Hostname: page-component-6898cfcb55-th4rj Total loading time: 0 Render date: 2026-11-11T05:05:47.251Z Has data issue: false hasContentIssue false
  • English
  • Français

Protein–protein interactions as targets for small-molecule therapeutics in cancer

Published online by Cambridge University Press:  19 March 2008

Alex W. White,
Andrew D. Westwell and
Ghali Brahemi
Alex W. White*
Affiliation:
Welsh School of Pharmacy, King Edward VII Avenue, Cardiff University, Cardiff, CF10 3NB, UK.
Andrew D. Westwell
Affiliation:
Welsh School of Pharmacy, King Edward VII Avenue, Cardiff University, Cardiff, CF10 3NB, UK.
Ghali Brahemi
Affiliation:
Welsh School of Pharmacy, King Edward VII Avenue, Cardiff University, Cardiff, CF10 3NB, UK.
*
*Corresponding author: Alex W. White, Welsh School of Pharmacy, King Edward VII Avenue, Cardiff University, Cardiff, CF10 3NB, UK. Tel: +44 (0)29 2087 6308; Fax: +44 (0)29 2087 4149; E-mail: whiteaw@cf.ac.uk
Rights & Permissions [Opens in a new window]

Abstract

Small-molecule inhibition of the direct protein–protein interactions that mediate many important biological processes is an emerging and challenging area in drug design. Conventional drug design has mainly focused on the inhibition of a single protein, usually an enzyme or receptor, since these proteins often contain a clearly defined ligand-binding site with which a small-molecule drug can be designed to interact. Designing a small molecule to bind to a protein–protein interface and subsequently inhibit the interaction poses several challenges, including the initial identification of suitable protein–protein interactions, the surface area of the interface (it is often large), and the location of ‘hot spots’ (small regions suitable for drug binding). This article reviews the general approach to designing inhibitors of protein–protein interactions, and then focuses on recent advances in the use of small molecules targeted against a variety of protein–protein interactions that have therapeutic potential for cancer.

Information

Type
Review Article
Information
Expert Reviews in Molecular Medicine , Volume 10 , October 2008, e8
Copyright
Copyright © Cambridge University Press 2008

Fundamental processes in living cells, such as intracellular signal transduction and maintenance of cytoskeletal architecture, are largely controlled by proteins, often acting in concert with other protein partners through protein–protein interactions (PPIs). Inappropriate protein–protein recognition can contribute fundamentally to many diseases, including cancer. Selective, small-molecule modulation of PPIs is therefore an area of growing interest to pharmaceutical science (Refs Reference Toogood1, Reference Sharma, Ramsey and Bair2, Reference Arkin and Wells3, Reference Arkin4, Reference Fry and Vassilev5, Reference Yin and Hamilton6, Reference Chene7, Reference Lindsley8, Reference Wells and McClendon9).

Despite the pivotal role of PPIs in many processes relevant to cancer development, targeting PPIs as a therapeutic strategy in cancer is still in its infancy, and examples of modulators of protein complexes as potential therapeutic agents are few relative to the more widely studied area of small-molecule enzyme (e.g. kinase) inhibitors (Ref. Reference Collins and Workman10). The reasons for the dominance of kinase (and other enzyme) inhibitors as targets for modern cancer therapeutics are centred on the enzyme substrate-binding site. In the case of enzyme inhibition, the design methodology for these compounds is well established, and active sites are usually contained within deep clefts that clearly define the ligand-binding site and contain the critical amino acid residues required for ligand interaction. Furthermore, all molecules in a biological system are bathed in an aqueous environment, and enzyme active-site clefts tend to shield binding sites from water molecules that can otherwise hinder ligand interactions.

Small-molecule drug discovery against protein–protein targets is more challenging than for conventional small-molecule protein targets but represents an innovative direction for 21st century medicine. Although the field has already been the subject of several reviews (for a selection see Refs Reference Toogood1, Reference Sharma, Ramsey and Bair2, Reference Arkin and Wells3, Reference Arkin4, Reference Fry and Vassilev5, Reference Yin and Hamilton6, Reference Chene7, Reference Lindsley8, Reference Wells and McClendon9), in this article we aim to give a broad overview of emerging small-molecule PPI inhibitors in the anticancer field.

The concept of protein–protein interaction and ‘hot spots’

Among the reasons for the lack of modulators of protein complexes as therapeutic agents has been the notion that protein interfaces represent difficult or intractable targets because of the generally large and noncontiguous nature of PPI surfaces. This perceived problem is compounded by the lack of high-throughput screening technologies and paucity of small molecules in compound libraries with the size and functionality to modulate protein-complex targets selectively. In recent years, however, several reports have appeared that challenge the dogma that PPIs represent difficult targets to modulate using small molecules. In particular, the concept of PPI ‘hot spots’ has aided the development of small molecules that interact with a small area of an entire protein–protein interface to exert high-affinity and selective binding (Ref. Reference Bogan and Thorn11).

The initial challenge in developing PPI inhibitors is the discovery of specific PPIs and in turn identification of those that are ‘druggable’. Two types of PPI can be defined: transient and tight. Tight interactions can be identified more easily, through biochemical co-purification; in contrast, transient interactions have been characterised by site-directed mutagenesis and molecular modelling as it is often difficult to co-crystallise transient partners (Ref. Reference Rudolph12). Computational techniques have also been used to identify PPIs and their inhibitors, and have been recently reviewed (Ref. Reference Zhong, Macias and MacKerell13). Following the identification of a suitable PPI, the next step involves study of the binding interface and ligand design. This is also challenging compared with designing a ligand to interact with an active-site cleft. Interaction sites are typically relatively flat with a large surface area, although shallow grooves are sometimes seen. Ligand-binding sites may be exposed only after conformational changes concomitant with binding of protein partners (Ref. Reference Sundberg and Mariuzza14).

To understand better the nature of protein–protein interfaces, a study of 75 PPIs established a ‘standard size’ of 1600 ± 400 Å2 for an interaction surface. The smallest interfaces were approximately 1150 Å2 and the largest interfaces identified were up to 4660 Å2 (Ref. Reference Lo Conte, Chothia and Janin15). Given the large sizes involved, it would be difficult to target a small molecule to the whole interface. Similarly, large peptide-based inhibitors can be less effective since a single continuous peptide sequence may poorly match the binding surface (Ref. Reference Rudolph12). Fortuitously, within a binding interface only a small number of highly conserved amino acid residues – ‘hot spots’ – are crucial for the interaction, and thus these are often the target for drug design. Hot spot residues within a protein–protein interface are less obvious to identify than, for example, important residues within an enzyme active site. Although a laborious process, site-directed mutagenesis has been commonly employed to locate hot spot residues, leading to a definition of a hot spot as ‘a residue that, when mutated to Alanine, gives a distinct drop in the binding constant, typically ten-fold or higher’ (Ref. Reference DeLano16). Phage display has also been used to detect hot spots (Ref. Reference Sidhu, Fairbrother and Deshayes17).

Based on the analysis of currently identified hot spots, the amino acids tryptophan (Trp), tyrosine (Tyr) and arginine (Arg) occur frequently, often surrounded by hydrophobic regions. These amino acids are thought to exclude solvent from the important interacting residues and are favoured as they can form multiple interactions. For example, despite structural similarity, phenylalanine is three times less likely to occur in a hot spot than Tyr, presumably due to the ability of Tyr to form an additional hydrogen bond. Hot spots contain leucine, methionine, serine, threonine and valine residues less frequently (Ref. Reference Bogan and Thorn11). However, a review of the thermodynamic aspects of PPIs concludes that hydrophobic interactions are also a key driving force for PPIs, and that the complex thermodynamics of PPIs are ultimately a function of how protein–protein binding changes the internal structure of the molecule(s) (Ref. Reference Stites18).

Scope of the review

Over the past decade, the principal exploitation of PPIs has been through therapeutic antibodies, such as Herceptin, which are showing great promise as a new class of drugs in cancer treatment (Ref. Reference Nahta and Esteva19). Therapeutic antibodies are large proteins that interact directly, for example, with membrane-bound receptor proteins, such as growth factor receptors, to elicit a biological response. Although an important class of cancer therapeutic agents exploiting protein–protein recognition, therapeutic antibodies (or other ‘biologics’) are not included in this review; instead, we discuss the emerging class of small-molecule inhibitors of PPIs, focusing on their application to anticancer therapy. We consider established clinical agents that fall into the PPI modulator class (such as tubulin inhibitors), but concentrate on agents in preclinical development (such as the p53–MDM2 inhibitors). The mechanism of action of small-molecule PPI inhibitors differs from that of the antibodies in that a small, nonphysiological compound designed to bind preferentially to a PPI site blocks the binding of a further protein, preventing the formation of a protein dimer/multimer that would otherwise be essential for regulating a biological process.

Protein–protein interactions as therapeutic targets in cancer

Microtubules

One of the most successful therapeutic targets in cancer treatment is tubulin. There are several natural, semisynthetic and synthetic compounds that have significant clinical activity by affecting tubulin PPIs. They have been extensively reviewed elsewhere (Refs Reference Downing20, Reference Jordan21, Reference Jordan22, Reference Kavallaris, Verrills and Hill23), but we summarise the topic here for completeness.

Microtubules are large cylindrical proteins – 25 nm in diameter and up to several μm in length (Ref. Reference Jordan21). A microtubule is composed of copolymers of two monomers – α- and β-tubulin – binding in an alternating manner. Microtubules undergo constant dynamic polymerisation and depolymerisation at both ends of the filament (referred to as plus and minus ends), although the process is faster at the plus end (Refs Reference Downing20, Reference Jordan21, Reference Nogales24). Microtubule dynamicity is required for several biological processes including maintenance of cell shape, cellular signalling and movement; however, their best-studied role is in cell division (Ref. Reference Jordan21). Several drugs exert their antitumour effect by interfering with tubulin PPIs and are classified according to the site to which they bind on microtubules.

A crystal structure of the taxane analogue taxol (a natural product from the Pacific yew tree Taxus brevifolia) bound to tubulin indicates that it binds within a hydrophobic pocket. It is suggested that taxol stabilises microtubules by strengthening PPIs, thus inhibiting cell division (Ref. Reference Downing20). Several synthetic taxane derivatives are under clinical investigation. In addition, nontaxane compounds are known to bind at the taxane site and several are under preclinical investigation, including the epothilones (Ref. Reference Bollag25), eleutherobins (Ref. Reference Long26), discodermolides (Ref. Reference terHaar27), sarcodicyins and laulimalides (Ref. Reference Jordan21).

By contrast to taxol, colchicine (a natural product from the plant genus Colchicum) and its analogues destabilise microtubules. Although the crystal structure of the tubulin–colchicine complex is not available, studies suggest that colchicine interacts with β-tubulin at the interface between two tubulin monomers (Refs Reference Bai28, Reference Uppuluri29, Reference Downing and Nogales30).

Vincristine and vinblastine (natural products from the periwinkle plant Catbaranthus roseus), and new-generation analogues vinflunine and vinorelbine, also destabilise microtubules (Ref. Reference Erickson31). Vinblastine has been shown to interact with β-tubulin (Refs Reference Downing20, Reference Kavallaris, Verrills and Hill23, Reference Rai and Wolff32). Other natural products, including rhyzoxin, cryptophycins and dolastatin 10, have similar effects to the vincas (Ref. Reference Gupta and Bhttacharyya33).

p53

The tumour suppressor protein p53, known informally as ‘the guardian of the genome’, has been extensively studied in the field of cancer. It protects biological tissues from malignant transformation and forms a central part of the DNA-damage response. It is estimated that around 50% of all human tumours have mutations in the p53 gene (TP53), implying that cancer cells sustain viability by reducing the biological activity of p53 (Refs Reference Levin34, Reference Hollstein35, Reference May and May36).

Human p53 contains 393 amino acids arranged in four functional domains. The N-terminus interacts positively with transcription machinery and with MDM2, an E3 ubiquitin ligase. The second domain constitutes the DNA-binding region, and harbours 90% of the reported p53 mutations. The third domain is responsible for the formation of the p53 tetramer. The current model of p53 function suggests that DNA interactions occur with a tetramer form of p53 (actually a dimer of a dimer) (Refs Reference Levin34, Reference May and May36, Reference Bossi and Sacchi37).

Several events are known to induce the activity of p53, including DNA damage, hypoxia, oncogene overexpression and viral infections (Refs Reference May and May36, Reference Gudkov and Kmarova38, Reference Lu and Lane39). Numerous biological functions have been attributed to p53 such as apoptosis (Refs Reference Clarke40, Reference Yonish-Rouach41) and senescence (Ref. Reference Wynford-Thomas42), G1 arrest, cell cycle control, DNA repair (Ref. Reference Wang43) and angiogenesis inhibition. Among the most important of these functions is apoptosis (programmed cell death), a conserved process that mediates the elimination of a cell upon exposure to genomic stress. The complete mechanism through which p53 mediates apoptosis is still under study; it is mediated primarily but not exclusively by p53. Following a stressful event, p53 activates through its N-terminus many pro-apoptotic genes such as POXA and PUMA. Another suggested mechanism is linked to the mitochondria, whereby p53 binds to the mitochondrial membrane resulting in the formation of pores and the release of cytochrome c; this in turn activates caspases, which degrade the nucleus resulting in cell death (Refs Reference Oren44, Reference Marchenko, Zaika and Moll45).

Under normal conditions, cellular levels of p53 are barely detectable: it has a short half-life of 10–20 min as a result of a PPI with MDM2 that results in the negative regulation of p53 (Fig. 1) (Refs Reference Levin34, Reference Hollstein35). MDM2 catalyses the binding of ubiquitin to p53, thereby directing it to degradation by the proteasome (Ref. Reference Freedman, Wu and Levine46). In a typical feedback loop, p53 increases the transcription of MDM2 (Ref. Reference Wu47). Under stressful conditions, p53 concentrations are increased due to a reduction in its MDM2-mediated degradation and a concurrent increase in p53 transcription.

Figure 1

A conceptual diagram of a protein–protein interaction (PPI) inhibitor targeting a PPI (p53–MDM2) hot spot. In this diagram, the inhibition of the p53–MDM2 interaction by a nutlin inhibitor is used as an example to illustrate the generic concept of the action of all PPI inhibitors discussed in this article, since it is one of the best-characterised examples of PPI inhibition. MDM2 normally negatively regulates the antitumour protein p53 by inducing its proteasomal degradation; however, in response to cellular stress, particularly DNA damage, p53 levels rise as a result of increased p53 transcription and reduced MDM2-mediated degradation (not shown). The nutlin inhibitors are similarly thought to increase p53 levels, by blocking MDM2-mediated degradation. The PPI of p53 and MDM2 is mediated by the side chains of three p53 amino acids (Leu, Trp and Phe) that are recognised by MDM2. The nutlin inhibitors mimic the binding of these amino acid side chains and preferentially bind to MDM2. This physically blocks the interaction of the two proteins, thus releasing p53 that is free to exert a beneficial antitumour effect. This schematic illustration is derived from the X-ray data presented in Fig. 2a, and comparison of Figs 1 and 2a should be instructive. Abbreviation: MDM2, ‘mouse double minute 2’ homologue.

As increased p53 activity has an antitumour effect, there have been many attempts to elevate the biological activity of p53 by increasing its accumulation in the cell. The main strategies employed have been stabilisation of p53 or protection from ubiquitination and proteasomal degradation, primarily by blocking the p53– MDM2 interaction (Refs Reference Fujiwara48, Reference Scardigli49, Reference Vassilev50). Close examination of the p53–MDM2 interface reveals that it is the side chains of only three key p53 amino acids that make most of the interactions within a hydrophobic groove in the surface of MDM2 (Figs 1 and 2). In an X-ray-crystallography-derived model of the p53–MDM2 interaction shown in Figure 2a, a p53-derived peptide (green) is shown complexed with an MDM2 fragment (surface render) containing the hydrophobic binding cleft. The p53 fragment forms an α-helix from which the side chains of the three amino acids Leu26, Trp23 and Phe19 (shown left to right) project into the binding cleft, promoting the p53–MDM2 interaction (Refs Reference Kussie51, Reference Lin52). Further analysis of this hot spot revealed that the area of the interaction site is about 300 Å2 (Ref. Reference Vassilev50) The hot spot is a good target for drug design – the strategy being to design a drug that mimics the conformation and interaction of the three p53 residues.

Figure 2

Structural analysis of the interactions of MDM2 with p53 and with a nutlin inhibitor. (a) X-ray crystallographic models of the p53–MDM2 interaction (Ref. 51). A fragment of MDM2 is shown rendered as a solvent-accessible solid surface and coloured by atom type (carbon, grey; oxygen, red; nitrogen, blue; sulphur, yellow). The p53 fragment is rendered as a green ribbon showing only the side-chains of Leu26, Trp23 and Phe19 rendered as sticks and coloured by atom type (carbon, green; nitrogen, blue). These three nonpolar side-chains project into a hydrophobic (i.e. predominantly carbon-containing) pocket in the surface, thus stabilising the interaction between p53 and MDM2. (b) X-ray crystal structure of the MDM2–nutlin-inhibitor complex (Ref. 57). The MDM2 fragment is again shown as a solvent-accessible solid surface, from a similar perspective. Binding to the hydrophobic pocket is a nutlin inhibitor, rendered as sticks and coloured by atom type (carbon, green; nitrogen, blue; oxygen, red; bromine, purple). Note that the hydrophobic (i.e. mostly carbon-containing) groups project into the binding pocket, mimicking the p53 side-chains and that additional groups in the nutlin inhibitor form further interactions. Abbreviation: MDM2, ‘mouse double minute 2’ homologue.

Small-molecule inhibitors of p53 PPIs

Although a p53–MDM2 PPI inhibitor has yet to reach clinical trials, several small-molecule inhibitors with significant potency have been reported in the literature. Most of these molecules are useful experimental tools for research. Chalcones (compounds derived from 1,3-diphenylprop-2-en-1-one) were early-reported molecules with such activity. They interfere with p53–MDM2 interactions by binding to the p53 transactivation domain with relatively low potency: IC50s were in the range 50–250 µm (IC50 is the concentration of inhibitor required to achieve 50% inhibition) (Ref. Reference Stoll53). The natural product chlorofusin is a fungal metabolite containing a peptide segment that was discovered by screening over 53 000 microbial extracts. It inhibits the p53–MDM2 interaction with an IC50 of 4.6 µm; however, it is a large (molecular weight of 1363.7), structurally complex molecule that is less practical for drug development or clinical use than a smaller molecule (Ref. Reference Duncan54). A more drug-like molecule called RITA (‘reactivation of p53 and induction of tumour cell apoptosis’) (Fig. 3a) was similarly discovered by screening a molecular library against cancer cell lines. It too has micromolar potency and was reported to interfere with p53–MDM2 interactions by binding to p53 (Ref. Reference Issaeva55), although more-recent nuclear magnetic resonance (NMR) studies argue that RITA does not in fact block p53–MDM2 binding in vitro (Ref. Reference Grinkevich56).

The nutlins are a series of highly promising molecules for p53–MDM2 inhibition and were discovered by high-throughput screening followed by structure-based optimisation (for an example structure, see Fig. 3b). The nutlins interfere with p53–MDM2 interactions by mimicking the interaction of the three p53 residues within the MDM2 hot spot (Fig. 1), and are highly potent (activities range between 100 and 300 nm). The interaction of nutlin inhibitors with MDM2 has been studied using X-ray crystallography (Fig. 2b): the imidazoline substituents of the nutlin inhibitor occupy similar regions within the binding cleft as the p53 side chains (Refs Reference Vassilev50, Reference Vassilev57).

The BCL2 family

The B-cell lymphoma 2 (BCL2) family of proteins is composed of more than 20 members (Refs Reference Reed58, Reference Zornig59, Reference Reed60). Some members are anti-apoptotic [e.g. BCL2, BCL2L1 (Bcl-xL), BCL2L2 and MCL1], whereas others are pro-apoptotic [e.g. BAX, BAK1, BID and BCL2L11 (BIM)]. The family members cooperate through PPIs to mediate the intrinsic apoptotic pathway (Refs Reference Borner61, Reference van Delft and Huang62). Anti-apoptotic BCL2 family members protect cells from apoptosis by inhibiting the action of pro-apoptotic members (Ref. Reference Wang63). Three-dimensional NMR studies of anti-apoptotic BCL2 proteins revealed the presence of a hydrophobic groove that acts as a binding site for the BH3 peptide domain of pro-apoptotic BCL2 members (Ref. Reference Sattler64). Agents designed to target the binding grooves of anti-apoptotic BCL2 proteins are predicted to induce apoptosis in cancer cells (which overexpress these anti-apoptotic proteins) by antagonising their protective effect.

Over the past ten years, several nonpeptidic small-molecule inhibitors to anti-apoptotic BCL2 proteins have been introduced and some of them are already in clinical trials as single-agent chemotherapy. For example, GX015-070 (Obatoclax) is a small-molecule BCL2 inhibitor developed by Gemin X Biotechnologies Inc. currently being tested in multiple Phase I and Phase II clinical trials against both solid tumours and haematological malignancies (Ref. Reference Fesik66). The development of BCL2-modulatory small molecules as potential anticancer therapeutics has been reviewed in recent years (for examples, see Refs Reference Makin and Dive65, Reference Fesik66), and discussion here is limited to two examples that have pioneered interest in this field and provided a proof-of-principle for inhibition of BCL2 PPIs as a viable therapeutic strategy in cancer.

Small-molecule inhibitors of BCL2 PPIs

ABT-737 (Fig. 3c) was discovered by high-throughput NMR-based screening of a chemical library to identify small molecules that bind to the BH3-binding groove of BCL2L1 (Ref. Reference Fesik66). ABT-737 was found to bind to BCL2, BCL2L1 and BCL2L2 with high affinity (Ki ≤ 1 nm). ABT-737 did not itself induce release of cytochrome c from mitochrondria, but inhibited BCL2 protection at concentrations ≤10 nm by blocking the inhibition of BID-mediated cytochrome c release through binding to the BH3-binding groove of BCL2. This binding disrupted the PPI between BCL2L1 and pro-apoptotic BCL2 proteins.

ABT-737 was found to be active both as a single agent or when used in combination with other chemotherapies or radiation, where it reduced the EC50 (the half maximal effective concentration – the concentration that produces a 50% effect compared with the maximum) of several antitumour agents by 50% or more. As a single agent, it displayed potent cytotoxicity in several lymphomas and small-cell lung cancer (SCLC), where it had an EC50 of 0.13–0.85 µm when used in lymphoma cell lines carrying the t(14:18) chromosomal translocation. As an inhibitor of anti-apoptotic BCL2, ABT-737 exhibited pro-apoptotic activity and induced apoptosis in a concentration-dependent manner in patient-derived chronic lymphocytic B-cell leukaemia cells. In vivo, ABT-737 caused complete regression of established SCLC tumour xenografts as a result of apoptotic cell death. ABT-737 induced apoptosis selectively in cancer cells without affecting normal cells, where a significant increase in caspase-3 was noticed in tumour cells but not in liver, heart or intestinal normal cell lines (Ref. Reference Oltersdorf67).

Peptide inhibitors of BCL2 PPIs

The concept of hydrocarbon-stapled helical peptides that bind with increased affinity to protein targets such as BCL2 has recently been pioneered; these ‘stapled’ peptides provide enhanced protease-resistance and cell permeability compared with their nonstapled counterparts. The chemical ‘stapling’ strategy involves the insertion of α,α-disubstituted non-natural amino acids containing side-chains with an olefinic terminus at two carefully chosen peptidic positions to flank either three or six amino acids. The chemoselective ruthenium-catalysed olefin metathesis reaction was used to ‘staple’ together the two alkene tethers to provide peptides that maintain helical structure, but with enhanced stability. A stapled peptide of the BH3 domain from the pro-apoptotic BID protein was found to specifically activate the apoptotic pathway to kill leukaemia cells, and inhibit the growth of human leukaemia xenografts in vivo (Ref. Reference Walensky68).

XIAP–DIABLO

‘Inhibitors of apoptosis proteins’ (IAPs) are endogenous caspase inhibitors that bind and inhibit caspases-3, -7 and -9. Eight members of the IAP family are known so far (Ref. Reference Schimmer69), of which X-linked inhibitor of apoptosis protein (XIAP) is the best characterised in terms of its inhibitory mechanism. XIAP has three BIR domains (BIR-1–3) and a RING finger that binds and inhibits caspases through specific PPIs (Ref. Reference Deveraux70). The BIR-3 domain of XIAP inhibits caspase-9, while BIR-1 and -2 are more specific for caspases-3 and -7 (Refs Reference Chai71, Reference Huang72, Reference Riedl73).

Endogenous inhibition of the IAP family can occur via DIABLO (also known as Smac), a protein released from the mitochondria in response to apoptotic stimuli, which directly interacts with XIAP (Refs Reference Huang72, Reference Riedl73). This, together with the finding that the levels of different IAP family members were elevated in several tumours, indicated the possibility of targeting the IAP family to induce apoptosis and treat cancer (Refs Reference Du74, Reference Verhagen75, Reference LaCasse76). More-recent studies in this area have focused on the role of TNF-α signalling in mediating apoptotic cell death, since IAP antagonists were found to induce TNF-α-dependent apoptosis (Refs Reference Varfolomeev77, Reference Vince78, Reference Peterson79).

Since small-molecule inhibitors of XIAP have been recently reviewed (Refs Reference Rajapakse80, Reference Hunter, LaCasse and Korneluk81), we will focus here on one relevant example from the group of compounds derived from DIABLO peptides (DIABLO ‘mimetics’) that inhibit the interaction between the BIR-3 domain of XIAP and DIABLO. DIABLO-mimetic XIAP inhibitors act by binding more strongly to the BIR-3 domain of XIAP, inhibiting the caspase binding to XIAP at this site; they thus antagonise the inhibitory effect of XIAP and induce apoptosis. For example, a series of tripeptides that bind to the XIAP BIR-3 domain with nanomolar affinities have been reported (Ref. Reference Oost82). The most active compound of the series (Fig. 3d) showed a wide range of potencies against a diverse panel of human cancer cell lines, with the highest activity (nanomolar EC50) against the human breast cancer cell lines BT-549 and MDA-MB-231, the melanoma cell line SK-MEL-5, and the acute promyelocytic leukaemia cell line HL-60. Interestingly, there was no correlation between the potency of this compound and the expression of XIAP. Measuring caspase-3 activation in MDA-MB-231 cells treated with the tripeptide showed an up to eightfold increase in caspase-3 activity in a dose-dependent manner. Apoptosis was induced by the tripeptide in the absence of an identified apoptotic stimulus. In vivo testing (MDA-MB-231 flank breast cancer xenograft model) showed that a 20 mg/kg/day dose of the compound inhibited tumour growth rate by 50–60%.

SH2 domains

Many membrane-associated proteins are coupled to kinases, and their activation is essential for cellular signalling. Src-homology 2 (SH2) domains are noncatalytic, highly conserved domains of approximately 100 amino acids that are an essential feature in many cellular signalling proteins: they recognise hot spots that contain a phosphorylated tyrosine residue and adjacent residues, facilitating the assembly of transient protein pairs that subsequently participate in signal transduction. One of the best-studied SH2 domains is that of growth factor receptor bound protein 2 (GRB2). GRB2 links signalling from receptor proteins to the RAS signal transduction pathway, ultimately controlling the mitogen-activated protein (MAP) kinase cascade. This important regulatory pathway is vital for cell division and an excellent target for novel cancer therapies (Refs Reference Feller and Lewitzky83, Reference Shakespeare84). Increased GRB2 SH2 signalling is seen in cancer, and so inhibiting the PPI between GRB2 and tyrosine-phosphorylated signalling proteins will have antiproliferative effects. Several tyrosine-containing, peptide-based inhibitors have been reported but have stability and delivery problems. However, compounds containing a phosphorylated tyrosine residue were shown to be optimal for potent PPI inhibition. The development of nonpeptide small-molecule inhibitors has been reported; the phosphorylated compound in Figure 3e represents a typical member of a series of compounds designed to inhibit Grb2 SH2 PPIs (Ref. Reference Furet85).

Figure 3

Chemical structures of protein–protein interaction (PPI) inhibitors. (a) RITA is a simple and potentially drug-like thiophene-furan-thiophene derivative. (b) The nutlin series of MDM2–p53 PPI inhibitors is exemplified here by nutlin 3; the dihydroimidazole core (i.e. the central five-membered nitrogen-containing ring) is fixed and variation is introduced within the substituents – for example, nutlin 3 is a chlorine analogue. (c) ABT-737 is a prototypical compound of the PPI inhibitors that target anti-apoptotic BCL2 family members. (d) This inhibitor of the anti-apoptoic XIAP is a tripeptide composed of alanine, an unnatural cyclohexyl amino acid and proline; the N-terminus is modified with a methyl group and the C-terminus is modified with a bulky tetrahydronapthalene derivative. (e–g) The structures of these GRB2 SH2 inhibitors clearly show the β-turn-like nature of these compounds. Note the addition of a restraining olefin (double bond) bridge in (g). (h) This Raf–Raf inhibitor is derived from the nonsteroidal anti-inflammatory drug sulindac (i); both structures are provided to show the comparison. (j) FJ9 is a small-molecule indole-based inhibitor of the interaction between the receptor FRZ-7 and the PDZ domain of DVL, thereby inhibiting oncogenic WNT signalling. (k) ICG-001 is a small molecule that downregulates β-catenin–TCF signalling by specifically binding to the coactivator CBP. Abbreviations: BCL2, B-cell lymphoma 2; CBP, cyclic AMP response element-binding protein; DVL, dishevelled; FRZ, frizzled; GRB2, growth factor receptor bound protein 2; MDM2, ‘mouse double minute 2’ homologue; SH2, Src-homology 2; TCF, T-cell factor; WNT, ‘wingless-type MMTV integration site family’; XIAP, X-linked inhibitor of apoptosis protein.

However, the phosphotyrosine group of this compound led to practical difficulties, and a bioisosteric replacement of the phosphate group was investigated to improve the stability/delivery profile of this class of compounds. When the phosphate group was replaced with a dicarboxylic acid (Fig. 3f), reasonable activity was retained, suggesting that a similar number of hydrogen bonds with the target protein were formed. Related compounds within this series were shown to penetrate cells and inhibit downstream MAP kinase activation, establishing good rationale for these compounds as potential cancer therapeutics (Ref. Reference Gao86). Their potency is believed to be related to the ability to mimic secondary structure elements in the PPI. Several PPI hot spots have been found in β-turns (Ref. Reference de Vega, Martin-Martinez and Gonzalez-Muniz87), and the Grb2 SH2 domain recognises a phosphorylated tyrosine hot spot that adopts the conformation of a β-turn. The cyclohexyl amino acid residue in this inhibitor series (Fig. 3e–g) helps to restrain the molecule in a similar, β-turn-like conformation (Ref. Reference Rahuel88). The concept has been extended with the addition of an olefin (i.e. carbon–carbon double bond) link (Fig. 3g) to further stabilise the β-turn conformation by incorporating it into a macrocyclic ring. This compound produced a 100-fold enhancement in terms of binding to the protein target, but no cellular activity was observed (Refs Reference Gao89, Reference Wei90).

The RAS–RAF interaction

RAS is a GTPase important to several signal transduction cascades. Normally, RAS binds to RAF, a kinase that subsequently activates the MAP kinase signalling pathway that is an important proliferative pathway in the cell. RAS is mutated and hence deregulated in approximately 30% of human cancers (Ref. Reference Walker and Olson91). Inhibitors of the RAS–RAF interaction are thus likely to block cell division and be of interest in anticancer therapeutics. Small-molecule inhibitors have been identified – notably a furano-indene derivative (Fig. 3 h). This compound was derived by rational medicinal chemistry from a metabolite of the nonsteroidal anti-inflammatory drug (NSAID) sulindac (Fig. 3i), following the observation that it had anticancer properties associated with the RAS pathway. Although inhibition of the RAS–RAF interaction by this compound was modest (IC50 = 30 µm), this simple small molecule provides a significant advance over previous attempts to disrupt the interaction (Refs Reference Yin and Hamilton6, Reference Waldmann92). Related investigations reported that a 28 amino acid peptide comprising the hot spot region of RAS inhibited the RAS–RAF interaction with an improved IC50 of 3 µm. However, although an order of magnitude less active, the nonpeptide, sulindac-derived inhibitor (Fig. 3 h) remains a more drug-like molecule better suited for further development (Ref. Reference Warne, Viciana and Downward93).

WNT pathway PPI targets

Multiple mechanisms activate the WNT signalling pathway in cancer. For example, in response to WNT ligands, inactivation of GSK-3β leads to inhibition of degradation of the β-catenin-turnover complex, leading to inappropriate transcriptional activation. Inappropriate activation of the WNT pathway is associated with stimulation of proliferation and prevention of apoptosis in a number of human cancers (Ref. Reference Barker and Clevers94), and the discovery of inhibitors of WNT pathway PPIs is a growing area of interest. Two PPI targets within the WNT signalling pathway are considered here, where small-molecule inhibitors with antitumour properties have been described.

Frizzled–Dishevelled

The Frizzled (FRZ)–Dishevelled (DVL) interaction is of crucial importance in relaying WNT signalling to the β-catenin-turnover complex. Different WNT signalling molecules use different subtypes of FRZ receptor. Particularly promising with respect to cancer drug design is the observation that WNT signalling via FRZ-7 is associated with oncogenesis, and FRZ-7 is reported to be overexpressed in both tumour cell lines and tumour tissues. FRZ-7 interacts directly with the PDZ domain of the DVL family of proteins (overexpressed in a variety of cancer cells); hence specific interference of the FRZ-7–DVL PPI could represent a promising target for cancer therapy. A small-molecule inhibitor (FJ9; Fig. 3j) of the interaction between FRZ-7 and the PDZ domain of DVL has been identified, which was able to suppress β-catenin-dependent tumour cell growth (Ref. Reference Fujii95). Significant antitumour effects of FJ9 were observed, including induction of apoptosis in human cancer cell lines and in vivo inhibition of tumour growth of mouse xenograft (H460) models. FJ9 is one of the first small-molecule inhibitors to show antitumour activity through selective inhibition of PPIs mediated by PDZ domains.

β-Catenin–CBP

Mutations in one of the components of the β-catenin complex regulated by WNT signalling, the adenomatous polyposis coli (APC) gene product, occur in most colon cancers and lead to inappropriate activation of β-catenin-responsive genes in colon cancer. β-catenin regulates transcription in a complex with T-cell factor (TCF), whose formation involves the coactivator cyclic AMP response element-binding protein (CBP). ICG-001 (Fig. 3 k) is a small molecule that downregulates β-catenin–TCF signalling by specifically binding to CBP. ICG-001 was also found to selectively induce apoptosis in transformed (but not normal) colon cells, and have in vitro activity in colon carcinoma cells (growth reduction and induction of apoptosis). Notably, ICG-001 was also found to have efficacy in the Min mouse and nude mouse xenograft models of colon cancer (Ref. Reference Emami96).

Summary and conclusions

The study of PPIs has gathered considerable pace in recent years. Initial experimental difficulties have been overcome and practical applications, for example drug design, can now be considered. Drug design has traditionally been directed towards well-characterised targets, such as enzyme active sites; however, a greater understanding of PPIs and how to target them is opening up new horizons in the future of drug development. A large number of diseases could potentially be treated by PPI inhibition due to the myriad PPIs that occur within the cell. Progress in the cancer arena has been particularly strong. In most cases PPI inhibition has provided an alternative approach to exploit an existing, known target. However, it seems highly likely that entirely novel PPI-unique therapeutic targets may soon emerge. Many of the early successes targeting PPIs involved the use of peptide inhibitors, but peptides (and large molecules, such as chlorofusin, with molecular weights >500) have poor bioavailability and are generally unsuitable as drug molecules. Although it initially appeared unlikely that small molecules could successfully inhibit PPIs the findings reviewed in this article clearly demonstrate that low molecular weight compounds can successfully block PPIs and thus represent a new emerging class of therapeutic agents.

Acknowledgements and funding

The authors thank the Algerian Government for funding, in the form of a PhD scholarship, to G.B. We also thank the peer reviewers for their insightful comments and suggestions.

References

References

1Toogood, P.L. (2002) Inhibition of protein-protein association by small molecules: approaches and progress. J Med Chem 45, 1543-1558Google Scholar
2Sharma, S.K., Ramsey, T.M., and Bair, K.W. (2002) Protein-protein interactions: lessons learned. Curr Med Chem: Anticancer Agents 2, 311-330Google Scholar
3Arkin, M.R., and Wells, J.A. (2004) Small-molecule inhibitors of protein-protein interactions: Progressing towards the dream. Nat Rev Drug Discov 3, 301-317CrossRefGoogle ScholarPubMed
4Arkin, M. (2005) Protein-protein interactions and cancer: small molecules going in for the kill. Curr Opin Chem Biol 9, 317-324Google Scholar
5Fry, D.C. and Vassilev, L.T. (2005) Targeting protein-protein interactions for cancer therapy. J Mol Med 83, 955-963CrossRefGoogle ScholarPubMed
6Yin, H. and Hamilton, A.D. (2005) Strategies for targeting protein-protein interactions with synthetic agents. Angew Chem Int Ed Engl 44, 4130-4163Google Scholar
7Chene, P. (2006) Drugs targeting protein-protein interactions. Chem Med Chem 1, 400-411CrossRefGoogle ScholarPubMed
8Lindsley, C.W. (2007) Small molecule inhibition of protein-protein interaction: An emerging paradigm in drug design. Curr Top Med Chem 7, 921-921Google Scholar
9Wells, J.A. and McClendon, C.L. (2007) Reaching for high-hanging fruit in drug discovery at protein-protein interfaces. Nature 450, 1001-1009Google Scholar
10Collins, I. and Workman, P. (2006) New approaches to molecular cancer therapeutics. Nat Chem Biol 2, 689-700Google Scholar
11Bogan, A.A. and Thorn, K.S. (1998) Anatomy of hot spots in protein interfaces. J Mol Biol 280, 1-9CrossRefGoogle ScholarPubMed
12Rudolph, J. (2007) Inhibiting transient protein-protein interactions: lessons from the Cdc25 protein tyrosine phosphatases. Nat Rev Cancer 7, 202-211CrossRefGoogle ScholarPubMed
13Zhong, S.J., Macias, A.T. and MacKerell, A.D. (2007) Computational identification of inhibitors of protein-protein interactions. Curr Top Med Chem 7, 63-82CrossRefGoogle ScholarPubMed
14Sundberg, E.J. and Mariuzza, R.A. (2000) Luxury accommodations: the expanding role of structural plasticity in protein-protein interactions. Structure 8, R137-R142Google Scholar
15Lo Conte, L., Chothia, C. and Janin, J. (1999) The atomic structure of protein-protein recognition sites. J Mol Biol 285, 2177-2198Google Scholar
16DeLano, W.L. (2002) Unraveling hot spots in binding interfaces: progress and challenges. Curr Opin Struct Biol 12, 14-20Google Scholar
17Sidhu, S.S., Fairbrother, W.J. and Deshayes, K. (2003) Exploring protein-protein interactions with phage display. Chembiochem 4, 14-25CrossRefGoogle ScholarPubMed
18Stites, W.E. (1997) Protein-protein interactions: Interface structure, binding thermodynamics, and mutational analysis. Chem Rev 97, 1233-1250CrossRefGoogle Scholar
19Nahta, R. and Esteva, F.J. (2006) Herceptin: mechanisms of action and resistance. Cancer Lett 232, 123-138CrossRefGoogle ScholarPubMed
20Downing, K.H. (2000) Structural basis for the interaction of tubulin with proteins and drugs that affect microtubules dynamics. Annu Rev Cell Dev Biol 16, 89-111CrossRefGoogle Scholar
21Jordan, M.A. (2002) Mechanism of action of antitumour drugs that interact with microtubules and tubulin. Curr Med Chem Anticancer Agents 2, 1-17Google Scholar
22Jordan, A. et al. (1998) Tubulin as a target for anticancer drugs: agents which interact with the mitotic spindle. Med Res Rev 18, 259-296Google Scholar
23Kavallaris, M., Verrills, N.M. and Hill, B.T. (2001) Anticancer therapy with novel tubulin interacting drugs. Drug Resist Updat 4, 392-401Google Scholar
24Nogales, E. (2001) Structural insights into microtubules function. Annu Rev Biophys Biomol Struct 30, 397-420Google Scholar
25Bollag, D.M. et al. (1995) Epothilones, a new class of microtubule-stabilizing agents with taxol-like mechanism of action. Cancer Res 55, 2325-2333Google ScholarPubMed
26Long, B.H. et al. (1998) Eleutherobin, a novel cytotoxic agent that induces tubulin polymerization, is similar to paclitaxel (Taxol®). Cancer Res 58, 1111-1115Google Scholar
27terHaar, E. et al. (1996) Discodermolide, a cytotoxic marine agent that stabilizes microtubules more potently than taxol. Biochemistry 35, 243-250Google Scholar
28Bai, R.L. et al. (1996) Identification of cysteine 354 of beta-tubulin as part of the binding site for the A ring of colchicine. J Biol Chem 271, 12639-12645CrossRefGoogle ScholarPubMed
29Uppuluri, S. et al. (1993) Localisation of the colchicine binding site of tubulin. Proc Natl Acad Sci U S A 90, 11539-11597Google Scholar
30Downing, K.H. and Nogales, E. (1998) New insights into microtubule structure and function from the atomic model of tubulin. Eur Biophys J Biophys Lett 27, 431-436Google Scholar
31Erickson, H.P. (1975) Negatively stained vinblastine aggregates. Proc Natl Acad Sci U S A 253, 51-52CrossRefGoogle ScholarPubMed
32Rai, S.S. and Wolff, J. (1996) Localization of the vinblastine-binding site on tubulin. J Biol Chem 271, 14707-14711Google Scholar
33Gupta, S. and Bhttacharyya, B. (2003) Antimicrotubule drugs binding to vinca domain of tubulin. Mol Cell Biochem 235, 41-47CrossRefGoogle Scholar
34Levin, A.J. (1997) p53, the cellular gatekeeper for growth and division. Cell 88, 323-331CrossRefGoogle Scholar
35Hollstein, M. et al. (1994) Database of p53 gene somatic mutations in human tumors and cell lines. Nucleic Acids Res 22, 3551-3555Google ScholarPubMed
36May, P. and May, E. (1999) Twenty years of p53 research: structural and functional aspects of the p53 protein. Oncogene 18, 7621-7636Google Scholar
37Bossi, G. and Sacchi, A. (2007) Restoration of wild-type p53 function in human cancer: relevance for tumour therapy. Head Neck 29, 272-284CrossRefGoogle Scholar
38Gudkov, A.V. and Kmarova, E.A. (2007) Dangerous habits of a security guard: The two faces of p53 as drug target. Hum Mol Genet 16, R62-R72CrossRefGoogle ScholarPubMed
39Lu, X. and Lane, D.P. (1993) Differential induction of transcriptionally active p53 following UV or ionizing radiation: defects in chromosome instability syndromes. Cell 75, 765-778CrossRefGoogle ScholarPubMed
40Clarke, A.R. et al. (1993) Thymocytes apoptosis induced by p53 dependent and independent pathways. Nature 362, 849-852CrossRefGoogle ScholarPubMed
41Yonish-Rouach, E. et al. (1994) Induction of apoptosis by transiently transfected metabolically stable wt p53 in transformed cell lines. Cell Death Differ 1, 39-47Google Scholar
42Wynford-Thomas, D. (1999) Cellular senescence and cancer. J Pathol 187, 100-1113.0.CO;2-T>CrossRefGoogle ScholarPubMed
43Wang, X.W. (1994) Hepatitis B virus X protein inhibits p53 sequence specific DNA binding, transcriptional activity, and association with transcription factor ERCC3. Proc Natl Acad Sci U S A 91, 2230-2234Google Scholar
44Oren, M. (2003) Decision making by p53: life, death and cancer. Cell Death Differ 10, 431-442Google Scholar
45Marchenko, N.D., Zaika, A. and Moll, U.M. (2000) Death signal induced localization of p53 protein to mitochondria, a potential role in apoptotic signaling. J Biol Chem 275, 16202-16212Google Scholar
46Freedman, D.A., Wu, L. and Levine, A.J. (1999) Functions of the MDM2 oncoprotein. Cell Mol Life Sci 55, 96-107Google Scholar
47Wu, X. et al. (1993) The p53-mdm-2 autoregulatory feedback loop. Gene Dev 7, 1126-1132Google Scholar
48Fujiwara, T. et al. (1994) Therapeutic effect of a retroviral wild-type p53 expression vector in an orthotopic lung cancer model. J Natl Cancer Inst 86, 1458-1462Google Scholar
49Scardigli, R. et al. (1997) Expression of exogenous wt-p53 does not affect normal hematopoiesis: implications for bone marrow purging. Gene Ther 4, 1371-1378Google Scholar
50Vassilev, L.T. (2005) p53 activation by small molecules: application in oncology. J Med Chem 48, 4491-4499Google Scholar
51Kussie, P.H. et al. (1996) Crystal structure of the MDM2 oncoprotein bound to the transactivation domain of the p53 tumor suppressor. Science 274, 948-953Google Scholar
52Lin, J. et al. (1994) Functions of the p53 protein in growth regulation and tumor suppression. Cold Spring Harb Symp Quant Biol 59, 215-223Google Scholar
53Stoll, R. et al. (2001) Chalcone derivatives antagonize interactions between the human oncoprotein MDM2 and p53. Biochemistry 40, 336-344CrossRefGoogle ScholarPubMed
54Duncan, S.J. et al. (2001) Isolation and structure elucidation of chlorofusin, a novel p53-MDM2 antagonist from a Fusarium sp. J Am Chem Soc 123, 554-560Google Scholar
55Issaeva, N. et al. (2004) Small molecule RITA binds to p53, blocks p53 − HDM-2 interaction and activates p53 function in tumors. Nat Med 10, 1321-1328Google Scholar
56Grinkevich, V. et al. (2005) NMR indicates that the small molecule RITA does not block p53-MDM2 binding in vitro. Nat Med 11, 1135-1136Google Scholar
57Vassilev, L.T. et al. (2004) In vivo activation of the p53 pathway by small-molecule antagonists of MDM2. Science 303, 844-848CrossRefGoogle ScholarPubMed
58Reed, J.C. (1998) Bcl-2 family proteins. Oncogene 17, 3225-3236Google Scholar
59Zornig, M. et al. (2001) Apoptosis regulators and their role in tumorigenesis. Biochim Biophys Acta Rev Cancer 1551, F1-F37Google Scholar
60Reed, J.C. (2002) Apoptosis-based therapies. Nat Rev Drug Discov 1, 111-121Google Scholar
61Borner, C. (2003) The Bcl-2 protein family: sensors and checkpoints for life-or-death decisions. Mol Immunol 39, 615-647Google Scholar
62van Delft, M.F. and Huang, D.C.S. (2006) How the Bcl-2 family of proteins interacts to regulate apoptosis. Cell Res 16, 203-213Google Scholar
63Wang, J.L. et al. (2000) Cell permeable Bcl-2 binding peptides: A chemical approach to apoptosis induction in tumour cells. Cancer Res 60, 1498-1502Google Scholar
64Sattler, M. et al. (1997) Structure of Bcl-x(L) Bak peptide complex: recognition between regulators of apoptosis. Science 275, 983-986Google Scholar
65Makin, G. and Dive, C. (2003) Recent advances in understanding apoptosis: new therapeutic opportunities in cancer chemotherapy. Trends Mol Med 9, 251-255Google Scholar
66Fesik, S.W. (2005) Promoting apoptosis as a strategy for cancer drug discovery. Nat Rev Cancer 5, 876-885CrossRefGoogle ScholarPubMed
67Oltersdorf, T. et al. (2005) An inhibitor of Bcl-2 family proteins induces regression of solid tumours. Nature 435, 677-681Google Scholar
68Walensky, L.D. et al. (2004) Activation of apoptosis in vivo by a hydrocarbon-stapled BH3 helix. Science 305, 1466-1470CrossRefGoogle ScholarPubMed
69Schimmer, A.D. (2004) Inhibitor of apoptosis proteins: translating basic knowledge into clinical practice. Cancer Res 64, 7183-7190CrossRefGoogle ScholarPubMed
70Deveraux, Q.L. et al. (1998) IAPs block apoptotic events induced by caspase-8 and cytochrome c by direct inhibition of distinct caspases. EMBO J 17, 2215-2223Google Scholar
71Chai, J. et al. (2001) Structural basis of caspase-7 inhibition by XIAP. Cell 104, 769-780Google Scholar
72Huang, Y.H. et al. (2001) Structural basis of caspase inhibition by XIAP: differential roles of the linker versus the BIR domain. Cell 104, 781-790Google Scholar
73Riedl, S.J. et al. (2001) Structural basis for the inhibition of caspase-3 by XIAP. Cell 104, 791-800Google Scholar
74Du, C.Y. et al. (2000) Smac, a mitochondrial protein that promotes cytochrome c-dependent caspase activation by eliminating IAP inhibition. Cell 102, 33-42CrossRefGoogle ScholarPubMed
75Verhagen, A.M. et al. (2000) Identification of DIABLO, a mammalian protein that promotes apoptosis by binding to and antagonizing IAP proteins. Cell 102, 43-53Google Scholar
76LaCasse, E.C. et al. (1998) The inhibitors of apoptosis (IAPs) and their emerging role in cancer. Oncogene 17, 3247-3259Google Scholar
77Varfolomeev, E. et al. (2007) IAP antagonists induce autoubiquitination of c-IAPs, NF-κB activation, and TNFα-dependent apoptosis. Cell 131, 669-681Google Scholar
78Vince, J.E. et al. (2007) IAP antagonists target cIAP1 to induce TNFα-dependent apoptosis. Cell 131, 682-693Google Scholar
79Peterson, S.L. et al. (2007) Autocrine TNFα signalling renders human cancer cells susceptible to Smac-mimetic-induced apoptosis. Cancer Cell 12, 445-456Google Scholar
80Rajapakse, H.A. (2007) Small molecule inhibitors of the XIAP protein-protein interaction. Curr Top Med Chem 7, 966-971Google Scholar
81Hunter, A.M., LaCasse, E.C. and Korneluk, R.G. (2007) The inhibitors of apoptosis (IAPs) as cancer targets. Apoptosis 12, 1543-1568Google Scholar
82Oost, T.K. et al. (2004) Discovery of potent antagonists of the antiapoptotic protein XIAP for the treatment of cancer. J Med Chem 47, 4417-4426Google Scholar
83Feller, S.M and Lewitzky, M. (2006) Potential disease targets for drugs that disrupt protein-protein interactions of Grb2 and Crk family adaptors. Curr Pharm Design 12, 529-548Google Scholar
84Shakespeare, W.C. (2001) SH2 domain inhibition: a problem solved? Curr Opin Chem Biol 5, 409-415Google Scholar
85Furet, P. et al. (1998) Structure based design and synthesis of high affinity tripeptide ligands of the Grb2-SH2 domain. J Med Chem 41, 3442-3449Google Scholar
86Gao, Y. et al. (2000) Inhibition of Grb2 SH2 domain binding by non-phosphate containing ligands. 2. 4-(2-Malonyl)phenylalanine as a potent phosphotyrosyl mimetic. J Med Chem 43, 911-920Google Scholar
87de Vega, M.J.P., Martin-Martinez, M. and Gonzalez-Muniz, R. (2007) Modulation of protein-protein interactions by stabilizing/mimicking protein secondary structure elements. Curr Top Med Chem 7, 33-62Google Scholar
88Rahuel, J. et al. (1996) Structural basis for specificity of Grb2-SH2 revealed by a novel ligand binding mode. Nat Struct Biol 3, 586-589Google Scholar
89Gao, Y. et al. (2001) Macrocyclization in the design of a conformationally constrained Grb2-SH2 domain inhibitor. Bioorg Med Chem Lett 11, 1889-1892CrossRefGoogle ScholarPubMed
90Wei, C. et al. (2003) Macrocyclization in the design of Grb2-SH2 domain binding ligands exhibiting high potency in whole cell systems. J Med Chem 46, 244-254Google Scholar
91Walker, K. and Olson, M.F. (2005) Targeting Ras and Rho GTPases as opportunities for cancer therapeutics. Curr Opin Genet Dev 15, 62-68Google Scholar
92Waldmann, H. et al. (2004) Sulindac-derived Ras pathway inhibitors target the Ras-Raf interaction and downstream effectors in the Ras pathway. Angew Chem Int Ed Engl 43, 454-458Google Scholar
93Warne, P.H., Viciana, P.R. and Downward, J. (1993) Direct interaction of Ras and the amino-terminal region of Raf-1 in-vitro. Nature 364, 352-355CrossRefGoogle ScholarPubMed
94Barker, N. and Clevers, H. (2006) Mining the Wnt pathway for cancer therapeutics. Nat Rev Drug Discov 5, 997-1014Google Scholar
95Fujii, N. et al. (2007) An antagonist of dishevelled protein-protein interaction suppresses beta-catenin-dependent tumor cell growth. Cancer Res 67, 573-579Google Scholar
96Emami, K.H. et al. (2004) A small molecule inhibitor of beta-catenin/cyclic AMP response element-binding protein transcription. Proc Natl Acad Sci U S A 101, 12682-12687Google Scholar

Further reading, resources and contacts

For general information on all aspects of cancer (science, treatment and research) visit:

http://www.cancerresearchuk.org/Google Scholar
http://www.cancerhelp.org.uk/Google Scholar
http://www.rcsb.org/pdbGoogle Scholar
http://en.wikibooks.org/wiki/Proteomics/Protein_-_Protein_InteractionsGoogle Scholar
http://www.geminx.com (Phase 2 clinical trial of GX015-070 small-molecule BCL 2 inhibitor)Google Scholar
http://www.roche.com/ (Developing the p53–MDM2 inhibitors; detailed presentation at http://www.roche.com/med_mb200605lv.pdf)Google Scholar
Figure 0

Figure 1A conceptual diagram of a protein–protein interaction (PPI) inhibitor targeting a PPI (p53–MDM2) hot spot. In this diagram, the inhibition of the p53–MDM2 interaction by a nutlin inhibitor is used as an example to illustrate the generic concept of the action of all PPI inhibitors discussed in this article, since it is one of the best-characterised examples of PPI inhibition. MDM2 normally negatively regulates the antitumour protein p53 by inducing its proteasomal degradation; however, in response to cellular stress, particularly DNA damage, p53 levels rise as a result of increased p53 transcription and reduced MDM2-mediated degradation (not shown). The nutlin inhibitors are similarly thought to increase p53 levels, by blocking MDM2-mediated degradation. The PPI of p53 and MDM2 is mediated by the side chains of three p53 amino acids (Leu, Trp and Phe) that are recognised by MDM2. The nutlin inhibitors mimic the binding of these amino acid side chains and preferentially bind to MDM2. This physically blocks the interaction of the two proteins, thus releasing p53 that is free to exert a beneficial antitumour effect. This schematic illustration is derived from the X-ray data presented in Fig. 2a, and comparison of Figs 1 and 2a should be instructive. Abbreviation: MDM2, ‘mouse double minute 2’ homologue.

Figure 1

Figure 2Structural analysis of the interactions of MDM2 with p53 and with a nutlin inhibitor. (a) X-ray crystallographic models of the p53–MDM2 interaction (Ref. 51). A fragment of MDM2 is shown rendered as a solvent-accessible solid surface and coloured by atom type (carbon, grey; oxygen, red; nitrogen, blue; sulphur, yellow). The p53 fragment is rendered as a green ribbon showing only the side-chains of Leu26, Trp23 and Phe19 rendered as sticks and coloured by atom type (carbon, green; nitrogen, blue). These three nonpolar side-chains project into a hydrophobic (i.e. predominantly carbon-containing) pocket in the surface, thus stabilising the interaction between p53 and MDM2. (b) X-ray crystal structure of the MDM2–nutlin-inhibitor complex (Ref. 57). The MDM2 fragment is again shown as a solvent-accessible solid surface, from a similar perspective. Binding to the hydrophobic pocket is a nutlin inhibitor, rendered as sticks and coloured by atom type (carbon, green; nitrogen, blue; oxygen, red; bromine, purple). Note that the hydrophobic (i.e. mostly carbon-containing) groups project into the binding pocket, mimicking the p53 side-chains and that additional groups in the nutlin inhibitor form further interactions. Abbreviation: MDM2, ‘mouse double minute 2’ homologue.

Figure 2

Figure 3Chemical structures of protein–protein interaction (PPI) inhibitors. (a) RITA is a simple and potentially drug-like thiophene-furan-thiophene derivative. (b) The nutlin series of MDM2–p53 PPI inhibitors is exemplified here by nutlin 3; the dihydroimidazole core (i.e. the central five-membered nitrogen-containing ring) is fixed and variation is introduced within the substituents – for example, nutlin 3 is a chlorine analogue. (c) ABT-737 is a prototypical compound of the PPI inhibitors that target anti-apoptotic BCL2 family members. (d) This inhibitor of the anti-apoptoic XIAP is a tripeptide composed of alanine, an unnatural cyclohexyl amino acid and proline; the N-terminus is modified with a methyl group and the C-terminus is modified with a bulky tetrahydronapthalene derivative. (e–g) The structures of these GRB2 SH2 inhibitors clearly show the β-turn-like nature of these compounds. Note the addition of a restraining olefin (double bond) bridge in (g). (h) This Raf–Raf inhibitor is derived from the nonsteroidal anti-inflammatory drug sulindac (i); both structures are provided to show the comparison. (j) FJ9 is a small-molecule indole-based inhibitor of the interaction between the receptor FRZ-7 and the PDZ domain of DVL, thereby inhibiting oncogenic WNT signalling. (k) ICG-001 is a small molecule that downregulates β-catenin–TCF signalling by specifically binding to the coactivator CBP. Abbreviations: BCL2, B-cell lymphoma 2; CBP, cyclic AMP response element-binding protein; DVL, dishevelled; FRZ, frizzled; GRB2, growth factor receptor bound protein 2; MDM2, ‘mouse double minute 2’ homologue; SH2, Src-homology 2; TCF, T-cell factor; WNT, ‘wingless-type MMTV integration site family’; XIAP, X-linked inhibitor of apoptosis protein.