Polyclonal dissemination of tetracycline resistance among Streptococcus pyogenes paediatric isolates from Brazil

Introduction: Scarce data are available on Group A Streptococcus (GAS) antibiotic resistance in South America. Methodology: The antibiotic susceptibility patterns of GAS recovered from symptomatic children living in the central part of Brazil during a prospective epidemiological study were analyzed. Results: No isolates were resistant to penicillin or macrolides. Sixty-one percent of the isolates were highly resistant to tetracycline, of which 85% harboured the tetM resistance gene. Ninety-five percent of these tetracycline resistant isolates were also resistant to minocycline. Thirty different emm-types were associated with tetracycline resistance. Phylogenetic analysis indicates that tetracycline resistance arose independently in distantly related emm-types. Conclusions: A high level of GAS tetracycline resistance has been observed in the central part of Brazil due to the polyclonal dissemination of resistant emm-types.


Introduction
Streptococcus pyogenes, also known as Group A Streptococcus (GAS), is one of the most common paediatric pathogens.It is responsible for a wide spectrum of human diseases, ranging from localized (pharyngitis, impetigo) to lethal infections (toxic shock syndrome, necrotizing fasciitis) [1,2].Penicillin represents the drug of choice for treating GAS infections, with erythromycin being used in cases of beta-lactam allergy [3].GAS resistance to macrolides is increasing in many countries, ranging from 6.2% in North America to more than 20% in some European countries [4].However, great variability across different areas has been reported [5] and scarce data are available for GAS isolated in South America [6][7][8][9].GAS tetracycline resistance is also reported in many countries and data from South America show a proportion of resistant isolates ranging from 8 to 70% [10][11][12][13].None of these recent studies have analyzed the presence of the tetracycline resistance-conferring genes.
In Brazil, antibiotics can be purchased without any prescription in most of the pharmacies and selfmedication is common, even in children [14].A study performed in Santa Catarina (South Brazil) in 2002 showed that 74% of the pharmacies offered antibiotic without any prescription [15].This is a cause for concern as an increased consumption of antibiotics, such as erythromycin and tetracycline, correlates with an increase of resistance among GAS [16,17].
Two main well-described molecular mechanisms are responsible for macrolide resistance in GAS.The ermA, ermB and ermTR resistance genes encode enzymes that methylate the 23S ribosomal RNA [18].This target site modification confers cross-resistance to lincosamide, streptogramines B, and other macrolides [18].The other molecular mechanism is encoded by the mefA gene which constitutes an active drug efflux pump and confers resistance to 14-and 15-membered macrolides.A similar mechanism is encoded by the tetK and tetL genes which confer tetracycline resistance to GAS.Tetracycline and minocycline GAS co-resistance, on the other hand, relies mostly on the tetM gene, and to a lesser extent on the tetO gene, both encoding ribosomal protection proteins [19].
emm-typing is the GAS molecular typing gold standard and is based on the DNA sequence of the 5" end 180 nucleotides of the emm gene [1].Currently, more than 200 emm-types have been reported throughout the world [20].This method of classification correlates quite well with the expression of a streptococcal apoproteinase called the serum opacity factor (sof) [21].
We previously published the clinical and microbiological characterization (emm-typing) of a GAS paediatric collection recovered from symptomatic children in the central part of Brazil during the year 2004 [22].The aim of the present study was to characterize the penicillin, macrolide, and tetracycline susceptibility patterns in this collection, as well as the genetic determinants underlying said resistance.

Materials
GAS clinical isolates analyzed in this study were prospectively collected in 2004 in Brasília, Brazil, from symptomatic children [22].Patients of <1 to 15 years old attending three public hospitals of Brasília were included in this study.The ethical board of all participant hospitals approved this study.Isolates were mostly recovered from pharyngitis (44%) and cutaneous infections (48%).Beta-haemolytic streptococci were phenotypically identified by betahaemolysis on blood agar, colony morphology, Gram stain, catalase reaction, and susceptibility to 0.04 U bacitracin.GAS identification was performed with a positive latex agglutination test containing group A specific antisera (Slidex, BioMérieux, Paris, France).emm-typing was performed using the Centre for Disease Control (CDC) protocol as previously described in [22] (information are available at http://www.cdc.gov/ncidod/biotech/strep/strepindex.htm).The phylogenetic analysis of M protein was published previously [23,24].Briefly, the portion of the emm gene encoding the whole surface exposed portion of M protein (including hypervariable and conserved regions) was sequenced, translated into an amino acid sequence, and a multiple alignment was performed using Clustal W [25].The sequence alignments were loaded in MEGA version 4 [26] to generate a bootstrapped tree using the neighbourjoining algorithm [27].

Results and Discussion
One hundred and thirty GAS isolates were recovered in Brasilia, Brazil, in 2004 from both pharyngitis and skin infections in equal proportion [22].None of these isolates were resistant to penicillin.The MICs for penicillin were very low, ranging from 0.004 mg/L to 0.032 mg/L for 98.5% of the isolates (Table 1).All the isolates were susceptible to the tested macrolides, lincosamides, and ketolides (Table 1).On the contrary, 79 of 130 isolates (61%) were resistant to tetracycline, most of them (75/79; 95%) presenting a high level of resistance (MIC  32mg/L).No difference in antibiotic resistance patterns could be observed among the three participating centres.This high rate of tetracycline resistance and absence of macrolide resistance is similar to the results obtained in previous studies in other areas of South America, including Brazil [10][11][12] (Table 2).This may be related to the low prescription rate of macrolides in Brazilian public hospitals; less than 3% of the acute pharyngitis were treated with macrolides in our study [34].Ninety-five percent (75/79) tetracycline resistant strains were also resistant to minocycline (MIC  8 mg/L) and all the minocycline resistant isolates (n = 75) were also tetracycline resistant.The remaining four isolates resistant to tetracycline (MIC = 16 mg/L) showed an intermediate level of minocycline resistance (MIC = 2 mg/L).Forty-seven out of the 79 tetracycline-resistant isolates (60%) were skin isolates; the remaining isolates were from the throat.
The tetM tetracycline resistance gene was found in 87% of the resistant isolates (69/79).None of the resistant isolates were positive for any of the three other tetracycline resistance genes (tetL, tetK and tetO) tested in this study.The tetM positive GAS isolates displayed tetracycline MICs ranging from 32 to 256 mg/L and minocycline MICs ranging from 8 to 16 mg/L (Table 3).In the other 13% (10/79 isolates), none of the four resistance genes were detected, suggesting that others or yet undescribed genes might be involved in tetracycline and minocyclin resistance in these isolates.These isolates belonged to four different emm-types (emm 33, 44/61, 64, and 74).Tetracycline and minocycline MICs ranged from 16 to 32 mg/L and from 2 to 16 mg/L, respectively (Table 3).Thus, among the 10 tetracycline-resistant isolates negative for tetM, four were minocycline intermediate, which indicates that the tetracycline resistance mechanism might be via an efflux pump, while the remaining six were minocycline resistant, suggesting the presence of a ribosomal methylase encoding gene [19].
To gain insights on the tetracycline resistance distribution with respect to the genetic relationships shared by these GAS isolates, we took advantage of our previous studies describing the genetic relatedness of the extracellular portion of the M protein among our isolates [23,35].Figure 1 shows that the GAS isolates included in this study separated into two main clades reflecting their emm-pattern and sof status [28,29,36,37].The sof + A clade was mainly composed of emm-pattern E isolates, which are considered to be generalist GAS (skin and throat tropism) [36].This clade contained the majority of the isolates (75/130, 58%).Clade B, on the other hand, contained 51/130 isolates (39%) and was composed of sof + isolates belonging to emm-pattern A-C (throat) and D (skin) [36].Four isolates were not included in the phylogenetic analysis for technical reasons [23].The 51 tetracycline susceptible isolates belonged to 21 different emm-types, which were found both in clade A and B. Interestingly, the 79 tetracycline resistant isolates were also found in these two clades.They represented 30 different emm-types and were of the A-C, D and E emm-pattern groups.This observation illustrates that tetracycline resistance does not show any preferential genetic background and may arise in distantly related emmtypes.The Tn916 transposon is widely distributed in Gram-positive bacteria [38] and carries the tetM gene [39].Multiclonal dissemination of the resistance trait is likely to be caused by horizontal transfer of the transposon (like the Tn916) rather than the epidemic spread of a few resistant clones.Multiclonal dissemination of tetracycline resistance was also described in southeast Brazil [10,11] and in Iran [17], which is in accordance with our results.However, other studies have observed somewhat different results.In Italy, 80% of the tetracycline resistant GAS isolates were found in four closely related clusters, as defined by pulsed field gel electrophoresis (PFGE) [40].However, GAS epidemiological studies underlined the small number of circulating emmtypes in western countries as opposed to "developing/emerging countries" where a large number of different emm-types are usually recovered [41].The apparent clonality of tetracycline resistance observed in western countries might in fact be due to the reduced number of circulating emm-types rather than the non-susceptible status.A similar interpretation has been proposed for fluoroquinolone non-susceptibility in GAS [24].
Although extensive genetic diversity is found in tetracycline resistant GAS, the same emm-type is rarely associated with susceptible and resistant isolates.The closely related emm58 and emm118 emm-types (clade A) have representatives that are either susceptible or resistant to tetracycline.The unrelated emm74 (clade B) and emm44/61 (clade A) have representatives that are tetracycline and minocycline resistant or tetracycline resistant and minocycline susceptible.The remaining 42 emmtypes have representatives that are exclusively tetracycline resistant or tetracycline susceptible.Thus we observed a good correlation between defined emm-types and tetracycline susceptibility status.At least three previous studies (performed in Italy, Iran, and Brazil respectively) observed a somewhat different correlation.These authors found that susceptible and resistant isolates belong to the same clonal type, as determined by PFGE [10,17,40].Within a defined geographical location and a restricted period of time (i.e. the settings of this study), emm-typing usually correlates quite well with other typing methods, such as multilocus sequence typing or PFGE [28,42].
In contrast, when comparing different locations or longer periods of time (i.e. the settings of the three other studies), identical emm-types might be associated with different clonal types.This difference in the study design might therefore explain the different observations.Studies using different typing methods might help to clarify the correlation between emmtypes, clonal types, and tetracycline/minocycline susceptibility status.Studies including at least 20 GAS isolates recovered in South America and published after January 2000 were compiled.ND: not determined.None of these studies analysed the presence of tetracycline resistance genes.PFGE, Pulsed field gel electrophoresis; ery R, erythromycin resistant; tet R, tetracycline resistant.The evolutionary history was inferred using the neighbour-joining method [27].The tree is drawn to scale, with branch lengths proportional to the evolutionary distances.Bootstrap value higher than 70% are shown next to the branches (500 replicates).

Conclusions
Our epidemiological study shows that the GAS isolates included in our study are macrolide susceptible.However, a high frequency of tetracycline resistance was observed (61 %) and is likely to be caused by a polyclonal dissemination of the tetM resistance gene.Tetracycline resistance has been suggested to precede macrolide resistance [47].With the potential increase of macrolide prescriptions in the future, active epidemiological surveillance will be needed to monitor the evolution of GAS antibiotic resistance patterns.

Table 1 .
In vitro activities of 7 antibiotics

Table 2 .
Macrolide and tetracycline resistance among GAS isolated in South America

Table 3 .
Presence of tetM according to tetracycline and minocycline MICs and emm-type