Camel GBS isolates used in this study

All work described was in full compliance with national regulations. The work was approved by the ethical committee of the International Livestock Research Institute, which adheres to international standards and is accredited by the National Council of Science and Technology in Kenya (approval number ILRI-IREC2013-12).

Information about the 92 camel GBS isolates used in this study is provided in Additional file 1. The GBS were isolated using standard methods [17] and the species level (GBS) was determined using Lancefield serological grouping as described before [4, 5]. Briefly, specimen material was streaked out on Edwards agar plates (Oxoid No. CM0027) containing 5% defibrillated sheep blood. Plates were examined after 24 and 48 h of incubation at 37 °C. Small blue and beta hemolytic colonies were subcultured for further testing. Gram-positive and catalase-negative cocci that were unable to hydrolyze esculin and reacted positive when subjected to Lancefield B testing using Oxoid No. DR587 Latex Grouping Reagent B were considered to be Streptococcus agalactiae (GBS).

Isolation of genomic DNA from camel GBS

The isolates were grown in 10 mL Luria Broth (LB) over night at 37 °C. Culture material was centrifuged at 4 °C and 8000 g for 10 min, the supernatant was discarded and the cell pellet was resuspended in 275 μL H₂O, 10 μL RNAse A (10 mg/mL) and 275 μL TEN buffer (0.05 M Tris, pH 8.0, 0.001 M EDTA, 0.016 M NaCl, 100 mg/mL Lysozyme). The solution was incubated at 37 °C for 30 min, 15 μL 20% SDS and 10 μL proteinase K (20 mg/mL) were added, mixed and the solution was incubated at 37°C for 60 min. 125 μL of 4 M NaCl and 80 μL CTAB solution (10% in H₂O, preheated to 50°C) were added, mixed and the solution was incubated at 65°C for 10 min. A phenol/chloroform extraction followed by a DNA precipitation using ethanol according to standard protocols [18]. The DNA pellet was resuspended in TE (pH 8.0) buffer and stored at -80 °C for subsequent use.

In vitro antimicrobial susceptibility testing of camel GBS

Testing of the presence of the tetracycline resistance genes tetM, tetO, and Tn916-like elements in camel GBS

The presence of the common genes reported to confer resistance to tetracycline i.e. tetM and tetO, was investigated in camel isolates phenotypically resistant to tetracycline using PCR amplification as described previously [21]. Twenty nanograms of genomic DNA was used as template in a final volume of 50 μL of PCR master mix containing; 1 × PCR buffer (DreamTaq™including MgCl₂, Fermentas, Germany) 200 μM of dNTP; 600 nM each of the forward/reverse primers and 1.25 U of DreamTaq™ DNA Polymerase. The amplification conditions were as follows; denaturation for 3 min at 95 °C, followed by 35 cycles of 95 °C for 1 min, 48/55 °C for 1 min for tetM/ tetO respectively, and 72 °C for 1 min with a final extension of 72 °C for 10 min. Positive controls strains for tetM and tetO PCRs were S. agalactiae 2603 V/R [15] and Streptococcus anginosus MG23 [22], respectively. Selected PCR products were sequenced and trimmed from position 865 to 1075 in tetM of strain 2603 V/R (GenBank Accession Number: NC_004116). A possible location of the tetM gene within a Tn916-like element was evaluated using a PCR spanning from the tetM gene (tetM 3-end: 5′-ACTACCGGTGAACCTGTTTG-3′) to the transposase (Tn916 tnase: 5′-TGGCTCTCTCCAGTCTTTAAG-3′). The expected PCR product was 2,740 bp long. PCR conditions were as outlined above with the difference that the annealing temperature was 58 °C and the extension time was set to 3 min. Positive control strain for Tn916 was Staphylococcus rostri RST11 [23].

Molecular capsular typing of camel GBS

Capsular polysaccharide gene types (cps) were determined using a multiplex PCR assay as described previously [12] using DreamTaq™ DNA Polymerase. Amplified products were separated on a 1.5% agarose gel and the band pattern of the isolates were compared to the band pattern of the GBS reference isolates included in this study representing all known capsular types; Ia, Ib, II, III, IV, V, VI, VII, VIII and IX [12].

Multilocus Sequence Typing (MLST) of camel GBS

The MLST was performed as described previously [24]. Briefly, seven S. agalactiae housekeeping gene loci were amplified, including alcohol dehydrogenase (adhP), phenylalanyl tRNA synthetase (pheS), amino acid transporter (atr), glutamine synthetase (glnA), serine dehydratase (sdhA), glucose kinase (glcK) and transketolase (tkt). The PCR amplification was carried out in duplicate (50 μL reaction volume) using GoTaq® Green master mix polymerase (Promega, USA) according to manufacturers’ instructions. PCR products were purified using a QIAquick PCR purification kit (QIAGEN, Germany) and sequenced by Macrogen Inc. (Seoul, Korea). Sequence traces were assembled using the CLC workbench 6. Individual alleles of the camel isolates were compared to the MLST database entries and novel alleles were assigned new allele numbers. New combinations of alleles were also assigned new sequence type (ST) numbers. All camel isolates were submitted to the MLST database [25].

Minimum Spanning Trees (MST) of camel GBS

To visualize the genetic relationship between camel GBS, MSTs were generated from the allelic profiles of the different isolates using the predefined settings in BioNumerics® 6.6. The clonal complex/populations, capsular type (Ia, Ib, II, III, IV, V, VI, VII, VIII and IX), tetracycline resistance (resistant or susceptible) and clinical complex (chronic cough, mastitis, wound infection/abscess/peri-arthricular abscess, gingivitis, vaginal discharge and healthy camels) were plotted onto the minimum spanning tree using different color codes.

Phylogenetic analysis of East African GBS isolated from camels and humans

We compared the camel GBS with other published MLST-typed GBS from East Africa [16]. A concatenated sequence of each ST was generated using MLST data from this study and from elsewhere [16]. Jmodeltest 1.0 [26] was used to select the best fitting model of nucleotide substitution, which was found to be the Generalized Time Reversible model with invariant sites and gamma-distributed rate heterogeneity (GTR + I + G) [27]. A maximum likelihood phylogeny for all STs was estimated using this model in PhyML 3.0 [28]. To assess statistical support for the resulting phylogeny, we performed 1000 bootstrap replicates. The tree was drawn using the software FigTree v1.3.1 [29].

Determination of population structure of East African GBS isolated from camels and humans

For this analysis we included 169 human isolates [16], which were MLST-typed before in order to investigate a possible gene flow between camel and human GBS. The population structure was estimated using the linkage model in STRUCTURE v2.3.2 [30, 31]. This Bayesian approach uses multilocus genotypic data to define a set of populations with distinct allele frequencies, and assigns isolates probabilistically to defined populations without prior knowledge of sampling location or sampled host. This program identifies admixtures of isolates and provides an estimate of the percentage ancestry from ancestral population for each isolate. We performed eight replications of the test, in which we initially discarded 10 000 Markov Chain Monte Carlo (MCMC) iterations as burn-in and kept the subsequent samples from 30 000 MCMC iterations for analysis. We tested values of K between 1 and 10, where K is the number of inferred populations. The results of the eight independent runs were averaged for each K value to determine the model with the highest likelihood. Real and simulated data have shown that it is not straightforward to determine the optimal value of K when complex population structure is present [30, 32] so we also calculated ΔK, a measure of the second order rate of change in the likelihood of K [32] to estimate the appropriate K value for our data.

In vitro antimicrobial susceptibility of camel GBS

The susceptibility profiles for 90 out of 92 camel GBS isolates against 23 antimicrobials are presented in Table 1 and Additional file 2. Two strains showed poor growth in Mueller-Hinton Broth with lysed horse blood and were therefore excluded from analysis of minimal inhibitory concentrations (MIC). The interpretation of data is hampered by the absence of clinical breakpoints for camels. We included the epidemiological criterion (wild type cut-offs, ECOFF) for interpretation of data, which unfortunately was only available for a fraction of antimicrobials tested. This criterion was determined on the basis of human strains and a different methodology than the one applied in this study. However, on the basis of veterinary clinical breakpoints available for streptococci [19] and MIC₉₀ values determined in this study, only resistance to gentamicin and tetracycline is obvious (Table 1). In fact, 34% of the GBS isolates were resistant to tetracycline. One isolate (ILRI029, ST-617, capsular type VI) showed reduced sensitivity to all β-lactam antibiotics tested, probably due to a mutation in the gene pbp, encoding the penicillin binding protein. As expected most GBS isolates showed relatively high MIC to gentamicin in the range of 8–64 mg/L that is characteristic to the genus Streptococcus. For all other antimicrobial agents investigated, MIC distributions were normal.

Prevalence of tetM, tetO or Tn916-like elements

All tetracycline-resistant camel GBS harboured the tetM gene. All camel GBS were negative for tetO. The 211 bp sequence obtained from the tetM amplicon was 100% identical between the isolates investigated (N = 17) and 100% identical to other tetM sequences in the database from Staphylococcus aureus and S. agalactiae (e.g. GenBank accession number: NC_004116 and CP003808). In addition, we showed that, in all tetracycline-resistant camel GBS, tetM was linked to the Tn916 transposase, pointing towards a transfer of the resistance gene via a Tn916-like element.

Capsular typing

We assigned a capsular type to all but one camel GBS isolate (ILRI041). The capsular types detected included Ia, II, III, V and VI. The most common types were Ia (37%, 34 isolates), III (27%, 25 isolates) and VI (26%, 24 isolates). Type II and V were less commonly detected with only 4% of the isolates (4 isolates each) being positive for the respective types. Capsular types Ib, IV, VII, VIII and IX were not present among the camel isolates tested.

Multilocus Sequence Typing (MLST) of camel GBS

Relationship between camel GBS STs, clonal complexes, capsular types, tetracycline resistance and clinical complexes

According to the minimum spanning tree (MST) network and the number of shared alleles, the ten camel STs clustered in three clonal complexes/populations consisting of ST-609/ST-614, ST-616 and the remaining seven STs that had six alleles in common (Figure 1A). ST-609 and ST-614 were most distantly related to the other camel ST-609 and shared only 2 alleles with ST-616 (Figure 1A).

The ten STs that grouped into 3 clonal complexes were further plotted against capsular type, resistance to tetracycline and clinical complexes. Capsular type Ia was predominant in isolates with ST-618, ST-613 and ST-612, while capsular type VI was most common in isolates with ST-610 and ST-617 (Figure 1B). Isolates with capsular type II were of ST-612 and ST-615. All isolates within ST-616 belonged to capsular type III while all isolates with ST-609/ST-614 belonged to capsular type V (Figure 1B).

Tetracycline resistant isolates were found within ST-612 (4%; 1 out of 24) and ST-617 (30%; 8 out of 26), however the majority of the resistant isolates were grouped within ST-616 (92%; 22 out of 24) representing a total of 71% of all resistant GBS (Figure 1C). Interestingly, most of the GBS isolated from cases of mastitis (81% of mastitis isolates) belonged to ST-616 (Figure 1D). Wound infection and abscesses were found mainly in STs other than ST-616.

No association between STs and geographical origin of the GBS isolates could be detected (data not shown).

Comparison of camel GBS with other GBS from the region

To investigate the relationship between camel GBS and other GBS from the region, we compared our dataset to previously described human GBS from Kenya [16].

A maximum likelihood (ML) phylogeny was generated using a total of 33 STs (10 camel GBS STs and 23 human GBS STs [16]). An unrooted tree clearly showed that all camel GBS group in two well supported clusters that are phylogenetically distinct from the Kenyan human GBS (Figure 2). One cluster comprises ST-609/ST-614, the other one all the other isolates.

In order to look for evidence of genetic exchange between camel and human GBS, we performed a STRUCTURE analysis. Calculation of ΔK produced a modal value for K = 7, whereas the averaged highest likelihood for eight independent runs was obtained for K = 8. We therefore present results for both K = 7 and K = 8 in Figure 3. The seven populations in common for both analyses were; two populations of camel isolates only, corresponding to 2 of the clonal complexes identified on Figure 1A, containing 64 and 24 isolates respectively. All 24 isolates belonging to the second population are of capsular type III, whereas 63 of the 64 isolates belonging to the first population are of capsular type Ia, II, IV or VI. The remaining isolate could not be assigned to a distinct capsular type. Four populations consisted only of human isolates, corresponding to previously defined clonal complexes [16]. The last population consisted of seven isolates, four from camels (ST-609/ST-614) and three from humans, all belonging to capsular type V. These isolates are hybrids with mixed ancestry, the camel isolates having at least 36% shared ancestry with human isolates. The human clonal complexes CC1 and CC19 [16], formed one population for K = 7, but were split into two distinct populations for K = 8.

Conclusions

Camel GBS sequence types (STs) were distinct from STs reported from other hosts so far. Most mastitis causing GBS were associated with ST-616. Widespread resistance (34%) to tetracycline was most prominent in ST-616 and associated with acquisition of tetM carried on a Tn916-like element. The presence of tetM within different MLST clades suggests acquisition on multiple occasions. Wound infections and mastitis in East African camels associated with GBS should be treated with antimicrobials other than tetracycline in East Africa.