Mutations in Penicillin-Binding Protein (PBP) Genes and in Non-PBP Genes during Selection of Penicillin-Resistant Streptococcus gordonii

Abstract

Penicillin resistance in Streptococcus spp. involves multiple mutations in both penicillin-binding proteins (PBPs) and non-PBP genes. Here, we studied the development of penicillin resistance in the oral commensal Streptococcus gordonii. Cyclic exposure of bacteria to twofold-increasing penicillin concentrations selected for a progressive 250- to 500-fold MIC increase (from 0.008 to between 2 and 4 μg/ml). The major MIC increase (≥35-fold) was related to non-PBP mutations, whereas PBP mutations accounted only for a 4- to 8-fold additional increase. PBP mutations occurred in class B PBPs 2X and 2B, which carry a transpeptidase domain, but not in class A PBP 1A, 1B, or 2A, which carry an additional transglycosylase domain. Therefore, we tested whether inactivation of class A PBPs affected resistance development in spite of the absence of mutations. Deletion of PBP 1A or 2A profoundly slowed down resistance development but only moderately affected resistance in already highly resistant mutants (MIC = 2 to 4 μg/ml). Thus, class A PBPs might facilitate early development of resistance by stabilizing penicillin-altered peptidoglycan via transglycosylation, whereas they might be less indispensable in highly resistant mutants which have reestablished a penicillin-insensitive cell wall-building machinery. The contribution of PBP and non-PBP mutations alone could be individualized in DNA transformation. Both PBP and non-PBP mutations conferred some level of intrinsic resistance, but combining the mutations synergized them to ensure high-level resistance (≥2 μg/ml). The results underline the complexity of penicillin resistance development and suggest that inhibition of transglycosylase might be an as yet underestimated way to interfere with early resistance development.

Penicillin resistance was well studied in pathogenic Streptococcus pneumoniae as well as in some other Streptococcus spp. occasionally responsible for infections (21, 24, 34). In these microbes, modifications of the penicillin-binding proteins (PBPs) leading to a decreased affinity for the drug are at the forefront of the resistance mechanism. Those modifications include mutations and/or mosaics in PBP 2X, PBP 2B, and sometimes PBP 1A (25). DNA transformation experiments also indicated that full expression of resistance necessitates the cotransfer of both PBP mutations and mutations located outside the PBP genes, of which only a few have been determined, namely, in the ciaRH, cpoA, and murMN loci (10, 14, 17). In contrast, penicillin-resistant clinical isolates were never observed in phylogenetically related and highly pathogenic Streptococcus pyogenes isolates (28), in spite of long-lasting exposure to the drug. Penicillin-resistant S. pyogenes isolates are apparently also difficult to obtain in the laboratory. Thus, the success of developing penicillin resistance differs among various organisms belonging to the same genus.

In the experiments described below, we studied the spontaneous development of penicillin resistance in cultures of Streptococcus gordonii exposed to increasing drug concentrations in the laboratory. This organism is one of the pioneer colonizers of the oral cavity (11, 19, 20) and occasionally an opportunistic pathogen causing bacteremia or infective endocarditis (6). As a bystander of iterative infections, it is repeatedly exposed to antibiotics. Thus, it could contribute to the pool of PBP genes as well as unknown resistance determinants, exchangeable with other Streptococcus spp.

The experiments were facilitated by a parallel study that characterized the S. gordonii PBPs at both the genetic and the functional levels (15). S. gordonii carries five high-molecular-weight PBPs, which were named according to their homologues in S. pneumoniae. Three of them (PBPs 1A, 1B, and 2A) belong to the class A enzymes, which carry both transpeptidase and transglycosylase domains, and two (PBPs 2X and 2B) are class B enzymes, carrying a transpeptidase domain (15). We also took advantage of the availability of PBP-deleted mutants (15), which permitted the analysis of specific roles for class A and class B PBPs in the process. The results underline the critical role of non-PBP mutations, which could afford a substantial MIC increase (≥35-fold) without any associated PBP mutations and further associate with PBP mutations to confer high-level resistance (MIC > 2 μg/ml). Moreover, they disclosed the unexpected, yet critical, contribution of class A PBPs for the whole resistance development process, in spite of the fact that these class A PBPs were not mutated in the highly resistant mutant.

(Part of this work was presented at the 44th Interscience Conference on Antimicrobial Agents and Chemotherapy, Washington, D.C., November 2004 [abstract C1-1310].)

MATERIALS AND METHODS

Microorganisms and growth conditions.

S. gordonii Challis (29) was used as the model organism. It was grown at 37°C either in brain heart infusion broth (Oxoid Ltd, Hampshire, England) without aeration or on Columbia agar (Oxoid) supplemented with 3% blood. Growth of the cultures was followed both by determination of optical density at a wavelength of 620 nm, using a spectrophotometer (Novaspec II; Amersham Biosciences, Piscataway, N.J.), and by determination of viable counts on agar plates. Bacterial stocks were stored at −80°C in broth supplemented with 10% (vol/vol) glycerol.

Antibiotics and chemicals.

Penicillin G was purchased from Sigma (St. Louis, Mo.). All other chemicals were reagent-grade, commercially available products.

Antibiotic susceptibility.

The MICs of penicillin were determined by a previously described broth macrodilution method (1), with a final inoculum of ca. 10⁶ CFU/ml. The MIC was defined as the lowest antibiotic concentration that inhibited visible bacterial growth after 24 h of incubation at 37°C.

Population analysis profiles (PAPs).

The phenotypic expression of penicillin resistance was determined by spreading bacterial inocula (ca. 10⁸ CFU) as well as appropriate dilutions on plates containing increasing concentrations of the drug. The numbers of colonies growing on the plates were determined after 48 h of incubation at 37°C. The results were expressed by plotting the numbers of colonies growing on the plates against the concentrations of penicillin.

Selection for penicillin resistance by successive cycling.

Selection of penicillin-resistant mutants was performed in broth cultures by exposing bacteria to stepwise-increasing concentrations of antibiotics (7). In brief, a series of tubes containing twofold-increasing concentrations of the drug (ranging from 0.008 to 8 μg/ml) was inoculated with 10⁶ CFU/ml (final concentration), as was done for MIC determinations. After 24 h of incubation, 0.01-ml samples from the tubes containing the highest antibiotic concentrations and still showing turbidity were used to inoculate a new series of tubes containing antibiotic dilutions. The cycling of the susceptible parent was carried on over 115 cycles, and the increases in MICs were followed. This cycling allowed the generation of intermediate (MIC of 0.12 to 1 μg/ml) and fully resistant (MIC of ≥2 μg/ml) mutants. The whole cycling experiment was repeated with five independent cultures (PR1 to PR5), and individual mutants were isolated and stored at −80°C for further study.

Moreover, to assess whether specific PBPs were required for penicillin resistance development, mutants deleted in either PBP 1A, 2A, or 2B were cycled independently over 50 cycles as described above. The three PBP-deleted mutants were generated by allelic replacement according to Lau et al. (23), as described by Haenni et al. (15). In brief, a large internal fragment of each PBP (≥87% of the total protein) was replaced by an erythromycin resistance cassette derived from the streptococcal suicide vector pJDC9 (3). Mutants were selected for erythromycin resistance and purified, and the deletion of PBPs was controlled by PCR with primers external to the constructed fragment as well as by [³H]penicillin radiolabeling.

DNA preparation and genetic strategies.

Molecular techniques were performed by using standard methods (30) or by following instructions provided with commercially available kits and reagents. Genomic DNA was extracted with a QIAGEN DNeasy tissue kit (QIAGEN GmbH, Hilden, Germany). Resistance mutations were sought in the transpeptidase domains of PBP genes and in the murMN and ciaRH operons, which are known to harbor mutations in penicillin-resistant pneumococci (10, 32). No homologue of cpoA, which also carries mutations in S. pneumoniae (13), was found in S. gordonii.

The S. gordonii PBPs and the murMN and ciaRH genes were identified by amino acid homology (≥54%) with the genes of S. pneumoniae, as determined by BLAST comparisons between the nonannotated S. gordonii chromosomal sequence (www.tigr.org) and the annotated S. pneumoniae chromosomal sequence (NCBI accession number NC_003098) (15). The PBP transpeptidase domains as well as the murMN and ciaRH loci were amplified by PCR on a Px2 thermal cycler (Thermo Hybaid, Ashford, Middlesex, United Kingdom). Primers were purchased from Microsynth (Microsynth GmbH, Balgach, Switzerland) and are presented in Table 1. Cycling conditions consisted of 30 cycles at 94°C for 30 s, 52°C for 45 s, and 72°C for 1.5 min, followed by a 10-min delay period at 72°C after the last cycle. Amplicons were purified with a PCR TM DNA and gel band purification kit (GFX; Amersham Biosciences, Buckinghamshire, England) and sequenced in both directions (Synergene Biotech, Schlieren, Switzerland). Sequences were downloaded on Chromas (version 2. 3) and analyzed using the LALIGN program at the Infobiogen website (www.fr.embnet.org).

TABLE 1.

List of the primers used for PCR amplification^a

Gene	Forward sequence	Backward sequence	Size (bp)
pbp1A	5′-C917TGACGCTCAAAAGCAACTG	5′-G1834AACTGGTGTCAAGCGGTTT	917
pbp1B	5′-G1075CTGAGGATGCCATGTATCA	5′-A2168CGTTTGGATCGGCTTGATA	1,093
pbp2A	5′-A1055TCAGGCCAGTATGCAGGTT	5′-A1952CAGAAGCTTCGTTGGCACT	897
pbp2B	5′-A1085CCCGGAAACTGGAGCTATT	5′-C+165GCTTAGCTGCCAAAAGA	1,147
pbp2X	5′-A906ACTCAGCGCCCAAGTTTTA	5′-T1766CAGGCTGCTTCACTGTCAC	860
murM	5′-G-201CTGAAAAACTGGGCTTGAG	5′-C+107TCTGCAATTTCCACGGTCT	1,522
murN	5′-A-167CGACCTCAATGGTGGACTC	5′-G+135ATTCATGTGGACAGCATCG	1,537
CiaR	5′-G-152AAAACACTACAAAATCCTAATCCA	5′-A+57CACCAAAATAGCGAATGAAA	883
CiaH	5′-A-40TATGCAAACCCTTCGCAGT	5′-A+54GCCAGCCATATGTCATTCC	1,494

^a−, number of nucleotides upstream of the start codon; +, number of nucleotides downstream of the stop codon.

Transformation experiments.

S. gordonii competent cells were prepared and transformed by previously described techniques (29). Transforming DNA included (i) genomic DNA purified from various isolates, including S. gordonii (parent and resistant mutants), S. pneumoniae (strain WB4; MIC = 4 μg/ml) (4), and Streptococcus mitis (strain 531; MIC = 2 μg/ml) (8), and (ii) PCR products encompassing the transpeptidase domains of mutated PBP 2B and/or 2X genes, amplified with PR1_2evolved as the template. Competent cells (ca. 10⁶ CFU in 450 μl) were supplemented with saturating concentrations of DNA (1 μg of genomic DNA or 0.3 μg of the PCR products), incubated for 3 h at 37°C, and then treated with a final concentration of 1 mg/ml of DNase (Fermentas) before further processing. The MICs of the transformed cells were assessed by inoculating 10⁶ CFU of each transformed culture into a series of 10 tubes containing increasing concentrations of penicillin, followed by incubation at 37°C for 24 h, as described above. In parallel, 10⁶ transformed cells were spread on agar plates containing increasing concentrations of penicillin, in order to detect resistant subpopulations as determined by PAP investigations.

RESULTS

Selection for penicillin resistance.

The basal MIC of penicillin for the parent S. gordonii was 0.008 μg/ml. Figure 1 depicts the results of a representative experiment where twofold-increasing concentrations of penicillin were used for selective steps. Three phases of MIC increase can be distinguished: first, a rapid, ca. 125-fold increase from 0.008 to 1 μg/ml, encompassing seven 2-fold-increase steps, over the first 16 cycles; second, a unique 2-fold increase from 1 to 2 μg/ml over the next 20 cycles (cycles 17 to 36); and third, a prolonged plateau over 78 additional cycles (cycles 37 to 115), during which the MIC (2 μg/ml) did not increase to the next 2-fold dilution step, despite continuous cycling as described in Materials and Methods. The experiment was stopped after 115 cycles. The cycling procedure was repeated with five separate cultures, named PR1 to PR5, where PR stands for penicillin-resistant culture, followed by the number of the tested culture (see also Table 3). The general pattern of MIC increase was very reproducible.

FIG. 1.

Evolution of penicillin resistance in a culture of susceptible Streptococcus gordonii exposed to penicillin. Liquid cultures were inoculated in a series of tubes containing increasing concentrations of the drug, as with MIC determination. Bacteria from the last tube showing visible growth were reinoculated in a new series of tubes as described previously (7), and the increase in MIC was followed. Three major phases were identified, based on the MIC increase and PBP mutations. Numbers on the graph represent steps at which mutants were purified and stored for further characterization.

TABLE 3.

Mutations in class B PBPs from five independent cultures

Isolate	No. of cycles	MIC (μg/ml)	Mutation(s) in class B PBPa
			2B	2X
Wild type	0	0.008	—	—
PR1_0.25	8	0.25	—	—
PR1_1	16	1	—	Q548E
PR1_2	36	2	—	G545S/Q548E
PR1_2evb	115	2	T450A/V596F	G545S/Q548E
PR2_0.25	14	0.25	—	—
PR2_1	37	1	—	P531L
PR2_2	60	2	—	P531L
PR3_0.25	19	0.25	—	—
PR3_1	39	1	—	G397V
PR3_2	60	2	W429G	G397V
PR4_0.5	10	0.5	H667Y	Q548E
PR4_1	25	1	H667Y	Q548E
PR4_2	60	2	A394V/H667Y	S379L/Q548E
PR5_0.5	25	0.5	H667Y	Q548E
PR5_1	40	1	H667Y	R382H/Q548E
PR5_2	57	2	H667Y	R382H/Q548E
PR5_4	100	4	A440V/N537S/H667Y	R382H/Q548E

^a—, no detected mutation.

^bPR1_2ev, PR1_2evolved.

Homogeneity of resistant bacterial subpopulations during penicillin exposure.

Cyclic exposure in liquid cultures reveals the effect of penicillin on the majority of the bacterial populations but not on subpopulations that might display progressive MIC increases ranging between the last selection level and the following one. To seek such subpopulations, we performed PAP investigations on the cycled cultures at several distinct steps of MIC increases. Figure 2 depicts the PAPs determined for the parent and for mutants taken at MICs of 0.5 μg/ml and 2 μg/ml in the experiment whose results are presented in Fig. 1. Large bacterial numbers (10⁸ CFU) were spread on plates containing gradually increasing penicillin concentrations between the twofold selecting steps. It can be seen that the populations were not homogeneous and contained subpopulations able to grow on drug concentrations greater than that at which they had been selected but lower than that resulting from the following twofold MIC increase. These subpopulations were likely to contain the candidate mutations useful for the next resistance level. Indeed, when the few colonies growing at the highest penicillin concentration on PAP plates were regrown and retested on new PAP plates, they presented further shifts in MIC. However, these outlier colonies represented ≤1% of the total population growing at the selected MIC and thus were unlikely to bias the resistance mutations observed in the majority of the population.

FIG. 2.

Population analysis profiles of S. gordonii isolates presenting increasing resistance to penicillin G. The susceptible wild type (A) and an intermediately resistant (B) and two highly resistant (C, PR1_2; and D, PR1_2evolved) isolates were plated on increasing concentrations of penicillin. Large numbers of bacteria (ca. 10⁸ CFU) and appropriate dilutions were plated. Colonies were enumerated after 48 h of incubation at 37°C, and numbers of CFU/ml were plotted against concentrations of antibiotic. The graphs highlight the major decrease (≥10⁶ CFU) in resistant subpopulations between two twofold MIC selection steps.

The homogeneity of the population was tested in the PR1 culture. Five independent colonies were purified from agar plates at each of these selection steps, and their PBP genes were sequenced (see below). The five colonies carried the same set of PBP mutations.

Resistance mutations in penicillin-resistant derivatives from a single culture.

Table 2 presents the mutations found at distinct MIC levels in the experiment whose results are depicted in Fig. 1. Five isolates taken at various resistance levels were studied in detail, including (i) the susceptible parent, (ii) PR1_0.25 (MIC, 0.25 μg/ml), (iii) PR1_1 (MIC, 1 μg/ml), (iv) PR1_2 (MIC, 2 μg/ml), and (v) PR1_2evolved (MIC, 2 μg/ml). Mutations were sought in the transpeptidase domains of the five PBP genes as well as in two non-PBP loci implicated in penicillin resistance in S. pneumoniae, i.e., ciaRH and murMN (10, 14) (no homologue was found for the S. pneumoniae cpoA locus [13]; see Materials and Methods). Finally, the five PBP genes were sequenced over ≥95% of their entire length in PR1_2evolved. Although this high-level-resistance mutant had been exposed to penicillin for 115 consecutive days, it carried no mutations outside the transpeptidase domain of its PBP genes.

TABLE 2.

Relationship between the MICs of penicillin and PBP alterations

Isolateb	MIC (μg/ml)	Mutation(s) in indicated genea
		pbp1A	pbp1B	pbp2A	pbp2B	pbp2X	murM	murN	ciaR	ciaH
Parent	0.008	—	—	—	—	—	—	—	—	—
Mutants
PR1_0.25	0.25	—	—	—	—	—	—	—	—	—
PR1_1	1	—	—	—	—	Q548E	—	—	—	—
PR1_2	2	—	—	—	—	G545S/Q548E	—	—	—	—
PR1_2ev	2	—	H510Y	—	T450A/V596F	G545S/Q548E	—	—	—	A273T

^a—, no detected mutation.

^bPR1_2ev, PR1_2evolved.

The first PBP mutation, a Q₅₄₈E substitution located close to the conserved KSG motif of PBP 2X, was observed at the end of phase I of the experiment whose results are shown in Fig. 1 (mutant PR1_1). Nevertheless, this resistance mutation was preceded by a ≥32-fold stepwise increase in MIC, as exemplified in mutant PR1_0.25, which was not associated with detectable alterations in PBP genes or in the ciaRH and murMN loci (Table 2).

A second PBP mutation appeared in PBP 2X at the end of phase II of the experiment whose results are shown in Fig. 1 (mutant PR1_2), by a G₅₄₅S substitution located within the KSG motif. Eventually, three additional PBP mutations occurred by the end of phase III of the experiment whose results are shown in Fig. 1 (mutant PR1_2evolved), in spite of the fact that the MICs of the majority of the population had not increased to the next twofold dilution step during this prolonged period of time. Two of these mutations were in the PBP 2B gene (one T₄₅₀A substitution in the SSNT motif and one V₅₉₆F substitution 25 amino acids upstream of the KTG motif), and the third was in PBP 1B (one H₅₁₀Y substitution 18 amino acids upstream of the SWN motif). The exact role of these mutations is as yet unknown. Of note, no mutations were found in the genes of class A PBPs 1A and 2A.

PBP mutations in penicillin-resistant derivatives from independent cultures.

To test whether independently cycled cultures would yield different sets of PBP mutations, the cycling experiment was repeated with four additional independent cultures and the transpeptidase domains of their PBP 2B and PBP 2X genes were sequenced as described above. Table 3, presents the results for these four cultures, plus the result for the one presented in Table 2 for comparison. Figure 3 shows a schematic representation of the PBP 2B and 2X transpeptidase domains, with the localization of all mutations detailed in Table 3. Each of the separate resistant cultures carried a different set of PBP mutations. However, some mutations occurred in more than one mutant, whereas other mutations were unique. For instance, three of the cycled cultures (PR1, PR4, and PR5) presented an early Q₅₄₈E mutation in the PBP 2X gene (Table 3). Thus, the Q₅₄₈E mutation may prevail as observed in S. pneumoniae (5, 31). As the MIC increased further (2 to 4 μg/ml), this shared initial event was followed by different additional mutations in the PBP 2X and PBP 2B genes. In comparison, cycled culture PR2 acquired an early P₅₃₁L mutation in the PBP 2X (which might result in a conformational change), which was not followed by additional PBP 2X or PBP 2B mutations at higher resistance levels. Thus, the way to become resistant was clearly not unique, as with S. pneumoniae (16, 18).

FIG. 3.

Schematic representation of the transpeptidase domain of PBP 2X (A) and PBP 2B (B), as determined by homology with previously characterized PBPs in S. pneumoniae and as described for S. gordonii (12, 15, 26, 27). The three conserved motifs of the active site (SXXK, SSN, and KXG) are indicated at the top of each scheme. The mutations detailed in Table 3 are indicated at the bottom of each scheme. The isolates containing specific mutations are indicated in brackets.

Requirement of class A PBP 1A and 2A for resistance development.

Since class A PBPs 1A and 2A had no resistance mutation in the experiment whose results are presented in Fig. 1, their importance for resistance was tested by repeating the stepwise penicillin cycling with mutants deleted in these determinants (15). As a control, a mutant inactivated in class B PBP 2B was cycled in parallel. PBP 1B was not tested, and PBP 2X could not be inactivated because it is essential, as with S. pneumoniae (15, 22).

Deletion of PBP 1A or 2A did not alter the basal penicillin MIC (0.008 μg/ml). Nevertheless, it considerably decreased the speed of resistance development (Fig. 4). This occurred from the very beginning of the selection process, as the two PBP 1A and 2A mutants required >25 supplementary cycles to attain the MIC of 0.25 μg/ml, which did not increase above 0.5 μg/ml after 50 cycles. In contrast, inactivation of class B PBP 2B did not modify the parental selection profile (Fig. 4). This is not in contradiction with the appearance of mutations in PBP 2B during the development of resistance, because high-level resistance could be achieved without any PBP 2B mutation in certain mutants (Table 3, mutant PR2_2). This modified selection pattern was reproducible in three independent experiments and indicates that, although PBP 1A and 2A were not mutated in the resistant mutants, their presence facilitated resistance development.

FIG. 4.

Evolution of penicillin resistance in PBP-deleted isolates. The susceptible parent, as well as the mutants (deleted in class A PBPs 1A and 2A or class B PBP 2B), was exposed to penicillin as described in the legend to Fig. 1, and the increase in MIC was followed. Arrows indicate the cycles at which DNA was extracted from the mutants for PBP 2B and PBP 2X sequencing.

Effect of class A PBP-deletion on class B PBP mutations.

The question that arose then was whether the presence of class A PBPs was indispensable for the emergence and/or expression of resistance mutations in the class B PBPs, among which PBP 2X was essential for high-level resistance in the parent (Table 3). Mutants deleted in PBP 1A and 2A were exposed to penicillin for 50 cycles (up to an MIC of 0.25 to 0.5 μg/ml) (Fig. 4, arrows), and the transpeptidase domains of their PBP 2X and 2B genes were sequenced (Table 4). The PBP 1A-deleted mutant contained no additional mutations, whereas only one out of two PBP 2A-deleted mutants cycled independently presented a single I₆₁₂N mutation in PBP 2B. In contrast, the class B PBP 2B-inactivated mutant, which had kept a “normal” pattern of resistance selection, did allow multiple PBP 2X mutations in its resistant derivatives after 25 cycles with penicillin (Fig. 4, arrows). One of the cycled cultures carried a unique G₅₆₂D mutation, whereas the other exhibited the characteristic Q₅₄₈E as well as Y₃₅₇C and G₃₈₉V substitutions (Table 4).

TABLE 4.

Mutations in PBP 2B and 2X in PBP-deleted mutants

Mutationb	No. of cycles	MIC (μg/ml)	Mutation(s) in PBPa
			2B	2X
Δ1A	50	0.25	—	—
Δ2A_1	50	0.5	—	—
Δ2A_2	50	0.25	I612N	—
Δ2B_1	25	1	Deleted	G562D
Δ2B_2	25	1	Deleted	Y357C/G389V/Q548E

^a—, no detected mutation.

^bPBP-deleted mutants cycled independently with penicillin.

We next tested whether deletion of class A PBPs also affected the MICs of high-level-resistance mutants that already contained mutations in class B PBP 2X and 2B. PBP 1A and PBP 2A genes were inactivated in resistant mutants PR1_2 and PR1_2evolved (MICs 2 μg/ml) as described previously (15). In both recipients, inactivation of PBP 1A resulted in a fourfold decrease in MIC, whereas inactivation of PBP 2A did not affect susceptibility. Thus, while the presence of PBP 1A and PBP 2A facilitated the acquisition and/or expression of both PBP and non-PBP resistance mutations, directly or indirectly, their deletion could also affect, to a certain extent, the MICs in already resistant mutants.

Individualization of PBP and non-PBP mutations after a single round of transformation.

The above observations, as well as previous work (14, 18), suggest that resistance development in the laboratory progresses sequentially from non-PBP mutations to PBP mutations. Therefore, we tested whether directly providing the bloc of PBP mutations to naïve cells could help bypass the non-PBP steps. Three experiments were performed to address this issue. (i) The contribution of PBP mutations was tested by transforming parent cells with only the mutated transpeptidase domains of PBP 2X and PBP 2B from mutant PR1_2evolved (MIC, 2 μg/ml). (ii) The contribution of both PBP and non-PBP mutation was tested by transforming parent cells with whole chromosomal DNA from the same DNA donor. (iii) The contribution of only non-PBP mutations was tested by transforming parent cells with chromosomal DNA from the non-PBP mutant PR1_0.25 (MIC, 0.25 μg/ml), which did not carry PBP mutations and was a precursor of PR1_2evolved (Fig. 1). The MIC and the acquisition of PBP mutations were tested directly in the transformed cultures, and the cultures were further cycled with penicillin to test the dynamics of resistance development to higher MICs (see next paragraph). Control cultures were transformed with chromosomal DNA from the susceptible parent or from penicillin-resistant isolates of both S. pneumoniae (strain WB4; MIC = 4 μg/ml) (4) and S. mitis (strain 531; MIC = 2 μg/ml) (8).

When only the PBP 2B and/or PBP 2X domain of mutant PR1_2evolved was used, the MICs of the transformed cultures increased 2- to 16-fold (Table 5). The major increases were associated with the acquisition of the Q₅₄₈E mutation in PBP 2X alone (8-fold) or combined with the V₅₉₆F mutation in PBP 2B (16-fold). Thus, PBP mutations could confer increased MICs in the absence of cotransforming non-PBP mutation(s), but these MICs (0.064 to 0.128 μg/ml) were well below that of the DNA donor (2 μg/ml).

TABLE 5.

Transformation of the parent strain with mutated PBP 2B and/or 2X transpeptidase domains from the resistant mutant PR1_2evolved

Recipient	Donor	Mutations	MIC increase	Mutation in transformanta
Wild type	2B	T450A/V596F	2×	—
Wild type	2X	G545S/Q548E	8×	Q548E
Wild type	2B	T450A/V596F	8-16×	V596F
	2X	G545S/Q548E	8-16×	Q548E

^a—, no detected mutation.

When whole chromosomal DNA from the same PR1_2evolved donor or from two of its precursors that contained PBP mutations (PR1_1 and PR1_2) (Fig. 1 and Table 2) was used, the MICs of the transformed cultures increased 16-fold, and this increase also corresponded to the acquisition of the PBP 2X Q₅₄₈E mutation from the DNA donor (Table 6). Thus, in spite of the fact that non-PBP mutations were available in the donor DNA, the PBP 2X Q₅₄₈E mutation was selected first, probably because it conferred the highest MIC increase in a single round of selection.

TABLE 6.

Transformation of the parent strain with genomic DNA of penicillin-resistant isolates^a

Recipient	Donor	Mutation(s)		MIC increase	Mutation(s) in transformant
		PBP 2B	PBP 2X		PBP 2B	PBP 2X
Wild type	PR1_1	—	Q548E	16×	—	Q548E
Wild type	PR1_2	—	G545S/Q548E	16×	—	Q548E
Wild type	PR1_2evolved	T450A/V596F	G545S/Q548E	16×	—	G545S/Q548E

^a—, no detected mutation.

When non-PBP DNA was transformed alone, the initial MIC increase of the culture was eightfold, and no PBP mutations were observed.

Reconciling PBP and non-PBP mutations during further penicillin cycling.

Eventually, the genuine advantage of acquiring PBP and non-PBP mutations emerged only when the transformed cultures were further cycled with penicillin in parallel (Fig. 5). The cultures transformed either with the mutated transpeptidase domains alone (PCR products) or with genomic DNA from the non-PBP mutant alone could not rapidly reach high-level resistance (i.e., 2 μg/ml). In sharp contrast, the cultures transformed with genomic DNA containing both types of mutations took only five to seven cycles to reach high-level resistance (Fig. 5). Thus, transforming naïve cells with the whole panoply of PBP and non-PBP mutations provided them with the mutations critical for rapid attainment of high-level resistance. Yet, these mutations were not acquired in a single round and necessitated a few regrowth and drug selection steps, during which mutations acquired by different bacteria were likely to be shuffled and recombined in the most efficient way.

FIG. 5.

Evolution of penicillin resistance after transformation of the S. gordonii susceptible wild type with the mutated transpeptidase domains from high-level-resistance PR1_2evolved amplified by PCR (A) or with genomic DNA extracted from various resistant strains (B). Transformations with genomic DNA extracted from the penicillin-resistant S. pneumoniae and S. mitis strains displayed the same pattern of resistance development as the wild type. Bacteria were exposed to penicillin as described in the legend to Fig. 1, and the increase in MIC was followed. The PBPs from which the transpeptidase domain was amplified, and the strains from which genomic DNA was extracted, are indicated in the figure.

Thus, while the contribution of both PBP and non-PBP mutations could be genetically and phenotypically individualized, their coexistence in the transformed DNA conferred an incontestable advantage for further resistance development. As a control, parent cells transformed with their own DNA or with DNA extracted from penicillin-resistant S. pneumoniae and S. mitis did not show MIC increases and developed resistance slowly during further penicillin cycling. The absence of resistant S. gordonii transformants might be related to a limited homology between the donor and recipient species (PBP homologies between S. gordonii and S. pneumoniae were between 59% and 67% at the amino acid level) (15).

DISCUSSION

The present experiments highlight expected and unexpected events occurring during laboratory selection of penicillin resistance in S. gordonii. As with other streptococci, resistance developed progressively, starting with mutations located outside the PBP genes, followed by PBP mutations. Yet, the main portion of MIC increase (≥35-fold) (Fig. 1) was consistently associated with non-PBP mutations, thus underlining the importance of these mostly unknown background events. So far, only three penicillin resistance mutations located outside the PBP genes were described for S. pneumoniae (10, 17). Two were located in the ciaH and cpoA genes and occurred very early during penicillin resistance selection (13, 14). The third one was located in the murMN locus and occurred in high-level-resistance S. pneumoniae mutants already carrying PBP mutations (9, 33).

In S. gordonii, a mutation occurred in the ciaH gene, but only in the high-level-resistance isolate. On the other hand, no mutations were found in the murMN homologue, and no homologue of cpoA was identified. It is possible that ciaRH and murMN mutations are specific to S. pneumoniae and replaced by other mutations in penicillin-resistant S. gordonii. Alternatively, ciaRH and murMN mutations might not be unique ways to resistance, and other non-PBP mutations could support resistance as well.

After an initial MIC increase related to non-PBP mutations, a second phase was associated with mutations in class B PBP 2X and PBP 2B. These mutations were indispensable for high-level resistance (MIC = 2 to 4 μg/ml), by conferring a 4- to 8-fold MIC increase which came on top of the ca. 35-fold increase due to non-PBP mutations. Furthermore, while independent cultures cycled with penicillin developed resistance in parallel, the resistant mutants generated by each culture carried different sets of PBP mutations, indicating that several mutational arrangements could confer the same level of resistance. This is concordant with results obtained with S. pneumoniae (16, 18) and underlines the versatility of resistance development in these organisms. However, in spite of this plasticity, a Q₅₄₈E mutation in PBP 2X, which was also found in S. pneumoniae (31), tended to occur preferentially, suggesting that the system was not devoid of constraints. Thus, the initial PBP alterations or the appearance of particular mutations was likely to influence further mutational events. In turn, such constraints might depend on the original non-PBP mutations, which might also differ among various isolates. The nature of such non-PBP mutations is undetermined but may bear importance in the clinical setting, where certain mutants might be fitter than others for survival in the natural environment.

An unexpected feature of the PBP events was the relative absence of mutation in class A PBPs 1A, 1B, and 2A. An exception was a mutation in PBP 1B, a phenomenon that has never been observed in S. pneumoniae. This raised the question as to whether class A PBPs played any role in the process of resistance development. The results demonstrated that deletion of PBP 1A or PBP 2A strikingly hampered the development of resistance. This was not the case when class B PBP 2B was deleted, indicating that the impaired resistance development was related to the class A enzymes. The difference between class A and class B PBPs is the presence of a transglycosylase domain in the class A type, which is not inhibited by penicillin. Thus, although transglycosylases are not a penicillin target, their activity might be critical for resistance development. We propose the following hypothesis to explain this phenomenon. An interconnected peptidoglycan consists in a network of glycan chains cross-linked by peptide bridges. At high penicillin concentrations, the whole transpeptidase apparatus is blocked, and bacterial growth comes to a halt. At low penicillin concentrations, transpeptidase is only partially blocked and cross-linking proceeds at a reduced pace. In this situation, long glycan chains have a greater chance than shorter chains to undergo minimal cross-linking. While transpeptidase domains are the cause of the destabilization of the system in the presence of penicillin, transglycosylase domains ensure a structural stability which is critical for the viability of the bacteria. Thus, this function becomes limiting in the case of the class A PBP deletions.

If transglycosylase is limited to ensure peptidoglycan stability as the bacterium remodels its transpeptidation apparatus, then it should not remain as critical once a penicillin-resistant machinery has been reestablished. This was indeed the case, since deletion of PBP 1A or 2A only moderately altered the MICs of high-level-resistance mutants. The model underlines the importance of accessory determinants that are not directly involved in the resistance mechanism per se but provide the functional framework for resistance establishment. Moreover, it is indirectly supported by the fact that expression of class A PBPs is induced during the cell wall stress response of Staphylococcus aureus (35) and B. subtilis (2) as well as during treatment of resistant S. gordonii with subinhibitory concentrations of penicillin (M. Haenni, unpublished results).

Another intriguing observation was the individualization of PBP and non-PBP resistance mutations in DNA transformation. During spontaneous resistance selection, non-PBP mutations usually occur first, whereas PBP mutations take place later. Yet, transformation experiments indicated that PBP mutations could be transferred directly, both from PCR-amplified transpeptidase domains and from genomic DNA. Why then does spontaneous resistance not select for PBP mutations first? One possibility is that PBP mutations are rare and were not selected from the 10⁶ CFU (original inoculum in liquid cultures) to 10⁸ CFU used in the present experimental system. Alternatively, the expression of mutated PBPs may require combination with some kind of non-PBP mutation, which could develop spontaneously in competent cells transformed with PCR-amplified transpeptidase domains or be acquired along with the PBP mutation during transformation with genomic DNA.

Taken together, the present observations highlight the critical contribution of unmutated class A PBPs for resistance development and the preferential acquisition of PBP resistance mutations in DNA transformation. This may explain the frequent occurrence of PBP modifications in clinical isolates of penicillin-resistant streptococci. Acquiring modified PBP genes from the environment and adapting non-PBP mutations later might be easier than the reverse. Moreover, modifying the transpeptidase apparatus at first might help bypass the dependency on intact class A transglycosylase for early resistance development.