Epizootic ulcerative syndrome causes cutaneous dysbacteriosis in hybrid snakehead (Channa maculata♀ × Channa argus♂)

Sample collection and protocol

Specimens of hybrid snakehead with and without EUS infection were collected on April 12, 2016, from the pond of a hybrid snakehead aquaculture farm (113°07′53″E, 22°45′07″N; pond size 5,000 m² with a water depth of 1.8 m, water temperature 19.1 ± 2.9 °C, pH 6.60 ± 0.18) located in Junan Town, Foshan City, southern China. The hybrid snakeheads were naturally infected by Aphanomyces sp. and approximately 70% of individuals in the sampling pond were infected. Fish showing normal swimming and feeding behavior were also collected from the same pond. A total of 10 fish of each type were collected, and their body weight, body length, and area of ulceration were measured prior to collecting cutaneous microbiota samples. The areas of ulceration were measured using a two-scale grid-based method (Zhou et al., 2016; Jiang et al., 2018). For microbiota analysis, five fish of each type were sampled and analyzed. Sections of fish skin measuring 2 × 2 cm were collected from the ulceration caused by Aphanomyces infection, located on the right abdomen (near the fish tail) of the infected fish, under sterile conditions after washing three time using sterile distilled water. Skin samples were also collected from the same position in healthy control fish under sterile conditions. The skin samples were stored in five mL sterile centrifuge tubes at 4 °C and transferred to the laboratory for DNA extraction. We also collected two further fish skin samples from the areas of ulceration caused by Aphanomyces infection to identify fungal pathogens. In addition, five pond water and five sediment samples were collected from the same pond using a five-point sampling method for identification of the environmental microorganisms associated with the hybrid snakehead habitat. The pond water samples were collected at a depth of 50 cm using a one L water sampler and stored in one L sterile sampling bottles. The water temperature and pH were determined at each sampling site using a portable physicochemical analyzer (Hach, Loveland, CO, USA). Sediment samples were collected from the top zero to five cm layer of the sediment surface. All samples were transferred to the laboratory at 4 °C.

The experimental protocols used in this study were approved by the Animal Ethics Committee of the Guangdong Provincial Zoological Society, China, under permit number GSZ-AW002.

Histopathological analysis

Three specimens of each type of hybrid snakehead with and without EUS were anesthetized using 80 mg/L MS-222, and from these fish, 2 × 2 cm samples of the epidermis and muscle (to a depth of approximately one cm) were collected from the right abdomen. These samples were thereafter cut into pieces of approximately 0.5 cm³ and washed using cold (4 °C) deionized water to remove the mucus. Subsequently, the samples were fixed using 10% formalin solution and then stored in Davison’s fixation solution. Paraffin tissue sectioning and hematoxylin-eosin staining were conducted according to previous reports (Cardiff, Miller & Munn, 2014; Huang et al., 2018). Photomicrographs of the stained sections were taken using a Nikon 80i microscopic imaging system.

Pathogen identification

Two hybrid snakeheads with EUS were used to confirm the presence of Aphanomyces. A small piece from the area of ulceration of hybrid snakehead was used for DNA extraction according to a method described by Fang et al. (2015) and Ni et al. (2017). A primer pair for the amplification of Aphanomyces DNA, F (5′-CTA GCC GAA GGT TTC GCA AGA-3′) and R (5′-GTT GTT TCC CAA TTT GCT TCC G-3′), was designed based on sequences of the Aphanomyces 18S rRNA gene in the GenBank database. The target fragment was 581 bp in length. Polymerase chain reactions (PCRs) were performed using 100 µL reaction mixes containing 1 × PCR buffer, one U of Taq DNA polymerase, 0.8 mM of each dNTP, 4.0 µM of each primer, and 40 ng microbial genomic DNA. The thermal cycling procedure consisted of an initial pre-denaturation step at 95 °C for 3 min, followed by 35 cycles of 95 °C for 30 s, 68 °C for 30 s, and 72 °C for 30 s, and a final extension at 72 °C for 5 min.

Genomic DNA of Aphanomyces invadans was used as positive control and sterile water was used as a negative control. Amplicons were checked using 1.5% agarose electrophoresis, and the length of target fragment was determined. The target fragment was retrieved from the gel using a gel extraction kit (Omega, Doraville, GA, USA) and was subsequently sequenced using an ABI 3700 DNA sequencer (ABI, Foster City, CA, USA).

Cutaneous microbiota enumeration

Five fish of each type were sampled and used to enumerate the cutaneous microbiota. The cutaneous microbiome was eluted from fish skin by placing skin samples into two mL of sterile PBST (140 mM NaCl, 10 mM phosphate, three mM KCl, 0.05% Tween-20, pH 7.4) in a sterile five mL conical tube, and mixing by vortexing for 2 min. Bacteria were collected from the PBST suspension by centrifugation at 5,000×g for 5 min. The DNA was extracted and purified using a genomic DNA clean & concentrator kit (Zymo, Irvine, CA, USA).

Polymerase chain reactions amplification was carried out using the prokaryotic primer set 515F (5′-GTG CCA GCM GCC GCG GTA A-3′) and 806R (5′-GGA CTA CHV GGG TWT CTA AT-3′) for the V4 region of the 16S rRNA gene, as we have described previously (Tamaki et al., 2011; Li et al., 2016, 2017; Ni et al., 2018). The resulting PCR products were pooled and purified using an Axygen gel extraction kit (Axygen, Union City, CA, USA). After purification, the PCR products of the 16S rRNA V4 region were quantified using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). A mixture of the amplicons was then used for sequencing using the Illumina MiSeq platform at Beijing Novogene Technology Co., Ltd, Beijing, China.

The sequencing data were analyzed as described previously (Ni et al., 2017; Huang et al., 2018; Li et al., 2018) with appropriate modifications. Briefly, after sequencing, the paired-end reads were overlapped to assemble the V4 tag sequences using Flash software (Magoč & Salzberg, 2011). The primers and spacers were trimmed using QIIME 1.9.0 (Caporaso et al., 2010). To minimize the effects of random sequencing error, both the low-quality fragments and sequences shorter than 240 bp were removed. Prior to further analysis, we checked for the presence of chimeras, which were filtered out using UCHIME software (Edgar et al., 2011). The sequences were classified into operational taxonomic units (OTU) by setting a 0.03 distance limit using the CD-HIT program (Li & Godzik, 2006). Taxonomic assignments of each OTU were determined using the RDP classifier (Wang et al., 2007).

Weighed and unweighted UniFrac distances between samples were calculated and principal co-ordinates analysis was conducted using QIIME 1.9.0. Principal component analysis and non-parametric multivariate analysis of variance (PERMANOVA) (Anderson, 2001) were used to reveal differences in the microbiota of the different groups and were conducted using the vegan package of R 3.5.1 (Dixon, 2003). A non-parametric Kruskal–Wallis test was used to detect significant differences among the different groups, and box plots were drawn to depict the relative abundances of significantly different OTUs among groups using STAMP software (Parks et al., 2014). Statistically significant markers were added to the box plots using Adobe Illustrator CS5 software according to the post hoc test results. A network of samples based on the dominant genus relative co-abundance between samples was constructed using the igraph and psych packages of R 3.5.1. Linear discriminant analysis effect size (LEfSe) was conducted to screen the significantly different genera between EUS and healthy controls using the Galaxy platform (Segata et al., 2011).

Sequencing data obtained in this study have been submitted to the NCBI Sequence Read Archive database with the accession number PRJNA495472.

Skin damage and the presence of Aphanomyces

Muscle underlying the damaged skin of hybrid snakehead with EUS became ulcerated and were colonised by numerous filamentous fungi (Figs. 1A and 1B). Control muscle shown for comparison (Fig. 1C). Numerous mycotic granulomas and large amounts of mycelia, which are typical characteristics of EUS, were detected in the ulcerated muscle of hybrid snakehead with EUS (Fig 1B). Ulceration in hybrid snakehead with EUS covered 12.2% ± 3.9% of skin area (Table 1). Infected fish were observed to be sluggish and lacking in vitality. Moreover, we found that the OD_630nm of skin suspensions obtained from fish with EUS was significantly higher than that of suspensions from healthy controls (independent t-test, t = 2.44, p = 0.046), indicating that a greater number of bacteria probably parasitize the skin of fish with EUS. PCR results showed that target DNA fragments (approximately 580 bp) could be amplified from the samples collected from the ulcerated regions of hybrid snakehead with UES (Fig. 1D). Sequencing results for the amplified DNA fragments showed that the fragments were derived from Aphanomyces species, indicating that Aphanomyces was the causal pathogen of EUS in the hybrid snakehead we analyzed.

Changes in the cutaneous microbiota of hybrid snakehead with EUS

Following the removal of low-quality reads and chimeric sequences, we obtained a total of 999,539 sequences from the 20 samples used for analysis (five cutaneous microbiota samples from hybrid snakehead with EUS, five cutaneous microbiota samples from healthy hybrid snakehead, five pond water microbiota samples, and five pond sediment microbiota samples). To eliminate the influence of sequencing depth, 29,781 sequences of each sample were randomly re-sampled according to the amount of sample sequence to obtain the minimum effective sequences and used for further analysis. Although three archaeal and 56 bacterial phyla were detected from the 20 microbiota samples, we found that the microbiota primarily comprised 15 predominant phyla (their relative abundances were greater than 1% in at least one sample; Fig. 2A). Although the relative abundances of most of these dominant phyla differed significantly among the cutaneous microbiota of hybrid snakehead with EUS, and those of healthy hybrid snakehead, pond water microbiota, and sediment microbiota, only Actinobacteria and Firmicutes were significantly reduced in the cutaneous microbiota of hybrid snakehead with EUS compared with that of healthy hybrid snakehead (Fig. 2C, 2I). This could be attributable to the fact that the numbers of many unclassified bacteria were enhanced in the cutaneous microbiota of hybrid snakehead with EUS, representing up to 29.77% ± 14.92% of the analyzed sequences. In contrast, unclassified bacteria comprised only 6.33% ± 1.07% of sequences in the cutaneous microbiota of healthy fish (independent t-test, t = 5.303, p = 0.024).

In the present study, we detected a total of 7,108 OTUs classified into 969 genera, among which, 201 genera dominated the cutaneous microbiota and habitat microbiota. Streptococcaceae unidentified genus (24.48% ± 5.37%), genus Lactococcus (7.97 ± 1.46%), genus Clostridium (3.79% ± 0.37%), Comamonadaceae unidentified genus (3.70% ± 0.30%), and Betaproteobacteria unidentified genus (2.48% ± 1.19%) were the five most abundant genera in the cutaneous microbiota of healthy hybrid snakehead, whereas genus Clostridium (16.37% ± 6.71%), Peptostreptococcaceae unidentified genus (6.24% ± 7.25%), Comamonadaceae unidentified genus (2.73% ± 1.98%), JTB215 unidentified genus (2.69% ± 1.27%), and Betaproteobacteria unidentified genus (2.48 ± 2.52%) were the five most abundant genera in the cutaneous microbiota of hybrid snakehead with EUS. At both the OTU (PERMANOVA, F = 12.746, p < 0.001; Fig. 3A) and genus (PERMANOVA, F = 16.409, p < 0.001; Fig. 3B) levels, there were significant difference among the microbiota from the cutaneous microbiota of hybrid snakehead with EUS, cutaneous microbiota of healthy controls, pond water, and pond sediment (Figs. S1A and S1B). The unweighted UniFrac distances in the EUS group were significantly higher than those in the PW group (Wilcoxon test, p = 0.001; Fig. S1C; Table S2). The weighted UniFrac distances in the EUS group were significantly higher than those in the healthy group (Wilcoxon test, p < 0.001) and the PW group (Wilcoxon test, p = 0.001; Fig. S1D; Table S2). The weighted UniFrac distances between groups were significantly higher than those within groups (Fig. S1D; Table S2). The LEfSe results indicated that bacteria classified into the genera Anaerovorax, Anaerosinus, Dorea, and Clostridium were significantly enhanced in the cutaneous microbiota of hybrid snakehead with EUS, whereas those classified into the genera Arthrobacter, Dysgonomonas, Anoxybacillus, Bacillus, Solibacillus, Carnobacterium, Lactococcus, Streptococcus, Achromobacter, Polynucleobacter, Vogesella, and Pseudomonas were significantly reduced (Fig. 3C).

The dominant OTUs enhanced in the cutaneous microbiota of hybrid snakehead with EUS tended to be positively correlated with the dominant OTUs in the sediment, whereas dominant OTUs in the cutaneous microbiota of healthy hybrid snakehead tended to be positively correlated with those reduced in the cutaneous microbiota of hybrid snakehead with EUS or those in the sediment (Fig. 4). The dominant OTUs reduced in the cutaneous microbiota of hybrid snakehead with EUS tended to positively correlated with those in the pond water (Fig. 4). These results indicate that the dominant cutaneous microbiotas of hybrid snakehead with EUS were more closely related to the sediment microbiotas, whereas those of healthy hybrid snakehead were closer to the pond water microbiotas.