In June of 2013, samples were collected from AHPND-suspected shrimp farms in Guangxi, China. Hepatopancreas (HP) from diseased shrimp were aseptically disaggregated and streaked on thiosulfate citrate bile salts sucrose (TCBS) plates at 28°C for 12 h. After pure cultures were obtained, a partial 16S rRNA region was amplified with primers 16S_27F and 16S_1492R (Lane, 1991) (Table S1) and sequenced. Partial rpoD, rctB, and toxR genes were amplified and sequenced as described by Pascual et al. (2010) (Table S1). The concatenated sequences of 16S rRNA, rpoD, rctB, and toxR loci were aligned. Then the phylogenetic tree was constructed using neighbor-joining analysis with maximum composite likelihood model in MEGA 5 (Tempe, AZ, USA) with 1,000 bootstrap replications. The nucleotide sequences from strain Vp 2S01 have been submitted to the GenBank database under accession numbers MF621565 (16S rRNA), MF621566 (toxR), MF621567 (rpoD) and MF621568 (rctB). 16S rRNA, rpoD, rctB and toxR allele sequences of Vc 3S01 were described by our previously report (Dong et al., 2017a). All 16S rRNA, rpoD, rctB, and toxR allele sequences of reference strains were described by Pascual et al. (2010).

Bacterial isolates Vp 2S01 and Vc 3S01 were cultured overnight in Tryptic soy broth with 2% NaCl (TSB+) at 28°C. One milliliter (mL) of the broth culture was boiled for 10 min at 95°C, and the supernatant was obtained by centrifugation, diluted 10-fold with distilled water and used as the template for PCR. PCR was performed using primers VpPirA and VpPirB (Han et al., 2015) (Table S1) as previously described. Protein products were examined by SDS-PAGE and mass spectrometry as previously described (Laemmli, 1970; Wang et al., 2011).

Since the Ethical Principles and Guidelines for the Use of Animals of the National Research Council of China applies to vertebrates only, there is no official standard for invertebrates, we adapted its principles to shrimp.

Before the challenge test, ~1 g healthy white shrimp (L. vannamei) were acclimated in the laboratory for 3 days in 50 L seawater at salinity 30 with constant aeration in plastic tanks (density 20 shrimp/tank) at 27 ± 2°C. For the immersion challenge, 20130629002S01 and 20130629003S01 were cultured in TSB+ at 28°C until the OD₆₀₀ reached 0.8–0.9 (approximately 8–12 h). Immersion challenge was performed following the immersion bioassay protocol described by Tran et al. (2013). Simply, the bacterial suspension was adjusted to 1 × 10⁸ cfu·mL⁻¹ (with TSB+), and 20 shrimp in each group were immersed in 4 L of this suspension for 15 min. The shrimp and bacterial suspension were then poured into tanks containing 50 L of seawater to give a final bacterial density of 1 × 10⁶ cfu·mL⁻¹. The control shrimp were immersed in TSB+ medium. The mortality of each group was recorded every 6 h. Moribund shrimp were fixed with Davidson's alcohol-formalin-acetic acid (DAFA) fixative for histopathological examination. All experiments were done in triplicate.

All shrimp sampled for histopathology purposes were fixed with DAFA for 24 h and stained with hematoxylin and eosin (H&E) using routine histological methods described by Lightner (1996). The histological sections were analyzed and photographed by a light microscopy system.

Genomic DNA isolation, sequencing, assembly and annotation were performed as previously described (Dong et al., 2017b). Antibiotic resistance genes were searched against the ResFinder (Zankari et al., 2012). DNA-uptake genes were identified by Blast against the counterparts reported in Vibrio cholera (Seitz and Blokesch, 2013).

The complete reference genome sequences of V. parahaemolyticus strains which downloaded from NCBI were used for comparative genome analysis. The reference V. parahaemolyticus strains were M0605 (JALL00000000), D4 (MYFH00000000), FIM-S1708+ (JPLV00000000), NCKU_CV_CHN (JPKU00000000), TUMSAT_DE1_S1 (BAVF00000000), TUMSAT_DE2_S2 (BAVG00000000), NCKU_TV_3HP (JPKS00000000), NCKU_TV_5HP (JPKT00000000), TUMSAT_D06_S3 (BAVH00000000), 1,335 (MYFF00000000), 12297B (MYFG00000000) and A3 (JOKE00000000). ANI value (Richter and Rossello-Mora, 2009) among genomes of Vp 2S01 and reference strains was calculated using the JSpecies program. All strains were cut into fragments of 1,020 bp for calculating the ANI values by using the BLAST algorithm (Goris et al., 2007). Next, a distance dendrogram was constructed using the R program.

The complete sequences of pirAB^vp-bearing plasmids downloaded from NCBI were subject to multiple alignment with pVPGX1 and pVCGX1. Those plasmids were pLA16-2 (accession no. CP021148) harbored by VC_AHPND strain LA16-V1, pVHvo (accession no. KX268305) harbored by VO_AHPND strain SH14, pVPA3-1 (accession no. KM067908) harbored by VP_AHPND strain 13-028A3, pVA1 (accession no. KP324996) harbored by VP_AHPND strain 3HP, pVPE61a (accession no. AP014860) harbored by VP_AHPND strain VPE61, pV110 (accession no. KY498540) harbored by VP_AHPND strain v110. Multiple plasmid sequence alignment was performed using Mauve (Darling et al., 2004).

Antibiotic susceptibility was determined by the disk diffusion test as described elsewhere (Roque et al., 2001). Briefly, strain suspensions (0.5 MacFarland) of 20130629002S01 and 20130629003S01 were inoculated by lawn onto marine agar and the antimicrobial sensitivity discs positioned. For the present study, antibiotics tested were ampicillin (AMP, 10 μg), aztreonam (ATM, 30 μg), bacitracin (BAC, 0.04 U), cefazolin (CFZ, 30 μg), ceftazidime (CAZ, 30 μg), ceftriaxone (CRO, 30 μg), cephalexin (LEX, 30 μg), cephradine (Rad, 30 μg), ciprofloxacin (CIP, 5 μg), florfenicol (FLO, 30 μg), imipenem (TPM, 10 μg), nitrofurantoin (NIT, 300 μg), norfloxacin (NOR, 10 μg), penicillin (PEN, 10 U), streptomycin (EST, 10 μg), sulfamethoxazole (SMZ, 300 μg), sulfazotrim (SUT, 25 μg), and tetracycline (TCY, 30 μg). The plates were incubated for 24 h at 37°C and the inhibition halos were measured (mm) with vernier caliper. Breakpoints were defined by the guideline (M45-A2, 2010) of Clinical and Laboratory Standards Institute (CLSI). Breakpoints of the antibiotics not listed in the guideline were defined as described by Zhang et al. (2012).

All statistical analyses were performed with SPSS, version 17.0 (SPSS Inc., Chicago, IL, USA). Cumulative mortality was compared by using One-Way ANOVA test. A P-value less than 0.05 was considered statistically significant.

Complete genome sequences of 20130629002S01 (V. parahaemolyticus) have been deposited in GenBank under the accession CP020034-CP020037.

In June of 2013, an AHPND-suspected outbreak occurred in a shrimp farm in Guangxi. We have previously reported an AHPND-causing isolate 20130629003S01 (3S01 for short) from the farm, which has been identified as a pirAB^vp-positive V. campbellii (Dong et al., 2017a). In addition, we isolated another strain, 20130629002S01 (2S01 for short) from the same batch of diseased Litopenaeus vannamei in a same pond of the farm. We found strain 2S01 was a pirAB^vp-positive V. parahaemolyticus (Figure 1). Thus, for short, we used “Vp 2S01” for strain 2S01 and “Vc 3S01” for strain 3S01 in this paper.

Shrimp immersed with Vp 2S01 or Vc 3S01 suspension were observed to develop typical gross signs of AHPND within 6 h. AHPND shrimp from the Vp 2S01- and Vc 3S01-infected groups had a pale, atrophied HP, and an empty stomach (ST) and midgut (MG), compared to the normal ones that had a normal size HP with dark orange color and a full ST and midgut MG. The survival pattern of different groups was illustrated in Figure 2A. The control group showed no mortality whereas Vp 2S01- and Vc 3S01-infected groups showed 100% mortality within 24 h. As shown in Figure 2B, Vp 2S01 and Vc 3S01 had similar pathogenicity (P > 0.05). Histopathological examination of moribund shrimp samples from Vp 2S01- and Vc 3S01-infected shrimp were used to confirm AHPND, which also revealed similar presence of AHPND lesions in HP (Figure S1).

The genome of Vc 3S01 has been previously sequenced (Dong et al., 2017b). Genome sequences of Vp 2S01 were determined here using the PacBio RS II sequencing platforms with 189× coverage. Vp 2S01 contains two circular chromosomes (I and II) and two circular plasmids (pVPGX1 and pVPGX2), while Vc 3S01 contains two circular chromosomes (I and II) and four plasmids (pVCGX1 to pVCGX4) (Dong et al., 2017b). Plasmids pVPGX1 and pVCGX1 are homologous to V. paraheamolyticus plasmids pVA1, pVPA3-1 with overall 99% nucleotide identities and partial pFORC4 (accession no. CP009849) with a 94% identity.

Significantly, genes related to DNA transfer were found in both strains. Conjugative transfer were identified on plasmids pVPGX1, pVCGX1, pVCGX3 and pVCGX4 and the chromosome I of Vc 3S01. Plasmids pVPGX1 and pVCGX1 carry tra-trb genes that are highly similar to the counterparts on pVA1 (Lee et al., 2015). The plasmids pVCGX3, pVCGX4 and the chromosome I of Vc 3S01 contain virB/D clusters (Table S2). The ones on pVCGX3 and the chromosome I are near identical, but distinct to the one on pVCGX4. The virB/D genes encoding a type IV secretion system were well known on the Agrobacterium tumefaciens plasmid Ti as being responsible for effector translocation and DNA conjugation (Shirasu et al., 1990; Christie, 2004). In addition, many other homologous virB/D clusters were characterized to be self-transmissible modules of conjugative plasmids and integrative and conjugative elements (ICEs) (Bi et al., 2013). Moreover, intact sets of genes related to DNA uptake were identified on the chromosomes I of both strains (Table 1). Their products display 35.7–95.4% amino acid identities to the counterparts of V. cholerae O1 biovar El Tor str. N16961, which are also encoded by chromosome I. The DNA-uptake genes in V. cholerae N16961 has been proved to be required for efficient natural transformation (Seitz and Blokesch, 2013).

Antibiotic resistance genes were found in both strains, which will be described later. Neither Vp 2S01 nor Vc 3S01 contain genes (scrB) that encode the sucrose hydrolase. Both Vp 2S01 and Vc 3S01 formed green and round colonies on TCBS plate. However, genes encoding the sucrose hydrolase were found in the genome sequence of AHPND-causing strain SH14 of V. owensii (Liu et al., 2015).

The pirAB^vp genes located on plasmids pVPGX1 in Vp 2S01 and pVCGX1 in Vc 3S01. The pVPGX1 and pVCGX1 displayed a 99% nucleotide identity spanning the entire sequences, while each of the two plasmids displayed a 96–100% nucleotide identity to pVA1 (Figure 4A). Further comparative analysis found that the two plasmids shared high homologies with other pirAB^vp-bearing plasmids from AHPND-causing Vibrio strains (Figure S2). Importantly, pVPGX1 and pVCGX1 shared the closest relationship among all the compared plasmids from AHPND-causing Vibrio (Figure 3). More interestingly, different versions of the genetic contexts of pirAB^vp (pirAB^vp contexts) were observed in all plasmids when compared (Figure S2).

In pVA1, the pirAB^vp genes locate within a 5.5-kb fragment ended with two inversely oriented copies of ISVal1 which belonged to the IS903 group of the IS5 family (length includes the two IS elements). The pirAB^vp were located among some small ORFs between the two ISVal1. ISVal1 was 1,054 bp in length and demarcated by 18-bp perfect inverted repeats (5′-GGCTTTGTTGCGTAATTC-3′). It displayed a 92% nucleotide identity to ISVa2, which was initially found in V. anguillarum (Tolmasky and Crosa, 1995). In pVPGX1, an inversion of the 5.5-kb pirAB^vp fragment was seen. Transformation of two configurations was likely due to recombination between two opposite ISVal1 (Partridge, 2011). In contrast, the pirAB^vp contexts in pVCGX1 had a more complex structure. The pirAB^vp fragment had an extra copy of ISVal1, which divided the fragment into two parts. The left part was inversely syntenic to pVA1, while the right part was directly syntenic to pVA1 (Figure 4B). In addition, an ISVal1 located downstream of trbN and a 217-bp fragment within the trb gene cluster in pVA1 were absent in pVPGX1 and pVCGX1 (Figure 4A).

We also analyzed the pirAB^vp contexts in 12 other sequenced AHPND-causing Vibrio genomes (Table S3). Novel variant was not detected using the current data except that the pirAB^vp fragment in V. parahaemolyticus M0605 likely located in a different site of a pVA1-type plasmid among them. The site was close to the insertion site of the solitary ISVal1 in pVA1. Note that since most of the reported genomes were unfinished, there is a possibility that there are undetected novel variants lying in the undetermined gaps.

Antibiotic resistance of Vp 2S01 and Vc 3S01 were detected. The resistance gene profiles of the two strains were also revealed. Vp 2S01 harbored two antibiotic resistance genes tet(35) and tet(34) on chromosome I, while Vc 3S01 contains tet(35) on chromosome I and sul2, strAB, tet(A), strB-2 (extra copy) and floR on pVCGX2. The results of disk diffusion test showed that Vp 2S01 was only susceptible to florfenicol. It was intermediate to ceftriaxone, nitrofurantoin, norfloxacin. Significantly, Vc 3S01 was resistant to a wider spectrum of antibiotics as being resistant or intermediate to all test antibiotics (Table S4).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.