TEXT

Top Abstract Text References

Penicillin-resistant Streptococcus pneumoniae has become common in many areas of the world and accounts for nearly 25% of isolates in the United States (16). While there is considerable genetic diversity among penicillin-resistant S. pneumoniae in Africa (7, 11), three particular clonesthe Spanish-U.S. serotype 23F (1, 2, 6, 8, 13, 17), the Spanish-French serotype 9/14 (1, 2, 6), and the Spanish serotype 6B (12, 15, 17)have predominated in most epidemiological studies in Europe, Asia, and the Americas. Several fingerprinting techniques have been used, including pulsed-field gel electrophoresis (PFGE) (1, 2, 13, 15), multilocus enzyme electrophoresis (MLEE) (8, 12, 15), PCR-based genomic profiling (2, 6, 15), and ribotyping (7, 8, 13, 17). Comparatively few studies have compared these fingerprinting techniques; MLEE was similar to ribotyping in two studies (7, 8). PFGE, BOX fingerprinting with repetitive sequences of S. pneumoniae as the DNA probe, and restriction fragment end labeling provided the highest degree of discrimination for fingerprinting S. pneumoniae in one comparative study (5).

During two citywide surveillances conducted in 1997 and 1999, all unique patient isolates of S. pneumoniae were collected from 15 hospitals in Brooklyn, N.Y. Susceptibility testing was performed by using either the broth microdilution method with 4% lysed horse blood in cation-supplemented Mueller-Hinton broth (10) or by the E-test method with Mueller-Hinton agar with 5% sheep blood. Serotypes were determined commercially by using appropriate antisera (Statens Serum Institute [Denmark]).

Chromosomal DNA from each isolate was prepared according to established methods (12), with minor modifications. Briefly, 10 ml of an early-stationary-phase culture was centrifuged and then washed and resuspended with PIV buffer (10 mM Tris [pH 8.0], 1 M NaCl). The concentration was adjusted to an optical density at 620 nm of 5.0 and diluted 1:1 with low-melting-point agarose. The cells were placed in EC lysis solution (6 mM Tris [pH 8.0], 1 M NaCl, 0.1 M EDTA [pH 8.0], 0.2% deoxycholate, 5% Sarkosyl) with RNase (50 µg/ml) and then incubated with ES buffer (0.5 M EDTA [pH 9], 1% Sarkosyl) containing proteinase K (1 mg/ml) at 50°C overnight. The disks were washed and stored in TE buffer (10 mM Tris, 1 mM EDTA [pH 7.5]). The DNA was restricted with SmaI overnight at 21°C. The DNA was then separated by using contour clamp electrophoresis (CHEF III) for 20 h at 13°C at 6 V/cm and ramped pulse times with an initial switch time of 1 s and a final switch time of 30 s. Isolates were compared to known epidemic clones based on their PFGE patterns (1); isolates were considered related if they differed by 6 bands (14).

Ribosomal DNA from each isolate was also studied by using the Riboprinter Microbial Characterization System (Qualicon, Wilmington, Del.). Overnight cultures were treated with lysis buffer and placed into the automated Characterization System. Two restriction enzymes were studied, HindIII and PvuII. Digestion with PvuII was performed according to the manufacturer's recommendations. Digestion with HindIII (20,000 U/ml) was performed with an extended incubation (120 min at 37°C). In the automated system, restriction fragments are separated by electrophoresis and then transferred to a nylon membrane. After hybridization with a chemiluminescent DNA probe containing the rRNA operon from Escherichia coli, the banding patterns were imaged and stored in a computerized database. Isolates were immediately compared to those in the database and were considered related if they had similarity coefficients of 0.93. In this program, both band position and intensity are used to determine the similarity coefficients.

A total of 144 isolates were collected during the surveys: 29 were resistant (MIC  2 µg/ml), 24 were intermediate (MIC = 0.12 to 1 µg/ml), and 91 were susceptible to penicillin. Fingerprinting was performed with 52 isolates, including 29 resistant, 6 intermediate, and 17 susceptible isolates. Penicillin intermediate and susceptible isolates were selected to represent all of the geographic neighborhoods in the city. A total of 21 PFGE types, 25 HindIII ribotypes, and 21 PvuII ribotypes were identified. Eight isolates were identified by PFGE as belonging to cluster A, which resembled the Spanish-U.S. 23F clone (Table 1). Six were resistant to penicillin, and seven belonged to either the 23F or the 19F serotype. Six belonged to one HindIII ribotype, and two had unique ribotypes (Fig. 1). One of the unique HindIII ribotypes was a non-23 valent vaccine serotype and susceptible to penicillin, suggesting it may have had a different genetic ancestry. Seven isolates belonged to a single PvuII ribotype (Fig. 1).

View larger version (56K): [in this window] [in a new window]   FIG. 1.   Serotypes and fingerprint patterns of seven isolates of cluster A. PFGE patterns of chromosomal DNA fragments digested with SmaI (A) and ribotype profiles after digestion with HindIII (B) or PvuII (C) are shown. Nearly all isolates belonged to a single HindIII (H1) or PvuII (P1) ribotype. NV, non-23 valent vaccine serotype.

Eleven isolates (cluster B) had a PFGE pattern resembling the Spanish-French 9/14 clone (Table 1; Fig. 2); seven were resistant and two were intermediate to penicillin. Ten belonged to serotypes 9V or 14, and one was a nonvaccine serotype. Nine belonged to a single HindIII ribotype, and the remaining two each had unique ribotypes (Fig. 2). Seven belonged to a single PvuII ribotype, and the remaining four each belonged to unique ribotypes (Fig. 2).

View larger version (60K): [in this window] [in a new window]   FIG. 2.   Serotypes and fingerprint patterns of 11 isolates of cluster B. PFGE patterns of chromosomal DNA fragments digested with SmaI (A) and ribotype profiles after digestion with HindIII (B) or PvuII (C) are shown. Nearly all isolates belonged to a single HindIII (H4) or PvuII (P3) ribotype. NV, non-23 valent vaccine serotype.

Two other PFGE clusters were identified (Table 1). Cluster C included eight isolates; all were resistant to penicillin, and seven were serotype 19F. Five isolates belonged to one HindIII ribotype, and three belonged to a second ribotype. Three of these isolates belonged to one PvuII ribotype, two belonged to a second ribotype, two belonged to a third ribotype, and the remaining isolate belonged to a unique ribotype. Four isolates belonged to cluster D; three were serotype 6B and penicillin resistant. All four isolates belonged to one HindIII ribotype, but all four belonged to different PvuII ribotypes.

PFGE revealed four clusters with two isolates apiece; one pair was penicillin resistant and serotype 6B. Isolates from three of the four clusters were paired together by HindIII. Similarly, three were paired by PvuII ribotypes, but considerable overlap with other PFGE clones occurred.

Unique PFGE patterns were found for the remaining 13 isolates (Table 1). All 13 possessed unique HindIII ribotypes. Only eight unique ribotypes were identified by PvuII; the remaining five isolates were grouped with previously found ribotypes (including clusters A and B ribotypes).

PFGE (1, 2, 12) and ribotyping (5, 7, 13, 17) have been frequently used to characterize the important international clones of resistant S. pneumoniae. Ribotyping with PvuII was less discriminatory in one study (5), but only one ribosomal probe (16S) was employed. PFGE and ribotyping with PvuII were comparable in identifying 18 Spanish-U.S. 23F strains among 22 isolates (13). Ribotyping with EcoRI correlated with MLEE and penicillin-binding gene fingerprinting in a small number of isolates (7). Automated ribotyping with EcoRI was used to analyze more than 200 isolates, but a comparison to other fingerprint methods was not done (17). Ribotyping with HindIII was comparable to MLEE in distinguishing 15 resistant Spanish-U.S. 23F isolates from a small number of other multidrug-resistant isolates (8). Seven resistant serotype 23F isolates were grouped together by ribotyping with HindIII, MLEE, and PFGE (9). Similar ribotype patterns were seen among a small number of serotype 6B isolates, but comparison to other fingerprint methods was not performed (4). With PFGE as the "gold standard," we found that ribotype profiles obtained with PvuII digestion were less discriminatory than those obtained with HindIII. Ribotype patterns obtained after digestion with HindIII were comparable to and can complement those obtained by PFGE.

Fingerprinting by PFGE can be laborious, and analysis can be difficult; genetic relationships may not be evident by PFGE if only a small number of isolates are tested. Guidelines for PFGE interpretation may not be applicable to epidemiological studies involving diverse geographic areas (3). Automated ribotyping is technically undemanding, requiring only ca. 30 min for preparation. Fingerprints from as many as 32 isolates can be obtained within 24 h, providing timely results for epidemiological investigations. With the standardization of banding patterns, interinstitutional comparison of ribotype profiles can be easily and reliably performed, allowing for comparison of isolates from different geographic areas. The major disadvantage of automated ribotyping is the cost, approximately $45 per isolate.