High prevalence of group B streptococcus ST17 hypervirulent clone among non-pregnant patients from a Hungarian venereology clinic

Ninety-six randomly selected GBS strains, deriving from the Department of Dermatology, Dermatooncology and Venerology, Semmelweis University, Budapest, over a 1-year period (October 2010 – September 2011), were involved in the study. During the study period, a total of 10,519 specimens arrived at the laboratory. Out of these, 296 were S. agalactiae, which equals to a 2.8% prevalence. Among the urogenital specimens (n = 4327), 259 were positive for S. agalactiae (6.0%), while among the dermatology specimens (n = 6192), only 37 were positive for S. agalactiae (0.6%).The vast majority of the examined isolates (n = 76) derived from the STD ambulance (which examines patients from the whole country), 14 from the general ambulance, six from the psoriasis ambulance. Most specimens were either vaginal (n = 38), urethral (n = 30), or from skin (n = 13); the others included glans (n = 8), urine (n = 4), and one each from nose, anus and ejaculate. The etiological role of GBS in the skin infections is more likely, however, the pathogenic role of the bacterium in STD patients remains unclear, especially if other microbes were simultaneously detected. The age of patients ranged between 15 and 77 years, with a mean age of 42.4 years. The genders were nearly equalised: 50 females and 46 males.

The following bacteria were also isolated from the patients: Ureaplasma spp.: n = 29, Chlamydia trachomatis D-K serotypes: n = 12, Mycoplasma hominis: n = 5, Ureaplasma / Mycoplasma: n = 1. There was no ongoing Treponema pallidum infection, but eight patients had syphilis treatment in the anamnesis. Every patient was negative for HIV, HBV, and HCV; HSV1–2 was not tested. In the urogenital specimens, the following Candida species were identified: C. albicans (n = 11), C. krusei (n = 3), C. glabrata (n = 3), C. tropicalis (n = 1). No yeasts were found in the dermatological samples. Trichomonas was present in three cases.

Identification of S. agalactiae was primarily done with routine laboratory tests (β-haemolysis, catalase negativity), followed by the Lancefield grouping with the Pastorex STREP agglutination Kit (BioRad, California, US). The identity of the isolates was confirmed in every case by PCR detection of the species-specific dltR gene, as described by Lamy et al. [28].

Susceptibility to six antibiotics (penicillin, erythromycin, clindamycin, levofloxacin, moxifloxacin, and tetracycline) was done by agar dilution method, using an A400 multipoint inoculator (AQS Manufacturing Ltd., Southwater, UK), according to the EUCAST guidelines [31]. As controls, Streptococcus agalactiae ATCC 80200 and S. pneumoniae ATCC 49619 were used. The macrolide resistance genes ermB, ermTR, mef, and linB were detected by PCR [32‐34]. The distinction between mefA and mefE was carried out by BamHI digestion, which generates two fragments in mefA, but none in mefE, as described before for S. pneumoniae [35].

Inducible clindamycin resistance was tested with the double-disc method, using 15 μg erythromycin and 2 μg clindamycin discs, as described by EUCAST [31], for those isolates which were resistant to erythromycin but had low clindamycin MIC.

To distinguish first serotypes I, II and III, the Pastorex Strepto BI, BII, BIII latex agglutination test (Bio-Rad) was used. Subsequently, the serotype of every isolate was determined by multiplex PCRs for serotypes Ia-IV and V-VIII [36] and a separate PCR was used for serotype IX [37].

The genes of Alpha-C, Alp2/3, Alp4, Rib and Epsilon proteins were detected by multiplex PCR, using a universal forward primer and a type-specific reverse primer for each protein, as described by Creti et al. [20].

The strict specificity of the hvgA gene to the hypervirulent ST-17 MLST clone enables the rapid and cheap PCR detection of this clone instead of performing the real MLST [28].

A 345 bp fragment was amplified with the primers described by Tazi et al. [29]. The other variant of the same protein, Gbs2018A (bibA) was also detected by PCR [30].

In all cases, DNA was prepared by the ZR Fungal/Bacterial DNA MiniPrep (Zymo Research Corp., Irvine, CA, US). Briefly, approximately 100 mg bacterial cells (equivalent to 1–2 plateful ON cultures) were resuspended in 200 μl saline buffer and transferred to a specific Lysis Tube, containing beads. After 750 μl Lysis solution was added, the suspension was rigorously vortexed for 5 min on full speed, in order to disrupt the bacterial cell wall by bead beating. The mixture was then centrifuged at 10,000 g and the supernatant additionally filtered to remove cell debris. The filtrate – containing the DNA – was then purified by the usual column-binding method, also provided in the kit.

To determine the genetic relatedness of the isolates, pulsed-field gel electrophoresis (PFGE) was used, as described by Benson et al. [38]. The complete bacterial genome was embedded in agarose plugs and lysed in several steps (with lysozyme, mutanolysin and proteinase K) to purify DNA. For the digestion, Sma I restriction enzyme was used (3 h, 25 °C). The digested samples were run in a 1% agarose gel along with N0340S Lambda Ladder PFG Marker (New England Biolabs, Hitchin, Hertfordshire, UK), in a Bio-Rad CHEF-DR® II PFGE machine, for 21 h at 14 °C, with the following pulse times: block 1, 5 s/15 s for 10 h, and block 2, 15 s/60 s for 11 h. After the gel image was captured, dendrograms were created by the Fingerprinting II software (Bio-Rad, Marnes-la-Coquette, France) and the PFGE patterns were analysed according to the criteria of van Belkum et al. [39].

Multi-locus sequence typing (MLST) was performed on 20 representative isolates, based on the PFGE patterns and the serotype - surface protein combinations. Well-defined sections of seven housekeeping genes (adhP, pheS, atr, glnA, sdhA, glcK, tkt) were amplified by PCR, using the primers provided by the MLST website [40]. The products were purified by the QIAquick PCR purification kit (QIAGEN, Hilden, Germany) and sent for sequencing to BIOMI Ltd., Gödöllő, Hungary. The allele sequences were compared to the MLST database and the sequence types were identified.

For statistical analysis, Fisher’s exact test or Chi-square test was applied.

All isolates were fully susceptible to penicillin, and only three isolates were resistant to levofloxacin and moxifloxacin (Table 1). On the other hand, 82.3% of them were resistant to tetracycline, and the macrolide and lincosamide resistance was also high (41.7% to erythromycin and 35.4% to clindamycin). Among the 40 erythromycin-resistant strains, 27 possessed the ermB gene alone, two had ermTR alone, four had ermB and ermTR together, one carried ermB + linB together, and finally six had mef genes (five mefE and one mefA). Among the six isolates showing resistance to erythromycin, but sensitivity to clindamycin, the D-test was positive in four cases (meaning inducible MLS_B resistance), out of which two isolates expressed high-level (128 and 256 mg/L) and two isolates only low-level (16 and 32 mg/L) erythromycin resistance. These latter two isolates were mef-negative but possessed the ermTR gene. The association between M phenotype (i.e., low-level resistance to erythromycin, but sensitivity to clindamycin) and the presence of mef genes (n = 6/9) or the ermTR gene (n = 3/9) was statistically significant (p < 0.001 for both, Table 2). A further eight isolates, which were phenotypically macrolide sensitive, carried the ermB gene. The six ermTR+ isolates belonged to serotype V (n = 5) and serotype Ia (n = 1), and the D-test was positive in three cases. Among the 17 tetracycline sensitive isolates, 12 were fully macrolide sensitive and five had the M phenotype.

For the 96 isolates in this study, the serotype prevalence in ranking order was the following: V (29.2%, n = 28), III (27.1%, n = 26), Ia (22.9%, n = 22), IV (10.4%, n = 10), II (5.2%, n = 5) and Ib (4.2%, n = 4), and finally one isolate proved to be non-typable. Always only one surface protein could be detected for each isolate; rib (32.3%) and eps (29.2%) were most common, followed by alp2/3 (24.0%) and alpC (14.6%). A strong association was demonstrated between the surface protein genes and serotypes, such as alp2/3 with type V, rib with type III, or eps with types Ia and IV (Table 3).

Serotypes Ia, Ib, and II were usually sensitive to macrolides, except for a type Ib strain, which was linB + ermB positive. Among serotype III isolates, both macrolide-resistant and sensitive strains were found, while serotype V was typically associated with macrolide resistance. Interestingly, although 6/10 serotype IV isolates were phenotypically macrolide sensitive, the ermB gene was detected in four of these. The highest levofloxacin and moxifloxacin MICs were measured with serotype II.

Of note, there were differences observed between the isolates deriving from the urogenital tract of female (n = 42) or male patients (n = 40). Whereas the prevalence of serotype V and IV was much higher in female specimens than in males (38.1% versus 20.0, and 11.9% versus 7.5%, respectively), the opposite was true for serotype III (21.4% versus 32.5%). As the surface protein genes showed a strong correlation with the serotypes, their presence also showed differences: meanwhile Alp2/3 showed a prevalence of 28.6% versus 17.5% in the two gender groups; AlpC was present in 9.5 and 20.0%, respectively. However, none of these differences were statistically significant.

The ST-17 specific hvgA gene was detected in 21 isolates, which equals to 21.9% prevalence. Nineteen out of these were serotype III, one isolate was serotype IV and one was non-typeable. MLST analysis of the serotype III isolates confirmed that these were ST-17, while the serotype IV isolate proved to be ST-291, which differs from ST-17 in a single nucleotide in the pheS allele, therefore belongs to the CC-17 clonal complex as well. All remaining isolates (n = 75), belonging to several different serotypes, possessed the bibA gene (serotype V: n = 28, Ia: n = 22, IV: n = 9, III: n = 7, II: n = 5, Ib: n = 4).

However, there were two serotype V isolates (B48 and B64), which shared a common PFGE pattern and were rib+, but were highly dissimilar from members of the major serotype V clone (alp2/3+). B48 proved to belong to ST-110, but the other strain (B64) surprisingly contained a 44-bp deletion in the middle of the atr allele. Otherwise, its atr sequence showed identity with atr-3, which is present in ST-110 (Fig. 1). The atr gene fragment sequence of the isolate B64 has been deposited in the GenBank nucleotide sequence database under accession number MG675237.

Interestingly, the ST17 / serotype III isolates were quite heterogeneous based on the banding patterns. Sometimes different serotypes shared the same restriction pattern. We found one serotype IV isolate, which belonged to the ST-291 clone, which differs only in a single nucleotide in the pheS allele compared to ST-17, hence is a member of the CC-17 clonal complex. Similarly, one of the serotype IV isolates (eps +, macrolide-resistant) shared the same banding pattern of the major serotype V cluster, i.e. those belonging to ST-1, and the MLST analysis revealed that it was ST-196, which is a double-nucleotide variant of ST-1 (one nucleotide difference each in the atr and glcK alleles), and also part of CC-1. Two serotype IV isolates belonged to the ST-23 clone and one to the ST-24 clone, both being members of CC-23. Finally, two serotype Ib and one serotype II strain belonged to CC-10. A summary of the isolates of known sequence type is shown in Table 4.