Tracking of clinical and environmental Vibrio cholerae O1 strains by combined analysis of the presence of the toxin cassette, plasmid content and ERIC PCR

Abstract

Clinical and environmental Vibrio cholerae O1 strains associated with the cholera epidemic in the Luanda province of Angola from 1991 to 1994 were tracked by toxin distribution, plasmid content and chromosomal polymorphism of the enterobacterial repetitive intergenic consensus (ERIC) sequences by PCR fingerprinting. To follow the distribution of ace, zot and ctxA toxin genes, 6 specific PCR tests were applied to 100 Vibrio strains, after preliminary hybridization experiments. Clinical isolates of Vibrio cholerae O1 were characterized by high stability of the toxigenic cassette and the presence of a large conjugative multi-resistant plasmid of incompatibility class C. Such characteristics were present in all isolates during the four years of the epidemic. Environmental strains, isolated from the river supplying water to the Luanda population showed three different genetic profiles: the presence of both cassette and plasmid, the presence of cassette only or absence of both. To assess the clonal relationship between the clinical isolates and the three groups of environmental strains, the strains were analyzed by PCR ERIC polymorphism. This analysis, supported by the toxin and the plasmid content, suggested the stability of the epidemic strain in clinical cases during the epidemic and led to the finding that there was a strict genetic relationship of the epidemic strain with the environmental ones as characterized by the presence of the toxin cassette. The role of the water supply from Bengo River as a reservoir of the Vibrio epidemic strain is discussed.

Vibrio cholerae, Toxin cassette, Plasmid content, ERIC PCR

1 Introduction

Epidemic and endemic cholera is a main public health hazard for many developing countries, and control of the disease is still a problem; Vibrio cholerae O1, biotype El Tor, is the main causative agent of these epidemics, but other V. cholerae strains and other Vibrio spp. are more frequently reported to cause cholera-like enteritis in endemic areas [1–3]. In 1993, the emergence of a new epidemic strain of V. cholerae O1 39 Bengal was reported in Bangladesh and other Asiatic countries [4,5].

The main pathogenic mechanism of V. cholerae O1 is the production of three enterotoxins. In addition to the strong heat-labile cholera toxin (CT), V. cholerae O1 produces the zonula occludens toxin (Zot) [6,7] and the accessory cholera enterotoxin (Ace) [8]. The genes for these toxins are clustered with cep (core encoded pilin, coding for a colonization factor) and orfU (an open reading frame of unknown function), in a 4.5 kb region, flanked by directly repeated recombinogenic sequences (RS) of 2.4–2.7 kb [9].

The presence of this virulence cassette is highly variable since it is subject to amplification and deletion, under the control of recA gene, rearrangement and copy number of this genetic area play an important role in determining the pathogenicity level of isolates [10]. Waldor and Mekalanos [11] demonstrated that the toxigenic cassette was encoded by a filamentous bacteriophage (CTXF) able to integrate into the V. cholerae chromosome: the resulting strain could produce cholera toxin and cause disease, this mechanism provides evidence of toxigenicity transmission among V. cholerae strains.

In V. cholerae O1, as in other enteric pathogens, resistance plasmids are naturally conjugative, mostly belonging to incompatibility class C [12,13]. Multidrug resistant V. cholerae O1 also causes serious outbreaks and becomes endemic [14–19].

Characterization of toxin and resistance genes of V. cholerae O1 and non-O1 and other Vibrio spp. from human beings and from the environment is of great concern, especially in areas with compromised environmental and social conditions. Such areas, in fact, are persistently affected by epidemic and endemic acute diarrhoeas.

Molecular epidemiology of enteric bacteria outbreaks is done by analysing the plasmid content [20] and chromosomal polymorphism [21,22]. Among other markers, ribosomal RNA genes [23–25] and enterobacterial repetitive intergenic consensus (ERIC) sequences [26,27] were chosen as highly conserved sequences present as multiple copies in the genome of all bacteria. Such markers are highly effective in investigating the clonal origin of epidemic strains and the PCR fingerprinting method applied to these chromosomal sequences is the most rapid and reliable technique.

Our interest was focused on Angola. Since 1987 this country has been affected by recurrent seasonal cholera epidemics [28] caused by strains that later became endemic in many areas. In a large microbiological study of acute diarrhoeal diseases [28–30] and drug resistance conducted in 1991–1994 in Luanda province V. cholerae O1 and Vibrio spp. strains were isolated from patients and from the aquatic environment. In order to understand the role of different Vibrio species in morbidity and mortality, the distribution of plasmids conferring resistance to drugs, the presence of ctx, zot and ace toxin genes and the analysis of their PCR ERIC sequences in clinical and environmental isolates was investigated. Furthermore, the distribution of the genetic markers mentioned above was correlated with the spread of epidemic strains of V. cholerae from the environment to an urban population.

2 Materials and methods

2.1 Isolation of bacterial strains

From 1991 to 1994, 221 bacterial strains (Table 1), representative of the outbreaks, were isolated in Luanda province from stool samples and/or rectal swabs of cholera patients (208 strains) and from Bengo River surface water (15 strains). The isolation was performed by standard methods [31] and strains were identified by the API 20E test (bioMèrieux, France). V. cholerae O1 were serotyped by specific antisera (Murex Diagnostici Spa, Italy). To confirm the El Tor biotype, agglutination tests in a 2.5% suspension of sheep red cells in physiological saline solution were performed. In addition 27 different non V. cholerae O1 isolates were identified and examined. Cells were grown in Luria Broth and Nutrient Broth and/or agar plates. Culture media were purchased from Difco and Oxoid.

Table 1

Source and number of Vibrio spp. and other bacterial isolates, and occurrence of toxin genes by hybridization and PCR techniques

Strain	Source	No. of isolates	Presence of toxins by
			Hybridization	PCR
V. cholerae O1	CLIN	184	3/3	71/71
	ENV	12	2/6	4/12
V. parahaemolyticus	CLIN	17	0/2	0/10
	ENV	1	0/1	0/1
V. mimicus	CLIN	1	0/1	0/1
	ENV	2	0/1	0/2
V. alginolyticus	CLIN	1	0/1	0/1
Kl. pneumoniae	CLIN	1	1/1	0/1
Y. intermedia	CLIN	1	−	0/1
Aeromonas spp.	CLIN	1	0/1	0/1
Total		221	17	103

Bacterial isolates from patients.

Bacterial isolates from water river samples.

Anomalous hybridizations.

2.2 Drug resistance tests

Minimum inhibitory concentrations (MIC) were tested on solid medium, as previously described [32]. The following drugs were used: ampicillin, chloramphenicol, erythromycin, gentamicin, kanamycin, nalidixic acid, rifampicin, streptomycin, sulfonamide, tetracycline and trimethoprim (SIGMA). Serial drug dilutions in solid media (from 512 to 0.5 εg ml⁻¹), were inoculated with 10 εl spots of a bacterial suspension (10⁴ cfu ml⁻¹) obtained from a log phase culture.

2.3 Laboratory bacterial strains and plasmids

The V. cholerae O1 395 strain used as positive control in conventional bacteriology and molecular biology experiments, and the pBB241 and pCVD27 plasmids used to provide probes in hybridization experiments for the detection of ace-zot and ctxA genes, were kindly provided by Dr. Alessio Fasano, University of Maryland. Escherichia coli HB 101 was used as negative control in conventional bacteriology and in PCR experiments. The nalidixic acid resistant derivative of E. coli CSH26 was used for conjugation experiments. The E.M. Lederberg Plasmid Reference Center provided the pR55 plasmid, incompatibility group C (gentamicin resistant), in rifampicin resistant E. coli CSH26, used for incompatibility tests.

2.4 DNA preparation

The Qiagen (Genenco, UK) commercial kit for plasmid DNA purification was used to obtain vector plasmid DNA. The plasmid content of 40 strains was analysed, including 20 clinical V. cholerae O1 representative of the 3 years of the epidemic, 12 environmental V. cholerae O1, 6 V. parahaemolyticus, 1 V. mimicus and 1 V. alginolyticus. Since difficulties in extracting and digesting large plasmid DNA from Vibrio strains by traditional techniques were experienced, a modification of the Kado and Liu [33] alkaline rapid method was used. Briefly, 2 M Tris-HCl was added to alkaline-lysed cell suspension (1:3 proportion) to lower the pH. Proteins were removed by phenol (unbuffered) chloroform/isoamyl/alcohol (25:24:1) treatments. DNA was ethanol precipitated in sodium acetate and washed with 70% ethanol. RNAse was added before restriction enzyme digestions. Restriction enzymes AvaI, EcoRV, HindIII, PstI and SalI (Boehringer Mannheim, Germany) were used according to the manufacturer's instructions.

Chromosomal DNA for restriction enzyme digestions, Southern blots, PCR ribotyping and PCR ERIC sequences, was prepared as suggested by Ausbel et al. [34]. Chromosomal DNA used as template for PCR toxin analysis was prepared by a crude method as previously described [35].

2.5 DNA hybridizations

DNA hybridization experiments were performed with the following probes: ClaI-XbaI 1493 bp fragment of pBB241 for ace and zot detection, and XbaI-ClaI 541 bp fragment cloned in the EcoRI site of pCVD27 for ctxA detection (Fig. 1). Fragments were extracted from agarose gels according to Sambrook et al. [36] and radiolabelled by random priming with the BRL commercial kit. Total DNA from 17 bacterial strains was digested by EcoRV, electrophoresed in 1% (w/v) agarose gel, transferred to a nylon membrane for Southern blots, and submitted to hybridization (Table 1). Bacterial strains were 2 V. cholerae O1 Inaba (strains no. 83 and 579), V. cholerae O1 Ogawa (no. 582), 2 Vibrio parahaemolyticus (no. 571 and 616), Aeromonas spp. (no. 589), Vibrio alginolyticus (no. 590), Vibrio mimicus (no. 75) and Klebsiella pneumoniae (no. 584) from patients; 5 V. cholerae O1 Inaba (no. 617, 618, 621, 703 and 623), V. cholerae O1 Ogawa (no. 622), V. parahaemolyticus (no. 698) and V. mimicus (no. 620) from the environment (Table 1). Hybridizations were performed under high stringency conditions [36].

Figure 1

Genetic and physical map of the CTX virulence cassette. Top: restriction fragments from plasmids pBB241 and pCVD27 used for hybridization experiments; toxin genes. Bottom: primer pairs and PCR amplicons. A=AluI; C=ClaI; E=EcoRV; H=HaeIII; S=SauIIIA; X=XbaI. Sizes are expressed in base pairs.

2.6 PCR amplifications and cycling conditions

ACE1-ACE2, ZOT1-ZOT2, ZOT3-ZOT4 [30], CTX2-CTX3 [36] and ACE1-ZOT2, ZOT1-ZOT4 primer pairs (Table 2) were used to detect ace, zot and ctxA genes (Fig. 1) in 103 strains.

Table 2

PCR oligonucleotide primers, amplicon sizes, restriction enzymes and restriction fragment sizes for detection of ace, zot, and ctxA genes

Gene	Primer pair	Sequence position	Oligonucleotide	Amplicon size	Restriction
					Enzyme	Fragments
ace (GenBank a.n. Z22569):
	ACE1	1322–1343	GCTTATGATGGACACCCTTTA	284	AluI	222, 62
	ACE2	1585–1606	TTTGCCCTGCGAGCGTTAAAC
zot (GenBank a.n. M83563):
	ZOT1	275–304	ACGTCTCAGACATCAGTATCGAGTTTATTG	198	HaeIII	127, 53, 16
	ZOT2	452–474	ATTTGGTCGCAGAGGATAGGCCT
	ZOT3	1020–1043	CACTGTTGGTGATGAGCGTTATCG	243	AluI	139, 90, 14
	ZOT4	1240–1263	CAAGCGCTGTGGGTAGAAGTGAAA
	ZOT1			988	AluI	679, 204, 90, 14
	ZOT4				HaeIII	493, 298, 127, 53, 16
					EcoRV	741, 247
ace-zot:
	ACE1			639	AluI	361, 217, 62
	ZOT2				SauIIIA	584, 55
ctxA[37]:
	CTX2	73–94	CGGGCAGATTCTAGACCTCCTG	564	AluI	500, 64
	CTX3	614–636	CGATGATCTTGGAGCATTCCCAC

Oligonucleotides were synthesized with the ‘Gene Assembler Plus’, (Pharmacia), according to the user manual. A thermocycler (Perkin Cetus) and Taq polymerase (Promega) were used to amplify 1 εl of crude DNA preparation under study as directed by the manufacturers.

The PCR conditions consisted of a preincubation step of 3 min at 94°C and finished with incubation for 10 min at 72°C, with a middle step consisting of: 30 cycles of 1 min at 94°C, 1 min at 57°C, 2 min at 72°C (for CTX2-CTX3, ZOT1-ZOT2 and ZOT3-ZOT4); 35 cycles of 1 min at 94°C, 1 min at 54°C and 1 min and 30 sec at 72°C (for ACE1-ACE2); 30 cycles of 1 min at 94°C, 1 min at 57°C and 2 min at 72°C (for ZOT1-ZOT4); 30 cycles of 1 min at 94°C, 1 min at 56°C and 2 min at 72°C (for ACE1-ZOT2).

Oligonucleotide primers and conditions for PCR ribotyping were adopted as suggested by Kostman et al. [25] with 500 ng of DNA spectrophotometrically measured; PCR ERIC polymorphism detection was made according to Rivera et al. [27] with 100 ng of DNA; both tests were made on 16 clinical isolates representative of the 4 years of the epidemic and in all 12 environmental isolates of V. cholerae O1.

Appropriate negative controls (withholding template DNA and using E. coli DNA) and positive controls (with V. cholerae O1 395 DNA) were added to each set of PCR reactions.

Electrophoresis of amplicons was made in 1.5% (w/v) agarose (Bio-Rad) gels, 45 mM Tris-Borate, 1 mM EDTA. Gels were stained with ethidium bromide. Amplified DNA to be submitted to restriction analysis was purified by ‘MicroSpinTM Sephacryl Elmer Sample Pack’ column (Pharmacia) utilized as directed by the manufacturer. About 1.0 εg of amplified DNA was digested by AluI, EcoRV, HaeIII and SauIIIA (Boehringer Mannheim, Germany) according to the amplicon restriction map (see Fig. 1). Ribotyping PCR amplicons were digested by AluI, HindIII and HinfI. Lambda DNA, digested by HindIII restriction enzyme, and Marker VI (Boehringer Mannheim, Germany) were used as DNA size markers. Chemicals were purchased from Sigma.

2.7 Evaluation of restriction fragment and PCR amplicon sizes

Migration positions of gel electrophoresis bands of molecular mass standards and unknown fragments were plotted by a macro for Microsoft EXCEL 4.0 specifically developed for this purpose. This unpublished method, named EDE.XLM (EDE: EXCEL DNA size estimation, and XLM is the macro file extension recognized by EXCEL) is able to estimate DNA fragment length from gel electrophoresis.

3 Results

3.1 Drug resistance and plasmid content

All clinical V. cholerae O1 and three of the environmental V. cholerae O1 were multi-resistant to ampicillin, chloramphenicol, erythromycin, kanamycin, streptomycin, sulfonamide, tetracycline and trimethoprim. The same strains were susceptible to gentamicin, nalidixic acid and rifampicin. Other Vibrio spp. isolates were susceptible to all the drugs tested.

A large plasmid was found in all the clinical (20 isolates) and in three environmental V. cholerae isolates showing the multi-resistant pattern. The environmental V. cholerae strains which did not harbour this plasmid did not show the multi-resistance pattern. This plasmid conjugated with E. coli HB 101, V. cholerae O1 395 and with V. parahaemolyticus (wild-type no. 571) and transferred its resistance pattern; it was recalcitrant to common methods of curing (acridine orange and ethidium bromide). It was shown to be in the incompatibility class C by conjugation experiments.

Plasmid DNAs were submitted to restriction analysis with AvaI, EcoRV, HindIII and SalI enzymes. A sample of restriction profiles obtained by EcoRV and SalI is shown in Fig. 2. Transconjugant plasmids in E. coli showed the parental restriction patterns (data not shown). No physical differences were revealed by restriction analysis among the isolates, suggesting substantial homogeneity of the plasmid, estimated to have a size of about 100 kb.

Figure 2

Restriction analysis of V. cholerae O1 plasmid DNA. Lanes M: HindIII digested lambda DNA, molecular mass marker. A: Undigested DNAs and SalI restriction patterns of representative clinical (lanes 1–2 strain no. 83; lanes 3–4 strain no. 582) and environmental (lanes 5–6 strain no. 617; lanes 7–8 strain no. 623) V. cholerae O1 plasmids. Faint bands between 17.5 and 8.0 kb are partially digested fragments. B: EcoRV restriction patterns of nine representative V. cholerae O1 plasmids including the four shown in A. Clinical strains are no. 83, lane 1; no. 582, lane 2; no. 583, lane 3; no. 612, lane 4; no. 676, lane 5; no. 680, lane 6). Environmental strains are: no. 617, lane 7; no. 623, lane 8; no. 703, lane 9. *: double band. C: Undigested and EcoRV digested plasmid DNAs of a clinical strain (no. 579) with an unexpected additional plasmid: lane 1, undigested plasmid DNA reveals an additional DNA band suspected to correspond to an additional plasmid, absent in all the other isolates; lane 2, two additional bands, corresponding to 18.9 and 6.0 bp restriction fragments, characterize the plasmid. In order to obtain a better resolution of upper bands, the gel was underloaded and the two EcoRV lower bands (0.66 and 0.56 kb) are invisible; lanes 3 and 4, typical undigested and digested plasmid pattern of strain no. 83. Sizes are expressed in kb.

A single clinical isolate of V. cholerae O1 (strain no. 579) exhibited polymorphism in the undigested plasmid DNA pattern (Fig. 2C, lane 1) and the presence of two additional fragments of 18.9 and 6.0 kb upon EcoRV digestion (Fig. 2C, lane 2). The plasmid profile of this strain loses the two additional fragments after several passages in culture or through conjugation. The results suggest the presence of an unstable plasmid of about 25 kb in addition to the common large resistance plasmid.

None of the clinical and environmental Vibrio spp. under study harbour the epidemic multi-resistance plasmid.

3.2 Distribution of toxin genes

Two probes were used initially to detect the presence of the virulence cassette in hybridization experiments (Fig. 1). Fig. 3 shows a sample of 8 hybridizations. The zot probe revealed two hybridization bands corresponding to the 3.7 kb and 2.9 kb EcoRV fragments, which contain respectively the right and the left portions of the cassette (Fig. 3A). The ctxA probe revealed only the 3.7 kb EcoRV fragment containing the right portion of the cassette (Fig. 3B). The two flanking EcoRV sites are located in the RS sequences not shown in Fig. 1.

Figure 3

Toxin gene detection by hybridization. A: Hybridization with ClaI-XbaI 1493 bp probe (ace and zot genes) detecting both EcoRV fragments. B: Hybridization with the 541 bp ctxA probe revealing only the right EcoRV fragment. Lanes for two panels: lane 1, clinical V. parahaemolyticus (strain no. 571); lanes 2 and 3, clinical V. cholerae O1, Inaba (strain no. 579) and Ogawa (strain no. 582); lane 4, environmental V. cholerae O1 (strain no. 617); lane 5, clinical Kl. pneumoniae (strain no. 584); lane 6, V. cholerae O1 395 control strain; lane 7, environmental V. cholerae O1 (strain no. 621); lane 8: environmental V. parahaemolyticus (strain no. 698). Sizes are expressed in kb.

Only V. cholerae O1 strains were positive for toxin genes. DNA from the clinical isolate Kl. pneumoniae, strain no. 584 (Fig. 3, lane 5) anomalously showed a single band of approximately 10 kb with both probes.

As a more rapid screening method of higher specificity, 103 strains randomly selected from the epidemic patients and from the environment were submitted to PCR amplification of the three toxin genes. In order to test the presence of the cassette in other species, 20 Vibrio spp. and enterobacteria were included in the screening in addition to 83 V. cholerae O1 (Table 1).

The ACE1-2 amplicon, covering almost the entire ace gene sequence, and the ZOT1-2 and ZOT3-4 amplicons corresponding to the left and the right portions of the zot gene (Fig. 1) were tested extensively. To confirm the identity of the amplified areas, the ACE1-ZOT2 amplicon including the ace and part of the zot sequences, and ZOT1-ZOT4, corresponding to the central portion of zot gene, were also tested. Furthermore, the central portion of ctxA gene was investigated (CTX2-3 amplicon). Amplified DNAs were also submitted to restriction enzyme analysis for confirmation in specific cases.

Amplifications and restrictions obtained with the control strain V. cholerae O1 395 Inaba DNA are shown in Fig. 4. Congruent patterns were obtained with the natural isolates (data not shown). All 71 clinical V. cholerae isolates contained the virulence cassette with no detectable polymorphism. No amplifications were obtained with the other Vibrio spp. or other genera tested, including the Kl. pneumoniae isolate which showed homology to the ctx and zot probes in the Southern blot experiment of Fig. 3. Results are summarized in Table 1.

Figure 4

PCR amplicons of ace, zot and ctxA toxin determinants in V. cholerae O1 395 and their digestions. See Fig. 1 and Table 2 for details. Fragments smaller than 100 bp are undetectable. Lanes M: DNA molecular mass Marker VI (from 2176 to 154 bp, the last fragment is undetectable on the photograph). A: Lanes 1 and 2, ACE1-2 amplicon, AluI digested and uncut respectively; lanes 3, 4 and 5, ACE1-ZOT2 amplicon, AluI, SauIIIA digested and uncut; lanes 6 and 7, ZOT1-2 amplicon, HaeIII digested and uncut. B: Lanes 1, 2, 3 and 4, ZOT1-4 amplicon, AluI, HaeIII, EcoRV digested and uncut; lane 1, the upper band is a partial digestion artifact; lanes 5 and 6, ZOT3-4 amplicon, AluI digested and uncut; lanes 7 and 8, CTX2-3 amplicon, AluI digested and uncut. Sizes are expressed in base pairs.

A PCR test for the presence of the toxin cassette in a number of bacterial strains, including V. cholerae O1 (Table 1) isolated from the water samples indicated that four V. cholerae O1 isolates (including the two showing homology to zot and ctxA probes) were positive. On the other hand, a total of eight strains of V. cholerae with regular culturing, biochemical and serological features, including four showing no homology in hybridization experiments, were negative by PCR amplification.

The clinical isolates of V. cholerae O1 were characterized by high stability of the toxigenic cassette and of the multi-resistant plasmid during the four years of the epidemic (Table 3, group A), whereas the 12 environmental strains show three different genetic profiles: presence of both cassette and plasmid (group B: three strains: no. 617, no. 623 and no. 703), presence of the cassette alone (group C: one strain, no. 618), and absence of both cassette and plasmid (group D: the most common, with eight strains out of 12, including strain no. 621 shown in Fig. 3).

Table 3

Definition of four groups of epidemic strains on the basis of their sources, PCR analysis of the toxigenic cassette, plasmid content, and PCR ERIC pattern

Group	Source	No. of strains	Toxigenic cassette	Plasmid	PCR ERIC pattern
A	Human	16	+	+	T
B	Water	3	+	+	T
C	Water	1	+	−	T
D	Water	8	−	−	Various

T=specific profile showed by all toxigenic isolates.

3.3 Ribotyping and ERIC chromosomal polymorphism

In order to track the clonal relationship among the four groups of strains described above (Table 3), their chromosomal fingerprints by PCR ribotyping and analysis of PCR ERIC sequences were examined.

A sample of 16 clinical isolates of V. cholerae O1 (group A), representative of the different years of the epidemic, and all 12 environmental isolates (groups B, C and D), were analysed by PCR amplification with ribotyping primers. PCR ribotyping patterns of the four groups and of the reference V. cholerae O1 395, show no detectable differences, exhibiting the same fingerprint of 3 amplicons: 1.0, 0.8 and 0.7 kb (Fig. 5A). Digestion of the resulting amplicons with different restriction enzymes (data not shown), was done but RFLPs were not observed. This is the first report where PCR ribotyping was applied to V. cholerae molecular epidemiology.

Figure 5

A sample of 8 PCR ribotyping (A) and ERIC polymorphism (B) of genomic DNA from V. cholerae O1 isolates, representing the four groups of genetic profiles (Table 3). Lanes for both panels: lane M, molecular mass marker VI; lane 1, V. cholerae O1 395 reference strain; lane 2 and 3, V. cholerae O1 579 and 663 respectively, clinical isolates belonging to group A; lane 4 and 5, V. cholerae O1 703 and 617 respectively, environmental isolates belonging to group B; lane 6, V. cholerae O1 621, environmental isolate belonging to group D; lane 7, V. cholerae O1 618, environmental isolate belonging to group C. Sizes are expressed in kb.

Further investigation by PCR ERIC polymorphism (Fig. 5B) confirmed the substantial identity of toxigenic groups A, B and C, characterized by a common fingerprint of 8 ERIC amplicons: 3.0, 2.6, 1.72, 1.44, 1.14, 0.89, 0.78 and 0.58 kb. It revealed a difference with the reference toxigenic strain, characterized by the same number of amplicons but lacking the 7th fragment and showing an additional amplicon of 1.36 kb. The non-toxigenic group D showed a very different fingerprint characterized by only 6 amplicons: the common 3.0, 1.44, 1.14, 0.89 and 0.58 kb fingerprint, and an additional amplicon of 1.07 kb (Fig. 5B, lane 6).

4 Discussion

The insurgency of multi-resistant strains in Luanda was first reported in 1988 [38]. The V. cholerae O1 strain epidemic in Luanda province is still characterized by multi-resistance to drugs, coded by a specific plasmid in all the clinical strains and in some environmental strains, and absent in other species. The presence of large conjugative plasmids belonging to the incompatibility class C was found in almost all African epidemics [13,39] and it suggests that this plasmid was a stable character in these cholera outbreaks.

Multi-resistance to drugs is therefore a phenotypic character of clinical and epidemiological interest, but of overriding clinical concern is the ability to produce toxins, since this determines pathogenicity. During the preliminary phase of research the toxigenic potential of the strains was screened by testing a representative sample of Vibrio isolates by colony hybridization, with the zot and ctxA sequence probes. This method provides limited information, since it does not detect genetic rearrangements. Southern blot hybridization with the two probes was more informative; the hybridizing restriction fragments found in the V. cholerae O1 isolates were consistent with the published restriction map of the core region [9]. However, this technique was non-specific, it also revealed an anomalous homology of Kl. pneumoniae to the toxin region, a result that needs further investigation.

For higher throughput, it was decided to utilize the PCR method, because of its high sequence specificity and of the opportunity to further investigate resulting amplicons. Its reliability, in substitution of DNA hybridization and ELISA test for toxins, is well demonstrated [21, 35, 36, 40]. Primers (ACE1-2, ZOT1-2, ZOT3-4, CTX2-3) were used for ace and zot as previously described [28,36] and for the first time in a different combination (ACE1-ZOT2, ZOT1-ZOT4) in order to obtain larger amplicons to reveal possible polymorphism in the toxin cassette.

Primers for the ctxA toxin gene [36] were used to enable comparison with other studies. Six primers were used to systematically test a sample of 91 isolates and to study in detail 2.4 kb of the toxin area in the 4.5 kb core region, essential to the cholera pathogenic mechanism.

Our experiments revealed no polymorphism of the core toxin region among the isolates, and confirmed the co-occurrence of ace, zot and ctxA as a single unit, belonging to the CTX phage [41,42] in 71 V. cholerae O1 clinical isolates, including the rare Ogawa serotype and in 4 environmental toxigenic isolates.

Since the clinical isolates under study, randomly chosen to represent four years of epidemics, showed no polymorphism of the toxigenic cassette, ribotyping and ERIC pattern, it was concluded that Luanda province was affected by a single epidemic V. cholerae O1 strain, harbouring the same resistance plasmid selected by drugs (mainly tetracycline) used for the patient treatment.

The toxin genes were not uniformly present in V. cholerae O1 isolated from water samples. In eight of 12 isolates no amplification product was obtained with any primer pair (group D). It is known that not all environmental strains are toxigenic [41,42], since the toxin cassette can be lost by recombination between the left and right RS sequences, excising the entire CTX phage DNA [9,11].

Environmental toxigenic (group C) and non-toxigenic (group D) isolates without the resistance plasmid in spite of its high stability were also found, but non-toxigenic strain harbouring the epidemic plasmid wasn't found. This supports the idea of the resistance plasmid as an accessory feature strictly related to pathogenicity.

To investigate the clonal relationship among the clinical strains and the three environmental genetic groups, and to understand the dynamic of the spread of the epidemic through water, chromosomal polymorphism of ribosomal genes and ERIC sequences were tested as epidemiological markers in the isolates. PCR ribotyping did not discriminate the epidemic strains.

Rivera et al. [27] proved that ERIC PCR was able not only to differentiate toxigenic from non-toxigenic strains of V. cholerae O1, but also to reveal a clear distinction among different strains not clonally related, independently of the presence of the toxigenic cassette, since it examines the distribution of dispersed repetitive DNA sequences not linked to the toxigenic cassette [26,27]. This technique allowed not only discrimination of toxigenic from non-toxigenic Angolan strains, in accordance with PCR detection of toxin genes, but also allowed the discrimination of toxigenic Angolan strains from the reference strain. In fact V. cholerae O1 reference strain 395 differed from toxigenic clinical and environmental Angolan isolates by the absence of the 7th ERIC amplicon (0.78 kb) and the presence of an additional ERIC amplicon of 1.36 kb (Fig. 5B, lane 1). Certainly the 2 fingerprints included the 4 ERIC amplicons of group FP1 described by Rivera et al. [27] even if some slight differences in molecular masses, probably due to different methods of calculation, occurred.

The non-toxigenic environmental isolate (group D) fingerprint was characterized by the presence of 5 common ERIC amplicons of 3.0, 1.44, 1.14, 0.89 and 0.58 kb and was differentiated from the toxigenic strain by the substitution of the 1.72 kb amplicon by another amplicon of 1.07 kb. Moreover, the 2.6 and 0.78 kb amplicons were not detected (Fig. 5B, lane 6). This suggested the occurrence of a chromosomal rearrangement rather than a simple excision of the toxigenic CTX phage genome. In conclusion, a common ERIC fingerprint among all the strains under study included the 3.0, 1.44, 1.14, 0.89 and 0.58 kb amplicons.

Our results suggested that the toxigenic V. cholerae O1 environmental strains, in groups B and C, had the same clonal origin as the epidemic strain, independent of the resistance plasmid, and that the environmental non-toxigenic ERIC variant of group D was not clonally related to the epidemic strain.

Environmental strains were isolated from a river estuary which serves as the source of urban water, although the area itself is neither densely populated nor directly polluted by urban sewage. This particular environment, rich in salty lagoons, appears to support the endemic V. cholerae O1 after its penetration into the area as faecal contaminate. The persistence of cholera mostly in the coastal provinces of Angola, characterized by similar environments, reinforces the idea that this specific ecosystem is favourable to the survival and growth of different V. cholerae genetic variants, regardless of their toxigenicity and plasmid content. Thus, strains present in this environment may not all come directly from cholera patients' stools, and the Bengo River area may be defined as a natural Vibrio reservoir.

During the oral-faecal transmission in the recurrent epidemic of cholera, the toxigenic strains could acquire the resistance epidemic plasmid, or strains still harbouring the plasmid could be positively selected by drug, mostly tetracycline, treatment of patients.

In conclusion we recommend CTX cassette detection by PCR technique in at least one environmental V. cholerae O1 isolate, in order to define a certain environment as causative of cholera infections.

Acknowledgements

We are grateful to A. Fasano, A. Bortolan, G. Riva and P. Donini for helpful discussion; to the Provincial Commission of Struggle Against Cholera and the National Directorate of Public Health of Ministery of Health, Angola, and to the W.H.O. Geneva and Luanda offices for providing very useful information and logistic support; and to M. Romano, M. Francisco, A. Pinto de Andrade, G. Erre and C. Morciano for assistance. T. Picano helped to develop the EDE.XLM EXCEL macro. K. Williams made useful comments on the manuscript. This work was funded by: Ministero Università Ricerca Scientifica Tecnologica (60% and 40%), Italy; Fondaz. ‘Cenci Bolognetti-Ist. Pasteur’, Italy; ‘Cholera Control in Subsahara Africa’, ‘Support to National Laboratory of Public Health, Luanda’ and ‘Support to University A. Neto, Luanda’ programmes by Direzione Generale Cooperazione Sviluppo, Ministero Affari Esteri, Italy; Ministerio de Saude, Angola.