|Home | About | Journals | Submit | Contact Us | Français|
While allostery draws increasing attention, not much is known about allosteric mechanisms. Here we argue that in all proteins, allosteric signals transmit through multiple, pre-existing pathways; which pathways dominate depend on protein topologies, specific binding events, covalent modifications and cellular (environmental) conditions. Further, perturbation events at any site on the protein surface (or in the interior) will not create new pathways but only shift the pre-existing ensemble of pathways. Drugs binding at different sites or mutational events in disease shift the ensemble toward the same conformations; however, the relative populations of the different states will change. Consequently the observed functional, conformational, and dynamic effects will be different. This is the origin of allosteric functional modulation in dynamic proteins: allostery does not necessarily need to invoke conformational rearrangements to control protein activity and pre-existing pathways are always defaulted to during allostery regardless of the stimulant and perturbation site in the protein.
Proteins are the ‘workhorses’ of the cell. Their response to changes in the cellular environment is modulated by an effector. The effector perturbs one site and thereby leads to an altered activity in a second, substrate site. Allostery is regulation at a distance by conveying the information from the first site to the second. Allostery is relayed through the cellular membrane, the cytoplasm and into the nucleus amplifying signaling (Ma and Nussinov, 2009) and the signals are sent dynamically (Smock and Gierasch, 2009). Allostery can result from physical binding events (with proteins, DNA/RNA, small molecules, lipids); light, as the photoswitch LOV2-Jα illustrates, fluctuating between ‘dark’ inactive and ‘light’ active conformations (Yao et al., 2008); mutational events; covalent modifications (phosphorylation, glycosylation, tethering (Hardy et al., 2004); or changes in the environment (pH, temperature, ionic strength, etc). Allostery was originally described in oligomeric systems (Changeux and Edelstein, 2005). Two recent reviews (Cui and Karplus, 2008; Goodey and Benkovic, 2008) provide a historical perspective: the first highlights cooperativity; the second emphasizes that catalysis and allostery emerge via common communication routes. Key questions include comprehensive description of allosteric mechanisms (Tsai et al., 2009), prediction of allosteric sites whether on the protein surface or of residues whose mutation could lead to allosteric effects, the pathways through which signals travel, and allosteric drug discovery. Currently, observation of specific pathways experimentally is a challenging problem. However, NMR provides information on protein dynamics from which pathways can conceivably be inferred (Boehr et al., 2006; Kern and Zuiderweg, 2003; Swain and Gierasch, 2006; Volkman et al., 2001). Recently, using NMR relaxation dispersion techniques the dynamic process through which the KIX domain of CREB binding protein communicates allosteric information has been directly observed (Bruschweiler et al., 2009), revealing that information is transmitted through an evolutionarily conserved network of residues. Below, we first briefly describe the principles of allostery, the “old” view, the so-called “new” view and an “updated new” view (Tsai et al., 2008). The “updated new” view leads us to re-address some of these questions. In particular, it highlights an allosteric property which has largely been overlooked: allostery invariably involves multiple pathways. Multiple pathways can relate to mutational effects in disease; to modulation in signaling pathways and to drug discovery. However, key to allostery, regardless of the site and nature of the perturbation events propagation will take place via the same pathways. The different perturbation events will simply shift pre-existing populations. This is the origin of allosteric modulation in cellular pathways and in drug binding effects.
Over the years, static pictures of allosterically-regulated proteins indicated a change in the shape of the substrate binding site between the On/Off states depending on whether the effector molecule was bound at the allosteric site. This has led to the paradigm that allostery involves conformational change. The two classical models, the Monod-Wyman-Changeux (or, MWC for short) (Monod et al., 1965) and the Koshland-Nemethy-Filmer (KNF) (Koshland et al., 1966), described allostery as a binding event at one site altering protein activity via a conformational change at the second site. The MWC model (Monod et al., 1965) emphasized that the conformational transition is a concerted action between two co-existing, distinct states (relaxed and tense, or R and T); on the other hand, the KNF model (Koshland et al., 1966) formulated it as a sequential, induced conformational change by the binding at the first site. The classical “old” view rested on two assumptions: that there are two distinct conformations and in the absence of a ligand their ratio is governed by the equilibrium constant; and that allostery involves a change of shape. In contrast to the “old” view, the more recent “new” view recognized that the native states are ensembles of pre-existing populations; thus, an allosteric effector leads to an equilibrium shift of pre-existing conformational and dynamic states (Gunasekaran et al., 2004). Now, the “updated new” view posits that allostery does not necessarily even involve a change of shape (Tsai et al., 2008). Recent data (Daily and Gray, 2007; Popovych et al., 2006; Tsai et al., 2008) validate the theoretical proposition (Cooper and Dryden, 1984), indicating that even if there is no visual change in the shape of the backbone at the substrate binding site upon binding of the allosteric effector, there could still be an allosteric change in the substrate binding site, and in protein activity. Popovych et al. (Popovych et al., 2006) presented direct experimental evidence illustrating that allostery can be mediated solely by changes in protein dynamics without any conformational change. Daily and Gray assembled an allosteric protein benchmark of pairs of known inactive and active allosteric protein structures from the Protein Data Bank (Berman et al., 2002). The dataset includes structural-pairs with no or subtle conformational change, governed largely by entropy (Daily and Gray, 2007). This absence of conformational change leads to a definition of allostery in pure thermodynamic terms: allostery can be controlled by enthalpy; by enthalpy and entropy; or solely by entropy (Tsai et al., 2008; Tsai et al., 2009). This “updated new” view is important since it helps in understanding allosteric mechanisms and thus in prediction of allosteric sites, allostery-related residues, allosteric drugs, and allosteric modulation.
The “old” view defined allostery in terms of two discrete states. It sought a single, well-defined propagation pathway between the allosteric- and substrate-binding sites which would lead to a conformational change. All allosteric residues were on this pathway and the signals were assumed to be transmitted by mechanically forming and breaking noncovalent interactions between on-pathway residues (Figure 1A). The old view was unable to offer an explanation for disease-causing (or, preventing (Trible et al., 2007)) mutations which were not at the binding sites, retained the global protein conformation, and did not lie on that pathway. In contrast, in the “new” view (Figures 1B,C) it is not a specific propagation pathway but an ensemble of states (Gunasekaran et al., 2004) that is the key for allostery; thus multiple pathways, major and minor, between the allosteric and the substrate binding sites. At the allosteric site noncovalent binding, covalent modification such as tethering, phosphorylation, or a point mutation create local stress. Local stress perturbs the structure, and the perturbation propagates via multiple pathways from the allosteric to the substrate binding sites. The PDZ2 domain interacts with a C-terminal peptide ligand obtained from the Ras-associated guanine nucleotide exchage factor 2 (RA-GEF2). It is known that binding site histidine 71 (hPTP1E PDZ2 numbering) determines the PDZ2 specificity, defining PDZ2 as class I PDZ domain (Kozlov et al., 2002). For the PDZ domain family, Ranganathan and coworkers predicted and confirmed experimentally a set of energetically-coupled positions for the binding site residue His71. In Figure 2a, this set of residues has been mapped onto the PDZ2/RA-GEF2 structure, forming a pathway that starts at the RA-GEF2 binding site and ends on the opposite side of the domain. Side-chain methyl dynamics measurements on the PDZ2 domain showed that allosteric transmissions due to RA-GEF2 peptide binding may be dominated by side-chain dynamics (Fuentes et al., 2004). Residues whose side-chain dynamics changed significantly upon peptide binding are also illustrated in Figure 2a. These residues form an alternative pathway starting at the peptide binding site and ending at two distal surface regions from the peptide binding site (see legend to Figure 2). These two pathways share common residues belonging to the protein core and distal surface 1. These findings based on the PDZ domain suggest the existence of multiple pathways from the perturbation to the substrate binding sites. The KIX domain of the CREB binding protein CBP provides another example of an allosteric protein with experimental data supporting the presence of multiple allosteric communication pathways. The KIX domain binds the mixed lineage leukemia transcription factor (MLL) inducing binding of the activation domain from c-Myb (Goto et al., 2002). Recently, using NMR relaxation measurements, a pathway through which the allosteric information is transmitted upon MLL binding has been observed (Bruschweiler et al., 2009). This pathway, which is formed by hydrophobic amino acids experiencing significant structural rearrangements, proceeds from the MLL binding site toward the c-Myb binding site (Figure 2b). Interestingly, a previous NMR study identified a set of residues exhibiting significant backbone amide chemical shift changes upon MLL binding (Goto et al., 2002). These amino acids form a hydrophobic groove that contains MLL binding site residues and other solvent exposed residues belonging to a distinct protein surface and are possibly involved in binding of other ligands (Figure 2b). Thus, in this example experimental results suggest the existence of two pathways that transmit the information to different protein surface patches.
An allosteric network has also been recently detected in caspase 1 (Datta et al., 2008). Of particular interest, 21 hydrogen bonds from nine side chains belonging to this network which connects the active and the allosteric sites were observed to alternate between the on-state and the off-state. Alanine-scanning mutagenesis of these side chains has shown that only two of these (Arg286 and Glu390), which form a salt bridge have major effects on catalysis, reducing it by 100- to 200-fold. Two neighboring residues, Ser332 and Ser339, have minor effects, causing 4- to 7-fold reductions, suggesting that the salt-bridging residues are on the major allosteric pathway, whereas the neighboring residues may reside on a minor communication pathway (Figure 3). Further, even a homologous substitution of the salt-bridging residue R286K causes a large 150-fold reduction. The structures of these variants suggest that in addition to salt bridge formation, these residues play an important role in the coordination of solvent water molecules near the allosteric binding pocket. A critical salt bridge important for specificity in discriminating between binding partners, has also been detected in p160 coactivator ACTR (Demarest et al., 2002). Computationally, major and minor pathways have been detected in the tRNA synthetases (Sethi et al., 2009).
Though not formulated in allosteric terms, Ansari et al. (Ansari et al., 1985) pointed out that similar to an earthquake, stress released from the focus could dissipate through propagation of deformation, and in the form of waves. From the theoretical standpoint, as in protein folding and in earthquake, in protein allostery a realistic scenario involves preferred pathways under given conditions. In principle, which pathway dominates can be determined from mutational effects: a mutation with a significant effect lies on a major pathway; a minor effect suggests a minor pathway. Mutations can break native state interactions and make new interactions (Figure 3). While mutational effects can be assessed by the extent of the allosterically-related conformational change, this is not necessarily the case. The perturbation caused by the effector could be reflected solely in entropy changes with no observable conformational change (Tsai et al., 2008), as illustrated by Popovych et al. (Popovych et al., 2006) and in structural comparisons of different allosteric states (Daily and Gray, 2007) (Figure 1C). The changes in entropy reflect all pathways under given conditions; that is, the landscape. Further, the “new” view is based on pre-existing equilibrium, invoking transitions between protein conformational and dynamical states (Formaneck et al., 2006; Kumar et al., 2000; Ma et al., 1999; Ma et al., 2002). Thus, key allosteric residues, predicted via coupling between residue-pairs quantified by statistical coupling energies (Suel et al., 2003) or experimentally identified by double-mutant cycles (Horovitz and Fersht, 1990), are not only those lying on the communication paths (Horovitz and Fersht, 1992; Sadovsky and Yifrach, 2007) but also those important for conformational and dynamical transitions. Mutations affecting these transitions could get the protein trapped, hindering population shift. Topologically, protein structures can be computationally decomposed into energetically independent modules, the building blocks of protein domains (Hu et al., 2007). Inter-modular residues play a key role in signal transmission. These residues are rigid, sustaining key amino acid interactions for the communication between modules. Most of the inter-modular residue interactions form long-range contacts that are predominantly involved in mediating signaling. Multiple pathways between the perturbation- and the substrate-binding sites are likely to share such central residues, but to diverge in modules. Further, recently, comparisons of crystal structures of the inactive and active forms of protein pairs led to identification of ‘rigid bodies’ (Daily and Gray, 2009). Analysis of the interactions between these showed that they are cyclically connected, again associating protein motions with network organization. Thus, multiple pathways always exist even if undetected by experiment; they can be reflected in dynamics and in conformational changes. The question is whether they encode different functions. Currently, no such cases are known, probably since it is only a matter of the relative concentration of the populations thus not sufficiently robust and sensitive to conditions.
Above, we related to pathways through which the signal is communicated between sites. A pathway implies a specific sequence in time and space. A pathway consists of a set of residues which are in (dynamic) contact. Allosteric pathways in a protein can be visualized as strain energy, created by a perturbation at the effector's site, radiating out to ease the unbalanced energy through residue-residue interactions (Figure 3). Theoretically, in terms of such a description, allosteric pathways are the outcome of a summation of many individual micro-pathways. Here we define a micro-pathway as a specific linkage starting from a residue at the perturbation site which follows a specified residue-residue interaction path down to a residue at the second (substrate) binding site. If the weightings for all residue-residue interactions in a protein are known, a statistical evaluation of the frequency for each residue to be on the pathway that conducts the allosteric wave from the effector's site to the second (substrate) site can be computed through a complete enumeration of all micro-pathways. In our definition, the pre-existing (major and minor) pathways are the outcome of a visualization of the linkage based on selected residues with a high statistical frequency above certain scale. The caspase-1 functional ‘hot wire’ or ‘allosteric circuit’ (Datta et al., 2008) provides an excellent experimental example for major and minor pathways detected via mutagenesis studies illustrating our point.
Above, we further related to protein conformational and dynamical states. A conformational state is determined by its associated free energy. For example, a binding event at the allosteric site shifts the population of the unbound (conformational) state to a bound (conformational) state. It is determined purely by thermodynamics. The change at the substrate site, however, is determined dynamically via residue-residue interactions to release the created strain energy if there is a conformational change at the allosteric site. If however, there is no conformational change at the allosteric site, the stiffness (i.e. lose of entropy) will still propagate dynamically through residue-residue interactions down to the substrate site.
Above, signal transmission occurred between two sites: the effector and the substrate. This situation rarely occurs in vivo. In the cell, all dynamic proteins have more than a single effector perturbation site. Proteins generally function when in large assemblies, and when regulated by multiple coincident environmental changes, binding or modification events. Let us consider the following multiple co-occurring perturbation scenarios: (i) only two molecules bind; however, external conditions change; (ii) disease-related mutations occurred as for example in the pVHL which still needs to bind its substrate and partner proteins (Knauth et al., 2006); (iii) there are more than two binding sites, as in tumor suppressor protein p53 (Riley et al., 2008); pVHL (Ohh et al., 2000); or regulatory proteins in the ubiquitin ligase E3 complex (Hao et al., 2007); (iv) there are multiple modification sites in the protein, for example phosphorylation sites on the p53 (Riley et al., 2008), or acetylation (Luo et al., 2004); (v) enzymes or receptors which recognize a large number of substrates, such as the kinase which must recognize between one and a few hundred phosphorylation sites, while at the same time discriminating an order of magnitude more potential phosphorylation sites in the kinome (Ubersax and Ferrell, 2007), arguing for possible allosteric modulating effectors; (vi) dynamically co-regulated proteins which have to respond to complex cell signals (Komurov and White, 2007); (vii) and finally, binding sites shared by different proteins (Keskin and Nussinov, 2007; Tuncbag et al., 2009). The conformations and dynamic effects of these are condition- (i.e. perturbation)-dependent. Clearly, each phosphorylation, mutational, covalent ubiquitination, nucleotide-binding or change in the environment is a perturbation event.
Signaling proteins constitute examples of combinations of such events. They are multi-modular and the transmission of information can involve several mechanisms such as post-translational modification, assembly into complexes, changes in subcellular distribution, enzymatic activity, and multiple binding events. Since mechanisms generally co-occur, they all present multiple perturbation events. Socs suppressors of cytokine signaling (Yoshimura et al., 2007) provide one example of a system subject to different mechanisms (Bullock et al., 2007). Socs proteins are induced by cytokines acting in a negative-feedback loop to inhibit cytokine signal transduction; they are also induced by various stimuli, such as lipopolysaccharide (LPS), isoproterenol, statins and cyclic AMP, suggesting multiple interactions in different environments with various protein/small molecule partners. The Socs proteins form part of the E3 ligase machine tagging proteins for degradation. The Socs box domain binds to the adaptor proteins, elongin C and elongin B, and the substrate binding domain binds to the substrate. Socs also undergoes post-translational modifications. All these events perturb the structure, enhancing allosteric effects. A second example relates to the molecular chaperones in the assembly of cellular complexes (Morimoto, 2002). Molecular chaperones respond to hormones and stress, and different combinations govern the activities of intracellular hormone receptors and heat shock transcription factors. Molecular chaperones dynamically ensure tight control, and rapid reversible transcriptional response. Thus their functional roles implicate complex interactions and multiple allosteric perturbation sites. A third example is provided by the 14-3-3 proteins. This conserved regulatory family modulates the action of target proteins by sequestration, relocalization and conformational changes. They bind signaling proteins including kinases, phosphatases, transmembrane receptors, enzymes, and structural and cytoskeletal proteins and are involved in metabolic pathways, redox-regulation, transcription, RNA processing, protein synthesis, protein folding and degradation, cell cycle, cytoskeletal organization and cellular trafficking (Kjarland et al., 2006). Thus, in vivo, for all (dynamic) proteins, the strain energy created at each of the many perturbation sites radiates out propagating and merging similar to waves initiating at several perturbed locations, to enhance, dissipate or alter the features of the deformation reaching the target site. Experiment is unable to directly observe specific propagation routes. However, since the relative pathway utilization varies, there are different observable allosteric effects which can be inferred from functional expression. Examples include the homeodomain-interacting protein kinase 2 (HIPK2) (Sombroek and Hofmann, 2009) which depending on the phosphorylation state activates the apoptotic programme by engaging diverse downstream targets, including tumor suppressor p53 and the anti-apoptotic transcriptional corepressor C-terminal binding protein. For the p53, specific post-translational phophorylation events are well documented to regulate its functions (Kruse and Gu, 2009). An additional example concerns Mdm2. Mdm2 is a negative regulator of p53. p53 promotes the transcription of Mdm2; in turn, Mdm2 binds to p53 and stimulates the ubiquitination of the p53 carboxy terminus, marking it for degradation. Mdm2 also negatively regulates pRb, similar to p53, and inhibits its regulatory growth function. pRb interacts with the transactivation domain of the E2F family transcription factors, and blocks E2F dependent transcription. The Mdm2 interaction with pRb disrupts the pRb–E2F binding leading to an increase in the E2F-dependent transcription (Martin et al., 1995). The stabilization of the E2F1 protein reflects another p53-independent component of Mdm2-mediated tumorigenesis via direct Mdm2-E2F1 interaction (Zhang et al., 2005). Mdm2 further interacts with PCAF, which competes with E1A for binding sites on P300/CBP (Jin et al., 2002). PCAF also provides a good example: PCAF has separate acetyltransferase and E3 ubiquitin ligase domains and a bromodomain for interaction with other proteins. In addition it also possesses sites for its own acetylation and ubiquitination (Linares et al., 2007). Hence, to conclude, each binding and each post translational modification (like phosphorylation or acetylation) event alters the relative pathway utilization. Consequently, the ensuing binding events to other proteins may be altered, leading to (some) functional modifications. This alteration in the relative pathway utilization in a given protein - which can be enhanced or suppressed by other perturbation event(s) - is the origin of allosteric modulation in all dynamic proteins.
While scaffolding proteins lead to colocalization and thus amplification and higher efficiency of signal transfer, they are not passive; they apparently also modulate the interactions of particular signaling modules. Thus, even the seemingly ‘inert’ proteins, whose role was originally believed to bring other proteins together for them to interact, have highly complex regulatory functions requiring allosteric multi-site communication. Scaffold proteins are involved in processes such as intracellular trafficking and pathway sequestering, and several factors have been shown to influence their signaling function (Kolch, 2005). MAPK (mitogen-activated protein kinase) scaffolds were suggested to hold the kinases in a manner that directly enhances their mutual interactions, thus potentially enhancing the rate of the phosphate transfer. At the same time, the MAPK-pathway ‘scaffold’ prototype is likely not a passive docking site for multiple kinases. Ste5 is a yeast protein involved in the Fus3 MAPK pathway (Burack and Shaw, 2000). Shuttling of the Ste5 complex through the nucleus is required for maintenance of pathway-constituents in a state competent to participate in signaling: Ste5 that is unable to transit the nucleus cannot localize at the periphery and is unable to activate the pathway. On the other hand, Ste5 transiting the nucleus is able to localize and activate the pathway (Mahanty et al., 1999). While currently there are no direct amplification data for the scaffolding proteins, studies of Ste5 revealed multi-step, multi-component, multi-binding site scaffolding proteins that can allow fine-tuned regulation (Burack and Shaw, 2000). The ubiquity of these multileveled kinase cascades allow not only signaling amplification but also incorporate regulatory checkpoints (Kolch, 2005). Hence, the colocalization, amplification, and the multiple interactions of the scaffolding proteins argue for multiple perturbation sites. The different times, combinations and complex formation of these proteins inherently implies function modulated by multiple allosteric pathways originating at these perturbation sites and propagating like waves to the substrate binding sites.
Multiple pathways can be seen in allosteric modulators like hormones and neurotransmitters whose signals are recognized, amplified and transmitted. A hormone is released into the bloodstream; a neurotransmitter is released from a nerve terminal following an electrical impulse in direct apposition to its target cells to ensure rapid and specific delivery of the signal. Allosteric modulators enhance the signals; they do not compete with endogenous ligands and therefore can exert their influence even if an endogenous ligand is bound to another site on the same target. Allosteric modulators contrast orthosteric allosteric effectors which compete with endogenous ligands for the same site on a given target. Allosteric modulators are not limited to simply turning a receptor on or off; rather, they offer a control over the degree of activation or deactivation like a light dimmer switch. Small molecule drugs can mimic this effect by inducing allosteric activation (Conn et al., 2009; Kalatskaya et al., 2009; Kenakin, 2007; Langmead and Christopoulos, 2006; Pelkey et al., 2007). Modulating allosteric drugs can have a non-observable activity in the absence of endogenous ligands; as such, they offer a less disruptive way to influence the functioning of biological systems. Since on their own the perturbation they cause is small, they better preserve the regulation of cellular processes compared to orthosteric (same site) approaches. Targeting allosteric binding sites represents a powerful mechanism for selectively modulating receptor function. Such modulators enhance (or reduce) ‘traffic’ in innate propagation pathways.
The seven-transmembrane (7TM) receptors are the largest, most ubiquitous and versatile membrane receptor family; they also constitute the most common drug targets. The N-terminal is outside; the C-terminal inside the cell. Binding of small extracellular ligands shifts the ensemble from a resting to an activated state driving signal transduction pathways and cellular responses. Orthosteric interaction obstructs access of endogenous ligands; allosteric binding usually occurs away from this site. Both can be associated with conformational (or dynamical) change. The M(2) muscarinic receptor provides a specific recent example: it has two distinct sites, orthosteric and allosteric. The allosteric site is recognized by compounds such as gallamine. Muscarinic receptors form oligomers resulting in two or more orthosteric and allosteric sites per multimer, an arrangement allowing cooperative interaction. Redka et al. (Kenakin, 2007; Redka et al., 2008) have shown that orthosteric ligands like N-methyl scopolamine and oxotremorine-M bind to the allosteric site of the M(2) muscarinic cholinergic receptor although at higher concentrations. The allosteric effects of orthosteric compounds like N-methyl scopolamine and oxotremorine-M mimicking the effects of the allosteric gallamine raise the possibility that they can contribute to conformational or dynamical events assumed to initiate at the orthosteric site, thus suggesting coincidence of some propagation pathways. Of particular interest is that allostery was detected by the propensity of seven transmembrane receptors to form dimers, thus demonstrating allosteric effects through binding at the orthosteric site. Thus, in principle any part of the protein can be an allosteric site, and the thermodynamic outcome of the binding may be to bias the ensemble to a similar pharmacologically-relevant endpoint. The binding of a ligand at either site, orthosteric or allosteric, stabilizes a preferred conformational state.
While frequently modulating drugs enhance the activity, Quiniou et al (Quiniou et al., 2008) have recently obtained a negative allosteric peptide regulator of IL-1, a major proinflammatory cytokine which interacts with the IL-1 receptor I (IL-1RI) complex. Another interesting recent case is that of the orthosteric agonist neurokinin A (NKA). NKA interacts with the tachykinin NK2 receptors (NK2Rs) stabilizing the receptor in at least two different active conformations (A1L and A2L). A small molecule, LPI805, is a noncompetitive inhibitor of NKA binding to NK2Rs, which decreases the number of NKA-NK2R complexes in A2L conformation and increases those in A1L conformation. Analysis of signaling pathways of NK2Rs showed that LPI805 dramatically inhibits the NKA-induced cAMP response while slightly enhancing the NKA-induced calcium response, establishing that allosteric modulators can be used to promote functional selectivity (Maillet et al., 2007). Another high-profile, impressively potent drug, Gleevec, is an example of an allosteric inhibitor that alters protein kinase conformation to block productive ATP binding (Bogoyevitch and Fairlie, 2007). Since Gleevec does not bind at the conserved ATP-binding site and as such does not compete with ATP binding, it has fewer side effects. Gleevec binds to the inactive conformational state, shifting the equilibrium in this (inactive) direction (Formaneck et al., 2006; Kumar et al., 2000; Ma et al., 1999; Ma et al., 2002; Tsai et al., 1999). Finally, a small molecule, InsP6 (inositol hexakisphosphate) was recently shown to induce allosteric activation of a toxin Vibrio cholerae RTX (Repeats in Toxin). RTX is an actin-disrupting toxin which is autocleaved by an internal cysteine protease domain (CPD). The autocleavage is activated by InsP6 which binds to a conserved basic cleft, distant from the protease active site. CPD mutants indicate that InsP6 binding induces an allosteric change leading to the autoprocessing and intracellular release of toxin-effector domains (Lupardus et al., 2008).
Here we argue that allostery involves multiple pathways, and that all pathways - between all allosteric and all substrate binding sites pre-exist. As in protein folding, under any set of (cellular or in vitro) conditions, there will be major pathways and minor pathways. We posit that allosteric events at different sites will shift the ensemble; but they will not create new pathways. That is, we argue that all allosteric perturbation events such as binding of small molecules or other proteins on the protein surface or mutational or post-translational events occurring anywhere in the protein structure will only lead to a shift in the relative populations of the different states. The shift in the ensemble can lead to different observed conformational, dynamic and functional effects. That is, all pathways are already there; it is only a question of the population shift and which pathways dominate.
Allosteric perturbation creates stress. The strain energy created at the perturbation sites dissipates by radiating out, propagating and merging similar to waves initiating at several perturbed locations, merging to enhance, dissipate or alter the features of the deformation reaching target sites. This is the origin of allosteric modulation in all dynamic proteins. Further, allosteric perturbation at a newly discovered allosteric site will not lead to new conformational or dynamical states; it will however disrupt, enhance or modulate substrate binding. Thus, allosteric modulators are not limited to simply turning a receptor on or off; rather, they offer a control over the degree of functional activation or deactivation like a light dimmer switch. Allosteric effectors, including allosteric drugs, mimic this effect.
Such a scheme rationalizes allosteric functional modulation; it explains why binding at different sites will lead to similar functional effects, except - with different relative populations. Here, we provided a range of experimental examples substantiating such a theoretical proposition. We further note that the pathways form networks, and the networks have a modular organization with the modules connected via central residues.
This project has been funded in whole or in part with Federal funds from the National Cancer Institute, National Institutes of Health, under contract number HHSN261200800001E. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. This research was supported (in part) by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.