Table 1 summarizes the relevant characteristics of the Colombian isolates, including isolation date, geographic origin, antibiotic susceptibility patterns, PFGE patterns, and results of hybridization of SmaI-digested DNA from PFGE gels with DNA probes. An earlier study (5) had already established that of the 43 serotype 5 invasive pneumococcal isolates, all were susceptible to penicillin, erythromycin, cefotaxime, and vancomycin; the majority (88%) showed resistance to chloramphenicol. Chloramphenicol is one of the antimicrobial agents of choice for the treatment of invasive diseases in children in Colombia (5). Additional susceptibility testing demonstrated that most of the 43 isolates, including the 38 chloramphenicol-resistant isolates, were also resistant to tetracycline, and a variable proportion were resistant to TMP-SMZ. Three of the serotype 5 isolates (CLB 32, CLB 33, and CLB 35) were susceptible to all antimicrobial agents tested, and two isolates (CLB 34 and CLB 43) were resistant only to tetracycline and not to chloramphenicol. Differences in resistance to TMP-SMZ among the serotype 5 isolates listed in Table 1 may reflect differences in local antibiotic use. A single Ile 100-to-Leu mutation in dihydrofolate reductase is known to be sufficient for trimethoprim resistance in S. pneumoniae (1), and evidence for the rapid emergence of resistance upon the introduction of TMP-SMZ into clinical use has been described (17).

All 38 serotype 5 strains showing resistance to both tetracycline and chloramphenicol represented a “Colombian clone”: a homogeneous group of S. pneumoniae isolates which were susceptible to penicillin and erythromycin, had a common PFGE pattern (X1) after SmaI restriction (Fig. (Fig.1A),1A), and hybridized with the tetM and cat DNA probes in a common 340-kb DNA fragment (Fig. (Fig.1B1B and C). A conjugative transposon, Tn5253, containing the cat and tetM genes has been described for S. pneumoniae and may explain the colocalization of the two genes in the same SmaI fragment (2). The same PFGE pattern (X1) was also identified in isolate CLB 34, which was resistant only to tetracycline and not to chloramphenicol (lane 16 in Fig. Fig.1A).1A). The 38 tetracycline- and chloramphenicol-resistant isolates and isolate CLB 34 also had a common PFGE pattern after restriction of their chromosomal DNAs with the endonuclease ApaI (data not shown). Table 1 shows that the serotype 5 isolates classified as the Colombian clone were widely dispersed in Colombia: they were recovered in 13 hospitals located in five cities and were prevalent throughout the surveillance period from early 1994 until late 1996.

Because of the genetic homogeneity of the 38 isolates representing the Colombian clone, a group of additional serotype 5 S. pneumoniae isolates with different antibiotic susceptibility profiles and different geographic origins were also examined by PFGE after digestion of the chromosomal DNAs with SmaI and ApaI. This group of serotype 5 pneumococci included four Colombian isolates (CLB 32, CLB 33, CLB 35, and CLB 43), which are listed in Table 1, four isolates from Brazil, three isolates from Argentina, and one isolate from the United States. Three of the four Colombian isolates (CLB 32, CLB 33, and CLB 35) were susceptible to both chloramphenicol and tetracycline, while isolate CLB 43 was resistant to tetracycline only (Table 1). The four Brazilian and three Argentinian isolates were susceptible to both chloramphenicol and tetracycline but were resistant to TMP-SMZ; the U.S. isolate was susceptible to all antimicrobial agents tested.

A comparison of the PFGE patterns of these additional serotype 5 isolates to the uniform PFGE pattern characteristic of the Colombian clone is shown for SmaI digestion in Fig. Fig.11 (for Colombian isolates only) and for ApaI digestion in Fig. Fig.22 (for serotype 5 isolates from several countries). After ApaI digestion, all 38 isolates of the Colombian clone had a common PFGE pattern (Y1), which is illustrated by a single representative isolate in lane 2 of Fig. Fig.2.2. CLB 35 (and CLB 32; data not shown) in lane 3 of Fig. Fig.22 (PFGE pattern Y2), CLB 33 in lane 4 (PFGE pattern Y3), and CLB 43 in lane 5 (PFGE pattern Y4) differed from the Colombian clone in three, one, and two DNA bands, respectively. The three Argentinian isolates (lanes 6, 7, and 8 of Fig. Fig.2)2) and three of the four Brazilian isolates (lanes 11, 12, and 13) had an identical and unique PFGE pattern which differed in five bands from that of the Colombian clone. The last of the four Brazilian serotype 5 isolates (ST 204 in lane 10 of Fig. Fig.2)2) differed from the Argentinian-Brazilian group in one band only (and differed from the Colombian clone in six bands). The single type 5 isolate from the United States had a PFGE pattern unrelated to those of any of the other serotype 5 isolates (the number of DNA band differences was greater than 12).

Serotype 5 S. pneumoniae was common in the United States and Europe in the 1930s (16), but since then, its frequency of isolation has declined, leading to its virtual disappearance (14). In Spain, serotype 5 pneumococci ranked 6th in order of frequency between 1979 and 1989 but only 13th in 1990 to 1996 (12, 13). Antibiotic resistance in isolates of this serotype was seen most often for tetracycline, sometimes associated with resistance to chloramphenicol or general susceptibility to all antimicrobial agents. Penicillin resistance has been rare among serotype 5 isolates (for a review, see references 13 and 22).

In contrast to the findings in Europe, the United States, and Canada, serotype 5 (and serotype 1) remained the most frequent pneumococcal serotype in Africa (3, 7, 9) and India (19) in the 1980s and 1990s, and the continued importance of serotypes 5 and 1 in other developing countries, including countries in South America, was also documented in recent studies (31, 32). In particular, serotype 5 S. pneumoniae isolates were found to be the second most frequent causative agents of invasive disease in children in Argentina (28, 29), Chile (23), and Uruguay (18) and were ranked third in Brazil (4) and ninth in Mexico (11). These observations led to the suggestion that serotype 5 should be included in a future protein conjugate vaccine (21, 32).

Support for this work was provided by the CEM/NET initiative (CEM/NET Project 31 from IBET, contract PRAXIS XXI-2/2.1/BIO/1154/95, and contract PECS/C/SAU/145/95 from JNICT) and by a grant from Fundação Calouste Gulbenkian awarded to H. de Lencastre. R. Sá-Leão was supported by grant BD/4259/96 from PRAXIS XXI from Fundação para a Ciência e Tecnologia. The work of M. Tamayo in Portugal was supported by Fundação Calouste Gulbenkian, Lisbon, Portugal; Support for M. Tamayo in Colombia came from PAHO, the Canadian Agency for International Development, and the Instituto Nacional de Salud, Santa Fe de Bogotá, Colombia.

We acknowledge Clara Inês Agudelo and María Victoria Ovalle from the Instituto Nacional de Salud and Alejandra Corso from the Instituto Nacional de Enfermedades Infecciosas, Buenos Aires, Argentina, for stimulating discussions. We express our gratitude to Alexander Tomasz of The Rockefeller University, New York, N.Y., for help in the interpretation of data and writing of the manuscript. Idalina Bonfim from Instituto de Tecnologia Química e Biológica, Oeiras, Portugal, provided technical assistance in some of the antibiotic susceptibility testing.