Abstract
Brucella inopinata strain BO1 and B. sp. strain BO2 isolated from human patients, respectively, are genetically different from classical Brucella species. We isolated bacteria of the genus Brucella from two species of wild-caught tropical frogs kept in the facilities in Japan: White's tree frog, which inhabits Oceania, and Denny's tree frog, which inhabits Southeast Asia. Phylogenetic analyses based on 16S rRNA and recA gene sequences and multilocus sequence analysis showed that two isolates of Brucella spp. showed significant similarity to BO1, BO2, and the isolates from other wild-caught frogs. These results suggest that a variety of frog species are susceptible to a novel clade of Brucella bacteria, including B. inopinata.
Introduction
B
B. inopinata strain BO1 has been isolated from a human patient with clinical symptoms, such as fever, hypotension, leukocytosis, and inflammation around breast implant, consistent with classical brucellosis (De et al. 2008). Brucella strain, BO2, isolated from a lung biopsy in a patient with pneumonia, was initially thought to be closely related to B. inopinata (Tiller et al. 2010b). However, further studies showed that BO2 was clearly distinct from BO1 (Eisenberg et al. 2012). The primary hosts of these novel Brucella strains have not yet been elucidated.
Recently, undescribed Brucella strains were isolated from frogs, such as African bullfrog (Pyxicephalus adspersus) (Eisenberg et al. 2012), big-eyed tree frog (Leptopelis vermiculatus) (Fischer et al. 2012) from Africa, and White's tree frog (Litoria caerulea) from Oceania (Whatmore et al. 2015). Phylogenetic analyses based on 16S rRNA and eight-locus multilocus sequence analysis (MLSA) revealed that these isolates were closely related to BO1 and BO2, rather than the Brucella sp. pathogens in marine mammals and livestock (Eisenberg et al. 2012, Fischer et al. 2012, Whatmore et al. 2015), suggesting that wild frogs may also be hosts for BO1 and BO2.
In the present study, we investigated Brucella spp. infection among wild-caught and commercially available imported wild-caught frogs in Japan.
Materials and Methods
Sample collection and preparation
Four Japanese common toads (Bufo japonicus), two White's tree frogs (L. caerulea), six false tomato frogs (Dyscophus guineti), and a Denny's tree frog (Polypedates dennysi) were used in the present study. All of the 13 frogs were wild caught. Tissue samples of the frogs were collected and maintained at two independent private facilities in Japan at the time of outbreaks of cutaneous mycosis by Veronaea botryosa among the frogs (Hosoya et al. 2015). All the false tomato frogs used in the present study contracted cutaneous mycosis. Two Japanese common toads and one false tomato frog were alive, but other frogs were found dead. Skin samples were not available, so other available tissue samples (Table 1) were crushed and suspended in PBS using a bead beater crusher (IEDA TRADING CO., Tokyo, Japan).
Number of positive/number of samples.
NS, no samples.
Detection of Brucella DNA
DNA was extracted from the tissue homogenates and isolated bacterial cells using a DNA Mini Kit (QIAGEN, Valencia, CA), according to the manufacturer's instructions.
Brucella DNA was detected by PCRs specific for BCSP31, which encodes a cell surface protein, and omp2a, omp2b, and omp31 genes, which encode outer membrane proteins (Imaoka et al. 2007).
Culture and isolation of Brucella spp.
Tissue homogenates were cultured on ATCC® medium 488 broth containing Brucella Selective Supplement SR0083 (Oxoid, Hampshire, United Kingdom) and incubated for 3 days at 37°C with 5% CO2 in a model BR-23FP shaking incubator (TAITEC Co., Tokyo, Japan). Then aliquots of the cultures were inoculated on brain heart infusion (BHI) agar plates, and the plates were incubated for 1 day at 37°C.
Phylogenetic analysis based on 16S rRNA and recA genes
The 16S rRNA and recA gene fragments were amplified by PCR (Shimoda et al. 2000, Tiller et al. 2010a). The nucleotide sequences of the PCR amplicons were determined using the Big Dye Terminator Cycle Sequencing Kit ver. 3.1 (Thermo Fisher Scientific, Waltham, MA). Phylogenetic analysis was performed using the MEGA6 package (Tamura et al. 2013) and neighbor-joining phylogenetic trees inferred from 1000 bootstrap replicates were constructed based on the sequences of 16S rRNA and recA genes using equivalent sequences of classic, marine, and novel Brucella species, including B. inopinata strains, rodent strains, and frog strains, to those of Ochrobactrum species, the nearest phylogenetic relative of the genus Brucella, as outgroup sequences (Supplementary Figs. S1 and S2; Supplementary Data are available online at
Multilocus sequence analysis
Nine discrete gene fragments for the MLSA were amplified from bacterial DNA by PCRs using nine primer sets specific for the gap, aroA, glk, dnaK, gyrB, trpE, cobQ, omp25, and int-hyp genes, respectively (Whatmore et al. 2007). The nucleotide sequences of the PCR amplicons were determined using the Big Dye Terminator Cycle Sequencing Kit ver. 3.1. The nucleotide sequences of the PCR amplicons of the nine distinct gene fragments of the isolated bacteria were compared with those of other Brucella strains, including 27 distinct sequence types (STs) (Whatmore et al. 2007), BO1 strain, BO2 strain, rodent strains (strain 83/13, NF2653, and B. microti), and African bullfrog strains (09RB8471, 10RB9215) (Eisenberg et al. 2012). Since the nucleotide sequences of the int-hyp gene fragment of the two African bullfrog strains, 09RB8471 and 10RB9215, concatenated sequences data for the eight genes, except for the int-hyp gene, were aligned and neighbor-joining phylogenetic trees were constructed using the MEGA6 package (Fig. 1). Since the nucleotide sequences of these gene fragments of big-eyed tree frog strain 152 (Fischer et al. 2012) and White's tree frog strain UK8/14 (Whatmore et al. 2015) were not available, the two strains were not included for MLSA.

Phylogenetic tree based on the sequence data for the eight concatenated loci. The phylogenetic tree was inferred using the neighbor-joining method described in the MEGA6 package. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the p-distance method and are in the units of the number of base differences per site. Among the nine loci sequence data for MLSA, eight concatenated loci sequences with 3666 bp were subjected to the analysis. The classical Brucella species, B. abortus, B. melitensis, B. suis, B. ovis, B. canis, and B. neotomae, and novel species, B. microti and B. papionis were clustered within the same clade, whereas the two frog isolates in the present study (A105, A141) formed a distinct clade with B. inopinata, the African bullfrog isolates [09RB8471, 10RB9215 (Eisenberg et al. 2012)] and two rodent isolates [NF2653, 83/13 (Tiller et al. 2010a)].
Biochemical properties
The biochemical properties of the isolated bacteria were analyzed using the API 20NE (API Laboratory Products, Hampshire, GB) and BACTOLABO oxidase (Wako Pure Chemical Industries, Osaka, Japan) tests, in accordance with manufacturers' instructions and compared with those of the classical Brucella species and Ochrobactrum species. The included tests are listed in Table 2. The Gram status was determined using the Nissui Gram Stain Kit (Nissui Pharmaceutical, Tokyo, Japan), according to the manufacturer's instructions.
Isolate A105+, isolate A141−.
+, positive; −, negative.
Nucleotide sequence accession numbers
The DDBJ/GenBank accession numbers for the sequences of 16S rRNA, recA gap, aroA, glk, dnaK, gyrB, trpE, cobQ, omp25, and int-hyp genes are LC007087, LC186097, LC008000, LC007994, LC007089, LC007998, LC008002, LC008008, LC007996, LC008006, LC008004 for A105 and, LC007088, LC186098, LC008001, LC007995, LC007090, LC007999, LC008003, LC008009, LC007997, LC008007, and LC008005 for A141, respectively.
Ethics statement
The experiments with animals were performed in strict accordance with the Animal Experimentation Guidelines of the National Institute of Infectious Diseases. The protocol was approved by the Institutional Animal Care and Use Committee of the institute (Permit number: 113155).
Results
Detection of Brucella DNA and bacterial isolation
Recently, outbreaks of cutaneous mycosis by V. botryosa among the frogs in the two facilities were reported (Hosoya et al. 2015). All the frogs used in the present study were from the two facilities associated with the outbreaks. False tomato frogs tested in the present study were diagnosed as systemic chromomycosis (Hosoya et al. 2015). Other frogs used in the present study were not diagnosed as chromomycosis since the lesions were not observed. In the present study, we attempted to detect and to isolate Brucella species from the frog specimens and detection of other bacteria or parasites was not attempted. The tissue samples of White's tree frog and Denny's tree frog were tested for the presence of Brucella DNA by Brucella-specific PCR (Imaoka et al. 2007). Brucella DNAs, BCSP31, omp2a, omp2b, and omp31 genes were detected in bone marrow tissue of a White's tree frog and liver tissue of a Denny's tree frog (Table 1). Brucella bp26 gene was also detected in the two samples by PCR (Cloeckaert et al. 2000) (data not shown).
Brucella species were isolated from these Brucella DNA-positive tissues (Table 1). Both of the isolates, A105 from a White's tree frog and A141 from a Denny's tree frog, grew well on sheep blood agar and BHI agar with or without 5% CO2. The colonies of the two isolates were white to yellow-white in color and pinched-sized with a smooth appearance. The microscopic examination revealed two isolates were gram negative and coccobacilli shaped (data not shown). The isolates were tested positive by Brucella-specific PCR (data not shown).
Phylogenetic analysis based on 16S rRNA and recA genes
Pairwise comparison of the 16S rRNA gene sequences revealed BO1 to be the closest relative of the two frog isolates, with 100% and 99.9% nucleotide sequence identity to the A141 and A105 isolates, respectively; the sequence identity between the two frog isolates was 99.9%, whereas that between A105 and UK8/14 (Whatmore et al. 2015), both isolated from White's tree frogs, was 99.9%. Compared with that observed for the classical and marine Brucella species, the sequence identity was somewhat lower, at 99.2–99.5% for strain A105 and 99.3–99.6% for strain A141. The phylogenetic analysis based on the 16S rRNA sequences showed that the two frog isolates were closely related to BO1, BO2, and other four frog isolates, 4986/3, 5541/1 (Eisenberg et al. 2012), 152 (Fischer et al. 2012), and UK8/14 (Whatmore et al. 2015), and two rodent isolates, 83/13 and NF2653 (Cook et al. 1966, Tiller et al. 2010a), but relatively distantly related to the classical, marine Brucella species and the novel species of B. microti (Supplementary Fig. S1).
The recA gene has been shown to be highly conserved among the classical and marine Brucella species, however, unique variability in the nucleotide sequences of recA gene was observed among BO1, BO2, frog isolates, A141, A105, 4983/3, 152, rodent isolate, NF2653 (Supplementary Fig. S2), as pointed out for BO1 and BO2 (Tiller et al. 2010b). Nearly all nucleotide substitutions were synonymous so that amino acid identities among Brucella species were nearly 100% (data not shown).
Multilocus sequence analysis
The results of the MLSA are summarized in Supplementary Table S1. The concatenated sequences of the two frog isolates, African bullfrog isolates, B. inopinata and rodent isolates, did not fit any of the original 27 STs in the MLSA analysis (Whatmore et al. 2007), whereas the frog isolates, rodent isolates, and B. inopinata were shown to be in close relationship with each other based on the MLSA analysis.
To further assess the relationships of the frog isolates with the other Brucella species, concatenated 3666 bp-sequence data for eight loci evaluated using MLSA, not including the int-hyp locus, were employed for the phylogenetic analysis. The results further confirmed the close relationships of the two frog isolates to BO1, BO2, two rodent isolates (83/13 and NF2653), and the African bullfrog isolates (09RB8471, 10RB9215) (Fig. 1). The individual phylogenetic trees based on the sequence data for each of the nine loci assessed in the MLSA showed similar relationships for the Brucella species (data not shown).
Biochemical properties
The differentiating biochemical properties of the frog isolates, A105 and A141, are summarized in Table 2. The results using API 20NE and oxidase test demonstrated that both isolates exhibited excellent metabolic capabilities with respect to the classical Brucella species, sharing performance of biological assimilation with Ochrobactrum anthropi. Actually, the biochemical properties identified both isolates as O. anthropi, even though the possibility of “Brucella spp.” was suggested by the commercial identification system.
Discussion
In the present study, Brucella DNA was detected and an isolate recovered from the bone marrow of a White's tree frog and the liver of a Denny's tree frog, respectively. The two frogs from which the bacteria were isolated were derived from independent private frog facilities. Furthermore, Brucella DNA was not detected in the other frog specimens in these facilities. Fischer et al. (2012) and Whatmore et al. (2015) isolated Brucella spp. from a subcutaneous abscess and skin lesions of frogs, respectively. In the present study, we did not examine the frog skin samples since the skin and some tissue specimens were used for investigation of Veronaea botryose (Hosoya et al. 2015). The two frogs from which the bacteria were isolated were not diagnosed as mycosis (Hosoya et al. 2015), and no subcutaneous abscess or skin lesions were observed, thus, the association between these bacteria and cutaneous mycosis in the frogs is not clear at present.
Although the two frog isolates were biochemically primarily identified to be O. anthropi as reported previously (Elsaghir and James 2003, Fischer et al. 2012). However, genetic analyses showed that the two isolates were classified in the genus Brucella (Fig. 1, Supplementary Figs. S1 and S2).
The MLSA analysis revealed that the frog isolates, A105 and A141, shared a closer ST with novel Brucella species, including African bullfrog strains, rodent strains (NF2653 and 83/13), BO1 and BO2 than with other Brucella species (Supplementary Table S1).
The phylogenetic analyses of the 16S rRNA gene, recA gene and MLSA, revealed that the A105 and A141 were closely related to BO1, BO2, rodent strains (NF2653 and 83/13), and the other four frog strains. These results indicated that Brucella species may be grouped into two major clades: clade A, composed of the classical and marine Brucella species with two novel species of B. microti and B. papionis, and clade B, composed of other Brucella species, including B. inopinata, B. strain BO2, rodent strains (NF2653 and 83/13), and the frog isolates A105, A141, 09RB8471, 10RB9215, 4986/3, 5541/1, UK8/14, 152 (Supplementary Fig. S1 and Fig. 1). The phylogenetic tree based on the recA gene nucleotide sequence showed more distantly related topology among Brucella species in clade B (Supplementary Fig. S2). However, as has been pointed out for BO1 and BO2 (Tiller et al. 2010b), a majority of the nucleotide substitutions was synonymous so that amino acid identities among these Brucella species were nearly 100% (data not shown). The Brucella species in clade B may also be divided into several group phylogenetic analyses, and based on 1681 single-copy protein families conserved over 17 Brucella species, showed the bacteria to be classified into three clades, the classic clade, the BO clade containing human isolates BO1and BO2, and the N8 clade containing the rodent strains 83/13 and NF2653 (Wattam et al. 2012). Based on the results of the phylogenetic analyses in the present study, it is difficult to clarify whether the frog isolates form an independent clade apart from the BO and N8 clades. Further genome sequence analyses are thus required to answer this question.
We assume that frogs are reservoir hosts of Brucella species in the clade B, even though we cannot rule out the possibility that these frogs were susceptible for a novel clade of Brucella bacteria and infected in a captive environment. Interestingly, BO2 was isolated in Australia (Tiller et al. 2010b), and the rodent isolate 83/13 and NF2653 were reported in Queensland, Australia (Cook et al. 1966, Tiller et al. 2010a). Furthermore, White's tree frog, from which the Brucella spp. UK8/14 and A105 were isolated, is also naturally distributed throughout Queensland (Cogger 2014). However, Denny's tree frog, from which A141 was isolated, inhabits countries in Southeastern Asia, such as China, Laos, Burma, and Vietnam, whereas African bullfrog and Big-eyed tree frog originate from Tanzania. These frog-derived isolates are equally related to each other genetically, therefore it is unlikely that the bacteria recently genetically evolved in, but rather coevolved with, frogs. Sequence analysis of more isolates from each frog species and phylogenetic analyses of more isolates from different frogs are needed to clarify this. Furthermore, replication of these bacteria in cells and organs of frogs are also necessary.
The tropical frog species are imported and sold year-round as household pets in Japan. At the moment, it is not clear whether the Brucella spp. isolated from a variety of frogs are pathogenic in human. However, B. inopinata BO1 and B. inopinata-like BO2, isolated from human patients, are closely related to the frog isolates. Brucella spp. derived from household frogs might present a risk of human infection, especially to animal keepers. Therefore, further studies are required to clarify the detailed genotypic and phenotypic characteristics and pathogenicity of frog-derived Brucella species.
Conclusions
B. inopinata strain BO1 and B. sp. strain BO2 isolated from human patients were shown to be genetically closely related to Brucella species identified from African, Oceanian, and Asian frogs. The present study indicates that a variety of frogs from different countries are the natural hosts of or susceptible to a novel clade of Brucella bacteria, including B. inopinata.
Footnotes
Acknowledgments
This study was supported in part by a grant-in-aid from the Ministry of Health, Labor, and Welfare of Japan (H25-shinkou-ippan-007, H25-shinkou-ippan-008, H26-shinkou-jitsuyouka-ippan-019) and the Research Program on Emerging and Re-emerging Infectious Diseases from Japan Agency for Medical Research and Development, AMED.
Author Disclosure Statement
No competing financial interests exist.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
