Changes in the fecal microbiome of the Yangtze finless porpoise during a short-term therapeutic treatment

2.1 Background of animals

In December 2017, the Freshwater Fisheries Research Center began inspecting the population of YFPs in the Xijiang River. During the inspection, small areas of skin on both the head and tail fin of four YFPs were found to have ulcerations, scratches which were attributed to coastal beach rubble approached during predation (Figure 1). After slowly using net to surround four YFPs from outside water to near shore, the animals were safely held and lifted up from water by a specially made stretcher. To protect these injuries from infection by pathogens, a health examination, combining the on-site external application of drug therapy and a feeding regiment, was performed immediately. All four animals were male, and we named them A, B, C and D. According to the body length and the formula derived by Zhang [14], we estimated the age of the four animals afterward (Table 1). The age of YFP A, B, C and D was 10, 4, 13 and 5 years old, respectively.

Figure 1

The typical appearance of YFP before and after treatment. (a) YFP before treatment and (b) YFP after treatment.

The whole procedure has two phases: on-site health examination and follow-up treatment. During the health examination, we measured animals’ body length and weight and also collected fecal samples from the intestine. Besides, a mixture of Chinese traditional herbs, Yunnan Baiyao (a famous Chinese patent medicine made of precious herbs, which can relieve blood stasis, promote blood circulation and ease pain and swelling [15]), and erythromycin ointment (the mixture of Yunnan Baiyao and erythromycin ointment in a ratio of 1 to 5) was applied to the wound site, and then drug coating was implemented for half an hour on a large suitable water-soaked sponge pad (Table 2). Then, the animals were released back to water in a purse seine (the water surface area of about 500 m²) temporarily for the follow-up treatment. The whole health examination was carried out by professional veterinarians from beginning to end, and the animals were unharmed and handled safely during the process.

The follow-up treatment commenced from 8th February to 5th March and lasted for 25 days. YFPs with abrasions were temporarily fed and observed by animal-care experts while living in a purse seine. Healthy and fresh crucians (bought from Anqing Yichuan Aquaculture Farm) were fed to the YFPs three times a day (8:30 am, 12:30 pm and 5:30 pm) after soaking in 5% salt water for 30 min to inhibit potential pathogens [16]. Each YFP was fed 1–2 kg every time. To prevent infection and promote wound healing, ciprofloxacin hydrochloride tablets (5 days) and multivitamin tablets (25 days) were placed in the mouth of bait fish to be fed to the YFPs on the appropriate days (Table 2).

After aforesaid temporary treatment, all animals were observed to swim normally and to eat steadily. We found that the ulceration of animals had improved remarkably and tended to heal. On the last day, after applying the mixed drugs and conducting the health examination one more time, the YFPs were released into open water.

1. Ethical approval: The research related to animal use has been complied with all the relevant national regulations and institutional policies for the care and use of animals. The medical examination and the relevant experiments performed in this study were approved by the Anqing Fisheries Bureau of China, the department responsible for the protected area. The research complies with the Aquatic Animal Protection Act promulgated in 1993.

2.2 Sample collection

We collected three kinds of samples during the whole process: feces (intestinal content), water and bait fish. The fecal samples and water samples were collected through the health examination before and after the treatment. The bait fish samples were collected during the follow-up treatment (Table 3).

First, during the non-harmful, non-invasive fecal collection, each YFP was kept on a large suitable water-soaked sponge pad and held in a steady position. The anus was cleaned with 70% ethanol. All instruments used in the experiment were sterilized with 75% alcohol and UV treatment, which is also illustrated in the method. Before inserting, we lubricated the head of the hose and cleaned the anal attachment with a sterile cotton swab. Then, the tube was inserted about 20 cm into the anus. The tube was then removed with the feces and cut into sections. The sections were put in a sterilized 2 mL Eppendorf tube and stored at −198°C (liquid nitrogen) until DNA extraction. There were eight fecal samples in total, which were divided into two groups; four representing the free-hunting (FH) group (A1, B1, C1 and D1) collected before the treatment and four representing the feeding treatment (FT) group (A2, B2, C2 and D2) collected after the treatment (Table 3).

Then, we collected one water sample at each health examination individually (W1 and W2): we took 150 mL of water inside purse seine into a sterile bottle and then passed the water through a 0.22 µm filter [17]. The filters were stored in 100% alcohol and were frozen at below −20°C until DNA extraction.

We also collected three bait fish samples during the follow-up treatment (F1, F2 and F3): bait fish was sampled by homogenizing the representative daily animal ration in a commercial food-grade blender, then a subsample of the homogenate was taken [18]. The homogenate was put in a sterilized 2 mL Eppendorf tube and stored at −198°C (liquid nitrogen) until DNA extraction.

2.3 DNA extraction

Total microbial genomic DNA samples were extracted using the DNeasy PowerSoil Kit (QIAGEN, Inc, Netherlands), following the manufacturer’s instructions and stored at −20°C for further analysis. The quantity and quality of extracted DNAs were measured using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and agarose gel electrophoresis, respectively.

2.4 16S rRNA Gene amplicon sequencing

PCR amplification of the bacterial 16S rRNA genes V3–V4 region was performed using universal primers (341F [ACTCCTACGGGAGGCAGCAG] and 806R [GGACTACHVGGGTWTCTAAT]) [19]. Sample-specific 7 bp barcodes were incorporated into the primers for multiplex sequencing. The PCR components contained 5 μL of Q5 reaction buffer (×5), 5 μL of Q5 High-Fidelity GC buffer (×5), 0.25 μL of Q5 High-Fidelity DNA Polymerase (5 U/μL), 2 μL (2.5 mM) of dNTPs, 1 μl (10 µM) of each forward and reverse primer, 2 μL of DNA template and 8.75 μL of ddH₂O [20]. Thermal cycling consisted of initial denaturation at 98°C for 2 min, followed by 25 cycles consisting of denaturation at 98°C for 15 s, annealing at 55°C for 30 s and extension at 72°C for 30 s, with a final extension for 5 min at 72°C. PCR amplicons were purified with Agencourt AMPure Beads (Beckman Coulter, Indianapolis, IN, USA) and quantified using the PicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA, USA). After the individual quantification step, amplicons were pooled in equal amounts, and paired-end 2 × 300 bp sequencing was performed using the Illlumina MiSeq platform with MiSeq Reagent Kit v3 at Shanghai Personal Biotechnology Co, Ltd (Shanghai, China). All sequences, generated by high-throughput sequencing, were submitted to NCBI Sequence Read Archive under the accession numbers SRR8517807–SRR8517819.

2.5 Sequence analysis

The Quantitative Insights Into Microbial Ecology (QIIME, v1.8.0) pipeline was employed to process the sequencing data, as previously described [21]. Briefly, raw sequencing reads with exact matches to the barcodes were assigned to respective samples and identified as valid sequences. The low-quality sequences were filtered using the following criteria [22,23]: sequences that had a length of <150 bp, sequences that had average Phred scores of <20, sequences that contained ambiguous bases and sequences that contained mononucleotide repeats of >8 bp. Paired-end reads were assembled using FLASH [24]. After chimera detection, the remaining high-quality sequences were clustered into operational taxonomic units (OTUs) at 97% sequence identity by UCLUST [25]. A representative sequence was selected from each OTU using default parameters. OTU taxonomic classification was conducted by BLAST searching the representative sequences set against the Greengenes Database [26] using the best hit [27]. An OTU table was further generated to record the abundance of each OTU in each sample and the taxonomy of these OTUs. OTUs containing <0.001% of total sequences across all samples were discarded. To minimize the difference of sequencing depth across samples, an averaged, rounded rarefied OTU table was generated by averaging 100 evenly resampled OTU subsets under the 90% of the minimum sequencing depth for further analysis.

The alpha-diversity (the Chao1 estimator, the ACE estimator, Shannon diversity index and the Simpson index) and beta-diversity (principal coordinate analysis [PCoA] based on UniFrac distance [28]) of the samples were calculated by analyzing the species abundance, richness and clustering of samples. Using the Mothur software with the Metastats statistical algorithm [29,30], a pairwise comparison test was performed on the sequence size difference between groups at the phylum and genus levels. The significance of differentiation in microbial structure among groups was assessed by analysis of similarities (ANOSIM) [31,32] using R package “vegan”. P < 0.05 was considered a statistically significant difference. The Circos species relationship analysis was performed using the OmicShare tools (http://www.omicshare.com/tools).

A custom pathogen database was constructed by Apprill et al. [33], which included bacteria that had been identified as both marine mammal pathogens and human bacterial pathogens as recognized by the American Biological Safety Association. By comparing the result with the database, we tried to find the existence of all potential pathogens within the fecal samples of YFPs. Each sequence of uncultured and unclassified bacterial species was compared against the National Center for Biotechnology Information (NCBI) nucleotide collection (non-redundant nucleotide database) using Blastn with default parameters (https://blast.ncbi.nlm.nih.gov/Blast.cgi) on 18 December 2018. Then, we downloaded the 16S-sequence data of the known pathogens’ standard strains from the GenBank® nucleic acid sequence database of NCBI. The evolutionary history of potential pathogen OTUs and downloaded sequences was inferred using the neighbor-joining method [34]. The evolutionary distances were computed using the Kimura 2-parameter method and evolutionary analyses were conducted in MEGA 6.0.6 [35,36].

3.1 General analysis of sequences obtained by high-throughput sequencing

To explore the fecal bacteria structure and composition of the four YFPs during the short-term therapeutic treatment, 13 samples were analyzed with Illumina MiSeq. After removing the low-quality reads, a total of 1,148,664 effective high-quality reads were clustered into 1,998 OTUs within the 97% sequence similarity threshold. The number of OTUs at different annotated taxonomic levels is listed in Table 4. The stability of the rarefaction curves indicated that the sampling was reasonably representative, and the sequencing depth was adequate to cover the general bacterial diversity (Figure A1).

3.2 Microbial community composition at the phylum level

The phylogenetic classification of sequences from all samples includes 22 different phyla or groups (Figure 2). In water samples, the dominant phyla were Cyanobacteria (varying from 40.0% to 54.8%), Proteobacteria (24.7–24.8%), Actinobacteria (6.9–12.9%), Bacteroidetes (6.8–11.5%), Verrucomicrobia (2.7–5.4%), Planctomycetes (0.7–3.5%) and Chlamydiae (0.2–1.2%), which represented 97.9–98.2% of the total reads. (The lower number in the bracket represents the least relative abundance in a single sample this phylum can account for, and the higher one indicates the most relative abundance in a single sample. The numbers referred to relative abundance in the bracket below represent the same.) In fish samples, the bacterial communities were dominated by Firmicutes (43.3–72.8%) and Proteobacteria (25.3–55.4%). These phyla were detected in high abundance and accounted for 98.1–98.7% of the total reads. The predominant phyla in the feces of YFPs were Firmicutes (20.9–96.1%), Proteobacteria (3.8–78.7%), Actinobacteria (0.1–35.0%), Fusobacteria (2.3–5.9%) and Tenericutes (0.8–17.1%), which covered 99.5–100.0% of the total reads.

Figure 2

Relative abundance of bacterial phyla in fish, water and intestines. FH group represents feces of animals in the FH group, and FT represents feces of animals under human intervention. F represents fish samples, and W represents water samples.

A three-dimensional weighted PCoA based on UniFrac distance was obtained to measure the discrepancies and distances among all the samples (Figure 3). Bacterial samples displayed three clear clusters referring to fish, water and feces. With the PCoA, we observed the qualitative difference between each sample of feces. Each sample in the FT group deviates from the original (FH group) to a different degree (Figure 3a). In addition, from the FH group to FT group, we also detected that the tendency of the feces was associated with the fish and water samples (Figure 3b). A clear distinction in the bacterial community structure of different groups was also revealed by ANOSIM (R = 0.7923, P = 0.001).

Figure 3

Weighted PCoA based on UniFrac distance in different groups. (a) Grouped by individual YFPs; (b) grouped by whether the animal obtained the therapeutic treatment. F represents fish samples and W for water samples.

3.3 Shared and unique bacteria in fecal microbiome reflect composition shifts during the therapeutic treatment

Figure 4 shows that the fecal microbiome changes at the phylum level throughout the therapeutic treatment. The relative abundance of Firmicutes and Actinobacteria was higher in the FT group than in the FH group, whereas Proteobacteria and Tenericutes were relatively less abundant in the FT group. Among the two YFP groups, the relative abundance of Firmicutes differed significantly (Metastats, P = 0.031182). At the phylum level, both FH and FT groups were composed of seven shared phyla (Figure 5a). At the genus level, the top 10 shared bacterial genera in these two groups were Unclassified_Clostridiaceae, Unclassified_Peptostreptococcaceae, Clostridium, Unclassified_Aeromonadaceae, Mycobacterium, Cetobacterium, Enterobacteriaceae, Unclassified_Clostridiales, Plesiomonas and Epulopiscium (Figure 5b). In particular, Mycobacterium (Actinobacteria) and Phormidium (Proteobacteria) showed significant discrepancies, with P-values of 0.019026 and 0.036923, respectively (Table A1). Unique bacteria also existed in each group. For example, Bacteroidetes and Tenericutes only existed in the FH group, while Cyanobacteria only appeared in the FT group. At the genus level, unclassified_Mycoplasmataceae, Edwardsiella, Acinetobacter, Sarcina and unclassified_Clostridia only existed in the FH group, and Erwinia only appeared in the FT group.

Figure 4

Changes in microbiome phyla of fecal samples throughout the therapeutic treatment: (a) Firmicutes, (b) Proteobacteria, (c) Actinobacteria and (d) Tenericutes.

Figure 5

The Circos diagram of YFP sample groups: (a) phylum level and (b) genus level. UC represents unclassified. FH group represents feces of animals in the FH group, and FT represents feces of animals under human intervention.

In addition, the fecal microbiomes varied individually among the four YFPs. For instance, YFP-A and YFP-C shared similarities in microbial composition, as well as discrepancies. At the phylum level, their fecal microbiome showed an increase in Actinobacteria and a reduction in Firmicutes after the treatment. At the genus level, both YFP groups displayed an upward trend of Mycobacterium (Actinobacteria). However, Peptostreptococcaceae (Firmicutes) decreased from 37.94% to 7.05% in YFP-A after the treatment, and it rose significantly from 7.50% to 47.76% in YFP-C. Likewise, Aeromonadaceae (Firmicutes) was slightly augmented, 7.31%, in YFP-A and subtly reduced, 7.98%, in YFP-C after the treatment.

3.4 Identification of potential pathogens

To ensure the health of these YFPs during the short-term holding period, we identified the potential pathogens in all collected fecal samples. Through comparison with the pathogen database [33], 13 potential pathogen genera were screened out (Mycobacterium, Propionibacterium, Bacillus, Streptococcus, Clostridium, Sphingomonas, Edwardsiella, Serratia, Acinetobacter, Pseudomonas, Stenotrophomonas, Helicobacter and Plesiomonas). Then, we further screened the bacteria at species level through the pathogen database and found that there were many uncultured and unclassified species of Mycobacterium. The phylogenetic trees revealed that Mycobacterium species do not contain pathogenicity (Figure A2). Furthermore, we identified 10 potential pathogen species: Clostridium bifermentans, Clostridium difficile, Clostridium botulinum, Clostridium colinum, Clostridium perfringens, Edwardsiella tarda, Helicobacter pylori, Plesiomonas shigelloides, Propionibacterium acnes and Serratia marcescens (Figure 6). As it can be seen from Figure 6, both C. perfringens and P. shigelloides can be detected to various degrees in the total samples. C. difficile and E. tarda only appeared in YFP-D, and E. tarda disappeared after treatment. C. colinum only appeared in YFP-A. H. pylori was detected here in our fecal samples, and similarly, Helicobacter spp. and Helicobacter cetorum had been previously detected in YFP feces in an aquarium [37]. The potential pathogens detected in fish samples are as follows: C. perfringens, P. acnes and S. marcescens. The aforementioned potential pathogens were also detected in water samples, and C. bifermentans was also detected in W2 samples.

Figure 6

Heatmap analysis of potential pathogen genera of all samples. The color of the bar represents the abundance of each bacteria species in all samples. The longitudinal clustering indicates the similarity of all species among different samples.

3.5 Microbial diversity increased during the therapeutic treatment

Multiple alpha diversity indices, including the Chao, ACE, Simpson and Shannon indices, were calculated to analyze the microbial community richness and evenness in the bacterial samples (Table 5 and Table A2). The fish and water samples were more diverse than the animal fecal samples. This indicates that the microbial diversity in the feces of the four YFPs was significantly higher after the treatment than before (paired-sample t test, P < 0.05).