Abstract

Background

We identified erm(A)-harbouring Streptococcus pyogenes that expressed three variant phenotypes: (1) low-level resistance to erythromycin (MICs 1–4 mg/L) but high azithromycin MICs in absolute terms (16–64 mg/L; n = 6); (2) same as (1) but with a high clindamycin MIC (256 mg/L; n = 1); and (3) high-level constitutive MLS (cMLS) resistance (n = 1). Here we analysed the genetic basis of these novel phenotypes.

Methods

The presence of erm(A) and the absence of macrolide/lincosamide resistance genes erm(B), mef and cfr were confirmed by PCR. erm(A), 23S rRNA, L4 and L22 genes were sequenced. Mutant erm(A) genes were cloned and electrotransformed into the macrolide-susceptible Escherichia coli AG100A. Clonality was determined by emm typing and PFGE. Effects of the identified mutations on free energy changes (ΔG) and putative configurations of the leader sequence were studied in silico.

Results

Point mutations (G98A, A137C, C140T and G205A) were observed in the erm(A) regulatory region of all eight erm(A)-harbouring S. pyogenes. Five and two isolates belonged to emm77 and emm89 clones, respectively, and one isolate was an emm1. E. coli transformed with mutant erm(A) harbouring G98A, A137C or C140T mutations (phenotypes 1 and 2) did not express high-level azithromycin or clindamycin resistance. However, cMLS resistance was clearly observed in transformants with erm(A) harbouring both A137C and G205A mutations (phenotype 3). In silico analysis showed that ΔG was minor except for the G205A mutation. Secondary structure predictions further showed that the A137C and G205A mutations together abolished the hairpin sequestering the ribosome-binding and initiation sites of the erm(A) gene, explaining the cMLS phenotype 3.

Conclusions

We report point mutations in the erm(A) regulatory region leading to constitutive methylase expression and the presence of additional, as yet unidentified mechanisms mediating high-level azithromycin and clindamycin resistance in erm(A)-harbouring S. pyogenes.

Introduction

High efficacy and safety of macrolides have made them popular drugs for the treatment of pharyngitis and other infections caused by Streptococcus pyogenes. However, excessive macrolide use has been linked to the emergence and spread of macrolide-resistant strains.1,2 Macrolide resistance commonly occurs due to methylation of the macrolide-binding site on the ribosome by methyltransferases encoded by the erm group of genes.3 Expression of the erm genes can be constitutive or inducible, depending on the sequence of the regulatory domain upstream of the structural gene. For instance, the erm(A) gene encodes an inducible methylase commonly present in Staphylococcus aureus and also reported in S. pyogenes. Induction of erm(A) occurs by translational attenuation, in a manner similar to that of erm(C), the prototypic ‘inducible’ erm gene.4 However, the erm(A) regulatory region is more complex and is composed of two leader peptides (LPs) as opposed to one for erm(C). erm(A) can be induced by the 14-membered ring (erythromycin) and 15-membered ring (azithromycin) macrolides, but not by the lincosamides (clindamycin) that share an overlapping ribosomal binding site with the macrolides. However, in vitro experiments show that clindamycin can easily select for alterations in the erm(A) regulatory region, resulting in a constitutive expression.5 Nonetheless, clinical S. pyogenes harbouring constitutively expressed erm(A) have remained rare so far.6 Moreover, in vitro-generated erm(C) LP mutants have been shown to exhibit variable efficiencies of induction with macrolides and lincosamides.7 We report here erm(A)-harbouring S. pyogenes, isolated from skin and throat infections in Belgium and Italy, expressing divergent phenotypes where the role of erm(A) in mediating these phenotypes was further investigated.

Materials and methods

Strain collection and identification

S. pyogenes isolated from skin and throat infections during national surveillance in Italy and Belgium were screened for macrolide resistance by phenotypic and genotypic methods. Isolates exhibiting phenotypic macrolide resistance by the conventional double-disc diffusion test with erythromycin and clindamycin were grouped into constitutive (cMLS), inducible (iMLS) and macrolide (M) phenotypes. MICs of erythromycin, spiramycin, clindamycin (Sigma-Aldrich, St Louis, MO, USA), clarithromycin (Abbott Ottignies, Belgium), azithromycin (Pfizer, Groton, USA) and telithromycin (Aventis, Romainville, France) were determined by agar dilution, and results were interpreted according to CLSI guidelines.8

Macrolide-resistant S. pyogenes were investigated for the presence of erm(B), erm(A) and mef(A) by a previously described multiplex PCR protocol.9 Screening for the cfr methyltransferase that confers resistance to clindamycin but not to macrolides in S. aureus was performed by PCR using primers FW (5′-TGTGACATGGATACCAGCAG-3′) and REV (5′-TCTTTTATGGGAAT GGGTGA-3′). These investigations yielded eight macrolide-resistant erm(A)-harbouring S. pyogenes expressing variant phenotypes that were investigated further.

Alterations in erm(A) or in the macrolide-binding ribosomal sites

The entire erm(A) gene, including the regulatory region, was amplified from the eight variant isolates and from two erm(A)-harbouring S. pyogenes used as controls and exhibiting a ‘normal’ iMLS phenotype. Three sets of overlapping primers (sequences available on request) were designed based on the erm(TR) sequence (accession no. AF002716). PCR mix and cycling conditions were as described previously.10 L4, L22 and portions of 23S rRNA genes (two parts of domain V and one of domain II) were amplified with known primers.11 The amplified parts of domain V and domain II include regions where nucleotide modifications associated with erythromycin resistance have been described previously (G2057, A2058, A2062, G2505, C2611, A752 and A754, Escherichia coli numbering). PCR products were analysed by direct double-strand sequencing (3730 DNA Analyzer, Applied Biosystems, CA, USA), with the BigDye Terminator Version 3.1 Kit (Applied Biosystems). Nucleotide sequence alignment was performed using SeqMan (DNASTAR, Madison, WI, USA).

Cloning and transformation experiments

The entire erm(A) gene, including the regulatory region, from isolates IT1, IT2, B1, B2, B3 and one wild-type erm(A) control strain (Table 1) was amplified (primers on request) and cloned into pCR2.1, according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA). The recombinant plasmid was electroporated into competent E. coli AG100A (a kind gift from Patrice Courvalin), which is susceptible to macrolides due to disruption of the AcrAB pump, and selected on Luria–Bertani agar with ampicillin (50 mg/L) and azithromycin (4 mg/L). Phenotypic analysis (double-disc diffusion and MICs of erythromycin, clindamycin and azithromycin) and double-strand sequencing were performed on the transformants.

Table 1

Characteristics of the variant and wild-type strains of erm(A)-harbouring S. pyogenes and the free energy changes (ΔG) resulting from the point mutations/amino acid substitutions identified in the erm(A) regulatory region

   MIC (mg/L)
 
   
Country Isolates Mutations in erm(A) regulatory region (amino acid substituted) erythromycin clarithromycin azithromycin spiramycina clindamycin telithromycinb PFGE type emm type ΔG 
Italy IT1–IT5 C140T (P16L) 1–4 16 0.06–0.12 0.06–0.12 0.03–0.01 emm77 −49.2 kcal/mol 
Belgium B1 A137C (Q15P) 64 0.5 0.06 emm−49.2 kcal/mol 
B2 A137C (Q15P) G98A (G2N) 32 256 0.06 49 emm89 −47.6 kcal/mol 
B3 A137C (Q15P) G205A >512 >512 >512 512 >512 64 49 emm89 −43.4 kcal/mol 
WT1 (wild-type) — NDc 0.25 0.01 64 emm94 −50.9 kcal/mol 
WT2 (wild-type) — ND 0.06 0.01 emm58 −50.9 kcal/mol 
   MIC (mg/L)
 
   
Country Isolates Mutations in erm(A) regulatory region (amino acid substituted) erythromycin clarithromycin azithromycin spiramycina clindamycin telithromycinb PFGE type emm type ΔG 
Italy IT1–IT5 C140T (P16L) 1–4 16 0.06–0.12 0.06–0.12 0.03–0.01 emm77 −49.2 kcal/mol 
Belgium B1 A137C (Q15P) 64 0.5 0.06 emm−49.2 kcal/mol 
B2 A137C (Q15P) G98A (G2N) 32 256 0.06 49 emm89 −47.6 kcal/mol 
B3 A137C (Q15P) G205A >512 >512 >512 512 >512 64 49 emm89 −43.4 kcal/mol 
WT1 (wild-type) — NDc 0.25 0.01 64 emm94 −50.9 kcal/mol 
WT2 (wild-type) — ND 0.06 0.01 emm58 −50.9 kcal/mol 

aMIC breakpoints for spiramycin: susceptible, 1 mg/L; intermediate, 2 mg/L; resistant, 4 mg/L.

bMIC breakpoints for telithromycin: susceptible, ≤1 mg/L; resistant, ≥4 mg/L.

cND, not determined.

Typing of isolates

Clonality of the eight isolates was investigated by PFGE using SmaI, and the patterns generated were analysed by GelCompar software 4.0 (Applied Maths, Kortrijk, Belgium).1emm typing was performed by the method described on the CDC web site (http://www.cdc.gov/ncidod/biotech/strep/strepindex.html).

Secondary structure of erm(A)

Secondary structure predictions of the entire, wild-type erm(A) transcript in the uninduced state were made after comparisons with the erm(A) and erm(C) sequences in S. aureus using the DCSE tool.12 Free energy changes (ΔG) were calculated using M-fold.13 Drawings were performed using RnaViz 2.0.14

Results and discussion

erm(A)-harbouring S. pyogenes expressing variant phenotypes

Of the eight erm(A)-harbouring S. pyogenes studied, isolates IT1 to IT5 were clonal and exhibited a typical iMLS phenotype on double-disc diffusion with low-level resistance to erythromycin and susceptibility to clindamycin, although MICs of azithromycin were exceptionally high (16 mg/L) (Table 1). Isolate B1 also showed an iMLS phenotype, although with intermediate resistance to clindamycin and high-level resistance to azithromycin (64 mg/L). Isolate B2 showed a novel phenotype (low-level resistance to erythromycin and high-level resistance to clindamycin), which presented as a shadow ‘D’ on double-disc diffusion with complete overgrowth in the clindamycin zone. This phenotype could be interpreted as a constitutive phenotype.5 MICs of both azithromycin (32 mg/L) and clindamycin (256 mg/L) were high for isolate B2. Isolate B3 exhibited the constitutive phenotype with corresponding high MICs of all antibiotics tested (≥512 mg/L). All eight isolates showed point mutations in the erm(A) regulatory region (Table 1). No additional macrolide/lincosamide resistance genes or mutations in the 23S rRNA genes or in the ribosomal proteins L4 or L22 were identified in these isolates.

Role of point mutations in the erm(A) regulatory region in mediating the variant phenotypes

Cloning and electrotransformation of the entire erm(A) gene from isolates IT1, IT2, B1 and B2 yielded E. coli AG100A transformants that expressed an iMLS phenotype similar to that observed with the wild-type erm(A) transformants. MICs of erythromycin, clindamycin and azithromycin for these transformants ranged from 4 to 8, 1 to 2, and 1 to 4 mg/L, respectively, indicating that the high-level resistance to azithromycin observed in isolates IT1-5, B1 and B2, and additionally to clindamycin in isolate B2, was not related to the G98A, A137C and C140T mutations present in the erm(A) regulatory region. In concordance, the role of these mutations in destabilizing the erm(A) secondary structure was also found to be minor (discussed subsequently). In contrast, E. coli AG100A transformed with erm(A) from isolate B3 and harbouring the A137C and G205A mutations expressed a constitutive MLS phenotype, which correlated with the observed impact of these mutations on the erm(A) secondary structure (discussed subsequently).

Conformational isomerization of the erm(A) regulatory region and influence of point mutations

Two thermodynamically probable conformations for the wild-type 211-base erm(A) regulatory region in the uninduced state were predicted, and hairpins made by segments 4 and 5, and 6 and 7 were supported by predictions at both energy levels (Figure 1a and b). Neither structure can actively translate the methylase because the ribosome-binding site (SD3) and the initiation codon of erm(A) remain sequestered in both isomers. Induction of erm(A) begins in the presence of an inducing antibiotic by stalling of an antibiotic-bound ribosome in LP1, which frees the ribosome-binding site (SD2) of LP2. Further ribosomal stalling while translating LP2 destabilizes hairpin 6–7 and reconforms the regulatory region so that the initiation site of the erm(A) structural gene is available for an antibiotic-free ribosome to initiate methylase expression. Induction is regulated primarily by LP2 as deletion of LP1 renders the erm(A) regulatory region identical to that of erm(C) with no changes in inducibility.7 Expectedly, the LP2 coding sequence is highly conserved across the erm(A) and erm(C) genes. In fact, our attempts to predict the structure of the entire wild-type, uninduced erm(A) transcript in S. pyogenes by a comparative analysis with erm(A) and erm(C) in S. aureus were unsuccessful specifically because the regions of LP2 [LP for erm(C)] and those before the erm initiation site were too highly conserved to permit any accurate structure predictions. As illustrated in Figure S1 [available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/)], the conserved parts of the region ranging from the end of LP2 to the second codon of the methylase form a palindrome: sequence 6 and 7 is a repeat of 4 and 5 respectively which is also complementary. In the predicted structure for this region in erm(C) and for one of the structures for erm(A) in S. aureus, this palindrome forms one large hairpin by pairing 6 and 7 with 4 and 5 [Figure S2, parts A and C, available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/)]. In erm(A) in S. pyogenes and in the second structure for erm(A) in S. aureus, another configuration is predicted, in which region 4 is paired with region 5, and 6 with 7 [Figure 1 and Figure S2, part B]. Nonetheless, both these structures effectively sequester the start of the methylase. When region 4, which is part of the LP, is unavailable because of the stalling of the antibiotic-bound ribosome, a structure in which region 5 is paired with 6 is predicted for both erm(A) and erm(C). This structure is also supported by a compensating base change [C–G pair in erm(A) versus a U–A pair in erm(C)] (Figure S1, underlined in red). This structure leaves region 7 free and the gene unsequestered.

Figure 1

Predicted secondary structure of the erm(A) regulatory region in the uninduced state. (a) and (b) Conformational isomers of the wild-type erm(A) mRNA. The numbers 1–7 denote major segments of the transcript capable of forming stem-loop structures by intramolecular base pairing. The proposed Shine–Dalgarno (SD) sequences are coloured orange and SD1 and SD2 are the ribosome-binding sites for LP1 and LP2, respectively, and SD3 is the ribosome-binding site for the erm(A) structural gene. Open reading frames (ORFs) are in blue. The sites of identified point mutations are boxed in red. (c) A137C and G205A mutations (boxed and in red) present in isolate B3 abolish the hairpin formed by segments 6–7, which is consistently present in both wild-type isomers (a) and (b), freeing SD3 and the initiation codon of the erm(A) structural gene.

Figure 1

Predicted secondary structure of the erm(A) regulatory region in the uninduced state. (a) and (b) Conformational isomers of the wild-type erm(A) mRNA. The numbers 1–7 denote major segments of the transcript capable of forming stem-loop structures by intramolecular base pairing. The proposed Shine–Dalgarno (SD) sequences are coloured orange and SD1 and SD2 are the ribosome-binding sites for LP1 and LP2, respectively, and SD3 is the ribosome-binding site for the erm(A) structural gene. Open reading frames (ORFs) are in blue. The sites of identified point mutations are boxed in red. (c) A137C and G205A mutations (boxed and in red) present in isolate B3 abolish the hairpin formed by segments 6–7, which is consistently present in both wild-type isomers (a) and (b), freeing SD3 and the initiation codon of the erm(A) structural gene.

Point mutations detected here in the erm(A) gene all mapped to the highly conserved bases either in LP2 or in the vicinity of SD3 (Figure 1 and Table 1), and their role in mediating any free energy changes that might destabilize the erm(A) transcript was studied (Table 1).15 Free energy changes due to the A137C (Gln15Pro) and C140T (Pro16Leu) mutations were minor and probably insufficient to result in a stable conformation where the initiation sequence of the methylase is free, which is supported by the fact that E. coli transformed with mutant erm(A) genes from IT1, IT2, B1 and B2 expressed an iMLS phenotype. While the present study was ongoing, the C140T mutation was also reported independently in clinical erm(A)-harbouring S. pyogenes.6 Free energy changes due to the additional G98A (Gly2Asn) mutation in isolate B2 were also minor (Table 1). Previous data also show that a substitution at Gly2 does not affect inducibility with erythromycin,7 although its effect on induction with azithromycin or clindamycin remains unknown. In contrast, the G205A mutation in isolate B3 caused major changes in free energy (Table 1). In vitro studies on constitutively expressed erm(C) in S. aureus have also shown that the C192-G205 [C171-G185 in erm(C)] base pair is critical for maintaining erm inducibility and, therefore, is a binary hotspot for mutations leading to a constitutive methylase expression.15 Structure predictions of the erm(A) regulatory region in isolate B3, which exhibits a constitutive methylase expression, showed an abolishment of the hairpin 6 and 7 that released the initiation site of the erm(A) coding sequence (Figure 1c). However, surprisingly, the G205A mutation does not seem to be solely responsible for this change; this mutation alone resulted in a massively destabilized (a change in ΔG from −13.4 kcal/mol for the wild-type to −7.3 kcal/mol for the G205A mutant) although intact hairpin 6–7. The additive effect of A137C, present in hairpin 4–5, and the G205A mutations together abolished hairpin 6–7.

In conclusion, our study identified point mutations in the erm(A) regulatory region leading to constitutive methylase expression and the presence of additional, as yet unidentified mechanisms mediating high-level azithromycin and clindamycin resistance in erm(A)-harbouring S. pyogenes.

Funding

This study was partly funded by the Belgian Antibiotic Policy Coordination Committee (BAPCOC). S. M.-K. receives a post-doctoral fellowship from the Research Foundation-Flanders (FWO), Belgium.

Transparency declarations

None to declare.

Supplementary data

Figures S1 and S2 are available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/).

References

1
Malhotra-Kumar
S
Lammens
C
Chapelle
S
, et al.  . 
Macrolide- and telithromycin-resistant Streptococcus pyogenes, Belgium, 1999–2003
Emerg Infect Dis
 , 
2005
, vol. 
11
 (pg. 
939
-
42
)
2
Malhotra-Kumar
S
Lammens
C
Mallentjer
C
, et al.  . 
Effect of azithromycin and clarithromycin therapy on pharyngeal carriage of macrolide-resistant streptococci among healthy volunteers: a randomised, double-blind, placebo-controlled study
Lancet
 , 
2007
, vol. 
369
 (pg. 
482
-
90
)
3
Leclercq
R
Mechanisms of resistance to macrolides and lincosamides: nature of the resistance elements and their clinical implications
Clin Infect Dis
 , 
2002
, vol. 
34
 (pg. 
482
-
92
)
4
Weisblum
B
Insights into erythromycin action from studies of its activity as inducer of resistance
Antimicrob Agents Chemother
 , 
1995
, vol. 
39
 (pg. 
797
-
805
)
5
Fines
M
Gueudin
M
Ramon
A
, et al.  . 
In vitro selection of resistance to clindamycin related to alterations in the attenuator of the erm(TR) gene of Streptococcus pyogenes UCN1 inducibly resistant to erythromycin
J Antimicrob Chemother
 , 
2001
, vol. 
48
 (pg. 
411
-
6
)
6
Doktor
SZ
Shortridge
V
Differences in the DNA sequences in the upstream attenuator region of erm(A) in clinical isolates of Streptococcus pyogenes and their correlation with macrolide/lincosamide resistance
Antimicrob Agents Chemother
 , 
2005
, vol. 
49
 (pg. 
3070
-
2
)
7
Mayford
M
Weisblum
B
The ermC leader peptide: amino acid alterations leading to differential efficiency of induction by macrolide–lincosamide–streptogramin B antibiotics
J Bacteriol
 , 
1990
, vol. 
172
 (pg. 
3772
-
9
)
8
Clinical and Laboratory Standards Institute
Performance Standards for Antimicrobial Susceptibility Testing: Sixteenth Informational Supplement M100-S16
 , 
2006
Wayne, PA, USA
CLSI
9
Malhotra-Kumar
S
Lammens
C
Piessens
J
, et al.  . 
Multiplex PCR for simultaneous detection of macrolide and tetracycline resistance determinants in streptococci
Antimicrob Agents Chemother
 , 
2005
, vol. 
49
 (pg. 
4798
-
800
)
10
Malhotra-Kumar
S
Wang
S
Lammens
C
, et al.  . 
Bacitracin-resistant clone of Streptococcus pyogenes isolated from pharyngitis patients in Belgium
J Clin Microbiol
 , 
2003
, vol. 
41
 (pg. 
5282
-
4
)
11
Malbruny
B
Nagai
K
Coquemont
M
, et al.  . 
Resistance to macrolides in clinical isolates of Streptococcus pyogenes due to ribosomal mutations
J Antimicrob Chemother
 , 
2002
, vol. 
49
 (pg. 
935
-
9
)
12
De Rijk
P
De Wachter
R
DCSE, an interactive tool for sequence alignment and secondary structure research
Comput Appl Biosci
 , 
1993
, vol. 
9
 (pg. 
735
-
40
)
13
Zuker
M
Mfold web server for nucleic acid folding and hybridization prediction
Nucleic Acids Res
 , 
2003
, vol. 
31
 (pg. 
3406
-
15
)
14
De Rijk
P
Wuyts
J
De Wachter
R
RnaViz 2: an improved representation of RNA secondary structure
Bioinformatics
 , 
2003
, vol. 
19
 (pg. 
299
-
300
)
15
Horinouchi
S
Weisblum
B
The control region for erythromycin resistance: free energy changes related to induction and mutation to constitutive expression
Mol Gen Genet
 , 
1981
, vol. 
182
 (pg. 
341
-
8
)