Comparison of Direct Genome Restriction Enzyme Analysis and Pulsed-Field Gel Electrophoresis for Typing of Vibrio vulnificus and Their Correspondence with Multilocus Sequence Typing Data

Abstract

We compared the potential of direct genome restriction enzyme analysis (DGREA) and pulsed-field gel electrophoresis (PFGE) for discriminating Vibrio vulnificus isolates from clinical (23) and environmental (17) sources. The genotypes generated by both methodologies were compared to previous multilocus sequence typing (MLST) data. DGREA established clearer relationships among V. vulnificus strains and was more consistent with MLST than with PFGE. DGREA is a very promising tool for epidemiological and ecological studies of V. vulnificus.

Vibrio vulnificus is a bacterium ubiquitous in marine environments. From a genetic perspective, the species is highly diverse and numerous strains can be found simultaneously in a single contaminated water or oyster sample (11). However, only a single strain is typically isolated from the blood of an infected person, suggesting that among the diverse V. vulnificus population to which a human is exposed, only certain strains cause infection (11). It is unclear what makes a particular strain virulent, but virulence appears for only a small proportion of naturally occurring strains. A number of methodologies have been utilized to study the genetic diversity in V. vulnificus. One of them is based on the presence of a polymorphic sequence in the 16S rRNA genes (rrs) originally identified using terminal restriction fragment length polymorphism (16). Nilsson et al. reported that most clinical V. vulnificus strains possessed a sequence designated type B in the rrs genes (16). Subsequently, a real-time PCR assay was developed to determine rrs types in V. vulnificus and soon it was discovered that some strains possess both A and B rrs types (22). More recently, a 200-bp amplicon originally identified by random amplified polymorphic DNA (RAPD) analysis was also reported to be associated primarily with clinical (designated type C) V. vulnificus strains and a conventional PCR assay was developed to detect these strains (17). An independent group using the above-described PCR assays reported that there was a 100% correspondence between rrs type B and C isolates (4).

While PCR methods are able to group V. vulnificus isolates into several broad categories that seem to correlate fairly well with clinical and environmental sources, they do not permit finer discriminations that are needed for epidemiological and ecological studies. Among the methods employed for the differentiation of V. vulnificus strains are pulsed-field gel electrophoresis (PFGE), RAPD, arbitrarily primed PCR (AP-PCR), multilocus enzyme electrophoresis, ribotyping, repetitive extragenic palindromic PCR (rep-PCR), and multilocus sequence typing (MLST) (2, 4, 9, 10, 17, 21, 23). Some of these genotyping techniques are difficult to reproduce (e.g., RAPD and AP-PCR), and others are laborious and time-consuming (e.g., PFGE and MLST). PFGE may be particularly troubling, as there is typically a high proportion of untypeable V. vulnificus strains using this method (1). rep-PCR appears to be more reproducible than AP-PCR and RAPD, but recent results suggest that clinical strains typed by this method were distributed among multiple rep-PCR genogroups, suggesting a high degree of genetic diversity (4). This may perhaps limit the usefulness of this method for making epidemiological associations. MLST is considered the gold standard because it is sequence based and can be replicated easily among laboratories; however, MLST is considerably more resource intensive and expensive, making this methodology impractical for the analysis of a large number of V. vulnificus isolates.

A new methodology for genetic diversity analysis based on DNA restriction using an endonuclease with a high cutting frequency (much like PFGE) has become available. This methodology, direct genome restriction enzyme analysis (DGREA), produces small and discrete DNA fragments that can be resolved easily by using nondenaturing polyacrylamide gel electrophoresis (7). DGREA was employed successfully for the characterization of V. parahaemolyticus strains associated with an outbreak in Chile in 2006 (7). In the case of V. parahaemolyticus, the discriminatory power of DGREA was similar to that of PFGE but the low cost and speed of analysis made the technique very attractive for the study of genetic diversity for this bacterium. In the present study, we compared PFGE and DGREA methodologies for their abilities to type a selected group of V. vulnificus strains (clinical and environmental isolates). Furthermore, since the V. vulnificus strains employed in this study were characterized previously by MLST (2), the relative relationships between strain typing methods, including MLST, were compared.

Bacterial strains, media, and DNA extraction.

A total of 40 V. vulnificus strains, representing both clinical (n = 23) and environmental (n = 17) sources (Table 1), were used in this study. These strains were chosen because they were examined previously for rRNA (rrs) sequence type (ST) (16) and by MLST (2). Bacterial strains were grown overnight with shaking (200 rpm) at 37°C in Luria-Bertani medium supplemented with 2% NaCl. Bacterial DNA was extracted using the Wizard genomic DNA purification kit (Promega, Madison, WI) as recommended by the manufacturer except that 50 μl instead of 200 μl of elution buffer was used for resuspension of the precipitated DNA to achieve final DNA concentrations of approximately 1 μg/μl.

TABLE 1.

Characteristics of the V. vulnificus strains employed in this study^a

Strain	STb	rrs typec	Yr of isolation	State of origin	Source
DAL 7-9002 (CDC 9060-96)d	2	B	1996	Texas	Clinical
LOS 6966 (CDC 9342-95)d	2	B	1995	Texas/Lousiana	Clinical
DAL 7-9087 (CDC 9005-97)	15	A	1997	Louisiana	Clinical
ATL 6-1306 (CDC 9031-96)	16	AB	1996	Florida	Clinical
ATL 71503 (CDC 9075-96)	16	AB	1996	Florida	Clinical
NSV 5829 (CDC 9149-95)	16	AB	1995	Florida/Louisiana	Clinical
ORL 1506 (CDC 9030-95)	16	AB	1995	Florida	Clinical
99-624 DP-C10	17	A	1999	Texas	Shellfish
DAL 6-5000 (CDC 9345-95)	18	B	1995	Louisiana	Clinical
ATL 9579	19	A	1994	Texas	Clinical
DAL 79040 (CDC 9070-96)d	21	B	1996	Texas	Clinical
99-742 DP-A9	22	AB	1999	Mississippi	Shellfish
99-796 DP-E7	22	AB	1999	Florida	Shellfish
99-584 DP-B12	23	A	1999	Texas	Shellfish
98-640 DP-E9e	24	A	1998	Louisiana	Shellfish
99-743 DP-B6d	25	B	1999	Texas	Shellfish
98-783 DP-A1	26	A	1999	Louisiana	Shellfish
99-645 DP-C4e	27	A	1999	Texas	Shellfish
99-780 DP-E1	27	A	1999	Louisiana	Shellfish
99-738 DP-B5d	28	A	1999	Florida	Shellfish
ATL-9572d	29	A	1994	Florida	Clinical
99-540 DP-B6d	30	A	1998	Texas	Shellfish
99-23 DP-F5	31	AB	1998	Florida	Shellfish
ATL 71491 (CDC 9074-96)	32	B	1996	Texas/Louisiana	Clinical
ATL 71504 (CDC 9076-96)	32	B	1996	Louisiana	Clinical
LOS 7343 (CDC 9062-96)	32	B	1996	Louisiana	Clinical
NSV 5736 (CDC 9349-95)	33	B	1995	Alabama	Clinical
99-736 DP-C7	34	AB	1999	Florida	Shellfish
ATL-9824	36	B	1994	Texas	Clinical
ATL 9823 (CDC 9352-94)	37	B	1994	Louisiana	Clinical
ATL-9580	37	B	1994	Texas/Louisiana	Clinical
ORL 8324 (CDC 9340-95)	37	B	1995	Florida/Louisiana	Clinical
FLA 8869 (CDC 9053-96)	40	B	1996	Texas	Clinical
FLA 9509 (CDC 9003-97)	40	B	1997	Louisiana	Clinical
99-520 DP-B8	41	AB	1998	Rhode Island	Shellfish
DAL 7-9000 (CDC 9067-96)d	42	B	1996	Texas	Clinical
99-622 DP-E4d	54	A	1998	Texas	Shellfish
99-609 DP-A4	62	A	1999	Oregon	Shellfish
98-641 DP-G8	63	AB	1998	Louisiana	Shellfish
ORL 8074 (CDC 9032-95)	64	B	1995	Texas	Clinical

^aAll clinical strains were from primary septicemia cases.

^bST, sequence type by MLST (http://pubmlst.org/vvulnificus/).

^c16S rRNA genes. Data published in Vickery et al. (22).

^dStrains that were not typeable by PFGE using NotI or SfiI.

^eStrains that were not typeable by DGREA.

Application of DGREA to V. vulnificus strains.

DGREA was performed as described previously (7). Briefly, each reaction mixture consisted of 10 μg DNA digested with 10 U of NaeI endonuclease (six-base restriction endonuclease, GCC→GGC) (Promega) for 2 h at 37°C. Each digestion reaction mixture was treated with proteinase K (0.020 μg/μl) (QIAGEN, Valencia, CA) for 1 h at 37°C. Eight microliters of each digestion was electrophoresed in 8% nondenaturing polyacrylamide gels (1 mm thick) for 3 h at 100 V. Bands were visualized by silver staining as described previously (8) and photographed using a Canon PowerShot A620 digital camera. DGREA and PFGE fingerprints were analyzed using the BioNumerics software package (version 4.6; Applied Maths, Kortrijk, Belgium). The DGREA and PFGE pattern analysis and genetic similarity coefficients were calculated using Dice correlation at 1.5% and 1.0% band position tolerance, respectively. Dendrograms were constructed using complete linkage.

Thirty-eight of the 40 (95%) V. vulnificus strains employed in this study were typeable by DGREA (Table 1). DNA from only two strains (98-640 DP-E9 and 99-645 DP-C4) was indigestible by the restriction enzyme NaeI. NaeI is a six-base restriction endonuclease (GCC→GGC) that is sensitive to CpG methylation, which blocks its endonuclease activity (www.fermentas.com/catalog/re/pdii.html); this is a likely explanation for the failure to digest the DNA of these two strains. This explanation was confirmed by treatment of the same DNA samples with a different restriction enzyme (BamHI) for which endonuclease activity was not affected by any methylation pattern; this treatment resulted in successful DNA digestion (data not shown).

DGREA patterns of the 38 V. vulnificus strains clustered into two groups designated clusters I and II (Fig. 1). Cluster I contained most of the strains (68%) employed in this study. The genotypes of these strains (according to rrs type) were variable, since all three rrs types described previously for V. vulnificus were observed for this group (22). This cluster contained all of the environmental and 49% of the clinical isolates. Interestingly, cluster II contained only clinical V. vulnificus strains of rrs type B. This cluster contained 51% and 71% of all the clinical and rrs type B strains, respectively.

FIG. 1.

DGREA fingerprinting and corresponding dendrogram for 38 of 40 V. vulnificus strains used in this study. ST by MLST. S, shellfish; C, clinical; AL, Alabama; FL, Florida; LA, Louisiana; MS, Mississippi; OR, Oregon; RI, Rhode Island; TX, Texas.

Fuenzalida et al. showed that V. parahaemolyticus strains belonging to the pandemic clonal complex were differentiated accurately and rapidly from other V. parahaemolyticus strains using DGREA (7). The V. vulnificus strains used in the present study were more diverse, and the cohesive clusters observed for V. parahaemolyticus were generally not observed, with the possible exception of four closely related clinical strains (rrs type AB) isolated from Florida in 1995 and 1996. Other than for the recently identified V. vulnificus biotype 3 (1), a clonal complex similar to that seen for V. parahaemolyticus has not been described for V. vulnificus. Interestingly, V. vulnificus biotype 3 strains belong to the same ST by MLST but were not typeable by PFGE (1).

Application of PFGE to V. vulnificus strains.

PFGE was performed according to the CDC protocol for V. cholerae (www.cdc.gov/pulsenet/protocols/vibrio_May2006.pdf). Plugs were digested using either 40 U NotI or 50 U SfiI enzymes (Promega). The DNA digestions were run in 1% SeaKem Gold gel agarose (Cambrex, Rockland, ME) in 0.5× Tris-borate-EDTA. Electrophoresis run conditions were programmed in two blocks with 6 V and an included angle of 120° on a CHEF-DR III (Bio-Rad, Hercules, CA). Block 1 ran for 13 h with an initial and final switch time of 2 s and 10 s, respectively, while block 2 ran for 6 h with an initial and final switch time of 20 s and 25 s, respectively. The 0.5× Tris-borate-EDTA running buffer was maintained at a constant temperature of 14°C. After electrophoresis, bands were visualized by staining in ethidium bromide (1 μg/ml). Gels were photographed by using a Gel Doc 1000 (Bio-Rad, Hercules, CA).

Only 31 of the 40 (77.5%) V. vulnificus strains tested were typeable by PFGE (Table 1). Interestingly, the two strains that were untypeable by DGREA were typeable by PFGE, while all of the strains untypeable by PFGE showed discrete banding patterns when DGREA was used. The occurrence of nontypeable V. vulnificus strains is not uncommon and may be a consequence of the degradation of DNA during its preparation for PFGE or during electrophoresis (3, 18). In some cases, this has been solved by the addition of thiourea to the running buffer during electrophoresis (20). In our case, even the use of thiourea did not eliminate the problem, as only two more strains (ATL 9572 and 99-540) were typeable and those were typeable only with the SfiI enzyme. These were not included in the PFGE analysis (data not shown).

When NotI was used, two major clusters, I and II, were observed at the 30% similarity level. Cluster II contained 10 of the 12 typeable rrs type B strains (Fig. 2A). All strains in this cluster were clinical isolates with the exception of isolate 99-736, an environmental strain possessing rrs type AB. SfiI also produced two major clusters at the 30% similarity level; however, both clusters contained strains carrying any of the three rrs types described previously for V. vulnificus (22) (Fig. 2B). In other studies, PFGE has been applied to a variety of Vibrio species and, in some cases, strain types were shown to correlate with serotype and pathogenicity (7, 13, 19). Although PFGE is commonly used for the typing of clinical isolates of V. vulnificus, previous work suggested that environmental strains are quite heterogeneous, making the technique less useful for determining relationships between strains isolated from different sources (24, 21). Our study further confirms the high degree of heterogeneity for environmental strains of V. vulnificus even though we observed clustering associated with rrs type.

FIG. 2.

PFGE fingerprinting and corresponding dendrogram for 31 of 40 V. vulnificus strains used in this study. Only 31 of the 40 strains were typeable by using either restriction enzyme employed (NotI or SfiI). (A) Fingerprints with NotI. (B) Fingerprints with SfiI. ST by MLST. S, shellfish; C, clinical; AL, Alabama; FL, Florida; LA, Louisiana; MS, Mississippi; OR, Oregon; RI, Rhode Island; TX, Texas.

Comparison of V. vulnificus strain typing data obtained using DGREA, PFGE, and MLST.

The MLST data of the V. vulnificus strains were retrieved from the public V. vulnificus database (http://pubmlst.org/vvulnificus/). The concatenated sequences of 10 housekeeping genes were used for phylogenetic distance calculation. The phylogenetic tree was constructed using the BioNumerics software package (Applied Maths, Kortrijk, Belgium). Dendrograms were constructed using complete linkage as clustering criteria.

MLST data correlated well with the rrs type, as illustrated by the fact that all but one of the rrs type B strains fell into one homogenous cluster (Fig. 3A). This one rrs type B strain (99-743) was isolated from shellfish, which may explain why it was so different from the rrs type B clinical strains. Among the strains that shared the same ST, this relationship was upheld by DGREA for 15 of the 16 typeable strains. For example, ATL 71491, ATL 71504, and LOS 7343 all had the same ST (ST 32), but only ATL 71491 and ATL 71504 possessed highly similar DGREA patterns (95% similarity); LOS 7343 fell in a different subcluster (Fig. 1). Typing by PFGE upheld the relationship in 10 and 12 of the 16 typeable strains using restriction enzymes SfiI and NotI, respectively (Fig. 2).

FIG. 3.

Dendrograms generated by MLST (A) and DGREA (B) for the 40 V. vulnificus strains used in this study. PFGE data are not shown since their correspondence with MLST was lower than that displayed by DGREA.

When the dendrograms generated by DGREA and MLST are compared, some findings warrant further discussion (Fig. 3A and B). For example, some V. vulnificus strains belonging to the same ST showed identical DGREA patterns, including ST 40 and ST 37. However, some other STs showed variable degrees of homology when typed using DGREA (e.g., for ST 37, 76% to 95%) and some were even located in different clusters (e.g., ST 32). These results could be explained in part because MLST analysis employs the sequencing of housekeeping genes that are characterized by a low rate of variation relative to those of other targets (5, 14, 15). It was reported that the native marine environment may provide a habitat promoting high degrees of gene transfer by transduction (12), which could increase genome variation even among Vibrio isolates with the same ST. These types of events will not be reflected in MLST, but they could be seen with DGREA or PFGE since these two methodologies analyze the entire genome.

Further analysis of the V. vulnificus MLST, PFGE, and DGREA data revealed that some V. vulnificus strains were related geographically and genetically. A cluster consisting of four clinical isolates associated with oysters harvested in Florida in 1995 and 1996 with MLST ST 16 (Table 1) were also highly related by PFGE and DGREA. In previous studies, these were the only isolates possessing a 29-kb plasmid (6) and also were the only isolates with the rrs type AB of clinical origin (22). These same strains were also shown to be closely related using rep-PCR (4). We consider it significant that six different genotyping methodologies were able to distinguish this group of isolates, suggesting that the V. vulnificus population causing oyster-associated primary septicemia in Florida during the mid-1990s was largely distinct from the populations in other Gulf Coast states that were comprised primarily of rrs type B (16 of 18 isolates). A greater number of strains that include more-recent clinical isolates should be examined to determine whether this trend still persists. A potential role for the 29-kb plasmid also merits further investigation. There were several instances in which three (ST 37) or two (ST 22) clinical isolates with the same ST and similar PFGE and DGREA types were observed, but the geographical association between these strains was less clear.

Concluding remarks.

Although it would be desirable to use typing methods that provide separate and perfect clustering for clinical versus environmental sources of V. vulnificus, this is rarely achieved. Among the methods used in this study, only MLST delineated clinical strains as well as PCR for rrs type and the C and E types. However, given a need to make discriminations more refined than those from using PCR-based methods, we believe that DGREA is a good candidate in this regard. For example, the DGREA cluster II consisted exclusively of rrs type B isolates and all these strains were from clinical sources. With rep-PCR and PFGE, the relationship between clinical origin and rrs B type was less clear, as the strains were clustered into multiple groups or were more mixed with environmental isolates. In addition, like rep-PCR, MLST, and PFGE, DGREA was able to identify and cluster four isolates with rrs type AB that were nearly identical, distinct from other environmental isolates with the AB genotypes, and temporally associated with human disease occurring in Florida in 1995 and 1996. Taken together, the data suggest that DGREA effectively tracks recent genetic trends and complements the genome evolutionary information identified by MLST to provide a powerful means of determining the population structure of V. vulnificus.

Similar to previous reports for rep-PCR (4), we found DGREA to be a low-cost and rapid molecular typing method for typing V. vulnificus. For instance, 48 strains could be analyzed in a single day, compared to results with PFGE, which required about 1 week to screen 20 strains; MLST took almost a month to test as many strains. Additionally, the equipment and supplies for DGREA are much less expensive than those required for PFGE and MLST. Method failure was much less frequent with DGREA than with PFGE. Further, DGREA clustering data corresponded more closely with strain source and virulence markers and were more congruent with MLST data than were PFGE data. Based on these findings, it appears that DGREA is a very promising tool for making epidemiological and ecological associations in studies of V. vulnificus.