Classical enzymology has been used for generations to understand the interactions of inhibitors with their enzyme targets. Enzymology tools enabled prediction of the biological impact of inhibitors as well as the development of novel, more potent, ones. Experiments designed to examine the competition between the tested inhibitor and the enzyme substrate(s) are the tool of choice to identify inhibitors that bind in the active site. Competition between an inhibitor and a substrate is considered a strong evidence for binding of the inhibitor in the active site, while the lack of competition suggests binding to an alternative site. Nevertheless, exceptions to this notion do exist. Active site–binding inhibitors can display non-competitive inhibition patterns. This unusual behavior has been observed with enzymes utilizing an exosite for substrate binding, isomechanism enzymes, enzymes with multiple substrates and/or products and two-step binding inhibitors. In many of these cases, the mechanisms underlying the lack of competition between the substrate and the inhibitor are well understood. Tools like alternative substrates, testing the enzyme reaction in the reverse direction and monitoring inhibition time dependence can be applied to enable distinction between ‘badly behaving’ active site binders and true exosite inhibitors.
Enzyme inhibitors have been used for decades as valuable tools in the study of enzyme mechanism of action, cell biology and physiology. Furthermore, enzyme inhibitors form one of the most successful classes of drugs (1). An important feature of inhibitor interaction with its enzyme target is the location of the inhibitor binding site on the enzyme. Knowledge of the binding site is critical for rational inhibitor optimization and for the understanding of its biological impact. Frequently, the enzyme active site also serves as the inhibitor-binding pocket. This is especially common in the case of small-molecule inhibitors for which the solvent-isolated nature of the active site and the availability of multiple interactions within the confines of a small binding pocket enable high binding affinity in spite of the limited number of interactions.
Usually, inhibitor binding to the active site will displace the substrate(s) and, vice versa, excess of the substrate can prevent inhibitor binding. The mutual exclusivity of substrates and active site–directed inhibitors is an important consideration during assay design for the purpose of inhibitor identification and evaluation. Moreover, it affects the activity of inhibitors in biological milieu. For example, the potency of protein kinase inhibitors that bind in the ATP-binding pocket is significantly reduced in cellular environment because of the presence of ATP concentration well above the Km for most kinases (2).
The competition between substrate(s) and active site–directed inhibitors is frequently utilized to determine whether an inhibitor binds in the active site by measurement of enzyme activity at multiple substrate and inhibitor concentrations. This review uses a functional definition for the classification of inhibition mode (3). Thus, competitive inhibitors increase substrate Km without affecting the Vmax, while inhibitor IC50 increases infinitely with an increase in substrate concentration (Figure 1). Classical non-competitive inhibitors, on the other hand, decrease the Vmax without affecting the Km and inhibit the enzyme independently of substrate concentration (Figure 1). Occasionally, non-competitive inhibitors can be affected by substrate concentration. These inhibitors are commonly termed mixed-type inhibitors. Substrate binding, while not mutually exclusive with a mixed-type inhibitor binding, appears to reduce inhibitor affinity (Figure 1B). The impact of substrate on inhibitor binding is defined by a factor, α, which represents the fold change in the Ki of the inhibitor for the substrate-bound form of the enzyme relative to the Ki towards free enzyme ((3), see Figure 1B legend for equation). In the literature, the product of α and Ki is also referred to as Ki’ or Kis. Because of the lack of mutual exclusivity with the substrate, this review will refer to mixed-type inhibitors as non-competitive.
Functional competition between an inhibitor and a substrate should not always be taken at face value. While many biochemists recognize that competitive inhibition can result from allosteric interference, fewer researchers are aware of the possibility that an active site–directed inhibitor can yield non-competitive inhibition patterns. The aim of this article is to review the mechanisms that cause active site–binding inhibitors to display non-competitive behavior. The review focuses on the biochemical mechanism rather than structural details which, in most cases, are not necessary for the understanding of the phenomenon.
Enzymes with exosites for substrate binding
Enzymes with high molecular weight substrates, such as proteases, derive most of the affinity to their substrates from regions outside of the active site where catalysis takes place. The substrate recognition site, in this case, is termed an exosite. This arrangement enables selective proteolysis in spite of the presence of sequences similar to the substrate cleavage site in other proteins. Among the best-characterized examples for exosite enzymes are the coagulation cascade serine proteases (4). The prothrombinase complex, which cleaves prothrombin and activates it, is composed of factor Va and the serine protease factor Xa and assembles on the surface of activated platelets. Surprisingly, para-aminobenzamidine that binds in the S1 pocket of serine proteases [(5), Figure 2A] is a non-competitive inhibitor of prothrombin cleavage by prothrombinase (6). However, when an alternate short chromogenic peptide is used as a substrate, instead of prothrombin, para-aminobenzamidine behaves as a competitive inhibitor (6).
The presence of an exosite for prothrombin binding on factor Xa was demonstrated using an exosite-binding antibody. This antibody showed competitive inhibition of prothrombin cleavage but did not inhibit cleavage of the alternate short peptide substrate (7). Thus, in exosite-containing enzymes, binding of the physiological substrate is not affected by an active site–directed inhibitor, with a resultant non-competitive inhibition pattern (Figure 2B). In this case, the use of an alternate small substrate that binds exclusively in the active site can help in the identification of the inhibitor-binding site (Figure 2C). Another example for an exosite-containing coagulation enzyme is the extrinsic Xase complex consisting of tissue factor and the serine protease, factor VIIa. As in the case of prothrombinase, para-aminobenzamidine is a non-competitive inhibitor of Xase when cleavage of the physiological substrate, factor X, is monitored but a competitive inhibitor with an alternate short chromogenic peptide (8). A somewhat different case is the inhibition of coagulation factor XIa by the active site–directed peptidic inhibitors, leupeptin and aprotinin. These inhibitors are competitive inhibitors of factor XIa in reactions with an alternate short peptide but show a mixed-type inhibition pattern with the physiological substrate, coagulation factor IX (9). In accordance with this observation, a factor IX–binding exosite was identified on the third apple domain of the non-catalytic heavy chain of factor XIa (10). The mixed-type inhibition pattern suggests an interaction between the exosite and the active site. Alternatively, leupeptin and aprotinin may be large enough to partially hinder the binding of factor IX to the non-catalytic exosite.
Exosite enzymes are not limited to the coagulation cascade. γ-Secretase, a membrane-bound multiprotein aspartyl protease, was shown to be non-competitively inhibited by the transition-state isosteres, pepstatin and L685458 (11). This study used the physiological substrate C100, a one hundred amino acid long peptide derived from the amyloid β-precursor protein. Unfortunately no shorter, alternate, substrate has been identified yet for γ-secretase. However, the observation that C100 co-purifies with the subunits of γ-secretase on an inhibitor column (12) suggests that the non-competitive inhibition by the transition-state isosteres is a result of C100 binding to an exosite.
Many enzymes undergo multiple transitions to complete a single catalytic cycle and return to a catalytically competent state. When these isomerization steps are sufficiently slow to become rate limiting for catalytic turnover, the enzyme is considered an ‘isomechanism’ enzyme (13). Under this scenario, it is possible that certain enzyme transition state(s) are capable of binding an inhibitor but not the substrate, thus yielding non-competitive inhibition patterns. Accordingly, while the inhibitor and the substrate bind to the same site, they do it at different time points and conformational states of the catalytic cycle (Figure 3), thereby no competition between them is observed. Indeed, Rebholz and Northrop reported this type of behavior with dead-end inhibitors (substrate mimics) of fumarase and carbonic anhydrase (14). Carboxylic acids such as acetate and formate are non-competitive inhibitors of carbonic anhydrase–mediated hydration of CO2. Re-hydration of the active site to replenish the water molecule that is used for catalysis is most likely the rate-limiting isomerization step of this reaction.
One characteristic feature of non-competitive, active site–binding, inhibitors of isomechanism enzymes is that, in many cases, changing reaction direction, using the product of the forward reaction as a substrate, can yield a fully competitive inhibition pattern. This interesting behavior is the result of binding of the forward reaction product to transition state(s) that are capable of binding the inhibitor (Figure 3). Thus, acetate and formate that were found to be non-competitive inhibitors of carbonic anhydrase–mediated hydration of CO2 are fully competitive inhibitors of the reverse reaction, bicarbonate conversion to CO2. Similarly, trans-aconitate is a competitive inhibitor of fumarase for fumarate hydration but a non-competitive inhibitor of the reverse direction reaction, with malate as a substrate (14).
The aspartyl protease, β-site amyloid precursor protein-cleaving enzyme (BACE), is inhibited non-competitively by the mechanism-based, hydroxyethylene-containing peptides, OM99-2 and statine valine (15). This non-competitive inhibition was attributed to an isomechanism involving the reprotonation of the catalytic aspartates. The strong solvent isotope effect on kcat supports the proposed reaction mechanism. The lack of isotope effect on the onset of inhibition by statine valine suggests that the inhibitor is able to bind to all the catalytic intermediates of BACE, thus evading competition from the substrate.
Two-step or induced-fit mechanisms
Another important mechanism involved in non-competitive inhibition pattern observed with several of the most successful enzyme inhibitor drugs is the two-step binding mechanism [reviewed extensively by Swinney as well as by Tummino and Copeland (16,17)]. In this case, inhibitors often interact with the enzyme target active site in two successive steps (Figure 4). The first step usually has rapid association–disassociation kinetics and is fully competitive with the substrate. In the second step, however, inhibitor binding induces a slower conformational change in the enzyme, resulting in stabilization of the enzyme–inhibitor interaction to form the EI* complex (Figure 4). Disassociation of the inhibitor from the EI* complex does not take place until the EI* complex reverts back to EI, which occurs very slowly. The small magnitude of k−2 relative to substrate binding and disassociation is responsible for the apparent lack of competition in two-step inhibitors. This mechanism has also been referred to as insurmountable antagonism as the substrate cannot overcome inhibition because of the stability of the EI* complex that prevents true equilibrium between substrate and inhibitor binding (16). For example, the active site–directed inhibitors of cyclo-oxygenase 1 and 2, valdecoxib and celecoxib, appear to deviate from competitive inhibition patterns. Binding of these inhibitors follows a two-step and, in some cases, a three-step mechanism with the third step being an irreversible inactivation of the enzyme (18). This multistep-binding mechanism was implicated not only in augmenting inhibitor activity but also in enhancing selectivity toward cyclo-oxygenase 1. The most extreme form of two-step, insurmountable, mechanism involves irreversible covalent attachment of the inhibitor to the enzyme. In the case of irreversible, covalent inhibitor binding, k−2 is zero and k2 is termed kinact. This was demonstrated to be the mechanism of the CYP 17 inhibitor, abiraterone, currently in clinical testing (19).
In several cases, the structural basis for the EI to EI* transition has been established. Aryl diketoacid inhibitors of protein tyrosine phosphatase 1B bind in the active site but display a non-competitive inhibition pattern. The X-ray structure of inhibitor–enzyme complexes reveals that when the inhibitor occupies the active site, the WFD loop assumes an ‘open’ conformation that prevents alignment of the scissile bond with the catalytic residues (20). A somewhat related situation was described for the complexes of the kinase inhibitors, Gleevec, BAY43-9006 and BIRB-796 with their cognate kinase targets. All of these inhibitors induce the inactive ‘DFG out’ conformation, although this does not always lead to non-competitive inhibition patterns with respect to ATP [reviewed in (21)].
One of the hallmarks of two-step inhibitors is time-dependent inhibition. Longer incubation of the enzyme with the inhibitor favors the accumulation of EI* complex that is ‘resistant’ to competition with the substrate. Thereby, the observed IC50 decrease as the incubation time of the enzyme and the inhibitor increases. This was indeed demonstrated for the cyclo-oxygenase 2 and CYP17 lyase inhibitors discussed previously (18,19). In cases where it is not feasible to monitor enzyme activity over short periods of time, binding assays can be utilized to establish a two-step mechanism (18).
Non-competitive inhibition of bisubstrate/product enzymes
Non-competitive inhibition by active site–binding inhibitors was also reported with bisubstrate enzymes that follow a compulsory ordered substrate addition or product release. 11β-Hydroxysteroid dehydrogenase (11β-HSD1) utilizes NADPH to reduce and activate the glucocorticoid cortisone. NADPH has to bind 11β-HSD1 before cortisone can bind, and following catalysis, the reduced steroid, cortisol, is released prior to the disassociation of NADP [Figure 4, (22,23)]. Inhibition of cortisone reduction by compound C, a small molecule inhibitor of 11β-HSD1 (Figure 5), was found to be non-competitive with both substrates, cortisone and NADPH. However, compound C is fully competitive with cortisol when the reverse reaction is monitored (23), indicating that compound C binds in the steroid-binding pocket of 11β-HSD1. The observed non-competitive inhibition pattern of compound C with cortisone is the result of preferential binding of compound C to the NADP-bound form of the enzyme that is incapable of binding cortisone (Figure 5). Interestingly, a more recent study reported similar behavior of an unrelated inhibitor of 11β-HSD1 (24), implying that preferential binding to the NADP-bound form of 11β-HSD1 is not limited to compound C.
A related example involves aldose reductase which catalyzes the NADPH-mediated reduction of a wide variety of carbonyl-containing compounds. The crystal structure of aldose reductase shows that the non-competitive inhibitor zopolrestat binds in the active site of the NADP(H)-bound enzyme (25). While not demonstrated directly, it is likely that the non-competitive inhibition by zopolrestat is a result of preferential binding to the NADP-bound form of aldose reductase. Similarly, non-competitive inhibition patterns observed with hypoxanthine phosphoribosyltransferase (PRPP) inhibitors identified via active site–docking models (26) may also be the result of preferential binding to enzyme–product complex. Indeed, PRPP was reported to follow compulsory ordered substrate binding and product release (27).
A somewhat different case involves inhibitors binding to the purine site (p-site) of adenylate cyclase. This enzyme has only one substrate (ATP) but two products (cyclic AMP and pyrophosphate, PPi). The dead-end inhibitor, 2-deoxyadenosine, is uncompetitive with ATP (inhibition increase with increased ATP concentration) and binds to adenylate cyclase only when one of the products, PPi, is bound. Although the order of product release appears to be random, the binding of 2-deoxyadenosine stabilizes the enzyme–PPi complex and is not competed by ATP which is unable to bind to enzyme–PPi complex (28).
Non-competitive inhibition of protein kinases
The recent clinical success of several kinase inhibitors for cancer therapy increased the interest in identifying non-competitive kinase inhibitors. Such inhibitors are expected to have increased potency in vivo as they evade competition with the cellular ATP pool, which is in large excess to the reported Km for most kinases (2). Furthermore, inhibitors utilizing an alternative site are expected to be more selective because they presumably bind to a site less conserved than the ATP-binding pocket. Although several kinase inhibitors were shown to bind sites that are distinct from the ATP-binding pocket (21), there are multiple reports of broad specificity kinase inhibitors that display non-competitive inhibition pattern with respect to ATP. The non-selective kinase inhibitor staurosporine is an ATP non-competitive inhibitor of protein kinase C (29). Nevertheless, crystal structure of its complex with another form of protein kinase C (30) as well as with protein kinase A (31,32) indicates binding in the ATP-binding pocket (Figure 6). The tyrphostin inhibitor, PP1, was found to be an ATP non-competitive inhibitor of pp60c-src (33) but was shown to bind to the ATP-binding pocket of another kinase, Hck (34). An intriguing example involves another tyrphostin, AG1296. This inhibitor is a mixed-type inhibitor with respect to ATP when the activated form of platelet-derived growth factor receptor tyrosine kinase (PDGF) is tested (35). However, AG1296 is fully ATP-competitive with the inactive form, consistent with binding to the ATP-binding pocket demonstrated for this class of inhibitors. As protein kinases are bisubstrate/biproduct enzymes, the non-competitive inhibition patterns observed with active site–binding kinase inhibitors could be the result of inhibitor binding to enzyme–product complex in an ordered substrate-binding/product-release mechanism. It is also possible that an iso or a two-step mechanism may be involved in mediating non-competitive inhibition patterns of kinase inhibitors binding in the ATP-binding pocket.
Several different mechanisms account for non-competitive inhibition patterns of inhibitors binding in the active site. Caution should be exercised in the interpretation of non-competitive inhibition patterns, especially when structural information indicates binding in the active site of the tested enzyme or a related one. Using alternate small substrates can help differentiating between inhibitors binding to an exosite and apparent non-competitive patterns resulting from the existence of an exosite for substrate binding. For isomechanism and multiple substrate enzymes, examining the effect of inhibitors on reactions performed in the reverse direction can facilitate distinction between active site and true exosite inhibitors. Two-step inhibitors can be identified by performing time-dependence inhibition studies or by monitoring inhibitor-binding kinetics.
I thank Donna Pedicord and Dr. Dietmar Seiffert for their helpful comments on this manuscript and Dr. Anzhi Wei and Dr. John Tokarski for preparing Figures 2A and 6.