|Home | About | Journals | Submit | Contact Us | Français|
Asymmetric diaroyl phosphates (ArCOOPO2−OCOAr′, where Ar = Ph, Ar′ = 4-biphenyl, 2-benzothiophenyl and 2-benzofuranyl), have been prepared, evaluated as serine (classes A, C, D) β-lactamase inhibitors, and compared with respect to the latter with their symmetric “parents”, where Ar = Ar′. The asymmetric compounds, in general, were found to react with the β-lactamases in two modes, corresponding to different orientations with respect to the active site, whereby either of the two aroyl groups may acylate the enzyme to form two different inert acyl-enzymes, E-COAr and E-COAr′. In all cases, the asymmetric compounds, in one orientation, react more rapidly with the enzymes tested than one symmetrical “parent” but not both. From comparisons of activation free energy differences, it was found that the changes in free energy on changing from one aryl group to another, in either the acyl group or the leaving group, were not additive, i.e. that the effect of changing one aroyl group to another depended on the leaving group and vice versa. Thus, intramolecular cooperativity between the aroyl groups must exist, arising either from direct interaction between them or from protein-mediated interaction, or from a combination of both. Such cooperativity brings fresh opportunities and challenges to the search for novel ß-lactamase inhibitors.
Aroyl phosphates, 1, have proven to be very effective inhibitors of serine ß-lactamases (1–4). The phosphate leaving group appears to interact with polar resides in a way as to enhance active site acylation by both specific amidoacyl groups and non-classical aroyl groups (2–4, 5, 6). Acylation by the latter leads to hydrolytically refractive acyl-enzymes and thus to significant inhibition (Scheme 1). A series of diaroyl phosphates 2 have proven to be particularly effective inhibitors (3,4). Structure-activity studies have shown that hydrophobic substituents entrance inhibition, largely through enhancement of acylation
rates, while electron-donating substituents enhance inhibition by depression of deacylation rates (3).
It is obvious that the two aroyl groups of 2 must interact with the enzyme active site differently and thus contribute differently to the inhibitory activity of 2. In order to understand the way in which the two aroyl groups separately contribute, asymmetric diaroyl phosphates 3 were needed. Such compounds would also allow more versatile modulation of pharmacological properties for any practical application of these compounds. ß-Lactamase inhibitors are of value in protecting ß-lactam antibiotics from ß-lactamases and thereby extending the clinically useful lifetime of the latter molecules (7).
The functional asymmetry of 3 means that two modes of reaction are possible, acylation by ArCO with an Ar′COOPO3= leaving group and vice versa (Scheme 2). In general, the rates of the two possible reactions will be different and one mode of reaction will lead to more effective inhibition i.e. to a higher proportion of that acyl-enzyme in the steady state. Another issue of interest stemming from the reactivity of 3, is that of additivity. This can be seen in the comparison of the reactivity of 3 with that of the symmetrical “parents” 2 and 4 (Scheme 3). In Scheme 3, 3a and 3b, which obviously are representations of the same molecule, correspond to the alternative orientations of 3 bound at the active site. The issue then is whether δΔG‡ (the change in activation free energy of enzyme acylation) between 2 and 3a (δΔG1‡, the effect of changing the leaving group form Ar to Ar′ with a common acyl group) is equal to that between 3b and 4, δΔG4‡? Similarly, is δΔG2‡ (the effect of changing the acylating group from Ar to Ar′ with a common leaving group) equal to δΔG3‡? If not, there must be an intramolecular cooperativity between the acylating group and the leaving group, which could, in principle, be exploited in inhibitor design.
Intramolecular cooperativity is, of course, well known in the reactions of a variety of enzymes with their substrates and inhibitors. The phenomenon is probably best known in proteinases where cooperative interactions between Sn and S−n residues via their interaction with the Pn and P−n sites, respectively, have been clearly demonstrated (8–11). Similar observations have been made with glycohydrolases (12) and nucleases (13). Cooperative interactions between aryl substituents have recently been observed in noncovalent phosphonate inhibitors of serine ß-lactamases (14). In the present paper, we describe the synthesis of asymmetric diaroyl phosphates 3, an analysis of their reactions with typical class A, C and D serine ß-lactamases in order to determine the dominant mode of reaction (i.e. is ki or ki′ of Scheme 2 greater?), and an analysis of the degree of cooperativity. The diaroyl phosphates evaluated were 5–12.
The purified Enterobacter cloacae P99 and Escherichia coli W3310 TEM-2 ß-lactamases were purchased from the Centre for Applied Microbiology and Research (Porton Down, Wiltshire, UK) and used as received. The OXA-1 ß-lactamase was a generous gift of Dr. Michiyoshi Nukaga of Jyosai University, Japan. Cephalothin was a gift of Eli Lilly and Co. Benzylpenicillin was purchased from Sigma-Aldrich.
The preparation of the symmetrical diacyl phosphates 5,6, 8, 10 and 12 was as described in the preceding paper (4). The asymmetric compounds 7, 9, and 11 were prepared in the same way as 6, 8 and 10 (4), i.e. generated in the same reaction mixtures, and separated from the symmetrical compounds and purified by hplc (Macherey-Nagel SS 250/0.5 in/10-nucleosil 300–7 C18 reverse phase column; 7: retention time 19.8 min from a 0–100% MeOH/ H2O gradient with flow rate 3ml/min; 9: retention time 24.4 min, 20–80% MeOH/H2O gradient, 3ml/min; 11: retention time 14.7 min, 20–80% MeOH/H2O gradient, 3ml/min.) and characterized as noted below.
1H NMR (DMSO-d6) δ 7.28 (t, J = 6.6 Hz, 1H), 7.32 (t, J = 6.0 Hz, 1H), 7.35 (t, J = 8.1 Hz, 2H), 7.52 (t, J = 9.0 Hz, 2H), 7.55 (d, J = 8.4 Hz, 2H), 7.61 (d, J = 6.9 Hz, 2H), 7.89 (d, J = 9 Hz, 2H), 7.95 (d, J = 7.5 Hz, 2H). 31P NMR (DMSO-d6) δ −21.12. FTIR (KBr, cm−1) 1725.1. ES(−)MS m/z 381.20
IH NMR (DMSO-d6) δ 7.36 (t, J = 7.6 Hz, 1H), 7.43 (t, J = 7.2 Hz, 2H), 7.44 (t, J = 7.8 Hz, 1H), 7.60 (t, J = 7.8 Hz, 1H), 7.86 (d, J = 6.6 Hz, 1H), 7.88 (d, J = 7.2 Hz, 1H), 7.97 (d, J = 7.8 Hz, 2H), 8.14 (s, 1H). 31P NMR (DMSO-d6) δ −18.37. FTIR (KBr, cm−1) 1719.2. ES(−)MS m/z 361.07.
1H NMR (DMSO-d6) δ 7.26 (t, J = 7.8 Hz, 2H), 7.41 (d, J = 7.8 Hz, 1H), 7.44 (t, J = 6.9 Hz, 1H), 7.50 (t, J = 8.4 Hz, 1H), 7.59 (t, J = 7.5 Hz, 1H), 7.69 (d, J = 7.8 Hz, 1H), 7.73 (s, 1H), 7.96 (d, J = 7.2 Hz, 2H). 31P NMR (DMSO-d6) δ −18.4. FTIR (KBr, cm−1) 1731.0. ES (−) MS m/z 345.02
All kinetics experiments were performed at 25°C in a buffer at pH 7.5 containing 20mM MOPS. When the OXA-1 β-lactamase was studied, 50mM NaHCO3 was included in the buffer. Stock solutions of the diacyl phosphates in DMSO were prepared as described in the preceding paper (4), except for 8 and 9, which were unstable in this solvent, possibly from oxidation. Stock solutions of 8 and 9 were therefore prepared in DMF, which at concentrations up to 4 % v/v, did not affect the activity of the enzymes. Kinetics data were fitted to Scheme 4 in all cases. The slow turnover rate constant k2 could be determined spectrophotometrically at 300 nm from steady state experiments carried out at saturating values of [I] (4). Effective steady state inhibition constants, Ki (= k2/ki), or, equivalently, effective Km values if I is considered a substrate, were determined either directly spectrophotometrically at 300 nm from a Henri-Michaelis-Menten analysis or by competitive inhibition of hydrolysis of a suitable substrate (4). Experimental details are expanded in the Supplementary Information (Table S1).
Mass spectra of OXA-1 and P99 β-lactamase complexes with 7 were determined as previously described for 5 (3). Thus. a reaction mixture containing the OXA-1 β-lactamase (25 μM) and 7 (520 μM) was quenched with trichloroacetic acid after 15 min. A similar mixture of the P99 enzyme (15.7 μM) and 7 (4 mM) was treated in the same way after 5 min. The washed and dried precipitates were subjected to electrospray mass spectroscopy at the Mass Spectroscopy Laboratory, School of Chemical Sciences, University of Illinois.
The synthesis of asymmetric diaroyl phosphates 7, 9, and 11 could not be achieved by the method of Chantrenne (15) that we had previously employed to prepare most of the symmetrical compounds (3,4). All attempts by this method to prepare asymmetric compounds appeared to lead only to symmetrical compounds. Synthesis of the required asymmetric diaroyl phosphates was, however, achieved by a one-pot two-step acylation process involving a HATU-mediated first step and a second acylation by aroyl chloride (Scheme 5). This process yielded a mixture of three products, the symmetrical diaroyl phosphates 2 and 4, and the asymmetric molecule 3 in comparable amounts, as shown clearly in 31P NMR spectra. Equilibration of the three compounds may have occurred via triaroyl phosphates. The three could be separated and isolated by means of hplc, and identified by NMR and mass spectra. The method employed here was also found to be an efficient method of synthesis of symmetric diaroyl phosphates where Ar = Ar′.
The previously described 5, 6, 8, 10 and 12 react with serine ß-lactamases, as inhibitory substrates, as described (3, 4). Rapid acylation of the enzyme active site is followed by slow deacylation (Scheme 1). The new asymmetric compounds 7, 9 and 11 appear to behave qualitatively in the same manner with typical examples of class A (TEM-2), class C (Enterobacter cloacae P99) and class D (OXA-1) enzymes. Apparent values of k2 and Ki (=k2/ki) were determined as described in the Experimental Section and the preceding paper (4). Figure 1A, for example, shows data for the determination of k2 for the P99 ß-lactamase on reaction with 7. The essentially linear initial rates, independent of the concentration of 7, indicate Kiapp << 10 uM and directly yield the k2app value. These are apparent values because of the uncertainty at that stage as to the mode of inhibition (Scheme 2). Figure 1B shows data for the experimental determination of Kiapp for the OXA-1 ß-lactamase on reaction with 7. A final example, Figure 1C, shows the data employed for the direct determination of k2app and Kiapp for reaction of the TEM-2 ß-lactamase with 9, by non-linear least square fitting of the Henri-Michaelis-Menten equation.
Steady-state kinetics data from the experiments described above, for slow turnover of 7, 9 and 11 by the various enzymes, is shown in Table 1. Also shown, for comparison, are previously determined values of these parameters for 5, 6, 8, 10 and 12 (3,4). Values for 8 and 10 for the OXA-1 enzyme were also determined. The apparent values for 7, 9 and 11 should be composites of the two possible modes of reaction (Scheme 2). This proposition was supported by mass spectrometry. For example, incubation of the P99 β-lactamase with 7 yielded a protein with mass peaks at 39,189, 39,295, and 39,376 amu. These correspond to the free enzyme (M), benzoyl-E (expected, M+104), and 4-phenylbenzoyl-E (expected, M+180), respectively. Similarly, with the OXA-1 enzyme, peaks at 28,130, 28,234, and 28,311 amu were observed. It thus seems likely that the two reaction modes of Scheme 2 will, in general, be observed in the reactions of 7, 9 and 11 with β-lactamases.
The quantitative kinetic parameters of the two modes of reaction were deconvoluted by means of Scheme 6 and equations 1 and 2 which are derived from it. This is possible when it is realized that k2 for 7 will be the same as that for 5 (already determined - Table 1) and k2′ will have the value of k2 for 6 (Table 1), if it is assumed, reasonably, that the asymmetric compound 7 will produce the same acyl-enzymes as from the symmetrical compounds 5 and 6. Simultaneous solution of equations 1 and 2 with the data of Table 1 yielded the values of Table 2.
where Ki = k2/ki and Ki′ = k2′/ki′.
Immediately noticeable from Table 1 for 7 is the fact that, within experimental uncertainty, the k2 values for 6 and 7 are equal for all three enzymes while that for 5 is smaller. Thus, in terms of Scheme 6, k2app ≈ k2′ and k2 < k2′. This will be possible (eq 1) if Ki >> Ki′ and, therefore, ki′ >> ki. It also follows from Scheme 6 that the steady state ratio of acyl enzymes [EI′]/[EI] is equal to Ki/Ki′, which is thus much greater than unity. Thus, the dominant acyl-enzyme generated by reaction of all three enzymes with 7 must be the 4-phenylbenzoyl species rather than the benzoyl, i.e. for all three enzymes, the 4-phenylbenzoyl group prefers the acylation site and benzoylphosphate the leaving group site, rather than vice versa. Within experimental limits, therefore, k2 and ki for 7 are indeterminable and k2′ and ki′ (Table 2) are equal to the apparent values (Table 1).
For the other asymmetric compounds 9 and 11, the k2app values lie between those of k2 and k2′ (the relevant values for the two cognate symmetrical compounds, 5 and 8 for 9, and 5 and 10 for 11) (Table 1). In these cases, individual k2, ki, k2′ and ki′ values could be calculated from eqs 1 and 2 (see Table 2). It can be seen from Table 2 that, in all cases but one, that of the OXA-1 enzyme with 9, the asymmetric compounds react more rapidly where the aroyl group acylates the enzyme rather than benzoyl (ki′ > ki). In most cases, the asymmetric compound, in one orientation, reacts more slowly than the dibenzoyl compound 5, and in one case, that of the OXA-1 enzyme with 11, both do. It is noticeable also that in no case, taking into account experimental uncertainties (notice the case of the TEM β-lactamase with 11) is an asymmetric compound superior to both of the related symmetric compounds, both with respect to ki and to k2. Although, in principle, there is no reason why this should necessarily be so, it does seem to be true with these particular enzymes and diacyl phosphates. There is, however, quite noticeable non-additivity in most cases between the aroyl groups as one is changed in the presence of another, i.e. certainly some degree of cooperativity, positive or negative, between the groups is evident. This is best seen from the changes in free energy of activation of enzyme acylation, δΔG‡, on replacement of one aroyl group with another (Scheme 3), calculated from equation 3, where ky/kx would contain the relevant ki and ki′ values for the various compounds and reactions. These calculations lead to the free energy diagrams of Figure 2.
It is clear from the diagrams of Figure 2 that the energetic effect of changing a benzoyl group to another aroyl group in the acylation site depends on the nature of the leaving group and vice versa, i.e. the contributions of each type of substitution to the activation free energy are not additive and therefore, in general, there must be some degree of cooperativity between the two aroyl groups in the acylation transition state. In most cases, the change in activation energy from replacement of benzoyl by another aroyl group in the acylation site leads to faster acylation when the leaving group is aroyl phosphate (A, B, C, F, H, I) rather than benzoyl phosphate, but there is one counter-example (E); two cases are uncertain (D, G). The converse is also, of course, true, i.e. the change from benzoyl phosphate to aroyl phosphate in the leaving group site usually leads to faster acylation when the acylating group is aroyl rather than benzoyl (with the same counter-example, E). In most cases, therefore, there is a greater positive cooperativity between the larger aroyl groups than between a benzoyl group and another aroyl group, irrespective of whether the single aroyl group is in the acylating or leaving group site. The differences are most dramatic, perhaps, in A and F. In F, for example, replacement of benzoyl by benzofurancarbonyl leads to a decrease in activation energy of 1.77 kcal/mole if the leaving group is benzoyl phosphate but a decrease of 3.0 kcal/mole if the leaving group is benzofurancarbonyl phosphate. On the other hand, a change in leaving group from benzoyl phosphate to benzofurancarbonyl phosphate leads to an increase in activation energy of 0.84 kcal/mole when the acyl group is benzoyl but a decrease of 0.38 kcal/mole when the acyl group is benzofurancarbonyl.
There is a general qualitative similarity between the various diagrams of Figure 2, dictated by the observation, noted above, that in most cases, the most reactive species is the diaroyl phosphate, although, even in the comparison of benzothiophenecarbonyl (B,E,H) with benzofurancarbonyl (C,F,I) which have very similar molecular shapes, there are considerable differences in the enzymes' quantitative responses. These must arise from differences in very specific interactions, dictated, for example, by the differences in dipole and quadrupole moments between these heterocycles (16). There are clear differences, also, in how the different enzymes interact with a particular aroyl group.
The molecular mechanism(s) of the cooperativity effects are not known at present. They could arise from direct interaction between the aroyl groups when at the active site of the OXA-1 enzyme, for example, where molecular modeling suggests that the two aroyl groups may be in close proximity to each other in the acylation transition state (3). In other cases, where the aroyl groups may be further apart in the acylation complex, such as with the P99 and TEM- 2 enzymes (4), the cooperative effects may arise from modulation of protein structure by the substituents. A combination of the two is the likely general scenario.
We have showed that it is possible to prepare and employ asymmetric diaroyl phosphates as β-lactamase inhibitors. These have reactivities with typical β-lactamases not linearly predictable from their symmetric “parents”. A combination of this fact and the likely variability in physical properties of those asymmetric compounds, may allow diaroyl phosphates to be even more versatile β-lactamase inhibitors than previously indicated.
This research was supported by National Institutes of Health, Grant R01 AI-17986