Comparative Proteomics and Secretomics Revealed Virulence and Antibiotic Resistance-Associated Factors in Vibrio parahaemolyticus Recovered From Commonly Consumed Aquatic Products

Vibrio parahaemolyticus is a seafoodborne pathogen that can cause severe gastroenteritis and septicemia diseases in humans and even death. The emergence of multidrug-resistant V. parahaemolyticus leads to difficulties and rising costs of medical treatment. The bacterium of environmental origins containing no major virulence genes (tdh and trh) has been reported to be associated with infectious diarrhea disease as well. Identification of risk factors in V. parahaemolyticus is imperative for assuming food safety. In this study, we obtained secretomic and proteomic profiles of V. parahaemolyticus isolated from 12 species of commonly consumed aquatic products and identified candidate protein spots by using two-dimensional gel electrophoresis and liquid chromatography tandem mass spectrometry techniques. A total of 11 common and 28 differential extracellular proteins were found from distinct secretomic profiles, including eight virulence-associated proteins: outer membrane channel TolC, maltoporin, elongation factor Tu, enolase, transaldolase, flagellin C, polar flagellin B/D, and superoxide dismutase, as well as five antimicrobial and/or heavy metal resistance-associated ABC transporter proteins. Comparison of proteomic profiles derived from the 12 V. parahaemolyticus isolates also revealed five intracellular virulence-related proteins, including aldehyde-alcohol dehydrogenase, outer membrane protein A, alkyl hydroperoxide reductase C, phosphoenolpyruvate-protein phosphotransferase, and phosphoglycerate kinase. Additionally, our data indicated that aquatic product matrices significantly altered proteomic profiles of the V. parahaemolyticus isolates with a number of differentially expressed proteins identified. The results in this study meet the increasing need for novel diagnosis candidates of the leading seafoodborne pathogen worldwide.


INTRODUCTION
Vibrio parahaemolyticus is a gram-negative bacterium that thrives in marine, estuarine, and aquaculture environments worldwide (Ghenem et al., 2017). The bacterium is frequently detected from seafood such as shellfish, shrimp, and fish. Consumption of raw or inadequately cooked seafood, primarily oysters, can cause acute gastroenteritis in humans (Ghenem et al., 2017). Vibrio parahaemolyticus was initially identified in an outbreak of infectious diarrhea disease in 1950 in Osaka, Japan, caused by contaminated semidried juvenile sardines, which sickened 272 and killed 20 people (Fujino et al., 1965). Henceforth, outbreaks and prevalence of the infectious disease are reported in many countries in the world (Ghenem et al., 2017;Baker-Austin et al., 2018). In the United States, there were approximately 35,000 cases of acute gastroenteritis infected by V. parahaemolyticus per year from 2000 to 2008 (Baker-Austin et al., 2018). Data from foodborne disease outbreak reporting system in China showed that 42.3% of biohazard cases reported were attributed to V. parahaemolyticus from 2011 to 2016 . Vibrio parahaemolyticus has been the leading cause of infectious diarrhea disease, especially among adults in coastal regions in China (Yang C. et al., 2019). Recently, V. parahaemolyticus was reported to be associated with acute hepatopancreatic necrosis disease, a newly emerging shrimp disease, which severely damaged the global shrimp industry . The global spread of V. parahaemolyticus underscores the need for a better understanding of virulence traits of the bacterium.
Most pathogenic V. parahaemolyticus strains have two major virulence genes (tdh and trh) encoding thermostable direct hemolysin (TDH) and TDH-related hemolysin. The former is a heat-resistant and pore-forming toxin composed of 156 amino acids, whereas the latter contains 189 amino acids and shares 54.8-68.8% identify in nucleotide level with the TDH (Zhang and Orth, 2013). Both these toxins have hemolytic activity, enterotoxin activity, cardiotoxicity, and cytotoxicity . Nevertheless, approximately 90-99% of V. parahaemolyticus isolates of environmental origins were detected negative for the toxic factors (Raszl et al., 2016;Park et al., 2018). Previous studies have revealed some other cytotoxic factors to human gastrointestinal cells secreted by type III and type VI secretion systems (T3SS and T6SS) in V. parahaemolyticus (Makino et al., 2003). Two types of T3SSs (T3SS1 and T3SS2) were identified, of which T3SS1 located in chromosome I is necessary for the bacterial survival in the environment (Makino et al., 2003;De Nisco et al., 2017). Recent studies indicated that T3SS1 appeared to inject effectors, for example, VopQ, VopR, VopS, and VPA0450, directly into human gastrointestinal cells, which lead to the induction of rapid autophagy followed by cell rounding, eventually cell lysis (Osorio, 2018). The T3SS2 located on chromosome II plays a critical role in the enteropathogenicity of the bacterium (Makino et al., 2003). The effectors of T3SS2 can cause enterotoxicity by destroying the cell cytoskeleton, for example, VopC, VopL, VopV, and VopO, or manipulating cell signaling transduction, for example, VopA, VopT, VopZ, andVPA1380 (De Souza Santos et al., 2017;Osorio, 2018;Luo et al., 2019). Additionally, T6SS is a complex secretory device capable of secreting effectors into host mammalian cells as well as target bacterial cells . It consists of a series of components including structural proteins, translocators, secreted proteins, and some other proteins with auxiliary function, for example, DotU, IcmF, ClpV, Hcp, VgrG, and PAAR (Salomon et al., 2014). Similar to T3SS, V. parahaemolyticus encodes two T6SSs (T6SS1 and T6SS2) located on chromosomes I and II, respectively. T6SS1 is active under warm marine-like conditions (3% NaCl, 30 • C), whereas T6SS2 is active under low salinity and low temperature (1% NaCl, 23 • C) (Salomon et al., 2013). Recent studies have indicated that T6SSs are tightly regulated in pathogenic bacteria, for example, Vibrio cholerae, Pseudomonas aeruginosa, and V. parahaemolyticus, and induced by external conditions and cues such as quorum sensing, salinity, temperature, mucin, chitin, surface sensing, and cell membrane damage (Ben-Yaakov and Salomon, 2019).
There is growing evidence to support that V. parahaemolyticus isolates lacking the virulence factors can cause infectious diarrhea disease in humans (Castillo et al., 2018). For example, Thongjun et al. (2013) reported that 9-10% of clinical isolates recovered from diarrhea patients were identified as non-toxigenic V. parahaemolyticus in South Thailand between 2001 and 2010. On the other hand, non-toxigenic V. parahaemolyticus isolates of environmental origins may possess additional pathogenicity mechanisms. The emergence of multidrug-resistant (MDR) V. parahaemolyticus increases the difficulty and cost of clinical treatment (Elmahdi et al., 2016). For instance, Hu and Chen (2016) reported that 74.5% of the V. parahaemolyticus strains (n = 208) isolated from 10 species of commonly consumed aquatic products in Shanghai, China, were resistant to more than three antimicrobial agents. Yang et al. (2017) isolated 98 V. parahaemolyticus strains from 504 seafood samples in 11 provinces of China and found that 68.38% of the isolates showed MDR phenotypes. Recently, Lopatek et al. (2018) reported that 55.8% of the V. parahaemolyticus isolates (n = 104) recovered from 595 samples collected from Denmark, France, Germany, Italy, the Netherlands, Norway, Poland, Spain, Sri Lanka, and Turkey had resistance to ampicillin and streptomycin and one isolate resistant to ampicillin, streptomycin, and ciprofloxacin. Therefore, identification of virulence and resistance-associated factors in V. parahaemolyticus is imperative for food safety systems, particularly in developing countries.
The combination of two-dimensional gel electrophoresis (2D-GE) and liquid chromatography tandem mass spectrometry (LC-MS/MS) techniques is widely used in current proteomics researches (Kim and Cho, 2019). The 2D-GE couples isoelectric focusing (IEF) in the first dimension and sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) in the second dimension, which separate proteins according to isoelectric points and molecular masses, respectively. Several hundreds of individual protein abundances separated by 2D-GE can be quantified in the cell population or sample tissues. The LC-MS/MS is also a powerful tool that can efficiently separate and identify constituents in protein mixtures with high sensitivity and specificity (Aslam et al., 2017). For instance, based on the LC-MS/MS technique, Zhong et al. (2019) compared V. parahaemolyticus proteomes between the resuscitation state and viable but non-culturable state and found a total of 429 differentially expressed proteins mainly involved in cellular process, establishment of localization, and metabolic process. Pang et al. (2020) analyzed 102 acetylation modified proteins of Vibrio alginolyticus HY9901 by LC-MS/MS and identified five virulence factors, including HemL, FabB, FabD, FabF-3, and PhoR.
The People's Republic of China is the world's largest producer, consumer, and exporter of aquatic products, including fish, crustaceans, and shellfish . Data from China Fishery Statistical Yearbook (2018) showed that shellfish production (14,371,304 tons) is an important part of maricultural production in China, and the main species include Ostrea gigas thunberg (known as oyster) (34.0%), Ruditapes philippinarum, Mactra veneriformis, and Paphia undulate (known as clam) (29.1%), Placopecten magellanicus (known as scallop) (14.0%), Perna viridis (known as mussel) (6.5%), and Solen strictus (known as razor clam) (6.0%). Additionally, Aristichthys nobilis (known as bighead carp) and Ctenopharyngodon idellus (known as grass carp) are commonly consumed freshwater fish in China, accounting for 18.4% (5,345,641 tons) and 10.7% (3,097,952 tons) of the total freshwater aquaculture production (29,052,930 tons), respectively. The production of Litopenaeus vannamei (known as white-leg shrimp) was 1,672,287 tons, which was the most predominant among the crustaceans in 2017 . Oratosquilla oratoria (known as mantis shrimp) is one of the most economically important species of Pacific shrimp, and its output was 219,087 tons in 2017 . Continuous monitoring of food contaminants and identification of risk factors are crucial to safeguard the food supply chain (Chen and Alali, 2018). In our prior researches, V. parahaemolyticus contamination in 10 species of aquatic products was surveyed (He et al., 2016;Hu and Chen, 2016). Secretomic profiles of seven V. parahaemolyticus strains of two clinical and five aquatic product origins (Haliotis asinina, Moerella iridescens, Metapenaeus ensis, Exopalaemon carinicauda, and L. vannamei) were compared, and six virulence-associated proteins involved in bacterial pathogenicity were identified (He et al., 2015). Recently, a number of V. parahaemolyticus isolates from 25 species of commonly consumed shellfish, crustacean, and fish samples were isolated and characterized (Su et al., unpublished). Among these, 12 V. parahaemolyticus isolates recovered from eight shellfish (M. veneriformis, O. gigas thunberg, P. undulate, P. viridis, P. magellanicus, R. philippinarum, Sinonovacula constricta, and S. strictus), two crustacean (L. vannamei and O. oratoria), and two fish (A. nobilis and C. idellus) had multiple resistance to antimicrobial agents and heavy metals. The objectives of this study were (1) to characterize genetic diversity of the V. parahaemolyticus isolates recovered from 12 species of aquatic products; (2) to take advantage of 2D-GE and LC-MS/MS technologies to determine secretomic and proteomic profiles of the 12 V. parahaemolyticus isolates and to identify common and/or differential extracellular and intracellular proteins by comparative secretomic and proteomic analysis of the isolates from three kinds of aquatic products; and (3) to study implications of various aquatic product matrices on proteomics of the V. parahaemolyticus isolates. The secretomes and proteomes between the V. parahaemolyticus isolates and pathogenic reference strains were compared as well. To our knowledge, very little of such information in V. parahaemolyticus recovered from all the species (except L. vannamei) was available to date. The results in this study support the increasing need for novel sites and targets for food safety control of V. parahaemolyticus contamination in aquatic products.

V. parahaemolyticus Strains and Culture Conditions
Vibrio parahaemolyticus strains used in this study are listed in Table 1, and their genotypes and phenotypes were characterized by Su et al. (under review). Vibrio parahaemolyticus strains were individually inoculated from −80 • C stock in our laboratory in Shanghai Ocean University and incubated in tryptic soy broth (TSB) medium (pH 8.5, 3.0% NaCl; Beijing Land Bridge Technology Co., Ltd., Beijing, China) at 37 • C. Bacterial cell cultures grown at mid-logarithmic phase with OD 600nm values ranging from 0.5 to 0.6 without shaking were collected for extracellular protein extraction, whereas those at late-logarithmic phase with OD 600nm values ranging from 0.8 to 1.0 with shaking at 180 revolutions/min (rpm) were used for intracellular protein extraction (He et al., 2015). Vibrio parahaemolyticus ATCC33847 (tdh + trh − ) and ATCC17802 (tdh − trh + ) were used as standard strains (Table 1).

Extraction of Extracellular and Intracellular Proteins of V. parahaemolyticus Isolates
Extracellular proteins of V. parahaemolyticus isolates were extracted according to the method described previously (He et al., 2015) with minor modifications. Briefly, growth curves of V. parahaemolyticus isolates incubated in the TSB medium (pH 8.5, 3% NaCl) at 37 • C were determined using Bioscreen Automatic Growth Curve Analyzer (BioTek Instruments, Inc., Winooski, VT, United States). A 20 µL of Bacterial Protease Inhibitors Complex (Sangon) was added to each 100 mL of bacterial cell culture at midlogarithmic phase, which was then centrifuged at 6,000 g for 10 min at 4 • C. The supernatant was filtered through 0.22-µm-pore-size membrane filters (Millipore, Bedford, MA, United States) to remove residual bacterial cells. The filtrate was precipitated by adding ice-cold trichloracetic acid (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) to a final concentration of 10% (vol/vol) and then incubated on ice for 5 h. The extracellular proteins were collected by centrifugation at 12,000 rpm for 30 min at 4 • C. The resulting protein pellet was washed with ice-cold acetone (Sinopharm) for three times, air-dried and stored at −80 • C for further analysis. For the extraction of intracellular proteins, V. parahaemolyticus isolates were individually incubated in the TSB medium (pH 8.5, 3.0% NaCl) to late-logarithmic phase at 37 • C with shaking. Bacterial cell cultures were collected as described above, and intracellular proteins were extracted using the Bacterial Protein Extraction Kit (Sangon) containing protease inhibitors, according to the manufacturer's instructions. Protein concentration was measured using Bradford Protein Assay Kit (Sangon) with bovine serum albumin as the standard protein.
For intracellular proteins, an aliquot of each 400 µL of protein sample was individually applied to the strips (pH 4-7, 17 cm) (Bio-Rad) and passive rehydrated for 17 h at 17 • C. After rehydration, IEF was run with a 9-step program: 50, 100, 500, and 1,000 V for 1 h with slow ramping, respectively; 2,000 and 4,000 V for 1 h with linear ramping, respectively; 6,000 and 8,000 V for 1 h with rapid ramping, respectively; and 10,000 V with rapid ramping until 80,000 V-hour was reached. The strips were individually transferred onto 12.5% separation gel using PROTEAN II XL electrophoresis cell (Bio-Rad) with a 2-step program: 5 mA for 5 h and 15 mA for 6 h.
After electrophoresis, the gels were stained using Protein Stains K (Sangon), according to the manufacturer's instructions. Silver-stained gels were scanned using GenoSens 1800 Series Gel Documentation and Analysis System (Clinx Science Instruments Co. Ltd., Shanghai, China). Protein spot detection, spot matching, and quantitative intensity analysis were performed using PDQuest Advanced-8.0.1 software (Bio-Rad).

LC-MS/MS Analysis
The LC-MS/MS analysis was carried out at HooGen Biotech, Shanghai, China. The visible and discriminative protein spots were individually excised from 2D-GE gels, and gel pieces were dried under vacuum. A 15 µL of digestion solution (25 mM ammonium bicarbonate) containing trypsin (12.5 ng/mL; Promega, Madison, WI, United States) was added to each protein spot sample tube and incubated with gentle shaking at 37 • C for 16 h. The resulting peptides were identified by Q Exactive Mass Spectrometer [Thermo Fisher Scientific (TFS), Waltham, MA, United States] coupled with Easy nLC 1200 Chromatography System (TFS). The peptide mixture was loaded onto C18-reversed phase column (15 cm long, 75-µm inner diameter) packed in-house with RP-C18 5 µm resin in buffer A [0.1% formic acid (Sigma-Aldrich, St. Louis, MO, United States) in high-performance liquid chromatography-grade water] and separated with a linear gradient of buffer B [0.1% formic acid in 84% acetonitrile (Sigma-Aldrich, United States)] at a flow rate of 250 nL/min over 60 min. The collected LC-MS/MS data files were converted to Mascot generic format (mgf) and then imported into Mascot version 2.2 server (Matrix Science, London, United Kingdom) for automated peptide identification using UniProt Vibrio parahaemolyticus database (download in September 2019; 89189 sequences). The criteria for peptide matching and protein calls included enzyme: trypsin; max missed cleavage: 2; fixed modifications: carbamidomethyl (C); variable modification: oxidation (M); peptide mass tolerance: ± 20 ppm; fragment MS/MS tolerance: 0.1 Da; Mascot score: ≥20; ion score: >20; and false discovery rate: <0.01 at peptide and protein level.
Common protein spots on secretomic profiles were marked with blue circles, whereas differential protein spots on secretomic and proteomic profiles were marked in red numbers. The identified proteins were subjected to Generic Gene Ontology (GO) Term Finder 1 for GO analysis (Yang et al., 2015).

Preparation of Aquatic Product Matrix Media
Aquatic product matrices were prepared according to the method described by Chaitiemwong et al. (2010) with minor modifications. Briefly, 12 species of fresh aquatic products were collected in two local fish markets [Guzong road fish market (30 • 53 11.34 N, 121 • 55 3.09 E) and Luchaogang seafood market (30 • 51 34.47 N, 121 • 51 3.15 E)] in Shanghai and transferred in sterile sealed plastic bags (Nanjing Maojie Microbial Technology Co., Ltd., Nanjing, China) in an ice box (700 × 440 × 390 mm) to the laboratory in Shanghai Ocean University for analysis immediately. A 10 g (wet weight) fresh meat of each sample was aseptically placed in a homogeneous bag (Sangon) with 40 mL sterile water and homogenized for 3 min using a laboratory blender BagMixer (Shanghai Jingxin Industry Development Co., Ltd., Shanghai, China). The mixture was centrifuged at 8,000 g for 20 min, and the supernatant was filtered through 0.22-µm-poresize membrane filters (Millipore). The filtrate of each aquatic product sample was supplemented into the TSB medium (pH 8.5, 3% NaCl) to produce corresponding aquatic product matrix medium with a final concentration of 1% (vol/vol).

Measurement of Compositions of Aquatic Product Matrices
Total protein content was measured using Bradford Protein Assay Kit (Sangon) according to the manufacturer's instructions. Crude fat content was determined by the automated Soxhlet method described by Shin et al. (2013) with minor modifications. Briefly, a 5 mL of aquatic product matrix was individually weighed and dried in a Petri dish at 104 • C for 3 h. The dried products were wiped with cotton balls moistened with petroleum ether (Shanghai Titan Scientific Co., Ltd., Shanghai, China). Crude fat on cotton balls was extracted by an automated Soxtherm Fat Extraction System (Gerhardt, Bonn, Germany) according to the manufacturer's instructions. The extracted products remaining in a weighed beaker were dried at 104 • C for 30 min and cooled in a dessicator to room temperature. Crude fat in the beakers was weighed and calculated. Carbohydrate (saccharides) content was measured by the phenol-sulfuric acid method described by Albalasmeh et al. (2013) with minor modifications. A 1 mL of each sample was mixed with 1 mL of 5% (wt/wt) aqueous solution of redistilled phenol (Shanghai Macklin Biochemical Co., Ltd., Shanghai, China) in a glass tube ( 15 mm, 150 mm). A 5 mL of sulfuric acid (95-98%; Shanghai Kuling Fine Chemical Co., Ltd., Shanghai, China) was added into the mixture and then incubated in a boiling water bath for 20 min. After cooling in ice for 2 min and at room temperature for 15 min, the absorption of the reaction mixture at 490 nm was measured using BioTek Synergy 2. Carbohydrate concentration was calculated according to a standard curve prepared with glucose (Sinopharm).

Reverse Transcription-PCR Assay
The reverse transcription (RT)-PCR assay was performed according to the method described by Zhu et al. (2017). Briefly, total RNA was extracted using RNeasy Mini Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. The DNA was removed from the samples with the RNase-Free DNase Set (Qiagen). The RT reactions were performed using PrimeScript TM RT reagent Kit with gDNA Eraser [Takara Biomedical Technology (Beijing) Co., Ltd., Beijing, China]. Relative quantitative PCRs were performed using TB Green R Premix Ex Taq TM II kit (Tli RNaseH Plus, Beijing, China) according to the manufacturer's instructions. All RT-PCRs were carried out in a 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, United States) under the following conditions: initial denaturation at 50 • C for 2 min and at 95 • C for 10 min, followed by 40 cycles of denaturation at 95 • C for 15 s and primer annealing at 60 • C for 1 min. The 16S RNA was used as the internal reference gene as described previously. The primers (Supplementary Table S1) were designed by Premier 5.0 software 2 and synthesized by Sangon (Shanghai, China). In a 20 µL of RT-PCR solution, 1,000, 200, 20, 2, 0.2, and 0.0 ng of cDNA template was individually added for standard curve preparation. The PCR efficiency was 90.0-100.0%, and r 2 > 0.980 for the primers. All tests were performed in triplicate in this study.
Evolutionary relatedness of the V. parahaemolyticus strains was examined by the ERIC-PCR assay. This analysis revealed that all the tested strains were classified into 14 different 2 http://www.premierbiosoft.com/ ERIC genotypes, and the UPGMA algorithm grouped the ERIC genotypes into five distinct clusters (Clusters A-E) (Figure 1). Vibrio parahaemolyticus ATCC33847 (tdh + trh − ) and ATCC17802 (tdh − trh + ) strains of clinical origin fell into Cluster B, whereas the remaining of aquatic product origins fell into Clusters A, C, D, and E. Vibrio parahaemolyticus isolates recovered from three shellfish (O. gigas thunberg, P. undulate, and S. constricta), two crustaceans (L. vannamei and O. oratoria), and two fish (A. nobilis and C. idellus) were grouped into the largest Cluster D with 7 ERIC genotypes, whereas Cluster A contained only the V. parahaemolyticus CHN-N3-2 isolate that had tolerance to 4 antimicrobial agents and 5 heavy metals. The Hg and AMP/KAN/RIF/STR resistance profile was the most predominant among the strains tested (Figure 1). These results demonstrated genetic diversity of the V. parahaemolyticus isolates recovered from the 12 species of commonly consumed aquatic products.

Distinct Secretomic Profiles of the V. parahaemolyticus Isolates
The supernatant of the V. parahaemolyticus cultures at midlogarithmic phases was collected, and extracellular proteins were extracted and analyzed by 2D-GE. This analysis uncovered distinct secretomic profiles of the 12 V. parahaemolyticus isolates of various aquatic product origins (Figure 2). The patterns yielded from three independent 2D-GE gels of each isolate were consistent (figures not shown). Based on the consensus patterns, 11 extracellular protein spots (marked with blue circles, Figure 2) were observed at similar locations on all the 2D-GE patterns derived from the 12 V. parahaemolyticus isolates. Notably, the V. parahaemolyticus CHN-B2-28 isolate recovered from R. philippinarum appeared to secret more extracellular proteins (35) than the other isolates (17-30). Additionally, a total of 28 differential extracellular proteins (marked with different numbers in red, Figure 2) were found from the distinct secretomic profiles. All common protein spots and the remaining differential protein spots among the 12 V. parahaemolyticus isolates were excised from the 2D-GE gels and digested with the trypsin. The resulting peptides were further identified by the LC-MS/MS analysis.

Identification of Common and Differential Extracellular Proteins of the V. parahaemolyticus Isolates
In this study, the LC-MS/MS technique was used to identify all common and differential extracellular proteins of the V. parahaemolyticus isolates. Based on the secretomic profiles, 11 common extracellular proteins shared among the 12 isolates were successfully obtained ( Table 2). Of these proteins, the majority (n = 7) were cell membrane and periplasm components, including TolC protein (Spot a), maltoporin (Spot b), maltodextrin-binding protein (Spot d), putrescine-binding periplasmic protein (Spot f), gram-negative porin family protein (Spot h), outer membrane protein W (OmpW, Spot j), and basalbody rod modification protein FlgD (Spot i). For the others  (n = 4), the Spot c was identified as elongation factor Tu (EF-Tu). Spot e was phosphoglucosamine mutase (PNGM) involved in carbohydrate metabolism and peptidoglycan biosynthetic pathway, whereas Spot g was fructose-bisphosphate aldolase that also acts as a transcriptional regulator in pathogenic Francisella (Ziveri et al., 2017). Additionally, the Spot k matched an uncharacterized protein encoded by the H320_01320 gene of V. parahaemolyticus 49 strain (GenBank: AONA00000000.1) with currently unknown function in public databases ( Table 2).
Remarkably, in this study, several identified extracellular proteins of the V. parahaemolyticus isolates have been reported to be involved in bacterial pathogenesis, including Spots a, b, c, S2, S5, S7, S8, and S26. The former three proteins TolC (Spot a), maltoporin (Spot b), and EF-Tu (Spot c) were secreted by all the isolates, and more than half of the isolates also secreted the enolase (Spot S2) and transaldolase (Spot S5). In contrast, only a few isolates secreted the flagellin C (Spot S7), polar flagellin B/D (Spot S8), and superoxide dismutase (Spot S26). For instance, the latter was only observed on the secretomic profiles derived from the CHN-B5-29 and CHN-B6-62 isolates recovered from P. magellanicus and S. constricta, respectively. Among all the V. parahaemolyticus isolates of aquatic product origins, the CHN-B2-28 isolate from R. philippinarum appeared to secret most of the extracellular virulence-associated proteins compared to the other isolates.

Secretomic Comparison of the V. parahaemolyticus Isolates With Pathogenic Reference Strains
The secretomic profiles of the V. parahaemolyticus ATCC33847 and ATCC17802 strains of clinical origin were also obtained for comparative secretomic analysis, although they have been reported in our prior research (He et al., 2015). One interesting observation was that the toxic ATCC33847 and ATCC17802 strains appeared to secret more extracellular proteins (59-62) than the V. parahaemolyticus isolates recovered from the 12 species of aquatic products (17-35), consistent with our prior results (He et al., 2015). Comparative secretomic analysis revealed that approximately 36.4% (4/11) of the common and 42.9% (12/28) of the differential extracellular proteins secreted by the 12 V. parahaemolyticus isolates were observed at similar locations on the secretomic profiles of the ATCC33847 and the ATCC17802 strains (Supplementary Figure S1). For instance, four extracellular proteins secreted by all the V. parahaemolyticus strains of aquatic product and clinical origins included the maltoporin (Spot b), EF-Tu (Spot c), maltodextrin-binding protein (Spot d), and PNGM (Spot e).

Distinct Proteomic Profiles of the V. parahaemolyticus Isolates
The 2D-GE and LC-MS/MS techniques were also used to analyze proteomics of the V. parahaemolyticus isolates recovered from the 12 species of aquatic products, and the obtained proteomic profiles are presented in Figure 3. The patterns yielded from three independent 2D-GE gels of each isolate were consistent (figures not shown). This analysis also revealed distinct proteomic profiles showing various visible differential protein spots (343-312) among the 12 V. parahaemolyticus isolates (Figures 3A-L). The V. parahaemolyticus CHN-N4-18 isolate recovered from P. viridis appeared to express the highest number of intracellular proteins (343) among all the isolates (Figure 3F). In contrast, approximately 312 intracellular proteins were observed from the proteomic profile of the CHN-B8-26 isolate from S. strictus ( Figure 3D). Notably, for the crustaceans origin, the CHN-N1-56 isolate from L. vannamei expressed much more intracellular proteins (342, Figure 3I) than the CHN-N2-5 isolate from O. oratoria (314, Figure 3J).

Identification of Differential Intracellular
Proteins of the V. parahaemolyticus Isolates Differential intracellular proteins of the 12 V. parahaemolyticus isolates were identified by LC-MS/MS analysis and summarized in Table 4. A total of 97 protein sequences were obtained and classified into three major GO categories in the GO       Figure S2A).

Proteomic Comparison of the V. parahaemolyticus Isolates From Three Kinds of Aquatic Products
The proteomic profiles of the shellfish, crustaceans, and fish origins were also different (Figure 3). Interestingly, approximately 28.9% (28/97) of the differential Intracellular proteins only appeared on the proteomic profiles of the V. parahaemolyticus isolates from the eight species of shellfish, but absent from the crustaceans and fish. Of these proteins, the glycerol dehydrogenase (GldA, Spot P12), 5 -nucleotidase surE (Spot P21), and putative glucose-6-phosphate 1-epimerase (G6PE, Spot P30) were shown only on the proteomic profile of the CHN-B2-28 isolate from R. philippinarum. The GldA (Spot P12) is required to catalyze the first step in fermentative glycerol metabolism, and its product is then funneled into the glycolytic pathway for further degradation. The 5 -nucleotidase surE (Spot P21) has nucleotidase activity and is involved in nucleotide metabolism in Escherichia coli (Proudfoot et al., 2004). The G6PE (Spot P30) catalyzes glucose-6-phosphate to fructose-6-phosphate that converted to fructose-1,6-bisphosphate under the catalysis of PFK (Spot P16). In this study, seven intracellular proteins were found only on the proteomic profile of the CHN-B5-29 isolate from P. magellanicus, including the periplasmic serine endoprotease DegP-like (Spot P33), GAPDH (Spot P40), oxidoreductase (Spot P42), threonine aldolase (Spot P43), azurin (Spot P47), s-ribosylhomocysteine lyase (LuxS, Spot P51), and tRNA-binding protein (Spot P52). Among these proteins, for instance, the oxidoreductase (Spot P42) facilitates the detoxification of xenobiotic organic compounds by various microorganisms (Khatoon et al., 2017). The threonine aldolase (Spot P43) catalyzes the cleavage of threonine into glycine and acetaldehyde and involves in threonine decomposition and glycine synthesis (Liu et al., 2015). Azurin (Spot P47) is necessary for bacterial protection from oxidative stress (electron donor to nitrate reductase) and copper toxicity in P. aeruginosa (Mohammadi-Barzelighi et al., 2019). The LuxS (Spot P51) is highly conserved among Aeromonas hydrophila, E. coli, V. cholerae, and Vibrio harveyi and has a critical role in regulation of genes associated with the transport of nucleotides, metabolism, and synthesis of cell walls or membranes. In this study, the cysteine desulfurase IscS (Spot P57) was observed only on the proteomic profile of the CHN-B6-62 isolate from S. constricta, which was involved in iron-sulfur cluster biogenesis and oxidative stress defense in Mycobacterium tuberculosis (Rybniker et al., 2014). Additionally, the CHN-N4-18 isolate from P. viridis expressed the outer membrane β-barrel domain protein (Spot P69), OmpW (Spot P70), and putative 4-hydroxy-4-methyl-2-oxoglutarate aldolase (Spot P71) ( Table 4).
Notably, among the differential intracellular proteins identified in this study, some were reported to be involved in the virulence of pathogenic bacteria, including the AdhE (Spot P1), OmpAs (Spots P15 and P78), AhpC (Spot P23), PtsA (Spot P36), and phosphoglycerate kinase (PGK) (Spot P95). The latter catalyzes the transfer of a phosphate group from 1,3-diphosphoglycerate to ADP to produce 3-phosphoglycerate and ATP (Smith et al., 2011).

Effects of Aquatic Product Matrix on Proteomic Profiles of the V. parahaemolyticus Isolates
To get insights into implications of aquatic product matrices on the proteomes of the 12 V. parahaemolyticus isolates, we incubated these isolates in their corresponding matrix media and identified a total of 71 differential intracellular proteins by the 2D-GE and LC-MS/MS analysis (Figures 4-6 and Table 5). These proteins were classified into three major GO categories, in which the most abundant GO term was catalytic activity (62.0%, 44/71), followed by cell (50.7%, 36/71) and binding (49.3%, 35/71). The opposite patterns were biological regulation (1.4%, 1/71), organelle (1.4%, 1/71), and structural molecule activity (1.4%, 1/71) (Supplementary Figure S2B).
Likewise, for the fish matrix media, for instance, approximately 329 and 333 intracellular protein spots were observed on the proteomic profiles derived from the CHN-Q5-1 isolate grown in the TSB and C. idellus matrix media, respectively (Figures 6L 1 ,L 2 ). The aspartate carbamoyltransferase (Spot L1-1) was expressed by the CHN-Q5-1 isolate grown in the former medium (Figure 6L 1 ), whereas four were produced in the C. idellus matrix medium (Figure 6L 2 ), including the 3-chlorobenzoate-3,4-dioxygenase dehydrogenase (Spot L2-1), OmpA family protein (Spot L2-2), GrpE (Spot L2-3), and 50S ribosomal protein L9 (Spot L2-4) ( Table 5). Additionally, to validate the differential proteins induced by the aquatic product matrices, we examined gene expression of several representative proteins by the quantitative RT-PCR assay. The resulting data were generally consistent with the proteomic analysis (Supplementary Figure S4).

Aquatic Product Matrix Composition
As shown in Supplementary Figure S5, the protein (3.73 -0.43 ), carbohydrate (1.23 -0.01 ), and crude fat (1.00 -0.01 ) contents of the 12 types of aquatic product matrices were remarkably different. The protein concentration in the        Figure S5).

DISCUSSION
Vibrio parahaemolyticus is the leading seafoodborne pathogenic bacterium worldwide. Nevertheless, the information in V. parahaemolyticus proteomics is minimally available to date (Fu et al., 2014;He et al., 2015;Perez-Acosta et al., 2018;Tang et al., 2018;Zhong et al., 2019). Based on our prior studies, in this study, we obtained and compared the secretomic and proteomic profiles of the V. parahaemolyticus isolates recovered from 12 species of commonly consumed aquatic products by 2D-GE and LC-MS/MS analysis. Secreted proteins have a major role in the pathogenesis of bacterial infection of host cells. He et al. (2015) investigated secretomic profiles derived from V. parahaemolyticus strains isolated from five species of aquatic products and identified six extracellular virulence-associated proteins involved in the pathogenicity of bacteria, such as EF-Tu, pyridoxine 5phosphate synthase, σ54 modulation protein, dihydrolipoyl dehydrogenase, transaldolase, and PGK. Among these proteins, the EF-Tu (Spot c) and transaldolase (Spot S5) were also identified by the comparative secretomic analysis of the 12 V. parahaemolyticus isolates in this study. The EF-Tu (Spot c) secreted by all the V. parahaemolyticus isolates was reported to be exposed on the cell surface of Streptococcus, Neisseria, and Mycoplasma and under certain conditions on E. coli (Mikhalchik et al., 2019). The transaldolase (Spot S5) was secreted by the V. parahaemolyticus CHN-B2-28, CHN-B6-62, CHN-B8-26, CHN-N4-18, CHN-N10-18, CHN-L7-40, and CHN-Q5-1 isolates. Gonzalez-Rodriguez et al. reported that the transaldolase (Spot S5) recruited on the cell surface via a non-classical secretion mechanism or an uncharacterized translocation pathway and acted as an important colonization factor for the survival of Bifidobacterium bifidum in host intestinal tract (Gonzalez-Rodriguez et al., 2012). In this study, some other extracellular proteins involved in bacterial virulence were also identified. For instance, the superoxide dismutase (Spot S26) secreted by the V. parahaemolyticus CHN-B5-29 and CHN-B6-62 isolates is an important virulence factor of Vibrio alginolyticus and contributes to the antioxidative stress with potential application for live attenuated vaccine .
The initial contact and anchoring of bacteria to a host cell are essential during the process of infection . Adhesions are present at the bacterial cell surface or released into extracellular space to form a contact platform for bacterial attachment onto a host cell (Zhang and Orth, 2013). In this study, our secretomic data derived from the 12 V. parahaemolyticus isolates revealed such proteins involved in the adhesions of pathogenic bacteria. For example, the TolC (Spot a), secreted by all the V. parahaemolyticus isolates, is a major adhesin in V. harveyi (Zhu et al., 2019). The enolase (Spot S2), secreted by the V. parahaemolyticus CHN-B2-28, CHN-B5-29, CHN-B6-62, CHN-N1-56, CHN-N3-2, CHN-N4-18, CHN-N8-5, and CHN-Q5-1 isolates, is an adhesion-related factor that binds plasminogen and allows bacteria to acquire surface-associated proteolytic activity that facilitates invasion and dissemination in the infected host . Bacterial membrane proteins can act as adhesion factors or adhesion enhancers (Gordon et al., 2015). For example, the GAPDH (Spot S13), like many housekeeping proteins, has been presumed to exist only in the cytoplasm to involve in glycolysis. However, it has been reported that GAPDH can be recruited on the cell surface and secreted via a non-classical secretion mechanism, and therefore it is a suitable vaccine candidate for protection against bacterial and parasitic diseases (Perez-Casal and Potter, 2016). The nitrogen regulatory protein P-II (Spot S28) is one of the most widely distributed families of signal transduction proteins widespread among bacteria, archaea, and plants (Radchenko and Merrick, 2011) and control the activities of a very diverse range of enzymes, transcription factors and some membrane transport proteins by direct interaction with their target hosts (Merrick, 2014). The maltoporin (Spot b), secreted by all the V. parahaemolyticus isolates, belongs to the outer membrane porin family of Gram-negative bacteria (Thoma et al., 2017) and is a versatile vaccine candidate in Vibrio species (Lun et al., 2014). Recently, Yang B. et al. (2019) reported that the maltoporin (Spot b) also contributed to the adhesion and invasion ability of Aeromonas veronii TH0426 to epithelioma papulosum cyprini cells. Additionally, bacterial flagellins, potent immunomodulatory agents, contribute to bacterial adhesion and invasion of host cells as well (Hajam et al., 2017). In this study, the flagellin C (Spot S7) was secreted by the V. parahaemolyticus CHN-B2-28, CHN-B8-26, CHN-N4-18, CHN-L7-40, and CHN-Q5-1 isolates, whereas the polar flagellin B/D (Spot S8) was secreted by the CHN-B2-28 and CHN-B8-26 isolates. These identified extracellular proteins could be the main targets of vaccine development because of their exposed epitopes on the cell surface.
The increase in MDR pathogenic bacteria has raised a serious public health and economic concern (Elmahdi et al., 2016). One common mechanism for bacteria to obtain antibiotic resistance is to actively pump drugs from bacterial cells by employing ABC transporters (Wilson, 2016). The functions of ABC transporters are very diverse, ranging from importing essential nutrients to conferring drugs in bacteria, archaea, and eukaryote (Beis, 2015). In this study, comparative secretomic analysis revealed several ABC transporters of the V. parahaemolyticus isolates with resistance phenotypes. For instance, the maltodextrinbinding protein (Spot d), secreted by all the V. parahaemolyticus isolates, is part of the maltose ABC complex MalEFGK (Machtel et al., 2019). The D-ribose ABC TSBP (Spot S15) and arginine ABC TSBP (Spot S18) were secreted by most of the isolates, except the CHN-B6-62 and CHN-Q5-1, as well as CHN-B6-62 and CHN-N1-56 isolates, respectively. The choline ABC TSBP (Spot S14) was secreted by the CHN-B2-28, CHN-B5-29, CHN-N1-56, CHN-N2-5, and CHN-N8-5 isolates. The glycine/betaine ABC TSBP (Spot S12) was observed only on the secretomic profile derived from the CHN-B2-28 and CHN-B8-26 isolates, whereas the peptide ABC TPPBP (Spot S24) was observed from the CHN-B2-28 and CHN-N2-5 isolates. The other possible mechanism of bacterial resistance is the ribosome protection (Wilson, 2016). Bacterial ribosome, being one of the main antibiotic targets in bacterial cells (Wilson, 2014), is a large protein-RNA complex that consists of two major subunits (a small 30S subunit and a large 50S subunit), each of which is composed of a variety of proteins. In this study, two translation-associated proteins EF-Tu (Spot c) and EF-Ts (Spot S22) were identified by the secretomic and proteomic analysis. The former existed in all the V. parahaemolyticus isolates, whereas the latter was secreted by the CHN-B2-28, CHN-B8-26, CHN-N3-2, CHN-N4-18, CHN-N10-18, CHN-L7-40, and CHN-Q5-1 isolates. The EF-Ts is a guanosine nucleotide exchange factor for EF-Tu (He et al., 2015), and the EF-Tu catalyzes the binding of aminoacyl-tRNA (aa-tRNA) to a site of the ribosome during protein synthesis (Daviter et al., 2003). Agarwal and O'Connor (2014) reported that a ribosomal protein S12 binding the EF-Tu contributed to streptomycin resistance in E. coli MC323. Overall, these identified proteins may serve as an explanation for the resistance phenotypes of the V. parahaemolyticus isolates with aquatic product origins.
In this study, all the V. parahaemolyticus isolates (except the CHN-B6-62 isolate) produced more intracellular protein spots on their proteomic profiles responding to aquatic product matrices. A total of 71 differential intracellular proteins were identified by the LC-MS/MS analysis, most of which were involved in biosynthesis process (e.g., diaminopimelate, lysine, ADP-Lglycero-β-D-manno-heptose, S-adenosylmethionine, GMP, riboflavin, and fatty acid biosynthesis), metabolic processes (e.g., gluconeogenesis, glycerol, one-carbon, glycerol-3-phosphate, alcohol, carbohydrate, D-ribose, deoxyribonucleotide, and amino acid metabolism), and cell membrane composition. The growth of microorganism may vary with the available carbon and nitrogen sources (Chiang and Chou, 2008). In this study, our comparative proteomic analysis showed that the crustacean matrices changed the proteomes of the V. parahaemolyticus CHN-N1-56 and CHN-N2-5 isolates, recovered from L. vannamei and O. oratoria, respectively, more than the other 10 types of aquatic product matrices. Wang et al. (2018) evaluated the influence of food matrices (shrimp, oyster, freshwater fish, pork, chicken, and egg fried rice) on extracellular products of V. parahaemolyticus and found that V. parahaemolyticus expressed significantly higher activity (p < 0.05) of gelatinase, caseinase, urease, DNase, and amylase in shrimp matrix than freshwater fish. In this study, our data also showed that the crude fat contents of the crustacean matrices were approximately 101-to 85-fold higher than those of the fish and shellfish matrices. Lipases expressed by Vibrio species can hydrolyze fats into glycerol and fatty acids (Beshiru and Igbinosa, 2018). Moravec et al. (2017) reported the Vibrio's ability to acquire fatty acids from environmental sources. Exogenous fatty acids can affect bacterial metabolism, modification of membrane lipids, alteration of protein function, regulation of gene expression, and stress responses (Moravec et al., 2017;Yao and Rock, 2017). In this study, our comparative proteomic data highlighted the significance of monitoring V. parahaemolyticus contamination in fat-rich aquatic products in the future research.
In this study, the other interesting finding was that some V. parahaemolyticus isolates produced virulence-associated proteins when incubated only in aquatic product matrices media. For instance, the OmpA family proteins (Spots D2-1, D2-2, D2-3, H2-1, and L2-2) were expressed by the V. parahaemolyticus CHN-B8-26 and CHN-N10-18 isolates when grown in the S. strictus and O. gigas thunberg matrices media, respectively. Recent research has indicated that proteins and lipids can form complexes (called liprotides) that assist in folding of outer membrane proteins, for example, OmpA, especially optimal folding at pH 8-9 (Nedergaard Pedersen et al., 2018). Protein folding is the essential process by which a polypeptide chain acquires its functional, native 3D structure. In this study, our data suggested that the aquatic product matrices with higher contents of protein and fat may facilitate the expression of virulence-associated factors (e.g., the OmpA family proteins) in the V. parahaemolyticus isolates.
Although gel-based proteomes are laboring and timeconsuming, images of 2D-GE gels can be compared so as to quantify each protein spot from different samples, and these protein spots can subsequently be excised, sequenced, and identified with MS, especially LC-MS/MS (Lee et al., 2019). Each protein has multiple forms as a result of genetic variations, splicing, truncation, degradation, and posttranslational modifications; therefore, innovated proteomic technologies such as MS-based shotgun proteomics (Lee et al., 2018) should be employed to explore more virulence and resistance-associated factors in V. parahaemolyticus. Also, new proteins during culture handing/extraction processing should be prevented by adding a compound (e.g., chloramphenicol), and the function of the virulence and resistance-related proteins identified in the V. parahaemolyticus isolates should be further pursued by cell and animal infection mode analysis in the future research.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

AUTHOR CONTRIBUTIONS
ZZ, LY, PY, YW, XP, and LC participated in the design and discussion of the study. ZZ carried out the experiments. LY performed the qRT-PCR assay. ZZ and LC wrote the manuscript. All the authors read and approved the final version to be published.

FUNDING
This study was supported by the grants from the Shanghai Municipal Science and Technology Commission (No. 17050502200) and the National Natural Science Foundation of China (No. 31671946).

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb.2020. 01453/full#supplementary-material FIGURE S1 | The secretomic profiles of the V. parahaemolyticus ATCC33847 and ATCC17802 strains by the 2D-GE analysis. (A) ATCC33847. (B) ATCC17802. The protein spots marked with blue circles and the numbers in red were the same at similar locations on the secretomic profiles of the 12 V. parahaemolyticus isolates. pI, isoelectric point.

FIGURE S2 |
The GO functional classification of the differential intracellular proteins. (A) The differential intracellular proteins of 12 V. parahaemolyticus isolates. (B) The differential intracellular proteins of the 12 V. parahaemolyticus isolates incubated between in the TSB and aquatic product matrix media.