JOURNAL OF CLINICAL MICROBIOLOGY,0095-1137/97/$04.0010

May 1997, p. 1151–1156 Vol. 35, No. 5

Copyright q 1997, American Society for Microbiology

Molecular Evolution of Vibrio cholerae O1 Strains Isolatedin Lima, Peru, from 1991 to 1995

A. DALSGAARD,1* M. N. SKOV,1† O. SERICHANTALERGS,2 P. ECHEVERRIA,2

R. MEZA,3 AND D. N. TAYLOR3

Department of Veterinary Microbiology, Royal Veterinary and Agricultural University, Frederiksberg, Denmark1;Armed Forces Research Institute of Medical Sciences, Bangkok, Thailand2; and Departments of Microbiology

and Medical Ecology, United States Naval Medical Research Institute Detachment, Lima, Peru3

Received 7 October 1996/Returned for modification 19 December 1996/Accepted 29 January 1997

Following the emergence of cholera in Lima, Peru, in 1991, isolates of Vibrio cholerae O1 biotype El Torrecovered from patients in various parts of Lima were selected and characterized. Ribotyping and pulsed-fieldgel electrophoresis (PFGE) revealed four BglI ribotypes and eight NotI PFGE types among 50 V. cholerae O1strains recovered from patients with cholera in Lima from 1991 to 1995, with certain genotypes appearing tocluster geographically. While differences in ribotype and PFGE type patterns suggest that genetic changes areoccurring in the strain responsible for the Latin American cholera epidemic, more frequently than previouslyreported, 40 (80%) O1 strains showed an identical ribotype pattern and 41 (82%) strains showed closely relatedPFGE types, types 1, 2, or 3, that differed by less than three restriction fragments. All strains were susceptibleto nine antibacterial agents studied. In 1991, more than 95% of the clinical V. cholerae O1 strains were serotypeInaba, whereas from 1992, serotype Ogawa began to predominate, with more than 90% of the isolates being ofthe Ogawa serotype in 1995. The small differences in genotypes of V. cholerae O1 is remarkable because cholerais highly seasonal in coastal areas of Peru and support the hypothesis that the epidemic strain reemerges froman environmental source. However, the relative high rate of genetic changes within V. cholerae O1 as shown byribotyping and PFGE should be taken into consideration when typing patterns of V. choleraeO1 associated withcholera in Latin America are evaluated.

In January 1991, cholera appeared in several cities along thecoast in Peru, including the capital Lima, and spread rapidly toother countries in Latin America (14). The outbreak of choleramarked the first epidemic in Latin America in this century.Among Latin American countries, Peru has reported the high-est number of cholera cases, with more than 600,000 casesreported from 1991 to 1993 (14). From 1993, cholera has beenendemic in most areas of Peru, a pattern often observed incountries affected by significant morbidity (14).A variety of phenotypic and molecular typing methods, such

as multilocus enzyme electrophoresis, ribotyping, and pulsed-field gel electrophoresis (PFGE), have been used to study theepidemiology of Vibrio cholerae O1 isolates from Latin Amer-ica (32). Previous studies of V. cholerae O1 strains isolatedfrom 1991 to 1993 found that all isolates were essentially iden-tical, as shown by ribotyping and PFGE (1, 21, 31). Popovic etal. (21) found that all isolates related to the Latin Americanepidemic in 1991 and 1992 had an identical ribotype, which wasalso shown by some V. cholerae O1 isolates associated with theseventh pandemic in other parts of the world. Cameron et al.(1) found that all isolates related to the current epidemic inLatin America were identical by PFGE with the restrictionenzyme NotI. However, in a recent study of the diversity ofelectrophoretic types of V. cholerae O1, two isolates from Perushowed a PFGE pattern which differed by a single band fromthose of all isolates previously associated with the Latin Amer-ican cholera epidemic (9). Thus, following its introduction in

1991, the Latin American strain appears to have undergonegenetic changes, although the rate of changes remains to bedetermined (9).Whereas the Latin American epidemic appears to be asso-

ciated with a single strain of V. cholerae O1, distinct O1 strainswere reported to be associated with cholera in Mexico andBrazil (4, 9). The ability of these and other possibly introducedstrains to compete with the Latin American epidemic strain isunknown.Within this epidemic setting, we studied the evolution of V.

cholerae O1 biotype El Tor isolates recovered in Lima, Peru,from 1991 to 1995. A collection of 50 isolates were character-ized by antibiotic susceptibility testing, plasmid profiling, ri-botyping, restriction fragment length polymorphism analysis ofcholera toxin (CT) genes, and PFGE typing. Molecular typingshowed that the isolate initially introduced into Peru in 1991has undergone several genetic changes. Furthermore, a distinctnew V. cholerae O1 serovar Ogawa strain isolated in 1994 wasidentified.

MATERIALS AND METHODS

Bacteria. A total of 50 V. cholerae O1 biotype El Tor strains recovered frompatients from 1991 to 1995 in various parts of Lima, Peru, were included in thestudy (Table 1). Five strains each which agglutinated with antisera to serotypeOgawa and to serotype Inaba were selected from each year. Strains with desig-nations CHO originated from Callao, the port of Lima or north of Lima, at amilitary training area; BAB strains were isolated at city hospitals; strains with thedesignations VIG, HMA, and ECC were recovered in south Lima during acholera surveillance to evaluate the efficacy of a V. cholerae vaccine (5, 25);strains with number designations only were isolated in north Lima; and CITstrains were isolated at a military base in Callao. Isolates O51A serotype Ogawaand O51B serotype Inaba were recovered from the same stool sample of a singlepatient, as were isolates O19A and O19B.Fecal specimens were analyzed for V. cholerae by enrichment for 6 h in alkaline

peptone water (pH 8.6) and were then plated onto thiosulfate-citrate-bile salt-sucrose agar (Difco, Detroit, Mich.). On the basis of standard biochemicalreactions (24), all suspected V. cholerae isolates were tested by agglutination tests

* Corresponding author. Mailing address: Department of Veteri-nary Microbiology, The Royal Veterinary and Agricultural University,Bulowsvej 13, 1870 Frederiksberg C, Denmark. Phone: 45-35-282720Fax: 45-35-282757. E-mail: Anders.Dalsgaard@vetmi.kvl.dk.† Present address: National Veterinary Laboratory, Aarhus, Den-

mark

1151

on January 11, 2016 by guesthttp://jcm

.asm.org/

Dow

nloaded from

with polyvalent O1, monospecific Ogawa and Inaba, and O139 polyclonal goatantisera (Naval Medical Research Institute, Bethesda, Md.).Antibiotic susceptibility testing. The 50 V. cholerae O1 isolates were tested for

their susceptibilities to nine antibacterial agents by the disk diffusion methodrecommended by the National Committee for Clinical Laboratory Standards(18) with disks (Sensi-disc; BBL, Becton Dickinson, Cockeysville, Md.) contain-ing cephalothin at 30 mg/disk, chloramphenicol at 30 mg/disk, doxycycline at 30mg/disk, furazolidone at 100 mg/disk, gentamicin at 10 mg/disk, kanamycin at 30mg/disk, streptomycin at 10 mg/disk, sulfisoxazole at 250 mg/disk, and tetracyclineat 30 mg/disk. In addition, the O1 isolates were tested for their susceptibilities to

a low level of tetracycline by using disks containing 10 mg/disk and following theinstructions of the disk manufacturer (Neo-Sensitabs; Rosco, Taastrup, Den-mark). Isolates were recorded as either sensitive or resistant.Isolation of plasmid DNA. Plasmid preparation was carried out by the method

of Kado and Liu (11), modified by incubating the cells at elevated pH (pH 12.75)for 30 min at 568C during the lysis step. Following electrophoresis, the plasmidswere visualized essentially as described previously (19). V. cholerae O1 V1075/25containing an approximately 150-kb plasmid was used as the control strain (27).Plasmid sizes were estimated from the migration distance in the agarose gelsrelative to the migration distance of reference plasmids in Escherichia coli V517and 39R861 (12, 29) by the method of Rochelle et al. (23). Repeated extractionof plasmid DNA was carried out for all isolates.Ribotyping. Total bacterial DNA was extracted by the method of Murray and

Thompson (17). On the basis of previous studies (6, 7, 21), the restriction enzymeBglI was used to digest chromosomal DNA. Ribotyping was performed by theprocedure described by Dalsgaard et al. (7) with digoxigenin-labeled 16S and 23SrRNA probes. A 1-kb DNA molecular size standard (GIBCO BRL, Gaithers-burg, Md.) was used as a size marker. Each isolate was ribotyped at least twice.Ribotype patterns were considered to be different when there was a difference ofone or more bands between isolates. Each ribotype was designated R followed byan arbitrary number.CT genotyping. Restriction fragment length polymorphism analysis of DNA

sequences encoding CT genes was performed by hybridization of nylon mem-branes with BglI-digested DNA with a digoxigenin-labeled probe consisting of a950-bp fragment of an XbaI-HincII digest of plasmid DNA isolated from E. coliMS 371(pJM17) (20). The hybridization procedures and detection of the probewere performed as described by the manufacturer (Boehringer Mannheim).PFGE. Strains were grown as described previously (26), and DNA was pre-

pared directly in a solid agarose plug (Bio-Rad, Hercules, Calif.) as described byCameron et al. (1). For restriction endonuclease digestion, thin slices were cut offthe agarose plugs, equilibrated in the appropriate nuclease buffer for 1 h, andthen digested for 4 h with 20 U of enzyme per plug (8). On the basis of previousPFGE studies of V. cholerae O1, the enzyme NotI (Amersham, ArlingtonHeights, Ill.) was selected for restriction analysis, allowing for comparison withpreviously published PFGE types (1, 9, 16). PFGE types were considered to bedifferent if they differed by one or more bands and were designated P followedby an arbitrary number.The samples were loaded as plugs into the wells of 1.0% agarose (Litex LSL

4000; FMC A/S, Vallensbaek Strand, Denmark) gels prepared in 0.53 TBEbuffer (103 TBE buffer is 89 mM Tris base, 90 mM boric acid, and 2.5 mMdisodium EDTA [pH 8.3]). PFGE was carried out by using a modified contour-clamped homogeneous electric field system (3) (Pulsaphor Plus; PharmaciaLKB, Uppsala, Sweden). The running conditions were 12 V/cm at 148C for 22 h.The pulse times were increased as follows: 5 s for 3 h, 9 s for 5 h, 12 s for 5 h,20 s for 4 h, 25 s for 3 h, and 30 s for 2 h. Multimeric phage lambda (48.5 kb)DNA (Pharmacia LKB) was used as molecular mass standard. Following elec-trophoresis, the gels were stained for 15 min in ethidium bromide (2 mg/ml inwater; Sigma), destained in distilled water for 15 min, and visualized on a UVlight box.

RESULTS

Specimens and antibiotic susceptibility patterns. Each ofthe 50 isolates included in the present study all showed bio-chemical and serological reactions typical of those of V. chol-erae and were identified as V. cholerae O1 biotype El Tor andserotype Inaba or Ogawa (Table 1). Most isolates were foundto be susceptible to the antibiotics tested, including low-dosetetracycline; exceptions were strains CHO 471, BAB 2292, andV. cholerae strains isolated in 1994 and 1995, which were re-sistant to kanamycin. Strain VIG 1398 was the only isolateshowing resistance to furazolidone and sulfisoxazole.Plasmid analysis. Plasmid analysis revealed that 7 of 50

(14%) strains carried plasmids, including 6 strains that con-tained a 37-kb plasmid and 1 strain (strain CHO 239) thatcarried a 48-kb plasmid (Fig. 1; Table 1). Among the 30 V.cholerae strains isolated from 1991 to 1993, plasmids werefound only in strain CHO 239, whereas 6 of 20 strains isolatedfrom 1994 to 1995 carried the 37-kb plasmid. Strain CHO 239was among the first V. cholerae O1 serotype Ogawa strainsisolated in Lima. The presence of plasmids did not appear tocorrelate with antibiotic susceptibility, because isolates with orwithout plasmids showed similar susceptibility patterns. Thecontrol strain showed a plasmid of approximately 150 kb (Fig.1).

TABLE 1. Plasmid profiles, ribotypes, and PFGE types of50 clinical V. cholerae O1 strains isolated in

Peru from 1991 to 1995

Strain Date of isolation(day, mo, yr) Serotype Plasmid

size (kb)Ribo-typea

PFGEtypeb

CHO 467 26 Nov 1991 Ogawa —c R1 P1CHO 474 26 Nov 1991 Ogawa — R1 P1CHO 471 26 Nov 1991 Ogawa — R1 P1CHO 468 26 Nov 1991 Ogawa — R1 P1CHO 239 11 June 1991 Ogawa 48 R1 P1BAB 2305 20 Nov 1991 Inaba — R1 P1BAB 2303 20 Nov 1991 Inaba — R1 P1BAB 2292 14 Nov 1991 Inaba — R1 P1BAB 2307 21 Nov 1991 Inaba — R1 P1BAB 2301 16 Nov 1991 Inaba — R1 P1CHO 563 6 Feb 1992 Ogawa — R1 P1CHO 561 29 Jan 1992 Ogawa — R1 P2CHO 569 8 Feb 1992 Ogawa — R1 P1CHO 567 8 Feb 1992 Ogawa — R1 P1O19Ad 12 Dec 1992 Ogawa — R1 P2O19Bd 12 Dec 1992 Inaba — R1 P1CHO 539 15 Jan 1992 Inaba — R1 P3CHO 549 16 Jan 1992 Inaba — R1 P3CHO 548 16 Jan 1992 Inaba — R1 P3CHO 552 16 Jan 1992 Inaba — R1 P3O70 18 Jan 1993 Ogawa — R2 P5O59 5 Jan 1993 Ogawa — R2 P6O54 5 Jan 1993 Ogawa — R2 P6O912 16 July 1993 Ogawa — R1 P1O51Ad 5 Jan 1993 Ogawa — R1 P3O51Bd 5 Jan 1993 Inaba — R1 P3O58 5 Jan 1993 Inaba — R1 P3O24 2 Jan 1993 Inaba — R1 P3O61 7 Jan 1993 Inaba — R1 P3O880 9 July 1993 Inaba — R1 P4VIG 707 24 Jan 1994 Ogawa — R1 P2VIG 705 24 Jan 1994 Ogawa — R1 P2VIG 795 24 Jan 1994 Ogawa — R3 P2VIG 753 25 Jan 1994 Ogawa — R1 P2VIG 1490 4 Feb 1994 Ogawa 37 R4 P7VIG 1398 4 Feb 1994 Inaba — R1 P1HMA 34 15 Dec 1994 Inaba — R1 P3CIT 171 3 Feb 1994 Inaba 37 R1 P1CIT 175 3 Feb 1994 Inaba 37 R1 P1CIT 178 3 Feb 1994 Inaba 37 R1 P1VIG 10917 24 Mar 1995 Ogawa — R3 P2VIG 10852 21 Mar 1995 Ogawa — R3 P2VIG 10989 28 Mar 1995 Ogawa — R3 P2HMA 159 21 Mar 1995 Ogawa — R3 P2ECC 1091 24 Mar 1995 Ogawa 37 R3 P8VIG 9562 26 Jan 1995 Inaba — R1 P4VIG 9835 8 Feb 1995 Inaba — R1 P3VIG 10179 20 Feb 1995 Inaba 37 R1 P4VIG 10371 28 Feb 1995 Inaba — R1 P4VIG 9834 7 Feb 1995 Inaba — R1 P3

a Ribotype was established by using the enzyme BglI.b PFGE type was established by using the enzyme NotI.c—, no plasmids were found.d Isolates O51A and O51B were recovered from the same patient, as were

isolates O19A and O19B.

1152 DALSGAARD ET AL. J. CLIN. MICROBIOL.

on January 11, 2016 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Ribotyping. Altogether, four BglI ribotypes were observedamong the 50 V. cholerae O1 isolates studied (Table 1). Anexample of the ribosomal banding patterns is presented in Fig.2. The chromosomal fragments containing rRNA genes rangedin size from 2 to 10 kb. All strains except strain VIG 1490contained common fragments of 2.3, 4.1, 4.6, 5.9, 6.1, 6.2, 6.7,7.0, and 9.6 kb and appeared to be closely related. Variationsin ribotype patterns were shown by serotype Ogawa strainsonly. All strains isolated in 1991 and 1992 showed an identicalribotype, ribotype R1. Ribotype R2 was shown only by threeOgawa strains isolated from the same location in the northernpart of Lima in 1993. Strain VIG 795 isolated in south Lima in1994 showed ribotype R3, which demonstrated a 5.6-kb frag-ment not seen among the other ribotypes. In addition, each ofthe five Ogawa strains isolated in south Lima in 1995 wasribotype R3. Strain VIG 1490 was isolated in 1994 and showeda unique ribotype, ribotype R4, which differed from the otherribotypes by four to five fragments. R2 isolates showed a 4.3-kbfragment not seen among isolates of the R1, R3, and R4ribotypes and, furthermore, lacked a 4.0-kb fragment demon-strated by isolates of the other ribotypes (Fig. 2). All V. chol-erae serotype Inaba strains belonged to ribotype R1. Repeatedstudies of all strains showed no variation in ribotype patterns,although differences in band intensity and the degree of back-ground were observed for a few strains.CT genotypes. Southern blot hybridization of BglI-digested

genomic DNA with the CT probe revealed an identical patternfor all strains which showed three fragments of approximately15, 9, and 7 kb. However, variations in the band intensity of the9- and 7-kb fragments were found for the majority of strains.PFGE. PFGE analysis with the restriction enzyme NotI gave

a suitable distribution of fragments, with about 20 fragmentsbeing .50 kb. The analysis of 50 V. cholerae O1 isolatesshowed that the isolates displayed eight different banding pat-terns (Fig. 3; Table 1). With the exception of strain VIG 1490,which showed a unique type (type P7) differing from the otherPFGE types by more than seven fragments, all isolates ap-peared to be closely related. Strain VIG 1490 was recovered on4 February 1994 from a 10-year-old boy with diarrhea. Theisolate showed an unique ribotype (ribotype R4) and appears

to represent a distinct O1 strain compared with the straincausing the present epidemic (28). The mother of the 10-year-old boy was also hospitalized for cholera. However, an isolaterecovered from the mother’s stool sample showed PFGE typeP2 (data not shown).Several different PFGE types were shown among V. cholerae

strains isolated from 1991 to 1995. All isolates from 1991showed an identical PFGE type (type P1). PFGE type P2,which differed from type P1 by having a 400-kb fragment in-stead of a 375-kb fragment, was represented by Ogawa isolatesonly and appeared for the first time in 1992. PFGE type P3lacked the 175-kb fragment shown by type P1 and P2 isolatesand was shown for the first time in 1992 by four serotype Inabaisolates. With the exception of isolate O51A, which aggluti-nated with serotype Ogawa antiserum, all type P3 isolates wereof the Inaba serotype. One and three serotype Inaba isolatesrecovered in 1993 and 1995, respectively, were type P4. MostPFGE type P1, P2, P3, and P4 isolates belonged to ribotypeR1, but four type P2 isolates recovered in 1995 belonged toribotype R3. One and two serotype Ogawa isolates recoveredin 1993 were types P5 and P6, respectively. PFGE types P5 andP6 had a 150-kb fragment not seen among the other types, andeach of the three isolates was ribotype R2. PFGE type P8,which had an unique 80-kb fragment, was shown by a singleisolate, isolate ECC 1091, recovered in 1995. No isolate recov-ered in 1995 was PFGE type P1.A comparison of the typing results indicated that PFGE

typing and ribotyping were in general agreement, with the fourribotypes being further differentiated by PFGE. However, iso-lates within ribotypes R1 and R3 showed an identical PFGEtype (type P2). From 1992, an association appeared betweenserotype and PFGE types because serotype Ogawa isolatesshowed PFGE types P1, P2, P5, P6, P7, and P8, whereas sero-type Inaba isolates showed types P3 and P4. However, excep-

FIG. 1. Examples of plasmid profiles of V. cholerae O1 isolated in Peru.Lanes: A, E. coli 39R 861 (four plasmids ranging from 147 to 6.9 kb); B, E. coliV517 (eight plasmids ranging from 54 to 2.0 kb); C, strain V1075/25 (controlstrain); D, strain CHO 239; E, strain VIG 10179.

FIG. 2. Examples of BglI ribotypes of 50 V. cholerae O1 strains isolated inLima, Peru, from 1991 to 1995. Unless indicated otherwise, the following expla-nations for the contents of the lanes indicate ribotype, strain designation, andyear of isolation. Lanes: a, 1-kb molecular mass standard; b, ribotype R1, CHO467, 1991; c, type R1, CHO 239, 1991; d, type R1, BAB 2307, 1991; e, type R2,O70, 1993; f, type R2, O59, 1993; g, type R2, O54, 1993; h, type R1, CHO 569,1992; i, type R1, O19A, 1992; j, type R1, CHO 539, 1992; k, type R1, 912, 1993;l, type R1, O51A, 1993; m, type R1, O24, 1993; n, type R1, VIG 1398, 1994; o,type R1, CIT 175, 1994; p, type R1, VIG 9562, 1995; q, type R3, VIG 795, 1994;r, type R1, VIG 10371, 1995; s, type R4, VIG 1490, 1994; t, 1-kb molecular massstandard.

VOL. 35, 1997 MOLECULAR EPIDEMIOLOGY OF V. CHOLERAE O1 1153

on January 11, 2016 by guesthttp://jcm

.asm.org/

Dow

nloaded from

tions were found because PFGE type P1 was also demon-strated by serotype Inaba isolates from 1994 and isolate O19B.V. cholerae O1 isolate O51A serotype Ogawa and serotype

Inaba isolate O51B were isolated from the same patient andshowed an identical PFGE type (type P3). Repeated aggluti-nation tests and PFGE typing confirmed the serotypes and thePFGE types of the two isolates. By using the PC/GENE, ver-sion 6.5, software (IntelliGenetics, Mountain View, Calif.), norestriction sites were identified in the rfbT gene for the enzymeNotI used for PFGE typing. The rfbT operon is responsible forserotype conversion in V. cholerae O1, and the rfbT sequencesfor the Inaba and Ogawa serotypes have been deposited inGenBank (13). With the absence of restriction sites for NotIwithin the rfbT operon of the two serotypes, identical PFGEtypes may occur. On the basis of the identical characteristics ofstrains O51A and O51B, a serotype conversion may have oc-curred within the patient during the course of infection. V.cholerae O1 strains O19A serotype Ogawa and O19B serotypeInaba were isolated from the same stool samples but showedPFGE types P2 and P1, respectively.Digested plasmids migrate according to the sizes of the frag-

ments in PFGE gels, whereas uncut plasmids have been re-ported to have aberrant mobilities (15, 22). In the presentstudy plasmids of approximately 37 and 48 kb were found

among isolates, as shown by one-dimensional electrophoresis.When extracts of plasmids of the two sizes were subjected toPFGE, fragments were located close to the 194-kb fragmentsof the phage lambda DNA size marker (data not shown).However, since none of the eight different PFGE types hadfragments with similar sizes, the presence of plasmids did notinfluence the interpretation and stability of PFGE typing.

DISCUSSION

A total of four BglI ribotypes and eight NotI PFGE typeswere found among 50 clinical V. cholerae O1 strains isolated inLima, Peru, from 1991 to 1995. These results indicate thatgenetic changes are occurring in the strain responsible for theLatin American cholera epidemic more frequently than haspreviously been reported (1, 9). While differences in ribotypeand PFGE type patterns occurred, 40 (80%) O1 strains showedan identical ribotype pattern (type 1) and 41 (82%) strainsshowed closely related PFGE types (type 1, 2, or 3) that dif-fered by less than three restriction fragments. Furthermore,genotyping with a CT probe revealed an identical pattern forall strains. The small differences in ribotypes and PFGE typesare consistent with relatively minor genetic variations within alargely clonal pool from which the epidemic strain arises everyyear. This is remarkable because cholera is highly seasonal inLima and other coastal areas of Peru. Most cases occur in thesummer months of January to March. Cholera does not occurat all on the coast from June until December. Despite this6-month hiatus, our data indicate that the nearly identicalcholera strain reemerges the next year. Our data support thehypothesis that the epidemic strain reemerges from an envi-ronmental source.In 1991, more than 95% of V. cholerae O1 biotype El Tor

isolates recovered from patients in the capital, Lima, and thesecond largest city, Trujillo, were serotype Inaba, whereas from1992 serotype serotype Ogawa began to predominate, withmore than 90% of the isolates being of the Ogawa serotype in1995 (30). A similar change in serotype was reported for iso-lates from most other Latin American countries (9). Theclosely related ribotype patterns and PFGE types found in thepresent study demonstrate that the two serotypes are derivedfrom the same clone of V. cholerae O1.Among isolates from Lima, certain ribotype and PFGE types

appeared to cluster geographically. For example, isolates ofribotype 2 and PFGE types P5 and P6 were only isolated innorth Lima. The strains from each year were found at locationswhere we were conducting surveillance for cholera. Thus,strains from any one year represent clusters of cholera cases orsmall outbreaks associated with a common location and date ofisolation. Our data indicate that within one of these clustersthere is marked genetic similarity of V. cholerae O1.However, there were some notable variations. Strain VIG

1490 isolated in 1994 from a boy represents a distinct PFGEtype that occurred at the same time that his mother was in-fected with a standard PFGE type strain. It is possible thatstrain VIG 1490 underwent genetic variation after beingpassed through one individual. Similarly, there were two in-stances (strains O19A and O19B and strains O53A and O53B)in which Ogawa and Inaba strains isolated from the same stoolspecimen showed nearly identical typing patterns. These dataalso provide evidence for point mutations in the rfbT operonwithin the epidemic O1 clone.Our data provide further support for the clonality of the

South American cholera strains. PFGE types P1 and P2 fromthis study appear to be identical to PFGE types 38 and 66,respectively, described by Evins et al. (9) among clinical V.

FIG. 3. PFGE banding patterns of NotI-digested total cellular DNAs fromrepresentative V. cholerae strains isolated in Lima, Peru, from 1991 to 1995.Unless indicated otherwise, the following explanations of the contents of thelanes indicate PFGE type, strain designation, and year of isolation. Lanes: A,multimers of phage lambda DNA (48.5 kb) as molecular mass markers; B, PFGEtype P1, CHO 467, 1991; C, type P2, CHO 561, 1992; D, type P3, CHO 539, 1992;E, type P4, VIG 9562, 1995; F, type P8, ECC 1091, 1995; G, type P5, O70, 1993;H, type P6, O59, 1993; I, type P7, VIG 1490, 1994; J, multimers of phage lambdaDNA (48.5 kb) as molecular mass markers.

1154 DALSGAARD ET AL. J. CLIN. MICROBIOL.

on January 11, 2016 by guesthttp://jcm

.asm.org/

Dow

nloaded from

cholerae strains isolated from several South American coun-tries from 1991 to 1993. An apparently identical PFGE type(type 38) was shown by O1 strains isolated from seafood andnonpotable water (ballast, bilge, and sewage) (9). Our ribotypeR1 pattern shown by all O1 strains isolated in 1991 and 1992appears to be identical to ribotype 5 of Popovic et al. (21).However, our ribotypes R2, R3, and R4 have not been de-scribed previously among South American O1 strains (9, 21).Although more than half of all cholera cases reported in

Latin America have been reported in Peru, significant antimi-crobial resistance has not occurred (2). In the present study,the majority of V. cholerae strains were susceptible to all anti-biotics tested; the exceptions were O1 strains isolated in 1994and 1995, which were resistant to kanamycin. However, a highprevalence of multiple-antibiotic-resistant V. cholerae O1strains was found in Ecuador in 1991 (33). In addition, antibi-otic resistance to furazolidone, sulfisoxazole, and streptomycinwas found among V. choleraeO1 strains in Mexico, Guatemala,and Brazil (9, 21). This resistance pattern was associated witha specific ribotype pattern 6a which we did not find amongisolates in Peru (21). These data indicate that antibiotic-resis-tant V. cholerae O1 strains are involved in the Latin Americancholera epidemic and that some O1 strains may develop anti-microbial resistance at a higher rate than others. The spread ofantibiotic-resistant O1 strains should be monitored.Although an increase in the number of strains containing

plasmids was registered from 1991 to 1995, plasmids did notappear to encode antibiotic resistance. However, a study car-ried out in Ecuador in 1991 demonstrated a 100-MDa plasmidin four multiple-antibiotic-resistant V. choleraeO1 strains stud-ied (33). The importance of plasmid analysis in the character-ization of Latin American V. choleraeO1 isolates remains to bedetermined.For future comparisons of ribotypes, we suggest that the

method originally described by Popovic et al. (21) be used andthat ribotype figures of high photographic quality be includedin any reports. Several studies have shown the usefulness of therestriction enzyme NotI in PFGE typing of V. cholerae O1isolates (1, 9, 16). We therefore suggest that NotI be used inPFGE typing of V. cholerae O1 isolates in the future, althoughthe enzyme ApaI was able to differentiate two clinical O1strains which were indistinguishable by NotI PFGE typing (16).Computer-assisted numerical analysis, including scanning ofribotyping membranes and PFGE gel patterns, is increasinglybeing used for quantitative assessment of genetic similaritiesand differences among V. cholerae O1 strains (10). Such as-sessments look promising, although the typing methods needfurther standardization because the ribotyping membranes andgels used for analysis are too often of poor quality. In thefuture, certain laboratories may establish databases of bacterialtyping patterns which scientists could access through the In-ternet for comparison analysis.The results of our study indicate the usefulness of ribotyping

and especially PFGE for studying genetic changes within the V.cholerae O1 strain responsible for the epidemic in Latin Amer-ica. The higher number of different genotypes shown in thepresent study compared with those shown in previous studies(9, 21) should be taken into consideration when the typingpatterns of V. cholerae O1 isolates associated with cholera inLatin America are evaluated.

ACKNOWLEDGMENTS

We are grateful for the technical assistance provided by AnitaForslund and Marianne Jacobsen at the Royal Veterinary and Agri-cultural University in Denmark and by Maruja Bernal at the NavalMedical Research Institute Detachment in Peru.

Anders Dalsgaard was supported by the Danish Council for Devel-opment Research (Danida grant 90810).

REFERENCES

1. Cameron, D. N., F. M. Khambaty, K. I. Wachsmuth, R. V. Tauxe, and T. J.Barrett. 1994. Molecular characterization of Vibrio cholerae O1 strains bypulsed-field gel electrophoresis. J. Clin. Microbiol. 32:1685–1690.

2. Carrillo, C., J. Tolmos, and J. Adachi. 1992. In vitro antimicrobial suscep-tibility testing of 319 toxigenic V. cholerae O1, El Tor, strains isolated duringthe cholera outbreak in Peru, comparing with 204 isolates of Shigella sp.,abstr. 793, p. 242. In Programs and abstracts of the 32nd Interscience Con-ference on Antimicrobial Agents and Chemotherapy. American Society forMicrobiology, Washington, D.C.

3. Chu, G., D. Vollrath, and R. W. Davis. 1986. Separation of large DNAmolecules by contour-clamped homogeneous electric fields. Science 234:1582–1583.

4. Coelho, A., J. R. C. Andrade, A. C. P. Vicente, and C. A. Salles. 1995. Newvariant of Vibrio cholerae O1 from clinical isolates in Amazonia. J. Clin.Microbiol. 33:114–118.

5. Dalsgaard, A., M. J. Albert, D. N. Taylor, T. Shimada, R. Meza, O. Serichan-talergs, and P. Echeverria. 1995. Characterization of Vibrio cholerae non-O1serogroups obtained from an outbreak of diarrhea in Lima, Peru. J. Clin.Microbiol. 33:2715–2722.

6. Dalsgaard, A., P. Echeverria, J. L. Larsen, R. Siebeling, O. Serichantalergs,and H. H. Huss. 1995. Application of ribotyping for differentiating Vibriocholerae non-O1 isolates from shrimp farms in Thailand. Appl. Environ.Microbiol. 61:245–251.

7. Dalsgaard, A., H. F. Mortensen, K. Mølbak, F. Dias, O. Serichantalergs, andP. Echeverria. 1996. Molecular characterization of Vibrio cholerae O1 strainsisolated during cholera outbreaks in Guinea-Bissau. J. Clin. Microbiol. 34:1189–1192.

8. Dalsgaard, A., M. N. Skov, O. Serichantalergs, and P. Echeverria. 1996.Comparison of pulsed-field gel electrophoresis and ribotyping for subtypingof Vibrio cholerae O139 isolated in Thailand. Epidemiol. Infect. 117:51–58.

9. Evins, G. M., D. N. Cameron, J. G. Wells, K. D. Greene, T. Popovic, S.Giono-Crezo, I. K. Wachsmuth, and R. V. Tauxe. 1995. The emerging diver-sity of the electrophoretic types of Vibrio cholerae in the Western Hemi-sphere. J. Infect. Dis. 172:173–179.

10. Faruque, S. M., S. Roy, A. R. M. A. Alim, A. K. Siddique, and M. J. Albert.1995. Molecular epidemiology of toxigenic Vibrio cholerae in Bangladeshstudied by numerical analysis of rRNA gene restriction patterns. J. Clin.Microbiol. 33:2833–2838.

11. Kado, C. I., and S. T. Liu. 1981. Rapid procedure for detection and isolationof large and small plasmids. J. Bacteriol. 145:1365–1373.

12. Macrina, F. L., D. J. Kopecko, K. R. Jones, D. F. Ayers, and S. K. McCowen.1987. A multiple plasmid-containing Escherichia coli strain: convenientsource of size reference plasmid molecules. Plasmid 1:417–420.

13. Manning, P. A., U. H. Stroeher, and R. Morona. 1994. Molecular basis forO-antigen biosynthesis in Vibrio cholerae O1 Ogawa-Inaba switching, p.77–94. In I. K. Wachsmuth, P. A. Blake, and Ø. Olsvik (ed.), Vibrio choleraeand cholera. Molecular to global perspectives. ASM Press, Washington, D.C.

14. Mata, L. 1994. Cholera El Tor in Latin America 1991–1993. Ann. N. Y.Acad. Sci. 740:55–68.

15. Mathew, M. K., C.-F. Hui, C. L. Smith, and C. R. Cantor. 1988. High-resolution and accurate size determination in pulsed-field gel electrophoresisof DNA. Biochemistry 27:9222–9226.

16. McCarthy, S. A., and F. M. Khambaty. 1994. International dissemination ofepidemic Vibrio cholerae by cargo ship ballast and other nonpotable waters.Appl. Environ. Microbiol. 60:2597–2601.

17. Murray, M. G., and W. F. Thompson. 1980. Rapid isolation of high molec-ular weight DNA. Nucleic Acids Res. 8:4321–4325.

18. National Committee for Clinical Laboratory Standards. 1993. Methods fordilution antimicrobial susceptibility test for bacteria that grow aerobically.National Committee for Clinical Laboratory Standards, Villanova, Pa.

19. Olsen, J. E., and J. L. Larsen. 1990. Restriction fragment length polymor-phism of the Vibrio anguillarum serovar O1 virulence plasmid. Appl. Envi-ron. Microbiol. 56:3130–3132.

20. Pearson, G. D. N., and J. J. Mekalanos. 1982. Molecular cloning of Vibriocholerae enterotoxin genes in Escherichia coli K-12. Proc. Natl. Acad. Sci.USA 79:2976–2980.

21. Popovic, T., C. A. Bopp, Ø. Olsvik, and K. Wachsmuth. 1993. Epidemiologicapplication of a standardized ribotype scheme for Vibrio choleraeO1. J. Clin.Microbiol. 31:2474–2482.

22. Prevost, G., B. Pottecher, M. Dahlet, M. Bientz, J. M. Mantz, and Y. Pie-mont. 1991. Pulsed field gel electrophoresis as a new epidemiological tool formonitoring methicillin-resistant Staphylococcus aureus in an intensive careunit. J. Hosp. Infect. 17:255–269.

23. Rochelle, P. A., J. C. Fry, M. J. Day, and M. J. Bale. 1985. An accuratemethod for estimating sizes of small and large plasmids and DNA fragmentsby gel electrophoresis. J. Gen. Appl. Microbiol. 132:53–59.

24. Sakazaki, R. 1992. Bacteriology of vibrio and related organisms, p. 37–55. In

VOL. 35, 1997 MOLECULAR EPIDEMIOLOGY OF V. CHOLERAE O1 1155

on January 11, 2016 by guesthttp://jcm

.asm.org/

Dow

nloaded from

D. Barua and W. R. Greenough (ed.), Cholera. Plenum Publishing Com-pany, New York, N.Y.

25. Sanchez, J. L., B. Vasquez, R. E. Begue, R. Meza, G. Castellares, C. Cabezas,D. M. Watts, A. M. Svennerholm, J. C. Sadoff, and D. N. Taylor. 1994.Protective efficacy of oral, whole cell/recombinant B subunit cholera vaccinein Peruvian military recruits. Lancet 344:1273–1276.

26. Skov, M. N., K. Pedersen, and J. L. Larsen. 1995. Comparison of pulsed-fieldgel electrophoresis, ribotyping, and plasmid profiling for typing of Vibrioanguillarum serovar O1. Appl. Environ. Microbiol. 61:1540–1545.

27. Tabtieng, R., S. Wattanasri, P. Echeverria, J. Seriwatana, L. Bodhidatta, A.Chatkaeomorakot, and B. Rowe. 1989. An epidemic of V. cholerae El TorInaba resistant to several antibiotics with a conjugative group C plasmidcoding for type II dihydrofolate reductase in Thailand. Am. J. Trop. Med.Hyg. 41:680–686.

28. Tenover, F. C., R. D. Arbeit, R. V. Goering, P. A. Mickelsen, B. E. Murray,D. H. Persing, and B. Swaminathan. 1995. Interpreting chromosomal DNArestriction patterns produced by pulsed-field gel electrophoresis: criteria forbacterial strain typing. J. Clin. Microbiol. 33:2233–2239.

29. Threlfall, E. J., B. Rowe, J. L. Ferguson, and L. R. Ward. 1986. Character-ization of plasmids conferring resistance to gentamycin and apramycin instrains of Salmonella typhimurium phage type 204c isolated in Britain. J. Hyg.Camb. 97:419–426.

30. Vugia, D. J., M. Rodriguez, R. Vargas, C. Ricse, C. Ocampo, R. Llaque, J. L.Seminario, K. D. Greene, R. V. Tauxe, and P. A. Blake. 1994. Epidemiccholera in Trujillo, Peru 1992: utility of a clinical case definition and shift inVibrio cholerae O1 serotype. Am. J. Trop. Med. Hyg. 50:566–569.

31. Wachsmuth, I. K., G. M. Evins, P. I. Fields, Ø. Olsvik, T. Popovic, C. A.Bopp, J. G. Wells, C. Carrillo, and P. A. Blake. 1993. The molecular epide-miology of cholera in Latin America. J. Infect. Dis. 167:621–626.

32. Wachsmuth, K., Ø. Olsvik, G. M. Evins, and T. Popovic. 1994. Molecularepidemiology of cholera, p. 357–370. In I. K. Wachsmuth, P. A. Blake, andØ. Olsvik (ed.), Vibrio cholerae and cholera: molecular to global perspectives.ASM Press, Washington, D.C.

33. Weber, J. T., E. D. Mintz, R. Canizares, et al. 1994. Epidemic cholera inEcuador: multidrug-resistance and transmission by water and seafood. Epi-demiol. Infect. 112:1–11.

1156 DALSGAARD ET AL. J. CLIN. MICROBIOL.

on January 11, 2016 by guesthttp://jcm

.asm.org/

Dow

nloaded from