New Mutations and Horizontal Transfer of rpoB among Rifampin-Resistant Streptococcus pneumoniae from Four Spanish Hospitals

A total of 103 (0.7%) of 14,236 Streptococcus pneumoniae isolatescollected in four Spanish hospitals from 1989 to 2003 were resistantto rifampin (MICs, 4 to 512 µg/ml). Only sixty-one (59.2%)of these isolates were available for molecular characterization.Resistance was mostly related to human immunodeficiency virus(HIV) infection in adult patients and to conjunctivitis in children.Thirty-six different pulsed-field gel electrophoresis patternswere identified among resistant isolates, five of which wererelated to international clones (Spain^23F-1, Spain^6B-2, Spain^9V-3,Spain¹⁴-5, and clone C of serotype 19F), and accounted for 49.2%of resistant isolates. Single sense mutations at cluster N orI of the rpoB gene were found in 39 isolates, while double mutations,either at cluster I, at clusters I and II, or at clusters Nand III, were found in 14 isolates. The involvement of the mutationsin rifampin resistance was confirmed by genetic transformation.Single mutations at clusters N and I conferred MICs of 2 µg/mland 4 to 32 µg/ml, respectively. Eight isolates showedhigh degrees of nucleotide sequence variations (2.3 to 10.8%)in rpoB, suggesting a recombinational origin for these isolates,for which viridans group streptococci are their potential genedonors. Although the majority of rifampin-resistant isolateswere isolated from individual patients without temporal or geographicalrelationships, the clonal dissemination of rifampin-resistantisolates was observed among 12 HIV-infected patients in thetwo hospitals with higher rates of resistance.

INTRODUCTION

Top

Introduction

A global increase in the rates of resistance of Streptococcuspneumoniae to penicillin and multiple antibacterial agents emergedworldwide in the 1980s and 1990s, severely complicating thetreatment of pneumococcal infections (12, 17, 20, 23, 24, 25).The use of rifampin (RIF) combined with either ß-lactamsor vancomycin is recommended for the treatment of meningitiscaused by multiresistant pneumococcal isolates (5, 26, 33).The rates of RIF resistance among pneumococci are low and rangefrom 0.1% in the United States (12) to 0.4% in Spain (23) andto 1.4% in Italy (25). RIF resistance is usually preceded byRIF therapy, either for tuberculosis (18) or for prophylaxisor treatment for multidrug-resistant pneumococcal infections(36, 38). RIF is also used in combined therapy to treat staphylococcalinfections, and it is extensively used in the prophylaxis ofNeisseria meningitidis exposure.

The bactericidal properties of RIF are due to its high affinityof binding to the bacterial DNA-dependent RNA polymerase andinhibit its function (6), which is essential for bacterial growth(21). Structural and biochemical studies of the essential catalyticcore of the RNA polymerase (subunit composition, ${alpha}$ 2ßß' ${omega}$ )of Thermus aquaticus have revealed that RIF interacts with apocket of the RNA polymerase ß subunit within theDNA-RNA channel and blocks the path of the elongating RNA whenthe transcript becomes 2 or 3 nucleotides long (6, 40). RIFresistance has been described in gram-positive and gram-negativebacteria. The mutations responsible for this phenotype are localizedin highly conserved regions, termed clusters N, I, II, and III,of the rpoB gene, which encodes the ß subunit (6)(Fig. 1). The residues involved in RIF resistance in severalbacteria (2, 3, 9, 19, 31) form part of the RIF-binding pocket,and 12 of these residues interact directly with the RIF molecule(6). Few studies describing RIF-resistant pneumococcal clinicalisolates have been reported (7, 15, 29, 32, 38), and all themutations identified were localized in clusters N, I, and II.In this study we report on the epidemiological and molecularcharacteristics of 61 RIF-resistant S. pneumoniae clinical isolatescollected during a 15-year period (1989 to 2003) in four Spanishhospitals.

View larger version (25K):
[in this window]
[in a new window]
FIG. 1. Regions of S. pneumoniae RpoB and mutations conferring RIF resistance. RpoB is represented as a bar showing clusters N, I, II, and III (black boxes) and sequenced regions (striped area). The PCR fragments used for transformation (black bars) and primers (black arrows) are indicated above. Amino acids that constitute clusters N, I, and II of S. pneumoniae (SPN), S. aureus (SAU), Mycobacterium tuberculosis (MTB), E. coli (ECO), and T. aquaticus (TAQ) are indicated. Amino acid changes found in resistant isolates are shown above the S. pneumoniae sequence, identical residues are indicated with an asterisk below the T. aquaticus sequence, residues that changed in RIF-resistant isolates are underlined, and the 12 residues involved in RIF binding are shaded.

MATERIALS AND METHODS

Top

Materials and Methods

Bacterial isolates and susceptibility tests. The bacteria were identified by the standard methodology, andthe serotypes were determined by the Quellung reaction. TheMICs of penicillin, erythromycin, clindamycin, tetracycline,chloramphenicol, co-trimoxazole, and RIF were determined bythe microdilution method (Sensititre commercial plates), accordingto National Committee for Clinical Laboratory Standards methods(30). The rifampin MICs of the transformants were determinedby microdilution (30) and by a macrodilution method (1) witha casein hydrolysate-based medium with 0.2% sucrose (AGCH) (22).S. pneumoniae ATCC 49619 and S. pneumoniae R6 were used forquality control.

Genetic transformation. S. pneumoniae strain R6 was grown in AGCH and was used as therecipient in transformation experiments, performed as describedpreviously (27). Two DNA fragments were used as donors: a 1,662-bpfragment (RpoB residues M1 to L554, in which the first residueof RpoB is taken as residue 1) and a 1,038-bp fragment (RpoBresidues A428 to M773). These fragments were obtained by PCRamplification from the RIF-resistant isolates and from R6, whichwas used as a control. Colonies were counted after 24 h growthat 37°C in a 5% CO₂ atmosphere in AGCH with 1% agar containing1 µg/ml of RIF.

PFGE and MLST. Genomic DNA embedded in agarose plugs was digested with SmaI,and the fragments were separated by pulsed-field gel electrophoresis(PFGE) (28). The PFGE patterns were compared to those of 26representative international clones from the Pneumococcal MolecularEpidemiology Network (28). Isolates with patterns that variedby three or fewer bands were considered to represent the samePFGE type (37). Multilocus sequence typing (MLST) was carriedout as described previously (14) with one representative RIF-resistantisolate of each dominant PFGE pattern.

PCR amplification and DNA sequence determination. The RpoB L1 to M773 region was amplified with oligonucleotidesrpob1 (5'-TTGACAAGGCTTGGAACTTAT-3') and rpob773R (5'-GTCATGTAGGCAACGAATTGGG-3').To amplify the 1,662-bp and the 1,038-bp fragments used in thetransformation experiments, oligonucleotides rpob1 and rpob554R(5'-CAAGTGTCCGTAAGATGACAAG-3') and oligonucleotides rpob428(5'-CGGTTGGTGAATTGCTTGCCAACCA-3') and rpob773R, respectively,were used. Amplifications were performed with 0.5 U of Thermusthermophilus thermostable DNA polymerase (Biotools, Madrid,Spain), 0.1 µg of chromosomal DNA, 1 µM (each) ofthe synthetic oligonucleotide primers, and 0.2 mM of each deoxynucleosidetriphosphate in the buffer recommended by the manufacturers.Amplification was achieved with an initial cycle of 1 min ofdenaturation at 94°C; 25 cycles of 30 s at 94°C, 45s at 55°C, and a 90-s or 180-s polymerase extension stepat 72°C; and a final 3-min extension step at 72°C. PCRfragments were purified with MicroSpin S400 HR columns (AmershamPharmacia Biotech, Pistcatway, NJ) and sequenced by using theoligonucleotides used in the PCR experiments and the internalprimers with an Applied Biosystems Prism 377 DNA sequencer bythe protocols provided by the manufacturer.

Nucleotide sequence accession numbers. The nucleotide sequences submitted to GenBank were assignedthe following accession numbers: AY695455 to AY695495, AY695497to AY695516, and AY785246 (for the RIF-resistant isolates);AY695496 (for Streptococcus oralis ATCC 10557); and AY785247(for S. oralis NCTC 11427).

RESULTS AND DISCUSSION

Top

Results and Discussion

Epidemiological characterization of S. pneumoniae isolates. The overall prevalence of RIF resistance among 14,236 pneumococciisolated from clinical specimens in four Spanish hospitals from1989 to 2003 was 0.7% (103 isolates with MICs $≥$ 4 µg/ml).However, geographical variations in resistance rates were found:0.3% in Hospital Central de Asturias (HCA) in the northwestof Spain, 0.4% in Hospital Universitario de Bellvitge (HUB)in the northeast of Spain (Barcelona), 1.1% in Hospital GregorioMarañón (HGM) in the central part of Spain (Madrid),and 1.1% in Hospital de Donostia (HD) in the north of Spain(Guipuzcoa). The frequency of RIF-resistant isolates in eachhospital from 1989 to 2003 is shown in Fig. 2. Two hospitals,HCA and HUB, presented low rates during the study period (0to 1.1%), whereas HD and HGM had higher rates, 6% in the 1993-1994period and 3.2% in the 1995-1996 period, respectively. Theseincreases in the rates of resistance were associated with thedissemination of four RIF-resistant clones among human immunodeficiencyvirus (HIV)-infected patients in HD and HGM. From 1999 to 2003the rates of RIF resistance were lower than 1.1% in the fourhospitals (Fig. 2).

View larger version (15K):
[in this window]
[in a new window]
FIG. 2. Frequencies of RIF-resistant isolates in four hospitals during the period from 1989 to 2003. HD, n = 4,211 isolates; HUB, n = 5,381 isolates; HCA, n = 1,056; HGM, n = 3,588.

Only 61 (59.2%) of 103 RIF-resistant isolates (1 per patient)were recovered for further studies from the stock cultures,accounting for more than 50% of RIF-resistant isolates collectedannually in each hospital. Among the 61 RIF-resistant isolatesstudied, 47 were isolated from adults (age, 21 to 76 years)and 14 were isolated from children (<15 years). The sourcesof the isolates from adults were sputum (n = 28), a conjunctivalswab (n = 1), blood (n = 11), pleural fluid (n = 2), bronchoalveolarfluid (n = 2), a protected catheter brush specimen (n = 1),pus (n = 1), and ascitic fluid (n = 1). The majority of pediatricisolates (12 of 14) were from children who had conjunctivitisand who were younger than 1 year of age, and the remaining 2isolates were from the blood of an 8-year-old HIV-infected girland from a 14-year-old child with conjunctivitis, respectively.More than half (32 of 61; 52.4%) of the RIF-resistant isolateswere from HIV-infected adult patients. Moreover, the majorityof invasive isolates (15 of 19; 79.0%) and isolates from sputum(18 of 28; 64.3%) were also from HIV-infected patients.

The in vitro activities of seven antimicrobial agents againstthe 61 RIF-resistant isolates was determined, and the resultsare summarized in Table 1. Forty-seven (77.1%) isolates wereresistant to penicillin (29 intermediate resistant and 18 resistant),and 30 (49.2%) were erythromycin resistant. Multidrug resistance(defined as resistance to three or more chemically unrelateddrugs) was detected in 41 isolates (67.2%), and 14 of them wereresistant to six drugs (penicillin, erythromycin, clindamycin,tetracycline, chloramphenicol, and co-trimoxazole). Eight ofthe RIF-resistant isolates (13.1%) were susceptible to the othersix antibiotics studied.

View this table:
[in this window]
[in a new window]
TABLE 1. In vitro activities of seven antimicrobial drugs against 61 RIF-resistant S. pneumoniae isolates

Next, the serotypes of the 61 RIF-resistant isolates was determined.Eighteen different serotypes were identified, with the followingdistribution: 6B (12 isolates), 23F (11 isolates), 14 (8 isolates),19F (6 isolates), 11 (4 isolates), 6A (3 isolates), 4 (2 isolates),9N (2 isolates), 9V (2 isolates), and 1 isolate of each of theremaining serotypes (3, 7F, 15F, 18C, 20, 21, 23A, 31, and 34).Two isolates were nontypeable.

The 61 RIF-resistant isolates were also characterized by theirPFGE patterns. A total of 36 different PFGE patterns were identified;5 PFGE patterns accounted for 30 (49.2%) isolates: Spain^23F-1(11 isolates), Spain^6B-2 (9 isolates), Spain^9V-3 (2 isolates),Spain¹⁴-5 (2 isolates), and clone C of serotype 19F (6 isolates).These four international multiresistant epidemic clones havebeen common in Spain since the 1980s (8, 16, 24, 34). The associationbetween PFGE patterns and global international clones (28) wasconfirmed by MLST of one representative isolate of each PFGEtype listed in Table 2: Rif-36 (Spain^23F-1) had sequence type81 (ST81); Rif-19 (Spain^6B-2) had ST90; Rif-55 (Spain^9V-3) hadST156; Rif-15 (Spain¹⁴-5) had ST17, which is a single-locusvariant of the reference strain with ST18; and Rif-3 (cloneC of serotype 19F) had ST89. Clone C of serotype 19F (ST89)has been identified in Spain among isolates from patients withmeningitis (16) and among ciprofloxacin-resistant isolates (10),and it has also been found sporadically in Italy (11) and Denmark(http://www.mlst.net).

View this table:
[in this window]
[in a new window]
TABLE 2. Phenotypic characteristics and changes in RpoB among the most prevalent RIF-resistant pneumococcal clones

In general, those isolates that shared the same PFGE patternhad the same rpoB polymorphisms with respect to the sequenceof strain R6 (Table 2). However, there were four exceptions:two isolates with rpoB mosaic genes (Rif-13 of the Spain^6B-2clone and Rif-31 of clone C of serotype 19F), one isolate (Rif-40of Spain^6B-2 clone) which had those polymorphisms correspondingto the Spain^23F-1 clone, and one isolate (Rif-75 of the Spain^23F-1clone) with an additional polymorphism at codon S498.

Twelve RIF-resistant isolates were isolated from 12 HIV-infectedpatients and grouped in four clusters (Table 2). One clusterincluded five isolates of the Spain^23F-1 clone (Rif-34, -35,-36, -38, and-39) isolated from invasive samples in HGM (Madrid)in 1996. The characteristics of these isolates and the presenceof 2 amino acid changes in RpoB (M488I and H499Y) suggest thecross-transmission of RIF-resistant pneumococci among HIV-infectedpatients. The remaining three clusters were identified at HD(Guipuzcoa) from 1993 to 1995 and included two isolates of theSpain^23F-1 clone (Rif-14 and -22), three isolates of the Spain^6B-2clone (Rif-18, -19, and -30), and two isolates of the Spain¹⁴-5clone (Rif-15 and -16). A temporal or geographical relationshipcould not be found among the remaining isolates, suggestingthat RIF resistance among pneumococci is mainly a sporadic eventthat occurs in individual patients.

Mapping of mutations involved in RIF resistance. Two RpoB regions, L42 to V175 (including cluster N) and Q464to T700 (including clusters I, II, and III), were sequenced(Fig. 1). Nucleotide sequence comparisons of the Q464 to T700region among RIF-resistant isolates and strain R6 revealed 53isolates with low nucleotide sequence variations ( $≤$ 0.7%) and8 isolates with high nucleotide sequence variations (2.0 to10.9%). Among the 53 isolates with low variations, 39 had singlesense mutations and 14 double sense mutations (Table 3). Singlemutations would produce amino acid changes at cluster N (Q150)or cluster I (S481, S482, Q486, D489, S495, H499, or L506) (Fig.1). The only amino acid change found at cluster N was Q150R,which is involved in low-level resistance, as reported before(29). Twelve residues of clusters I and II (shadowed in Fig.1) that are conserved in pneumococci and in other bacteria aredirectly involved in the interaction with RIF in the T. aquaticusenzyme (6). Most RIF-resistant isolates had changes at fiveof these residues (Q486, D489, H499, R501, and L506; Table 3).Position H499 was the most frequently affected (26 of 39 singlemutants), as previously found in pneumococci, either clinicalisolates (15, 32) or laboratory mutants (27), as well as inother bacteria (2, 3, 9, 19, 31). The prevalence of substitutionsat H499 could be due to the low biological cost imposed by thepresence of changes at this residue, as has been reported forthe H499N change in Staphylococcus aureus (39) and for the H499Ychange in S. aureus (39) and Escherichia coli (35). Isolateswith the highest MICs (128 µg/ml) carried the Q486L, D489V,or H499Y change (Table 3). Two of the single changes, L506Sand H499S, have not been reported before. Double mutations wouldproduce changes at one residue of cluster N and one residueof cluster III (isolate Rif-52; see below), two residues ofcluster I, or one residue of cluster I plus a residue of clusterII (Table 3). Although S481, S482, S485, M488, S495, and P537may not be in direct contact with the RIF molecule (Fig. 1),these residues are located in the vicinity of the RIF-bindingpocket. Alteration of these residues may modify the conformationof the pocket and, consequently, the binding of the antibiotic.

View this table:
[in this window]
[in a new window]
TABLE 3. Relationship between RIF MICs and amino acid changes in RpoB in 53 nonrecombinant RIF-resistant isolates

To establish the contribution of the mutations identified aboveto RIF resistance, genetic transformation experiments were performed.PCR products from the RIF-resistant isolates were able to transformthe susceptible R6 strain (MIC, 0.015 µg/ml) to resistanceat a high frequency (about 1 x 10^–2), whereas PCR productsfrom R6 transformed at frequencies 10³-fold lower (about 2 x10^–5). Transformants containing single mutations had MICsequivalent to that of the corresponding isolate, with a one-or twofold dilution margin (Table 3). The Q150R change at clusterN conferred a MIC of 2 µg/ml (T^Q150R/Rif-42), equivalentto the MIC of 4 to 8 µg/ml for the clinical isolates carryingthe same mutation (Rif-42, -52, -61, and -67). Although Rif-52carried the Q150R change plus the V638G change at cluster III,no RIF-resistant transformants with PCR products carrying V638Gwere obtained. Moreover, since the MIC of T^Q150R/Rif-42 is equivalentto that of Rif-52, it could be assumed that V638G is not involvedin RIF resistance. On the other hand, single mutations at clusterI conferred MICs of 4 to 32 µg/ml (Table 3). To determinethe contribution of the double mutations to resistance, theMICs for transformants containing one or two mutations werecompared (Table 3). These comparisons confirmed the contributionsof S485P, M488I, and L506V to resistance: the MIC of T^{S481P,S485P/Rif-59} was eightfold higher than that of T^S481P/Rif-59,and the MIC of T^{M488I, H499Y/Rif-34} was eight-fold higher thanthat of T^H499Y/Rif-12. However, the N547S and D489G changeswould not be involved in resistance, given that the MIC forT^{H499Y, N547S/Rif-11} was twofold higher than that for T^H499Y/Rif-12and the MIC for T^{D489G, L506V/Rif-18} was equal to that for T^L506V/Rif-18.On the other hand, the contributions of D489N, R501H, and P537Sto RIF resistance could not be discerned, since no transformantswith those single changes were obtained (Table 3).

Furthermore, the deduced amino acid sequences of the eight RIF-resistantrecombinant isolates contained several additional changes, besidesthe H499N and D489V changes involved in resistance (Fig. 3B).All recombinant isolates except Rif-65 shared the Y589F changepresent in Streptococcus mitis NCTC 12261, S. oralis NCTC 11427,and S. oralis ATCC 10557 RIF-susceptible isolates (Fig. 3B),indicating that this residue is not involved in resistance.Likewise, the I624V, Q671K, N623E, and N669D changes would notbe implicated in resistance since they are also present in RIF-susceptibleS. mitis and/or S. oralis strains. However, although it is notpossible to assign the contribution of each residue to resistance,the comparison of the RIF MICs conferred by the D489V and H499Nchanges to the nonrecombinant pneumococcal isolates (Table 3)suggests that these mutations are indeed responsible for theRIF resistance phenotypes of the recombinant isolates.

View larger version (48K):
[in this window]
[in a new window]
FIG. 3. Nucleotide (A) and amino acid (B) sequence variations at the RpoB T464 to T700 region of RIF-resistant recombinant isolates. The nucleotides and amino acids present at each polymorphic site are shown in full for the R6 strain, but for the other isolates, only sites that differ from those of R6 are shown. Codon numbers are indicated vertically above the sequences. Positions 1, 2, and 3 refer to the first, second, and third nucleotides in the codon, respectively. Sense mutations and amino acid changes involved in RIF resistance are shown in boldface and underlined. SPN R6, S. pneumoniae R6; SMI 12261, S. mitis NCTC 12261; SOR 10557, S. oralis ATCC 10557; SOR 11427, S. oralis NCTC 11427.

Phylogeny of isolates with high levels of nucleotide sequence variations. As pointed out above, a majority (86.9%) of the RIF-resistantisolates showed low-level nucleotide sequence variation ( $≤$ 0.7%)at their Q464 to T700 rpoB sequences with respect to the sequenceof strain R6. Similar intraspecific variations have been determinedfor atpC-atpA (<0.7%) and for the quinolone resistance-determiningregions of parC, parE, gyrA, and gyrB ( $≤$ 1%) (4). However, eightRIF-resistant isolates showed high levels of nucleotide sequencevariation (2.0 to 10.9% in the Q464 to T700 fragment and 2.3to 10.8 in the L42 to T700 fragment). The existence of S. pneumoniaeRIF-resistant isolates with high levels of nucleotide sequencevariations have been described previously (7, 15, 32). Thisvariation was in accordance with the divergence found betweenS. pneumoniae and S. oralis for the amylomaltase gene (4 to6%) (13) and between S. pneumoniae and viridans group streptococcifor the parE gene ( $≥$ 8.5%) (4). High levels of nucleotide sequencevariations (Fig. 3) suggested that these isolates would havea mosaic structure in rpoB as a consequence of recombinationwith viridans group streptococci. To establish the mosaic structure,a 1,977-bp region (residues L42 to T700) was sequenced and comparedwith the sequence of R6. Five isolates (Rif-13, -15, -16, -25,and -56) showed a continuous block of divergence, suggestingthat the recombination points are located outside the regionanalyzed (Fig. 4). Isolates Rif-24 and Rif-65 showed two blocks,whereas isolate Rif-31 showed three blocks. The blocks withdivergence lower than 1% could represent the recombination sites.

View larger version (25K):
[in this window]
[in a new window]
FIG. 4. The pneumococcal recombinant isolates interchanged parts of their rpoB genes with viridans group streptococci. (A) Mosaic structure of a 1,977-bp region (L42 to T700) of rpoB. The divergence of each block with respect to the strain R6 sequence is indicated. (B) Phylogenetic tree of a 357-bp region that includes RpoB residues A474 to A592 in which all isolates showed nucleotide sequence variations in the 4% to 9% range. Phylogenetic and molecular evolutionary analyses were conducted with the MEGA program (version 2.1) by the neighbor-joining method. Only bootstrap confidence intervals exceeding 90% are shown.

To establish the origin of the gene donor to the Rif-65 isolate(the isolate showing the highest divergence with respect tothe sequence of R6 strain, 10.8%), several WU-Blast2 analysesof the EMBL database were performed. The sequences more similarto the rpoB sequence of Rif-65 were chosen to perform a phylogeneticanalysis. Since the block structure of the rpoB genes revealeda common variable region of 357 bp (A474 to A592) for the eightrecombinant isolates (Fig. 4A), this fragment was used for thedetection of different clusters within the isolates. All recombinantisolates except Rif-65 formed a monophyletic group that includedS. pneumoniae strains and viridans group streptococcal strains(S. oralis and S. mitis) (Fig. 4B). These results suggest thatthe donors in the horizontal transfer of rpoB genes to S. pneumoniaeare the viridans group streptococci belonging to S. mitis orS. oralis species, as described for the penicillin- and fluoroquinolone-resistancedeterminants (10, 13). Three of these recombinant isolates belongedto the Spain¹⁴-5 and Spain^6B-2 clones (Table 2), suggestinga possible dissemination of RIF resistance through these clones.

In conclusion, the incidence of RIF resistance among S. pneumoniaeisolates is rare in Spain and was mainly related to HIV-infectedpatients and children with conjunctivitis, suggesting priortreatment with this drug. Although cross-transmission of RIF-resistantisolates was demonstrated among HIV-infected patients, the majorityof RIF-resistant isolates were isolated from individual patientswithout temporal or geographical relationships. This resistancewas acquired either by point mutations in the rpoB gene or byrecombination with viridans group streptococci. Continuous surveillanceof resistance to rifampin among invasive pneumococci is important,because in serious pneumococcal infections, a combination ofrifampin and broad-spectrum cephalosporins or vancomycin isrecommended.

ACKNOWLEDGMENTS

This work was supported by Red Temática de InvestigaciónCooperativa G03/103 from Fondo de Investigación Sanitariaand by grant BIO2002-01398 from the Ministerio de Ciencia yTecnología. We acknowledge the use of the pneumococcalMLST database, which is located at Imperial College London andwhich is funded by the Wellcome Trust.

FOOTNOTES

* Corresponding author. Mailing address: Unidad de Genética Bacteriana, Centro Nacional de Microbiología, Instituto de Salud Carlos III, 28220 Majadahonda, Madrid, Spain. Phone: (34) 91-5097057. Fax: (34) 91-5097919. E-mail: agcampa{at}isciii.es.

Spanish Pneumococcal Infection Study Network G03/103: generalcoordination, Román Pallarés; participants andcenters, Ernesto García (Centro de Investigaciones Biológicas,Madrid); Julio Casal, Asuncion Fenoll, and Adela G. de la Campa(Centro Nacional de Microbiología, Instituto de SaludCarlos III, Madrid); Emilio Bouza (Hospital Gregorio Marañon,Madrid); Fernando Baquero (Hospital Ramón y Cajal, Madrid);Francisco Soriano and José Prieto (Fundación JiménezDíaz y Hospital Clínico, Madrid); RománPallarés and Josefina Liñares (Hospital Universitaride Bellvitge, Barcelona); Javier Garau and Javier MartínezLacasa (Hospital Mutua de Terrassa, Barcelona); Cristina Latorre(Hospital Sant Joan de Deu, Barcelona); Emilio Pérez-Trallero(Hospital Donostia, San Sebastian); Juan García de Lomas(Hospital Clínico, Valencia); and Ana Fleites (HospitalCentral de Asturias).

REFERENCES

Top