Klebsiella pneumoniae Carbapenemase-2 (KPC-2), Substitutions at Ambler Position Asp179, and Resistance to Ceftazidime-Avibactam: Unique Antibiotic-Resistant Phenotypes Emerge from β-Lactamase Protein Engineering

ABSTRACT The emergence of Klebsiella pneumoniae carbapenemases (KPCs), β-lactamases that inactivate “last-line” antibiotics such as imipenem, represents a major challenge to contemporary antibiotic therapies. The combination of ceftazidime (CAZ) and avibactam (AVI), a potent β-lactamase inhibitor, represents an attempt to overcome this formidable threat and to restore the efficacy of the antibiotic against Gram-negative bacteria bearing KPCs. CAZ-AVI-resistant clinical strains expressing KPC variants with substitutions in the Ω-loop are emerging. We engineered 19 KPC-2 variants bearing targeted mutations at amino acid residue Ambler position 179 in Escherichia coli and identified a unique antibiotic resistance phenotype. We focus particularly on the CAZ-AVI resistance of the clinically relevant Asp179Asn variant. Although this variant demonstrated less hydrolytic activity, we demonstrated that there was a prolonged period during which an acyl-enzyme intermediate was present. Using mass spectrometry and transient kinetic analysis, we demonstrated that Asp179Asn “traps” β-lactams, preferentially binding β-lactams longer than AVI owing to a decreased rate of deacylation. Molecular dynamics simulations predict that (i) the Asp179Asn variant confers more flexibility to the Ω-loop and expands the active site significantly; (ii) the catalytic nucleophile, S70, is shifted more than 1.5 Å and rotated more than 90°, altering the hydrogen bond networks; and (iii) E166 is displaced by 2 Å when complexed with ceftazidime. These analyses explain the increased hydrolytic profile of KPC-2 and suggest that the Asp179Asn substitution results in an alternative complex mechanism leading to CAZ-AVI resistance. The future design of novel β-lactams and β-lactamase inhibitors must consider the mechanistic basis of resistance of this and other threatening carbapenemases.


RESULTS AND DISCUSSION
Substitutions at position Asp179 alter KPC-2 ␤-lactamase expression in Escherichia coli. All 19 amino acid variants at Ambler position 179 in KPC-2 ␤-lactamase were engineered. To determine whether these single substitutions alter protein expression, immunoblots using whole-cell preparations and periplasmic extracts were probed with an anti-KPC-2 polyclonal antibody (Ab) that is sensitive and specific and maps to three main linear epitopes of KPC-2 (9). Single amino acid substitutions at position Asp179 in the ⍀-loop generally result in decreased levels of expression (Fig. 2), possibly due to differences in overall protein stability. However, certain variants (Asp179Asn) maintain notable levels of expression during the exponential phase of the growth curve.
Microbiological analysis. (i) Asp179 variants of the ⍀-loop of KPC-2 and ␤-lactam resistance. The impact of site-saturation mutagenesis at Ambler position 179 of KPC-2 on antibiotic resistance was next assessed using whole-cell viability assays. Twenty-four different ␤-lactam and ␤-lactam-␤-lactamase inhibitor combinations were tested for susceptibility against KPC-2 and the 19 variants expressed in E. coli DH10B cells (Tables 1 and 2). The Klebsiella pneumoniae KPC-2-containing positive-control strain maintained resistance (as defined by Clinical and Laboratory Standards Institute [CLSI] criteria) against all the commercially available ␤-lactams tested ( Table 1). The     E. coli strain containing the KPC-2 construct exhibited resistance to the same panel of tested antibiotics, including cephalosporins, monobactams, and carbapenems (Table 1). In contrast, the Asp179 variants (except Asp179Asn) expressed in E. coli generally showed an increase in susceptibility to the ␤-lactam antibiotics, possibly attributable to the attenuated protein expression of the variants (Fig. 2). Notably, striking resistance to ceftazidime was maintained by all the variants, including the least-expressed Asp179Lys and Asp179Arg variants (MICs of 64 g/ml for both variants) ( Table 1).
(ii) Addition of avibactam overcomes the ceftazidime resistance mediated by KPC-2 but not that mediated by the Asp179 variants. The addition of the ␤-lactamase inhibitor avibactam abrogated ceftazidime resistance in the KPC-2containing strain but, alarmingly, was insufficient to restore susceptibility to any of the strains harboring the Asp179 variants ( Table 2). This supported an earlier observation that the Asp179Ala, Asp179Gln, and Asp179Asn variants of KPC-2 expressed in E. coli conferred resistance to ceftazidime-avibactam (12).
(iii) The Asp179Asn variant exhibits a resistant antimicrobial profile. The strain harboring the Asp179Asn variant stood out among the other 18 variant strains for conferring levels of resistance to all commercially available ␤-lactams tested as monotherapies except meropenem (the breakpoint for BAL30072 is not yet defined), similarly to the resistance profile of KPC-2 (Table 1). Notably, the Asp179Asn variant strain demonstrated elevated resistance to ceftazidime (KPC-2 was measured at 128 g/ml compared to Asp179Asn measured at 512 g/ml) ( Table 1) and to the ceftazidimeavibactam combination (KPC-2 measured at 1 g/ml versus Asp179Asn measured at 16 g/ml) ( Table 2).
Aztreonam-avibactam and ceftaroline-avibactam, two combinations currently in clinical trials, effectively showed lower MICs for the Asp179Asn strain. To gain insight into the therapeutic potential of clinically relevant avibactam combinations, susceptibility testing against commercially available avibactam (Advanced ChemBlocks) combined with aztreonam and ceftaroline was conducted on strains containing KPC-2 and each of the variants. Both the KPC-2 and Asp179Asn-containing strains were resistant to each of these ␤-lactams in the absence of avibactam (Table 1). Avibactam restored the susceptibility of KPC-2 to aztreonam and ceftaroline based on the breakpoint for the each ␤-lactam alone. In contrast to the resistance to ceftazidimeavibactam of the Asp179Asn strain, ceftaroline-avibactam decreased the resistance of the Asp179Asn strain to an intermediate level based on the breakpoint of ceftaroline alone ( Table 2). The aztreonam-avibactam combination was even more effective, rendering the Asp179Asn strain susceptible to aztreonam. These data suggest that ceftaroline and aztreonam may be attractive and viable therapeutic partners for avibactam against Asp179Asn variants of KPC-2.
Avibactam in combination with imipenem and ceftriaxone lowers the drug MICs of the Asp179Asn strain; the structure of the R1 side chain in ceftazidime contributes to ceftazidime-avibactam resistance. To further explore the structureactivity factors (particularly Asp179Asn) responsible for the increased ceftazidime resistance of the Asp179 variant ␤-lactamases, experiments using avibactam in combination with a representative carbapenem (imipenem), a cephalosporin structurally similar to ceftazidime (ceftolozane; tested in combination with tazobactam), and a cephalosporin structurally distinct from ceftazidime (ceftriaxone) were conducted with strains containing KPC-2 and each of the variants. Ceftolozane, a novel cephalosporin, is the ␤-lactam most similar in structure to ceftazidime (Fig. 3). Ceftazidime and ceftolozane differ by only one atom in the R1 side chain, with ceftazidime possessing a carbon atom (aminothiazole) and ceftolozane a nitrogen atom (aminothiadiazole). Ceftriaxone is an expanded-spectrum oxyimino-cephalosporin that has been in commercial use for decades. Ceftazidime and ceftriaxone share the same aminothiazole group but differ on the oxyimino end of R1 (Fig. 3). Ceftriaxone has a less bulky oxyimino group and lacks the acidic carboxylate group (like the oxyimino group of ceftaroline). c Ceftolozane-tazobactam was tested at a ratio of 2:1.
Barnes et al.
Both the KPC-2 and Asp179Asn-containing strains were resistant to imipenem, ceftolozane-tazobactam, and ceftriaxone in the absence of avibactam (Table 1). With results mimicking the susceptibility to ceftazidime-avibactam, avibactam restored susceptibility of KPC-2 to imipenem, ceftolozane-tazobactam, and ceftriaxone based on the breakpoints for the respective ␤-lactams. The Asp179Asn strain maintained resistance to ceftolozane-tazobactam-avibactam (based on the breakpoint of ceftolozanetazobactam), similarly to the ceftazidime-avibactam resistance results (Table 2). In contrast, avibactam restored the susceptibility of the Asp179Asn strain to ceftriaxone and imipenem. Ceftriaxone-avibactam was the most effective cephalosporin combination tested ( Table 2), suggesting that the bulkiness and/or the carboxylate group on R1 is the structural moiety that may be primarily responsible for ceftazidime resistance.
Biochemical analysis. (i) The Asp179Asn variant hydrolyzes ceftazidime slowly but demonstrates a lower K i for ceftazidime. To elucidate the mechanistic differences between Asp179Asn and KPC-2, each was purified for biochemical analysis. In previous comparisons of KPC-2 to Asp179Asn performed using periplasmic extracts, we noted similar rates of ceftazidime hydrolysis (12). However, the previous analysis was performed using amounts of ␤-lactamase normalized for nitrocefin (NCF) hydrolysis. Here, we used 1 M KPC-2 and Asp179Asn to measure the hydrolysis of ceftazidime and used a 0.5 M concentration of each enzyme in the assessment of imipenem hydrolysis. The Asp179Asn variant hydrolyzed both ␤-lactams at a much lower rate than KPC-2 ( Fig. 4A and B).
We previously reported a burst in hydrolysis of ceftazidime by KPC-2 (7). Thus, the conditions were optimized for assessment of early time points in the hydrolysis of ceftazidime by KPC-2 compared to Asp179Asn. A burst amplitude of 0.16 M Ϯ 0.02 M was obtained with 2 M KPC-2 and 25 M ceftazidime (similarly to our previous results with KPC-2) (Fig. 4C). A higher burst amplitude of 0.57 M Ϯ 0.06 M was obtained with Asp179Asn under the same conditions. A burst that occurs prior to the establishment of the linear steady-state reaction reflects rapid acylation followed by a relatively slow rearrangement or product formation (equation 1) (16) as follows: In such a reaction, the burst decay constant, k burst , is given by k burst ϭ k ac ϩ k 2 ; the burst amplitude, A, is given by A ϭ [Eo] · (k ac /k ac ϩ k 2 ) 2 , where [Eo] is the concentration of active enzyme; and the steady-state rate, k ss , is given by k ss ϭ k 2 /(1 ϩ k 2 /k ac ).
For the experiment represented in Fig. 3, the values were calculated to be k ac ϭ 0.043 Ϯ 0.004 s Ϫ1 and k 2 ϭ 0.097 Ϯ 0.010 s Ϫ1 for KPC-2 and k ac ϭ 0.038 Ϯ 0.004 s Ϫ1 and k 2 ϭ 0.030 Ϯ 0.003 s Ϫ1 for the Asp179Asn variant. Thus, multiple steps along the reaction coordinate are likely affected by the substitution, although the effect on the hydrolysis of the acyl intermediate or product release (k 2 ) appears to be the more profound.
To compare the apparent affinity of ceftazidime for KPC-2 to its affinity for the Asp179Asn variant, we used various concentrations of ceftazidime to inhibit hydrolysis of a reporter substrate, nitrocefin ( Fig. 4D and E). More ceftazidime was required with the wild type (apparent K i , 3.5 mM) to reach the same level of inhibition as that seen with the Asp179Asn variant (apparent K i , 0.13 mM), as is expected in a reaction such as that described in equation 1 in which the k 2 value is lower for the variant.
(ii) The Asp179Asn variant is a "trap" for ␤-lactams. To further support our kinetic analysis, timed mass spectrometry was used to probe for mechanistic differences between KPC-2 and the Asp179Asn variant. The ␤-lactamases (E) were incubated with a substrate (S [ceftazidime, imipenem, or aztreonam]) and an inhibitor (I [avibactam]) at a molar ratio of 1:1:1 (E:S:I), thus establishing a direct competition between the ␤-lactam (avibactam) and the ␤-lactamase.
Several characteristics of the Asp179Asn variant that are distinct from those of the wild-type enzyme were revealed. KPC-2 preferentially bound avibactam compared to the tested ␤-lactams ( Fig. 5A; see also Fig. S1 in the supplemental material). This observation was not surprising as avibactam is a potent inhibitor of KPC-2. Also, ceftazidime, imipenem, and aztreonam are substrates for KPC-2 and could therefore be hydrolyzed before they could be detected as mass adducts. In contrast, under the same conditions, Asp179Asn preferentially bound all the tested ␤-lactams compared to avibactam ( Fig. 5B and S1).
Asp179Asn binds imipenem for the longest time, with the acyl-enzyme remaining the predominant species of Asp179Asn at 15 min, unlike ceftazidime or aztreonam (the acyl-enzymes are undetectable by 15 min). These data suggest that the asparagine substitution at the 179 aspartate position allows the Asp179Asn ␤-lactamase to "trap" the substrate. The decrease in the rate of deacylation, predicted by the transient kinetic analysis described above, may be sufficient to explain the preferential trapping of ␤-lactams compared to avibactam. This step is not involved in the interaction of avibactam with the ␤-lactamase (17); therefore, its reaction kinetics are not so strongly affected.
(iii) Desulfation of avibactam is unique to KPC-2 and is not observed with the Asp179Asn variant. Desulfation of avibactam is known to occur with KPC-2 (18,19), and we show here that it does not occur in the Asp179Asn variant ( Fig. 5A and S1). Possible explanations for the distinct ability of KPC-2 to desulfate avibactam among the ␤-lactamases include a lack of hydrogen bonds with the N6 atom and a necessary water molecule that desulfates avibactam (18). Both the avibactam and desulfated avibactam bound to KPC-2 are stable complexes still present at 48 h, although the apo-KPC-2 form becomes more predominant (data not shown).
(iv) Mass spectrometry and the detection of a unique mass adduct. Interestingly, the reaction of Asp179Asn with ceftazidime produced two protein charge envelopes; one was the expected primary charge envelope at approximately 1,100 to 2,000 m/z (29,187 Da mass, corresponding to the predicted mass of 28,719 Da for Asp179Asn in addition to the 468 Da for ceftazidime minus R2), and the second was a more highly charged species (at approximately 800 to 1,200 m/z) with a different overall deconvoluted mass that was not observed with KPC-2 ( Fig. S4; Table 3). The formation and elimination of this secondary envelope were time dependent, deconvoluted to a single protein peak with a mass of 29,124 Da, and were unique to Asp179Asn reacted with ceftazidime or the ceftazidime-avibactam combination. The identity of this ϩ405 adduct with a particular chemical rearrangement in ceftazidime is not obvious. We did not observe this peak with avibactam in combination with aztreonam or imipenem (data not shown), eliminating the likelihood of a mass spectrometry artifact. We take this observation to suggest that the Asp179Asn-ceftazidime complex may undergo an alternative or additional conformational change during hydrolysis, resulting in an altered surface charge and a corresponding shift in the mass-to-charge ratio.
In an attempt to understand the mechanistic basis for the formation of the ϩ405 adduct, structurally similar cephalosporins (ceftriaxone, ceftolozane with tazobactam, and ceftaroline) were tested with the Asp179Asn variant. The mass adducts for Asp179Asn and ceftazidime, ceftriaxone and ceftolozane, which have good leaving groups in R2, were consistent with the molecular weights of the antibiotics after elimination of the R2 group ( Fig. S2; Table 3). Ceftaroline, which does not have a good R2 leaving group, bound to Asp179Asn with an intact R2 group (Fig. S2). Ceftriaxone and Asp179Asn revealed the expected primary envelope with a deconvoluted mass of 29,114 Da (predicted mass of Asp179Asn of 28,719 Da in addition to the 395 Da of ceftriaxone minus R2) and a second charge envelope which corresponded to a unique  Table 3). This adduct equates to ceftriaxone minus 61 Da, which parallels the mass of ceftazidime missing 63 Da in the Asp179Asnceftazidime adduct. These data suggest that a time-dependent modification to the antibiotic occurs and is selective for Asp179Asn with ceftazidime and ceftriaxone. The nature and clinical impact of this modification are being explored.
(v) BAL30072, a monosulfactam, lowers the drug MICs of the ceftazidimeavibactam (CZA)-resistant Asp179 variants. Given the levels of resistance attributed to the single-amino-acid substitutions at position Asp179 in KPC-2, we were compelled to test the novel monosulfactam BAL30072 for its efficacy against these variants (20). BAL30072, like aztreonam, consists of a monocyclic ␤-lactam scaffold with an R1 group containing a siderophore moiety and a thiazole ring similar to that in the R1 group of ceftazidime (Fig. 3). Aztreonam was previously shown to have a 927-fold-higher k cat value and a 3,031-fold-higher k cat /K m value for KPC-2 than BAL30072 (20). Bypassing KPC-2 through a lack of positive interactions (high K m ) and targeting penicillin-binding proteins (PBP) is therefore likely responsible for the potent activity of BAL30072 against strains expressing KPC-2.
Here, we found potent activity of BAL30072 against the Asp179 variants of KPC-2 (Table 1). In addition, we conducted mass spectrometry with KPC-2 and the Asp179Asn variant with BAL30072 (Fig. S3). The Asp179Asn variant bound BAL30072 for longer than an hour (longer than a typical bacterial division cycle) and longer than any other antibiotic tested, while KPC-2 did not bind BAL30072 at all. MIC analyses of BAL30072 against KPC-2 and Asp179Asn resulted in the lowest MICs of all the antibiotics tested (see Table 1). In addition, BAL30072 inhibits penicillin-binding proteins (PBP 1a, PBP 1b, and PBP 3) (20). On the basis of these observations, BAL30072 could be effective with strains containing Asp179Asn and is also "trapped" in the active site of Asp179Asn. The Asp179Asn trapping of this monosulfactam (aztreonam), cephalosporins, and carbapenems suggests that trapping is not reserved for select classes or activities of ␤-lactams.
(vi) Molecular modeling of KPC-2 and Asp179Asn with and without ceftazidime; the Asp179Asn ⍀-loop is more flexible and mobile. To investigate how the structural differences induced by the aspartate-to-asparagine substitution in the variant enzyme could explain the mechanism of ceftazidime resistance and antibiotic "trapping," molecular modeling of the Asp179Asn variant and KPC-2 in the presence of ceftazidime was performed ( Fig. 6 and S6 to S8).  in KPC-2 is disrupted in the Asp179Asn variant ( Fig. 1A and B). The Asp179 residue in KPC-2 forms a hydrogen bond network with Pro67, Leu68, Arg161, Asp163, and Arg164, but most of these interactions are absent in the variant (Fig. 1B). The disruption of the hydrogen bonds and salt bridge between Asp179 and Arg164 in the variant generates an "open channel" in the middle of the ⍀-loop, enhancing the flexibility of this structure ( Fig. 1A and B). Most of the side chains preserve their conformations (RMSD ϭ Յ1 Å). However, the side chains of active-site residues are shifted by 0.5 to 1.5 Å and the active site in a Connolly representation is clearly deeper and wider, showing that the Asp179 variant has a more "open" conformation than KPC-2 (Fig. S6). The analysis of trajectory generated during the 0.55-ns simulation of KPC-2 and Asp179Asn acyl-enzyme complexes with ceftazidime revealed increased mobility of individual residues of the ⍀-loop for the Asp179Asn variant (average RMSD of 2 Å for KPC-2 compared to a 6 Å average for the variant) (Fig. 6A). The RMSD for the initial trajectory conformation of the KPC-2 ⍀-loop increased from 1.5 Å for the first 120 ps to 3 Å (Fig. 6B). Impressively, the predicted movement of the Asp179Asn variant ranged from 1.4 Å for the first 40 ps to a 9 Å RMSD (Fig. 6B), showing in real time the increased mobility of the variant (Fig. 6C [KPC-2] vs. Fig. 6D [variant]).
(viii) Structural impact on acylation. Notably, a significant difference in hydrogen bond networks in Asp179Asn profoundly alters the position of the catalytic Ser70 in the oxyanion hole (Fig. S7). Ser70 in KPC-2 is oriented toward the oxyanion hole, forming hydrogen bonding interactions with Thr237 and a catalytic water; thus, the active site is "primed" and ready for catalysis. However, the hydroxyl group of Ser70 in the variant is shifted more than 1.5 Å and rotated more than 90°toward Asn170 and Glu166 (Fig. S7). The altered position of Ser70 supports alternative new hydrogen bond interactions with Lys73 and Asn170, which could make acylation more challenging for the variant.
(ix) Structural impact on deacylation. The heat map of the hydrogen bonds generated during the molecular dynamics simulation (MDS) (Fig. S8A to F) suggests two possible pathways for ceftazidime deacylation by KPC-2. In the first pathway, Glu166 could act as a general base and activate the water molecule for proton transfer (for the first 10 ps, Glu166 forms a hydrogen bond with Ser70:O␥) (Fig. S8A). Subsequently, the water is positioned between Glu166:O and Ser70:O␥, allowing the hydrogen bond between Glu166 and the oxygen O and between Ser70 and the O␥ to alternate between the two hydrogen atoms of the water (Fig. S8B). Alternatively, K73 could serve as a proton donor (for t Ͼ 20 ps, the Lys73 is at a hydrogen bond distance from Ser70:O␥) (Fig. S8C). The trajectory of the KPC-2 acyl-enzyme for the first 20 ps shows the catalytic water positioned at a hydrogen bond distance from Glu166 and Ser70:O␥ ( Fig. 6E and S8B). Lys73 forms a hydrogen bond with Glu166 and is less than 3 Å distant from Ser70:O␥.
In the Asp179Asn variant, the water molecule makes hydrogen bonds with Glu166:O but not with Ser70:O␥ ( Fig. S8D and E). Instead, for the first 240 ps, Glu166 forms hydrogen bonds with Asn170 and water ( Fig. 6E and S8D). Glu166 is 5 Å from S70:O␥, and the water is positioned at 4.8 Å from Ser70:O␥, unfavorable positioning for deacylation (Fig. S8E). Lys73 is oriented toward Asn132 and makes hydrogen bonds with Glu166 (Fig. S8F). After 240 ps, Glu166 and the catalytic water are favorably repositioned to make interactions with Ser70:O␥ to participate in deacylation (Fig. S8E).
Overall, these analyses suggest that Asp179Asn has structural perturbations in the active site and ⍀-loop and associated hydrogen bond networks that result in decreased catalytic efficiency, which is consistent with the biochemical analysis that indicates that the Asp179Asn variant deacylates less rapidly and therefore acts as a "trap" for ␤-lactams.
Conclusions. Many single-amino-acid substitutions at positions 164 and 179 in KPC ␤-lactamase as well as other class A ␤-lactamases result in increased ceftazidime resistance and represent a clinical threat as a potential evolutionary adaptation to the widespread use of ceftazidime and other cephalosporins. One variant ␤-lactamase that is particularly notable is the KPC-2 Asp179Asn variant, as E. coli expressing bla KPC-2 Asp179Asn demonstrated resistance not only to ceftazidime but also to other ␤-lactams and ␤-lactam-␤-lactamase inhibitor combinations, including ceftazidime-avibactam. This laboratory analysis closely recapitulates the clinical observations being reported showing the emergence of KPC variants resistant to ceftazidime-avibactam (15). Strikingly, our data show that all strains containing variants at position 179 had elevated ceftazidime-avibactam MIC values. The clinical appearance of these variant ␤-lactamases poses a serious threat to ceftazidime-avibactam (2,15). In this study, we also found that the drug MIC values of the KPC-2 variants in an isogenic background were decreased significantly with BAL30072 used as a monotherapy or with avibactam (as a model DBO) combined with imipenem, aztreonam, ceftaroline, or ceftriaxone. Although there was a significant reduction in MIC values when the carbapenems alone were tested, the Asp179Asn variant tested resistant to each of the carbapenems except meropenem (nonsusceptible).
Mass spectrometry showed a potential additional step in the enzymatic scheme for ceftazidime hydrolysis. Moreover, mass spectrometry data permit us to advance an explanation for why Asp179Asn shows enhanced ceftazidime resistance but not increased imipenem or aztreonam resistance. We propose that the 5-min "trap" of the acyl enzyme species serves as a "sink" for ceftazidime while still able being to hydrolyze it. However, this mechanism requires that the enzyme concentration be sufficient to "trap" all the ceftazidime, eliminating the intracellular pool of free ceftazidime available to bind to penicillin-binding proteins. In contrast, imipenem is "trapped," but to a greater extent, and is stable at least three times longer, with the Asp179Asn-imipenem complex persisting as the most predominant species at 15 min. Aztreonam is also "trapped," but less so, and the enzyme is subsequently inactivated by avibactam. Structural studies designed to identify the location and orientation of ceftazidime intermediates bound to Asp179Asn are needed to tease apart the details of this complex mechanism. However, it is very clear that the chemical nature of the ceftazidime adducts bound to Asp179Asn changes with time and that secondary-reaction chemistry is under way during catalysis.
Consistent with the increase in ceftazidime resistance for the Asp179Asn substitution in KPC-2, enhanced ceftazidime resistance in the Asp179Asn variant has also been found with TEM-1 (21) and SHV-1 (22) ␤-lactamases, indicating that the increased ceftazidime resistance caused by perturbations at the 179 position is a relatively global phenomenon for class A ␤-lactamases. However, the resulting effects seen in the kinetic characteristics of the Asp179 variants are substrate and enzyme specific. For example, the increased catalytic efficiency of the Asp179Asn variant in TEM was selective for ceftazidime among the nine substrates tested and correlated to increased affinity (21). We found that, relative to KPC-2 ␤-lactamases, the Asp179Asn variant demonstrated decreased hydrolysis of ceftazidime and imipenem and enhanced affinity for all substrates tested (aztreonam, ceftazidime, ceftaroline, ceftriaxone, ceftolozane, imipenem, and BAL30072).
Reduced activity in the Asp179Asn variant of KPC-2 parallels that of the Asp179Asn variant called P54 in PC-1 in Staphylococcus aureus due to disorder of the ⍀-loop induced by dissociation of the salt bridge with Arg164 (23). That study found that the variant ␤-lactamase has an alternative interaction between Asp179Asn and Ala69 not found in PC-1. In addition, Stojanoski et al. showed that mutations in the TEM-1 ⍀-loop induced conformational changes that permitted the subsequent enlargement of the active site to accommodate the large size of ceftazidime (increased burst kinetics) (24). Likewise, we found a correlation between a larger size in the R1 side group of the cephalosporins and increased drug MIC values for the Asp179Asn variant. These data are instrumental in designing effective inhibitors and pairing them with the most efficacious partner.
We found that BAL30072 shows promising activity against E. coli containing KPC-2 and all variants at position 179. These data raise the possibility that antibiotics such as monosulfactams may be optimally suited to pairing with DBO inhibitors for resistant strains. Indeed, several studies have already assessed the use of BAL30072 in combination therapy (25)(26)(27)(28).
In closing, the analyses of Asp179 variants of KPC-2 showed that novel and catalytically versatile ␤-lactamases are emerging in the clinic and present an unprecedented challenge to drug development. The results from our biochemical and molecular studies reveal the basis of this unwelcome phenotype and point to rational approaches to overcome this resistance.
Site-saturation mutagenesis. Escherichia coli containing bla KPC-2 in pBR322-catI vector was a gift from Fred Tenover (previously of the Centers for Disease Control and Prevention, Atlanta, GA) (29). For 16 of the 19 amino acid substitutions, mutagenesis was performed at nucleotides corresponding to position 179 in bla KPC-2 in the pBR322-catI plasmid using degenerate primers and a QuikChange site-directed mutagenesis kit (Agilent Technologies; catalog no. 200518-5) per the manufacturer's instructions. Resulting plasmids were transformed into E. coli DH10B Electromax cells (Invitrogen).
The bla KPC-2 Asp179Glu , bla KPC-2 Asp179Iso , and bla KPC-2 Asp179Ser genes with a ribosomal binding site (nucleotides 236 to 279) from pET24aϩ vector positioned upstream of the bla gene and flanked by XbaI and BamHI restriction sites in the pBluescript II SK vector were purchased from Celtek Genes (Franklin, TN). The bla KPC-2 gene was amplified by PCR (Promega master mix) using T7 and M13 primers and was subsequently cloned into pCR-XL vector using a Topo XL PCR cloning kit (Invitrogen catalog no. 1647751). After electroporation into E. coli DH10B Electromax cells, plasmid from these cells was digested with XbaI and BamHI and ligated into pBC SK(ϩ) vector. All nucleotide sequences corresponding to amino acid substitutions at position 179 in the bla KPC-2 gene were confirmed by sequencing (Molecular Cloning Laboratories, McLab, South San Francisco, CA) using bla KPC-2 primers.
Molecular modeling and docking. A structural representation of the Asp179Asn variant of KPC-2 ␤-lactamase was generated using the crystal coordinates of KPC-2 (PDB: 2OV5) and Discovery Studio 4.1 (DS 4.1; Acclerys Inc., San Diego, CA) molecular modeling software as previously described (7,8). The crystallographic water molecules were maintained during modeling. The KPC-2 ␤-lactamase structure and the variant model were solvated and minimized to an RMS value of 0.03 Å using a conjugate gradient method. To assess the stability of the models and possible conformational changes, molecular dynamics simulation (MDS) was conducted on the apo-enzymes for 0.55 ns.
Ceftazidime and acyl-ceftazidime were constructed using the Fragment Builder tools and minimized using a Standard Dynamics Cascade protocol of DS. The intact ceftazidime and acylated ceftazidime were automatically docked into the active site of KPC-2 and the Asp179Asn variant using the CDOCKER module of DS. To obtain acyl-enzyme complexes, the most favorable pose of ceftazidime demonstrating anticipated active-site contacts (such as a short distance [2 to 3 Å] between Ser70:O and C7 of ceftazidime) was chosen. MDS was conducted for 550 ps on KPC-2 and Asp179Asn as apo-enzymes and acyl-ceftazidime complexes. The trajectories were saved every 2 ps and analyzed for hydrogen bond heat maps, distances, etc.