ReviewsREVIEW

Waste disposal—An attractive strategy for cancer therapy

See allHide authors and affiliations

Science  17 Mar 2017:
Vol. 355, Issue 6330, pp. 1163-1167
DOI: 10.1126/science.aam7340

Abstract

Targeted therapies for cancer are typically small molecules or monoclonal antibodies that act by inhibiting the activity of specific proteins that drive tumor growth. Although many of these drugs are effective in cancer patients, the response is often not durable because tumor cells develop resistance to the drugs. Another limitation of this strategy is that not all oncogenic driver proteins are “druggable” enzymes or receptors with activities that can be inhibited. Here we describe an alternative approach to targeted therapy that is based on co-opting the cellular quality-control machinery—the ubiquitin-proteasome system—to remove specific cancer-causing proteins from the cell. We first discuss examples of existing cancer drugs that work by degrading specific proteins and then review recent progress in the rational design and preclinical testing of small molecules that induce selective degradation of specific target proteins.

Cancer continues to be a leading global health problem; it has been estimated that by 2025 there will be nearly 20 million new cancer cases diagnosed each year (1). Notable progress is being made in cancer drug development, particularly in the areas of immunotherapy and targeted therapy,but the enormity of the cancer problem requires a variety of therapeutic strategies. Many targeted therapies for cancer are small molecules or monoclonal antibodies that inhibit the activity of proteins driving tumor growth. Tumor cells often develop resistance to these drugs through overexpression of the target protein and/or through the acquisition of new mutations in the target protein that allow it to escape the inhibitory effect of the drug.

Over the past 15 years, researchers have begun to explore an alternative therapeutic approach that aims to control protein function by regulating protein expression levels rather than activities. These efforts to harness controlled proteostasis as a therapeutic strategy evolved from the discovery that proteasome inhibitors that block protein degradation have anticancer activity. Carfilzomib and bortezomib are two examples of such proteasome inhibitors approved by the U.S. Food and Drug Administration (FDA) for the treatment of multiple myeloma (MM) (2). Investigators have explored other ways in which the ubiquitin-proteasome system (UPS) can be manipulated to stabilize or promote the degradation of disease-causing proteins (Table 1). For example, efforts have been made to disrupt the interaction between proteins and the ubiquitin E3 ligases responsible for their degradation (3). The interaction between the tumor suppressor p53 and its ubiquitin E3 ligase, MDM2 (mouse double minute 2 homolog), has been an attractive oncology target: A potent inhibitor of this interaction (RG7112) was found to kill wild-type p53-expressing cancer cells and to inhibit tumor growth in preclinical models of cancer. A reduction in the levels of oncoproteins can also be achieved by inhibiting enzymes that function to stabilize these proteins (3). For example, inhibition of USP7 (the ubiquitin-specific protease 7), a deubiquitinating enzyme (DUB) that deubiquitinates and stabilizes MDM2, reduces MDM2 levels and consequently increases p53 levels. To that end, the USP7 inhibitor P5091 has shown promising antitumor activity in MM xenograft models (4).

Table 1 Representative cancer drugs and drug candidates that work by a controlled proteostasis mechanism.

This list is not intended to be comprehensive but rather is an illustrative selection of the many compounds that work by this mechanism and are currently in the cancer drug development pipeline. ERα, estrogen receptor alpha; PML-RARα, promyelocytic leukemia–retinoic acid receptor alpha; CRBN, cereblon; USP7, ubiquitin-specific protease 7; MDM2, mouse double minute 2 homolog; HER3, human epidermal growth factor receptor 3; BET, bromodomain and extra-terminal; BRD4, bromodomain-containing protein 4; APL, acute promyelocytic leukemia; AML, acute myeloid leukemia; CK1α, casein kinase 1α; SNIPER, specific and nongenetic IAP-dependent protein eraser; PROTAC, proteolysis targeting chimera; IMiD, immunomodulatory drug; MM, multiple myeloma; DUB, deubiquitinating enzyme.

View this table:

More recently, interest has focused on directly using the UPS to induce degradation of specific target proteins, especially proteins for which there are no established DUBs or E3 ligases. These newer strategies involve the pharmacological hijacking of the cellular quality-control system to posttranslationally eliminate disease-causing proteins (Table 1). In this Review, we discuss the initial applications of this concept of targeted protein degradation to achieve controlled proteostasis and the strategies employed to generate protein degraders. We highlight the progress achieved to date, as well as the some of the challenges inherent in this approach.

The limitations of “occupancy-driven” pharmacology

In this postgenomic era, a better understanding of the molecular drivers of cancer has led to the development of several successful cancer therapies that inhibit the activity of enzymes such as protein kinases (e.g., imatinib, erlotinib, and palbociclib), histone deacetylases (e.g., belinostat), and poly(ADP-ribose) polymerase (e.g., olaparib). These small-molecule inhibitors generally work by occupying a binding pocket or active site, resulting in the loss of protein function. However, given that most enzyme inhibitors bind noncovalently (and thus reversibly), high drug concentrations must be maintained to ensure active-site occupancy and to sustain the clinical benefit (5). This is understood as an “occupancy-driven” pharmacological paradigm: one that necessitates that the binding pocket remain occupied to maintain effectiveness of the small molecule (Fig. 1). Unfortunately, achieving and maintaining high systemic drug levels is one of the major challenges in drug development. Moreover, these high dosages can lead to undesirable off-target effects. In the case of kinases [which all bind ATP (adenosine triphosphate)], active-site inhibition by ATP analogs is challenging because many kinases show structural similarity. However, this promiscuity can sometimes be advantageous, especially when targeting cancers that rely on the activation of multiple cellular signaling pathways.

Fig. 1 Occupancy-driven pharmacological paradigm.

(A) In the absence of inhibitor, the target-protein active site is unoccupied and the protein remains active. (B) At low concentrations of inhibitor, the activity of the protein is only partially inhibited because the active site is not maximally occupied. (C) At high concentrations of inhibitor, excess compound ensures that the active site is occupied and protein enzymatic activity is disrupted, yielding maximum efficacy. Some proteins have additional functions that are independent of their active-site enzymatic activities (e.g., scaffolding). These functions will remain unaffected despite high inhibitor concentrations. (D) In certain mutated proteins in cancer [e.g., the Phe876→Leu876 mutation in the androgen receptor (49)], the inhibitor compound can behave as an agonist and have the adverse effect of further activating the target protein.

GRAPHIC: ADAPTED BY K. SUTLIFF/SCIENCE

Creating a new paradigm: Induced protein degradation

Targeted protein degradation has emerged as an alternative paradigm that requires only brief interaction between a small molecule and its target protein to elicit the desired loss of protein function. Under this “event-driven” paradigm, loss of function is due to the removal of the target protein as a result of the transient binding event (Fig. 2). The strategy is to engage the cellular quality-control machinery (the UPS) and thereby tag unwanted proteins for destruction via ubiquitination. Because the small-molecule drug survives the target-protein tagging and destruction steps, it is free to engage in multiple cycles of target-protein degradation, resulting in substoichiometric activity. This event-driven approach is catalytic in nature, eliminating the need to maintain high levels of drug. For the inhibitory effect of degradation to be reversed, the protein must be resynthesized.

Fig. 2 Event-driven pharmacological paradigm.

(A) Active-site binding and/or enzymatic proteome engagement. A chimeric molecule binds to the active site of the target protein and inhibits its activity. This molecule also recruits the cellular protein degradation machinery to tag (ubiquitinate) the target for proteasomal degradation. This molecule displays catalytic (i.e., processive) activity over repeated cycles of induced degradation, thus eliminating the need for high concentrations to achieve efficacy. It also disrupts nonenzymatic functions of the target protein. Restoration of target-protein function requires resynthesis of the protein. Mutated proteins that retain the ability to bind the ligand and would otherwise become activated are now vulnerable. (B) Non–active-site binding and/or nonenzymatic proteome engagement. A chimeric molecule functions in the same manner as in (A), except that the molecule binds to any crevice (as opposed to an active site) on the surface of the protein. This enables targeting of proteins that lack an active site (e.g., transcription factors and scaffolding proteins).

GRAPHIC: ADAPTED BY K. SUTLIFF/SCIENCE

Protein degradation by the UPS involves a series of steps culminating in the conjugation of the 8.5-kDa protein ubiquitin to the targeted protein by an E3 ligase (68). This monoubiquitinated protein then undergoes multiple cycles of ubiquitination to become polyubiquitinated, which allows it to be recognized and then degraded by the 26S proteasome.

Small molecules that both inhibit protein activity and induce protein degradation

Early literature reports suggested that certain small molecules developed to inhibit protein activity could also induce the degradation of their targets, although the mechanism was often unknown. Canertinib (CI-1033) is one example of an irreversible tyrosine kinase inhibitor that also induces the polyubiquitination and degradation of ErbB-2 (a receptor tyrosine-protein kinase also known as HER2/neu), an important therapeutic target in several human cancers (9). Fulvestrant is another example of a small molecule that inhibits the signaling function of its target, ERα (estrogen receptor alpha), and induces its degradation. Fulvestrant produced better outcomes in patients with ERα-positive breast cancers in comparison with tamoxifen, another ERα antagonist that acts primarily by inhibiting ERα signaling (10). Arsenic trioxide is a highly effective therapy for acute promyelocytic leukemia (APL), and it works in part by degrading PML-RARα (promyelocytic leukemia–retinoic acid receptor alpha), the fusion protein that is characteristic of this disease and whose activity disrupts normal myeloid differentiation (11). More recently, lenalidomide, an effective immunomodulatory drug (IMiD) used in the treatment of MM, was found to cause the selective ubiquitination and degradation of CK1α (casein kinase 1α) (12) and two important transcription factors in MM, Ikaros and Aiolos (13). Other examples of small molecules that promote the degradation of specific proteins include the cancer-preventive isothiocyanates, which selectively bind and induce the degradation of α- and β-tubulins in human cancer cells (14), and CaCCinh-AO1, which causes the degradation of the calcium-activated chloride channel ANO1. CaCCinh-AO1 reduces the proliferation of ANO1-dependent cancer cells by inducing an endoplasmic reticulum–associated proteasomal degradation of its target (15). The success of these drugs has spurred increased interest in the development of additional degrader compounds. However, because the ability of these small molecules to induce target-protein degradation was discovered fortuitously, the challenge has been to develop a strategy that would allow for the rational design of degraders that can target any specific protein of choice.

Proteolysis targeting chimeras (PROTACs)

Typically, an E3 ubiquitin ligase requires a specific recognition signal to recruit and ubiquitinate its natural substrate (8). However, the proteolysis targeting chimera (PROTAC) technology is one approach that can achieve the same result for any protein target, even those not naturally ubiquitinated. By using a heterobifunctional molecule to form a complex between the target protein of interest and a recruited E3 ligase, PROTACs induce the ubiquitination and subsequent proteasomal degradation of a target protein. The PROTAC concept was introduced in a study in which an aliphatic linker was used to connect two protein ligands: ovalicin, which covalently binds the enzyme methionine aminopeptidase 2, and a phosphopeptide that associates with the ubiquitin ligase complex SCFβ-TrCP (Skp1/Cullin/F boxβ-TrCP) (16). Incubation of this PROTAC with Xenopus egg extract resulted in ubiquitination of methionine aminopeptidase 2 and its degradation by the proteasome (16). The same approach was subsequently used to ubiquitinate and degrade the estrogen and androgen receptors (17), protein targets that play major roles in the pathogenesis of breast and prostate cancers (18, 19), respectively. These first-generation PROTACs had low activity in cells, presumably because of the poor cell permeability of the peptide employed.

“…achieving and maintaining high systemic drug levels is one of the major challenges in drug development. Moreover, these high dosages can lead to undesirable off-target effects.”

The next generation of PROTACs moved away from SCFβ-TrCP and instead engaged the von Hippel–Lindau (VHL) tumor suppressor protein to ubiquitinate recruited proteins. The E3 ligase binding moiety of these PROTACs consisted of a short, hydroxyproline-containing peptide sequence derived from the VHL protein’s natural substrate, the transcription factor HIF1α (hypoxia-inducible factor 1α). When coupled to a targeting ligand and with incorporation of a cell-penetrating peptide, the resulting PROTACs proved selective for degradation of their target proteins in intact cells (20).

Though these early PROTACs were cell-permeable, they were far from drug-like because they still required peptide sequences for E3 ligase recognition and cell penetration. Thus, the challenge was to design a nonpeptidic E3 ligase ligand to allow for the creation of “all-small-molecule” PROTACs. This was achieved with the development of a compound consisting of a selective androgen receptor modulator tethered to nutlin. Nutlin is a ligand for the E3 ubiquitin ligase MDM2; it disrupts MDM2 binding to its natural substrate, p53. This PROTAC was shown to recruit androgen receptors (ARs) to MDM2, leading to the ubiquitination and degradation of ARs in HeLa cells (21). A key step in the evolution of the PROTAC technology was the development of hydroxyproline-based small-molecule VHL ligands that could replace the HIF1α peptide used in earlier studies (22). Bondeson et al. used these ligands to design two all-small-molecule PROTACS that targeted the nuclear hormone receptor, ERRα (estrogen-related receptor alpha), and the serine-threonine kinase RIPK2. At a concentration 1000 times lower than that of previously reported compounds, PROTAC_ERRα and PROTAC_RIPK2 induced degradation of their target proteins to below 10% of levels present in untreated cells (23). The activity of these PROTAC molecules was catalytic and specific for the intended protein targets. Several research groups have developed additional PROTAC molecules that are likewise showing promise in preclinical studies. Some of these examples will be discussed later in the Review.

Further PROTAC technology development

As discussed above, there are three basic components of a small-molecule PROTAC: the target-binding ligand, the E3 ligase-binding ligand, and the linker that holds these two moieties together. A successful PROTAC molecule will require optimization of each of these components. Fortunately, a plethora of high-affinity small-molecule ligands and drugs have been developed to target the many proteins of interest in the cancer research field. However, despite an estimated >700 endogenous human E3 ligases (8), the use of PROTAC technology has largely been limited to the few E3 ligases for which researchers have successfully developed selective small-molecule ligands: VHL, CRBN (cereblon), MDM2, and cIAP1 (cellular inhibitor of apoptosis protein 1). The VHL ligand was described above (22). Three small molecules have been identified as CRBN ligands; these include thalidomide, lenalidomide, and pomalidomide (a class of immunomodulatory antitumor compounds in myeloma cells) (24). Methyl bestatin (MeBS) and nutlin are small-molecule ligands for cIAP1 (25) and MDM2 (21), respectively. In addition to target-protein and E3 ligase moieties, the connecting linker is an important component of successful PROTACs. For each PROTAC, the length, hydrophilicity, and rigidity of the linker must be optimized for improved cell permeability and optimal presentation of a target to the E3 ligase for ubiquitination.

Recent work highlights the importance of each of these components within the context of a PROTAC molecule and illustrates how small changes can substantially affect outcomes (26). The study focused on PROTACs targeting BCR-ABL, an oncogenic fusion tyrosine kinase that drives the development of chronic myelogenous leukemia. Small-molecule inhibitors of the tyrosine kinase activity are highly effective therapies for this disease and are now the clinical standard of care, although the development of drug resistance can limit their efficacy in a subset of patients. The PROTACs were designed using a variety of linkers to attach imatinib, bosutinib, or dasatinib (structurally dissimilar small molecules that inhibit the c-ABL kinase domain) to the VHL small-molecule ligand (22) or to pomalidomide, a thalidomide derivative (24). Of the four linkers tested, only one showed a considerable loss of PROTAC affinity for the target, suggesting a wide latitude of flexibility in linker design for these PROTACs. Whereas the imatinib PROTACs failed to degrade BCR-ABL or c-ABL despite verifiable target binding and inhibition in cells, the dasatinib-CRBN and bosutinib-CRBN PROTACs successfully degraded both BCR-ABL and c-ABL. This implies that, beyond affinity considerations, target engagement alone is not sufficient for degradation and that the identity of the target ligand is important in determining the efficacy of PROTAC molecules in cells. The observation that the bosutinib-VHL compound does not degrade BCR-ABL or c-ABL and that the dasatinib-VHL compound degrades only c-ABL suggests that changing the E3 ligase recruited for ubiquitination and subsequent degradation can considerably alter the ability to degrade a target, as well as the specificity of this degradation. In essence, it is possible to tune degradation specificity—from no degradation to selective degradation (only c-ABL) to promiscuous-degradation (BCR-ABL and c-ABL)—by testing different target-ligand and E3 ligase combinations. Conceivably, this type of “specificity dial” would allow researchers to use promiscuous inhibitors as a starting point for the design of more-selective degraders.

Other strategies for targeted protein degradation

A number of other promising approaches have been developed that, like the PROTAC strategy, use small molecules to induce targeted protein knockdown. For example, Itoh et al. developed a hybrid compound that fuses MeBS, a ligand for cIAP1, to all-trans retinoic acid (ATRA), a ligand for retinoic acid receptors and an effective drug for APL (27). Through recruitment of cIAP1, the compound degraded cellular retinoic acid–binding proteins (CRABPs), based on their ability to bind ATRA (27). Furthermore, this compound inhibited migration of IMR-32 neuroblastoma cells, a process that requires CRABP-II. This class of PROTACs is known as SNIPER (specific and nongenetic IAP-dependent protein eraser) PROTACS (28) and consists of MeBS linked to ligands for other protein targets, thus recruiting cIAP1 to induce ubiquitination and proteasomal degradation of the target protein. Using a SNIPER with 4-hydroxy tamoxifen as the ERα ligand, Okuhira et al. showed successful degradation of ERα and resultant necrotic death of ERα-expressing MCF-7 breast cancer cells (28). Another strategy involves treating cells with two halves of a PROTAC molecule (as opposed to a larger single compound with less optimal physiochemical properties) that then self-assembles intracellularly (29). This approach was used to recruit the E3 ligase cereblon to successfully degrade BRD4 (bromodomain-containing protein 4) and ERK1/2 (extracellular signal–regulated kinases 1 and 2), both important targets for cancer therapy.

Similar to the PROTAC technology, hydrophobic tags have been used to hijack the UPS to degrade proteins of interest. The underlying concept is that partially unfolded or misfolded proteins expose hydrophobic patches that are otherwise buried, serving as a recruiting signal for E3 ubiquitin ligases that then catalyze ubiquitination and subsequent degradation. Researchers have successfully mimicked this protein unfolding for specific targets by appending a low–molecular weight hydrophobic tag to the target’s small-molecule ligand; this results in recruitment of the UPS to degrade the target protein (30). This strategy has been applied to degrade the HER3 (human epidermal growth factor receptor 3) pseudokinase, a currently “undruggable” cancer target, by appending a “greasy tag” to the potent and selective HER3 ligand TX1-85-1 to generate the bifunctional TX2-121-1 compound (31). The resulting TX2-121-1–induced HER3 knockdown led to inhibition of downstream signaling and reduced proliferation of HER3-dependent cell lines. Gustafson et al. generated selective androgen receptor degraders (SARDs) that degrade the AR via hydrophobic tagging (32). SARD279 was shown to be as effective as enzalutamide (an FDA-approved inhibitor of AR signaling) in suppressing proliferation of human prostate cancer cells; this drug also suppressed proliferation of enzalutamide-resistant prostate cancer cells (32).

Protein degradation strategies in preclinical models of cancer

Within the past 2 years, considerable progress has been made in the advancement of protein degradation technology to selectively and effectively degrade key cancer targets. The BET (bromodomain and extra-terminal) proteins, such as BRD4, play important roles in the progression of various cancers, including acute myeloid leukemia (AML), MM, Burkitt’s lymphoma (BL), ovarian cancer, and prostate cancer (33, 3439). To that end, potent small-molecule inhibitors of BET proteins have been developed and are in clinical trials (40, 41). The experience to date suggests that the effectiveness of these inhibitors may be limited by incomplete suppression of the downstream oncogene c-MYC (39) and a compensatory increase in BRD4 protein levels as a way to circumvent inhibition (42).

Several labs have reported PROTACs that degrade the BRD4 protein in cells (33, 4244). For example, Lu et al. designed ARV-825, which degrades BRD4 by engaging the CRBN E3 ligase and incorporates the potent BET inhibitor OTX015 as a recruiting moiety (42). At subnanomolar concentrations, ARV-825 induced near-complete BRD4 degradation in BL cell lines. It suppressed c-MYC expression and was a more potent inducer of apoptosis than conventional BET inhibitors such as JQ1 and OTX015. Winter et al. designed a PROTAC called dBET1 by appending JQ1 to a phthalimide moiety, which hijacks the cereblon E3 ubiquitin ligase complex (43). dBET1 showed greater antiproliferative effects in AML and lymphoma cells when compared with JQ1 inhibition. Furthermore, dBET1 inhibited leukemia progression in a mouse xenograft model of AML. Raina et al.’s ARV-771 compound recruits BET proteins to the VHL E3 ligase for degradation at subnanomolar concentrations in several prostate cancer cell lines. ARV-771 inhibited the proliferation of enzalutamide-resistant prostate cancer cells and inhibited tumor growth in a castration-resistant prostate cancer mouse xenograft model (33). Zengerle et al. designed a BRD4 PROTAC in which JQ1 is tethered to a ligand for the VHL E3 ligase and showed that this compound selectively induces degradation of BRD4 in cultured cells (44).

Conclusion

The most promising aspect of targeted degradation as a therapeutic strategy may be its potential for targeting proteins for which there currently is no drug. Undruggable proteins such as scaffolding proteins, pseudokinases, and transcription factors make up ~80% of the human proteome; these proteins are neither enzymes nor receptors and lack an enzymatic activity or functional interaction that can be compromised by an inhibitor (45). The ability to target any of these would require the identification of a specific ligand. However, because event-driven protein degradation can be mediated via any binding site on the surface of the target protein rather than restricted to a single, identifiable active site (Fig. 2), the development of simple but potent and selective ligands may be easier. The targeted degradation approach to eliminate such proteins has the potential to render these otherwise-undruggable proteins pharmaceutically vulnerable.

Targeted degradation may also prove useful in drug-resistance mechanisms that involve a compensatory increase in the expression of inhibited proteins or mutations that result in the loss of inhibition despite maintained target engagement. Given the encouraging preclinical studies on targeted degradation of BET proteins and the AR, it appears that tools to win this pharmacological “arms race” may be available. One possible application is the subgroup of cetuximab-resistant non–small cell lung cancers that show increased expression of the epidermal growth factor receptor, the protein targeted by cetuximab (46). In this case, substantial loss of target-protein levels and activity would still be achieved because of the catalytic nature of the PROTAC mechanism of action. Finally, although this Review has focused specifically on applications in cancer therapy, other disease contexts may also benefit from this emerging drug paradigm.

References and Notes

Acknowledgments: C.M.C. is the chief scientific adviser of and is a shareholder in Arvinas, a biotechnology company focused on developing protein degradation therapeutics for cancer and other diseases. C.M.C. is an inventor on patent US7041298 B2 and patent applications PCT/US2013/040551, PCT/US2013/021136, EP20150180508, and PCT/US2011/063401 (submitted by Yale University) and PCT/US2015/025813 (submitted by Arvinas), which cover targeted protein degradation. C.M.C. acknowledges support from the Leukemia and Lymphoma Society and the NIH (grant R35CA197589). C.M.C. also receives research funding from Arvinas.
View Abstract

Navigate This Article