|Home | About | Journals | Submit | Contact Us | Français|
6-β-Halogenopenicillanates are powerful irreversible inhibitors of various β-lactamases and penicillin binding proteins. Upon acylation of these enzymes, the inhibitors are thought to undergo a structural rearrangement associated with the departure of the iodide and formation of a dihydrothiazine ring, but, to date, no structural evidence has proven this. 6-β-Iodopenicillanic (BIP) is here shown to be an active antibiotic against various bacterial strains and an effective inhibitor of the class A β-lactamase of Bacillus subtilis BS3 (BS3) and the D,D-peptidase of Actinomadura R39 (R39). Crystals of BS3 and of R39 were soaked with a solution of BIP and their structures solved at 1.65 Å and 2.2 Å, respectively. The β-lactam and the thiazolidine rings of BIP are indeed found to be fused into a dihydrothiazine ring that can adopt two stable conformations at these active sites. The rearranged BIP is observed in one conformation in the BS3 active site and in two monomers of the asymmetric unit of R39, and is observed in the other conformation in the other two monomers of the asymmetric unit of R39. The BS3 structure reveals a new mode of carboxylate interaction with a class A β-lactamase active site that should be of interest in future inhibitor design.
The antibacterial activity of penicillin rests upon its ability to inhibit the enzymatic activity of penicillin-binding proteins (PBPs), which are responsible for the late stages of peptidoglycan biosynthesis. PBPs are DD-peptidases that cleave the peptide bond between the C-terminal D-Ala-D-Ala of the peptidoglycan stem pentapeptide. In a similar way to that in which the active serine of PBPs attacks the peptidic amide bond, it also attacks the endocyclic amide bond of the penicillin β-lactam ring, leading to a long-lived acyl-enzyme PBP-penicillin that impairs the DD-peptidase activity 1. Penicillin derivatives and other β-lactam antibiotics (cephalosporins, cephamycins, carbapenems and monobactams) or γ-lactam antibiotics (lactivicin, pyrazolidinones) with enhanced antibacterial activity were either discovered as natural metabolites produced by bacteria and fungi or developed from them by direct chemical elaboration 2–4.
A major mechanism of resistance of bacteria is the production of β-lactamases, which are able to hydrolyse the endocyclic amide bond of the β-lactam ring and release the hydrolysed product 5. Finding β-lactamase inhibitors has been a considerable field of research leading, for example, to the discovery and development of clavulanic acid, tazobactam, sulbactam, BRL42715, and 6-β-halogenopenicillanates 6–11. 6-β-Iodopenicillanates and 6-β-bromopenicillanates are powerful inhibitors of β-lactamases 10–14. Inactivation of the class A β-lactamase from Bacillus cereus by BIP was shown to be accompanied by the formation of a new chromophore 15,16 and analysis of the absorption, circular dichroic and NMR spectra of the protein-bound chromophore or the isolated chromophore gave strong evidence that BIP covalently binds to the enzyme and undergoes a rearrangement of the penicilloyl-enzyme intermediate, possibly via ring opening of the thiazolidine and reaction of the thiolate anion thereby formed 17.
Inactivation of class A β-lactamases by BIP can be described by successive steps: Michaelis complex formation, acylation, iodide departure and rearrangement of the penicilloyl moiety into a dihydrothiazine ring. Hydrolysis of the acyl-enzyme sometimes competes with this rearrangement (scheme I) 12,18. Iodide departure was proposed to be the rate limiting step in the process. The strong influence of the ionic strength on the ratio of product turnover versus inactivation provided the rationale for a different rearrangement pathway proposal where opening of the β-lactam ring is followed by the transient formation of an episulfonium ion intermediate (scheme I) 19. A detailed physical organic analysis of the rearrangement independently came to the same conclusion concerning the mechanism 20
Because early experiments showed that they generally had weak antimicrobial activity, 6-β-halogenopenicillanates are usually considered to be β-lactamase inhibitors. As such, they can be used to lower the minimum inhibitory concentration of various β-lactams as antibiotics 13,21,22. Few experiments, however, have been conducted to explore the inhibitory effect of 6-β-halogenopenicillanates on purified PBPs. BIP actually does inhibit the DD-peptidase activity of Actinomadura R39 (R39), but not the DD-peptidase from Streptomyces R61 23.
R39 is a low molecular weight type-4 PBP with a structure homologous to E. coli PBP4 and B. subtilis PBP4a 24. R39 is a multidomain water-soluble enzyme, loosely associated with the bacterial cell membrane. The penicillin-binding domain of R39 contains the active site responsible for the DD-peptidase activity and acylation by β-lactam antibiotics. The overall fold of the DD-peptidase domain is very similar to the fold of class A β-lactamases such as BS3 (figure 1). The active site is at the interface of two subdomains, an all α and an α/β domain, and is defined by three motifs common to all PBPs and serine β-lactamases. The main difference between PBPs and class A β-lactamases is the presence in the latter of a loop bearing an asparagine and a glutamic acid responsible for the rapid deacylation of most β-lactam antibiotics. In contrast, PBPs form long-lived acyl-enzymes with β-lactams.
In this paper, we show that kinetic experiments, mass spectrometry and the X-ray structure of the complexes between the BIP and two enzymes, BS3 and R39, give direct evidence for the rearrangement of BIP in a dihydrothiazine ring (Scheme I). To explore the inhibitory properties of BIP on the DD-peptidase activity of PBPs, minimum inhibitory concentrations of BIP on different gram negative and gram positive strains of bacteria have been determined.
The interaction of the PBPs (E) and BIP (I) can be described by model 1.
If k3 =0 and k2 k4 the rate of the formation of EI* is the rate-limiting step and the determination of the pseudo-first order rate constants ki allows the calculation of k2/K by using the following equation 25:
S is the reporter substrate S2d for the R39 and nitrocefin for BS3, respectively. The Km-values for S2d 26 and nitrocefin were determined initially by initial rate measurements. R39 can hydrolyse the thioester S2d. The activity is increased in the presence of D-alanine because deacylation is rate-determining and D-alanine intercepts the acyl enzyme. The activity towards the thioester was determined spectrophotometrically in the presence of 4,4′-dithiodipyridine, a sensitive thiol-specific reagent, that has an absorption maximum at 324 nm (ε324: 20.000M−1s−1) after reaction with two equivalents of thiol. Assays were conducted at 30°C in 10 mM sodium phosphate buffer with 100 mM NaCl pH 7.2, 100 μM S2d (Km = 0.3 mM), 100 mM D-alanine and 0.5 mM 4,4′-dithiodipyridine (Acros Organics). R39 (0.4 nM) was mixed with various concentrations of BIP (20 to 125 nM) in the assay described above; S2d hydrolysis was followed by absorbance at 324 nm using an Uvikon 860 spectrometer linked to a microcomputer through an RS232 interface. Once the time course was complete, values of pseudo-first order rate constants of inactivation, ki, were determined from total progress curve analysis and the k2/K value were determined as described above.
For BS3, nitrocefin was used as reporter substrate. BS3 (0.7 nM) was mixed with various concentrations of BIP (10 to 70 nM) in 50 mM sodium phosphate buffer pH 7.2 with 5 μM nitrocefin (Km = 11μM) at 30°C. Nitrocefin hydrolysis was followed by absorbance at 482 nm (ε482: 15.000M−1s−1) and the ki values and the k2/K value were determined as described above.
The product of BIP turnover is 2,3-dihydro-2,2-dimethyl-1,4-thiazine-3,6-dicarboxylate (Scheme I, 4), which has a maximum of absorbance at 305nm (ε305: 8150 M−1s−1) 12. In order to show the presence of 4 the hydrolysis of BIP was followed by absorbance at 305nm Therefore 10 μM BS3 was incubated with 100 μM BIP in 50 mM sodium phosphate buffer pH 7.2 at 30°C.. The same experiment was done with 1 μM R39 in the presence of 10 μM BIP in 10 mM sodium phosphate buffer with 100 mM NaCl pH 7.2. In the presence of stoichiometric amount of BIP and protein, a new absorption band can be observed at 323nm (ε323: ~10500 M−1cm−1), which can be attribute to the formation of the inactivation complex (EIi) 27. Difference spectra between the complex formed in the presence of BIP and the free enzyme were measured in the presence of 10 μM protein (R39 or BS3) in buffer (10mM sodium phosphate buffer with 100 mM NaCl pH 7.2 with R39 and 50 mM sodium phosphate buffer pH 7.2 with BS3, respectively) in the presence of different BIP concentrations (10μM, 20μM, 30μM and 100μM).
Covalent adducts were identified by mass spectrometry. 100 μM BIP was incubated with 1 μM protein in buffer (R39: 10 mM sodium phosphate buffer with 100 mM NaCl pH 7.2; BS3: 50 mM sodium phosphate buffer pH 7.2) at 25°C for 60 min (volume: 1.2 ml). The volume was decreased to 10 μl by ultra filtration using Microcon YM-10 (Millipore). Sample preparation was done by two additional ultra filtration steps with 500 μl of water with R39 and by two additional ultra filtration steps with 500 μl of 500 mM ammonium acetate and 500 μl of 25 mM ammonium acetate with BS3, respectively. The samples were concentrated to 30 μl. They were mixed with acetonitrile (50 % v/v) and trifluoroacetic acid (0.1 % v/v). Mass spectrometry analysis was then conducted using ESI-Q-TOF micro mass (Waters) operated in positive mode, with a capillary voltage of 3.0 kV, a sample cone voltage of 100 V and a temperature of 100° C, RF lens 1 was set at 100 V, and a 500 l h−1 desolvation gas flow. Sample was injected into the source for desolvation at 4 μl min−1 flow.
The geometry of the 2,2-dimethyl-2,3,-dihydro-1,4-thiazine-6-methoxycarbonyl-3S-carboxylic acid (figure 2) was fully optimized with the B3Lyp functional 28 using the double ζ basis set 6-31G(d) 29. All the calculations were performed with the Gaussian 03 program 30.
The expression and purification of the BS3 enzyme were performed as described previously 31. Crystals were grown at 20°C by hanging drop vapor diffusion with drops containing 5 μl of a protein solution (38 mg ml−1), 1 μl of ethanol 30% and 4 μl of 10% PEG 6000 in 100 mM sodium citrate buffer (pH 3.4), equilibrated against 1 ml of the latter solution at 20 °C. The BS3-BIP adduct was obtained by diffusing into the crystals increased concentrations of BIP over the course of 24 h at room temperature. A fresh solution of BIP was continuously added up to a final concentration of 68 mM.
Data were collected at room temperature at LURE (Orsay, France) on beamline DW32 at a wavelength of 0.9474 using a MarResearch Mar345 Imaging Plate. Indexing and integration were carried out using Mosflm 32, and the scaling of the intensity data was accomplished with SCALA of the CCP4 program suite 33. Refinement was carried out using REFMAC5 34, TLS 35, and Coot 36. The structure of the BS3 β-lactamase bound to the rearranged BIP (BIPr) was refined to 1.65 Å with Rcryst and Rfree values of 19.0 % and 21.1 % respectively. 92.7 % of the residues are in the most favored region of the Ramachandran plot and 6.8 % of the residues in the additionally allowed region. The ligand occupancy was refined as 1.0. Data statistics and refinement are given in Table 1.
The R39 DD-peptidase was expressed and purified as described previously 37. Crystals were grown at 20°C by hanging drop vapor diffusion. Crystals were obtained by mixing 4 μl of a 25 mg ml−1 protein solution (also containing 5 mM MgCl2 and 20 mM Tris, pH 8), 2 μl of well solution (2.0 M ammonium sulfate and 0.1 M MES, pH 6), 0.5 μl of 0.1 M CoCl2 solution, and 0.2 μl of BIP 0.1 M.
Data were collected at 100 K on an ADSC Q315r CCD detector at a wavelength of 0.9763 Å on beamline BM3OA at the European Synchrotron Radiation Facility (ESRF, Grenoble, France). X-Ray diffraction experiments were carried out under cryogenic conditions (100° K) after transferring the crystals into 100% glycerol. Intensities were indexed and integrated using Mosflm32. Data were scaled with SCALA of the CCP4 program suite 33. Refinement was carried out using REFMAC5 34, TLS 35, and Coot 36. The structure of the R39 DD-peptidase bound to BIPr was refined to 2.3 Å with Rcryst and Rfree values of 20.6 % and 26.1 % respectively. 91.6 % of the residues are in the most favored region of the Ramachandran plot and 7.7 % of the residues in the additionally allowed region. The ligand occupancy was refined as 1.0. Data statistics and refinement are summarized in Table 1.
Tests were made using microtitration plates, in 200 μl (final volume) of Müller-Hinton Broth (MHB), following the EUCAST (European Committee on Antimicrobial Susceptibility testing)/CLSI (Clinical and Laboratory Standard Institute) recommended procedure. BIP was solubilized in MHB, just before utilization. Inocula were prepared for each strain, resuspending isolated colonies from 18h cultured plates. Equivalents of 0.5 McFarland turbidity standard (approximately 1.108 CFU ml−1) were prepared in saline solution (NaCl 0.085%) and then diluted 200 fold in MBH (initial population). MIC was determined as the lowest dilution of product showing no visual turbidity.
The product of BIP turnover is 2,3-dihydro-2,2-dimethyl-1,4-thiazine-3,6-dicarboxylate (Scheme I, 4), which has a maximum of absorbance at 305nm 12. No increase of the absorbance at 305 nm was observed during an incubation of 60 min of R39 and BS3 with 10 equivalents of BIP, respectively. Difference spectra between the complex formed in the presence of BIP and the free enzymes show a maximum at 323nm which is characteristic for the formation of this complex (Scheme I, 7). A maximum absorption at 323nm was observed directly after the first addition of BIP (1/1 ratio of BIP to enzyme) indicating the rapid formation of the inactivation complex. No shift of the maximum of absorbance to lower wavelength after further additions of BIP was observed as it has been described for the Streptomyces albus G β-lactamase 12. This indicated the absence of turn-over of the inactivator. In the absence of turn-over (k3 = 0) and a rapid formation of the inactivation complex (k4 k2) the formation of the acylenzyme is the rate-limiting step 27 and the second-order rate constants of the inactivation of R39 (k2/K = 52300 ± 6400 M−1s−1) and BS3 (k2/K = 58900 ± 7900 M−1s−1) were determined. These values indicated that BIP is a good inactivator of both proteins.
The presence of 2,2-dimethyl-2,3-dihydro-1,4-thiazine-6-carbonyl-3S-carboxylate (BIPr) in the active site of both proteins has been confirmed by mass spectrometry. The mass increments between R39 (47759 ± 2 Da) and BS3 (29470 Da ± 2) and the inhibitor complexes (R39-complex: 47953 ± 2 Da and BS3-complex: 29665 ± 2) were respectively 194 and 195, consistent with the formation of the covalently attached dihydrothiazine (7).
Computations suggest that in the model compound 2,2-dimethyl-2,3,-dihydro-1,4-thiazine-6-methoxycarbonyl-3S-carboxylic acid (rearranged molecule with the methyl ester mimicking the serine bond), the dihydrothiazine group can adopt two stable conformations (figure 2). The C-up conformer (in which the carbon C2, bonded to the sulfur atom, is above the plane formed by the other atoms of the ring) has an axially oriented carboxylate group. The S-up conformer (in which the sulfur atom is above the plane formed by the other atoms of the ring) is rather flat with an equatorially associated carboxylate. The C-up conformer is 3.51 kcal mole−1 more stable. By reference to the Sup conformer, the energy barrier between them is 7.70 kcal mole−1. For both minima, a different conformation of the ester mimicking the serine bond can be trapped after rotation of 180 degrees around the C6-C12 bond. The energy stabilization is roughly 0.82 kcal.
The electron density in the BS3-BIP crystal is well defined except for that of some glutamic acids that suffered radiation damage, lysines exposed to the solvent, loop 253–257 in monomer A and loops 86–100, 200–209, and 253–257 in monomer B. The rms deviation between the monomers is 0.25 Å. The overall fold of the BS3-BIP complex is similar to the previously reported BS3 structure (Apo-enzyme, PDB: 1I2S 38) with an r.m.s. deviation between the apo and complex structure of 0.25 Å for the Cα atoms of BS3 monomer A.
The asymmetric unit of BS3 crystals contains two protein molecules. The electron density map calculated in the absence of ligand provides a very clear density into which BIPr is covalently linked to the enzyme active serine (Figure 3a). The density in each monomer shows a covalent adduct resulting from the acylation of the active site serine by BIP, and followed by the intramolecular rearrangement of the latter. A chloride ion lies between Ser130Oγ, Lys234Nζ, Thr235Oγ, and Ser70Oγ. The density cannot account for the iodide ion released during catalysis and the iodide remains unobserved. The β-lactam-derived ester carbonyl of the covalent adduct hydrogen bonds to Ser70 and Ala237 nitrogens that define the oxyanion hole. BIPr adopts the C-up conformation with the sulfur oriented on the same side that the ester carbonyl.
One BIPr carboxylate oxygen is hydrogen bonded to Asn132Nδ and Asn104Nδ and its partner is hydrogen bonded to Asn170Nδ. Asn170 normally interacts with a nucleophilic water molecule conserved in the structures of class A β-lactamases. The interactions between BS3 and BIPr induce a slight displacement of the Ω-loop that eventually reduces the space left for the deacylating water molecule. In monomer A, the total absence of electron density clearly precludes the presence of a water molecule. In monomer B, a small amount of density suggests the presence of a water molecule between Glu166, Asn170 and BIPr. Comparison of the position of this water molecule with its conserved position in other class A β-lactamases shows that the water molecule in BS3 monomer B is shifted some 0.4 Å, which might reflect a deficient positioning of the deacylation machinery of BS3 towards BIPr.
The electron density in the R39-BIP crystal is well defined except the N-terminal residue and some arginine, lysine and glutamic acid side chains pointing into the solvent. The overall fold of the R39-BIP complex is similar to the previously reported R39 structure with an r.m.s. deviation between the apo and complex structures of 0.43 Å for the 466 Cα atoms of R39 monomer A (PDB: 1W79) 39.
The asymmetric unit of R39 crystals contains four protein molecules. The environment is identical for monomers A and D and for monomers B and C but different for both pairs. Although not preventing BIP from acylating the enzyme (see below), the active sites of monomers A and D in the apo-enzyme structure are obstructed by the loop 175–178 of a symmetric molecule, whereas monomers B and C show a freely accessible active site.
In the R39-BIP complex, the electron density map calculated in the absence of ligand shows clear density for the rearranged BIP (BIPr), which is covalently linked to the enzyme active serine in all four monomers (figure 3b and 3c). As expected, the carbonyl oxygen lies in the oxyanion hole. In monomers A and D, BIPr displays the C-up conformation and the dihydrothiazine ring is oriented with its sulfur on the side of the acyl-serine carbonyl (figure 3b). The BIPr conformation is, therefore, similar to that in the BS3 β-lactamase. The dihydrothiazine carboxylate is oriented towards the phenyl group of Tyr147 but doesn’t make any hydrogen bond with the protein. The carboxylate O8 distance to Asn300Nδ (from the SxN motif) is 3.9 Å in monomer A and 4.5 Å in monomer D. The sulfur atom of BIPr is at 3.5 Å from the backbone oxygen of Thr413. In monomers B and C, BIPr displays the S-up conformation and the dihydrothiazine ring is oriented with its sulfur on the side opposite to acyl-serine carbonyl (figure 3c). The dihydrothiazine carboxylate points out from the active site without making interactions with the surrounding amino-acids. The sulfur atom of BIPr is at 3.6 Å from the backbone oxygen of Ser298 and 3.5 Å (monomer B; 3.7 Å in monomer C) from the Asn300Oδ. In all four monomers, a water molecule is found instead of the sulfate ion observed in the apo R39 structure between residues Ser298, Lys410, Thr411 and Thr413. As in BS3, the iodide atom released during catalysis remains unobserved in R39.
Superposition of R39 monomer B onto monomer A shows that BIPr cannot adopt in monomer A the S-up conformation found in monomer B because of steric hindrance between the BIPr carboxylate and the side chain methyl group of Ala175, which belongs to a symmetry loop slightly inserted in the active site of monomer A. As well, superposition of the structure of BS3-BIPr with monomer B of R39-BIPr shows that BIPr cannot adopt in BS3 the S-up conformation found in R39 monomer B because of steric hindrance between the two BIPr methyl groups and the side chain of Tyr105. Conversely, BIPr could adopt in R39 monomer B the C-up conformation found in R39 monomer A or in BS3, indicating that BIPr conformation in R39 monomer B (S-up) is the preferred one in an unhindered active site.
BIP shows antimicrobial activity when tested on a broad range of bacterial strains (table 2), indicating that one or more essential PBPs is probably inhibited. No significant differences in MIC values could be detected between Gram positive and Gram negative bacteria, suggesting that the external membrane of the latter organisms does not constitute a barrier for BIP to reach its target. Sensitive Gram positive Staphylococcus aureus are inhibited at low BIP concentration (< 2 μg/ml). The relatively high concentrations of BIP are needed to inhibit bacterial growth in the other tested strains, however, may be due to PBP-unrelated factors needed for triggering bacterial killing. It is important to note that BIP does not act as a good resistance inducer in the S. aureus PL1 strain. This strain produces, in the presence of β-lactams, inducible low affinity PBP2a, but remains sensitive to BIP mostly because PBP2a is not expressed in the presence of BIP. On the contrary, S. aureus ATCC 43300, which constitutively expresses low affinity PBP2a shows a high MIC value. The others tested strains constitutively producing low affinity PBPs (Enterococcus faecalis, E. faecium, E. hirae) show similar high MIC values. This result is in good agreement with the low inhibitory activity measured against purified low-affinity S. aureus PBP2a and E. faecium PBP5 (data not shown).
Kinetic results have previously evidenced that 6-β-iodopenicillanates and 6-β-bromopenicillanates are good irreversible inhibitors of class A β-lactamases. Our kinetic, mass spectrometry and crystallographic results assess the nature of the final product formed from BIP upon its reaction with the β-lactamase BS3. The BIP-BS3 structure provides a clear view of the acyl-enzyme formed between the enzyme and the inhibitor. It shows a dihydrothiazine ring, which is likely formed from the rearrangement of Scheme I, in its energetically most favorable conformation with the sulfur-linked carbon lying over the plane formed by the other ring atoms, and the ring oriented with the sulfur on the side opposite to the acyl-serine carbonyl.
The formation of the Michaelis complex and acylation of the enzyme active site serine that have been described for penam derivatives of β-lactams should be similar for BIP 40–42. The presence of the iodine atom induces a positive charge on C6 and, after hydrolysis of the β-lactam amide bond, a 90° rotation around the C6-C5 bond, enforced by steric interactions across this bond, would allow the sulfur to get close to the positive charge on C6. Formation of the episulfonium ion S-C5-C6 would then be concerted with the departure of iodide. The labile S-C5 bond breaks and C5 becomes sp2. The rearrangement involves a large movement of the carboxylate group from its likely position in the preacylation state, where it probably hydrogen bonds to Thr235, to its observed position in the crystal structure, close to the Ω-loop where it interacts with Asn170 and Asn132. These interactions lead to the observed axial C-up conformation of the rearranged BIP.
The interaction between the BIP-dihydrothiazine carboxylate and Asn170 induces a slight shift in the Ω-loop position that shortens the distance between Glu166 and the acyl-enzyme ester bond. Deacylation is impaired because the space left to the water molecule necessary for deacylation is reduced, which results in the absence of water in monomer A and, in monomer B, a shift of the water molecule compared to its position in the apo-structure of BS3 or in other class A β-lactamases.
The class A β-lactamase inhibitory property of the BIP results from its rearrangement after enzyme acylation and the interaction of the dihydrothiazine carboxylate with both Asn132 and Asn170. These residues are conserved as essential amino-acids for the β-lactam hydrolysis activity of class A β-lactamases. The mechanism should therefore be common to all class A β-lactamases. The chemical stability of the vinylogous amide moiety in 6 is another contributor to the inhibition 43. Moreover, the few interactions of BIP with residues not critical for catalysis reduce the possibilities of resistance development.
The most interesting feature of the BS3 structure may be the above-mentioned positioning of the inhibitor carboxylate. This represents a new mode of interaction of a carboxylate with the class A β-lactamase active site, a motif that could be incorporated into future inhibitors.
The mechanism of inhibition of class A β-lactamases by BIP is different from the mechanism used by penicillin sulfones such as tazobactam. In the reaction pathway of tazobactam, Ser70 acylation is followed by thiazolidine ring opening between the sulfur atom and carbon C5. Attack on the thus generated reactive imine by Ser130 results in a vinyl carboxylic acid that is covalently attached to Ser130Oγ and thereby inactivates the enzyme 44,45.
Improvement of BIP inhibitory efficiency against class A β-lactamases may be achieved, in principle, by substitution on C6-α. In cefoxitin or imipenem, the C6-α substituent is essential for inhibition in preventing a water molecule from approaching the acyl-enzyme ester bond for deacylation. But a C6-α substituent may clash with the side chain of Asn132 and reduce the affinity of the enzyme for the ligand. A C6-α substituent may thus be detrimental to BIP inhibition. Attempts to improve 6-bromopenicillanic acid by incorporation of such substituents proved unsuccessful 46. For the same reason, the low efficiency of 6-α-halogenopenicillanates as inhibitors of class A β-lactamases could result from a poorer affinity due to steric hindrance between the halogen atom and the side chain of Asn132. As an exception, 6-α-halogenopenicillanates could be inhibitors of the class A β-lactamase of Mycobacterium tuberculosis, which bears an Asn132Gly substitution. As well, the Enterobacter cloacae NMCA class A β-lactamase, a carbapenemase with enlarged cavity and additional space near Asn132 47, could perhaps accommodate 6-α-halogenopenicillanates or 6-β-halogenopenicillanates substituted on position α. The rearrangement of 6-α-halogenopenicillanates may, however, also be inhibited by intramolecular steric interactions 20. Another possibility, not yet explored, would be substitution at the 2-methyl groups where further interaction with the protein may be induced; such functionalization has proved effective with penicillin sulfones such as tazobactam and other 2′β-substituted penam sulfones 48.
R39 is a type-4 PBP not essential to bacterial survival in laboratory conditions but essential PBPs may also be sensitive to BIP, as shown by the MIC tests on a wide range of bacteria (Table 1). The structural results (Figures 3b, 3c) suggest that PBPs react with BIP in the same way as β-lactamases, yielding the same rearranged inhibitor. The structure of R39 in complex with BIP reveals two conformations of the dihydrothiazine. In monomers B and C, the active site is free of steric hindrance by the symmetric enzyme molecule, and the conformation of the dihydrothiazine is likely the one that should be observed in solution. The more stable dihydrothiazine conformation is observed in monomers A and D, where a symmetry loop prevents BIP from adopting the conformation found in the free active site. The difference between both conformations not only involves the puckering of the dihydrothiazine ring but also its rotation around the C6-C12 bond with regard to the enzyme structure. The S-up conformer of the dihydrothiazine ring that is observed in monomers B and C may be stabilized by the interaction between the sulfur atom and the backbone oxygen of Ser298 and the side chain oxygen of Asn300 49.
Observation of the calculated less stable conformation of the BIP dihydrothiazine ring in the unhindered active site of R39 monomer B resembles the situation of the thiazolidine ring of penicillins, which can adopt an axially oriented carboxylate group or an equatorially disposed conformation. Quantum chemical calculations have shown that the axial conformation is the most stable, and the X-ray structures available in the Cambridge Structural Database reveal that the majority of the penicillins are in the axial orientation. Crystal structures of acyl-enzymes derived from penicillins, however, present the thiazolidine in its equatorial conformation. Those crystallographic results suggest that the equatorial conformer of the BIP thiazolidine ring is the biologically active conformer. The results show that PBPs are inhibited by BIP. Its efficiency as an inhibitor could probably be improved, however, as in the case of class A β-lactamases, by substitution on C6-α or substitution at the 2-methyl groups. The C3′ substituent of ceftobiprole is responsible for the good antibacterial activity of this third generation cephalosporin against methicillin resistant S. aureus 50, and one could take advantage of the information obtained with the C3′ substituent of cephalosporins to substitute one of the 2-methyl groups. Such information is available with the crystal structure of the complex between ceftidoren and Streptococcus pneumoniae PBP2x 51. Highly resistant PBPs appear to be only weakly sensitive to BIP, but BIP could represent a lead compound to more effective inactivators of these PBPs
Structural study of the BS3-BIP acyl-enzyme has lead to direct proof of the unique rearrangement involved in inhibition of class A β-lactamases by BIP (Scheme 1) and the nature of the interactions of the rearranged inhibitor with conserved active site residues have been elucidated. The position of the inhibitor carboxylate and its interactions with Asn132, Asn104 and Asn170 are especially noteworthy.
PBPs were shown to be sensitive to BIP to various degrees. The crystal structure of the R39-BIP complex is the first of a PBP acylated by a 6-halogenopenicillanate. The detailed observation of the two possible conformations of the dihydrothiazine ring of the rearranged BIP linked to the enzyme active serine provides a structural basis for the development of substituted 6-halogenopenicillanate as inhibitors of PBPs. BIP-based inhibitors of PBPs not only would escape the hydrolysing action of β-lactamases but are potential inhibitors of β-lactamases themselves.
We thank the staff of beamline BM30a at ESRF for assistance in X-ray data collection. This work was supported in part by the European Commission Sixth Framework Program grants LSMH-CT-EUR-INTAFAR 2004-512138, by the Belgian Program on Interuniversity Poles of Attraction initiated by the Belgian State, Prime Minister’s Office, Science Policy programming (IAP no. P6/19), by the Actions de Recherche Concertées (grant 03/08-297), by the Fonds de la Recherche Scientifique (IISN 4.4505.00, FRFC 2.4.528.01.F, FRFC 9.4.538.03.F) and the University of Liège (Fonds spéciaux, Crédit classique, 1999). G.D. is research associate of the FRS-FNRS. He thanks the FRS-FNRS for the financial support of the high performance computing systems installed in Liège and Louvain-la-Neuve. Financial support to R.F.P from the U.S. National Institutes of Health is acknowledged.