Identification of Genetic Bases of Vibrio fluvialis Species-SpecificBiochemical Pathways and Potential Virulence Factors byComparative Genomic Analysis

Xin Lu,a,b Weili Liang,a,b Yunduan Wang,a,c Jialiang Xu,a,d Jun Zhu,c Biao Kana,b

State Key Laboratory for Infectious Disease Prevention and Control, Department of Diarrheal Diseases, National Institute for Communicable Disease Control andPrevention, Chinese Center for Disease Control and Prevention, Beijing, Chinaa; Collaborative Innovation Center for Diagnosis and Treatment of Infectious Diseases,Hangzhou, Zheijiang, Chinab; Department of Microbiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USAc; School of Foodand Chemical Engineering, Beijing Technology and Business University, Beijing, Chinad

Vibrio fluvialis is an important food-borne pathogen that causes diarrheal illness and sometimes extraintestinal infections inhumans. In this study, we sequenced the genome of a clinical V. fluvialis strain and determined its phylogenetic relationshipswith other Vibrio species by comparative genomic analysis. We found that the closest relationship was between V. fluvialis andV. furnissii, followed by those with V. cholerae and V. mimicus. Moreover, based on genome comparisons and gene complemen-tation experiments, we revealed genetic mechanisms of the biochemical tests that differentiate V. fluvialis from closely relatedspecies. Importantly, we identified a variety of genes encoding potential virulence factors, including multiple hemolysins, tran-scriptional regulators, and environmental survival and adaptation apparatuses, and the type VI secretion system, which is indic-ative of complex regulatory pathways modulating pathogenesis in this organism. The availability of V. fluvialis genome se-quences may promote our understanding of pathogenic mechanisms for this emerging pathogen.

Vibrio fluvialis is a Gram-negative, oxidase-producing, halo-philic bacterium that is normally found in coastal water and

seafood. Clinically, V. fluvialis has been implicated as a cause ofgastroenteritis with diarrhea (1) and even wound infection withprimary septicemia in immunocompromised individuals (2). Inan epidemic situation, V. fluvialis behaved more aggressively thanVibrio cholerae O1, having a higher attack rate and a differentclinical picture in the population (3). The gastrointestinal illnesscaused by this pathogen is usually associated with the consump-tion of raw or improperly cooked seafood. Additionally, this bac-terium causes significant economic loss, like in the lobster indus-tries of the eastern coasts of the United States and Canada (4).Therefore, V. fluvialis has gained importance as an epidemic-caus-ing Vibrio, especially in coastal areas.

V. fluvialis was broadly defined as a nonagglutinating Vibriospecies (5). It was first isolated in 1975 from patients with severediarrhea (6) and was originally called “group F Vibrios” by Furnisset al. and “EF-6 Vibrios” by the Centers for Disease Control (6, 7).Both phenotypic tests and DNA relatedness indicated that theseorganisms were much closer to the genus Vibrio than to the genusAeromonas (8). In some taxonomic studies, group F was separatedinto two subgroups based on gas production during glucose fer-mentation (9, 10). The aerogenic group F strains were in a differ-ent DNA relatedness group from the anaerogenic strains, and thetwo biogroups were separated into two species within the genusVibrio (8). The name V. fluvialis was proposed for both aerogenicand anaerogenic strains of group F and the synonymous groupEF-6. However, only anaerogenic strains have been isolated fromhuman samples, even though both aerogenic and anaerogenicstrains of V. fluvialis have been found in the environment (11).Subsequently, the aerogenic strains of group F were confirmed tobe a separate species from V. fluvialis and were named a new spe-cies, Vibrio furnissii (12). V. furnissii produces gas from D-glucose,while V. fluvialis does not.

The clinical symptoms of gastroenteritis caused by V. fluvialisare quite similar to those caused by V. cholerae, except for thefrequent occurrence of blood in stools. In spite of several virulencefactors, including reports of enterotoxin-like substances, lipase,proteases, cytotoxins and hemolysin, their definitive contribu-tions to the clinical manifestations remain to be characterized (2).We recently have identified a quorum-sensing regulatory cascadethat regulates several potential virulence genes (13); however,questions regarding the microbiological characteristics, mecha-nism of pathogenicity, and ecology of the organism remain mostlyunanswered.

Whole-genome sequencing has served as a routine tool in bac-teriological studies by revealing the basic genetic information ofmicroorganisms, which provides the foundation for further func-tional studies. To date, seven whole-genome sequences of Vibriospp. (including V. cholerae, V. parahaemolyticus, V. splendidus, V.vulnificus, V. anguillarum, V. harveyi, and V. furnissii) have beenreleased on GenBank, in addition to several draft genome se-quences of Vibrionales. V. cholerae, V. parahaemolyticus, and V.vulnificus are among the intensively studied and high-risk human-pathogenic species. The whole-genome sequence of V. furnissii

Received 1 November 2013 Accepted 15 January 2014

Published ahead of print 17 January 2014

Editor: C. A. Elkins

Address correspondence to Jun Zhu, junzhu@upenn.edu, or Biao Kan,kanbiao@icdc.cn.

X.L., W.L., and Y.W. contributed equally to this article.

Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.03588-13.

Copyright © 2014, American Society for Microbiology. All Rights Reserved.

doi:10.1128/AEM.03588-13

March 2014 Volume 80 Number 6 Applied and Environmental Microbiology p. 2029 –2037 aem.asm.org 2029

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

(strain NCTC 11218), which is considered one of the “emergingVibrio species” potentially implicated in human infection (14),was recently reported as well (15). During our preparation of thisarticle, two draft genome sequences of V. fluvialis were reported(16). In this study, we sequenced and annotated the genome of aclinical strain of V. fluvialis, which was isolated from a patient withdiarrhea. The genome sequences outlined herein provide a globalview of the genome of V. fluvialis and a clear map of the phyloge-netic relationship with other Vibrio spp. The molecular mecha-nism that underlies the biochemical differences that discriminateV. fluvialis, V. furnissii, and V. cholerae is also demonstrated.

MATERIALS AND METHODSStrains and genome sequencing. V. fluvialis strain 85003, which was iso-lated from an adult diarrheal patient in 1985 in Xinjiang Province, China,was used to construct shotgun libraries for genomic sequencing. Thegenomic DNA of strain 85003 was extracted from its overnight Luria-Bertani broth culture using the TIANamp bacterial DNA kit (TiangenBiotech, Beijing, China). In addition, V. cholerae O1 El Tor strain N16961was used in the biochemical tests and the genetic complement experi-ments.

An initial V. fluvialis shotgun genome sequence was obtained from amixture of Solexa sequences from the pair-end Solexa sequencing of a500-bp insert-size library to 160� coverage. The low-quality regions ofraw reads were trimmed using the fastq quality trimmer of FASTX-Tool-kit (version 0.0.13.2) (http://hannonlab.cshl.edu/fastx_toolkit/index.html), with 20 as the quality threshold. After trimming, reads shorterthan 50 bp were removed as low-quality reads. High-quality reads wereassembled to contigs with Velvet (version 1.2.07) (17). Contigs were thenscaffolded by SSPACE (18).

Annotation. Protein coding genes were predicted by Glimmer (ver-sion 3.02) (19) with the default parameters. Repeats were identified usingRepeatMasker (version open-3.2.7) (http://repeatmasker.org). tRNAswere predicted by tRNAScan (version 1.23) (20). rRNAs were predictedby RNammer (version 1.2) (21).

The genes were annotated using the Swiss-Prot, Clusters of Ortholo-gous Groups of proteins (COG), and Kyoto Encyclopedia of Genes andGenomes (KEGG) databases. BLAST (22) was used to compare sequencesto those in the Swiss-Prot database (downloaded from the EuropeanBioinformatics Institute on 8 August 2012) with E values of �1e�10 (23).They were assigned functional annotations via a sequence similarity com-parison against the COG database with BLAST with E values of �1e�10.A Perl script was written to assign the functional classification to genes(24). Genes were first compared with the KEGG database (release 58)using BLASTx at E values of �1e�10. A Perl script was used to retrieveKO information from the BLAST result, and pathway associations be-tween unigene and KEGG were then established (25).

Phylogenetic tree construction. Ten Vibrio genome sequences re-trieved from GenBank were used to determine the genome similarities of V.fluvialis strain 85003 to these species. The strains included V. anguillarum775 (GenBank accession no. NC_015633.1 and NC_015637.1), V. chol-erae IEC224 (CP003330.1 and CP003331.1), V. furnissii NCTC 11218(NC_016602.1 and NC_016628.1), V. harveyi ATCC BAA 1116(NC_009784.1 and NC_009783.1), V. parahaemolyticus RIMD 2210633(NC_004603.1 and NC_004605.1), V. splendidus LGP32 (NC_011753.2and NC_011744.2), V. vulnificus YJ016 (NC_005139.1 andNC_005140.1), V. mimicus MB451 (ADAF00000000), V. ordalii ATCC33509 (AEZC00000000), and V. campbellii DS40M4 (AGIE00000000).The genome sequences of two Photobacterium spp. (including P. damselae[ADBS00000000] and P. angustum [AAOJ00000000]) were used as theoutgroup. First, the single-copy ortholog clusters within 11 Vibrio spp.were identified; OrthoMCL (version 2.0.3) (26) was then used to clusterthe newly predicted genes with genes from other references. Default pa-rameters were used for OrthoMCL, but clusters were formed based on a

shared sequence similarity of 70% instead of the OrthoMCL default pa-rameter value of 50%. We performed a phylogenetic analysis following thedescribed methods (45). Briefly, protein sequences of approximately 10%of the identified single-copy ortholog clusters (100 genes from 961 single-copy ortholog clusters in this study) were randomly selected as a sampleand used to construct a maximum likelihood tree. Three independentsamplings were performed from these single-copy ortholog clusters. Eachsample was used as a basis for construction of a maximum likelihood treeand compared each other subsequently. ClustalW was used to indepen-dently align the sequence members of each ortholog cluster to minimizegene rearrangement within the multiple-sequence alignment (27). A max-imum likelihood method was used to generate a phylogenetic species treewith 100 replicates for bootstrapping (28) using phyML 3.0 and MEGA5(29). All three produced trees were of highly similar topologies.

Comparative genome analysis. The assembled genome was com-pared to that of V. furnissii NCTC 11218 (15) using CGView (CircularGenome Viewer) (30) at both the nucleotide and protein levels.

Biochemical pathway complement tests. In the biochemical tests,D-glucose fermentation and three amino acids utilizations (arginine,lysine, and ornithine) were used to discriminate V. fluvialis from V. furnis-sii and V. fluvialis from V. cholerae, respectively. The genes associated withthese carbohydrate fermentation pathways were screened and listed ac-cording to the KEGG database (http://www.genome.jp/kegg/). All of thesegenes were correspondingly aligned to those of V. furnissii and V. choleraewith BLAST with an overlap of less than 80%. The absent genes were thenidentified in the species with negative phenotypes.

To investigate whether the lack of these homologs in the pathways isresponsible for the negative fermentations, the genes were amplified fromthe genomes of the species with positive phenotypes, cloned into the plas-mid vectors, confirmed by sequencing, and transformed into the negativestrains by electroporation (31). The 500-bp upstream fragments of thecloned genes (operons) were included. The primers, restriction sites forcloning, and recipient strains are listed in Table S2 in the supplementalmaterial.

In the biochemical tests, the wild-type strains and the complementedstrains were tested using a microbial biochemical identification tube(073260; HuanKai, China) for gas from glucose, API20E (bioMérieux,France) for the lysine decarboxylase test and ornithine decarboxylase test,and microbial biochemical identification tubes (073130 and 0713140;HuanKai, China) for the arginine dihydrolase test. All procedures wereperformed following the manufacturer’s instructions.

Nucleotide sequence accession number. The genome sequences of V.fluvialis reported in the present study have been deposited in the SequenceRead Archive under accession no. SRX397301.

RESULTS AND DISCUSSIONGeneral genome features of V. fluvialis strain 85003. A total of5,135,920 reads (770,388,000 nucleotide bases) were assembledinto 132 large contigs. The contig N50 was 193,015 bp in length,and the longest assembled contig was 532,968 bp. The draft ge-nome of V. fluvialis 85003 consists of 4,689,093 bp with 50.0% GCcontent and contains 4,596 predicted coding sequences (CDSs)(Table 1), just the same as the genome characteristics of V. fluvialisstrains PG41 and 121563 (16). The putative functions could beassigned to 60.7% of the CDSs, while 14.4% possessed similarity togenes of unknown function, and no function could be proposedfor 25.0% of the CDSs. A total of 53 tRNA operons are predicted inthe draft genome. The assembled contigs from the newly se-quenced V. fluvialis 85003 strain are shown aligned to the V.furnissii NCTC 11218 genome in Fig. 1. The general features of thegenome are summarized in Table 1 and Fig. 1.

Genome comparison revealed the phylogenetic position ofV. fluvialis in Vibrio. We compared the genome sequences of 11Vibrio spp. and two Photobacterium spp. from GenBank with the

Lu et al.

2030 aem.asm.org Applied and Environmental Microbiology

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

identified 961 common single-copy ortholog clusters. The pat-terns of the phylogenetic tree showed that all Vibrio spp. clusteredtogether and formed a phylogroup separate from Photobacteriumspp. and indicated the taxonomic position of V. fluvialis amongthe Vibrionaceae (Fig. 2), which were similar to those observed inthe16S rRNA gene gene-based tree. The taxonomic position wasalso similar to those recorded in Bergey’s Manual of SystematicBacteriology, 2nd ed. (32). As expected, V. fluvialis 58003 and V.furnissii NCTC 11218 clustered together, which suggested thatboth species may have evolved recently from a common ancestor.Both species formed a sister taxon with V. cholerae IEC224 andVibrio mimicus MB451.

Within the phylogenetic tree, V. cholerae, V. mimicus, V. flu-vialis, and V. furnissii were distributed into the Cholerae clade,which was recognized as the nearby branch of the Anguillarumclade (33). V. parahaemolyticus, V. campbellii, and V. harveyi wererecognized as members of the Harveyi clade; they formed a dis-tinct sublineage in our genomic analysis (Fig. 2). V. ordalii waspreviously recognized as biotype 2 of V. anguillarum; the genome-based phylogenic tree showed that they were in the same lineage(Fig. 2). The phylogenetic analysis showed the relationshipsamong Vibrio species and confirmed the taxonomy of these spe-cies, which will be useful in an evolution context.

We further calculated the core genome of V. fluvialis and otherVibrio spp. The pairwise comparisons of core genes between V.fluvialis 85003 and the other 10 Vibrionaceae species used in theabove phylogenetic analysis showed that V. fluvialis 85003 sharedthe most core genes (3,754 genes) with V. furnissii NCTC 11218and less than 2,810 genes with the rest of the Vibrionaceae species.V. fluvialis 85003 shared the fewest core genes (2,388 genes) withV. ordalii ATCC 33509 (Fig. 3A) because V. ordalii ATCC 33509has the smallest genome of these Vibrionaceae species. Figure 3B

TABLE 1 Global genomic view of V. fluvialis 85003 and V. furnissiiNCTC 11218

Parameter

Result for:

V. fluvialis V. furnissii

Length of sequence, bp 4,689,093 4,916,408GC content, % 50.03 50.67

Repeat content length, bp (% coverage) 5,608 (0.12) 19,104 (0.39)Simple repeat length, bp (% coverage) 557 (0.01) 1,067 (0.02)Low-complexity length, bp (% coverage) 1,438 (0.03) 1,881 (0.04)Repeat element length, bp (% coverage) 504 (0.01) 1,131 (0.02)Small RNA no. (bp length [% coverage]) 27 (3,292 [0.07]) 73 (15,537 [0.32])

ORF length in genome, bpa 4,016,205 4,220,358Features

ORFs, no. 4,596 4,455Genome coverage, % 85.65 85.84Genome density, genes/kb 0.98 0.91Avg gene length, bp 874 947Maximum gene length, bp 4,596 9,450

FunctionsORFs with assigned functions, no.b 2,788 2,928Function prediction only or unknownb 660 633ORFs without COG match, no. 1,148 894

RNArRNAc

5S rRNA, no. (length, bp) 2 (115) 8 (115)16S rRNA, no. (length, bp) 1 (1,541) 7 (1,541)23S rRNA, no. (length, bp) 1 (2,906) 7 (2,889)

tRNA, no. (total bp length)d 53 (4,205) 95 (7,474)

a Coding genes were predicted by Glimmer version 3.02.b BLAST against the COG database with an E value of �1e�10.c rRNA genes were predicted by RNammer version 1.2.d tRNA genes were predicted by tRNAscan version 1.23.

FIG 1 Genome comparison of V. fluvialis 85003 and V. furnissii NCTC 11218. From the outside in, the first circle (pink) represents the nucleotide level (BLASTn)alignment of V. fluvialis contigs against the reference genome. The second circle (green) represents the amino acid level (tBLASTx) alignment of V. fluvialiscontigs against the reference genome. Circle 3 shows the GC content, and circle 4 shows the GC skew. The figure was generated using CGView.

Genome of Vibrio fluvialis

March 2014 Volume 80 Number 6 aem.asm.org 2031

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

showed the profile curves for the core gene number. Vibrionaceaespecies strains were added successively into the analysis by thedistance between them and V. fluvialis in the phylogenetic tree.The distributions of the clusters of orthologous groups (COG) inthe core genes were examined. Chi-square tests showed that thenumbers of core genes in the same COG groups did not signifi-cantly differ between V. fluvialis and V. furnissii, which furtherdemonstrated their close relationship. When V. fluvialis was pair-wise compared with the other nine Vibrionaceae species, the sig-nificant differences in gene number were found mainly in threeCOG groups: (i) transcription; (ii) translation, ribosomal struc-ture, and biogenesis; and (iii) signal transduction mechanisms.

Genome comparison of V. fluvialis and V. furnissii. In thisstudy, we found that the mean BLASTp identity of the homolo-gous genes between V. fluvialis 85003 and V. furnissii NCTC 11218was 86.8%. The complete genome sequence of the V. furnissii sp.nov. strain NCTC 11218 was reported (15). It consists of twocircular chromosomes (3.2 Mb and 1.6 Mb). The GC content ofchromosome I is 50.7%, and that of chromosome II is 50.5%.Chromosome I contains 3,013 open reading frames (ORFs) and95 tRNA sequences, and chromosome II has 1,448 ORFs and 5tRNA sequences. Using V. furnissii as the reference, the V. fluvialis85003 genome contains 141 more genes and shares 3,948 homol-ogous genes (see Fig. S2 in the supplemental material). The GCcontent of the V. fluvialis 85003 genome is 50.0%. Approximately25.0% (1,148/4,596) of the chromosomal ORFs lacked a COGmatch in nature, and these genes account for the majority of genesthat are unique to the V. fluvialis 85003 genome. For the remain-ing functionally known genes, we examined the distributions indifferent COG groups (see Table S1 and Fig. S1 in the supplemen-tal material). Amino acid transport and metabolism and energyproduction and conversion accounted for 8.9% (308/3,448) and6.4% (219/3,448), respectively, of the genes in this major COGgroup. Compared with V. furnissii, carbohydrate transport and

FIG 3 Comparison of the numbers of core genes shared between V. fluvialis85003 and other Vibrio strains used in phylogenetic analysis. (A) Pairwisecomparisons of core genes between V. fluvialis 85003 and 10 other Vibrion-aceae species were used in the above phylogenetic analysis; (B) profile curvesfor the core gene number. Vibrionaceae species were successively added intothe analysis in the following order (from 2 to 11): V. furnissii, V. cholerae, V.mimicus, V. anguillarum, V. ordalii, V. splendidus, V. vulnificus, V. parahaemo-lyticus, V. campbellii, and V. harveyi. The black diamonds denote the core genesize for each species.

FIG 2 Phylogenetic analysis of the common single-copy orthologs supports the taxonomic location of V. fluvialis among Vibrionaceae. Ten Vibrio genomesequences (including V. anguillarum 775 [GenBank accession no. NC_015633.1 and NC_015637.1], V. cholerae IEC224 [CP003330.1 and CP003331.1], V.furnissii NCTC 11218 [NC_016602.1 and NC_016628.1], V. harveyi ATCC BAA 1116 [NC_009784.1 and NC_009783.1], V. parahaemolyticus RIMD 2210633[NC_004603.1 and NC_004605.1]), V. splendidus LGP32 [NC_011753.2 and NC_011744.2], V. vulnificus YJ016 [NC_005139.1 and NC_005140.1], V. mimicusMB451 [ADAF00000000], V. ordalii ATCC 33509 [AEZC00000000], and V. campbellii DS40M4 [AGIE00000000]) and two genome sequences of Photobacteriumspp. (including P. damselae [ADBS00000000] and P. angustum [AAOJ00000000]) were retrieved from GenBank.

Lu et al.

2032 aem.asm.org Applied and Environmental Microbiology

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

metabolism and cell motility accounted for a small fraction of thegenes in V. fluvialis 85003. COG function classification identifiedindividual genes involved in chromatin structure and dynamics,as well as RNA processing.

Genetic bases of metabolic analysis to differentiate V. fluvia-lis from other closely related Vibrio spp. (i) V. fluvialis and V.furnissii. Because they are two closely related Vibrio species, thebiochemical phenotype of V. fluvialis is very similar to that of V.furnissii, except for the gas production during the D-glucose fer-

mentation test. V. furnissii produces gas from the fermentation ofD-glucose, while V. fluvialis does not (12). The gas productionfrom glucose is associated with activity of formate dehydrogenase(34). The pathway comparison between V. fluvialis 85003 and V.furnissii NCTC 11218 revealed that the vfu_A02711 gene of V.furnissii NCTC 11218, annotated as producing an NADP-depen-dent formate dehydrogenase alpha subunit, was absent in V. flu-vialis strain 85003 (Table 2 and Fig. 4A). Formate dehydrogenase(NADP�) (EC 1.2.1.43) catalyzes the chemical reaction with for-

TABLE 2 Identification of genes responsible for amino acid metabolism whose absence in the recipient strain results in a negative phenotype

Biochemicaltests Gene(s)

Strain species Test result for:

Original RecipientV. fluvialis85003 Recipient

D-Glucose vfu_A02711 V. furnissii V. fluvialis � �, gas

Lysine cadAB operon V. cholerae V. fluvialis � �VC0278 V. cholerae V. fluvialis � � (with cadAB)

Ornithine speF-potE operon V. cholerae V. fluvialis � �

Arginine arcABCD operon V. fluvialis V. cholerae � � (slower)argR V. fluvialis V. cholerae � � (faster, with

arcABCD)

FIG 4 Genome structure of the absent genes identified in the species with negative phenotypes. Panels A to D show the absent genes identified that are associatedwith the D-glucose fermentation pathway (A), the lysine utilization pathway (B), the ornithine utilization pathway (C), and the arginine utilization pathway (D).

Genome of Vibrio fluvialis

March 2014 Volume 80 Number 6 aem.asm.org 2033

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

mate and NADP� as the substrates and CO2 and NADPH as prod-ucts. We introduced vfu_A02711 of V. furnissii via a recombinantplasmid into V. fluvialis strain 85003 and reproduced the gas pro-duction of V. furnissii from glucose (Fig. 5A). The results experi-mentally demonstrated that the deletion of the NADP-dependentformate dehydrogenase alpha subunit gene accounts for the an-aerogenic phenotype of glucose fermentation in V. fluvialis.

(ii) V. fluvialis and V. cholerae. V. cholerae is the second clos-est species to V. fluvialis. Species-specific minimal biochemicaltests are used to identify V. fluvialis; without these tests, V. fluvialismay be confused with V. cholerae (5). Utilizations of arginine,

lysine, and ornithine are the major biochemical tests used to dif-ferentiate the species V. fluvialis and V. cholerae. V. cholerae is��� for utilization of these three amino acids, respectively, butV. fluvialis is ���. We searched for enzymes and genes involvedin their metabolism in the genomes of both species. Lysine is usedby the enzyme lysine decarboxylase, whose genes are carried by theoperon containing VC0281 (cadA) and VC0280 (cadB) in V. chol-erae N16961. The transcription of this operon is controlled by theupstream membrane-bound transcription activator CadC, whichis encoded by VC0278 (35). These three genes are absent in the V.fluvialis 85003 genome (Table 2 and Fig. 4B). Introduction ofcadBA of V. cholerae into V. fluvialis 85003 with a plasmid stillreproduced a negative result for the lysine decarboxylase test,whereas introduction of both cadBA and cadC of V. cholerae intoV. fluvialis 85003 yielded a positive result, as for V. cholerae (Fig.5B). Therefore, a lack of the three genes results in a defect in thecatabolism of lysine by V. fluvialis.

Ornithine decarboxylase catalyzes ornithine to form pu-trescine. The speF-potE operon encoded ornithine decarboxylaseand the polyamine transport PotE protein, which are induced atan acidic pH (36). VCA1063 and VCA1062 encode SpeF and PotEin V. cholerae but were not found in the V. fluvialis 85003 genome(Table 2 and Fig. 4C). Introduction of the intact speF-potE operonof V. cholerae into V. fluvialis strain 85003 reproduced the positiveresult (Fig. 5C).

The arginine deiminase pathway enables V. fluvialis to growanaerobically on arginine. The arc genes appear clustered in anoperon-like structure, including arcA (arginine deiminase), arcB(ornithine carbamoyltransferase), arcC (carbamate kinase), andarcD (putative arginine-ornithine antiporter). arcC and arcD wereabsent in the V. cholerae N16961 genome (Table 2 and Fig. 4D).Introduction of the arcCD of V. fluvialis into the V. choleraeN16961 strain did not reproduce the positive result for the argi-nine dihydrolase test, and introduction of the whole arcABCDoperon into the N16961 strain reproduced a positive but time-delayed result. ArgR is an essential local transcriptional regulatorof the arc operon (37). The argR in strain N16961 only shares 41%of its identity with V. fluvialis. We then introduced both the argRgene and arcABCD operon of V. fluvialis into V. cholerae strainN16961; the transformed strain showed a positive result over thetime frame examined (Fig. 5D).

Taken together, the above data indicate that the differences inthe hexose and amino acid metabolic tests used to differentiate V.fluvialis from V. furnissii and V. cholerae are due to an absence ofthe genes that are involved in the metabolic pathways. The genedeletions may endow these species with different phenotypes.From the genome comparisons and the complement experiments,the genetic mechanisms of the metabolic differences of these spe-cies are revealed. These data also present the genetic evidence tostudy the differentiation and evolution of these closely relatedspecies within Vibrio. Based on the gene differences, nucleic acid-based tests could be developed to replace the biochemical testsused to identify these similar species.

Virulence-related genes identified in the genome of V. flu-vialis 85003. (i) Hemolysin. V. fluvialis is an enteric pathogen;however, its pathogenesis has not been assigned to one or sev-eral virulence factors, such as the clear role of cholera toxin inthe pathogenesis of V. cholerae. V. fluvialis reportedly producesan extracellular hemolysin (38). In this study, several hemolysin-related genes, which showed high similarity to the genes identified

FIG 5 Complementation of absent genes endows strains with amino acidmetabolic capacity, as shown through biochemical tests. (A) A complementa-tion experiment with the vfu_A02711 gene in strain 85003 showed that thecomplement strain displayed similar positive results for gas production fromglucose to those observed in V. furnissii. A-1, V. fluvialis 85003; A-2, V. furnissiiICDC001; A-3, V. fluvialis 85003 carrying the vfu_A02711 gene in pBR322;A-4, V. fluvialis 85003 carrying plasmid pBR322 as a control. (B) A comple-mentation experiment introducing both cadBA and cadC of V. cholerae into V.fluvialis 85003 enabled it to process lysine (pink) like V. cholerae. B-1, V.fluvialis 85003; B-2, V. fluvialis 85003 carrying the cadAB genes in pBR322 andthe cadC gene in pACYC184. (C) A complementation experiment with thewhole speF-potE operon of V. cholerae in V. fluvialis strain 85003 showed thatthe complement strain was positive for ornithine decarboxylation (red). C-1,V. fluvialis 85003; C-2, V. fluvialis 85003 carrying the speF-potE operon inpBR322. (D) In a complementation experiment in which both argR and thearcABCD operon of V. fluvialis were transformed into V. cholerae strainN16961, the transformant strain was positive for arginine utilization (blue).D-1, V. cholerae strain N16961; D-2, V. cholerae strain N16961 carrying thearcABCD operon in pBR322; D-3, V. cholerae strain N16961 carrying the argRgene in pACYC184; D-4, V. cholerae strain N16961 carrying the arcABCDoperon in pBR322 and the argR gene in pACYC184.

Lu et al.

2034 aem.asm.org Applied and Environmental Microbiology

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

in other Vibrio species, were found in the genome of V. fluvialis85003. The gene VFL_002885 of V. fluvialis 85003 shares 98% ofits identity with that of V. furnissii and 78% with the V. choleraeEl Tor hemolysin A (hlyA) gene. Genes VFL_003095 andVFL_003869 are also predicted to be hemolysin genes. VFL_000876 is 91% identical to the thermostable hemolysin gene of V.furnissii and similar to the thermostable direct hemolysin gene(tdh) of V. parahaemolyticus. Four other hemolysin genes werealso predicted in the V. fluvialis 85003 genome, which agrees withthe multiple hemolysin genes present in the genomes of patho-genic and nonpathogenic Vibrio spp. (39).

(ii) T6SS. Secretion systems are implicated in bacterial survivaland pathogenesis (40). We searched the genome of V. fluvialisstrain 85003 and found an intact cluster of the type VI secretionsystem (T6SS) homologous genes, but no homologous genes ofother secretion systems or single homologous genes were identi-fied. The homologs (Table 3) of the vas (virulence-associated se-cretion) genes (vasA, vasF, vasK, and vasH) were found, which areproposed to encode a prototypic T6SS in the non-O1/non-O139V. cholerae strains (41). The genome also contains homologs(VFL_001426 to VFL_001440 and VFL_002824 to VFL_002840)

of genes (e.g., imp, vgrG, vas, clp, and hcp) that constitute a T6SSidentified in V. cholerae. The Impk, IcmF, ClpV, Hcp, and VgrGorthologs, which are described as T6SS core components, were allfound in the V. fluvialis 85003 strain (Table 3). VasH is the globalregulator of the V. cholerae T6SS, which is essential to initiate thetranscription of T6SS genes and has a universal role in endemicand pandemic V. cholerae strains (42). In the V. fluvialis 85003genome, VFL_002830 was also annotated for its homolog withvasH. T6SS is implicated in interbacterial relationships, biofilmformation, cytotoxicity, and survival in the phagocytic cells ofbacteria. These T6SS homologs may serve similar roles in the vir-ulence of V. fluvialis.

Other virulence-related genes. Based on the virulence factordatabase (VFDB; http://www.mgc.ac.cn/VFs/), the genes of othervirulence factors were annotated in the V. fluvialis 85003 genome,including the genes coding for the ferric uptake regulator (Fur),heat shock protein 90 (HtpB), flagellar biosynthetic proteins(FliA, FliC, FliD, FliE, FliF, FliG, FliH, FliI, FliM, FliN, FliP, FliQ,FliR, and FliS), toxin-coregulated pilus biosynthesis protein I(TcpI), and accessory colonization factor (AcfB). In addition tofur, other genes essential for iron transport were also predicted,including the fbp family for ferric iron, the fhu family for fer-richrome transport, and the feo family for ferrous iron. All of thesegenes are homologues to those of V. cholerae. The mannose-sen-sitive hemagglutinin (MSHA) pilus, which belongs to the family oftype IV pili, is directly associated with the adherence of V. choleraeto environmental surfaces (43) and biofilm formation (44). In theV. fluvialis 85003 genome, VFL_003720 to VFL_003740 werefound similar to the MSHA synthesis genes of V. cholerae, whichsuggests the presence of an MSHA-like pilus with a similar role inV. fluvialis.

Conclusion. In this study, we sequenced the V. fluvialis ge-nome and determined its genetic relationships within Vibrio spp.The data provide details of the genomic sequences to show theclose relationship between V. fluvialis and V. furnissii and theirphylogenetic position in Vibrio, which is similar to the relation-ship revealed by traditional microbiological strategies. The com-bination of whole-genome sequencing and comparison also pro-vides a rapid and comprehensive approach to identify themechanisms of the biochemical phenotyping differences, whichare used to characterize these closely related species. The approachrevealed that gene deletion contributed to their phenotype diver-sity. From genome sequence searching, multiple virulence-relatedgenes and gene clusters were found, demonstrating the compli-cated pathogenesis of this enteric pathogen. Therefore, the ge-nome data promote the understanding of the genetic basis ofphysiology, virulence, evolution, and the environmental survivalmechanisms for V. fluvialis. Furthermore, for environmental anddisease surveillance purposes, the gene markers and even meta-bolic pathways could be searched for through genome compari-sons and used to develop specific detection methods for this ma-rine microbe.

ACKNOWLEDGMENTS

We thank Karl-Gustav Rueggeberg for critically reviewing the manu-script.

This work was supported by the Priority Project on Infectious DiseaseControl and Prevention (2012ZX10004215 and 2008ZX10004-008) of theMinistry of Health, China, and by the National Basic Research Priorities

TABLE 3 Homologs of T6SS-associated genes identified in the V.fluvialis 85003 isolate and BLASTp identity results

V. fluvialisgene

Homolog in other Vibrio spp.(% identity by BLASTp)

Product by BLASTxKEGGV. furnissii V. cholerae

VFL_001426 vfu_B00795 (82) ImpAVFL_001427 vfu_B00794 (100) HcpVFL_001428 vfu_B00793 (99) VCA0107 (35) ImpBVFL_001429 vfu_B00792 (97) VCA0108 (34) ImpCVFL_001430 vfu_B00791 (98) VCA0108 (25) ImpDVFL_001431 vfu_B00790 (100) ImpFVFL_001432 vfu_B00789 (98) VCA0110 (24) ImpGVFL_001433 vfu_B00788 (99) Hypothetical proteinVFL_001434 vfu_B00786 (99) VCA0116 (42) VasGVFL_001435 vfu_B00785 (94) Hypothetical proteinVFL_001436 vfu_B00784 (92) Hypothetical proteinVFL_001437 vfu_B00783 (93) VasDVFL_001438 vfu_B00782 (99) VCA0114 (27) ImpJVFL_001439 vfu_B00781 (81) ImpKVFL_001440 vfu_B00780 (90) VCA0120 (20) ImpLVFL_002824 vfu_B01011 (89) VCA0018 (72) VgrGVFL_002825 vfu_A01957 (98) VCA0018 (75) VgrGVFL_002826 vfu_B01190 (93) VCA0121 (56) VasLVFL_002827 vfu_B01189 (98) VCA0120 (84) ImpLVFL_002828 vfu_B01188 (90) VCA0119 (73) VasJVFL_002829 vfu_B01187 (86) VCA0118 (61) VasIVFL_002830 vfu_B01186 (95) VCA0117 (75) �54-dependent

transcriptionalregulator

VFL_002831 vfu_B01185 (93) VCA0116 (86) VasGVFL_002832 vfu_B01184 (94) VCA0115 (85) ImpKVFL_002833 vfu_B01183 (98) VCA0114 (90) ImpJVFL_002834 vfu_B01182 (91) VCA0113 (84) VasDVFL_002835 vfu_B01181 (82) VCA0112 (64) ImpIVFL_002836 vfu_B01180 (97) VCA0111 (81) ImpHVFL_002837 vfu_B01179 (96) VCA0110 (84) ImpGVFL_002838 vfu_B01178 (100) VCA0109 (89) Hypothetical proteinVFL_002839 vfu_B01177 (98) VCA0108 (92) ImpCVFL_002840 vfu_B01176 (89) VCA0107 (87) ImpB

Genome of Vibrio fluvialis

March 2014 Volume 80 Number 6 aem.asm.org 2035

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

Program (2009CB522604) of the Ministry of Scientific Technology,China.

REFERENCES1. Tacket C, Hickman F, Pierce G, Mendoza L. 1982. Diarrhea associated

with Vibrio fluvialis in the United States. J. Clin. Microbiol. 16:991–992.2. Igbinosa E, Okoh A. 2010. Vibrio fluvialis: an unusual enteric pathogen of

increasing public health concern. Int. J. Environ. Res. Public Health7:3628 –3643. http://dx.doi.org/10.3390/ijerph7103628.

3. Bhattacharjee S, Bhattacharjee S, Bal B, Pal R, Niyogi S, Sarkar K. 2010.Is Vibrio fluvialis emerging as a pathogen with epidemic potential incoastal region of eastern India following cyclone Aila? J. Health Popul.Nutr. 28:311–317. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2965321/.

4. Tall B, Fall S, Pereira M, Ramos-Valle M, Curtis S, Kothary M, Chu D,Monday S, Kornegay L, Donkar T, Prince D, Thunberg R, Shangraw K,Hanes D, Khambaty F, Lampel K, Bier J, Bayer R. 2003. Characteriza-tion of Vibrio fluvialis-like strains implicated in limp lobster disease. Appl.Environ. Microbiol. 69:7435–7446. http://dx.doi.org/10.1128/AEM.69.12.7435-7446.2003.

5. Chowdhury G, Pazhani G, Dutta D, Guin S, Dutta S, Ghosh S, IzumiyaH, Asakura M, Yamasaki S, Takeda Y, Arakawa E, Watanabe H,Mukhopadhyay A, Bhattacharya M, Rajendran K, Nair G, RamamurthyT. 2012. Vibrio fluvialis in patients with diarrhea, Kolkata, India. Emerg.Infect. Dis. 18:1868 –1871. http://dx.doi.org/10.3201/eid1811.120520.

6. Huq M, Alam A, Brenner D, Morris G. 1980. Isolation of Vibrio-likegroup, EF-6, from patients with diarrhea. J. Clin. Microbiol. 11:621– 624.

7. Furniss A, Lee J, Donovan T. 1977. Group F, a new Vibrio? Lancetii:565–566.

8. Seidler R, Allen D, Colwell R, Joseph S, Daily O. 1980. Biochemicalcharacteristics and virulence of environmental group F bacteria isolated inthe United States. Appl. Environ. Microbiol. 40:715–720.

9. Lee JV, Donovan TJ, Furniss AL. 1978. Characterization, taxonomy, andamended description of Vibrio metschnikovii. Int. J. Syst. Bacteriol. 28:99 –111. http://dx.doi.org/10.1099/00207713-28-1-99.

10. Lee JV, Shread P, Furniss AL. 1978. The taxonomy of group F organisms:relationships to Vibrio and Aeromonas. J. Appl. Microbiol. 45:ix.

11. Lee J, Shread P, Furniss A, Bryant T. 1981. Taxonomy and descriptionof Vibrio fluvialis sp. nov. (synonym group F vibrios, group EF6). J. Appl.Bacteriol. 50:73–94.

12. Brenner D, Hickman-Brenner F, Lee J, Steigerwalt A, Fanning G,Hollis D, Farmer GR, Weaver R, Joseph S, Seidler R. 1983. Vibriofurnissii (formerly aerogenic biogroup of Vibrio fluvialis), a new speciesisolated from human feces and the environment. J. Clin. Microbiol. 18:816 – 824.

13. Wang Y, Wang H, Liang W, Hay A, Zhong Z, Kan B, Zhu J. 2013.Quorum sensing regulatory cascades control Vibrio fluvialis pathogenesis.J. Bacteriol. 195:3583–3589. http://dx.doi.org/10.1128/JB.00508-13.

14. Igbinosa E, Okoh A. 2008. Emerging Vibrio species: an unending threat topublic health in developing countries. Res. Microbiol. 159:495–506. http://dx.doi.org/10.1016/j.resmic.2008.07.001.

15. Lux T, Lee R, Love J. 2011. Complete genome sequence of a free-livingVibrio furnissii sp. nov. strain (NCTC 11218) J. Bacteriol. 193:1487–1488.http://dx.doi.org/10.1128/JB.01512-10.

16. Khatri I, Mahajan S, Dureja C, Subramanian S, Raychaudhuri S. 2013.Evidence of a new metabolic capacity in an emerging diarrheal pathogen:lessons from the draft genomes of Vibrio fluvialis strains PG41 and I21563.Gut Pathog. 5:20. http://dx.doi.org/10.1186/1757-4749-5-20.

17. Zerbino D, Birney E. 2008. Velvet: algorithms for de novo short readassembly using de Bruijn graphs. Genome Res. 18:821– 829. http://dx.doi.org/10.1101/gr.074492.107.

18. Boetzer M, Henkel C, Jansen H, Butler D, Pirovano W. 2011. Scaffold-ing pre-assembled contigs using SSPACE. Bioinformatics 27:578 –579.http://dx.doi.org/10.1093/bioinformatics/btq683.

19. Delcher A, Harmon D, Kasif S, White O, Salzberg S. 1999. Improvedmicrobial gene identification with GLIMMER. Nucleic Acids Res. 27:4636 – 4641. http://dx.doi.org/10.1093/nar/27.23.4636.

20. Lowe T, Eddy S. 1997. tRNAscan-SE: a program for improved detectionof transfer RNA genes in genomic sequence. Nucleic Acids Res. 25:955–964.

21. Lagesen K, Hallin P, Rødland E, Staerfeldt H, Rognes T, Ussery D.2007. RNAmmer: consistent and rapid annotation of ribosomal RNA

genes. Nucleic Acids Res. 35:3100 –3108. http://dx.doi.org/10.1093/nar/gkm160.

22. Altschul S, Gish W, Miller W, Myers E, Lipman D. 1990. Basic localalignment search tool. J. Mol. Biol. 215:403– 410.

23. Brenner S, Chothia C, Hubbard T. 1998. Assessing sequence comparisonmethods with reliable structurally identified distant evolutionary relation-ships. Proc. Natl. Acad. Sci. U. S. A. 95:6073– 6078. http://dx.doi.org/10.1073/pnas.95.11.6073.

24. Tatusov R, Fedorova N, Jackson J, Jacobs A, Kiryutin B, Koonin E,Krylov D, Mazumder R, Mekhedov S, Nikolskaya A, Rao B, Smirnov S,Sverdlov A, Vasudevan S, Wolf Y, Yin J, Natale D. 2003. The COGdatabase: an updated version includes eukaryotes. BMC Bioinformatics4:41. http://dx.doi.org/10.1186/1471-2105-4-41.

25. Kanehisa M, Goto S, Hattori M, Aoki-Kinoshita K, Itoh M, KawashimaS, Katayama T, Araki M, Hirakawa M. 2006. From genomics to chemicalgenomics: new developments in KEGG. Nucleic Acids Res. 34:D354 –D357. http://dx.doi.org/10.1093/nar/gkj102.

26. Li L, Stoeckert CJ, Roos D. 2003. OrthoMCL: identification of orthologgroups for eukaryotic genomes. Genome Res. 13:2178 –2189. http://dx.doi.org/10.1101/gr.1224503.

27. Thompson J, Higgins D, Gibson T. 1994. CLUSTAL W: improving thesensitivity of progressive multiple sequence alignment through sequenceweighting, position-specific gap penalties and weight matrix choice. Nu-cleic Acids Res. 22:4673– 4680. http://dx.doi.org/10.1093/nar/22.22.4673.

28. Guindon S, Gascuel O. 2003. A simple, fast, and accurate algorithm toestimate large phylogenies by maximum likelihood. Syst. Biol. 52:696 –704. http://dx.doi.org/10.1080/10635150390235520.

29. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. 2011.MEGA5: molecular evolutionary genetics analysis using maximum likeli-hood, evolutionary distance, and maximum parsimony methods. Mol.Biol. Evol. 28:2731–2739. http://dx.doi.org/10.1093/molbev/msr121.

30. Stothard P, Wishart D. 2005. Circular genome visualization and explo-ration using CGView. Bioinformatics 21:537–539. http://dx.doi.org/10.1093/bioinformatics/bti054.

31. Hamashima H, Nakano T, Tamura S, Arai T. 1990. Genetic transfor-mation of Vibrio parahaemolyticus, Vibrio alginolyticus, and Vibrio chol-erae non O-1 with plasmid DNA by electroporation. Microbiol. Immunol.34:703–708. http://dx.doi.org/10.1111/j.1348-0421.1990.tb01047.x.

32. Garrity GM, Bell JA, Lilburn TG. 2004. Taxonomic outline of the pro-karyotes. Bergey’s manual of systematic bacteriology, 2nd ed. Springer,New York, NY.

33. Sawabe T, Kita-Tsukamoto K, Thompson F. 2007. Inferring the evolu-tionary history of vibrios by means of multilocus sequence analysis. J.Bacteriol. 189:7932–7936. http://dx.doi.org/10.1128/JB.00693-07.

34. Popov V, Lamzin V. 1994. NAD�-dependent formate dehydrogenase.Biochem. J. 301:625– 643.

35. Kanjee U, Gutsche I, Alexopoulos E, Zhao B, El Bakkouri M, ThibaultG, Liu K, Ramachandran S, Snider J, Pai E, Houry W. 2011. Linkagebetween the bacterial acid stress and stringent responses: the structure ofthe inducible lysine decarboxylase. EMBO J. 30:931–944. http://dx.doi.org/10.1038/emboj.2011.5.

36. Kashiwagi K, Watanabe R, Igarashi K. 1994. Involvement of ribonu-clease III in the enhancement of expression of the speF-potE operon en-coding inducible ornithine decarboxylase and polyamine transport pro-tein. Biochem. Biophys. Res. Commun. 200:591–597. http://dx.doi.org/10.1006/bbrc.1994.1489.

37. Fulde M, Willenborg J, de Greeff A, Benga L, Smith H, Valentin-Weigand P, Goethe R. 2011. ArgR is an essential local transcriptionalregulator of the arcABC operon in Streptococcus suis and is crucial forbiological fitness in an acidic environment. Microbiology 157:572–582.http://dx.doi.org/10.1099/mic.0.043067-0.

38. Kothary M, Lowman H, McCardell B, Tall B. 2003. Purification andcharacterization of enterotoxigenic El Tor-like hemolysin produced byVibrio fluvialis. Infect. Immun. 71:3213–3220. http://dx.doi.org/10.1128/IAI.71.6.3213-3220.2003.

39. Chun J, Grim C, Hasan N, Lee J, Choi S, Haley B, Taviani E, Jeon Y,Kim D, Lee J, Brettin T, Bruce D, Challacombe J, Detter J, Han C,Munk A, Chertkov O, Meincke L, Saunders E, Walters R, Huq A, NairG, Colwell R. 2009. Comparative genomics reveals mechanism for short-term and long-term clonal transitions in pandemic Vibrio cholerae. Proc.Natl. Acad. Sci. U. S. A. 106:15442–15447. http://dx.doi.org/10.1073/pnas.0907787106.

Lu et al.

2036 aem.asm.org Applied and Environmental Microbiology

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

40. Records AR. 2011. The type VI secretion system: a multipurpose deliverysystem with a phage-like machinery. Mol. Plant Microbe Interact. 24:751–757. http://dx.doi.org/10.1094/MPMI-11-10-0262.

41. Pukatzki S, Ma A, Sturtevant D, Krastins B, Sarracino D, Nelson W,Heidelberg J, Mekalanos J. 2006. Identification of a conserved bacterialprotein secretion system in Vibrio cholerae using the Dictyostelium hostmodel system. Proc. Natl. Acad. Sci. U. S. A. 103:1528 –1533. http://dx.doi.org/10.1073/pnas.0510322103.

42. Kitaoka M, Miyata S, Brooks T, Unterweger D, Pukatzki S. 2011. VasHis a transcriptional regulator of the type VI secretion system functional inendemic and pandemic Vibrio cholerae. J. Bacteriol. 193:6471– 6482. http://dx.doi.org/10.1128/JB.05414-11.

43. Marsh J, Taylor R. 1999. Genetic and transcriptional analyses of theVibrio cholerae mannose-sensitive hemagglutinin type 4 pilus gene locus.J. Bacteriol. 181:1110 –1117.

44. Chiavelli D, Marsh J, Taylor R. 2001. The mannose-sensitive hemagglu-tinin of Vibrio cholerae promotes adherence to zooplankton. Appl. Envi-ron. Microbiol. 67:3220 –3225. http://dx.doi.org/10.1128/AEM.67.7.3220-3225.2001.

45. Hasan N, Grim C, Haley B, Chun J, Alam M, Taviani E, Hoq M, MunkA, Saunders E, Brettin T, Bruce D, Challacombe J, Detter J, Han C, XieG, Nair G, Huq A, Colwell R. 2010. Comparative genomics of clinical andenvironmental Vibrio mimicus. Proc. Natl. Acad. Sci. U. S. A. 107:21134 –21139. http://dx.doi.org/10.1073/pnas.1013825107.

Genome of Vibrio fluvialis

March 2014 Volume 80 Number 6 aem.asm.org 2037

on June 18, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from