Altered PBP 2A and Its Role in the Development of Penicillin, Cefotaxime, and Ceftriaxone Resistance in a Clinical Isolate of Streptococcus pneumoniae

Abstract

We report the unusual involvement of altered PBP 2A in the development of β-lactam resistance in Streptococcus pneumoniae. This was investigated amid three identical serotype 14 isolates (designated isolates 1, 2, and 3, respectively) of pneumococci cultured successfully from the blood of a human immunodeficiency virus-seropositive child with recurrent pneumonia. The passage of this strain through its human host induced several changes in the bacterium, which is typical of the adaptive and evolving nature of the pneumococcus. An efflux resistance mechanism, which conferred increased ciprofloxacin resistance, was induced in isolates 2 and 3. In addition, faster growth rates and larger capsules were also observed for these isolates, with respect to isolate 1. Notably, compared to isolates 1 and 2, isolate 3 showed a decrease in penicillin, cefotaxime, and ceftriaxone resistance. This change was associated with the replacement of an altered PBP 2A for an unaltered PBP 2A. In all likelihood, these events produced a strain which evolved into a fitter and more virulent type, isolate 3, that resulted in an aggravated pneumococcal infection and ultimately in the patient's death.

β-Lactam antibiotics inhibit the growth of pneumococci by inactivation of cell-wall-synthesizing penicillin-binding proteins (PBPs). Pneumococcal resistance to β-lactams is essentially due to a complex production of altered PBPs with decreased affinities for these antibiotics (6, 19, 23, 30). A functional MurMN operon is required for expression of PBP-mediated β-lactam resistance in the pneumococcus (4, 29). In conjunction with altered PBPs, alteration in MurM may assist in the development of β-lactam resistance (22). Pneumococci contain a set of six PBPs. High-molecular-weight (80- to 90-kDa) PBPs 2X, 2B, 1A, 1B, and 2A catalyze the essential transpeptidase reaction (peptidoglycan cross-linking) and are the lethal targets for β-lactams. Low-molecular-weight PBP 3 (45 kDa) only has carboxypeptidase activity, and its inhibition is tolerated. Altered PBPs 2X, 2B, and 1A are the major players in the development of β-lactam resistance in clinical isolates (1, 15). These altered PBPs are encoded by genes of a mosaic organization in which parts of the genes are replaced by allelic variants that differ by up to 25% in DNA sequence (3, 10, 14). On the other hand, very little data exist for PBPs 1B, 2A, and 3 as β-lactam resistance determinants. Hakenbeck and coworkers (5) showed that pneumococcal PBPs 1B and 2A were changed into variants of low penicillin affinity by successive transformation with chromosomal donor DNA from a high-level β-lactam-resistant Streptococcus mitis isolate. When a cloned pbp2A gene was used as donor DNA, it was able to transform a strain from a cefotaxime MIC of ±0.2 μg/ml to ±1.5 μg/ml, therefore proving that altered PBP 2A can contribute to cefotaxime resistance (5). In addition, other studies have reported low-affinity PBP 1B in penicillin-resistant transformants (2), while low-affinity PBP 2A has been reported in penicillin-resistant isolates (10) and cefotaxime-resistant transformants (19). Furthermore, a mutation in the carboxypeptidase domain of PBP 3 has been shown to confer cefotaxime resistance in a laboratory-generated strain (9). Therefore, the potential clearly exists for PBPs 1B, 2A, and 3 to play a role in the development of β-lactam resistance in the clinical setting.

In this work, we show the unusual involvement of altered PBP 2A in the development of β-lactam resistance in the pneumococcus. This was investigated amid three identical serotype 14 isolates of pneumococci cultured successfully from the blood of a human immunodeficiency virus-seropositive child with recurrent pneumonia. We further report on the adaptive and evolving nature of this strain as it passaged through its human host. This was displayed by changes in antibiotic susceptibility, growth rate, and capsule size.

MATERIALS AND METHODS

Origin and properties of isolates.

On 12 February 2002, a human immunodeficiency virus-seropositive 7-year-old female child was admitted to a university-affiliated hospital in Johannesburg, South Africa, and treated for pneumonia. Following a complicated course of recurrent episodes of pneumonia, she eventually died after her last hospital admission on 24 March 2002. Antibiotic therapy during the course of her recurrent infections, prior to the last hospital admission, included ampicillin (12 to 14 February), cefuroxime (14 to 19 February), ciprofloxacin (19 February to 2 March), and cefotaxime (2 March to 12 March). Blood cultures taken on 12 February, 27 February, and 23 March during each recurrent episode of infection yielded a strain of serotype 14 pneumococcus, which were designated isolates 1, 2, and 3, respectively. Pneumococci were routinely cultured at 37°C in 5% CO₂ on Mueller-Hinton agar (Oxoid, Hampshire, England) supplemented with 5% lysed horse blood (blood agar). Serotyping was performed with the Quellung reaction using capsule type-specific antiserum provided by the Statens Seruminstitut, Copenhagen, Denmark. For the Quellung reaction, bacterial cells were taken from a fresh blood agar plate, resuspended in saline, and air dried on a microscope slide followed by the placement of a coverslip coated with antiserum and methylene blue. Capsular reactions were observed with phase-contrast microscopy using a Nikon E-400 microscope equipped with a Plan-Apo ×100 oil lens. Photographs were taken with a Nikon DS-5 M-U1 camera. Images were processed using the MediaCybernetics Image-Pro plus software. Antibiotic MICs for isolates were determined by the Etest (AB BIODISK, Solna, Sweden) and the agar dilution method (16). The MICs for isolates are shown in Table 1.

TABLE 1.

Antibiotic MICs for clinical isolates

Antibiotic	MIC (μg/ml) for isolate
	1	2	3
Penicillin	1.5	1.5	0.38
Cefotaxime	1	1	0.25
Ceftriaxone	0.5	0.5	0.19
Cefuroxime	8	8	8
Ciprofloxacin	0.5	1	2

Genotyping, DNA isolation, PCR, and DNA sequencing.

Pneumococcal isolates were genetically fingerprinted using previously described methods of pulsed-field gel electrophoresis (PFGE) (11) and BOX-PCR fingerprinting (8). Chromosomal DNAs were extracted from bacterial cells and pbp genes were amplified by PCR from the chromosomal DNAs with methods that have been described previously (22-25). For PCR of gyrA, gyrB, parC, and parE genes, the PCR was 25 cycles of denaturation at 95°C for 1 min, primer annealing at 52°C for 2 min, and primer extension at 72°C for 2 min. PCR primer sequences were as follows: gyrA-forward (ACCGTCGCATTCTCTACG), gyrA-reverse (CAGTTGCTCCATTAACCA), gyrB-forward (TTCTCCGATTTCCTCATG), gyrB-reverse (AGAAGGGTACGAATGTGG), parC-forward (TGGGTTGAAGCCGGTTCA), parC-reverse (TGCTGGCAAGACCGTTGG), parE-forward (AAGGCGCGTGATGAGAGC), and parE-reverse (TCTGCTCCAACACCCGCA). PCR products were purified from agarose gel using a Geneclean kit (Bio 101 Inc., La Jolla, CA) and sequenced using a BigDye Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, CA) and an Applied Biosystems model 310 automated DNA sequencer.

Growth studies.

For physiological studies, isolates were inoculated from glycerol stocks into tryptone soy broth (TSB; Oxoid), prewarmed at the appropriate temperature, and incubated at 37°C or 40°C in an atmosphere of 5% CO₂. Growth was monitored turbidimetrically at 650 nm every 30 min for 9 h with an Ultrospec 3100 (Amersham Pharmacia Biotech). The statistical significance of differences between mass doubling times of isolates was calculated using the paired t test with P values interpreted at the 95% level of confidence. For the purpose of standardization, the viable counts at time zero were determined by serial dilution of the initial cultures and spot plating 20 μl onto tryptic soy agar (Oxoid) supplemented with 5% defibrinated sheep blood. At first, the cell number during growth was also determined by viable counting every 90 min, as described above, in parallel with the spectrophotometric measurements. At selected times during growth, culture samples (100 μl) were transferred to microcentrifuge tubes and fixed with 1% formaldehyde for 15 min at room temperature. Aliquots (10 μl) were transferred to poly-l-lysine-coated slides (Sigma, St. Louis, MO) and observed with phase-contrast microscopy using a ZEISS Axioskop HBO 50 microscope equipped with a Plan-NEOFLUAR ×100 oil lens. Photographs were taken with a ZEISS MC100 Spot camera.

Transformation studies.

Transforming genes were cloned into plasmid pGEM-3Zf (Promega, Madison, WI) using standard techniques, followed by DNA sequencing to confirm the nucleotide sequence of cloned genes. Therefore, for transformation of pneumococci, all transforming DNA was in the form of plasmid DNA. Pneumococcal isolate 3 was made competent and transformed as follows. Bacteria were cultured in C medium (27) until the mid-exponential phase (optical density at 620 nm, 0.17) and, after addition of glycerol to 10%, were frozen at −70°C in 500-μl aliquots. For transformation, 1 μg of transforming DNA and 200 ng of synthetic competence stimulatory peptide (amino acid sequence EMRLSKFFRDFILQRKK) were added to 500 μl of competent cells, which were then incubated at 30°C for 45 min and at 37°C for 90 min. Eighty-microliter amounts were then plated onto Mueller-Hinton blood agar medium containing increasing concentrations of penicillin, and the plates were incubated at 37°C for 48 h. Bacterial cells that were transformed to increased penicillin resistance grew on agar medium with a higher penicillin concentration than that of the original recipient, isolate 3, and could therefore be selected. Transformants were picked and further analyzed.

Nucleotide sequence accession number.

The nucleotide sequence of the pbp2A gene from isolates 1 and 2 has been submitted to GenBank under the accession number AY461842.

RESULTS AND DISCUSSION

Genotypic relatedness of isolates and antibiotic resistance.

PFGE (Fig. 1) and BOX-PCR (Fig. 2) fingerprinting revealed that the successive isolates 1, 2, and 3 were clonally related and were the identical pneumococcal strain. The PFGE fingerprint pattern of isolate 1 showed one band difference to that of isolates 2 and 3, indicative of a single genetic event. However, based upon the guidelines of interpretation of PFGE patterns (26), isolate 1 can still be considered clonally related to isolates 2 and 3. Clonal relatedness of isolates was further supported by nucleotide sequence analysis of pbp2X, pbp2B, and pbp1A genes, which revealed identical altered genes for all three isolates. For the remaining discussion and interpretation of our data, we assume the most likely scenario, which is based on the hypothesis that the patient was infected with a pure strain of pneumococcus (isolate 1) and that at the moment in time of infection with isolate 1, the patient was also free of any other strains of infecting pneumococci. On the other hand, we cannot completely exclude the possibility that the patient was infected with a mixed population of clonally related strains and/or that the patient carried strains clonally related to the infecting strain at the time of infection.

FIG. 1.

PFGE fingerprint patterns (SmaI restriction enzyme digestion) of pneumococcal isolates. Lane 1, lambda ladder (49 to 970 kilobase pairs); lane 2, unrelated pneumococcal strain; lane 3, isolate 1; lane 4, isolate 2; lane 5, isolate 3.

FIG. 2.

BOX-PCR fingerprint patterns of pneumococcal isolates. Lane 1, isolate 1; lane 2, isolate 2; lane 3, isolate 3; lanes 4 and 5, unrelated pneumococcal strains; lane 6, size markers (in base pairs).

Surprisingly, isolate 3 showed decreased penicillin, cefotaxime, and ceftriaxone resistance (MICs, 0.38, 0.25, and 0.19 μg/ml, respectively) compared to that of isolates 1 and 2, with MICs of 1.5, 1, and 0.5 μg/ml, respectively. All three isolates showed the same cefuroxime MIC at 8 μg/ml. MIC data suggested that isolate 3 had lost a crucial β-lactam resistance determinant as a result of a recombinational event, most likely transformation. Nucleotide sequence analysis of pbp2X, pbp2B, and pbp1A genes revealed identical altered genes for isolates 1, 2, and 3. These altered genes were similar to genes typical of β-lactam-resistant isolates. Their sequences revealed numerous nucleotide substitutions resulting in numerous amino acid substitutions in the PBPs compared to that of susceptible strains. Altered PBPs 2X, 2B, and 1A were therefore clearly not responsible for the loss of β-lactam resistance in isolate 3. We then decided to investigate whether MurM or the remaining PBPs (PBPs 2A, 1B, and 3) were playing any part in this loss of resistance. Genes encoding these proteins were PCR amplified from isolate 1 and used in transformation experiments with recipient isolate 3. Of these genes, pbp2A was able to transform isolate 3 to increased resistance (penicillin MIC, 1.5 μg/ml; cefotaxime MIC, 1 μg/ml; ceftriaxone MIC, 0.5 μg/ml) to levels identical to that of isolates 1 and 2. Nucleotide sequence analysis of pbp2A from isolate 3 revealed an essentially unaltered gene, while isolates 1 and 2 revealed an identically altered pbp2A gene with 5.4% nucleotide sequence divergence, resulting in 23 amino acid mutations in altered PBP 2A. These included six mutations in the penicillin-sensitive transpeptidase domain. Six transformants resulting from the transformation of isolate 3 were investigated and showed introduction of this altered transpeptidase domain of PBP 2A, which accompanied the phenotypic restoration of penicillin, cefotaxime, and ceftriaxone resistance. Altered PBP 2A, therefore, clearly participated as a β-lactam resistance determinant in isolates 1 and 2, and the loss of altered PBP 2A was responsible for decreased resistance in isolate 3. Altered PBP 2A is rarely reported as a β-lactam resistance determinant. Some reports have shown the involvement of altered PBP 2A in the development of cefotaxime resistance (5, 19). To the best of our knowledge, this is the first report showing the essential role that altered PBP 2A plays in the development of penicillin, cefotaxime, and ceftriaxone resistance in the pneumococcus. Usually, altered PBPs 2X, 2B, and 1A are sufficient to increase penicillin, cefotaxime, and ceftriaxone MICs to ≥0.5 μg/ml. This was clearly not the case for isolate 3, for which loss of altered PBP 2A led to the loss of resistance. These data suggest a new mechanism for the emergence of β-lactam resistance in the pneumococcus. Hakenbeck and coworkers (5) reported a scenario which suggested that an altered PBP 2A was a prerequisite for the transfer a particular altered PBP 1A. Our analysis extends this observation to suggest that altered PBP 2A may be a prerequisite for the expression of altered PBP 1A as a β-lactam resistance determinant, at least in this particular strain. We hypothesize that in the passage from isolate 2 to isolate 3, a DNA fragment housing an unaltered PBP 2A gene was acquired by transformation followed by recombinational replacement of genes. This may have resulted in a strain of increased fitness in its ecological niche. The fitness determinant was either unaltered PBP 2A itself or some other compensatory determinant. With regards to PBP 2A and fitness, perhaps an unaltered PBP 2A is the preferred choice compared to an “uncomfortable” altered PBP 2A. In addition to changes in β-lactam susceptibility, isolates 1, 2, and 3 revealed a sequential decrease in susceptibility to ciprofloxacin, with MICs of 0.5, 1, and 2 μg/ml, respectively. The quinolone-resistance-determining region of the three isolates was sequenced, and all revealed no amino acid mutations in the GyrA, GyrB, ParC, and ParE proteins. To determine whether an efflux system was responsible for this development of low-level ciprofloxacin resistance, we took advantage of the chemical reserpine, an inhibitor of multidrug efflux systems in gram-positive bacteria (13). When ciprofloxacin MIC tests were performed using the agar dilution method with reserpine (10 μg/ml) incorporated into blood agar, the ciprofloxacin MIC for isolate 1 was unchanged at 0.5 μg/ml, while for isolates 2 and 3, the MIC decreased to 0.5 μg/ml. This suggests that the passage of the pneumococcus through its human host, while under the selective pressure of ciprofloxacin, induced expression of efflux-mediated resistance in this strain.

Growth rates and capsule production.

This strain of pneumococcus epitomizes the nature of this bacterial species, that of an adaptive and highly evolving bacterium. Our further experiments suggest that following invasive colonization by isolate 1, the strain evolved into a more virulent type. Growth studies using TSB medium showed that isolate 1 evolved to produce a strain (isolates 2 and 3) with a shorter mass doubling time and therefore a faster growth rate (Table 2 and Fig. 3 and 4). At 37°C, isolate 1 revealed an average doubling time of 29.08 min compared to the shorter average doubling time of isolates 2 and 3, at 26.81 min and 26.56 min, respectively. Statistically, isolate 3 had a significantly shorter mass doubling time compared to isolate 1 (P = 0.0383), while the difference between isolates 1 and 2 was at the limit of significance (P = 0.0517). The results of increased growth rate were even more evident at a higher temperature of 40°C, where isolate 1 revealed an average doubling time of 34.80 min compared to the shorter average doubling time of isolates 2 and 3, at 29.47 min and 29.15 min, respectively (P = 0.0037 and P = 0.0052, compared to isolate 1, respectively). The change in cell number, determined by counting the viable cells every 90 min during growth, confirmed the results obtained spectrophotometrically. Moreover, it showed that at 40°C, both isolates 2 and 3 started decreasing in number before and faster than isolate 1, consistent with the faster growth rates observed. Phase-contrast microscopy showed typical pneumococcal morphology (diplococci and short chains) for all three isolates grown at 37°C (Fig. 3). Consistent with its slower growth rate, some irregular cellular forms were visible for isolate 1 grown at 40°C (Fig. 4). We do not know what genetic and/or biochemical events resulted in this increased cellular growth rate. One of the outcomes of an increased growth rate is that the stationary phase of pneumococcal infection would be reached in a shorter period of time, which may contribute to increased virulence. At stationary phase, pneumococci start undergoing autolysis. This contributes to virulence through the rapid release of virulence factors, which include pneumolysin, peptidoglycan, teichoic acid, and lipoteichoic acid (17). The Quellung reaction and phase-contrast microscopy revealed that isolates 2 and 3 had larger capsules compared to isolate 1 (Fig. 5). This larger capsule was particularly striking in isolate 3, which consistently demonstrated the largest capsule of the three isolates. Measurements were made across the width of the pneumococcal cell. These are reported as a cell diameter measurement (membrane to membrane) divided by a measurement flanked by capsule boundaries. For isolates 1, 2, and 3, these ratio values averaged at 0.50, 0.33, and 0.19, respectively. A smaller value reflects a larger capsule. The capsule is a major virulence determinant, which enables the pneumococcus to inhibit complement-mediated opsonophagocytosis by polymorphonuclear neutrophils (28). Studies have shown that pneumococcal virulence can directly be associated with the quantity of polysaccharide capsule that is produced (7, 12). Furthermore, Ogunniyi and coworkers (18) showed that the expression of genes encoding pneumococcal capsule biosynthesis is up-regulated around fourfold in the blood of infected mice compared to in vitro broth culture. Our results strongly suggest that passage of the pneumococcal strain through its human host resulted in a gain of virulence, with isolate 3 being the most virulent of our three successive isolates. This phenomenon has previously been documented in animal models. Saladino and coworkers (21) found that passage of various serotypes of pneumococci through rats uniformly resulted in strains with increased virulence. Rieux and coworkers (20) created penicillin-resistant laboratory strains by transforming a susceptible strain with altered PBPs 2X and 2B and found that these resistant strains were significantly reduced in virulence; however, they were able to recover full resistance following passage of the strains through mice.

TABLE 2.

Growth rates of clinical isolates at different temperatures

Isolate	Mass doubling time (min)a
	37°C	40°C
1	29.08 ± 1.48	34.80 ± 4.54
2	26.81 ± 0.56	29.47 ± 3.20
3	26.56 ± 0.56	29.15 ± 3.91

^aThe mass doubling times are expressed as the averages of at least four different experiments.

FIG. 3.

Growth and morphology of the three clinical isolates cultivated in TSB at 37°C. (A) Growth curves of isolate 1 (•), isolate 2 (▪), and isolate 3 (▴). (B) Phase-contrast micrographs of the different strains taken in exponential phase. The scale bar corresponds to 1.5 μm. OD, optical density.

FIG. 4.

Growth and morphology of the three clinical isolates cultivated in TSB at 40°C. (A) Growth curves of isolate 1 (•), isolate 2 (▪), and isolate 3 (▴). (B) Phase-contrast micrographs of the different strains taken in exponential phase. The scale bar corresponds to 1.5 μm. OD, optical density.

FIG. 5.

Phase-contrast microscopy of pneumococcal cells and the capsular Quellung reaction with serotype 14 antiserum. Visualization of the capsule is made possible by capsular swelling, which results in greater light refraction. (A) Isolate 1; (B) isolate 2; and (C) isolate 3.

Concluding remarks.

In our case, we observed a pneumococcal strain that, through the passage in its human host, evolved into a more virulent type, determining a severe pneumococcal infection and ultimately the patient's death. It is interesting that the gain of virulence in this successful invasive strain occurred despite the induction of fluoroquinolone resistance and furthermore was associated with decreased β-lactam resistance. This corroborates numerous reports that suggest that invasive strains are largely penicillin susceptible.