Correspondence: Jesús L. Romalde, Departamento de Microbiología y Parasitología CIBUS, Universidad de Santiago de Compostela, Campus Sur s/n. 15782, Santiago de Compostela, Spain. Tel.: +34 981 563100, ext. 16961; fax: +34 981 528085; e-mail: email@example.com
Vibrio tapetis is the etiological agent of brown ring disease (BRD) in clams. Phenotypic, antigenic and genetic variability have been demonstrated, with three groups being established associated with host origin. In this work we analyze the variability of representative strains of these three groups, CECT 4600T and GR0202RD, isolated from Manila clam and carpet-shell clam, respectively, and HH6087, isolated from halibut, on the basis of the whole proteome analysis by 2D-PAGE and multilocus sequence analysis (MLSA). A quantitative analysis of the proteome match coefficient showed a similarity of 79% between the clam isolates, whereas fish isolate showed similarities lower than 70%. A preliminary mass spectrometry (MS) assay allowed the identification of 27 proteins including 50S ribosomal protein L9, riboflavin synthase β subunit, ribose-phosphate pyrophosphokinase and succinyl-CoA synthase α subunit. The MLSA approach gave similar results, showing a 99.4% similarity of the clam isolates, which was higher than that observed between the fish isolate and either clam strain (98.2%). The topology of the maximum parsimony tree, obtained from 2D-PAGE analysis, and the phylogenetic tree, constructed with the maximum likelihood algorithm from concatenated sequences of 16S rRNA gene and five housekeeping genes (atpA, pyrH, recA, rpoA and rpoD), was very similar, confirming the closer relationship between the two clam isolates.
Vibrio species are extensively distributed in marine environments, associated with a wide range of marine organisms; some of the species are pathogenic to humans (Thompson et al., 2006; Beaz-Hidalgo et al., 2010).
Genotyping strategies such as restriction fragment length polymorphism and pulse field gel electrophoresis have been used traditionally for epidemiological analysis of Vibrio isolates (Castro et al., 1997; Romalde et al., 2002). PCR typing methods have also been widely used, including randomly amplified polymorphic DNA analysis and repetitive-sequence-based polymerase chain reaction based on polymorphic, repetitive extragenic palindromic sequences and enterobacterial repetitive intergenic consensus (Rodríguez et al., 2006). More recently, amplified fragment length polymorphism and multilocus sequence analysis (MLSA) (Maiden, 2006) have allowed a more precise identification of Vibrio species (Beaz-Hidalgo et al., 2008, 2010; and references therein).
Proteomics could complement and extend the nucleic acid analytical technologies, being an experimental link between the expressed product and the genome (Lester & Hubbard, 2002; Phillips & Bogyo, 2005; Norbeck et al., 2006; Cash, 2009; Zhang et al., 2010). 2D-PAGE has been successfully applied for the discrimination of closely related isolates (Cash et al., 1995; Dumas et al., 2008), revealing even more variability than with DNA–DNA hybridization, as protein content reflects dynamic changes produced in the cells as a response to changes in the environment (Andersen et al., 1984; Cash, 2009; Zhang et al., 2010).
Vibrio tapetis is the causative agent of an epizootic infection in adult clams called brown ring disease (Borrego et al., 1996). The first studies indicated that strains of this pathogen constituted a homogeneous group. However, as new strains were isolated from different hosts, including different mollusk and fish species, some variability on the basis of their antigenic, phenotypic and genotypic characteristics has been demonstrated, leading to the description of three main groups within this species that correlate with the type of host (Castro et al., 1996, 1997; Romalde et al., 2002; Rodríguez et al., 2006).
In this work, a proteomic method, 2D-PAGE, was used to study the intraspecific variability of representative strains of the three groups described for V. tapetis, as well as an additional indication of their phylogenetic relationship. The results obtained were compared with those of MLSA, a well established genetic technique to infer bacterial phylogeny.
Materials and methods
Representative strains of the three previously described groups of V. tapetis (Rodríguez et al., 2006) with different phenotypical, serological and genetic profiles as well as different host origin were used in this study: CECT 4600T, type strain of the species isolated from Manila clam (Ruditapes philippinarum), GR0202RDRD obtained from carpet shell clam (Ruditapes decussatus) and HH6087 isolated from halibut (Hipoglossus hipoglossus) (Borrego et al., 1996; Novoa et al., 1998; Reid et al., 2003). The bacteria were routinely grown aerobically on marine agar (MA) (Pronadisa, Spain) at 15 °C for 72 h. Stock cultures were maintained frozen at −80 °C in marine broth (MB) (Pronadisa) supplemented with 15% glycerol (v/v).
Bacterial inocula with 109 cells mL−1 were prepared by diluting the bacterial suspension to an OD of 1 (OD580 nm). For each strain, 1 L of sterile MB was inoculated to achieve a final concentration of 105 cells mL−1 and was aerobically incubated in a Innova 4340 rotary shaker (70 r.p.m.) (New Brunswick Scientific) at 15 °C for 72 h.
Bacteria were harvested and washed with Tris–buffered sucrose (10 mmol Tris, 250 mmol sucrose pH 7) and lyophilized. Proteins were extracted by suspending 40 mg of lyophilized bacteria in 1 mL standard lysis buffer – 7 M urea, 2 M thiourea, 4% CHAPS [3-(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate] – and 65 mM dithiothreitol (DTT) for 3 h at 27 °C and sonication (three cycles of 10 pulses). Next, samples were centrifuged at 12 000 g for 30 min and supernatants were collected and subjected to protein precipitation using the Clean-up kit (GE Healthcare, Sweden). After suspension of the pellet in 1 mL of lysis buffer, the protein concentration was measured with a CB-X protein assay kit (Gbiosciences). Finally, samples were stored at −80 °C prior to use.
Isoelectrofocusing (IEF) was performed using a Protean IEF cell (Bio-Rad) and 24 cm pH 4-7 IPG strips (GE Healthcare). For each sample, 400 μg of protein was resuspended in 390 μL of rehydration buffer (7 M urea, 2 M thiuorea, 4% CHAPS, 0.6% DTT, 1% IPG buffer 4-7 and bromophenol blue traces). IEF was carried out at 20 °C in the following steps: active rehydration (50 V) for 12 h, 250 V for 30 min, 500 V for 1 h, 1000 V for 1 h, 4000 V for 2 h, 8000 V for 2 h and 10 000 V, to achieve 65 kVh.
Prior to running the second dimension, strips were equilibrated at room temperature for 15 min with an equilibration solution [6 M urea, 50 mM Tris–HCl pH 8.8, 30% glycerol, 2% sodium dodecyl sulfate (SDS)] with the addition of 1% DTT, and for other 15 min in the same solution supplemented with 2.5% iodoacetamide. Strips were placed on top of a 21 ×26 cm 12.5% polyacrylamide gel and fixed with sealing solution (25 mM Tris, 192 mM glycine, 0.1% SDS, 0.5% agarose, 0.01% bromophenol blue). The second dimension was performed according to Laemmli (1970), in an EttanDalt-Six electrophoretic system (GE Healthcare) at 30 °C overnight (5 mA per gel for 1 h, 10 mA per 1 h, followed by 16 mA per gel until bromophenol blue dye reached the bottom of the gel). Three independent cultures as well as protein extractions and 2D-PAGE were performed to assess the reproducibility of the experiment.
Protein visualization and image analysis
Gels were stained with Coomassie Brilliant Blue (CBB). CBB staining was carried out according to Neuhoff et al. (1988) with minor modifications and scanned in a Microtek 9800XL densitometer (Microtek) at 300 dpi resolution. Gels were stored in vacuum-sealed plastic bags at 4 °C. pdquestadvance software version 8.0 (Bio-Rad) was used for spot detection and quantitation, and to assess reproducibility.
In-gel digestion and peptide extraction
Protein spots chosen for mass spectrometric analysis (MS) were excised from the gels and manually digested. The gel pieces were rinsed three times with AmBic buffer (50 mM ammonium bicarbonate in 50% HPLC grade methanol (Scharlau, Spain) and once with 10 mM DTT (Sigma-Aldrich). The gel pieces were rinsed twice with AmBic buffer and dried in a SpeedVac before alkylation with 55 mM iodoacetamide (Sigma-Aldrich) in 50 mM ammonium bicarbonate. Once again, the gel pieces were rinsed with HPLC grade AmBic buffer (Scharlau), before being dehydrated by the addition of HPLC grade acetonitrile (Scharlau) and dried in a SpeedVac. Modified porcine trypsin (Promega) was added to the dry gel pieces at a final concentration of 20 ng μL−1 in 20 mM ammonium bicarbonate, incubating them at 37 °C for 16 h. Peptides were extracted three times by 20 min incubation in 40 μL of 60% acetonitrile in 0.5% HCOOH (formic acid). The resulting peptide extracts were pooled, concentrated in a SpeedVac and stored at −20 °C.
Mass spectrometric analysis and protein identification
A combination of matrix-assisted laser-desorption-ionization time-of-flight mass spectrometry (MALDI-TOF) (MS) and MALDI-TOF/TOF (MS/MS) was used for protein identification according to the following procedure. Dried samples were dissolved in 4 μL of 0.5% formic acid. Equal volumes (0.5 μL) of peptide and matrix solution, consisting of 3 mg α-cyano-4-hydroxycinnamic acid (CHCA) dissolved in 1 mL of 50% acetonitrile in 0.1% trifluoroacetic acid, were deposited using the thin-layer method onto a 384 Opti-TOF MALDI plate (Applied Biosystems). Mass spectrometric data were obtained in an automated analysis loop using a 4800 MALDI-TOF/TOF analyzer (Applied Biosystems). MS spectra were acquired in reflectron positive-ion mode with an Nd:YAG, 355-nm wavelength laser, averaging 1000 laser shots, and at least three trypsin autolysis peaks were used as internal calibration. All MS/MS spectra were performed by selecting the precursors with a relative resolution of 300 full width at half maximum and metastable suppression. Automated analysis of mass data was achieved using the 4000 Series explorer software V3.5. Peptide mass fingerprinting (PMF) and peptide fragmentation spectra data of each sample were combined through the GPSexplorer Software v3.6 using mascot software v2.1. (Matrix Science) to search against a non-identical protein database (NCBInr release data 20100526), with 30 p.p.m. precursor tolerance, 0.35 Da MS/MS (analysis of the tandem mass) fragment tolerance, carbamydomethyl cysteine (CAM) as fixed modification, and oxidized methionine as variable modification, allowing one missed cleavage. All spectra and database results were manually inspected in detail using the above software. Protein scores greater than 56 were accepted as statistically significant (P < 0.05), and the identification was considered positive when the protein score confidence interval (CI) was above 98%. In the case of MS/MS spectra, the total ion score CI was greater 95%.
Similarity percentages between V. tapetis isolates were calculated on the basis of protein profile similarities calculated between pairs of isolates using the simple matching co-efficient (Sneath & Sokal, 1973).
Phylogenetic analysis based on proteins
bionumerics 5.1 2D software (Applied-Maths) was used to construct a maximum parismony tree based on the different protein content of V. tapetis isolates.
Phylogenetic analysis based on MLSA
Genomic DNA extraction and amplification of the 16S rRNA gene was performed as previously described (Beaz-Hidalgo et al., 2008). Sequences for five protein-coding housekeeping genes, atpA (α subunit of ATPase), pyrH (uridyl monophosphate kinase), recA (recombinase A), rpoA (α subunit of RNA polymerase) and rpoD (RNA polymerase sigma factor), were performed according to Thompson et al. (2004, 2005, 2007) and Pascual et al. (2010). Sequencing reactions were performed with the GenomeLab DTCS-Quick Start kit (Beckman Coulter, Ireland). Sequence data analysis was performed with the dnastarseqman program (Lasergene). The percentage of similarity of concatenated sequence of genes was calculated using the dnastarmegaling program (Lasergene).
For maximum-likelihood (ML) analysis, the optimal model of nucleotide substitution was estimated with the program jmodeltest 0.1.1 (Posada, 2008) using the Akaike information criterion. The ML estimation was implemented in phyml (Guindon & Gascuel, 2003), using the GTR model as recommended by jmodeltest 0.1.1. Bootstrap analyses were performed using 1000 replications.
The three strains yielded different numbers of spots in 2-DE gels, despite loading the same quantities of protein. There were 729 (± 13 standard deviation), 681 (± 2) and 556 (± 6) spots for CECT 4600T, GR0202RD and HH6087, respectively (Fig. 1). Technical replicates showed a high degree of congruence (0.91 for CECT 4600T and GR0202RD, 0.85 for HH6087) (Fig. 1).
Visual inspection of gels showed that the majority of proteins detected were localized in the acidic part of the pH range studied and they also showed similar or different protein profiles depending on a specific molecular weight region (Fig. 2). Thus, the high molecular weight region was very similar in all strains, whereas the low molecular weight region was more similar between CECT 4600T and GR0202RD strains than between CECT 4600T and HH6087 strains. The protein profile of the middle region seems to be specific for each strain, although some common spots are also present in this region (Fig. 2).
Sixty representative proteins (common and unique for each strain) of the three strains were selected and sequenced by MS but only 27 of these proteins were identified (Table 1). Interestingly, two proteins selected as unique for CECT 4600T and GR0202RD were the same, representing a hypothetical protein pVT1_26.
Table 1. Selected proteins of the three strains of Vibrio tapetis identified by MS
50S ribosomal protein L9
Peptidyl-prolyl cis–trans isomerase B (rotamase B)
The level of protein profile similarity within V. tapetis was calculated between pairs of strains applying the simple matching co-efficient formula. Results showed a 79% similarity between CECT 4600T and GR0202RD strains, 69% similarity between CECT 4600T and HH6087 strains, and 60% similarity between GR0202RD and HH6087 strains. These results were used to construct an un-rooted tree (Fig. 3), which showed that the GR0202RD strain was clearly more similar to CECT 4600T than to HH6087.
Fragments of the 16S rRNA gene (1531 bp) and five coding-protein housekeeping genes, rpoD (535 bp), rpoA (863 bp), pyrH (540 bp), atpA (1194 bp) and recA (789 bp), were sequenced to yield a concatenated sequence of 4090 nucleotides, which corresponded to more than 80% of the coding regions of each gene. Identities between concatenated sequences of the three isolates were 99.4% between CECT 4600TT and GR0202RD, 98.2% between CECT 4600TT and HH6087, and 98.2% between GR0202RD and HH6087. These results indicate a higher similarity between clam isolates (CECT 4600TT and GR0202RD) than between either clam and the fish isolate (HH6087). This similarity can also be seen in the phylogenetic tree, in which clam isolates are closer to each other than to the fish isolate (Fig. 3).
Automatic software analysis revealed differences in protein spot number, ranging from 729 spots for strain CECT 4600T to 556 spots for strain HH6087. The similarity of protein profiles was higher between strains isolated from clam species (CECT 4600T and GR0202RD) than between these strains and the fish isolate (HH6087). Spot number and the similarity percentages between the V. tapetis strains are in agreement with those reported in previous studies for other bacterial species (Gormon & Phan-Thanh, 1995; Govorun et al., 2003; Dopson et al., 2004).
The majority of proteins detected, regardless of the strain, were localized in the acidic part of the pH range studied. This finding agrees with results of other authors who observed a predominance of proteins with low pI over high pI in halophilic bacteria (Kiraga et al., 2007). The identified proteins could be related to important functions in the cells, such as 50S ribosomal protein L9, metabolic pathways, including riboflavin synthase β subunit, ribose-phosphate pyrophosphokinase and peptidyl-prolyl cis–trans isomerase B (rotamase B), as well as integrases, transcriptional regulators and ABC transporter. Two nonequivalent spots in the profiles of strains CECT 4600T and GR0202RD were identified as the same hypothetical protein pVT1_26, indicating the possible existence of charge variants of the same protein between isolates, as previously described for other bacterial species (Cash, 2009). The lack of a complete genome sequence for V. tapetis and, therefore, the unavailability of an appropriate database is reflected in our study, where only 27 of the 60 proteins sequenced by MS were identified, and indicates the necessity for further studies to characterize the proteome of this pathogen.
In comparison with proteomics, genetic procedures such as MLSA have the advantage that the information is fairly consistent; the procedure is unaffected by the growth conditions of bacteria and can generate highly reproducible and portable data, which enables the comparison of results between laboratories using the public online databases. MLSA has been demonstrated to be a powerful, both intra- and interspecific, discriminative tool within the Vibrio genus (Thompson et al., 2004, 2005, 2007, 2009; Pascual et al., 2010). The choice of the protein encoding genes for the MLSA is the most important aspect in a correct MLSA analysis. This choice is particularly difficult in the case of a set of strains belonging to the same species or to closely related taxa, due for the need for genes that are able to measure such low variability. In our case, each selected gene has been used previously for Vibrio species (Thompson et al., 2004, 2005, 2007) and the results obtained were in agreement with those reached using genotyping methods (Castro et al., 1996, 1997; Romalde et al., 2002; Rodríguez et al., 2006).
Both methods, 2D-PAGE and MLSA, rendered trees with similar topology, the clam isolates appearing to be more closely related than those from fish. In addition, the relative branching order is clearly in agreement with the three genetic groups previously described on the basis of typing methods (Romalde et al., 2002; Rodríguez et al., 2006). The congruence between the results obtained in the phylogenetic study of housekeeping genes (conservative approach) and the analysis of the whole proteome of the isolates (dynamic approach) provide an inter-validation of the techniques.
In conclusion, the proteomic approach using 2D-PAGE can be a useful complementary tool for the study of the intraspecific variability of V. tapetis. In addition, the method does not require prior information about the genome sequence and possesses the added value of describing gene expression at protein level, which can furnish helpful information on host–pathogen interaction and pathogenic processes.
This work was partially supported by Grants AGL2006-13208-C02-01 and AGL2010-18438 from the Ministerio de Ciencia e Innovación (MICINN) (Spain). The kind donation of strains by Drs J.J. Borrego (University of Málaga, Spain) and T.H. Birkbeck (University of Glasgow, UK) is gratefully acknowledged. S.B. and J.B.C. thank the Ministerio de Ciencia Innovación (MICINN) and Xunta de Galicia (Spain) for research fellowships.
In loving memory of J.L. López who died of cancer during the course of this work.