Targeting the untargetable: RB1-deficient tumours are vulnerable to Skp2 ubiquitin ligase inhibition

Abstract

Proteins that regulate the cell cycle are accumulated and degraded in a coordinated manner during the transition from one cell cycle phase to the next. The rapid loss of a critical protein, for example, to allow the cell to move from G1/G0 to S phase, is often regulated by its ubiquitination and subsequent proteasomal degradation. Protein ubiquitination is mediated by a series of three ligases, of which the E3 ligases provide the specificity for a particular protein substrate. One such E3 ligase is SCF^Skp1/Cks1, which has a substrate recruiting subunit called S-phase kinase-associated protein 2 (Skp2). Skp2 regulates cell proliferation, apoptosis, and differentiation, can act as an oncogene, and is overexpressed in human cancer. A primary target of Skp2 is the cyclin-dependent kinase inhibitor p27 (CDKN1b) that regulates the cell cycle at several points. The RB1 tumour suppressor gene regulates Skp2 activity by two mechanisms: by controlling its mRNA expression, and by an effect on Skp2’s enzymatic activity. For the latter, the RB1 protein (pRb) directly binds to the substrate-binding site on Skp2, preventing protein substrates from being ubiquitinated and degraded. Inactivating mutations in RB1 are common in human cancer, becoming more frequent in aggressive, metastatic, and drug-resistant tumours. Hence, RB1 mutation leads to the loss of pRb, an unrestrained increase in Skp2 activity, the unregulated decrease in p27, and the loss of cell cycle control. Because RB1 mutations lead to the loss of a functional protein, its direct targeting is not possible. This perspective will discuss evidence validating Skp2 as a therapeutic target in RB1-deficient cancer.

Introduction

Tumorigenesis is a complex, multi-step process that entails the reprogramming of cellular signalling networks and primarily originates with pro-oncogenic gene mutations. Technological advances now allow for the routine clinical analysis of these tumour-specific alterations, primarily at the genomic and transcriptomic levels, and when one or more of these are actionable, to new drug therapies. The proteome, however, is the ultimate driver of the cellular functions that define the many changes occurring in malignant transformation [1, 2]. The signalling networks are regulated, and dysregulated in cancer, by the abundance, sub-cellular localisation, and activity of key proteins. These in turn can be controlled by site-specific post-translational modifications, of which ubiquitination is second only to phosphorylation in abundance [2, 3]. Despite their importance in protein degradation, protein function, and cellular homoeostasis, drugs targeting the ubiquitin ligases have not had the same impact on cancer therapy as have inhibitors of protein kinases. This review will highlight a specific ubiquitin E3 ligase, SCF-Skp2, and the pre-clinical studies that suggest it would be an attractive target for treating the most highly aggressive and drug-resistant cancers.

E3 ubiquitin ligases are therapeutic targets in cancer

Protein degradation is an essential cellular process that entails the elimination of damaged and unwanted proteins to maintain cellular homoeostasis [4]. Proteins that regulate the cell cycle, in particular, are rapidly accumulated and then equally rapidly eliminated in a timely manner as the cell progresses from one cell cycle phase to the next. The process of protein degradation is controlled by the ubiquitin-proteasome system (UPS) that mediates the ubiquitination of proteins at specific amino acids, a critical post-translational modification. The ubiquitination process is tightly regulated by the sequential actions of three enzymes: ubiquitin-activating E1, ubiquitin-conjugating E2, and ubiquitin-protein E3 ligase [4–7]. Mechanistically, the UPS-mediated protein degradation involves three steps. The cysteine component of the E1 enzyme forms a thioester bond with the carboxyl group of the glycine component on ubiquitin. This results in an ATP-dependent activation of ubiquitin, which then forms a thioester linkage with the cysteine component of the E2 enzyme. Lastly, the E3 enzyme attaches ubiquitin to the amino group of the lysine component on a substrate protein through a covalent bond. Subsequent recruitment of ubiquitin molecules leads to the polyubiquitination of the substrate protein, which is then degraded by the 26S proteasome [4–7].

Among the three enzymes, E3s are the most studied and over 600 E3s, encoded in the human genome, have been identified to date. In contrast, only two E1s and approximately 40 E2s have been found. All E3 ligases are multiprotein complexes, and they have been divided into three categories: RING-in-between-RING (RBR) family, homologous to E6-associated protein C terminus (HECT) family, and Really Interesting New Gene (RING) family [5, 6]. The RING family of E3s comprises the largest number of E3s, and has a characteristic zinc-binding RING domain or a U-box domain. A distinct class of RING E3s known as CULIN-RING, accounts for over 20% of the proteins degraded through the UPS. The Skp1/Cullin 1/F-box (SCF) protein complex, consisting of S-phase kinase-associated protein 1 (Skp1), cullin 1, an F-box protein, and Ring box protein 1 (Rbx1), belongs to the CULIN-RING class of E3s (Fig. 1). The F-box protein controls the recognition of protein substrates for ubiquitination, whereas Skp1 is essential for recognition and interaction of F-box proteins [8]. Cullin 1 functions as a linker between Skp1 and RBX1 and lastly, RBX1 contains a zinc-binding domain to which E2 binds and transfers ubiquitin to the lysine component of the substrate protein [4–6].

Fig. 1. Schematic of the SCF E3 ligase Skp2.

a The RB1 protein (pRb) binds to the Skp2-F-box component protein, preventing Skp2 protein substrates from binding. Under these conditions, the Skp2 substrate p27 is neither ubiquitinated nor degraded, allowing it to restrict and regulate cell cycle progression. b With the loss of RB1, p27 is able to bind to Skp2, where it is ubiquitinated and degraded by the proteasome. c Small-molecule inhibitors of Skp2 can bind to the pockets formed by the interaction of Skp2, its accessory protein Cks1, and p27, and restore p27’s functions.

The E3 ubiquitin ligase Skp2 regulates the cell cycle in normal and transformed cells

Out of the 69 F box proteins identified in the human genome, S phase kinase-associated protein 2 (Skp2), discovered in 1995, is the best characterised. Skp2 is the substrate recruiting subunit of the SCF^Skp1/Cks1 complex (Fig. 1) [4–7]. Skp2 regulates many cellular processes including the cell cycle, cell proliferation, apoptosis, and cell differentiation. Among the substrates of Skp2 are the tumour suppressor proteins p27, p21, FOXO1, Tob1, p57, and p130; thus, increased expression or dysregulation of Skp2 activity can have oncogenic properties [9–11]. One of these proteins, p27 (also known as CDKN1b and KIP1), is a member of cyclin-dependent kinase inhibitor (CDKI) family and regulates the cell cycle at multiple points (G1, G2, and G2/M) by binding to a cyclin-CDK complex [12]. p27 is rarely mutated or deleted in human cancer, rather its protein levels are often deregulated and reduced, and this has been identified as a negative independent prognostic factor in many human cancers [12]. Although Skp2 has several protein substrates, studies indicate that p27 is the critical substrate in the regulation of tumorigenesis in animal models, as the phenotypes observed in Skp2^−/− mice are no longer seen in Skp2^−/−p27^−/− double KO mice [12–14]. Structurally, p27 has an N-terminal domain, where its Cdk-inhibitory property resides, and a C terminal domain which controls its nuclear localisation [15]. Studies have shown that the increases and decreases in p27 levels are mainly regulated post-translationally, mediated by ubiquitin-proteasome degradation [14–16]. In a normal cell cycle, the activation and deactivation of CDKs are the keys to the controlled transition through the different phases of the cell cycle. The Cdk2/cyclin E or Cdk2/cyclin A-mediated phosphorylation of p27 at threonine 187 marks p27 to be recognised by Skp2. The cyclin-dependent kinase subunit 1 (Cks1) is an adapter protein that links phosphorylated p27 to Skp2 and increases its binding affinity [17]. Thus, Skp2 often acts as an oncoprotein by targeting p27 for its degradation, which results in uncontrolled cell cycle progression and cell proliferation.

Skp2 overexpression has been detected across a wide spectrum of human solid tumours, and in all cases, its overexpression was correlated with poorer patient prognosis [18–21] (Table 1). In most, but not all of these studies, the levels of Skp2 expression were inversely correlated with p27 expression, as would be anticipated from the role of Skp2 in mediating p27 protein stability. Interestingly, in some tumour models, Skp2 overexpression alone was not sufficient to cause transformation in vitro or in vivo. Rather its pro-tumorigenic actions occur primarily in the presence of other oncogenic drivers (e.g., mutant H-Ras and N-Ras) [22, 23].

Table 1.

Human cancers with Skp2 overexpression.

	Skp2 expression inversely correlated with p27	Skp2 expression correlated with poor prognosis	References
Breast	Yes	Yes	[74–79]
Cervical	No	Yes	[80]
Colorectal	Yes	ND	[81–83]
Endometrial	ND	Yes	[84]
Oesophageal	Yes	Yes	[85]
Gastric and GIST	Yes	Yes	[86, 87]
Glioblastoma	ND	Yes	[88]
Kaposi sarcoma	No	Yes	[89]
Lung SCLC	Yes	ND	[90]
Lung NSCLC	Yes	Yes	[91–95]
Lymphoma	Yes	Yes	[20, 22, 96, 97]
Melanoma	Yes	Yes	[98–100]
Neuroblastoma	ND	Yes	[101, 102]
Oral squamous	Yes	Yes	[23]
Oro-nasopharyngeal	ND	Yes	[103–107]
Osteosarcoma	ND	Yes	[65, 67, 108–110]
Ovarian	ND	Yes	[111]
Pancreatic	ND	Yes	[112]
Prostate	Yes	Yes	[60, 113, 114]
Renal call	ND	Yes	[115]
Soft-tissue sarcoma	No	Yes	[116]

ND not determined.

Conversely, loss of Skp2 profoundly restricts tumorigenesis, despite the presence of pro-oncogenic conditions (discussed below), providing a rationale for developing Skp2 inhibitors for therapy. For Skp2 inhibitors to be successful clinically, however, their anti-tumour effects must be balanced with potential adverse effects on normal cells, so as to achieve a manageable therapeutic index. Genetic studies support the therapeutic viability of a Skp2 inhibitor: male and female mice with a germ-line knockout (KO) of Skp2 were viable, fertile, and tumour free [15]. No illnesses were evident in the Skp2 KO mice for up to 10 months of age. The mice did exhibit reduced body weight, compared to their Skp2 wild-type littermates; however, the weight differences originated during embryogenesis, suggesting it would not be a major concern when treating patients. Skp2 KO can induce cellular senescence; however, this occurred only in the presence of pro-oncogenic signals [21]. These observations contribute to the appeal of Skp2 as a therapeutic target, as its inhibition by small-molecule inhibitors would have a degree of selectivity towards tumour cells, compared to normal cells. In fact, a number of small-molecule inhibitors of Skp2, with varying mechanisms of action and varying degrees of specificity, have been identified and found to have anti-tumour activity in pre-clinical models [24–28].

Skp2 inhibition is a validated target in tumours with an inactivated RB1 tumour suppressor gene

Over the years, a large body of literature has reported that RB1 regulates cellular functions including cell proliferation, DNA damage and repair mechanisms, chromosomal instability, autophagy, angiogenesis, cell cycle, and apoptosis [29, 30]. Hypo-phosphorylated RB1 protein (pRb) binds to cellular proteins, of which the most widely studied are the E2F transcription factors, where its binding represses transcription. The loss of pRb’s functions, primarily due to hyperphosphorylation of pRb or inactivating mutations in the RB1 gene, causes increased and sustained activation of E2F1 and the loss of control of the G1/G0 to S phase transition [31].

While the ability of pRb to bind to E2Fs has been the focus of much research, protein interaction databases indicate that there are more than 300 proteins that might interact with pRb [30]. For example, pRb exerts significant cell cycle control that is transcription-independent, due to its well-characterised regulation of protein stability by direct effects on the ubiquitin ligase proteasomal degradation pathway and the regulation of the stability and function of proteins, particularly the cyclins and cyclin-dependent kinase inhibitors. Our group identified Skp2 as an early repression target of the pRb protein [32]. pRb regulates Skp2 by two mechanisms: it binds to the Skp2 substrate recruiting subunit, effectively blocking the binding of Skp2’s substrates (including p27) and their subsequent ubiquitination (Fig. 1a); and it represses Skp2 mRNA expression [33, 34]. In cells that have lost functional pRb, therefore, there is both increased Skp2 expression and the loss of restraint on Skp2-mediated p27 degradation, which leads to the loss of key cell cycle restriction points and promotes cell proliferation (Fig. 1b). In fact, some studies of the role of RB1 in cell cycle control have shown better correlations with p27 protein levels than with changes in proteins encoded by E2F-regulated genes [4, 35–37].

RB1 mutation and loss of function is common in aggressive tumours

Cancer genome studies have shown the frequent genetic inactivation of RB1 and loss of its expression in solid tumours. The likelihood of inactivation increases in more advanced, metastatic, and chemotherapy-resistant cancers [38–42]. An analysis of the TCGA pan-cancer atlas expression dataset, containing RNA expression data for 11,007 tumours from 33 different cancer types, found that 698 (6.3%) were annotated as having two or more RB1 biallelic DNA alterations (mutations and copy-number alterations), 1514 (13.8%) as having one RB1 alteration, and 7727 (70%) as having no RB1 DNA alterations [38]. Clinical tumour specimens with inactivating RB1 mutations include lung, prostate, bladder and metastatic breast cancers, sarcomas, glioblastomas, and advanced gastro-enteropancreatic neuroendocrine carcinomas [39–42].

The loss of RB1 is closely associated with, and may in fact drive the acquisition of a neuroendocrine phenotype across a broad range of advanced cancers, regardless of their anatomical origin [43]. For example, primary small cell carcinomas of the oesophagus are a lethal neuroendocrine carcinoma, and whole-exome and RNA sequencing of clinical specimens found multiple mechanisms of RB1 disruption in 98% of cases. Furthermore, the transcriptomic landscape of these RB1-deficient tumours more closely resembled RB1-deficient neuroendocrine small cell lung cancers (SCLCs), rather than oesophageal carcinomas with an adenocarcinoma phenotype [44]. In addition to gene mutations, DNA methylation and gene expression analysis of TCGA detected additional tumours with RB1 loss through mechanisms not detected by DNA sequencing [38]. Interestingly, recent proteogenomic analysis of the paradoxical amplification of RB1 seen in some human colon cancers was found to be associated with pRb hyperphosphorylation and the consequent inactivation of pRb, suggesting that other mechanisms may also contribute to the loss of RB1 function [45]. Regardless of the mechanism and the specific tumour type, RB1 loss of function was generally associated with shorter progression-free and overall patient survival [38].

In this review, we focus on studies validating the targeting of Skp2 in three types of cancers, notable for RB1 mutations and loss of function as a driver of malignancy: SCLC, prostate cancer (PrCa), and sarcomas. Based on the information in Table 1, however, it is likely that the use of Skp2 inhibitors could have an impact on a broad range of advanced and intractable cancers, particularly those with RB1 loss. RB1 mutations are routinely assessed in the genetic analysis of some types of cancer, and they could be readily added to others to assist in treatment decisions.

Small cell lung cancer

SCLC is an aggressive, fast growing, and frequently metastasising form of lung cancer that occurs primarily in smokers [46]. The median survival time of patients with metastatic disease is about 10 months with a 5-year survival of less than 5% [46, 47]. Although it represents only 15% of all lung cancer patients, SCLC causes over 30,000 deaths per year in the United States, surpassing the death toll of PrCa and many other malignancies. Over the past three decades, more than 40 drugs have been tested and have failed in phase III clinical trials, and a study investigating >500 FDA-approved oncology drugs failed to produce any with significant activity in 63 SCLC cell lines [48, 49]. Greater than 95% of SCLC patient tumours have a biallelic inactivation of tumour suppressor genes RB1 and TP53, which is considered a hallmark of the disease [46, 47, 50]. The functions of the genetically deleted RB1 and TP53 cannot be directly restored clinically, and their absence makes it difficult to develop targeted therapies for SCLC.

Our group has shown that the KO of Skp2 prevents tumorigenesis in RB1-deficient mouse models of pituitary tumours [51, 52]. Based on this promising finding, we recently reported that the loss of Skp2 in the lung prevents SCLC development in Rb1/Trp53- deficient mice [53]. In addition to blocking tumorigenesis in genetic mouse models, induced knockdown of Skp2 significantly slowed the growth of established subcutaneous mouse tumours. When the SCLC cells were implanted orthotopically in the lungs of mice and allowed to become established, subsequent Skp2 knockdown reduced the size of the lung tumours and inhibited the development of liver metastases [53]. In addition, a small-molecule inhibitor of Skp2 called C1 was investigated for its anti-tumour activity in vitro and in vivo [24, 25]. C1 was identified by in silico modelling and virtual library screening based on the crystal structure of the SCF-Skp2-p27 complex, which defined the interaction of Skp2 with its accessory protein Cks1 to form a pocket to which p27 binds [24, 25]. C1 binds to the pocket, thereby preventing p27 from binding and preventing its subsequent ubiquitination and proteasomal degradation (Fig. 1c). C1 inhibited the proliferation, increased p27 levels, induced apoptosis and slowed tumour growth in vivo in human and murine SCLC cells [53].

Prostate cancer (PrCa)

Large-scale genomic studies of cohorts of primary, hormone-naive, and metastatic castration-resistant prostate cancer (mCRPC) have found a high degree of molecular heterogeneity among PrCa patients [54–56]. Nevertheless, certain molecular aberrations in PrCa have been closely associated with aggressive disease, especially inactivating mutations in RB1, which are common in clinically aggressive primary PrCa, become more frequent with progression to mCRPC, and may be a common second hit gene [56]. Skp2 is highly overexpressed in PrCa cells and tissue and its expression correlates with the histological grade and tumour stage of PrCa. Studies have shown that overexpression of Skp2 leads to dysplasia, hyperplasia, and low-grade carcinoma in the prostate gland [57, 58]. Although the exact mechanism by which Skp2 promotes prostate tumour growth is still under investigation, a number of upstream regulators, such as phosphatidylinositol 3-kinase (PI3K), Akt, Pten, and the androgen receptor, and downstream regulators such as p27 and BRCA2, are known to interact with Skp2 in PrCa cells [59, 60]. Due to the interplay between a number of signalling molecules and Skp2, it can modulate cell proliferation, migration, metastasis, cell cycle, and apoptosis in PrCa cells. Our group has shown that the genetic deletion of Skp2 in the prostate of Rb1/Trp53-deficient mice completely blocked tumour formation [52]. All of Rb1/Trp53-deficient mice with wild-type Skp2 developed invasive prostate carcinomas and died by 10 months, while the Rb1/Trp53-deficient mice with Skp2 KO lived for normal lifespans and none had tumours that progressed beyond the prostatic intraepithelial neoplasia stage [52].

As was done for SCLC, researchers have attempted to exploit the interplay of Skp2’s interaction with pro-tumorigenic factors to develop pharmacological strategies for treating PrCa. In an early study, a proteasome inhibitor, MG-132 was reported to upregulate the expression of p27 by inhibiting its Skp2-mediated degradation, thereby decreasing cell proliferation and increasing apoptosis [61]. However, the clinical use of non-specific inhibitors of the UPS can cause adverse effects such as nausea, anaemia, neutropenia, neuropathy, and fever. Thus, efforts have focused on the development of small-molecule inhibitors of Skp2 with improved selectivity, specificity, and limited side effects. Two such compounds, SMIP001 and SMIP004, were identified in a screen of >7000 compounds and were found to increase p27 expression, cause cell cycle arrest, inhibit colony formation in soft agar, and exhibit preferential cytotoxicity in LNCaP PrCa cells, relative to normal human fibroblasts [62]. Our group has also studied the highly specific inhibitor of Skp2’s ligase activity, C1, in mouse and human PrCa cells and organoids, and found that it upregulated p27, inhibited the proliferation of Rb1/Trp53 double deficient PrCa cells, and caused the death and disintegration of PrCa organoids [63].

Osteosarcoma (OS) and synovial sarcoma (SS)

Osteosarcoma (OS), a bone malignancy, and synovial sarcoma (SS), a soft-tissue malignancy, affect children and young adults [64, 65]. In patients with aggressive and highly metastatic disease, the 5-year survival rate is less than 30% [65]. Although decades of research have searched for new treatment options for OS and SS, surgery, cytotoxic chemotherapy and radiation remain the only available treatment modalities [66]. Recently, immunomodulatory agents are being employed clinically but with caution due to their limited specificity and high adverse effects [66].

RB1 and TP53 are the most frequently inactivated tumour suppressor genes found in OS [64]. Knockdown of Skp2 in two OS cell lines, MG-63 and SW1353, caused an increase in the p27 protein expression and cell cycle arrest in G0/G1 phase, significantly reduced cellular proliferation, and increased apoptosis [66]. Recently, it was reported that primary OS cells isolated from an Rb1/Trp53 double KO OS mouse model showed increased tumour stemness [65]. Interestingly, p27^T187A/T187A knock-in in a DKO OS model delayed tumour progression and decreased tumour initiating properties, as compared to DKO alone. The T187A variant of p27 is not able to bind to Skp2 and hence its degradation is prevented. Our recent study suggests that Skp2 is frequently overexpressed in OS and serves as a prognostic marker for disease progression [67]. Genetic knockdown of Skp2 effectively inhibits OS invasion and metastasis [65, 67]. Studies using transgenic models of OS also revealed that blocking the interaction of Skp2 with p27 promotes apoptosis and inhibits cancer stemness [65]. These data suggest that Skp2 inhibition, if it can target the cancer stem cell subpopulation, may have an advantage over conventional chemotherapies in preventing relapses in OS. Besides OS, an increased expression of Skp2 has also been reported in SS cell lines and mouse tumours, where it correlates with an increase in proliferation, invasion, migration, and stemness of SS [68]. Moreover, genetic knockdown or pharmacological inhibition of Skp2 suppresses Twist1, a transcription factor that modulates stemness and epithelial-mesenchymal transition. As a result, there is an increase in apoptosis and a decrease in cell proliferation and stemness in SS models [68–70].

Other solid tumours

Documenting its overexpression and correlation with more aggressive tumours is the first step in validating Skp2 as a viable target for drug development. As discussed above, evidence that Skp2 KO or inhibition reverses tumorigenesis and is effective against established tumours in PrCa, SCLC, and OS is the next step, providing pre-clinical support for further investigation and for the discovery of new potent and specific Skp2 inhibitors. There are supporting data from other tumour types as well: the growth of breast cancer cells and mutant B-Raf-driven human melanoma cells in vitro and as tumours in vivo were inhibited by Skp2 knockdown [71, 72]. Pten^+/- mice developed adrenal tumours and lymphadenopathy with 100% penetrance, and these were completely blocked in mice with concurrent Skp2 loss [21]. Given the importance of Skp2 in cell cycle regulation, it is likely that other vulnerable tumours will be identified.

Conclusions

Skp2, an E3 ubiquitin ligase, causes proteasome-mediated degradation of a number of substrate proteins, some of which critically regulate cell cycle progression and cell proliferation. Skp2’s oncogenic properties have been well documented and thus its genetic or pharmacological suppression could be a useful therapeutic strategy. The upregulation of Skp2 expression and activity has been linked mechanistically to the loss of the tumour suppressor RB1; the possibility that Skp2 inhibitors could be active in tumours with other oncogenic drivers remains to be explored. RB1 loss is also causally linked to neuroendocrine differentiation and tumour plasticity in a range of tumour types, and it often leads to more aggressive, metastatic and therapy-resistant cancer. The study of Skp2’s interaction with other signalling pathways could also open new avenues in understanding its broader role in cancer. Hence there is a strong rationale for continued ongoing efforts to develop small-molecule inhibitors of Skp2 for clinical use. Our discussion here represents only the first step towards this goal; lead optimisation, proof of concept in pre-clinical studies, and clinical testing are some of the next steps in drug development. Issues of efficacy, specificity and potential on-target and off-target side effects of Skp2 inhibitors would need to be addressed. Despite these challenges, biomarker-matched drugs historically have a much higher FDA approval rate than other small-molecule anti-cancer drugs [73]. Because they target tumours with ‘untargetable’ RB1 mutations, Skp2 inhibitors would be a novel and valuable addition to the treatment of these recalcitrant cancers.

Author contributions

PG, BH and ELS wrote the manuscript; PG, HZ and ELS edited the manuscript. Revising it critically: all. Final approval of the version to be published: all.

Funding

This work was supported by NIH/NCI grants RO1 CA230032, RO1 CA201458, RO1 CA255643, and an AACR-Bayer Innovation and Discovery grant.

Competing interests

The authors declare no competing interests.