D-Amino acids in peptide linkage are a noteworthy exception from the universal homochirality of proteins and peptides.1 The first animal peptide that was found to contain a D-amino acid as the second residue was dermorphin, isolated from the skin of a South American tree frog.2 This peptide binds with high affinity to μ-opiate receptors. The D-residue in this and related amphibian opioid peptides is essential for the biological function, whereas the all-L-isomers lack activity.
Further studies led to the discovery of additional diastereomeric peptides from numerous sources,3 which include two components, a natriuretic peptide and a β-defensin, from the venom of the male platypus, a primitive mammal.4 All of these peptides contain a single D-amino acid substitution in a well-defined position, which is the second residue of the mature product in all vertebrate peptides known to date. Whenever the structure of the precursor polypeptides was deduced from cloned cDNAs, a codon for the corresponding L-amino acid was present in the mRNA at the position where a D-residue occurs in the mature form. This fact shows that the D-residue is enzymatically introduced during a post-translational event.5
Fire-bellied toads (Bombina variegata, Bombina bombina as well as Bombina orientalis) contain in their skin secretions bombinins H, peptides with antibacterial and hemolytic properties,6 some of which contain a D-amino acid substitution; indeed, both isomers coexist in the skin glands. The D-amino acid at the second position gives rise to subtle changes in biophysical properties and folding propensity.7 More recently, the D-amino acid was observed to affect both the target microbial cell selectivity and membrane-perturbing activity of bombinin H.8 The peptidyl-aminoacyl-L/D-isomerase (henceforth briefly referred to as isomerase), which catalyses this post-translational stereo-inversion, has been purified and characterized.9 Surprisingly, this protein, a 52 kDa glycoprotein, is apparently excised from a polyprotein containing several isomerase domains separated by short individual spacer sequences. Genes coding for related proteins are present in several vertebrate species.
In particular, the amino-terminal domain H of the human IgG-Fcγ binding protein10 is related to the frog skin enzyme, and thus, could have isomerase activity. The similarity is highest in a central region, which contains two conserved histidine residues as well as a pair of cysteines. Further studies have shown that the Bombina isomerase is inactivated by diethylpyrocarbonate, which indicates that one or two histidine residues indeed are part of the active site. The pH dependence of the reaction rate, which shows a steep decline below pH 5.5, supports this notion. If the isomerization reaction, which proceeds in both directions, is carried out in tritiated water, radioactivity is incorporated into the second residue, which is an Ile in bombinin H. This demonstrates that the change of chirality proceeds through a deprotonation/protonation mechanism at the α-carbon. Such a mechanism is also operative in other isomerases, for example, an epimerase from spider venom,11 which is structurally unrelated to the Bombina isomerase, and an isomerase from the venom of male platypus.12 The latter could also involve at least one His as a catalytic base;13 its sequence, however, is not known at present. The enzymes from hylid frogs, conus snails, and crustaceans as well as the isomerase activity, which is likely present in mouse heart, are not yet studied to sufficient detail.14
Here, we have extended these studies in order to obtain a closer insight into the reaction mechanism. For this purpose, we performed the reaction catalysed by the frog isomerase in 2H2O. A comparison of the reaction rate with the rates of deuterium incorporation revealed that the sole product is deuterated during early reaction stages (Figure S1 in the Supporting Information). These observations are consistent with a classical two-hydrogen-acceptor (two-base) mechanism (Scheme 1).15 In D2O, the reaction was significantly retarded (by about three times when compared to water). This effect suggests that proton delivery to the α2-carbon also significantly contributes to the reaction rate. Moreover, the symmetry of the reaction is shifted at the cost of the L→D reaction as indicated by the equilibrium concentrations. This fact indicates that the action of the bases is differentially impaired by the isotope effect.
Next, we synthesized two specifically designed planar substrate analogues and tested their effect on the isomerization reaction catalysed by the frog isomerase. Peptides Ile-Δ2, 3-Ala-Gly-Pro-Val-Leu-amide (IΔA) and Ile-Δ2, 3-Phe-Gly-Pro-Val-Leu-amide (IΔF) both contain at the second position an α,β-unsaturated amino acid with a planar sp2 hybridised α-carbon, which eventually could mimic potential enolate anion intermediates (Scheme 2) or the planar transition states in a SN2-type mechanism (concerted deprotonation/protonation). The corresponding saturated peptides, IA and IF, are good substrates of the isomerase, yet with vastly diverse catalytic efficiencies ((kcat/KM)IF/(kcat/KM)IA≈50). Compounds IΔA and IΔF exerted a strong inhibitory effect independently of the direction of the enzymatic reaction (Figure 1), that is, either L→D or vice versa (Figure S2 in the Supporting Information).
Surprisingly, the experimental data indicated that the inhibition was not competitive; high substrate concentrations did not restore enzymatic activity as it is the characteristics of competitive inhibition (a Michaelis–Menten analysis of the data in Figure 1 is shown in Figure S3 in the Supporting Information). Furthermore, no evidence of reactivation of partly inhibited enzyme was observed after 100-fold dilution into buffer. Conversely, their action was virtually irreversible, and therefore, probably indicative of tight, yet noncovalent binding of the inhibitors.16 The determination of the release rates (off-rate) of the inhibitors from the enzyme was, however, hampered by the rapid loss of the enzymatic activity in a diluted solution. Therefore, we cannot fully exclude the possibility that the Δ2, 3-peptides could act as suicide substrates. Such a covalent enzyme inactivation could possibly proceed by a Michael-type addition of a nucleophilic group in the enzyme to the α,β-unsaturated carbonyl.17 However, a considerably larger enzyme preparation will be needed in order for the enzyme–inhibitor complex to be directly detected and its nature delineated.
In view of this binding behaviour, the dissociation constants (KD) were derived from simple titration experiments assuming that the enzyme–inhibitor complex is catalytically inactive.18 For both compounds, within the experimental error, the KD was found to be 0.94(±0.1)×10−6 M (Figure 2). It is interesting that peptides IΔA and IΔF are almost equally potent isomerase inhibitors, even though the saturated substrate peptides, IA and IF, differ strongly in their catalytic efficiencies. Given that the similar hydrophobicities of the Z/E isomers in IΔF account for similar active site binding, this fact suggests that the side chain substituent at the second position contributes only little to the transition state binding.
The proposed deprotonation/protonation mechanism necessarily implies the abstraction of a weak acidic proton by a weak basic catalytic site of the enzyme, such as the imidazole group of histidine, from the α-carbon of an amino acid residue in peptide linkage. The pKa value of an α-proton has been estimated to be higher than 20 but might well approach the upper limit of enzyme-catalysed proton transfer reactions.19 For the removal of such an α-proton, an enzyme could stabilize a planar enolate anion intermediate of the bound substrate (Scheme 2). Such a strategy has been demonstrated for the mandelate racemase, a member of the enolate superfamily, which engages the carbonyl group of mandelate in a low-barrier hydrogen bond and coordinates it to Mg2+.20 Moreover, recent investigations on an isomerase from the venom of a spider, which catalyses the stereoinversion of a Ser within a peptide chain,11 make an enolate anion a likely intermediate in this reaction21 (Scheme 2).
Studies of the racemization of free amino acids have indicated that the reactant side chain group also contributes to the stability of the enolate anion. Aryl side chain groups, such as phenylglycine (Phg), can stabilize an intermediate enolate anion primarily by resonance.22 In a similar manner, Met enhances the stability of the enolate by orbital overlap.23 Indeed, studies on the enzyme-catalysed isomerization of various model peptides have shown that the highest observed relative reaction rates were obtained with peptides containing either of these residues at position two.24
A deprotonation/protonation mechanism is also utilized by several cofactor-independent amino acid racemases. Consistent with the idea of a planar transition state, these reactions are well known to be inhibited by planar intermediate analogues. An example is the inhibition of proline racemase by the planar compound pyrrole-2-carboxylic acid.25 Furthermore, planar enamine and imine species have been shown to serve as potent inhibitors of the enzymes diaminopimelate epimerase and glutamate racemase.26 Finally, the spider isomerase is also inhibited by a planar substrate analogue, which contains a Δ2, 3-Ala at the isomerization site.21
The platypus enzyme, on the other hand, was shown to be inhibited by amastatin but not by the related bestatin, both of which are competitive aminopeptidase inhibitors.12, 27 The N-terminal residue of these compounds is either an aliphatic (amastatin) or an aromatic (bestatin) α-hydroxy-β-amino acid. Therefore, the different side chains were suspected to also determine the activity of the inhibitors and their binding to the active site.27 Both compounds only have a similar, relatively weak effect (Ki>10−4 M) on the reaction catalysed by the frog isomerase (Figure 3). It seems likely that those inhibitors with a β-amino acid bind to the active site in a substrate-like manner, albeit with the α2-carbon in an improper position for catalysis. Conversely, in the case of the unsaturated analogues, their tight binding to the enzyme as well as a high tolerance against the side chain of residue 2 suggest that they bind to the transition state conformation of the enzyme, once more emphasising the importance of geometry for catalysis.
Taken together, peptides containing an unsaturated amino acid as the second residue effectively inhibit the stereoinversion catalysed by the Bombina isomerase; this supports the notion of a planar transition state during this reaction. These are the first specific inhibitors to be described for this enzyme. These specific inhibitors supplement the set of tools available for studying this isomerase, which at present is very limited, mostly because attempts to express the enzyme in different over-expression systems have been unsuccessful. Therefore, these inhibitors could prove even more useful in future investigations of the post-translational isomerization of natural peptides.