Background
Streptococcus pyogenes (group A streptococcus) is an important pathogen causing pharyngitis, scarlet fever, impetigo, erysipelas, cellulitis and necrotizing fasciitis. S. pyogenes throat infection can lead to suppurative complications like peritonsillar cellulitis and abscesses 1. Nonsuppurative complications of streptococcal pharyngitis, acute rheumatic fever and glomerulonephritis are rare 2. S. pyogenes can also colonize the throats of asymptomatic persons 1. The treatment of S. pyogenes infections relies on penicillin 3, which has retained its antimicrobial activity and is the drug of choice 4. Macrolide antibiotics are used to treat patients with allergies to penicillin. In contrast to penicillin, resistance to macrolides among S. pyogenes strains has become an increasing problem 56.
There are two main acquired macrolide resistance mechanisms in S. pyogenes, posttranscriptional target site modification and macrolide efflux 6. Target site modification is mediated by erythromycin resistance methylase (erm) genes. In S. pyogenes, two different Erm methylases have been found. Erm(B) methylase encoded by the erm(B) gene belongs to the Erm(B) class of rRNA methylases, and causes resistance to macrolides, lincosamides and streptogramin B antibiotics, i.e. the MLSB type of resistance. Erm(TR) methylase belongs to the Erm(A) class of rRNA methylases, and also causes the MLSB type of resistance 7. However, the type of resistance depends also on the expression type of the Erm methylase. erm(B) and erm(TR) genes of the inducible type usually cause resistance only to erythromycin and other 14- and 15- membered-ring macrolides but not to lincosamides and streptogramin B antibiotics 5. After induction of the expression of the gene by sub-inhibitory concentration of erythromycin, these genes cause resistance also to lincosamides and streptogramin B antibiotics 8. The other main acquired resistance mechanism, macrolide efflux, is mediated by mef(A) resistance genes 7. mef(A) genes cause the M type of resistance, i.e. resistance against 14- and 15- membered-ring macrolides 9. In addition to mef(A) mediated macrolide efflux, Giovanetti et al. 10 have demonstrated the presence of a putative novel efflux system, associated with erm(TR) gene in S. pyogenes. The gene or genes behind this erythromycin efflux system are not yet known.
Mutations at the ribosomal proteins (L4 and L22) and 23S rRNA are known to cause macrolide reistance in streptococci 1112. In S. pyogenes, mutations at the ribosomal protein L4 and a mutation at the position 2611 of the 23S rRNA as a cause of macrolide resistance have been described before 1314. In this work, we describe one clinical S. pyogenes strain resistant to macrolides, lincosamides, streptogramin B antibiotics (quinupristin) and telithromycin with none of the known acquired erythromycin resistance methylases or macrolide efflux pumps, but having a mutation at the position 2058 of the 23S rRNA. This mutation has not been described before to cause macrolide resistance in S. pyogenes.
Material and methods
Bacterial strains
Macrolide-resistant Streptococcus pyogenes strain ni4277 was isolated from a throat specimen in the clinical microbiology laboratory of the Helsinki University Central Hospital. Identification of the strain was based on the colony morphology and hemolysis on Blood Agar Base (Oxoid Ltd., Basingstoke, Hampshire, England) plate supplemented with 7,5% sheep blood as well as resistance to bacitracin (A/S Rosco, Taastrup, Denmark) and positive reaction in group A antigen test (Streptex ZL51, Murex Biotech Ltd., Dartford, England). The identification was further confirmed by sequencing the 23S rRNA genes (see results and discussion). S. pyogenes strain ATCC 700294, the whole genome of which has been sequenced, was purchased from the American Type Culture Collection and was used as a macrolide-susceptible control strain. Other bacterial strains with different macrolide resistance determinants were Escherichia coli with plasmid pJIR (erm(B)), Staphylococcus simulans 13044 with plasmid pPV142 (erm(C)), E. coli with plasmid pEM9592 (erm(A)), S. pyogenes A200 (erm(TR)), S. pyogens A569 (mef(A)) and Staphylococcus epidermidis A33 (msr(A)). These strains were used as positive controls in different PCRs 12.
MIC testing and phenotyping
MIC testing was done using the agar-dilution technique 15. The bacteria were cultured 20 hours in air at 35°C on Mueller-Hinton II (Becton Dickinson Microbiology Systems, Cockeysville, MD, USA) agar plates supplemented with 5% sheep blood. The antibiotics used were: clindamycin, josamycin, telithromycin, HMR3004 (Aventis Pharma, Romainville Cedex, France), azithromycin (Pfizer, Groton, CT, USA), quinupristin, spiramycin (Rhône-Poulenc Rorer, Vitrysur-Seine, France), erythromycin, lincomycin (Sigma-Aldrich Chemie, Gmbh, Steinheim, Germany). The double-disk method with erythromycin (diffusible content 78 μg) and clindamycin (diffusible content 25 μg) (Rosco Neo-Sensitabs, A/S Rosco, Taastrup, Denmark) discs was used for classification of macrolide resistance phenotypes. The disks were placed 15–20 mm apart on Mueller-Hinton II agar plates supplemented with 5% sheep blood. Bacteria were cultured 20 hours at 35°C. After incubation, blunting of the clindamycin zone of inhibition proximal to the erythromycin disk was interpreted to indicate inducible MLSB resistance. NCCLS breakpoints, if available, were used for interpretation of the MIC results 16.
PCR detection of macrolide resistance determinants
The presence of macrolide resistance genes erm(B), erm(C), erm(A), erm(TR), mef(A) and mrs(A) were determined by PCR as described earlier 12. Amplification of the six ribosomal RNA (rRNA) operons were done using long-range PCRs. Six different PCRs were needed for amplification of six different rRNA operons. The PCRs were identical except for the operon-specific primer. For each rRNA operon, there was one common primer, located at the beginning of the 23S rRNA molecule, and one operon-specific primer (table 1). To obtain high quality DNA for long-range PCR, bacterial cells were cultivated overnight in 10 ml of BHI broth (BD Bioscience, Sparks, MD, USA). Cells were collected by centrifugation and the DNA was isolated as described before 8. Long-range PCRs were done using DyNazyme Ext DNA polymerase and the buffer supplied with the polymerase (Finnzymes OY, Espoo Finland). The reaction volume was 50 μl, containing 200 ng of the template DNA, 0,5 μM of operon-specific primer and 0,5 μM of common 23S rRNA primer (table 1), 3 U of the polymerase and 200 μM of each nucleotide. The MgCl2 concentration was 1,5 mM. PCR conditions were as follows: 1 minute at 94°C (one cycle), 20 seconds at 94°C, 30 seconds at 58°C and 3 minutes at 72°C (35 cycles). Amplification was performed using a PTC-200 (MJ Research, Massachusetts, USA). The rRNA operon specific PCR primers were designed using GeneComposer v. 1.1 17 and the genome sequence of the S. pyogenes (accession number AE004092) strain SF370 (same as ATCC 700294) 18.
<p>Table 1</p>
PCR and sequencing primers
Primer
Target gene
Position
Use
Sequence (5'-3')
Reference
23Scom a
23S rRNA
6-26b
PCR/Seq
AGTTAATAAGGGCGCACGGTG
This work
23Sg1 c
23S rRNA gene1
1577119-157741d
PCR/Seq
AAGACGTATTGAAGCTTACTCTA
This work
23Sg2 c
23S rRNA gene2
22879-22857 d
PCR/Seq
AGCTACTTCCCGAACTGATGCAC
This work
23Sg3 c
23S rRNA gene3
29128-29106 d
PCR/Seq
GATAACCGTCTTCTTTCCCCTTT
This work
23Sg4 c
23S rRNA gene4
85192-85170 d
PCR/Seq
CATATTTCTAACACGGGCAGTAG
This work
23Sg5 c
23S rRNA gene5
270607-270582 d
PCR/Seq
CACTGCCAAGCTATCTAAACGTA
This work
23Sg6 c
23S rRNA gene6
1330037-1330059d
PCR/Seq
CAATTGAATAGCCTGCACGTTCG
This work
MS2 e
23S rRNA
550-573b
Sequencing
CGTGTGCCTACAACAAGTTCGAGC
This work
MS3 e
23S rRNA
573-550b
Sequencing
GCTCGAACTTGTTGTAGGCACACG
This work
MS4 e
23S rRNA
1016-997b
Sequencing
TTTCGACTACGGATCTTAGC
[11]
MS5 e
23S rRNA
867-885b
Sequencing
CTGTTTGGGTGAGGGGTCC
[11]
MS6 e
23S rRNA
1424-1446b
Sequencing
GGACGCAGTAGGCTAACTAAACC
This work
MS7 e
23S rRNA
1446-1424b
Sequencing
GCTTTAGTTAGCCTACTGCGTCC
This work
MS8 e
23S rRNA
2021-2040b
Sequencing
ACCTGCATCTTCACAGGTAC
[11]
MS9 e
23S rRNA
1902-1921b
Sequencing
CGGCGGCCGTAACTATAACG
[11]
MS10 e
23S rRNA
2415-2437b
Sequencing
GCCATCGCTCAACGGATAAAAGC
This work
MS11 e
23S rRNA
2437-2415b
Sequencing
GCTTTTATCCGTTGAGCGATGGC
This work
MS12 e
23S rRNA
2904-2880b
Sequencing
TTGGATAAGTCCTCGAGCTATTAG
[11]
L4fu
L4
64481-64498c
PCR/Seq
GCAAACACAGCTCCTAAG
This work
L4rd
L4
65751-65734c
PCR/Seq
CTGTGTAACGACCAACGC
This work
L4f156
L4
65074-65091c
Sequencing
AACCGTTCAGCAGTATCA
This work
L4r444
L4
65307-65290c
Sequencing
TGAAAGGCCTTCTACAGC
This work
L22-1
L22
67079-67098c
PCR/Seq
GCAGACGACAAGAAAACACG
[11]
L22-2
L22
67500-67481c
PCR/Seq
ATTGGATGTACTTTTTGACC
[11]
a A PCR primer common to all 23S rRNA genes b Position in the 23S rRNA molecule (E. coli accession number J01695, numbering). c rRNA operon spesific PCR primers, artificial numbering of the 23S rRNA genes of the S. pyogenes d Position in the S. pyogenes genome (accession number AE004092). e23S rRNA gene sequencing primers
Amplification of the genes coding for ribosomal proteins L4 and L22 was done using DyNazyme DNA polymerase and buffer supplied with the polymerase (Finnzymes OY, Espoo Finland). The reaction volume was 100 μl, containing 200 ng of the template DNA, 0,2 μM of each primer (table 1), 1 U of the polymerase and 200 μM of the each nucleotide. MgCl2 concentration was 1,5 mM. PCR conditions were as follows: 1 minute at 94°C, 1 minute at 55°C and 1 minute at 72°C (35 cycles). Amplification was performed using the PTC-200.
Sequencing of the 23S rRNA, L4 and L22 genes
Sequencing was performed using an ABI Prism BigDye Terminator Kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's instruction. Sequencing primers are presented in table 1. Sequences were handled using Vector NTi software (InforMax Inc, Bethesda, MD, USA) and the GeneComposer software. Comparisons to bacterial genome sequences were done with the sequence similarity and homology searching tools (Proteomes & Genomes Fasta) using a www-interface to the European Bioinformatics Institute genome database 19.
Accession numbers for the 23S rRNA gene sequences of the strains ni4277 and ATCC 700294, submitted to the EMBL nucleotide sequence database are AJ544681 and AJ544682, respectively. Accession numbers for the genes coding for L4 and L22 proteins of the strains ni4277 are AJ544683 and AJ544684, respectively.
Results
The S. pyogenes strain ni4277 was highly resistant to all macrolides and azalides. MICs were: erythromycin >256 μg/ml, spiramycin 64 μg/ml, josamycin 64 μg/ml, azithromycin >256 μg/ml. This strain had elevated MICs for ketolides (telithromycin MIC 16 μg/ml and HMR3004 MIC 4 μg/ml). At the moment, NCCLS does not have breakpoints for telithromycin 16. However, the high MIC value suggests that this strain is resistant to telithromycin. It was also resistant to lincosamides (clindamycin MIC 16 μg/ml and lincomycin MIC > 64 μg/ml), and it showed elevated MIC for streptogramin B antibiotic, quinupristin MIC 4 μg/ml. When tested with the erythromycin-clindamycin double-disk method, the resistance phenotype in the strain ni4277 was non-inducible. The S. pyogenes strain ATCC 700294 was used as a macrolide susceptible control strain. The susceptibility of this strain was confirmed by MIC testing. MICs for various antibiotics were: erythromycin 0,063 μg/ml, spiramycin 0,25 μg/ml, josamycin 0,25 μg/ml, azithromycin 0,125 μg/ml, telithromycin 0,008 μg/ml, HMR3004 0,008 μg/ml, clindamycin 0,031 μg/ml, lincomycin 0,063 μg/ml and quinupristin 0,25 μg/ml. Thus, the MIC values for macrolides, lincosamides, ketolides and quinupristin were much higher in the strain ni4277 than in the strain ATCC 700294. The strain ATCC 700294 has no macrolide resistance determinants in its genome. Neither has it known macrolide-causing mutations in the ribosomal proteins (L4 or L22) or 23S rRNA 18.
In S. pyogenes, three macrolide resistance determinants have been described thus far: erm(B), erm(TR) and mef(A) 6. Using PCR conditions that gave positive results from positive control strains, none of these known macrolide resistance genes was found in the S. pyogenes strain ni4277. In addition, PCRs for macrolide resistance genes (erm(A), erm(C) and msr(A)), known to cause resistance in other gram-positive bacteria gave negative results, although PCRs from corresponding control strains were positive.
S. pyogenes has six operons coding for ribosomal RNA molecules 18. Based on the genome sequence data, primers specific for each rRNA operon were designed (table 1). In order to confirm the specificity of each operon-specific PCR, a short fragment of each PCR product was sequenced using the corresponding operon-specific primer. The sequences were compared to the genome sequence of S. pyogenes. These confirmations were done for macrolide-resistant strain ni4277 and for the strain ATCC 700294, the genome sequence of which is available 18. These sequences confirmed that each operon specific PCR amplified a different ribosomal operon.
The operon-spesific PCRs were used to amplify all the six rRNA operons of the macrolide-resistant strain ni4277. Both DNA strands of the 23S rRNA genes of each six operons were sequenced using 23S rRNA specific sequencing primers (table 1). At least 2854 nucleotides between positions 52 and 2904 of the 23S rRNA gene (E. coli, accession number J01695, numbering) from each of the six 23S rRNA coding genes were sequenced. When the six 23S rRNA gene sequences of the different rRNA operons of the strain ni4277 were compared to each other, five of the genes were found to be identical. The remaining 23S rRNA gene differed from the other five genes in one position. All the five 23S rRNA genes had guanine residue at the position 2058 (E. coli numbering), whereas adenine residue was found at the corresponding position in the one operon. So, in five of the six 23S rRNA genes of the macrolide-resistant strain ni4277, there was adenine to guanine mutation at the position 2058. This mutation is known to cause macrolide resistance, for example, in S. pneumoniae 11. The sequencing of all the six 23S rRNA genes of the strain ni4277 was done twice. Both times identical sequences were obtained from different operons (five operons with guanine at the position 2058 and one operon with adenine at the corresponding position).
Operon-specific PCRs were also used to amplify 23S rRNA genes of the macrolide-susceptible control strain ATCC 700294. Both DNA strands of all the six 23S rRNA genes of this strain were sequenced. When the sequences (at least 2854 nucleotides between positions 52 and 2904) of these six 23S rRNA genes were compared to each other, they were found to be identical. All the six 23S rRNA genes of this macrolide-susceptible strain had adenine at the position 2058 (E. coli numbering). These sequences confirmed that the macrolide-susceptible strain ATCC 700294 has no known macrolide causing mutation at the 23S rRNA, and that the mutations found at the 23S rRNA genes of the strain ni4277 were not caused by a systematic sequencing error.
In addition to the difference at the position 2058 of the 23S rRNA, another difference between the 23S rRNA gene sequences of the macrolide-resistant strain ni4277 and the macrolide-susceptible strain ATCC 700294 was observed. At the position 2163 (E. coli numbering) there was a cytosine in all the six 23S genes of the strain ni4277, but thymine at the corresponding position in all the six 23S genes of the strain ATCC 700294. This position is not known to be linked to macrolide resistance 20.
Genes coding for ribosomal proteins L4 and L22 were sequenced, and the sequences were compared against complete genome sequences of S. pyogenes using Fasta. Comparisons showed that both ribosomal proteins of the strain ni4277 were identical to the published wild-type L4 and L22 sequences at the amino acid level 18. No mutations known to cause macrolide resistance in S. pyogenes were found 1314.
Discussion
We studied the presence of efflux pumps in the strain ni4277 by PCR. mef(A) or mrs(A) genes were not found. This does not exclude, a previously unknown efflux mechanism as the cause of resistance. However, the macrolide resistance phenotype in the strain ni4277 is not typical for efflux pumps, which are rather specific and do not cause resistance against all antibiotics that bind to 23S rRNA 9. When compared to control strain ATCC 700294, the strain ni4277 had high MIC values for all macrolides, including 14- and 16- membered-ring macrolides, one azalide antibiotic, two different lincosamides, one streptogramin B antibiotic and two different ketolides. These antibiotics have different structures, but they all have common target site, the bacterial ribosome. All these antibiotics bind to or at least are in close contact with the 23S rRNA. So, most likely, target site modification rather than efflux mechanism is the cause of resistance in the strain ni4277.
The S. pyogenes strain ni4277 carries two mutations in the 23S rRNA genes (positions 2058 and 2163) when compared to the corresponding genes from a macrolide-susceptible S. pyogenes strain ATCC 700294. The other mutated position, 2058, is located at the very conservative part of the domain V of the 23S rRNA molecule. Adenine at the corresponding position can be found in 23S rRNA molecules of almost all bacteria 2021. The domain V and the adenine at the position2058 are known to be important for the binding of the macrolides to ribosomes. In bacteria, mutations at this position are always linked to the MLSB type of resistance or at least erythromycin resistance 20. The macrolide resistance pattern of the strain ni4277 was of the MLSB type. It was resistant to macrolides and clindamycin, and it had elevated MIC for streptogramin B (quinupristin) when compared to control strain ATCC 700294. In addition, the strain ni4277 had elevated MIC values for ketolides. In general, the phenotype of the strain ni4277 is very similar to the phenotype of S. pyogenes strains with constitutively expressed erm(B) methylase: highly resistant to all macrolides, lincosamides, resistant to streptogramin B antibiotics and, the most importantly, resistant to ketolides 5. This phenotypic similarity supports our finding. If the adenine at the position 2058 is mutated, this mutation produces similar phenotype as if the adenine at the corresponding position were methylated. Erm(B) and erm(TR) methylases as well as other known erythromycin resistance-causing methylases are known to methylate adenine at the position 2058 in streptococci and in other bacteria 522. In the strain ni4277 neither erm(B) gene nor any other methylase was found by PCR. Also, it is unlikely that a methylase would be the cause of the resistance in the strain ni4277, because the most of the target sites of the putative methylase were mutated. So the observed mutations at the position 2058 of the 23S rRNA genes are the most likely reason for MLSB type of resistance and for elevated MICs for ketolides. The other mutated position 2163 is not evolutionary conserved. According to the Comparative RNA Web Site, 31,40%, 13,64%, 52,48% of the bacteria have adenine, cytosine or guanine at this position, respectively, indicating that variation at this position is a normal phenomenon among bacteria 21. Nor is this position known to interact with macrolides or been associated with macrolide resistance 20. So it is unlikely that mutation at the position 2163 is the cause of the macrolide resistance.
The strain ni4277 had elevated MICs for ketolides, when compared to the control strain ATCC 700294. Telithromycin MIC 16 μg/ml was so high, that although there are no breakpoints available for telithromycin, this strain is likely to be telithromycin resistant. This finding indicates that a S. pyogenes strain can become telithromycin resistant, without acquiring external genetic material. The previously described point mutation, cytosine to thymine, at the position 2611 of the 23S rRNA of S. pyogenes 14 does not result in telitromycin resistance. Based on our knowledge, this is the first description of a S. pyogenes strain with adenine 2058 to guanine mutation as a cause of macrolide resistance. This strain was isolated from a throat specimen. Thus, our finding also indicates that this kind of mutation in S. pyogenes 23S rRNA genes can arise in a clinical situation.
Conclusions
Known acquired erythromycin resistance methylases or macrolide efflux pumps were not found in the S. pyogenes strain ni4277. The ribosomal proteins L4 and L22 in this strain were identical to corresponding genes of macrolide-susceptible S. pyogenes strain. The macrolide resistance phenotype of this strain was similar to that of other bacteria with adenine at the position 2058 modified or mutated. Based on this data we conclude that the detected mutation, adenine to guanine at the position 2058, in five of the six 23S rRNA-coding genes, was the most probable cause of macrolide resistance in the strain ni4277. Based on our knowledge, this is the first description of this mutation as the cause of macrolide, lincosamide and ketolide resistance in a clinical isolate of S. pyogenes.
List of abbreviations
MIC – minimal inhibitory concentration; erm – erythromycin resistance methylase; mef – macrolide efflux; rRNA – ribosomal RNA
Authors' contributions
JJ carried out the MIC and molecular genetic studies. MV carried out the isolation and identification of the strain. PH participated in the design and coordination of the study. All authors read and approved the final manuscript.