Mutations in the Active Site of Penicillin-binding Protein PBP2x from Streptococcus pneumoniae

Penicillin-binding protein 2x (PBP2x) isolated from clinical β-lactam-resistant strains of Streptococcus pneumoniae (R-PBP2x) have a reduced affinity for β-lactam antibiotics. Their transpeptidase domain carries numerous substitutions compared with homologous sequences from β-lactam-sensitive streptococci (S-PBP2x). Comparison of R-PBP2x sequences suggested that the mutation Gln552 → Glu is important for resistance development. Mutants selected in the laboratory with cephalosporins frequently contain a mutation Thr550 → Ala. The high resolution structure of a complex between S-PBP2x* and cefuroxime revealed that Gln552 and Thr550, which belong to strand β3, are in direct contact with the cephalosporin. We have studied the effect of alterations at positions 552 and 550 in soluble S-PBP2x (S-PBP2x*) expressed in Escherichia coli. Mutation Q552E lowered the acylation efficiency for both penicillin G and cefotaxime when compared with S-PBP2x*. We propose that the introduction of a negative charge in strand β3 conflicts with the negative charge of the β-lactam. Mutation T550A lowered the acylation efficiency of the protein for cefotaxime but not for penicillin G. Thein vitro data presented here are in agreement with the distinct resistance profiles mediated by these mutations in vivo and underline their role as powerful resistance determinants.

The ␤-lactam antibiotics are powerful inhibitors of the transpeptidase (TP) 1 activity of the bacterial penicillin-binding proteins (PBPs). These enzymes catalyze the formation of cross-linked peptidoglycan (1,2). The sensitivity of PBPs to penicillins is related to the structural similarity between the ␤-lactam ring of penicillin and the carboxyl-terminal D-alanyl-D-alanine residues in the natural substrates of these enzymes.
Streptococcus pneumoniae, one of the major human pathogens of the upper respiratory tract, has developed resistance to ␤-lactam antibiotics via modification of the target enzymes, PBPs. Altered PBPs have a reduced affinity for the ␤-lactams, and increased drug concentrations are required for their in vivo inhibition. Low-affinity PBPs in resistant clinical strains of pneumococci (R-PBPs) are encoded by mosaic genes (3)(4)(5)(6)(7)(8)(9) and carry many amino acid substitutions when compared with PBPs from ␤-lactam-sensitive strains.
In resistant clinical strains of S. pneumoniae, up to five PBPs are phenotypically altered in ␤-lactam low-affinity variants: PBPs 1a, 1b, 2a, 2x, and 2b (4, 10 -12). PBP2x and PBP2b are essential for cellular growth (13) and are primary targets for ␤-lactams (14). Therefore, a detailed understanding of the structural modifications conferring the ␤-lactam resistance properties of these two PBPs is a crucial step in designing new ␤-lactam antibiotics. A prime candidate for such an analysis is the PBP2x from S. pneumoniae strain R6 whose three-dimensional structure is available (15).
By comparing R-PBP2x sequences with sequences from sensitive strains of streptococci (S-PBP2x), we have established that the resistant character of PBP2x is often associated with substitution at position Thr 338 . This position is located immediately after the active site Ser 337 , but does not contact the antibiotic (15). Using site-directed mutagenesis we have confirmed the role of this position in modulating the reactivity of Ser 337 toward ␤-lactams (16).
During the course of this work we have identified other amino acid positions that were preferentially altered in R-PBP2xs. Herein, we extend our initial PBP2x sequence analysis and we present evidence linking the resistant phenotype to amino acid replacement at position 552 and 550. These positions are part of the active site groove strand ␤3 and defines one of the main contact region with the antibiotic as shown by the high resolution structure of a complex between PBP2x and the ␤-lactam cefuroxime (Fig. 1). Changes at both positions display different patterns of antibiotic acylation efficiency but a similar pattern of deacylation. These functional results highlight the role of the strand ␤3 residues of the active site groove in conferring the antibiotic-specific resistant phenotype.
Construction of the Expression Plasmid-A DNA fragment encoding soluble PBP2x (PBP2x*), corresponding to Gly 49 to Asp 750 , was ampli-fied by polymerase chain reaction from pCG31 (19). The amplification was carried out using the following cycle conditions: 95°C for 3 min once and then 30 cycles (95°C for 1 min; 55°C for 1 min; 72°C for 2.5 min). Polymerase chain reaction mixture (50 l total volume) contained 5 ng of pCG31 template DNA, 50 pmol of primers, 0.2 mmol of 2Јdeoxynucleotide 5Ј-triphosphates, and 1 unit of Vent DNA polymerase (New England Biolabs, Beverly, MA). The 2125-base pair polymerase chain reaction product was purified by gel electrophoresis on a 1% agarose gel using the GeneClean kit (Bio 101 Inc., Vista, CA) and inserted into the pCR-Script SK(ϩ) plasmid (Stratagene, La Jolla, CA). The resulting plasmid was digested simultaneously with EcoRI and XhoI restriction enzymes. The 2114-base pair purified EcoRI-XhoI DNA fragment was ligated into the EcoRI-XhoI sites of pGEX-4T1 (Pharmacia Biotech, Uppsala, Sweden) resulting in plasmid pGEX-S-PBP2x* encoding the glutathione S-transferase region fused to the NH 2 terminus of S-PBP2x*. The complete nucleotide sequence encoding S-PBP2x* was determined using the T7 Sequenase version 2.0 DNA sequencing kit (Amersham France SA). No unexpected mutation was detected, an observation confirmed by the mass spectrometry measurement of the encoded protein.
Site-directed Mutagenesis-The expression vector pGEX-S-PBP2x* was modified by introducing the replication origin of phage f1 as follows. First, a small DNA cassette resulting from the hybridization of oligonucleotides K7ori1 (5Ј-CAAGGTACCGCATGCAAGCTTTCTA-GACCGACGT-3Ј) and K7ori2 (5Ј-CGGTCTAGAAAGCTTGCATGCGG-TACCTTGACGT-3Ј) was introduced into the unique AatII restriction site of pGEX-S-PBP2x*. The resulting plasmid vector, pGEX-S-PBP2x*KX, was digested with a combination of KpnI and XbaI restriction enzymes and the large DNA fragment was purified as described above. This DNA fragment was ligated with the 525-base pair KpnI/ XbaI DNA fragment from pUC-f1 (Pharmacia Biotech, Uppsala, Sweden) containing the replication origin of phage f1. The resulting expression vector pGEX-S-PBP2x*-f1 was used to transform CJ236 Escherichia coli (dut-1, ung-1, thi-1; relA-1; pCJ105 (Cm r )). The transformed strain was infected with the M13K07 helper phage (Bio-Rad). Site-directed mutagenesis was performed using the single-stranded DNA produced from the resulting phagemid. Each mutant was controlled by DNA sequencing in the region of the expected mutation.
Protein Expression, Purification, and Characterization-Recombinant S-PBP2x* and all mutants derived from it described here were expressed in E. coli and purified by affinity chromatography as described before (16). The proteins were analyzed by SDS-polyacrylamide gel electrophoresis and their molecular masses were measured by electrospray ionization-mass spectrometry (ESI-MS) using a PE Sciex (Toronto, Canada) APIIIIϩ triple quadrupole mass spectrometer equipped with an ionspray source (16). Functional homogeneity of proteins was determined by titrating the active sites present in the preparation using cefotaxime as a reporter. PBP2xs (100 M) were incubated for 15 min at 37°C in 50 mM ammonium acetate, pH 7.0, with or without a 5-fold excess of cefotaxime, and analyzed by ESI-MS.
Determination of Hydrolytic Activity-The hydrolytic activity, defined by k cat /K m , of the TP was assayed in the presence of S2d, a synthetic thiol ester analogue of the cell wall stem peptide (Table I), according to the method described in Ref. 20.
Determination of Acylation Efficiency (k 2 /K)-The interaction between PBPs and ␤-lactams is classically described by, where E is the PBP enzyme, I is the ␤-lactam, EI represents the noncovalent complex, EI* stands for the acyl enzyme covalent complex, and P is the reaction product (cleaved ␤-lactam). k 2 /K, where K ϭ k -1 /k 1 , accounts for the acylation step efficiency. k 2 /K was determined by following the rapid decrease of the intrinsic fluorescence of the protein in the presence of ␤-lactam using spectrofluorometric measurements linked to a stopped-flow apparatus (16). The apparent rate of decay k app (with (k 2 /K) ϭ k app /[␤-lactam]) was determined from the exponential decrease of the fluorescence using the Pade-Laplace method following the simplex method.

Determination of Deacylation Rate (k 3 )-The deacylation reaction obeys the following equation,
where [EI*] 0 is the initial concentration of acyl enzyme and [EI*] t is its concentration at time t. The acyl enzyme was quantitated by fluorography (21).

Identification of Residues Putatively Involved in
Modulating the Affinity of ␤-Lactams for PBP2x-The sequences of 25 resistant PBP2x produced by S. pneumoniae clinical isolates were compared with those of penicillin-sensitive S. pneumoniae, S. mitis, and S. oralis. Overall, 73 sites within the 351-residue TP domain of PBP2x (residue 266 to 616) were changed in at least one of the isolates, and 29 positions were affected in at least 20% of the strains (Fig. 2). The number of amino acid differences between one R-PBP2x and the TP PBP2x from S. pneumoniae R6 strain ranged from 11 to 45. A substitution at Thr 338 occurred in 20 R-PBP2xs (80%). There was only one substitution Gln 552 3 Glu common to the remaining five sequences. This mutation was found, furthermore, in combination with an Pro 338 in one, and with Ala 338 in another three cases. In summary, at least one of the two positions, 338 or 552, were changed in all 25 sequences examined. Most interestingly, both amino acid changes are known to confer a resistance phenotype in vivo 2 and thus appear to be first candidates for a close in vitro investigation. The distance between the ␣ carbon of the 15 remaining amino acid positions and the active site Ser 337 ␣ carbon was measured. Two positions, 338 and 552, are within 10 Å of Ser 337 and are part of or near to conserved sequence motifs of the ASPRE family (active site serine penicillin recognizing enzymes) (Fig. 3) (22). Therefore, positions 338 and 552 are putative determinants for resistance to ␤-lactams. Indeed, position 338 was shown to be a key position in modulating the affinity to ␤-lactams, even thought this side chain is not in direct contact with either the substrate or the ␤-lactams (16).
Gln 552 in S-PBP2x from R6 is replaced by a Glu in 9 R-PBP2xs (36%). This position is located in the active site groove defined by strand ␤3 (Figs. 1 and 3). This element of secondary structure also includes position 550. A Thr 550 3 Ala mutation has been observed in five independently obtained laboratory mutants after selection with extended spectrum cephalosporins (cefotaxime or cefpodoxime) (Table I), and a mutation to Gly 550 has also been reported (14,23,24). The mutation T550A occurs in one R-PBP2x, most likely also the result of selection with cephalosporins (17). The contribution of positions 552 and 550 to the resistance character was probed by site-directed mutagenesis in S-PBP2x*, a molecular context deprived of all the other amino acid changes found in R-PBP2xs.
Expression, Purification, and Stability Measurement of S-PBP2x* Mutants-Six S-PBP2x* mutants were expressed in E. coli as glutathione S-transferase fusion proteins. The recombinant proteins were purified to homogeneity by affinity chromatography following cleavage by thrombin. The yield of purified S-PBP2x* mutants is comparable to the one reported for S-PBP2x* and ranges from 20 to 50 mg/liter of bacterial culture (16). The molecular mass for all mutants measured by ESI-MS corresponds to the calculated mass deduced from the amino acid sequence (Table II). ESI-MS was also used to demonstrate that all mutants and S-PBP2x* are fully acylated by cefotaxime when incubated with an excess of antibiotic (data not shown).
These results show functional homogeneity of all mutants. Accurate determination of enzyme kinetic parameters might be affected by variations in stability between mutants. This is especially true for k 3 measurement which requires long incubation periods at 37°C. Therefore, we have measured the residual activity of the mutants following prolonged incubation at 37°C which is the temperature used to measure all kinetic parameters (k 2 /K, k 3 , and k cat /K m ). Since the enzymatic activity remained constant over time, the kinetic parameters of all mutants can be directly compared.
Characterization of Single S-PBP2x* Mutants-The effect on S-PBP2x* of mutations most frequently found in ␤-lactamresistant laboratory or clinical isolates, was determined. The acylation efficiency, represented by k 2 /K, of S-PBP2x* mutants was measured by following the rapid accumulation of the acyl enzyme reported by the quenching of the protein intrinsic fluorescence (Fig. 4). The acylation efficiencies of S-PBP2x* Q552E decrease by 72% for both Pen G and cefotaxime when compared with S-PBP2x*. The pattern of acylation efficiency of the S-PBP2x* T550A mutant is very different with a k 2 /K reduction over 90% for cefotaxime whereas acylation by Pen G is not affected. To distinguish the influence of the hydroxyl group from the steric property of the side chain, the T550S S-PBP2x* mutant was constructed. This mutation reduces the acylation efficiency of enzyme for both antibiotics by about two-thirds.
The reduction of ␤-lactam inhibition efficiency on laboratory and clinical resistant PBPs results mostly from a large decrease of the k 2 /K, the influence of the slow deacylation rate being negligible (16,25). The deacylation rate of the three single mutants was measured by fluorography (Table III). This rate ranges from 1.3 ϫ 10 Ϫ5 (s Ϫ1 ) to 3.8 ϫ 10 Ϫ5 (s Ϫ1 ). These values remain very low, thus the effect of the mutations on the efficiency of acylation by ␤-lactams dominates.
To monitor the effect of the mutations on the enzymatic properties of S-PBP2x*, we have determined the hydrolytic activity of the mutants for the S2d substrate analogue of the cell wall stem peptide (Table III). The hydrolytic activities of the S-PBP2x* Q552E and T550A mutants were reduced by 32 and 46%, respectively, when compared with the S-PBP2x* activity. A marked decrease of k cat /K m for the S2d thiol ester is observed for the S-PBP2x* T550S mutant, with less than 10% remaining activity of the wild type soluble enzyme. Thus, the S-PBP2x* T550S mutant, which is absent in the set of clinical or laboratory isolates, has a much reduced level of hydrolytic activity than the Q552E and T550A mutants.

Combining Mutations at Gln 552 , Thr 338 , and Ser 571 Mimic
Clinical Resistant PBP2x-In previous work (16) we have shown that positions 338 and 571 were critical determinants for resistance to ␤-lactam antibiotics. In our set of 25 clinical isolate PBP2x sequences, three of them contain both the T338A and Q552E mutations and one displays the combined substitutions T338A, Q552E, and S571P. These two combinations of mutations were introduced in the context of S-PBP2x* as a mean to evaluate their effect in the enzymatic properties (Table  III). The acylation efficiencies of S-PBP2x* T338A,Q552E double mutant decrease by over 90 and 80% for cefotaxime and Pen G, respectively, when compared with S-PBP2x*. The deacylation rate of the mutant-Pen G complex decreases by 30% over the wild type enzyme. This slow rate remains negligible over the decrease of acylation efficiency. The k cat /K m for the hydrolysis of S2d for the double mutant is reduced 8-fold when compared with S-PBP2x*.
The S-PBP2x* T338A,Q552E,S571P mutant is active since it can be fully acylated when incubated with a five times excess of cefotaxime as determined by ESI-MS (data not shown). However, the accurate kinetic parameters of this triple mutant cannot be determined because their values are too low. Based upon previous experiments (16), we have estimated the minimal measurable kinetic parameters as below 0.1 and 3.8% of the wild type value for k 2 /K for cefotaxime and Pen G, respectively. Similarly, the minimal measurable k cat /K m for S2d is 0.2% of the wild type value.  PBP2x is one of the main determinants for ␤-lactam resistance in the Gram-positive pathogen S. pneumoniae. The 351residue long TP domain from S. pneumoniae R-PBP2x contains between 11 and 45 amino acid changes when compared with homologous genes from sensitive streptococci. Only a few of those changes are likely to be linked to the resistance phenotype. Two positions, located within 10 Å of the active site Ser 337 , display a restricted pattern of substitutions in R-PBP2x but are strictly conserved in PBP2x from sensitive strains. These amino acid changes in PBP2x are likely to be relevant to the acquisition of the resistance phenotype.
The first position, Thr 338 in S-PBP2x from R6 strain was shown to be a major determinant for resistance to ␤-lactam (16). The second position, Gln 552 in S-PBP2x from R6 strain, follows the Lys-Ser-Gly conserved motif (Fig. 3) in the ASPRE family and is part of strand ␤3 which borders the enzyme active site (15). We have replaced this position by a Glu, a side chain frequently found in R-PBP2xs. This mutation reduced the efficiency of acylation by over two-thirds for both cefotaxime and Pen G but had only a marginal effect on the deacylation step considering that this value remains very low. This result unambiguously identifies position 552 as another key position for the modulation of PBP2x sensitivity to ␤-lactams. The PBP2x active site displays a global positive charge favoring binding of ␤-lactams which present a global negative charge. Introduction of a negative charge in active site via Glu 552 conflicts with ␤-lactams and consequently leads to the observed reduction in acylation efficiency. Electrostatic steering effects are important in class A ␤-lactamases, which also belong to the ASPRE family (26). Since the net charge of inhibitors and thiol substrate is Ϫ1, the driving force for ligand binding is affected by the mutation Q552E.
Besides this long-range electrostatic effect, a direct contact between Gln 552 and the ␤-lactams was shown by the recent 2.8-Å resolution x-ray structure of the complex between S-PBP2x* and cefuroxime. 3 The cefuroxime lies close and parallel to strand ␤3 (Fig. 1), a mode of binding of ␤-lactams to the active site different from the one reported for the Streptomyces R61 PBP complexed to cefotaxime and cephalothin (27). The furan ring of cefuroxime is close to the S-PBP2x* Gln 552 side chain. Cefuroxime and cefotaxime have very similar structures ( Table I). Assuming that their mode of binding to PBP2x is very close, then the introduction of a negative charge in strand ␤3 might conflict with the cefotaxime. The effect is much less pronounced when the hydrolytic activity of the Q552E mutant against the S2d substrate analogue is considered. S2d is a shorter molecule than cefotaxime and Pen G and the S2d phenyl group is less likely to be affected by Glu 552 as it could be more readily accommodated by the protein.
Thr 550 in R6 strain, which is also located in strand ␤3 at a distance of 6.6 Å from active site Ser 337 , is often altered to Ala in PBP2x isolated from ␤-lactam-resistant S. pneumoniae selected in the laboratory using a single cephalosporin antibiotic. The T550A mutation in S-PBP2x* is neutral for Pen G but reduces the acylation efficiency for cefotaxime almost 20-fold. The same mutation was found in one PBP2x from a clinical isolate. The mutation increases the resistance of S. pneumoniae to cephalosporin but reduces its resistance to penicillin (17). Taken together, these results show that, contrary to mutation Q552E, substitution T550A in S-PBP2x* affects the acylation efficiency for cefotaxime and Pen G differently. Our experimental data are in agreement with the observed differences between clinical and laboratory PBP2x mutants. Multiple selective pressure leads to PBP2x with a reduced acylation efficiency toward a larger spectrum of ␤-lactams than when a single antibiotic is used under laboratory conditions. Interestingly, some natural variants of TEM-1 ␤-lactamases have been found to contain an A237T substitution in a position equivalent to residue 550 in PBP2xs (28,29). These class A ␤-lactamases, generally penicillinases, have the ability to hydrolyze thirdgeneration cephalosporins such as cefotaxime.
Preserving the hydroxyl group at position 550 while altering the steric property of the side chain (T550S mutation) leads to a PBP2x* mutant unable to discriminate between the Pen G and cefotaxime. This mutant displays acylation efficiency values very close to the Q552E mutant (Fig. 4). T550S mutation is not found in our collection of R-PBP2x, and this could be explained by the marked reduction in hydrolytic activity of the mutant. Thus, it is likely that the TP activity of S-PBP2x* T550S mutant is affected to a greater extent than it is for the Q552E mutant, which might be detrimental to the host bacteria. These differences might explain the selection of Glu at position 552 over Ser at position 550 in clinical isolates.
All R-PBP2x sequences of the considered set contain substitutions at position 338 or 552; in four cases, the R-PBP2x sequences contain both substitutions. Combining T338A with Q552E further reduces the acylation efficiency over the single mutants. This reduction is more pronounced in the triple mutant T338A,Q552E,S571P reproducing a combination of side chains found in one sequence from our set of clinical isolates. Thus a limited number of mutations is sufficient to mimic the enzymatic properties of R-PBP2x.