Introduction

Streptococcus agalactiae, also designated group B streptococcus (GBS), has long been recognized as a mammalian pathogen and is an emerging concern in fish. GBS infections have been reported in many fish species, such as seabream Sparus auratus, silver seabream Sparus sarba, rainbow trout Oncorhynchus mykiss (Walbaum), channel catfish Ictalurus punctatus, hybrid striped bass Morone chrysops × M. saxatilis and tilapia Oreochromis spp. [1–5]. Tilapia is native to Africa and the Middle East. It is cultured worldwide nowadays, with total production of over 2 million tons per year. Tilapia is also an important freshwater cultured fish in China, with annual production of 1.2 million tons. Guangdong and Hainan Provinces in Southern China are two major tilapia-culturing areas. Guangdong Province alone produces over 500,000 tons of tilapia each year [6]. Strong disease resistance is considered one of the most important characteristics of tilapia, which helped it to become a worldwide culture species. However, in recent years Streptococcus infection of tilapia has occurred in various countries, including the USA, Israel, Kuwait, Brazil and Thailand [1, 4, 5, 7, 8], overturning the myth that tilapia is hardy and disease resistant. In fact, Streptococcus disease may become a major threat to the future of tilapia industry.

The first case of GBS-related tilapia disease in China was reported in Fujian Province, where it caused 20–30% fish mortality in limited areas [9]. However, in the summer of 2009 and 2010, disease outbreaks spread to tilapia farms in major cultivation areas of Southern China, especially in Guangdong and Hainan Provinces. Diseased fish showed typical clinical signs of streptococcal infection, including septicaemic infection, exophthalmia, corneal opacity and various swimming abnormalities (such as swimming on side and erratic surface or bottom swimming). Compared with earlier reported cases, the disease was more widespread and detrimental. High infectivity (>50%) and mortality rate (>95%) were observed in some farms. To effectively control the disease, it is essential to recognize the pathogens. Therefore, in the present study, we identified the pathogens through artificial infection experiments, biochemical assays, 16S rRNA analysis and sequencing of the GBS-specific gene CAMP-factor specific gene (cfb).

To date, 10 GBS serotypes (Ia, Ib and II–IX) have been characterized based on their capsular polysaccharide antigens (cps) [10]. Serotype identification is necessary for both epidemiological studies and vaccine preparations. However, there is a lack of GBS-specific serotype identification methods for animals, especially for fish. A large proportion of GBS isolated from animals are nonserotypable by conventional serotyping (CS) methods, presumably because the typing antisera were initially raised against GBS isolates from humans [11]. Molecular GBS serotyping techniques have been shown to possess potential for high discriminatory power and reproducibility. Therefore, in this study, we combined the multiplex PCR assay of GBS ACP genes (alpha C protein gene, a surface protein antigen gene) with PCR analysis of cps gene to identify the molecular serotype (MS) of the isolated strains. Multi-locus sequence typing (MLST) was also applied to further characterize the isolated pathogens. Data from this study will enrich our current knowledge in epidemiology and help the establishment of effective disease prevention methods, particularly in vaccine development.

Materials and methods

Fish, bacteria and vectors

Diseased Nile tilapia (Oreochromis niloticus) were collected from major tilapia cultivation areas in Guangdong and Hainan Provinces of China in 2009 and 2010. A total of ten farms in five districts of Guangdong Province and six farms in three districts of Hainan Province were investigated and sampled. The sampling was random from different fish farm each year (different farms for 2009 and 2010). Six to twenty diseased fish (body weight between 10 and 575 g) with typical clinical symptoms from each farm were collected. Diseased fish showed clinical signs of exophthalmia, corneal opacity, swimming abnormalities (such as swimming on side and erratic surface or bottom swimming), off-feeding, abdominal fluid (in some fish) and haemorrhagic ulcers inside the operculum and at the base of the pectoral. Tilapias used for artificial infection weighted 20–26 g and were collected from Gaoyao Aquatic Engineering Centre of the Pearl River Fishery Research Institute (PRFRI), where water temperature was maintained at 29–31°C. The standard bacterial strain of S. agalactiae (CVCC 1887) was provided by China Institute of Veterinary Drug Control. Escherichia coli DH5α was provided by the Biotech Laboratory of PRFRI.

The rapid ID 32 Strep test kit is a product of Biomerieux Company (Lyon, France). Lancefield analysis reagents, Brain heart infusion (BHI) cultivation media, blood agar plates and other reagents were purchased from Huankai Microorganism Reagent Company (Guangzhou, China).

Isolation of pathogenic strains and artificial infection

Tissue samples were directly and aseptically collected from the brain, liver and eyes of diseased tilapia and inoculated onto blood agar, BHI agar or nutrient agar plates. Plates were incubated at 28°C for 24 h before single colonies of the dominant morphotype were selected for further culturing. These further purified bacterial strains were used for subsequent artificial infection, physiobiochemical analyses and molecular identification. For artificial infection, cells cultured in nutrient broth for 24 h were collected and washed with 0.65% sterile saline before they were resuspended in saline to four different concentrations: 1 × 108 CFU/mL, 1 × 107 CFU/mL, 1 × 106 CFU/mL and 1 × 105 CFU/mL. About 0.2 mL bacterial suspension was intraperitoneally injected into each healthy tilapia in a group of 20. The control group was injected with equal amount of sterile saline. After injection, fish were monitored daily for 7 consecutive days. Symptoms and mortality rate were recorded each day. At the end of the experiment, diseased fish with apparent symptoms induced by regressive infection were collected for a second round of bacterial isolation and fish artificial infection using the same method as described above. Strains with strong virulence were selected for further morphological analysis and identification.

Morphological observation and physiobiochemical identification of the pathogen

Morphology of bacteria and bacterial colonies: Bacteria cultured at 28°C for 24 h on blood agar plates or in BHI liquid media were observed under a microscope after smearing on a slide, fixation and Gram staining. Following the identification methods by Evans et al. [12] and Bergey’s Manual of Systematic Bacteriology (9th ed) [13], a series of physiobiochemical tests were carried out (Table 1). Bacteria were finally identified using the Rapid ID 32 Strp identification system (BioMérieux, Lyon, France) in conjunction with other bacterial biochemical micro-tube tests.

Table 1 Results of the ID 32 Strep test and other biochemical assays of the isolated strains
Full size table

Molecular identification

Pure bacterial strains were inoculated to nutrient broth and incubated overnight at 28°C in a shaker incubator. Cultures were centrifuged at 4,000 rpm for 25 min the next day to collect bacterial pellets of ~1 mL. The genomic DNA of each isolate was then extracted using a bacterial genomic DNA extraction kit (Boer BioTechnologies, Guangzhou, China) following the manufacturer’s instructions. Primers for amplification of the full-length Streptococcus agalactiae 16S rRNA (PLS and PLA, Table 2) [14] were synthesized and used to generate PCR products of about 1,500 bp. After gel electrophoresis on a 1.0% agarose gel, PCR bands were recovered using the Omega gel extraction kit (Omega Guangzhou, China). Recovered DNA fragments were ligated to the pMD19T vector (TaKaRa, Dalian, Japan) and transformed to competent E. coli cells following standard protocols. Positive colonies were identified via PCR and restriction digestion. Subsequent sequencing of the inserts was performed by Shanghai Yingjun Biotechnology Co. Ltd. (China). The Vector NTI suite 9.0 software and BLAST program (http://blast.ncbi.nlm.nih.gov/blast) were then used for sequence homology analysis. The MEGA4 software and neighbour-joining (NJ) method were employed to construct the phylogenetic tree [15].

Table 2 Primers for identification and molecular typing of isolated strains
Full size table

To obtain the complete cfb sequence, primers (Table 2) were designed based on the conserved sequences upstream and downstream of the GBS cfb genes registered in GenBank (accession nos. NC_004116.1, NC_00743.1 and X72754.1). The estimated amplicon size is ~900 bp. Methods for cloning, sequencing and homology analysis were the same as above.

Molecular serotyping

Primers for multiplex PCR assay of the GBS ACP gene were the same as previously described (Table 2) [16–18]. The 20-μL PCR reactions contained 10 μL 2× ExTaq buffer (for GC-rich templates), 0.4 μL dNTP mixture (2.5 mM each), 0.4 μL each of 20 μM primers (forward primer: PAS; reverse primer: PACA, PribA, PepsA, PA2/3A or PA4A), 50 ng template DNA, 0.15 μL 5 U/μL ExTaq and ddH2O. The amplification conditions were as follows: initial denaturation at 96°C for 5 min; 32 cycles of denaturation for 30 s at 95°C, annealing for 45 s at 58°C and extension for 45 s at 72°C; and final extension for 10 min at 72°C. PCR products were separated by electrophoresis on a 1.2% agarose gel followed by 0.5 μg/mL ethidium bromide staining and ultraviolet (UV) transillumination. Bands of the right size were excised and extracted using the gel extraction kit (Omega Inc., USA). The purified DNA fragment was sub-cloned into the pMD18-T vector. Positive colonies were identified via PCR and restriction enzyme digestion.

Primers (cpsES1 and cpsGA1) for amplification of a 790-bp partial cps gene cluster (from the 3′ end of cpsE-cpsF to the 5′ end of cpsG) were the same as previously reported (Table 2) [11, 19] and synthesized by Shanghai Sangon Biotech Company (China). The PCR reaction mixture was the same as described above. The amplification was carried out with initial denaturation at 96°C for 5 min, followed by 32 cycles of denaturation (94°C for 30 s), annealing (63°C for 30 s) and extension (72°C for 1 min), and final extension at 72°C for 10 min. Purification of the PCR products, sub-cloning and positive clone identification followed the same protocols as described above.

Primers for the supplementary PCR based on the difference between the cpsH and cpsI genes to distinguish MS Ia from MS III were designed according to Kong et al. [19]. Sequences of the forward (IacpsHS1) and reverse (IIIcpsHS or cpsIA) primers are listed in Table 2. The same PCR program as for amplification of the 790-bp cps gene was used, except that the annealing temperature was changed to 53°C and the extension time was changed to 45 s. Sub-cloning and colony identification methods were the same as mentioned above.

Sequencing of the PCR inserts was performed by Invitrogen Co. Ltd. (Shanghai, China). Multiple-sequence alignments were performed with the DNA Vector NTI Suit 8.0 and by the BLAST program (http://www.ncbi.nlm.nih.gov). Methods for MS analysis followed Kong et al. [19] and Slotved et al. [10].

Multi-locus sequence typing

Primers for seven housekeeping genes of GBS were designed following instructions on the Streptococcus agalactiae MLST website (http://pubmlst.org/sagalactiae/) [20, 21]. These genes include alcohol dehydrogenase (adhP), phenylalanyl tRNA synthetase (pheS), glutamine transporter protein (atr), glutamine synthetase (glnA), serine dehydratase (sdhA), glucose kinase (glcK) and transketolase (tkt). Genomic DNA preparation was the same as described above. The PCR amplification conditions were as follows: initial denaturation at 95°C for 5 min; 32 cycles of 94°C for 35 s, 53°C for 30 s and 72°C for 55 s; and final extension at 72°C for 7 min. PCR amplicons were gel-purified and sequenced. Sequence types were determined using tools on the Streptococcus agalactiae MLST website following online instructions. Briefly, sequences were submitted to the profile database through a “multiple locus query” to generate an allelic profile, and then an “allelic profile query” was selected to find the best match of sequence type (ST) that was either identical or similar to the profile.

Results

Isolation of pathogenic strains and artificial infection

Tissue samples from diseased fish from different tilapia farms were each smeared onto 3 different agar plates for pathogen isolation. Individual colonies on all 3 types of agar plates appeared uniform for different samples. Compared with colonies on nutrient agar plates, colonies on blood agar plates and BHI agar plates grew faster with larger size. Dominant colony morphotypes were picked for further pure culture isolation. A total of 10 pure strains were obtained from diseased fish from each farm of different districts. These strains were used subsequently for artificial infection experiments by injection.

After artificial infection, acute death was observed in the experimental group injected with high concentrations (1 × 108 CFU/mL, 1 × 107 CFU/mL and 1 × 106 CFU/mL) of isolated bacteria. The first death occurred 2 days after infection, and within 4 days mortality reached 100%. When 1 × 105 CFU/mL bacteria was injected into the fish, symptoms and death appeared on the 5th day and the death rate was 60-80%. Artificially infected fish showed major symptoms similar to those of naturally diseased fish, such as bulging eyes, corneal opacity, abnormal swimming and spinning. When high-dose bacteria was used for infection, the tested fish did not show typical symptoms such as bulging eyes as these fish died rapidly, however liver bleeding was found during autopsy. Infection results using the second-round isolated strains from diseased fish caused by the first-round artificial infection were similar to results of the first-round artificial infection, while fish in the blank control group remained healthy. The artificial infection-induced symptoms were similar among different isolated strains. Therefore, these isolated strains were considered as the pathogenic source of the tilapia disease outbreaks in 2009 and 2010.

Morphological and biochemical analyses of the pathogenic strains

A total of 14 strains with strong virulence were selected for further morphological analysis and identification. Results showed that colonies from different tissues of diseased fish were uniform with similar morphology, colour and size. Pure isolates on the blood agar plate formed white round smooth mucoid raised colonies with entire edges. When examined under a microscope, bacterium appeared spherical or ovoid with diameter of 0.5–2.0 μm. In BHI liquid medium, the majority of the bacterial cells were doublets or in short chains. Cells were Gram positive and produced rings of β-haemolysis on the blood agar plate. All isolated strains had physiobiochemical characteristics similar to the standard strain, with the majority of ID32 Strep test reactions being identical except for 1 different reaction and a few questionable results. Results of the supplementary biochemical tests were the same among the standard strain and different isolates. The identification results suggested that all 14 isolated strains were Streptococcus agalactiae (ID >99.0% and 99.9%, respectively, for strains isolated in 2009 and 2010) (Table 1).

Molecular identification

The 16S rRNA gene sequence of each isolate was amplified using primers PLS and PLA. All PCR products showed unique bands of 1,471 bp (excluding the primer lengths) after gel electrophoresis. Sequencing results of the PCR products suggested that 16S rRNA genes of the 14 isolates were highly homologous (99.9–100%). The 16S rRNA gene sequences of two isolates (GDzl and HNwc) have been submitted to GenBank (accession nos. GU217535 and GU217531, respectively). BLAST analysis revealed that the 16S rRNA sequences of the isolates were highly homologous to that of the Streptococcus agalactiae strain registered at GenBank (≥99.9–100%), and less similar to those of Streptococcus iniae (95.1–96.9%) and Streptococcus dysgalactiae (97.5–97.6%). A phylogenetic tree was constructed based on the 16S rRNA gene sequences of the isolates and the GenBank Streptococcus entries, showing that Streptococcus agalactiae, Streptococcus iniae and Streptococcus pyogenes each formed a separate branch (data not shown).

The cfb gene of each isolate was amplified using primers PcfbS and PcfbA. PCR products of ~830 bp were clearly observed after gel electrophoresis. Sequencing results suggested that the cfb gene has an open reading frame (ORF) of 763 bp and that sequences of all isolates are identical (100%). The cfb gene sequence from the representative GDzl strain has been registered in GenBank (accession no. GU217532). BLAST analysis showed that the cloned cfb sequence is highly homologous (99.0%) to GenBank registered sequences from strains of S. agalactiae (CP000114.1, AL766855.1, AE009948.1, X72754.1 and EF694027.1), while less similar to those from S. uberis and S. pyogenes (66.1% and 63.6% similarity, respectively). Therefore, results from sequencing analysis of the 16S rRNA gene and cfb gene further indicated that the isolated strains were Streptococcus agalactiae.

Molecular serotyping

A single amplicon of 400 bp was produced through ACP multiplex PCR using the PAS forward primer and 5 different reverse primers for all 14 isolates, as well as the standard S. agalactiae strain (CVCC 1887). Sequencing results showed that all amplicons were identical to the sequences of GBS bca genes registered in GenBank (accession nos. M97256.1 and CP000114.1). The bca gene sequence of the representative GDzl strain has been submitted to GenBank with accession no. GU217536.

Amplification of the cps E–G gene with primers cpsES1 and cpsGA1 produced amplicons of the expected size (790 bp) for all tested isolates and the standard S. agalactiae strain (CVCC 1887). Sequences of all these strains isolated in 2009 and 2010 were highly similar. After comparison of the single-nucleotide polymorphisms (SNPs) in the 790-bp fragment with those of reported serotypes, the isolated strains were classified to MS Ia/III-3 and the standard strain was identified as MS ll/III-4, or VII (Table 3). Representative sequences from strains GDzl and HNwf have been submitted to GenBank with accession nos. GU217533 and GU217534, respectively.

Table 3 Serotype identification based on SNP sites within the partial cps sequence (from 3′ of cps E-cps F to 5′ end of cps Ga)
Full size table

Another supplementary PCR was conducted to distinguish MS Ia from MS III. Amplification of the cpsH–I region with primers IacpsHS1 and cpsIA produced amplicons of 354 bp for all fourteen isolates, suggesting that these strains belong to MS Ia. No amplicon was produced with primers IacpsHS1 and IIIcpsHS. Sequences of the fourteen isolates shared high similarities, and BLAST analysis showed that they were highly homologous to the cps genes of GBS MS Ia in GenBank (CP000114.1, AF332914, AF363060, AB028896.21 and AB017355.1), further confirming the PCR result.

Taking the three PCR results together, we conclude that the fish GBS strains isolated from different regions in China in 2009 and 2010 belong to GBS MS Ia.

Multi-locus sequence typing

Seven housekeeping genes from the isolated strains were cloned and sequenced. Sequences were assigned to ST-7 based on the Streptococcus agalactiae MLST web database (Table 4). The standard strain was classified to ST-19 with the corresponding capsular serotypes of III or II. Since no GBS capsular serotype in the database was found to correspond to ST-7, the fish GBS isolates possibly possess unique molecular features.

Table 4 MLST analysis of the isolated strains
Full size table

Discussion

In this study, we isolated, identified and further molecular-typed the pathogenic bacteria strain responsible for the tilapia disease outbreaks in main breeding areas of Southern China in summers of 2009 and 2010. This study paves the way for effective prevention measures for tilapia Streptococcus disease, such as appropriate drug selection and vaccine development. Diseased tilapia from each farm showed highly similar symptoms with the most typical symptoms being bulging eyes, bleeding in mouth and gill covers, swimming abnormalities and body spinning. Artificial infection of healthy tilapia using bacterial isolates from different fish farms, as well as second-round artificial infection using bacteria isolated from dying fish induced by the first-round artificial infection, both produced symptoms similar to those of naturally infected fish. Through API 32 Strep testing and supplementary physiobiochemical analysis, we preliminarily identified the isolated strains as Streptococcus agalactiae. Further sequence analysis of the 16S rRNA gene and cfb gene were in agreement with this conclusion. Due to the highly conserved sequence of 16S rRNA genes, 16S rRNA sequences are widely used for bacteria identification. This is suitable to determine the genetic relationship at genus level or higher hierarchies, however it is less sensitive to differences between closely related species [22]. In our study, the 16S rRNA sequence of each isolate was highly homologous to that of Streptococcus agalactiae registered in GenBank (99.8–99.9%), while less similar to those of Streptococcus dysgalactiae and Streptococcus iniae (albeit still at a level of 95.1–97.6%). The cfb gene is a specific gene for GBS which encodes an outer-membrane protein called CAMP factor or haemolysis-promoting factor cohemolysin. The encoded protein can work conjunctively with the β-toxin (nerve phospholipase: sphingomyelinase) secreted by S. aureus to promote haemolysis of sheep red blood cells and dissolution of membrane components, exhibiting a positive CAMP reaction [23]. Total haemolysis of CAMP factor is one of the important indices for clinical diagnosis and identification of Streptococcus agalactiae. The sequence of CAMP factor in GBS strains is highly conserved. CAMP-like factors were also found in group A, C, and G Streptococcus, however sequences of their coding genes shared relatively low homologies. Therefore, the coding gene of CAMP factor, cfb, can be used for molecular identification of GBS. In our study, the complete coding sequences of cfb of all isolated strains were highly homologous (99%) to that of S. agalactiae registered in GenBank, while the homologies were relatively low to those of Streptococcus uberis (SUU34322) and Streptococcus pyogenes. Therefore, these strains were further confirmed as GBS based on their cfb gene sequences. In short, combined physiobiochemical and molecular identification results as well as results of regression infections using isolated strains strongly suggested that the pathogenic bacteria in the tilapia disease outbreaks in main breeding areas of China in 2009 and 2010 was Streptococcus agalactiae.

Isolations of fish GBS have been reported in recent years, but serotype identification of GBS is under-reported despite its importance in epidemiological studies and effective vaccine preparation. To date, only one case of serotype Ib has been reported in fish through whole-cell protein electrophoretic analysis [24], and a few cases of serotypes Ia and III (III-4) were identified in Nile tilapia in Thailand via multiplex PCR-based reverse line blot hybridization (mPCR/RLB) [5]. In the present study, using three rounds of PCR targeting ACP and cps cluster genes, we identified the MS of isolated tilapia GBS as MS Ia.

ACP is one of the GBS virulence factors that interact with epithelial cells [16, 25, 26] and the prototype for a family of surface-expressed proteins [C alpha protein (bca), C alpha like proteins 2, 3 and 4 (alp2/3/4), Rib protein (rib) and the epsilon protein] [17, 18]. Only one member of the surface protein gene family can be present in any single isolate, with the exception of bca-containing isolates which can also contain rib [16]. MS methods based on GBS surface protein antigen genes have showed excellent specificity [16, 27], and it has been well demonstrated that serotypes Ia, Ib and II are associated with the bca, while serotype III-3 is associated with alp2 [17, 19]. In the present study, multiplex PCR produced one single amplicon for all isolates that corresponded to the bca gene, thereby assigning these isolates to MS Ia, Ib or II.

cps is another major virulence factor of GBS, and its expression is controlled by the cps cluster genes. The cps cluster comprises genes cpsA-O, cpsR, cpsS and cpsY, most of which are conserved across serotypes [28]. The MS identification method based on cps synthesis gene cluster is considered to be an alternative to conventional serotyping of GBS isolates, and a convenient method of sequencing the 790-bp cps E–G gene PCR amplicon has been previously established [19]. However, this method cannot distinguish MS Ia from III-3, nor can it distinguish subtype III-4 from MS II. In this study, amplification of the cps E–G gene produced a 790-bp fragment for all the isolates. Based on the sequence heterogeneity (two SNPs found in the 790-bp sequence), all the isolates were classified as MS Ia/III-3.

An additional PCR test was performed to distinguish MS Ia from III based on their nucleotide differences in the cps H–I fragment [19, 28]. An amplicon was produced by primers (IacpsHS1 and cpsIA) specific to Ia, but not by primers (IacpsHS1 and IIIcpsHS) specific to III, confirming that all the GBS isolates belonged to MS Ia.

Multi-locus sequence typing (MLST) is a sequence-based typing method that involves sequencing ~500-bp fragments of seven housekeeping genes, and it has been used successfully to type strains of a number of human bacterial pathogens [21, 29, 30]. It is particularly suitable for epidemiological studies because it provides data that can easily be compared with the online database collected from different laboratories. The MLST system for GBS was developed in 2003, and a website was set up for easy access (http://sagalactiae.mlst.net) by Jones et al. [20]. In the present study, all the isolated strains were assigned to ST-7, but no capsular serotype corresponding to ST-7 was found in the database, suggesting that the fish GBS isolates possessed unique molecular characteristics that are not found in GBS strains of human or other animals.

GBS is the leading cause of severe bacterial infections in human newborn infants. Investigations from the USA and Western Europe have shown that four serotypes (Ia, II, III and V) account for 86–90% of all clinical isolates [20, 31], and MS Ia, III and V are the most common isolates that cause disease in human [31]. Investigation on seafood products and blood from nonpregnant adults showed high rates of GBS serotypes Ia and Ib [32]. Other reports demonstrated that several S. agalactiae isolates from different nature hosts (fish, cows and humans) were able to infect fish and cause meningoencephalitis [33], and a GBS serotype Ia isolate from a human neonatal meningitis patient recently caused disease and death of infected Nile tilapia [34]. Two genotypes of GBS were isolated from infected red tilapia and Nile tilapia cultured in Thailand, one being serotype Ia and the other belonging to an uncommon subtype of serotype III (III-4) which has also been identified as an invasive human pathogen in Hong Kong [5]. These reports indicated that GBS is a common pathogen in human and fish, and serotype Ia is the main serotype that causes disease in both human and fish thus far. In this study, the tilapia GBS isolated from different districts of China in two successive years also belong to MS Ia, however they possess molecular characteristics different from the GBS found in human. Further investigation is required for the development of effective disease treatment measures and vaccines.