Identification of Austwickia chelonae as cause of cutaneous granuloma in endangered crocodile lizards using metataxonomics

Ethics statement

All experimental animal procedures in this study were approved by the Committee on the Ethics of Animal Experiments of the Guangdong Institute of Applied Biological Resources (GIABR180928 and GIABR201027) and followed basic principles.

Animals and sampling

The sick crocodile lizards were found in the Guangdong Luokeng S. crocodilurus National Nature Reserve (24°31′14″N, 113°20′18″E). All of the lizards were adults or sub-adults. They were raised in ecological simulation pools. Some of them were rescued from the local wild. The others were born and raised in captivity, and their mother or grandmother was rescued from the local wild. The Nature Reserve raised these lizards to reintroduce them to the wild.

Lesions that formed one or more nodules in the skin of the crocodile lizard were resected, as shown in Fig. 1. The skin nodules were collected and immediately stored in liquid nitrogen. Some nodules were collected from dead animals, and the others were collected by biopsies.

A total of 16 crocodile lizards were sampled and 33 nodules were collected in this study (Table 1). Five nodules were used for histological examination. Three nodules were used for scanning electronic microscopy (SEM). Four nodules were used for bacteriological cultivation. Eleven nodules were used for high-throughput sequencing to analyze the bacterial and fungal components of the nodules. Six soil samples and two water samples from the crocodile lizard living environment were also collected for sequencing to trace the source of the pathogen. The information of sequenced samples was provided in Table 2.

Histology and SEM

Five skin nodules were fixed in 4% paraformaldehyde for structural observation.

For light microscopy, 2–3 µm paraffin sections were prepared and stained with hematoxylin-eosin (H&E, Hematoxylin and Eosin Staining Kit C0105; Beyotime Biotechnology, Jiangsu, China) and Grocott-Gomori’s methenamine silver (GMS, Grocott-Gomori’s Methenamine Silver Staining Kit M052; Shanghai Gefan Biotechnology Co., Ltd., Shanghai, China) stains. Slides for light microscopy were examined using EVOS^® FL Auto Cell Imaging System (Thermo Fisher Scientific Inc., Shanghai, China).

For SEM, samples were dehydrated in a graded ethyl alcohol series from 20% to 100%. Subsequently, the samples were dried in a CO₂ critical point dryer (Leica EM CPD300; Leica Microsystems Inc., Allendale, NJ, USA), mounted onto aluminum stubs, coated with platinum and examined under a Hitachi S-3400N SEM (Hitachi Ltd.; Japan).

Metataxonomic high-throughput sequencing and analysis

Total DNA was extracted from the skin nodules, water samples and soil samples using a PowerFecal^® DNA Isolation Kit (MOBIO Laboratories, Inc., Carlsbad, CA, USA) with 20 mg/ml lysozyme. For bacterial community analysis, the V4 hypervariable region of the 16S rRNA gene was amplified with the primers 515F (5′-GTGCCAGCMGCCGCGGTAA-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′). The amplicon library was prepared using TruSeq^® DNA PCR-Free Sample Preparation Kit for Illumina (Illumina, Inc.; USA). Sequencing on an Illumina HiSeq platform (250 bp paired-end reads) was performed by the Novogene Corporation (China). In addition, to determine whether the fungal infection was present, the fungal ITS1 sequence was amplified. The primers for ITS1 sequence amplification were ITS5-1737F (5′-GGAAGTAAAAGTCGTAACAAGG-3′) and ITS2-2043R (5′-GCTGCGTTCTTCATCGATGC-3′), and the expected size of the amplicon was 250–300 bp. 2% agarose gel electrophoresis was used to detect the amplicons.

After sequencing, raw tags were filtered using the QIIME package (Caporaso et al., 2010) to remove low-quality and chimeric sequences. Sequences with ≥97% similarity were assigned to the same operational taxonomic units (OTUs) using Uparse (Edgar, 2013). A representative sequence for each OTU was annotated using Mothur by searching the SILVA database (Threshold = 0.8) (Quast et al., 2013; Schloss et al., 2009). For comparisons between samples, the OTU abundances were normalized by the number of OTUs obtained from the sample with the lowest counts. Multiple sequence alignment and phylogenetic tree were conducted to show the general view of the sequenced bacteria based on top 100 genera using QIIME (Caporaso et al., 2010). Flower plots were drawn using R software to compare the similarities and differences in microbes among samples.

Molecular detection of ranavirus

Co-infection of A. chelonae with ranavirus was reported previously (Tamukai et al., 2016). Therefore, the presence of ranavirus in crocodile lizards was examined to confirm the pathogen. Three primer sets were used to detect the ranavirus specific major capsid protein (MCP) gene: RanaM68F (5′-GCACCACCTCTACTCTTATG-3′) and BIVMCP154 (5′-CCATCGAGCCGTTCATGATG-3′), RanaJP556F (5′-GGTTCTTCCCCTCCCATTC TTCTT-3′) and RanaJP772R (5′-GGTCATGTAGACGTTGGCCTCGAC-3′), OlT1 (5′-GACTTGGCCACTTATGAC-3′) and OlT2R (5′-GTCTCTGGAGAAGAAGAAT-3′). The expected sizes of the amplicons were 230 bp, 217 bp and 500 bp, respectively (Stöhr et al., 2013; Une et al., 2014). A total of 2% agarose gel electrophoresis was used to detect the amplicons. Samples were considered positive if two or three primer sets were positive (Tamukai et al., 2016).

Bacterial isolation and cultivation

Four granulomas were cut open and spread on Columbia blood agar base plates. The plates were placed at 30 °C for 24–48 h to cultivate the bacteria, which were then isolated and purified using repeated plate streaking. The isolated bacteria were incubated in Columbia medium. The DNA of each bacterium was extracted using the TIANamp Bacteria DNA Kit DP302 (Tiangen Biotech Co., Ltd., Beijing, China). The 16S rRNA gene was amplified with the universal primers 27Fs (5′-GAAGTCATCATGACCGTTCTGCAAGAGTTTGATCMTGGCTCAG-3′) and 1492Rs (5′-AGCAGGGTACGGATGTGCGAGCCTACGGHT ACCTTGTTACGACTT-3′), and sequenced with the primers 1S (5′-GAAGTCATCATGACC GTTCTGCA-3′) and 2RS (5′-AGCAGGGTACGGATGTGCGAGCC-3′). The bacteria were annotated by matching the similarity results in NCBI database using BlastN (https://blast.ncbi.nlm.nih.gov).

Artificial infection

Because of their endangered status and legality issues, crocodile lizards cannot be experimentally inoculated with bacteria to confirm the pathogen. Instead, another widely distributed lizard species, the Chinese skink (Plestiodon chinensis), was used to verify the pathogenicity of Austwickia chelonae and fulfill Koch’s postulates (Byrd & Segre, 2016).

All Chinese skinks were adults, captured from wild and temporarily housed in the 57 cm*42 cm *30 cm cages in the laboratory for at least one week before inoculation. The bottom of the cages was covered with grass brought back from wild or wood chips. Each cage contained three or four skinks. The cages were kept in secluded rooms to minimize human interference. The skinks were fed with 4–5 cm Zophobas morio larvae and autodrinker.

The candidate pathogen, A. chelonae LK16-18, was incubated with Columbia medium for 48 h at 30 °C and re-suspended in sterile phosphate buffer saline (PBS, pH =7.4) at a concentration of 2.67 ×10⁸ CFU/ml. Ten experimental Chinese skinks were randomly assigned into two groups: the treated group (N = 5) and the negative control group (N = 5). The average weight and snout-vent length of the experimental animals was 26.65 g and 9.57 cm, respectively. All of the Chinese skinks were clinically healthy, and no nodules or other lesions were found before inoculation. After local disinfection with 75% ethanol, the treated group were subcutaneously inoculated with A. chelonae suspension. The inoculation dose was 20 µl/g (volume of bacterial suspension: weight of animal): 200 µl/30g for the left side of the trunk and tail, 100 µl/30g for the left forelimb and hind limb. The right lateral skin was used to comparative observation. The negative control group were hypodermically inoculated with the same dose of sterile PBS. The disease signs development was observed every day after inoculation for one month or until the experimental Chinese skinks died. When the animals died, the nodules were collected and examined for the presence of A. chelonae. The detection methods included 16S rDNA sequencing using the primers 27Fs and 1492Rs, H&E staining and bacterial isolation as described above.

Moreover, isolated bacteria corresponding to the other three common OTUs in crocodile lizard nodules (Salmonella sp., Acinetobacter sp., Pseudomonas sp.) were also subcutaneously inoculated into Chinese skinks to confirm the pathogeny of A. chelonae. Twelve Chinese skinks were randomly assigned into four groups, subcutaneously inoculated with Salmonella enterica LK18-19, Acinetobacter sp. Exi5-53, Pseudomonas protegens Exi5-13, and PBS, respectively, at the same dose and with the same method mentioned above. The average weight and snout-vent length of these skinks was 36.95 g and 10.78 cm, respectively.

Antibiotic sensitivity test

A standard disk diffusion test was used to test the antibiotic sensitivity of the candidate pathogen A. chelonae LK16-18. The bacterial suspension was spread onto Columbia blood agar base plates, and disks containing antibiotics (Hangzhou Microbial Reagent Co., Ltd., Hangzhou, China) were sterilely placed on these plates. The plates were incubated at 30 °C for 48 h, and the disk diffusion zone diameters were recorded. The sensitivity categories were interpreted according to the breakpoints provided in manufactural instruction and the Clinical and Laboratory Standards Institute (CLSI) document M100 (28th Edition) (CLSI, 2018). The antibiotic information was listed in Table 3.

Sequencing data availability

All raw sequences obtained from high-throughput sequencing were deposited into the NCBI Sequence Read Archive (SRA) under the accession number SRP152217. The complete genome sequence of A. chelonae LK16-18 has been deposited into NCBI GenBank and published (Jiang et al., 2018). The 16S rRNA gene sequences of A. chelonae LK16-18, S. enterica LK18-19, Acinetobacter sp. Exi5-53 and P. protegens Exi5-13 have been deposited into NCBI NR database under the accession numbers MK110377, MK235186, MK235211 and MK235212.

General description of the disease

This disease was characterized by one or more nodules (Fig. 1). The clinical signs were located on the head, limbs and tail but not on the dorsal or ventral skin (Fig. 1). The most common nodule location was under the tongue (Fig. 1D). Usually, the nodules were located in the skin system. However, there are two exceptions. One lizard was found to have nodules in the stomach and intestinal tract. The other lizard was found to have nodules in the heart (Table 1).

Usually, one nodule was located inside one lesion. Two or more nodules were rarely found, but some crocodile lizards with two or more lesions were observed. The nodules were light yellow or white (Fig. 2). Some advanced nodules were caseous necrotic.

No ectoparasites were found on the surface of lesions.

Results of histology and SEM

Five nodules used for histological staining showed the same characteristics. Microscopic images at low magnification showed that the nodules were histologically caseous necrotic with a membrane outside the nodule and layer structure inside the nodule (Figs. 3A, 3C). A lot of filamentous bacteria were observed inside the nodule (Figs. 3B, 3D). The filamentous bacteria occupied the whole nodule. Some mycelia have penetrated into the membrane (Fig. 3D). Scanning electron microscopy also showed a large number of filamentous bacteria across the necrotic tissue in the nodule (Fig. 4).

Detection of bacteria

16S rDNA based metataxonomic high-throughput sequencing was used to determine the bacterial compositions of 19 samples, including 11 nodules from the skin of crocodile lizards and 8 environmental samples (Table 2). Each sample contained at least 31,635 effective sequences (Fig. S1). The rarefaction curves showed that these sequencing depths were sufficient for capturing microbiota in each sample, especially in the nodule samples (Fig. S2). A total of 54 phyla were sequenced in this study. The phylogenetic relationship of top 100 genera was showed in Fig. S3.

The flower plots showed that five OTUs were common to all nodule samples: Austwickia chelonae, Salmonella sp., Acinetobacter sp., Pseudomonas sp. and Halomonas sp. (Fig. 5). Relative abundance analysis showed that these five OTUs were predominant, with relative abundances of 29.0%, 8.7%, 1.0%, 0.3% and 0.2%, respectively, in the total dataset of nodules. In addition to the five OTUs mentioned above, the dominant bacteria in nodules also included Bacillus, Fusobacterium, Bacteroides, Chryseobacterium and Morganella (Fig. 6). However, these genera were not present in every nodule.

According to Koch’s postulates, the pathogen must occur in every case of the disease (Byrd & Segre, 2016). Therefore, the five common bacteria, A. chelonae, Salmonella sp., Acinetobacter sp., Pseudomonas sp. and Halomonas sp., were identified as the candidate pathogens underlying nodules in crocodile lizards. Of these five bacteria, only A. chelonae was filamentous, which was consistent with the morphology observed in H&E staining, GMS staining and SEM (Figs. 3–4), while the others were rod-shaped. These results suggested that A. chelonae was a candidate pathogen for this crocodile lizard disease.

Detection of fungi

For fungal infection analysis, only two out of 11 nodule samples showed a clear band, and three samples showed a weak band at the expected size of the ITS1 sequence (Fig. 7). In other words, not all of the nodules contained fungi. Moreover, no fungi were observed at GMS staining and SEM micrographs.

Detection of ranavirus

For primer set RanaM68F/ BIVMCP154, only samples L.LK.17 and L.LK.20 showed a weak band at the expected size of ranavirus MCP gene (Fig. 8A). For primer set RanaJP556F/ RanaJP 772R, no expected sized amplicon was found in all samples (Fig. 8B). For primer set OIT1/ OIT2R, only L.LK.13 showed a band at the expected size of ranavirus MCP gene (Fig. 8C). Therefore, no ranaviruses were considered positively. Besides, no viruses were observed under SEM.

Confirmation of pathogenicity of A. chelonae

To verify the pathogenicity of A. chelonae, we isolated and purified the bacteria from nodules resected from crocodile lizard skin using Columbia blood agar base plates, and a pure A. chelonae culture (strain LK16-18) was obtained. This bacterium was filamentous under light microscopy and SEM. Colonies of A. chelonae on Columbia blood agar base plates were beta hemolytic, rough, adherent, and white (Fig. 9).

Half a month after hypodermic inoculation with A. chelonae, growing nodules were found on the inoculated sites of the tested Chinese skinks, as expected (Fig. 10). As the nodules growing, the experimental Chinese skinks appeared to eat less or did not eat. In addition, they had difficulty in moving. Moreover, all of the Chinese skinks inoculated with A. chelonae died in 21–44 days. The 16S rDNA sequencing results showed that the bacteria in all nodules resected from dead Chinese skinks were A. chelonae. Furthermore, A. chelonae colonies were re-isolated from the Chinese skink nodules at Columbia blood agar base plates. Filamentous bacteria were observed at H&E staining micrographs of Chinese skink nodules. Necropsy of the dead Chinese skinks showed that small white pellets were found at the surface of visceral organs. The 16S rDNA sequencing results showed that these pellets were bacterial colonies of A. chelonae. These results indicated that A. chelonae had spread into viscera.

No nodules were found in the skin of experimental Chinese skinks that inoculated with S. enterica LK18-19, Acinetobacter sp. Exi5-53, P. protegens Exi5-13, and PBS. Two Chinese skinks died in 4–5 days after inoculated with P. protegens Exi5-13. The left limbs were red and swollen, but no nodules were observed at the inoculated sites and internal organs of these two dead animals. The other experimental Chinese skinks were still alive at the end of the experiment.

All the above results led to the conclusion that A. chelonae caused the nodules in crocodile lizards reported in this study.

Source of A. chelonae

To trace the source of the pathogen six soil samples and two water samples from the crocodile lizard living environment were sequenced. The results showed that A. chelonae was found in some soil and water samples with very low relative abundances (Fig. 6, Table S1).

Sensitivities of A. chelonae to antibiotics

Austwickia chelonae was tested against twelve antibiotics to find effective drugs to control this disease in crocodile lizards (Table 3). The most sensitive antibiotic was cephalothin, followed by ampicillin and minocycline. Furthermore, A. chelonae was also sensitive to levofloxacin, rifampicin, erythromycin, ciprofloxacin, and piperacillin. However, this bacterium was not sensitive to kanamycin, gentamicin, streptomycin or clarithromycin.