pristane (2,6,10,14-tetraethyl-pentadecane; Sigma Chemical Co., St. Louis, Mo.), andascitic fluid was induced with the SP2/0 murine myeloma lineas described previously (5). Ascitic fluid was clarified bycentrifugation, and the supernatant was frozen at -20°Cuntil used.The agglutination titer of each polyclonal antiserum was

determined in 96-well microtiter plates (Flow Laboratories,McLean, Va.) by using freshly cultured bacteria. Briefly,15-h cultures of E. ictaluri isolates were washed three timesin phosphate-buffered saline (PBS) and standardized to anoptical density at 550 nm between 0.95 and 1.0. The poly-clonal antiserum was serially diluted twofold in the microti-ter plates, and an equal amount of the washed bacteria wasadded. The plates were agitated and kept at 4°C overnight.Agglutination titers were expressed as the reciprocal of the

last log2 dilution of antiserum which yielded positive agglu-tination.

Indirect fluorescent-antibody analysis of E. ictaluni iso-lates used both the anti-S85-1377 antiserum and the anti-GNF isolate antiserum. Briefly, fresh cultures of bacteriawere prepared as described above, and 100-,ul aliquots ofwashed cells were reacted with an equal amount of anti-serum that had been diluted 10-fold in PBS. After a 20-minincubation at 4°C the cells were washed and reacted for anadditional 20 min with fluorescein isothiocyanate (FITC)-conjugated rabbit anti-mouse immunoglobulin G (IgG) (Ac-curate Chemical, Westbury, N.Y.). The cells were washedand examined microscopically, and the relative fluorescentintensity of the cells was determined qualitatively as con-trasted with bacteria that had not been reacted with theprimary antisera.Sodium dodecyl sulfate-polyacrylamide gel electrophore-

sis (SDS-PAGE) was done by standard procedures as re-ported in detail elsewhere (6). Cell lysates of the E. ictalunistrains were prepared from fresh cultures as describedabove. Two milliliters of bacteria, standardized to an opticaldensity at 600 nm of 0.3, was pelleted in a microcentrifugetube and washed in Tris-buffered saline, and 200 ,ul ofSDS-PAGE sample buffer was added. The cells were heatedat 100°C for 2 min, and any insoluble matter was removedwith a sterile toothpick. Samples of cell lysates were ana-lyzed directly by SDS-PAGE or, alternatively, incubated for1 h at 60°C with 50 ,ug of proteinase K (Sigma) prior toSDS-PAGE analysis. Western blotting (immunoblotting) ofreplicate lysates was performed by standard methods (13)using 1:100 dilutions of primary polyclonal antiserum anddeveloped with alkaline phosphatase-conjugated goat anti-mouse IgG (Promega, Madison, Wis.).

RESULTS

Plasmid profile of different E. ictaluri isolates. PlasmidDNA was prepared from the GNF and danio isolates of E.ictaluni and compared with plasmid DNA prepared fromrepresentative isolates of E. ictaluri recovered from infectedchannel catfish (Fig. la, lanes A through G). Each of thechannel catfish isolates had previously been shown to harborboth pCL1 (5.7 kb) and pCL2 (4.9 kb) (7). The plasmidprofile of the danio isolate was identical to the plasmidprofile of the channel catfish isolates; two plasmids, similarin size to pCLl and pCL2, were observed. The plasmidprofile of the GNF isolate, however, was distinct. The GNFisolate contained four plasmids; the relative mobilities ofthese plasmids were 6.0, 5.7, 4.1, and 3.1 kb, as judged byinterpolation from the reference plasmid ladder. Only theGNF 5.7-kb plasmid had a mobility similar to that of one ofthe known plasmids (pCL1).To determine the relationship of the plasmids in the GNF

and danio isolates to pCL1 and pCL2, as well as to resolvequestions regarding the potential similarity of pCL1 andpCL2, pCL1 and pCL2 were separately purified from agar-ose gels and radiolabeled to the same specific activity. Theradiolabeled pCL1 and pCL2 probes were then hybridizedseparately to replicate Southern blots containing the plasmidDNA from the E. ictaluri isolates. In addition, plasmid poolsof pCL1 and pCL2 derived from strain S85-1377 were

restricted with enzymes predicted from the previously de-rived restriction maps to yield fragments for specific regionsof both plasmids. Hybridization analyses showed that pCL1or pCL2 did not cross-hybridize with one another, nor didthe probes cross-hybridize with the bacterial chromosome

VOL. 59, 1993

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 21

Feb

ruar

y 20

22 b

y 22

2.11

3.99

.190

.

2832 LOBB ET AL.

bABCD E FGH I J; $~~ ~~~itfs 0.00

_jEE.t~~~~~.it O,0

CA BC D E

i*~

Xho I

05

pCL 1.a "

-4.0 { l"

ago

e

4.0 1D u_

pCL2 _,N

&S1.5 M.E Eco R I

30'

FIG. 1. Plasmid profile and Southern blot analyses of different strains of E. ictaluri hybridized with probes specific for pCL1 and pCL2.(a) Lane Ml, plasmid ladder (the sizes of the reference plasmids [in kilobase pairs] are indicated on the left); lanes A to G, plasmids detectedin E. ictaluri ATCC 33202, S85-1381, S85-1377, AL84-142, and AL84-187, danio isolate 10.15, and GNF isolate 10.16, respectively; lanes Hto J, plasmids ofATCC 33202 restricted with EcoRV plus BamHI (lane H), BglI (lane I), and EcoRI (lane J), respectively; lane M2, PstI digestof lambda DNA (the sizes of the standard linear fragments [in kilobase pairs] are indicated on the right). The gel was stained with ethidiumbromide. (b) Replicate Southern blot of the gel depicted in panel a hybridized with pCL2. (c) Replicate Southern blot of the gel depicted inpanel a hybridized with pCL1. The lanes in panels b and c contained the same preparations as the corresponding lanes in panel a. Southernblot hybridization was performed under high-stringency conditions (see Materials and Methods). (d and e) Restriction maps of pCL1 andpCL2, respectively, as determined previously by Lobb and Rhoades (7).

(Fig. lb and c). The supercoiled and linearized plasmids, aswell as the restriction fragments derived from either pCLland pCL2, hybridized only with probes derived from therespective plasmid.

In addition, pCLl and pCL2 probes hybridized to the 5.7-and 4.9-kb plasmids, respectively, of the danio isolate.Because these danio plasmids have the same relative mobil-ity as pCL1 and pCL2 and hybridized with probes specificfor pCL1 and pCL2, respectively, under stringent hybridiza-tion conditions, it seems highly probably that these plasmidsare very similar to pCL1 and pCL2. The GNF isolate, whichcontained four plasmids of different mobilities, was alsofound to have two plasmids which shared similarity to pCLl

and pCL2. The GNF 5.7-kb plasmid hybridized to probesderived from pCL1, and the GNF 4.1-kb plasmid hybridizedto probes derived from pCL2. These results are consistentwith the conclusion that there is a high degree of similaritybetween pCL1 and the GNF 5.7-kb plasmid, as well as a highdegree of similarity between pCL2 and the GNF 4.1-kbplasmid. The GNF 6.0- and 3.1-kb plasmids did not stronglyhybridize to probes derived from either pCL1 or pCL2.The GNF isolate harbors two other plasmids besides those

similar to pCL1 and pCL2. It was important to determinewhether the 6.0- and 3.1-kb GNF plasmids, which did notstrongly hybridize with probes derived from either pCL1 orpCL2, were probably different from each other. Plasmid

a

14.2

12.1

10.1

FGH I J

4.

11.5

4.7

2.8

2.5

2.0

1.7

1.1

2.

0.8

APPL. ENvIRON. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 21

Feb

ruar

y 20

22 b

y 22

2.11

3.99

.190

.

PLASMID AND SEROLOGICAL DIFFERENCES IN E. ICTALURI 2833

TABLE 1. Agglutination titers and relative fluorescent intensitiesof isolates of E. ictaluri reacted with two antisera

Anti-S85-1377 Anti-GNF strain 10.16antiserum antiserum

E. ictaluriisolate Aggltt Fluores- Aggltint Fluores-~-~uma ion cent in- cent in-

titera tensityb titera tensity"ATCC 33202 64 +++ 4 +S85-1377 128 +++ 4 +Danio strain 10.15 64 +++ 4 +GNF strain 10.16 4 + >4,096 +++

a Agglutination titer reported as the reciprocal of the highest dilution ofantiserum able to cause agglutination of bacterial cells as determined inmicrotiter plate assays (see Materials and Methods).

b The fluorescent intensity of the bacterial cells was scored after reactionwith the primary antiserum and rabbit anti-mouse FITC-conjugated immuno-globulin G. Relative intensity was scored with +, as contrasted with negativecontrols, which were developed without the primary antiserum.

-1-1.1

FIG. 2. Restriction analysis of the plasmids in the GNF isolate ofE. ictalun. Lane M, PstI digest of lambda DNA (the sizes of thestandard fragments [in kilobase pairs] are indicated on the right);lane A, undigested plasmid DNA (the locations of the four plasmidsare indicated on the left with arrows); lanes A to G, plasmid DNAfollowing digestion with restriction endonucleases HindIll, SalIl,SstII, XhoI, BamHI, EcoRI, and ClaI, respectively.

pools of the GNF isolate were restricted either with enzymeswhich did not have sites in pCL1 or pCL2 or with enzymeswhich cut these reference plasmids only once. From theseanalyses, it appears that the GNF 6.0- and 3.1-kb plasmidsare distinct from one another. Of the seven enzymes used forrestriction analyses shown in Fig. 2, Sall was the onlyenzyme which restricted the GNF 3.1-kb plasmid. However,the same seven enzymes yielded a different restrictionprofile for the 6.0-kb plasmid. This plasmid did not contain a

SalI restriction site but had single sites for EcoRI and ClaIand at least two restriction sites for BamHI. Therefore, thereare at least five differences in the restriction sites of the 6.0-and 3.1-kb plasmids. These results suggest that the GNF 6.0-and 3.1-kb plasmids are probably distinct from each other.These results, coupled with the hybridization results de-scribed above, indicate that E. ictaluni isolates may harboradditional small plasmids which are not similar to pCL1 andpCL2.

Serological analysis. Polyclonal antiserum to the GNFisolate as well as polyclonal antiserum to a channel catfishisolate (S85-1377) were prepared to determine whether theseisolates were serologically different. Comparative analysesusing either microtiter agglutination assays or fluorescent-antibody analyses indicated that the GNF isolate appearedto be antigenically distinct from these other E. ictalunistrains (Table 1). The anti-GNF isolate antiserum reactedstrongly only against the GNF isolate; the serological reac-

tions with the other isolates were marginal. In contrast, theantiserum made to isolate S85-1377 strongly cross-reactedwith the type strain ATCC 33202, as well as other E. ictaluri

isolates recovered from infected channel catfish. In addition,this antiserum strongly reacted against the danio isolate.However, the anti-S85-1377 antiserum gave only a marginalreaction when reacted with the GNF isolate. These analysesindicate that the GNF isolate appears serologically distinctfrom these other E. ictalun isolates.

Structural and antigenic analysis of E. ictaluri isolates. Todetermine whether the serological difference observed be-tween the GNF isolate and the other E. ictaluri strains couldbe defined, SDS-solubilized cell lysates were analyzed bySDS-PAGE. Cell lysates stained with Comassie brilliant blueshowed that the channel catfish isolates as well as the danioisolate had similar protein profiles. The positions of thestained bands as well as their relative intensities were verysimilar. The GNF isolate also had an essentially similarprotein profile, with only a few minor structural differencesapparent (Fig. 3A).

Identical aliquots of the lysates used above were alsoanalyzed in Western blots using the two different anti-E.ictaluni antisera (Fig. 3B and C). Western blots developedwith the anti-S85-1377 antiserum showed that the antigenicprofile of the danio lysate was very similar to that observedwith the other two E. ictaluri strains obtained from infectedchannel catfish. In contrast, the antigenic profile of the GNFlysate which reacted with this antiserum was somewhatdifferent from that of these other strains. The anti-S85-1377antiserum did not detect the same number of bands, nor didthese bands exhibit the same relative staining intensity as

those derived from the danio or channel catfish isolates. Thegeneral conclusion that the antigenic profile of the GNFisolate was different from that of these other isolates was

supported when lysates from these strains were reacted withthe antiserum derived to the GNF isolate. With this anti-serum multiple bands were observed in the lysate of theGNF isolate, but many of the bands in the other strainseither were not reactive or reacted with less staining inten-sity than those from the GNF isolate. There were, however,some proteins which appeared to be antigenically similar ineach of these strains. As judged by the relative intensity ofthe staining reaction with both antisera, as well as theidentical relative mobility of the reactive proteins, threeproteins, exhibiting relative molecular masses of 97, 43, and37 kDa, appear to be antigenically similar in each of theseisolates.

Proteinase K-treated lysates were also examined in West-ern blots using these two different antisera (Fig. 3D and E).

-4.7

-2.8

-2.1

-1.7

VOL. 59, 1993

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 21

Feb

ruar

y 20

22 b

y 22

2.11

3.99

.190

.

2834 LOBB ET AL.

A. M 1 2 3 4

-,4 i Irtt e t*4y#i-

B1 2 3 4

C D

FIG. 3. SDS-PAGE and Western blot analyses of four strains of E. ictalun. (A) Commassie brilliant blue-stained SDS-PAGE gel ofreduced cell lysates. Lane M contained molecular weight standards (a mixture of myosin, P-galactosidase, phosphorylase b, bovine albumin,egg albumin, and carbonic anhydrase). The masses of these standards (in kilodaltons) are indicated on the left. In all panels lanes 1 through4 contained cell lysates of E. ictalun GNF strain 10.16, danio strain 10.15, strain S85-1377, and type strain ATCC 33202, respectively. (B andC) Western blots of cell lysates of E. ictaluri strains reacted with anti-GNF strain 10.16 and anti-S85-1377 antisera, respectively. (D and E)Western blots of proteinase K-treated cell lysates of E. ictaluri strains reacted with anti-GNF strain 10.16 and anti-S85-1377 antisera,respectively.

These reactions clearly distinguished the GNF isolate fromthe other strains. The anti-GNF E. ictaluni antiserum reactedwith the lipopolysaccharide (LPS) ladder of the GNF isolateand not with the LPS ladder of the other strains. Theconverse reaction was observed when these strains werereacted with the anti-S85-1377 antiserum. These results areconsistent with the conclusion that the 0 antigen of the GNFisolate is antigenically distinct from the 0 antigen of theseother isolates and hence probably accounts for the majorserological differences observed in the agglutination andfluorescent-antibody analyses.

DISCUSSION

The GNF isolate of E. ictaluri has a different plasmidprofile and is serologically distinct from strains of E. ictaluniroutinely isolated from infected channel catfish. It waspreviously shown that E. ictaluri strains isolated from in-fected catfish in the southeastern United States containedtwo different small plasmids, designated pCL1 and pCL2.Comparisons of the restriction maps of these plasmids, as

well as the inability to detect possible cointegrate structuresin plasmid pools, suggested that partial homology of theseplasmids was unlikely. Thus, on the basis of the consistencyof detecting both plasmids in channel catfish isolates, it wassuggested that these plasmids should be valuable tools in thepresumptive identification of this bacterium from infectedchannel catfish. These findings have been confirmed by otherinvestigators (8, 10).The use of hybridization analyses in which plasmid-

specific probes are used to detect E. ictaluri has obviouspotential, as related earlier (7). Because of this potential, it

was important to ensure that pCL1 and pCL2 did notcross-hybridize with each other. Another study indicatedthat these two plasmids cross-hybridized (11), although asubsequent study by this group indicated that pCL1 did notcross-hybridize with pCL2 (10). To resolve this question,pCL1 and pCL2 were twice purified from agarose gels andused as probes in hybridization analyses. These studiesclearly showed that these probes do not cross-hybridize witheach other and do not cross-hybridize with the bacterialchromosome.

Plasmid analysis of the isolates of E. ictalun from thedanio and the GNF support the evidence of Newton et al. (8)and Reid and Boyle (10) that the GNF isolate containedplasmids which are different in size from those found ineither the danio isolate or the E. ictalun strains routinelyisolated from channel catfish. Both of these earlier investi-gations showed that the GNF isolate contained three plas-mids which had molecular sizes of about 5.6, 4.0, and 3.0 kb.Reid and Boyle (10) showed that the 5.6-kb plasmid cross-

hybridized with probes derived from pCL1 and the 4.0-kbplasmid hybridized with probes derived from both pCL1 andpCL2. Plasmid profiles of the GNF isolate in the studies ofboth Reid and Boyle and Newton et al., however, did notappear to reflect the 6.0-kb plasmid which was identified inthis study. This difference might reflect several possibilities,including comigration of the 6.0- and 5.7-kb plasmids inthese earlier studies, the loss of the 6.0-kb plasmid fromthese cultures, or even integration of the 6.0-kb plasmid intothe chromosome. It is not clear which of these possibilitiesmight be correct. However, this study shows that the GNFstrain contains two plasmids in addition to plasmids similarto pCL1 and pCL2. The 6.0- and 3.1-kb GNF plasmids have

205-11 697-

66-

45-

29-

E1 2 3 4

APPL. ENvIRON. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 21

Feb

ruar

y 20

22 b

y 22

2.11

3.99

.190

.

PLASMID AND SEROLOGICAL DIFFERENCES IN E. ICTALURI 2835

different restriction patterns, and neither strongly hybridizeswith probes derived from either pCL1 or pCL2. In contrast,the GNF 5.7-kb plasmid hybridized with probes derivedfrom pCL1, whereas the GNF 4.1-kb plasmid hybridizedwith probes derived from pCL2. Because high-stringencyhybridization conditions were used in these analyses, theseplasmids share considerable sequence similarity with pCL1and pCL2, respectively.The possibility that strains of E. ictaluri may lose plasmids

during cell culture is of potential concern, although earlier invitro attempts to cure pCL1 and pCL2 from E. ictaluriproved difficult (7). Studies have shown that a strain of E.ictalun isolated in Maryland from white catfish had a 4.0-kbplasmid which, by hybridization analysis, was probablysimilar to pCL2. However, this strain apparently lacked a5.7-kb (pCL1-like) plasmid, and furthermore, hybridizationanalyses suggested that this plasmid was not integrated intothe chromosome (10). Whether this strain may have lost aplasmid similar to pCL1 is not known, but these observa-tions seem to caution that plasmid analyses should beconducted with strains before extensive cell passage.

Earlier studies had shown that different isolates of E.ictalun appear to be biochemically homogeneous (15). Sero-logical studies have also concluded that different strains ofE. ictaluri appear to be serologically homogeneous. Plumband Klesius (9) derived seven hybridoma cell lines from miceimmunized with E. ictaluni type strain ATCC 33202. Theiranalyses suggested that although not all monoclonal antibod-ies reacted uniformly against all isolates, there were con-served antigens in all of the strains analyzed. Bertolini et al.(1) have also done serological assessments on differentstrains of E. ictaluri. In their analyses they compared anantiserum raised against a channel catfish isolate with anantiserum raised against the GNF isolate. In both cases,antiserum was prepared from rabbits that were immunizedwith formalin-killed cells. Bertolini et al. showed that therewere antigens which were apparently specific to the GNFisolate but concluded that there was insufficient evidence tosuggest that the GNF isolate represented a different sero-type. In the present study murine polyclonal antisera ob-tained from mice immunized with live E. ictaluri from achannel catfish isolate or the GNF isolate were compared.By both agglutination reactions and the fluorescent-antibodytechniques, the GNF isolate was judged to be serologicallydistinct.Comparative SDS-PAGE analyses of cell lysates indicate

that the channel catfish isolates and the danio and GNFisolates have similar protein profiles, as judged by therelative staining pattern of similar-sized bands. However,when replicate Western blots were reacted with these twodifferent antisera, the antigenic profiles were different. It isalso interesting that there were several proteins which werejudged to be similar in their antigenic reactions with bothantisera. Although at this point little is known regarding thestructural proteins of E. ictaluri, proteins with relativemolecular masses of 97, 43, and 37 kDa appeared to stainwith equal intensity with both antisera. This observationsuggests that these proteins are major antigenic determinantswhen live whole cells are used for immunization.Western blots of proteinase K-treated lysates indicate that

the 0 antigen of the GNF isolate is antigenically differentfrom the 0 antigen of the other strains of E. ictaluriexamined in this report. Studies with the more commonmember of the genus, E. tarda, have shown that there aredifferent 0-specific chains in different isolates. This findinghas resulted in the serological classification of different

0-antigens in this species (12). The present study wouldindicate that E. ictaluri may also be classified into differentserotypes. Whether the diversity of 0 antigens representedin E. tarda will be reflected as more serological analyses areperformed with additional isolates of E. ictaluni must awaitfurther studies.

In conclusion, this study has identified a strain of E.ictaluri isolated from an infected GNF which has a differentplasmid profile and is antigenically distinct from isolates ofE. ictaluni recovered from channel catfish in the southeast-ern United States. Of the four plasmids identified in the GNFstrain, only the 5.7-kb plasmid was similar in size to one ofthe characterized plasmids (pCL1). Hybridization analyses,however, showed that the 5.7-kb plasmid and the 4.1-kbplasmid shared significant similarity to pCL1 and pCL2,respectively. The serological differences observed betweenthe GNF isolate and the other E. ictaluni isolates examinedwere determined by Western blotting to reflect significantdifferences in the 0 antigens of these two serological groups.These combined results suggest that isolates of E. ictaluriwhich have plasmid profiles different from those strainstypically isolated from channel catfish should be examined todetermine whether they are serological different from knownisolates. It may be possible that the cryptic plasmids of E.ictaluri may mediate changes in the 0 antigen. If newserotypes of E. ictaluri prove to be virulent in challengeexperiments with channel catfish, these strains may becomeof significant economic concern if they are introduced intomajor channel catfish aquaculture areas.

ACKNOWLEDGMENTS

Our appreciation is extended to Jim Bertolini for providing thedanio and GNF isolates of E. ictaluri used in this study.

This investigation was supported by Public Health Service grantAI23052 from the National Institutes of Health.

REFERENCES1. Bertolini, J. M., R. C. Cipriano, S. W. Pyle, and J. J. A.

McLaughlin. 1990. Serological investigation of the fish pathogenEdwardsiella ictaluri, cause of enteric septicemia of catfish. J.Wild. Dis. 26:246-252.

2. Ghaffari, S. H., and C. J. Lobb. 1989. Nucleotide sequence ofchannel catfish heavy chain cDNA and genomic blot analyses.Implications for the phylogeny of Ig heavy chains. J. Immunol.143:2730-2739.

3. Hawke, J. P. 1979. A bacterium associated with disease of pondcultured channel catfish, Ictalurus punctatus. J. Fish. Res.Board Can. 36:1508-1512.

4. Hawke, J. P., A. C. McWhorter, A. G. Steigerwalt, and D. J.Brenner. 1981. Edwardsiella ictaluri sp. nov., the causativeagent of enteric septicemia of catfish. Int. J. Syst. Bacteriol.31:396-400.

4a.Kent, M. L., and J. M. Lyons. 1982. Report. Fish Health News2:ii.

5. Lobb, C. J., and L. W. Clem. 1982. Fish lymphocytes differ inthe expression of surface immunoglobulin. Dev. Comp. Immu-nol. 6:473-479.

6. Lobb, C. J., and L. W. Clem. 1983. Distinctive subpopulationsof catfish serum antibody and immunoglobulin. Mol. Immunol.20:811-818.

7. Lobb, C. J., and M. Rhoades. 1987. Rapid plasmid analysis forthe identification of Edwardsiella ictaluri from infected channelcatfish (Ictalurus punctatus). Appl. Environ. Microbiol. 53:1267-1272.

8. Newton, J. C., R. C. Bird, W. T. Blevins, G. R. Wilt, and L. G.Wilt. 1988. Isolation, characterization, and molecular cloning ofcryptic plasmids from Edwardsiella ictalui. Am. J. Vet. Res.49:1856-1860.

9. Plumb, J. A., and P. Klesius. 1988. An assessment of the

VOL. 59, 1993

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 21

Feb

ruar

y 20

22 b

y 22

2.11

3.99

.190

.

2836 LOBB ET AL. APPL. ENVIRON. MICROBIOL.

antigenic homogeneity of Edwardsiella ictaluri using monoclo-nal antibody. J. Fish Dis. 11:499-509.

10. Reid, W. S., and J. A. Boyle. 1989. Plasmid homologies inEdwardsiella ictalun. Appl. Environ. Microbiol. 55:3253-3255.

11. Speyerer, P. D., and J. A. Boyle. 1987. The plasmid profile ofEdwardsiella ictaluri. J. Fish Dis. 10:461-469.

12. Tamura, K., R. Sakazaki, A. C. McWhorter, and Y. Kosako.1988. Edwardsiella tarda serotyping scheme for internationaluse. J. Clin. Microbiol. 26:2343-2346.

13. Towbin, H., T. Staehelin, and J. Gordon. 1979. Electrophoretic

transfer of proteins from polyacrylamide gels to nitrocellulosesheets: procedure and some applications. Proc. Natl. Acad. Sci.USA 76:4350-4354.

14. Waltman, W. D., E. B. Shotts, and V. S. Blazer. 1985. Recoveryof Edwardsiella ictaluri from danio (Danio devario). Aquacul-ture 46:63-66.

15. Waltman, W. D., E. B. Shotts, and T. C. Hsu. 1986. Biochemicalcharacteristics of Edwardsiella ictaluni. Appl. Environ. Micro-biol. 51:101-104.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 21

Feb

ruar

y 20

22 b

y 22

2.11

3.99

.190

.