Analysis of the plasticity of location of the Arg244 positive charge within the active site of the TEM-1 β-lactamase

Abstract

A large number of β-lactamases have emerged that are capable of conferring bacterial resistance to β-lactam antibiotics. Comparison of the structural and functional features of this family has refined understanding of the catalytic properties of these enzymes. An arginine residue present at position 244 in TEM-1 β-lactamase interacts with the carboxyl group common to penicillin and cephalosporin antibiotics and thereby stabilizes both the substrate and transition state complexes. A comparison of class A β-lactamase sequences reveals that arginine at position 244 is not conserved, although a positive charge at this structural location is conserved and is provided by an arginine at positions 220 or 276 for those enzymes lacking arginine at position 244. The plasticity of the location of positive charge in the β-lactamase active site was experimentally investigated by relocating the arginine at position 244 in TEM-1 β-lactamase to positions 220, 272, and 276 by site-directed mutagenesis. Kinetic analysis of the engineered β-lactamases revealed that removal of arginine 244 by alanine mutation reduced catalytic efficiency against all substrates tested and restoration of an arginine at positions 272 or 276 partially suppresses the catalytic defect of the Arg244Ala substitution. These results suggest an evolutionary mechanism for the observed divergence of the position of positive charge in the active site of class A β-lactamases.

Introduction

β-Lactamases provide bacteria with a catalytic mechanism to avoid β-lactam antibiotic mediated inactivation of transpeptidase enzymes [also known as penicillin binding proteins (PBPs)]. β-Lactamases can be divided into four classes based on amino acid sequence homologies.^1–3 The class B enzymes are zinc-dependent metallo-β-lactamases that are structurally and mechanistically independent of the other three classes of active-site serine β-lactamases (A, C, and D). Homology searches combined with multiple sequence alignments have identified several common structural elements between active-site serine β-lactamases and PBPs.^4,5 In contrast to the PBPs, active-site serine β-lactamases have acquired an efficient deacylation capacity that permits release of the inactivated β-lactam drug. Mechanism-based β-lactamase inhibitors are often used in conjuction with penicillins to treat bacterial infections that are drug resistant due to the presence of a class A β-lactamase such as TEM-1. Resistance to β-lactamase inhibitors has emerged in the last several years due to mutations that alter residues Met69, Ser130, or Arg244 in TEM-1 β-lactamase.

TEM-1 is a class A β-lactamase and is among the most common plasmid encoded β-lactamases in Gram negative bacteria. The TEM-1 β-lactamase has been the subject of extensive experimental investigations into the determinants of enzyme structure and function.^6,7 Structural information on acylated versions of TEM-1 β-lactamase as well as the Arg244Ser inhibitor resistant TEM-30 enzyme indicate that Arg244 forms hydrogen bonds with the carboxylate group common to both penicillins and cephalosporins.^8,9 In addition, site-directed mutagenesis and enzyme kinetic characterization as well as molecular dynamics simulation studies indicate that the interaction of the Arg244 guanidinium group with the carboxyl group of the antibiotic is important for both substrate and inhibitor binding.^10–14 Analysis of sequence alignments of class A β-lactamases indicate that Arg244 is not absolutely conserved. However, based on sequences and structures, there appears to be conservation of positive charge in this region that results from arginine extending from positions 220, 244, or 276 as has previously been noted.^14–17 Thus, a critical structural property appears to be conserved in the class A β-lactamase family without strict conservation of sequence at a given position. A phylogenic tree of class A enzymes suggests that divergence of the position of arginine in the 244 region was an early event in the diversification of class A β-lactamases. Therefore, an interesting question is whether the positive charge can be manipulated in the TEM-1 enzyme while retaining function or whether coevolution of the other substitutions is required to allow repositioning of the arginine. To test these possibilities, an arginine was substituted at either position 220, 272, or 276 either in the wild type TEM-1 enzyme or in the presence of an Arg244Ala substitution. The arginine substitutions were either neutral or deleterious in the wild type background but improved the catalytic efficiency of the Arg244Ala enzyme to various extents. The results indicate plasticity in the position of the arginine and suggest possible pathways for the divergence observed in this region among class A β-lactamases.

Results

A comparison of class A β-lactamase structures indicates a convergence of positive charge upon the active site via the extension of an arginine residue from either position 220, 244, or 276 (Fig. 1). The X-ray structures of 20 class A β-lactamases were analyzed, and it was found that all but one contained a positive charge in one of these three positions (Table I). The spatial conservation of positive charge in the β-lactamase active site suggests an accompanying functional role that has been preserved during the evolution of class A β-lactamases. A single exception is provided by the inhibitor resistant TEM-30 clinical variant of TEM-1 having an Arg244Ser substitution.⁹ Because of its medical relevance, the arginine at position 244 has been the subject of several studies in the TEM-1 and the closely related SHV β-lactamases. The results of these studies indicate an important role for Arg244 in β-lactam catalysis and inhibitor interactions.^10,11,14,18

Figure 1.

Alternative positioning of arginine in the active site of class A β-lactamases. Residues at positions 220, 244, 272/274, and 276 are shown in each panel in stick format. (a) TEM-1 β-lactamase has an arginine at position 244 directed toward the carboxyl group of the penicillin G substrate (PDB: 1FQG).⁸ (b) The structure of the Streptomyces albus G β-lactamase has arginines at positions 220 and 274 (PDB: 1BSG).¹⁹ (c) The structure of CTX-M-14 has an arginine at position 276 (PDB: 1YLT)²⁰. In each structure, the guanidinium group of Arg220, Arg244, Arg274, and Arg276 occupies a similar physical position in the active site.

Table I.

Positive Charge Position in the β-Lactamase Active Site

	220	244	276	272a/274
TEM-1 (1M40)b	Leu	Argc	Asn	Meta
TEM-30 (1LHY)	Leu	Ser	Asn	Meta
SHV-2 (1N9B)	Leu	Arg	Asn	Meta
Pse-4 (1G68)	Leu	Arg	Asn	Meta
Ges-1 (2QPN)	Thr	Arg	Asp	Alaa
L2 (1N4O)	Cys	Arg	Ala	Tyra
B. licheniformis (1I2S)	Leu	Arg	Asp	Tyr
PC1 (1BLC)	Leu	Arg	Asp	Pro
E. faecalis (3CJM)	Arg	Ile	Ala	Asna
Per-1 (1E25)	Arg	Thr	Glu	Sera
M. tuberculosis (3CG5)	Arg	Ala	Glu	Pro
S. albus G (1BSG)	Arg	Asn	Asp	Arg
SME-1 (1DY6)	Arg	Ala	Asp	His
Nmc-A (1BUE)	Arg	Ala	Asp	His
KPC-2 (2OV5)	Arg	Ala	Glu	His
M. fortuitum (2CC1)	Ser	Thr	Arg	Asn
CTX-M-14 (1YLT)	Ser	Thr	Arg	Ser
Toho-1 (1IYS)	Ser	Thr	Arg	Arg
K1 (1HZO)	Ser	Thr	Lys	Tyr
Sed-1 (3BFE)	Ser	Thr	Lys	Trp

^aExtension of α11 causes position 272 to occupy a position similar to 274.

^bPDB codes are provided in parenthesis.

^cPositively charged residues are in bold.

In the β-lactamase from Streptomyces albus G, an arginine extends from position 220 rather than from position 244.¹⁹ In support of an analogous role for Arg220 in substrate binding and catalysis, an R220L mutation of the Streptomyces albus G β-lactamase greatly impairs the catalytic ability of this enzyme.¹⁵ The Streptomyces albus G β-lactamase structure includes an additional positively charged residue at nearby position 274 (Fig. 1). However, the presence of positive charge at position 274 is neither highly conserved among β-lactamase crystal structures nor is there a clear correlation with the residues found at the other three positions (Table I). No investigations have been performed to determine what role, if any, this additional positive charge may have in β-lactamase activity. The CTX-M enzymes contain an arginine at position 276 (Table I). Site-directed mutagenesis of Arg276 in CTX-M-1 and CTX-M-4 resulted in a loss of resistance conferred to E. coli against some β-lactams. However, catalytic parameters of the Arg276 substituted enzymes were only moderately affected with the substrates tested.^21,22

Taken together, the site-directed mutagenesis experiments suggest that the arginines found at positions 220, 244, or 276 provide a contribution toward β-lactamase activity, and the variation in position from which this positively charged residue extends indicates a selective pressure for maintaining positive charge within this region of the active site.

To extend these observations, a phylogenetic tree of 156 class A β-lactamase sequences was assembled. Figure 2 provides a juxtaposition of charge position over the branches of the β-lactamase phylogenetic tree. Of the 156 sequences, only three lack positive charge at positions 220, 244, or 276. A single enzyme, BES-1, has arginines at both positions 220 and 276. A vast majority of sequences maintained positive charge at either position 220, 244, or 276. This nearly absolute conservation further indicates an important role for the positive charge in this region of the active site.

Figure 2.

Phylogenetic tree of 156 class A β-lactamase sequences. Branch colors correspond with enzymes having Arg/Lys220 (blue), Arg244 (black), or Arg/Lys276 (red). Three enzymes lacking positive charge at any of the three positions (orange) and a single enzyme having both Arg220 and Arg276 (green) are also indicated by branch color. Enzymes for which a crystal structure is available are labeled. Because of tight clustering, dashed edges connect the Nmc-A, Sme-1, and KPC-2 carbapenemase sequences to their labels.

The architecture of the phylogenetic tree permits inference into the evolutionary history of the positive charge in the β-lactamase active site. Although the majority of β-lactamase sequences have Arg244, the sequences with Arg220 are much more broadly distributed across the tree. The diverse and deeply rooted Arg220 branches suggest the existence of an Arg220 β-lactamase precursor from which the Arg244 and Arg276 β-lactamases emerged. Four evolutionarily distinct branches have an arginine at position 244 (Fig. 2). Each of the four Arg244 branches included in the phylogenetic analysis is more closely related to a branch having an Arg220 than to the other Arg244 branches. This suggests that multiple, independent mutational events occurred that resulted in a switch from an Arg220 β-lactamase to an Arg244 β-lactamase. In contrast, the Arg276 branch is a discrete unit suggesting a single divergence point (Fig. 2). The BES-1 β-lactamase sequence, having both Arg220 and Arg276, lies at the intersection of the Arg276 branch and the Arg220 branch that is composed of carbapenemases (e.g. SME-1, KPC-2). The BES-1 enzyme could represent a transition from an Arg220 β-lactamase to an Arg276 β-lactamase. Although perhaps by coincidence, the most closely related β-lactamase to BES-1 is ERP-1, a β-lactamase devoid of positive charge at all three positions (Supporting Information Fig. 1).

The phylogenetic tree presents two alternative evolutionary pathways in switching the location of positive charge. In one scenario, a precursor acquires a mutation yielding a second arginine (e.g. BES-1) followed by the loss of the original arginine by mutation. The second scenario is simply an inversion of the mutational steps whereby the precursor loses the original arginine (e.g. ERP-1) followed by the mutational gain of a second arginine. The two alternative evolutionary pathways were investigated using site-directed mutagenesis of the well-characterized TEM-1 β-lactamase.

In TEM-1 β-lactamase, the side chains of residues Leu220, Met272, and Asn276 are directed toward and restrict the movement of Arg244.^8,23 As a result of an extended helix α11, Met272 of TEM-1 occupies a similar physical space as the residue at position 274 in the CTX-M-14 and Streptomyces albus B β-lactamases (Fig. 1, Table I). Each of these residues is critical for wild type TEM-1 catalytic function in that amino acid substitutions at these positions result in decreased activity.⁶ β-lactamase inhibitor resistant mutants from clinical isolates with amino acid substitutions at residues Arg244 or Asn276 are less efficient at β-lactam hydrolysis but are capable of avoiding inactivation by inhibitors.^24,25 Examination of the TEM-1 structure suggests that relocating the arginine at position 244 to either positions 220, 272, or 276 will permit the guanidium group of arginine to approach the carboxyl group of the β-lactam substrate. Modeling the Leu220Arg mutation revealed only one favorable conformation as a result of the buried nature of this position. In contrast, all conformations of the modeled Met272Arg mutation were favorable and the Asn276Arg substitution had several high and low energy conformations possible. Site-directed mutagenesis was used to place an additional arginine at the target positions (Leu220Arg, Met272Arg and Asn276Arg). Next, a mutant was created that replaced the arginine at position 244 with an alanine (Arg244Ala). Finally, the Arg244Ala substitution was combined with each of the Leu220Arg, Met272Arg, or Asn276Arg substitutions with the intent of returning the guanidinium group to the area left vacant from the removal of arginine at position 244 (Leu220Arg:Arg244Ala, Arg244Ala:Met272Arg, and Arg244Ala:Asn276Arg). Based on the sequence and structure comparisons of enzymes in the class A family, the second site arginine is hypothesized to restore catalytic function to the Arg244Ala mutant.

As an initial assessment of function, in vivo resistance levels were measured by determining the minimum inhibitory concentration (MIC) of ampicillin for E. coli containing plasmids encoding the various mutants. The mutation of position 220 or 276 to arginine in TEM-1 resulted in very low ampicillin resistance levels for E. coli containing the mutants (Table II). The introduced arginine at position 220 or 276 may contact the naturally occurring arginine at position 244 and perturb its function via charge repulsion. This result suggests that structural constraints would prevent TEM-1 from acquiring an additional arginine at position 220 or 276. In contrast, the Met272Arg mutant maintains wild type levels of ampicillin resistance even in the presence of arginine at position 244 (Table II). This result agrees with modeling results in which none of the Met272Arg conformers exhibited steric clashes with Arg244. As previously mentioned, some class A β-lactamases have a positively charged residue at position 274 (structurally equivalent to position 272 in TEM-1) and another positively charged residue at either position 220 or 276. This indicates a positively charged residue at position 272/274 can also be compatible with a positively charged residue at position 244, at least with the TEM-1 enzyme.

Table II.

Minimum Inhibitory Concentrations of Ampicillin for TEM-1 and Mutants

Clone	MICa
BL21(DE3)	≤4
TEM-1b	2048–4096
L220R	4
R244A	256
M272R	2048
N276R	<4
L220R:R244A	16
R244A:M272R	256
R244A:N276R	512

^aμg/ml ampicillin.

^bTEM-1 and mutants used in this study are from pET-TEM1 and contain Glu28Gly. See Materials and Methods.

Substitution of the arginine at position 244 with an alanine in the TEM-1 enzyme eliminates positive charge in the region near the substrate carboxylate group. As expected, the Arg244Ala mutation significantly reduced the ampicillin resistance of E. coli containing the mutant (Table II). Similar results were obtained in previous studies examining TEM-1 Arg244 substitutions and the observation is consistent with the proposed role of the arginine side chain stabilizing substrate through interactions with the carboxylate group of β-lactam antibiotics.^10,26,27

Site-directed mutagenesis was used to combine each of the Leu220Arg, Met272Arg, and Asn276Arg mutations with the Arg244Ala mutation with the intent of returning a guanidinium group to the area left vacant from the removal of arginine 244. The effect of the arginine substitutions at 220, 272, and 276 in the context of the Arg244Ala mutant varied widely with the mutants (Table II). Although attenuated relative to Arg244Ala, the Leu220Arg:Arg244Ala clone provided ampicillin resistance at levels greater than the clone with only the Leu220Arg mutation. The Arg244Ala:Met272Arg mutant provided ampicillin resistance at the same level as Arg244Ala alone and, in contrast to the loss of function seen with the Asn276Arg mutant, the Arg244Ala:Asn276Arg clone provided ampicillin resistance levels higher than the Arg244Ala mutant. The loss of function seen with the Leu220Arg and Asn276Arg single substitutions and the subsequent recovery of ampicillin resistance in the Arg244Ala background (Leu220Arg:Arg244Ala and Arg244Ala:Asn276Arg) suggests the introduced arginines at positions 220 and 276 may partially assume the function of the missing arginine side chain at position 244. The data also suggests that structural constraints favor an evolutionary pathway from the TEM-1 Arg244 enzyme to a TEM-1 Arg276 enzyme that proceeds through an intermediate lacking positive charge.

To confirm and extend the MIC results, steady state kinetic parameters for a range of substrates were obtained with purified mutant enzymes (Table III, Supporting Information Fig. 2). In agreement with the MIC data, the Arg244Ala mutant exhibited reduced catalytic efficiency (k_cat/K_m) for hydrolysis of each substrate tested. For ampicillin, benzylpenicillin and nitrocefin, the loss of activity of the Arg244Ala mutant was largely due to an increase in K_m (Table III). Previous studies with various TEM-1 Arg244 mutants have shown a similar effect upon K_m with penicillin substrates.^28,29 The K_m values for cephalothin and cephalosporin C were too high to be reliably determined and so the relative contribution of K_m and k_cat upon catalytic efficiency of TEM-1 Arg244Ala is unknown for these substrates. However, the catalytic efficiency of the Arg244Ala enzyme, relative to wild-type TEM-1, was reduced >1000-fold for the cephalothin and cephalosporin C substrates (Table III). The Arg244Ser mutation of the inhibitor resistant TEM-30 β-lactamase results in displacement of a water molecule that likely provides the proton necessary for inactivation by the inhibitor clavulanic acid.^9,11,30 As expected, the Arg244Ala enzyme exhibits a large increase in the IC₅₀ value for clavulanic acid inhibition (Table III).

Table III.

Enzyme Kinetic Parameters of TEM-1 β-Lactamase and Mutant Enzymes.^a

	TEM-1a	R244A	L220R	M272R	N276R	L220R: R244A	R244A: M272R	R244A: N276R
Ampicillin
Km (μM)	65.2	1630	3666	272	871	536	296	161
kcat (s−1)	960	1000	1749	1578	625	207	1630	1610
kcat/Km (μM−1 s−1)	14.8	0.61	0.48	5.80	0.72	0.39	5.49	10.0
Benzylpenicillin
km (μM)	46.1	321	548	81	451	171	324	114
kcat (s−1)	534	658	310	749	414	38.8	504	676
kcat/Km (μM−1 s−1)	11.6	2.05	0.57	9.25	0.92	0.23	1.56	5.9
Nitrocefin
km (μM)	72.5	1980	ND	260	256	NDb	258	233
kcat (s−1)	464	782	ND	991	48.4	ND	711	978
kcat/Km (μM−1 s−1)	6.40	0.40	0.03	3.81	0.19	0.01	2.76	4.20
Cephalothin
km (μM)	228	ND	ND	375	ND	ND	ND	ND
kcat (s−1)	130	ND	ND	96	ND	ND	ND	ND
kcat/Km (μM−1 s−1)	0.57	5.4 × 10−4	0.01	0.26	0.023	5.5 × 10−4	2.1 × 10−3	8.0 × 10−3
Cephalosporin C
km (μM)	903	ND	ND	414	ND	ND	ND	ND
kcat (s−1)	53.9	ND	ND	90	ND	ND	ND	ND
kcat/Km (μM−1 s−1)	0.06	2.0 × 10−5	1 × 10−3	0.22	3.2 × 10−3	(−)c	4.5 × 10−4	8.5 × 10−4
Clavulanic acid
IC50 (μM)	0.05	5.56	0.11	0.07	0.04	0.72	1.73	0.28

^aTEM-1 and mutants used in this study are from pET-TEM1 and contain Glu28Gly. See Materials and Methods.

^bNot determined.

^cNo activity detectable.

Standard error on measurements is +/− 20%.

The impact of introducing an arginine at positions 220, 272, and 276 in the presence of the wild type Arg244 residue was evaluated by obtaining kinetic measurements for the purified Leu220Arg, Met272Arg and Asn276Arg TEM-1 enzymes. The kinetic data for these single mutants support the observation during phylogenetic analysis of interdependence among positions 220, 244, and 276. Where measurable, K_m was increased for all substrates tested for the Leu220Arg and Asn276Arg mutants resulting in greater than 10-fold loss of catalytic efficiency (Table III, Fig. 3). As observed with the Arg244Ala mutant for the cephalothin and cephalosporin C substrates, K_m becomes too large to be accurately measured with the Leu220Arg and Asn276Arg enzymes. In contrast, K_m values of the Met272Arg enzyme were only moderately affected for all substrates tested and were in fact reduced in the case of cephalosporin C.

Figure 3.

Percent catalytic efficiency (k_cat/K_m) of engineered TEM-1 β-lactamases relative to wild type TEM-1. Abbreviations in legend are for the substrates ampicillin (AMP), penicillin G (PENG), nitrocefin (NCF), cephalothin (CF), cephalosporin C (CPC) and clavulanic acid (CA). Clavulanic acid inhibition is indicated as percent 1/IC50 relative to wild type TEM-1.

Although the Leu220Arg and Asn276Arg mutants show deficiencies in substrate hydrolysis, inhibition by the suicide inhibitor clavulanic acid occurs at levels similar to that observed with wild type TEM-1. If the presence of an additional arginine at positions 220 or 276 were perturbing the function of Arg244 in substrate hydrolysis one might expect a concomitant effect upon binding and hydrolysis of the clavulanic acid inhibitor. However, the reaction pathway for substrate hydrolysis differs from the terminal reaction with clavulanic acid and introduction of a second arginine could favor the terminal pathway. Regardless of the precise enzymatic mechanisms, the overall reduction in substrate hydrolysis and maintenance of inhibitor sensitivity suggests that the Leu220Arg and Asn276Arg enzymes would be unfavorable intermediates in the evolutionary pathway that results in a relocation of positive charge in the TEM-1 active site.

The catalytic efficiency of the Leu220Arg:Arg244Ala enzyme is highly attenuated for all the substrates tested (Table III). However, K_m values for ampicillin and benzylpenicillin are lower relative to both the Arg244Ala and Leu220Arg mutants suggesting that the arginine side chain at position 220 is positioned to restore affinity with the substrate carboxylate group. The IC₅₀ for clavulanic acid inhibition was also reduced approximately eightfold for this enzyme relative to Arg244Ala (Table III). This result is likely due to interactions between the Arg220 side chain and the carboxylate group of clavulanic acid.

The Arg244Ala:Met272Arg mutant hydrolyzed benzylpenicillin with kinetic parameters similar to the Arg244Ala enzyme. For the other substrates tested, however, catalytic efficiencies were increased 4- to 23-fold relative to Arg244Ala. The catalytic efficiencies for ampicillin and nitrocefin hydrolysis were restored to 37% and 43% of wild-type levels as a result of improved K_m values (Table III). The clavulanic acid IC₅₀ value for the Arg244Ala:Met272Arg mutant was also significantly reduced compared to Arg244Ala. The increased level of inactivation is consistent with the lowered K_m values for β-lactam substrates and may reflect stabilization of the carboxylate of clavulanic acid by the arginine side chain at position 272. Thus, it appears the arginine at position 272 partially compensates for the loss of arginine at position 244.

The most striking effect in terms of restoration of Arg244Ala function was observed with the Arg244Ala:Asn276Arg enzyme (Table III, Fig. 3). All substrates were hydrolyzed with increased catalytic efficiency compared to the Arg244Ala enzyme (3- to 43-fold increase of k_cat/K_m) and nearly all substrates were hydrolyzed more efficiently than the Asn276Arg enzyme. However, cephalothin and cephalosporin C were still poorly hydrolyzed by the Arg244Ala: Asn276Arg enzyme when compared to hydrolysis of these substrates by either wild-type TEM-1 β-lactamase or the Asn276Arg mutant (Table III). Inhibition by clavulanic acid, as indicated by an IC₅₀ of 0.28 μM, was also more pronounced with this enzyme than seen with the other engineered β-lactamases. The success of the Arg244Ala:Asn276Arg enzyme to partially restore function may be related to increased amount of solvent exposure of the Arg276 side chain which could facilitate access to a side chain conformation that positions the guanidinium group for optimal interactions with the substrate carboxylate. The synergistic effect upon ampicillin and benzylpenicillin hydrolysis by combining the Arg244Ala and Asn276Arg substitutions suggests that removal of an arginine at position 244 allows an arginine placed at position 276 to act as a functional analog for antibiotic and inhibitor binding.

Taken together, the double mutant experiments with Arg244Ala indicate that a positive charge normally provided by residue 244 can be provided from alternate positions. However, the restoration of Arg244Ala enzyme function is partial and is strongly dependent on the specific residue position of the compensating arginine.

DISCUSSION

The role of the arginine at position 244 in substrate hydrolysis and inhibitor resistance has been well studied in several class A β-lactamases with regards to its role in substrate interactions.^{9,11,14,18,30,31} In TEM-1, the guanididium group of Arg244 makes a direct interaction with the C3/C4 carboxylate group and forms a hydrogen bond with an active site water molecule that also interacts with the C3/C4 carboxylate.^8,9,29 The importance of the arginine-substrate carboxylate interaction is supported by substitutions created by site-directed mutagenesis at Arg244 that exhibit reduced catalytic efficiency for β-lactam substrates. The effect on k_cat/K_m of Arg244 substitutions were found to be large due to increased K_m values and the results of this study with the Arg244Ala enzyme are consistent with those observations.¹⁴ The significance of K_m values for the β-lactamase acyl-enzyme mechanism depends on whether acylation is rate limiting. It has been reported that the similarity of first order rate constants makes K_m approximately equal to K_s and therefore increased K_m values would be reflective of decreased substrate binding, which is consistent with the loss of the guanidinium-carboxylate interaction.³²

Substitutions of Arg244 in the TEM-1 and SHV-1 β-lactamase have also been studied intensively due to their role in inhibitor resistance. The Arg244Ser and Arg244Cys substitutions in TEM-1 β-lactamase have been found in bacterial isolates resistant to β-lactam antibiotic-clavulanic acid combinations.^10,27,33 These substitutions have been shown to result in the loss of affinity for clavulanic acid in addition to the loss of a conserved water molecule that is involved in the inactivation mechanism.^8,9,30 Inhibitor-resistant clinical isolates containing SHV-1 β-lactamase with substitutions at Arg244 have not been found but it has been shown that substitution of Arg244 results in a loss of affinity for clavulanic acid with a resulting increase in IC₅₀.^8,18 Consistent with these previous studies on TEM-1 and SHV-1, the Arg244Ala substitution created in this study exhibited a 100-fold increase in clavulanic acid IC₅₀ suggesting greatly reduced affinity for the compound.

Analysis of a phylogenetic tree of 156 class A β-lactamase sequences identified residues 220 and 276 as alternatives to position 244 for the placement of positive charge in the active site. Such an alternate role has previously been suggested based on structural analysis of class A β-lactamases.¹⁷ In addition, analysis of the tree suggests Arg220 was the ancestral position of the positive charge which subsequently diverged to positions 244 and 276 in subgroups of enzymes. These observations, combined with the crystal structure data available for 19 independent β-lactamases, suggest that residues at positions 220, 244, 276 and, perhaps, 272/274 are acting as an interdependent unit that has undergone co-variation during the evolution of class A β-lactamases. The divergence in charge position from 220 to either 244 or 276 could have proceeded via a β-lactamase lacking positive charge at either position 220, 244, or 276. Among the 156 class A sequences aligned, it is noteworthy that the ERP-1 enzyme lacks a positively charged residue at positions 220, 244, and 276 indicating such an evolutionary intermediate could exist. Alternatively, divergence could have occurred from a β-lactamase having multiple positively charged residues in this region of the active site. In this regard, the multiple sequence alignment of β-lactamases contained only one enzyme, BES-1, which contained two positively charged residues in the 220, 244, and 276 cluster. The existence of the BES-1 enzyme suggests this pathway may be possible for some β-lactamases. It should also be noted that a structure is available for a low molecular weight class C penicillin binding protein that has arginines at positions equivalent to positions 220, 274, and 276 of class A β-lactamases.³⁴ The class A β-lactamases are more closely related to the low molecular weight class C PBPs than the other classes of β-lactamases⁵ and may have shared a common progenitor with multiple positively charged residues oriented toward the active site.

To determine if the modularity of positive charge seen in the family of class A β-lactamases could be applied to a single β-lactamase, the arginine at 244 of TEM-1 was transferred to positions 220, 272, or 276 by site directed mutagenesis. Steady state kinetic analysis revealed various degrees of enzyme activity for each engineered TEM-1 β-lactamase variant. Although moving the arginine to position 220 in the Arg244Ala enzyme failed to improve catalytic efficiency, the penicillin substrate K_m values and clavulanic acid IC₅₀ values decreased relative to the Arg244Ala parent enzyme suggesting a partial restoration of affinity for the substrate carboxylate group. Placing the arginine at position 272 in the Arg244Ala enzyme resulted in improved catalytic efficiency for the majority of substrates tested and increased inhibition by clavulanic acid compared to the Arg244Ala mutant. Molecular modeling of the Met272Arg substitution suggests the restoration of activity is due to the Arg272 side chain replacing the lost interaction between Arg244 and the substrate carboxylate group. The fact that Arg272 substitution acts by decreasing the K_m for substrates relative to the Arg244Ala parent also supports a role for improved substrate binding.

The introduction of the Asn276Arg substitution into the Arg244Ala enzyme resulted in the most striking improvement in antibiotic hydrolysis among the mutants constructed. The catalytic efficiency of the Arg244Ala:Asn276Arg enzyme relative to Arg244Ala alone was increased for all substrates tested and this effect was mediated through decreased K_m values. The IC₅₀ for clavulanic acid was also significantly reduced compared to Arg244Ala. Taken together, these findings strongly support a substrate binding role for the repositioned arginine. It is unclear why the Arg244Ala:Asn276Arg enzyme exhibits much higher catalytic efficiency for hydrolysis of β-lactam substrates than the Leu220Arg:Arg244Ala enzyme. One possibility is that the Leu220 side chain is largely buried and thus constrained in TEM-1 β-lactamase while the Asn276 residue is solvent exposed. The arginine substitution at position 276 may therefore be able to sample more conformations and thereby exhibit increased occupancy of conformations that stabilize the substrate carboxylate group.

This site-directed mutagenesis study addressed the question of whether the positive charge at position 244 could be relocated to positions 220, 272, or 276 in TEM-1 β-lactamase. The complementary question is whether the arginine found at positions 220, 272/274, or 276 in natural class A enzymes plays an analogous role as Arg244 in facilitating substrate and inhibitor binding. As discussed earlier, substitution of Arg220 in the S. albus G β-lactamase greatly reduces substrate binding, consistent with a role in stabilizing the carboxylate, however, inhibitor binding was not assessed.¹⁵ Site-directed mutagenesis studies in the CTX-M-4 and CTX-M-1 enzymes indicate Arg276 in these enzymes contributes to cefotaxime hydrolysis although the impact of substitutions is relatively low and K_m values are largely unaffected.^21,22 In addition, the IC₅₀ for clavulanic acid inhibition for substituted enzymes was similar to the wild type value. Therefore, despite the fact that an arginine at position 276 can largely compensate for the defect in the TEM-1 Arg244Ala mutant, Arg276 does not appear to play a similar role in the CTX-M enzymes. The CTX-M enzymes also contain a serine at position 237 (TEM-1 Ala237) and an arginine at position 274 (TEM-1 Met272) which differ from TEM-1. It is possible that the Ser237-Arg274-Arg276 triad together contributes to interactions with the substrate/inhibitor carboxylate group and mutation of any one residue does not result in a large defect. Additional studies are required to address this possibility as well as to define the residues critical for inhibitor interactions in the CTX-M enzymes.

It is important to note that, in natural enzymes, the position of the arginine at position 220, 244, or 276 could significantly influence the substrate specificity and therefore the antibiotic resistance profile provided by the enzyme. In addition, the position of the arginine may influence the evolutionary potential of a β-lactamase with respect to how the substrate specificity can be altered by mutation.

The β-lactamase phylogenetic tree is a sequence map of the protein family but it does not provide a detailed view of the functional sequence space accessible to a single member within the β-lactamase family. Although phylogenetic analysis indicates TEM-1 is more closely related to β-lactamases possessing an arginine at position 220, the Leu220Arg:Arg244Ala TEM-1 mutant provided the poorest β-lactamase activity of the engineered mutants. In contrast, the Arg244Ala:Met272Arg and Arg244Ala:Asn276Arg TEM-1 mutants possessed significant activity even though β-lactamases with an arginine at positions equivalent to 272 or 276 are more distantly related to TEM-1. Overall, these experiments indicate that phylogenetic analysis can guide protein engineering but an individual enzyme's plasticity is not limited to sampling the functional sequence space of its most closely related homologues.

Materials and Methods

Bacterial strains and pET-TEM-1 plasmid

E. coli K12 XL1-Blue strain (recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [F′proAB lacI^qZΔM15 Tn10 (Tet^r)]) was obtained from Stratagene (La Jolla, CA) and utilized in site-directed mutagenesis experiments. The E. coli BL21(DE3) strain was used in MIC determinations and for protein expression and purification. The pET-TEM-1 vector encodes an ompA leader-TEM-1 fusion gene driven by the T7 promoter.³⁵ The TEM-1 enzyme encoded in pET-TEM-1 contains a non-canonical Glu28Gly substitution as a result of a g→a nucleotide mutation present in the parent vector. All β-lactamases used in this study are derived from the pET-TEM-1 plasmid and therefore contain the Glu28Gly substitution. Residue 28 is the third N-terminal residue after the signal cleavage site. No detectable change in TEM-1 activity occurs as a result of this substitution³⁶ and NMR studies indicate the backbone atom chemical shifts are similar to wild type.³⁷

Site-directed mutagenesis

All TEM-1 mutants were created using the QuikChange kit according to the manufacturer's protocol (Stratagene, La Jolla, CA). Primers were obtained from Integrated DNA Technologies (Coralville, IA). Top strand primers introduced the following mutations (underlined) into pET-TEM-1: Leu220Arg (5′-GCAGGACCACGTCTGCGCTCGGCCC-3′), Arg244Ala (5′-GCGTGGGTCTGCTGGTATCATTGCAGCACTGGG-3′), Met272Arg (5′-GGGAGTCAGGCAACTCGTGATCAACG-3′) and Asn276Arg (5′-GGCAACTATGGATGAACGA CGTAGACAGATCGC-3′). All TEM-1 mutant constructs were sequenced using Applied Biosystems Instruments (Foster City, CA) Prism Big Dye DNA sequencing with an ABI 3100 automated sequencer.

Phylogenetic analysis of class A β-lactamases

The dataset of 156 class A β-lactamase sequences was constructed using the ET Viewer available online at http://mammoth.bcm.tmc.edu/traceview/index.html.^38–40 Because of the frequently deposited TEM and SHV sequences and to insure a representative sampling of class A β-lactamase sequences, multiple queries were performed using the sequences from the following PDB codes: 1AXB, 1BLC, 1BUE, 1DY6, 1IYS 1G68, 1I2S, 2H5S, 1N4O, 1E25, and 2CC1. Additional sequences were systematically added by utilizing the NCBI protein BLAST search. Sequences that created large gaps, insertions or having ≥95% identity with another sequence in the dataset were removed. The β-lactamase branches represented by the Mycobacterium and PER-1 crystal structures were excluded from the dataset because an active site misalignment was identified upon comparing crystal structure alignments to the multiple sequence alignment. A Supporting Information data file containing the list of β-lactamases and their amino acid sequences used for this analysis is included as Supporting Information Data File 1.

Modeling residue substitutions in TEM-1 β-lactamase

The DeepView/Swiss-PdbViewer (http://www.expasy.org/spdbv/) was utilized to model amino acid substitutions as positions 220, 244, 272, and 276.⁴¹ A PBD file for TEM-1 β-lactamase, 1M40, was loaded into Swiss-PdbViewer and the mutation function was utilized to determine possible rotamers of substituted side chains and the score of those rotamers as described in the Swiss-PdbViewer User Guide. Briefly, possible rotamers are calculated by DeepView/Swiss-PdbViewer and can be manually toggled by the user. Corresponding rotamer scores are displayed and account for potential clashing and bond formation with other residues in the given structure. Freedom of movement for the modeled residue substitutions was qualitatively judged by the number of rotamers showing low energy scores.

Minimum inhibitory concentration (MIC) determination

In vivo antibiotic resistance levels were determined to evaluate the function of TEM-1 β-lactamase mutants. Ampicillin MICs were determined by broth dilution in 96-well format. Twofold dilutions of ampicillin were tested in each well containing 1 × 10⁴ bacteria per ml in 100 μL of LB broth containing 33 μM IPTG. The 96-well plate was sealed and allowed to incubate overnight at 37°C before scoring for growth. Each assay was conducted in at least duplicate using independent overnight cultures.

Protein purification

The wild-type and mutant TEM-1 (E28G) β-lactamases were purified to >90% homogeneity. E. coli BL21(DE3) cells transformed with the relevant pET-TEM-1 construct were grown in LB broth with 300 mM sorbitol, 250 mM betaine and 25 μg/mL kanamycin to OD₆₀₀ 0.8 and protein expression was induced with 0.4 mM IPTG. The IPTG-induced culture was grown overnight with shaking at room temperature. Cells were harvested by centrifugation and the periplasmic contents were obtained by osmotic shock. The periplasmic fraction was dialyzed overnight at 4°C before loading onto a Hi-Trap zinc chelating column (Amersham, GE Healthcare, Piscataway, NJ) charged with ZnCl₂. The β-lactamase containing fractions were eluted using a pH gradient and purity was determined by SDS-PAGE gel electrophoresis. Buffer exchange into 50 mM PO4 pH 7.0 was facilitated by use of a Centricon centrifugal filter (Millipore, Billerica, MA). Protein concentrations were determined with the Bio-Rad Bradford protein assay reagent (Hercules, CA).

Enzyme kinetics

Michaelis-Menten steady-state kinetic parameters were determined on a Beckman-Coulter spectrophotometer model DU 800 (Fullerton, CA). Substrate hydrolysis was monitored at the following wavelengths: ampicillin, 235 nm; benzylpenicillin, 233 nm; nitrocefin, 482 nm; cephalothin, 262 nm; cephalosporin C, 280 nm. Reactions were performed at 30°C in 50 mM phosphate buffer pH 7.0 with 1 mg/mL BSA. BSA was excluded from ampicillin and benzylpenicillin reactions due to an overlap in protein absorbance with the measurement wavelength. K_m and k_cat values were determined by fitting initial velocity rates over a range of substrate of concentrations to a Michaelis-Menten curve using SigmaPlot. In cases were TEM-1 mutants had K_m values that prevented measurements to be made at saturation, catalytic efficiency (k_cat/K_m) was determined by the equation: V_i = [k_cat/K_m][E][S] where [S]<<K_m.⁴² Clavulanic acid IC₅₀ values were determined by measuring initial velocity of hydrolysis of 100 μM nitrocefin after incubation of the enzyme with various concentrations of inhibitor ranging from 1 pM to 10 mM for 20 min. Measurements were performed in at least triplicate and the values given are an average of these determinations. The values have standard error of <20%.