Function, mechanism and drug discovery of ubiquitin and ubiquitin-like modification with multiomics profiling for cancer therapy

Abstract

Ubiquitin (Ub) and ubiquitin-like (Ubl) pathways are critical post-translational modifications that determine whether functional proteins are degraded or activated/inactivated. To date, >600 associated enzymes have been reported that comprise a hierarchical task network (e.g., E1–E2–E3 cascade enzymatic reaction and deubiquitination) to modulate substrates, including enormous oncoproteins and tumor-suppressive proteins. Several strategies, such as classical biochemical approaches, multiomics, and clinical sample analysis, were combined to elucidate the functional relations between these enzymes and tumors. In this regard, the fundamental advances and follow-on drug discoveries have been crucial in providing vital information concerning contemporary translational efforts to tailor individualized treatment by targeting Ub and Ubl pathways. Correspondingly, emphasizing the current progress of Ub-related pathways as therapeutic targets in cancer is deemed essential. In the present review, we summarize and discuss the functions, clinical significance, and regulatory mechanisms of Ub and Ubl pathways in tumorigenesis as well as the current progress of small-molecular drug discovery. In particular, multiomics analyses were integrated to delineate the complexity of Ub and Ubl modifications for cancer therapy. The present review will provide a focused and up-to-date overview for the researchers to pursue further studies regarding the Ub and Ubl pathways targeted anticancer strategies.

Graphical abstract

This review summarizes and discusses the therapeutic targets in Ub and Ubl pathways with additional multiomics data, as well as advancements in drug discovery.

1. Introduction

Ubiquitination (or ubiquitylation) is a common and essential post-translational modification that regulates several biological processes by degrading or activating/inactivating numerous proteins. In the 1980s, Hershko et al.^1,2 described ubiquitination as an ATP-dependent enzymatic reaction cascade that induced the conjugation of a highly conserved 76-residue Ub to substrates. For their contribution to the field of ubiquitination-mediated proteolysis, they were awarded the Nobel Prize in 2004. The ubiquitination reaction is sequentially catalyzed by three enzymes: Ub-activating enzyme (E1), Ub-conjugating enzyme (E2) and Ub ligase (E3)^3,4. The first step of ubiquitination involves the ATP-dependent activation of Ub by an E1. Subsequently, the activated Ub is transferred to an E2 via a transthiolation reaction. Finally, a specific E3 delivers the activated Ub to the substrate via covalent attachment (Fig. 1A). To date, 2 E1s, 33 E2s, and >600 E3s have been identified in the Ub pathway. Furthermore, several Ubl proteins have been identified as post-translational modifiers, such as neural precursor cell expressed, developmentally down-regulated gene 8 (NEDD8), and small ubiquitin like modifier (SUMO)⁴. These proteins share a similar Ub-fold domain and the ability to conjugate to substrates via a cascade reaction driven by three evolutionarily related enzymes⁴.

Figure 1.

Process of ubiquitination and its associated functions. (A) Schematic diagram of the ubiquitination and deubiquitination process and their related enzymes. (B, C) Ubiquitination involves in almost all tumor associated pathways and cell death pathways.

Accumulating studies have revealed that Ub and Ubl modifications are related to the development of distinct human diseases, especially in the onset and progression of cancer. More than 90,000 references to Ub in PubMed highlight the significance of Ub and Ubl modifications for cellular homeostasis by 2022, August. Over 10,000 studies have shown that Ub and Ubl modifications play a crucial role in regulating tumor growth, proliferation, differentiation, metabolism, and the formation of tumor microenvironment³ (Fig. 1B). Furthermore, they are also involved in almost all cell death processes, such as apoptosis, autophagy, senescence, necrosis, lysosome-dependent cell death, necroptosis, and ferroptosis (Fig. 1B). These findings suggest the importance of Ub or Ubl pathways for tumorigenesis. In the present review, we have summarized the current efforts and prospects of the Ub and Ubl pathways in tumors, including oncogenic or tumor-suppressive functions, clinical significance, regulatory mechanisms, and targeted drugs. Moreover, we have summarized some important challenges that are faced by current targeting Ub or Ubl studies, and how advances in multiomics profiles can help us to deeply understand and address these challenges.

2. Advancements in the studies of Ub and Ubl-associated enzymes in cancer

Abnormality of Ub and Ubl pathways in tumors is closely related to the dysregulation of their catalytic enzymes, including E1s, E2s, E3s and deubiquitinases^3,5, 6, 7, 8. These anomalies are regulated at various layers, including overexpression, depletion, or mutation. Given that the best characterized function of Ub and Ubl pathways is to maintain cellular homeostasis, these pathways have been identified as attractive targets for cancer therapy. Therefore, a comprehensive overview of the anomalies and their roles of Ub- and Ubl-associated enzymes, including E1s, E2s, E3s and deubiquitination enzymes, in tumorigenesis is essential. In the following sections, we have summarized the latest advancements in the studies of Ub and Ubl-associated enzymes in tumors. We also discussed neddylation, a representative Ubl pathway via adding NEDD8 to substrates, to provide insight into the feature of the Ubl conjugation machinery in cancer.

2.1. Targeting E1s suppresses tumor initiation or progression

Two Ub-activating enzymes UBA1 (also known as UBE1) and UBA6 have been identified so far. In 1981, UBA1 was identified as the first E1, which catalyzes Ub activation in 99% cellular ubiquitination reactions^9,10. It was not until 2007 that UBA6 was identified as an alternative E1, which activated Ub with the same efficiency as that of UBA1 in approximately 1% cellular ubiquitination reactions¹¹. UBA1 and UBA6 possess three common domains, the adenylation domain that binds with ATP, catalytic cysteine domain that forms the thioester bond with Ub, and Ub-fold domain (UFD) that interacts with E2¹⁰ (Fig. 2). The difference in UFD determines the specificity of the E1–E2 pairs. UBA1 favors pairing with UBE2A, UBE2B, UBE2C, UBE2D_1–4, UBE2G_1–2, UBE2H, UBE2J_1–2, UBE2K, UBE2L3, UBE2Q2, UBE2R_1–2, UBE2S, and UBE2T, whereas UBA6 specificity activates UBE2Z (also known as Use1). Additionally, UBA6 also pairs with other UBA1 paired E2 enzymes, including UBE2G2, UBE2S, UBE2D_1–4, UBE2E3, UBE2T, and UBE2L3¹² (Fig. 2). Recently, a genome-wide pan-cancer CRISPR/Cas9 screening was conducted to investigate the difference between UBA1 and UBA6 in suppressing tumor cell growth¹⁰. The screening results revealed that UBA1 depletion inhibited the growth of the analyzed cancer cell lines (n = 582), whereas UBA6 knockdown suppressed the growth of only 10.9% of the analyzed cancer cell lines¹⁰, thereby indicating that UBA1 is more essential than UBA6 for tumor growth.

Figure 2.

Schematic structure of E1s (UBA1 and UBA6), and the specificity of the E1–E2 pairs.

As for Ubl E1, NEDD8 activating enzyme E1 (NAE), a heterodimer of NAE1 and UBA3, is significantly up-regulated in a myriad of tumor tissues. The overexpression of NAE promotes the tumorigenesis and tumor progression, while its inhibition via pharmacological (e.g., MLN4924) or genetic approaches (e.g., siRNA) suppresses tumor growth13, 14, 15. In mechanism, targeting NAE to the inhibition of neddylation 1) induces apoptosis via inhibiting the degradation of ATF4, IκBα and NOXA^13,16, 17, 18, 19, 20, 21; 2) promotes senescence via inducing the accumulation of p21 and p2722, 23, 24; 3) facilitates cell cycle arrest via suppressing the degradation of p21, p27 and Wee125, 26, 27; 4) induces the re-initiation of DNA replication (re-replication) by increasing the levels of CDT1 and ORC1^28,29; 5) inhibits the inflammation via inducing the accumulation of IKBα or RHBO^30,31. 6) suppresses the infiltration of immune suppressive cells (e.g., MDSCs and TAMs) via NFκB–CCL2/CXCL1 axis32, 33, 34; 7) inhibits tumor angiogenesis via inducing the accumulation of RhoA³⁵; and 8) suppresses the extravasation of cancer cells via disrupting actin cytoskeleton formation²⁵. Furthermore, the inactivation of neddylation also induces protective autophagy in tumor cells via 1) inducing the accumulation of DEPTOR (a well-known inhibitor of mTORCs)³⁶; 2) inhibiting the degradation of HIF1α to activate REDD1–TSC1 pathway³⁷; and 3) blocking the degradation of IκBα to activate the ROS–ATF3 pathway³⁸ (Fig. 3).

Figure 3.

Regulatory and pro-tumoral mechanisms of neddylation pathway. N8: NEDD8, neural precursor cell expressed, developmentally down-regulated gene 8; TSME: tumor suppressive microenvironment.

2.2. Functions and regulatory mechanisms of E2s in cancer

Up to 33 Ub-associated E2s have been identified, which share the common catalytic ubiquitin-conjugating (UBC) domain³⁹. Through the UBC domain, a free E2 interacts with E1 to afford an activated Ub from the E1–Ub adduct. After coupling with a specific E3, E2–Ub transfers Ub to the target substrates to form Ub chains³⁹. Notably, the type of Ub chains is determined by E2. For example, UBE2C and UBE2S are specific Lys11-linked ubiquitin-conjugating enzymes40, 41, 42, 43, whereas UBE2D and UBE2R1 catalyze the formation of Lys48-linked polyubiquitin chain^44,45.

UBE2C, also known as UbcH10, delivers Ub to the substrate, which is followed by UBE2S-driven Ub extension to form the Lys11-polyubiquitination chain40, 41, 42, 43. Both UBE2C and UBE2S are overexpressed in various cancer cell lines and tumor tissues^46,47 and facilitate malignant transformation of the cells or tumor growth by promoting the degradation of tumor-suppressing substrates (e.g., p27 and p53)40, 41, 42, 43^,48, 49, 50, 51. UBE2C transgenic mice is susceptible to developing spontaneous tumors when exposed to certain carcinogens (e.g., anthracene derivatives)⁴⁶. UBE2C and UBE2S are also activated by certain oncogenes, such as estrogen receptor 1 and androgen receptor (AR)^50,52, AKT1⁵³, APC/CCDH1, and Emi1^42,54, whereas they are suppressed by the tumor suppressor p53^50,52. Interestingly, there appears to be some divergence in the tumor-promoting functions of UBE2S and UBE2C in cancers. For instance, it has been observed that the deletion of UBE2C, but not UBE2S, regulates the degradation of DEPTOR, which in turn suppresses lung cancer cell growth⁵⁵.

UBE2D (also known as UBCH5) efficiently delivers Ub to the substrate, while UBE2R1 (also known as Cdc34) promotes the extension of the Lys48-linked polyubiquitin chain^44,45. According to clinical tumor sample analysis, UBE2D and UBE2R1 are overexpressed in numerous tumor tissues, such as lung cancer^56,57, hepatocellular carcinoma^58,59 and multiple myeloma⁶⁰, and regulate the ubiquitination and degradation of several essential tumor-suppressive substrates, including IKBα, cyclin-dependent kinase inhibitor p21 and p27^45,61, 62, 63. Additionally, UBE2R1 stabilizes the epidermal growth factor receptor by competing with UBE2D-matched Casitas B-lineage lymphoma (c-Cbl), thereby promoting the proliferation of lung cancer cells⁵⁶. This research reveals a previously unknown function of E2s, describing that they can directly stabilize substrates instead of acting as Ub carriers to promote the ubiquitination and degradation of substrates.

In contrast to the Ub pathway, neddylation involves only two E2 enzymes, UBE2M and UBE2F. In general, UBE2M pairs with RBX1 to promote the neddylation of cullin 1, 2, 3, 4A and 4B, while UBE2F pairs with RBX2 to enhance the neddylation of cullin 5. Our studies, as well as other studies, have demonstrated that UBE2M and UBE2F are up-regulated in various tumors^17,64,65. Consequently, the overexpression of UBE2M primarily leads to the accumulation of substrates on the cullin 1, 2, 3, 4A or 4B (e.g., ATF4, p21, p27, CDT1 and ORC1), whereas the overexpression of UBE2F promotes the accumulation of cullin5 substrates (e.g., NOXA) (Fig. 3). Interestingly, there is a cross-talk between these two independent E2s. The activated UBE2M serves as a dual E2 to induce the proteolysis degradation of UBE2F, thereby inducing the accumulation of pro-apoptosis protein NOXA, which promotes apoptosis in lung cancer cells⁶⁶.

2.3. Functions and regulatory mechanisms of E3s in cancer

More than 600 E3 ligases have been identified in humans so far, which contribute to the functional diversity and substrate specificity of ubiquitination. According to their structures and functions, E3 ligases are grouped into three main categories: really interesting new gene (RING), homologous to E6-associated protein C-terminus (HECT), and RING-in-between-RING (RBR) E3 ligases (Fig. 4). Especially, RING E3 ligases, the largest E3 ligases with more than 600 members, were summarized in details in this section.

Figure 4.

Structures and components of RING, HECT and RBR E3 ligases. According to the active form, RING E3s are divided into 2 groups, including the single-subunit RING E3 ligases and multiple subunit E3 ligases.

2.3.1. Representative RING E3s in cancer

RING E3 ligases share a common RING domain or U-box domain for binding with E2–Ub thioester. RING E3s typically transfer Ub directly from E2 to the targeted substrates. All RING E3 ligases have an E2 binding domain, but not every RING E3 ligase possesses the substrate recognition domain. Therefore, RING E3 ligases are divided into two groups: single-subunit RING E3 ligases (e.g., c-Cbl, MDM2, and IAP) and multi-subunit RING E3 ligases (e.g., Cullin–RING ligases [CRLs])⁶⁷ (Fig. 4). CRLs, also known as the largest RING E3 ligases, comprise several subunits, including RING, cullin, adaptor, and/or substrate receptor proteins⁶⁸ (Fig. 4). RBX1 and RBX2, also known as the two classical RING proteins of CRLs, deliver Ub or Ubl from the E2–Ub or E2–Ubl complex to the substrate⁶⁸. Cullins constitute the skeletal proteins of CRLs that assemble the RING, adaptor, and/or receptor proteins to deliver Ub to the recruited substrates⁶⁸. Eight distinct cullins paired with RBX1 or RBX2 form CRL1, CRL2, CRL3, CRL4A, CRL4B, CRL5, CRL7, and CRL9⁶⁸, which are primarily involved in regulating certain specialized cellular functions during tumor initiation and progression.

2.3.1.1. CRL1 E3 ligases

CRL1, also known as SKP1-Cullin1-F-box protein (SCF), comprises four subunits: RBX1, cullin1, adaptor protein Skp1, and substrate receptor F-box proteins. F-box proteins have two essential functional domains, the F-box motif for Skp1-mediated cullin1 recruitment and the C-terminal domain for substrate recognition. Based on the latter, F-box proteins are divided into three categories: 10 members of FBXWs with WD40 repeats, 22 members of FBXLs with leucine-rich repeats (LRR), and 37 members of FBXOs with a different domain or no recognizable motifs^69,70. Among them, S-phase kinase-associated protein 2 (Skp2), F-box and WD repeat domain-containing 7 (FBXW7), and β-transducin repeat-containing E3 ubiquitin protein ligase (β-TRCP) are well-characterized F-box proteins in cancer studies (Fig. 5).

Figure 5.

Representative receptors or adaptors of CRL E3 ligases and their role in tumor initiation and progression. Green: tumor suppressive role; Orange: tumor promoting role; Blue: dual roles in a context-dependent manner.

In 1995, Skp2 (also known as FBXL1, FBL1, and p45) was identified as an essential cytokinetic regulator of the cyclin A–CDK2 complex for S-phase entry⁷¹. In the past 26 years, Skp2 has been established as an oncogene that promotes the progression of various tumors, such as those of lung cancer⁷², nasopharyngeal carcinoma⁷³, hepatocellular carcinoma⁷⁴ and gastric carcinoma⁷⁵. Mechanistically, Skp2 recognizes substrates via its LRR domain in a phosphorylation kinase-dependent manner, subsequently promoting the formation of a Lys48-linked polyubiquitin chain for substrate degradation via the 26S proteasome or the formation of a Lys63-linked polyubiquitin chain for protein–protein interaction (PPI) and activation⁷. Therefore, Skp2 substrates are divided into two categories: proteolytic substrates, including p21⁷⁶, p27⁷⁷, p57⁷⁸, CDT1⁷⁹, ORC1⁸⁰ and FOXO1⁸¹ and nonproteolytic substrates, including AKT^73,82, YAP⁸³, MTH1⁸⁴, LKB1⁸⁵, NBS1⁸⁶ and Twist⁸⁷. Notably, proteolytic Skp2 substrates primarily exhibit tumor-suppressive activity, inhibiting tumor growth or promoting cell death pathways (e.g., apoptosis or senescence), whereas the nonproteolytic Skp2 substrates primarily promote tumor growth or metastasis⁷. Thus, Skp2 constitutes a promising pharmacological target in cancer therapy owing to its essential role in facilitating tumor initiation and progression.

FBXW7, also known as FBW7, CDC4, and SEL-10, is widely regarded as a tumor suppressor. Numerous FBXW7 substrates have been identified as oncoproteins that promote tumor growth, including Aurora A^88,89, BRAF⁹⁰, c-Jun^91,92, c-Myc^93,94, GFI1⁹⁵, HIF1α^96,97, SHOC2⁹⁸, KLF5^99,100, MCL1^101,102, NOTCH1103, 104, 105 and ZNF322A¹⁰⁶, further highlighting the tumor-suppressive property of FBXW7. FBXW7 encodes three isoforms, namely FBXW7α, FBXW7β, and FBXW7γ, which display distinct subcellular distributions. FBXW7α is a major isoform in human tissues and cells, located in the nucleoplasm; FBXW7β resides in the cytoplasm, whereas FBXW7γ is primarily found in the nucleolus^107,108. Therefore, different FBXW7 isoforms may cooperatively promote the ubiquitination of oncogenic proteins. Loss-of-function FBXW7 mutations have been frequently observed in tumors where they intensify tumor progression, chemoresistance, and radiation tolerance^{90,95,101,107,109,110}. Additionally, FBXW7 is activated by p53¹¹¹ and suppressed by some microRNA, such as miR-24¹¹², miR-27¹¹³, miR-32¹¹⁴, miR-223¹¹⁵ and miR-367¹¹⁶. Thus, targeting the negative regulators or substrates of FBXW7 can serve as an effective anticancer strategy.

β-TRCP1 (also known as FBXW1, FBXW1A, and FWD1) and β-TRCP2 (also known as FBXW11, FBXW1B, and HOS) are subordinate members of the β-TRCP family that exert context-dependent oncogenic or tumor-suppressive effects. For example, β-TRCP targets numerous tumor suppressors, including IκBα^117,118, Wee1¹¹⁹, p53¹²⁰, DEPTOR¹²¹, FBXW2¹²², BimEL¹²³, Mxi1¹²⁴, PDCD4¹²⁵, and pro-caspase3¹²⁶, for degradation to facilitate tumor growth, whereas β-TRCP ubiquitinates certain important oncoproteins, including HER2¹²⁷, YAP¹²⁸, MDM2¹²⁹, PD-L1¹³⁰ and MCL1¹³¹, to suppress tumor initiation or progression (Fig. 5). Clinical tissue analysis revealed that β-TRCP is upregulated in various tumors (e.g., colorectal cancer and hepatoblastomas) and is associated with decreased overall survival in patients with tumor^132,133, whereas somatic mutations of β-TRCP promote the progression of certain tumors (e.g., gastric and prostate cancers)^134,135.

2.3.1.2. CRL2 E3 ligases

CRL2 comprises RBX1, cullin2, adaptor protein Elongin B/C, and substrate recognition subunit¹³⁶. The most characteristic substrate recognition subunit of CRL2 is the Von Hippel-Lindau tumor suppressor (VHL). VHL was initially cloned in 1993¹³⁷, and subsequently its mutations were identified in most ccRCC cases in 1994¹³⁸. The classical substrates of VHL include hypoxia-inducible factors (HIFs)¹³⁶. Loss-of-function VHL mutations induce HIF accumulation, which further facilitates tumor angiogenesis by promoting HIF-mediated transcription of essential oncogenes, including VEGF and EPO¹³⁹. Loss-of-function VHL mutations and elevated HIF and VEGF levels have been frequently observed in tumors, especially in those of ccRCC140, 141, 142. Hence, targeting the VHL–HIF–VEGF axis constitutes a major therapeutic intervention strategy for ccRCC^142,143. In addition to HIFs, VHL substrates include Zinc fingers and homeoboxes 2 (ZHX2)¹⁴⁴, protein kinase C¹⁴⁵, β2-adrenergic receptor¹⁴⁶, and the RNA polymerase II subunit hsRPB7¹⁴⁷. ZHX2, an NF-κB transcriptional activator, offers an additional therapeutic option for patients with HIF-insensitive or -resistant ccRCC¹⁴⁴.

2.3.1.3. CRL3 E3 ligases

Unlike other CRLs, CRL3 comprises RBX1, cullin3, and broad-complex-tramtrack-bric-α-brac (BTB) adaptor protein. BTB proteins not only directly bind to cullin3 but also recognize substrates via their BTB domain (also known as the POZ domain). Functionally, BTB proteins facilitate the 26S proteasome-mediated substrate degradation by promoting the formation of Lys48-linked polyubiquitin chains. Conversely, BTB proteins catalyze Lys63-and Lys33-linked ubiquitination, affecting PPI and subcellular localization¹⁴⁸. To date, 188 putative BTB proteins have been identified in the human genome, which are grouped into five subfamilies: the BTB-kelch (n = 55), BTB–Zn finger (n = 49), K⁺ voltage-gated channel (n = 27), and KCTD (n = 25) subfamilies and others (n = 32)¹⁴⁸. Kelch-like ECH-associated protein 1 (Keap1) and speckle-type BTB/POZ protein (SPOP) constitute two classical substrate adaptor proteins of CRL3 (Fig. 5).

Keap1 was identified as a nuclear factor erythroid 2-related factor 2 (Nrf2) inhibitor. Nrf2 is a key regulator of cellular antioxidant response that stimulates the transcription of >200 antioxidant genes, including HO1 and NQO1¹⁴⁹. Previous studies have reported that Nrf2 is ubiquitinated by cytoplasmic Keap1 under normal conditions. Conversely, the oxidation of Keap1 by reactive oxygen species or electrophiles failed to ubiquitinate Nrf2, inducing its accumulation. The accumulated Nrf2 translocated into the nucleus, thereby increasing the transcription of antioxidant genes and preventing oxidative damage to normal cells and tissues. Thus, Nrf2-knockout mice are susceptible to spontaneous tumors^150,151. However, numerous studies have shown that Nrf2 hyperactivation protects cancer cells from chemotherapeutic agents or radiotherapy-induced oxidative stress¹⁴⁹.

SPOP regulates substrate ubiquitination and affects the progression of several human cancers, including lung cancer¹⁵², prostate cancer153, 154, 155, 156, RCC^157,158, endometrial cancer^155,159 and breast cancer¹⁶⁰. Generally, SPOP acts as a tumor suppressor in multiple cancers (e.g., lung and prostate cancers) by mediating the degradation of several oncoproteins, including ERG¹⁵⁶, PD-L1¹⁶¹, cyclin E1¹⁶², ATF2¹⁶³, c-Myc¹⁶⁴, AR¹⁶⁵, steroid receptor coactivator 3¹⁶⁶, BRD2, BRD3 and BRD4153, 154, 155. However, SPOP exhibits an oncogenic function in RCC by mediating the degradation of phosphatase and tensin homologue (PTEN), proapoptotic molecule Daxx, Gli2, DUSP6, and DUSP7¹⁵⁷. The subcellular localization of SPOP primarily affects its dual functions. SPOP accumulates in the nucleolus of cells to mediate oncoprotein degradation^154,167,168. Conversely, SPOP demonstrates predominantly cytoplasmic localization in ccRCC due to the VHL mutations and hypoxic conditions in ccRCC, consequently promoting the degradation of cytoplasmic tumor suppressors (e.g., PTEN, Daxx, Gli2, DUSP6, and DUSP7)¹⁵⁷.

2.3.1.4. CRL4 E3 ligases

CRL4 is subdivided into CRL4A and CRL4B based on the skeleton proteins of cullin 4A and cullin 4B, respectively. Damage-specific DNA-binding protein 1 (DDB1) is an essential adaptor protein of CRL4 that detects and repairs damaged DNA. Meanwhile, DDB1-cullin4 associated factors (DCAFs) serve as substrate receptors^169,170. Three representative DCAFs in anticancer therapy are damage-specific DNA-binding protein 2 (DDB2), denticleless E3 ubiquitin protein ligase homolog (DTL), and cereblon (CRBN) (Fig. 5).

DDB2 is a WD40 repeat-containing protein that heterodimerizes with DDB1 to repair DNA damage or inhibit the transcription of certain oncogenes. When DDB2 is directly linked to damaged DNA, it assembles the CRL4 complex to ubiquitylate histones H2A¹⁷¹, H3, or H4¹⁷², thereby facilitating DNA repair by weakening the histone–DNA interaction. Furthermore, DDB2 was ubiquitinated by CRL4A at the lesion site^173,174, thus inducing the accumulation of damage recognition factor XPC to remove damaged pyrimidine dimers^174,175. Therefore, DDB2 knockout mice are susceptible to skin carcinogenesis, and the upregulation of DDB2 suppresses the onset and progression of UV-induced squamous cell carcinoma176, 177, 178. In addition, DDB2 inhibits the transcription of HIF1α¹⁷⁹, ALDHA1¹⁸⁰, Snail and ZEB1¹⁸¹, thus inhibiting tumor growth or epithelial-to-mesenchymal transition.

As a potential oncoprotein, DTL (also known as CDT2, DCAF2, RAMP, and L2DTL) recognizes and ubiquitinates its substrates to master genome stability^182,183. Inducing the accumulation of CDT1¹⁸⁴, p21¹⁸⁵ and PR-set7/set8¹⁸⁶ in the S phase in a proliferating cell nuclear antigen (PCNA)-dependent manner, CDT2 knockdown inhibits DNA replication. Knockdown of CDT2 triggers G2 cell cycle arrest by preventing the degradation of CHK1 in a PCNA-independent manner¹⁸⁷. In contrast, overexpression of CDT2 accelerates the degradation of the substrate, CDT1, subsequently promoting cancer cell re-replication and impairing senescence or apoptosis¹⁸⁸.

CRBN, an important substrate recognition receptor of the CRL4, is the primary target of immunomodulatory drugs (IMiDs), such as thalidomide and its derivatives (e.g., pomalidomide, lenalidomide, CC-885) for hematological malignancy treatment. CRBN facilitates the ubiquitination and degradation of the Ikaros family (IKZF1 and IKZF3) in a lenalidomide-dependent manner189, 190, 191 and that of ARID domain-containing protein 2 (ARID2) in a pomalidomide-dependent manner to suppress the survival and proliferation of multiple myeloma cells¹⁹². In acute myeloid leukemia cells, CRBN promotes the degradation of G1 to S-phase transition protein 1 to exert a broad-spectrum anticancer effect in a CC-885-dependent manner¹⁸⁹. Furthermore, CRBN stabilizes CD147 and monocarboxylate transporter 1 (MCT1) to promote angiogenesis, tumor growth, or lactate export, whereas IMiDs interact with CRBN to disrupt the CD147–MCT1 complex and exert antitumor and teratogenic effects¹⁹³. Interestingly, CRBN also functions as a transmembrane protein-specific co-chaperone molecule of HSP90 to modulate HSP90–AHA1 activity, whereas IMiDs inhibit their interaction to suppress multiple myeloma cell growth¹⁹⁴.

2.3.1.5. CRL5 E3 ligases

CRL5 comprises the scaffold protein cullin 5, RING protein RBX2, adaptor protein Elongin B/C, and substrate recognition receptor protein SOCS. The primary substrate receptors of CRL5 include 37 members, categorized into 5 types: SOCS, ASB, SPSB, and WSB subfamilies and others (such as Rab40 and MUF1)¹⁹⁵.

Suppressor of cytokine signaling 1 (SOCS1, also known as SSI1) and SOCS3 (also known as SSI3) are two extensively studied SOCS proteins of CRL5 that suppress inflammatory reaction and exert context-dependent effects in tumorigenesis¹⁹⁵. SOCS1 depletion in macrophages¹⁹⁶ or dendritic cells¹⁹⁷ in conditional knockout mice enhances antitumor inflammation or antigen-specific antitumor immunity. Conversely, SOCS3 or cullin 5 depletion inhibits integrin β1 degradation, subsequently activating the focal adhesion kinase/SRC (FAK/SRC) signaling pathway, thereby promoting small-cell lung cancer metastasis¹⁹⁸. Similarly, SOCS1 inhibits STAT3 phosphorylation and limits granulocyte–macrophage colony-stimulating factor and interleukin-6 production, thus inhibiting myeloid-derived suppressor cell differentiation and increasing antitumor immunity¹⁹⁹. Additionally, elevated SOCS3 levels improve the susceptibility of castration-resistant prostate cancer cells toward natural killer cells²⁰⁰. Consistent with these findings, low SOCS1 and SOCS3 expression levels have been frequently observed and positively associated with poor prognosis in patients with small-cell lung¹⁹⁸ and breast cancer²⁰¹.

SplA/ryanodine receptor domain and SOCS box-containing 1 (SPSB1, also known as SSB1) ubiquitinates a key inflammatory effector, inducible NO synthase (iNOS)^202,203. iNOS has multiple cellular origins and harbors both tumor prosurvival and suppressive functions²⁰⁴. In addition to iNOS, SPSB1 promotes p21 proteasomal degradation to increase the viability and migration of ovarian cancer cells²⁰⁵. Furthermore, SPSB1 promotes breast cancer recurrence by preventing chemotherapy- or HER2/neu inhibitor-induced apoptosis in tumor cells²⁰⁶.

2.3.2. HECT E3s in tumorigenesis

In contrast to RING E3s, HECT ligases ubiquitinate their substrates in two steps. They first load the activated Ub onto themselves to produce the E3–Ub intermediate via the HECT domain, and then they transfer Ub onto the substrates⁵ (Fig. 4). HECT E3s are divided into three subfamilies based on the N-terminal structure of the HECT domain: 1) Nedd4 subfamily, also known as C2–WW–HECT, whose structures contain two to four tryptophan–tryptophan (WW) domains; 2) HERC (HECT and RCC1-like domain) subfamily, which contains one or more RCC1-like domains; and 3) other HECT E3s, such as E6AP, HECTD2, G2E3, TRIP12, EDD, HACE1, and HUWE1⁵ (Fig. 4). NEDD4, one of the brain development regulators, is the first Nedd4 subfamily member to be identified²⁰⁷. In tumors, NEDD4 ubiquitylates and promotes degradation of PTEN, an essential tumor suppressor, to promote tumor growth²⁰⁸, whereas CK1α competitively antagonizes NEDD4-mediated PTEN ubiquitination to inhibit lung tumor growth²⁰⁹. In addition, NEDD4 promotes the formation of K63-linked polyubiquitination chains of MDM2 to stabilize MDM2, thus promoting the degradation of p53 and inhibiting the DNA damage response²¹⁰. Furthermore, NEDD4 promotes the proteasomal degradation of VDAC2/3, the voltage-dependent anion channel, to suppress erastin-induced ferroptosis²¹¹. On the contrary, the most prominent feature of NEDD4-null mice is delayed embryonic development and growth as a result of a decrease in insulin and insulin-like growth factor 1 (IGF-1)²¹². Meanwhile, knockout of NEDD4 upregulates PDL1 and inhibits CD8⁺ T cell infiltration in bladder cancer cells to avoid immune surveillance²¹³. Besides, NEDD4 inhibits RBX2 to sensitize etoposide-induced apoptosis²¹⁴. This dual function of NEDD4 in tumors justifies further research into NEDD4 for future anticancer therapy.

2.3.3. RBR E3s in tumorigenesis

RBR E3s, possessing the characteristics of both RING and HECT E3s, comprise a RING1 domain, an in-between-RING (IBR) domain, and a RING2 domain^6,215. RBR E3s catalyze the ubiquitination reaction in three steps. They first recognize the E2–Ub complex with the RING1 domain, then transfer Ub onto RING2 to form the E3–Ub intermediate, and finally transfer Ub onto the substrates^6,215 (Fig. 4). Herein, 14 RBR E3s, including Parkinson protein 2 (also known as Parkin), RNF19A, RNF19B, RNF144A, RNF144B, RNF216, RNF217, RBCK1, RNF31 (also known as HOIP), RNF14, ARIH1, ARIH2, PARC, and Ankib1, identified in humans (Fig. 4)⁶.

Parkin is a well-known RBR E3 ligase with neuroprotective function²¹⁶. Parkin is often downregulated in various tumors and demonstrates tumor-suppressive activity217, 218, 219. Parkin downregulation is typically associated with poor overall survival of patients with tumors^217,218. Parkin null mice are susceptible to developing carcinoma²²⁰. In mechanism, Parkin facilitates apoptosis via promoting the degradation of the antiapoptosis protein MCL1 during mitochondrial depolarization²²¹, whereas Parkin depletion promotes the activation of PI3K–AKT pathway in a PTEN-dependent manner to facilitate tumorigenesis²¹⁸. In addition, Parkin suppresses Kras-driven pancreatic tumorigenesis and HIF1α-mediated Warburg effect by promoting the degradation of SLC25A37 and SLC25A28²²². Furthermore, Parkin promotes the degradation of phosphoglycerate dehydrogenase (PHGDH), the first rate-limiting enzyme of serine synthesis, to inhibit serine synthesis and tumor progression²²³. Additionally, Parkin promotes the formation of the K33-linked polyubiquitination chain of RIPK3 to inhibit necrosome assembly, hence preventing necroptosis and tumorigenesis²²⁴.

2.4. Functions of representative deubiquitinases (DUBs) in cancer

Protein deubiquitination is a reversal of ubiquitination that uses deubiquitinases (DUBs) to remove Ub from substrates. Based on their sequence and domain conservation, ∼100 identified DUBs have been classified into six families⁸. Five families of cysteine proteases include USPs, OTUs, UCHs, MJDs, and MINDYs^8,225. In addition, the JAB1/MPN/MOV34-domain-containing metalloproteases (JAMMs) comprise 16 zinc metallopeptidases^8,225. DUBs primarily remove Ub from substrates to modulate protein stability, enzymatic activity, and subcellular localization^8,225. Therefore, DUB disruption contributes to tumor initiation and progression by interfering with the dynamic equilibrium of ubiquitination.

Based on their roles in tumorigenesis, DUBs are classified into three types. The first type is prosurvival DUBs, such as USP2 and USP7 in the USP family and CSN5 in the JAMM family. USP2 is overexpressed in several human tumor tissues and is crucial for tumor growth as it inhibits the ubiquitination and degradation of oncoproteins, including Skp2²²⁶, Twist²²⁷, MDM2^228,229, TGFBR1²³⁰, fatty acid synthase (FAS)²³¹, PD-L1²³², and cyclin D1²³³. USP7 activation is associated with poor overall patient survival in multiple tumors. USP7 overexpression stabilizes numerous oncoproteins, such as EZH2²³⁴, PHF8²³⁵, MDM2²³⁶, LSD1²³⁷, β-Catenin²³⁸ and DNMT1^239,240, thereby stimulating tumor growth or drug resistance. CSN5 facilitates tumor growth by inhibiting the degradation of Snail²⁴¹, ZEB1²⁴², surviving²⁴³ and FOXM1²⁴⁴, whereas curcumin-induced CSN5 depletion decreases PD-L1 expression and enhances the anticancer efficacy of anti-CTLA4 drugs²⁴⁵.

Tumor-suppressive DUB families, CYLD lysine 63 deubiquitinase (CYLD) in the USP family and OTU deubiquitinase 1 (OTUD1) in the OTU family, constitute the second type of DUBs. CYLD inhibits the Wnt/β-catenin and TGF-β signaling pathways by regulating the Lys63-linked ubiquitination of ALK5, thereby suppressing tumor growth and cell invasion^246,247. OTUD1 deubiquitinates the TGF-β inhibitor SMAD7 and inhibits p53, thereby increasing the cleavage of caspase-3 and PARP-dependent apoptosis and enhancing tumor suppression^248,249.

Context-dependent DUBs, such as OTUD3 and USP10, constitute the third type of DUBs that exhibit contradictory functions in various tumors. OTUD3 regulates PTEN stability to inhibit PI3K/AKT signaling transduction and tumorigenesis in breast cancer²⁵⁰. However, OTU3 stabilizes GRP78 and promotes lung cancer cell growth and migration²⁵¹, suggesting that the tumor-suppressive or -proliferative activity of OTU3 depends on the tissue specificity of the tumor²⁵¹. USP10 typically functions as a tumor suppressor in lung cancer, RCC, and colon cancer, and as a tumor promoter in acute and chronic myeloid leukemia; however, in hepatocellular carcinoma, USP10 exerts multiple activities: a) it reverses the MDM2-induced nuclear export and degradation of p53 and reduces the progression of RCC²⁵²; b) it blocks the degradation of the canonical tumor suppressor PTEN and KLF4 and inhibits lung cancer growth^253,254; c) it antagonizes the transcriptional activity of the oncogene c-Myc by stabilizing SIRT6 and p53, thereby promoting cell cycle arrest and inhibiting colon cancer cell proliferation²⁵⁵; d) it stabilizes the oncoproteins FLt3 and Skp2 to facilitate the growth of myeloid leukemia^256,257; and e) in hepatocellular carcinoma, it removes the Ub chains on YAP/TAZ to promote the growth of the HepG2, SNU387, and Li7 cell lines²⁵⁸, whereas it stabilizes PTEN and AMPKα to suppress the growth of the Huh7, HCCLM3, MHCC97L, and Bel7402 cell lines²⁵⁹.

3. Multiomics analyses empower anticancer target identification in Ub and Ubl pathways

As mentioned above, Ub and Ubl pathways are attractive anticancer targets. However, the complexity of these pathways, comprising hundreds of Ub-associated enzymes that form a hierarchical task network to modulate substrates, complicates our understanding of their significance in cancer. Thus, there remains a need for a comprehensive approach to identify potential anticancer targets within the Ub and Ubl pathways. With the advancements of high-throughput sequencing and computational tools, multiomics analyses have emerged as valuable resources for unraveling the complexity of cancer. In the following sections, we will delineate the role of Ub and Ubl-associated enzymes in tumorigenesis by multiomics analyses. Furthermore, we will outline their advance in the identification of potential anticancer targets and application for precise treatment. The narrative order and associated subtitle are expanded in the order of E1, E2 and E3.

3.1. Multiomics analyses of E1s reveals that overexpression of UBA1 and UBA6 represents poor overall survival of tumor patients

First, we investigated the association between E1s in tumors and in normal tissues alongside their prognostic value using the Cancer Genome Atlas (TCGA) and Clinical Proteomic Tumor Analysis Consortium (CPTAC) databases from the UALCAN website^260,261. Compared with normal tissues, >13 tumor types expressed high UBA1 and UBA6 mRNA levels, whereas only 3 tumor types exhibited low UBA1 and UBA6 mRNA levels (Fig. 6A and B). At the protein level, UBA1 and UBA6 were overexpressed in colon cancer and head and neck squamous cancer (Fig. 6C and D). These findings align with the protumoral functions of UBA1 and UBA6, as evidenced by the following observations: 1) targeting UBA1 via pharmacological and genetic approaches significantly inhibits tumor cell growth, including lung cancer, colon cancer, liver cancer, acute myeloid leukemia, non-Hodgkin lymphoma, melanoma^11,262, 263, 264, 265; 2) inhibition of UBA1 overcomes drug resistance by inducing ER stress or apoptosis in myeloma²⁶⁶; 3) overexpression of UBA6 promotes tumor formation²⁶⁷. Kaplan–Meier analysis revealed that higher UBA1 in patients was associated with poorer overall survival in prostate adenocarcinoma, hepatocellular carcinoma, and lung adenocarcinoma (LUAD) (Fig. 7A). Similarly, patients with LUAD, kidney chromophobe, and ccRCC exhibited worse prognoses in case of high tumoral UBA6 levels (Fig. 7B).

Figure 6.

Multiomics analyses of E1s. (A–D) Gene expression levels of UBA1 and UBA6 in 24 tumors and normal tissues using TCGA database and CPTAC database in UALCAN website. Statistical significance was determined by the Mann–Whitney test (two-tailed): ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, n.s. indicates no significant difference.

Figure 7.

Kaplan–Meier analyses of UBA1 and UBA6 using TCGA database in UALCAN website.

3.2. Multiomics analyses of E2s reveals the potential of UBE2C as a broad-spectrum anticancer target in tumors

The expression levels and clinical importance of ubiquitination-related E2s were determined using transcriptomic, proteomic, and clinical data from patients with tumors^260,261. The mRNA levels of 27 E2s (81.81%) were significantly upregulated in the tumor tissues of patients with LUAD than that in normal tissues, whereas the mRNA levels of 23 E2s (69.70%) were significantly upregulated in the tumor tissues of patients with lung squamous cell carcinoma (P ≤ 0.05) (Fig. 8A and B). Conversely, less than six E2s were significantly downregulated in the tumor tissues of patients with LUAD and lung squamous cell carcinoma (P ≤ 0.05, respectively) (Fig. 8A and B). Kaplan–Meier analysis revealed that higher UBE2C, UBE2O, UBE2S, UBE2T, UBE2V2 and UBE2Z in patients was associated with poorer overall survival in LUAD (Fig. 8A and B).

Figure 8.

Multiomics analyses of E2s. (A, B) mRNA levels and Kaplan–Meier analyses of ubiquitination-associated E2s in lung adenocarcinoma and lung squamous cell carcinoma.

The UBE2C mRNA levels were 18- and 30-fold higher in the tumor tissue of patients with LUAD than that in normal tissues and those with lung squamous cell carcinoma tissue, respectively. An overall survival analysis revealed that UBE2C (HR = 2.50) was risk factors for LUAD (P ≤ 0.05), but not in lung squamous cell carcinoma through the analysis of the TCGA database (Fig. 9A and B). UBE2C has been identified as an essential factor in Kras^G12D-induced lung cancer⁵⁵. In addition, in our multiomics analyses, UBE2C was considerably and consistently upregulated in the tumor tissue of patients within the enrolled 10 tumors, including lung cancer, than in the normal tissues at protein levels (Fig. 9A). UBE2C, in conjunction with the APC/C E3 ligase, promotes ubiquitylation and degradation of tumor suppressive proteins (e.g., p27, DEPTOR and p53)40, 41, 42, 43^,48, 49, 50, 51^,55. The mRNA levels of UBE2S, which cooperate with UBE2C for Lys11-polyubiquitination chain extension (as discussed in detail in the previous section), were significantly upregulated in the tumor tissue of patients with LUAD and lung squamous cell carcinoma tissue than that in normal tissues (Fig. 8A and B). An overall survival analysis showed that the hazard ratio of UBE2S was up to 2.71 in LUAD (Fig. 8A). Consistently, Zhang et al.⁵⁵ reported that overexpression of UBE2C or UBE2S is related with the poor overall survival over 400 LUAD cases. Further protein level analysis showed that UBE2S was significantly upregulated in the tumor tissue of patients than in the normal tissues in ovarian cancer, head and neck cancer and glioblastoma, but not in lung cancer (Fig. 9B). Knockdown of UBE2C, but not UBE2S, inhibits the growth of lung cancer cells and Kras^G12D-induced lung carcinoma⁵⁵. This intriguing pattern suggests that UBE2C, but not UBE2S might be an attractive target for lung cancer.

Figure 9.

Gene expression levels of UBE2C and UBE2S in 10 tumors and normal tissues by exploring the CPTAC database in UALCAN website. Statistical significance was determined by the Mann–Whitney test (two-tailed): ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, n.s. indicates no significant difference.

3.3. Multiomics analyses of representative E3s reveals their clinical significance and regulatory mechanisms

F-box proteins, serving as substrate-recognition subunits of CRL1 E3 ligases, play important roles in tumor initiation and progression^69,70. To comprehensively delineate the role of F-box proteins in tumorigenesis, multiomics analyses were performed, encompassing all F-box proteins (n = 69), with the aim of providing insight into the characteristics of RING E3 ligases in cancer. Transcriptomic analysis showed that thirty-three F-box proteins were upregulated and twelve F-box proteins were downregulated (P < 0.05) at mRNA levels in LUAD tissues compared to normal tissues (Fig. 10A). At the mRNA level, the extensively studied F-box protein Skp2 showed a 2.72-fold increase in LUAD. This finding aligns with previous studies that have confirmed the upregulation of Skp2 in LUAD^72,268. These consistent results between our omics analyses and published data further support the reliability of our findings. The expression of FBXO45, a potential oncogene that inhibits tumor cell death by degrading tumor suppressor FBXW7²⁶⁹, was increased by 2.10-fold in LUAD (Fig. 10A). Similarly, the expression of FBXW5, an underlying oncogene that targets kinesin-13 and LATS1/2 for proteasomal degradation^270,271, was increased by 1.36-fold (Fig. 10A). At the protein level, FBXW5 was significantly increased in ccRCC and uterine corpus endometrial carcinoma compared to normal tissues, whereas FBXO45 was predominantly upregulated in breast cancer, ccRCC, head and neck squamous cell carcinoma, and glioblastoma (Fig. 10B and C). Consistently, FBXW5 and FBXO45 are overexpressed in numerous tumors and exhibit oncogenic properties^270,271. There were lower mRNA and protein levels of FBXW2 (Fig. 10D). The downregulation of FBXW2 leads to the inhibition of ubiquitylation and degradation of β-catenin and Skp2, thereby promoting lung cancer cell migration and invasion^122,272. In addition, the mRNA level of FBXL21 was 14.59-fold higher in LUAD tissues than in normal tissues, followed by FBXO43 (6.05-fold), FBXO32 (5.70-fold), CCNF (also known as FBXO1, 3.78-fold), FBXO41 (2.99-fold), FBXL13 (2.86-fold), and FBXL6 (2.08-fold) (Fig. 10A). These analyses offered a comprehensive perspective on the mRNA levels of F-box proteins in LUAD tissues compared to normal tissues.

Figure 10.

Multiomics analyses of F-box proteins (n = 69). (A) mRNA levels and prognostic values of F-box proteins in lung adenocarcinoma. (B–D) Protein expression levels of FBXW5, FBXO45 and FBXW2 in tumors and normal tissues by exploring the CPTAC database in UALCAN website.

High protein levels are regulated at multiple layers, including genetic amplification, transcriptional, translational, and post-translational levels. Consequently, we examined genetic alterations of Skp2, FBXW7, β-TRCP1 (gene name: BTRC), β-TRCP2 (gene name: FBXW11), FBXW5, FBXO45 and FBXW2 in thirty-three tumor types using the cBioPortal website (Fig. 11). Among these six F-box proteins, SKP2 and FBXO45 showed a higher amplification rate, whereas FBXW7 was the most frequently mutated gene (Fig. 11A). FBXW7, a well-known tumor suppressor that has been discussed in detail in the previous section, was exhibited the highest frequency of endometrial cancer (mutation rate of 19.97%), colorectal cancer (14.81%) and cervical cancer (11.78%) (Fig. 10B). These mutation patterns of FBXW7 were consistent with previous studies²⁷³. The mutations primarily occurred in the F-box domain and the substrate recognition WD40-repeat domain, particularly at three arginine residues (R465 C/H/G/L/P, R479Q/∗/G/P/L, and R505 G/C/H/L) in the WD40-repeat domain (Fig. 11C), which are essential for recognizing and binding substrates¹⁰⁷. Additionally, Skp2, a well-established oncogene, was found to be amplified in non-small cell lung cancer (amplification rate of 9.12%), bladder cancer (6.81%), and esophagogastric cancer (6.27%) (Fig. 11D). FBXO45 surpassed Skp2 with an amplification rate of 15.75% in ovarian epithelial tumor, followed by non-small cell lung cancer (14.72%) and cervical cancer (14.14%) (Fig. 10A and E). These results suggest that the overexpression of Skp2 and FBXO45 may be attributed to genetic amplification.

Figure 11.

Multiomics analyses of F-box proteins. (A) Gene alterations of 7 F-box proteins were determined in 33 tumors via the exploration of cBioPortal website. (B) Cancer type summary about altered SKP2, FBXO45 and FBXW7. (C) Mutation site analysis of FBXW7 by exploring the cBioPortal website. (D, E) Cancer type summary about altered SKP2 and FBXO45.

4. Overview of Ub and Ubl targeted drug discovery

The Ub and Ubl pathways are integral components of the ubiquitin–proteasome system (UPS) that is responsible for the degradation of over 80% of cellular proteins. In the past two decades, accumulated evidence confirmed that targeting 26S proteasome (the downstream of UPS) had been viewed as a desirable outcome of anti-cancer therapeutics. This endeavor resulted in U.S. Food and Drug Administration (FDA) approval of the first proteasome inhibitor (PI) bortezomib (also called velcade) in 2003, for treatment of multiple myeloma and relapsed mantle cell lymphoma. Subsequently, other PIs, such as ixazomib, delanzomib, and carfilzomib, were approved, further validating anticancer reliability of UPS suppression in clinic. However, the application of PIs is hindered by limitations, including targeting specificity, oral stabilization, and low penetration into solid tumors, consistently constrain the application of PIs. For instance, PIs have shown limited effectiveness in solid tumors, even when combined with other antitumor agents. Recent rapid advances in the functions and mechanisms of Ub and Ubl pathway, offer potential solutions to address these challenges.

In this section, we categorized the current reported small molecule inhibitors targeting essential components of the Ub and Ubl pathways, including E1s, E2s, E3s, DUBs, as well as their structural characteristics, clinical research progress and limitations. We also discuss the application of E3s ligands in the studies of target protein degradation (TPD). Especially, molecular glue degraders (MGD), as a classical TPD type, are considered as a prospective therapeutic strategy in cancer therapies, to be summarized the current progress in this section. Recognizing the growing importance of bioinformatics approaches in the pharmaceutical industry, we also discussed the utilization of bioinformatics resources and methods to facilitate the TPD drug discovery based on the multiomic data.

4.1. Representative E1 and E2 inhibitors

As mentioned above, ubiquitination has two essential E1s, UAE1 (also known as UAE) and UBA6. The first successful attempt at UAE inhibition dates back to the 1990s. A nonhydrolyzable ATP analog, adenosyl-phospho-ubiquitinol, was designed and synthesized to evaluate the inhibitory activity of the UAE–Ub adduct in an ATP-competitive manner (K_i = 50 nmol/L)²⁷⁴. This study first demonstrated the feasibility of selectively targeting the upstream of UPS, and developing small-molecular inhibitors of UAE. For a considerable amount of time, however, the majority of reported UAE inhibitors lacked targeting specificity and potent inhibitory effects against UAE (half maximal inhibitory concentration IC₅₀ > 1 μmol/L). Similar circumstances occurred during the development of Ubl inhibitors²⁷⁵. Not until 2009 did Soucy T.A. et al. develop the first NAE specific inhibitor, MLN4924 (also known as pevonedistat), with a new adenosine sulfamate (AdoS) skeleton (Fig. 12). MLN4924 is a potent and selective NAE inhibitor (IC₅₀ = 4 nmol/L) that is unrelated to other homologous enzymes, such as UAE, UBA6, and SUMO-activating enzyme (SAE) of SUMO (IC₅₀ = 1.50, 8.20, and 1.80 μmol/L, respectively)²⁷⁶. Furthermore, MLN4924 suppresses the growth of various cancer cell lines by inducing apoptosis, senescence, or sensitizing cancer cells to chemoradiation³³. The crystal structure of the MLN4924–NAE–NEDD8 ternary complex showed that MLN4924 not only binds with high affinity to the ATP-binding site of NAE but also forms a covalent cysteine thioester with NEDD8 via its terminal sulfamate moiety²⁷⁷.

Figure 12.

Chemical structures of AdoS-derived covalent inhibitors targeting E1 and their inhibitory mechanism.

The structure of MLN4924 inspired the researchers of Millennium Pharmaceuticals to successfully identify an AdoS analog, Compound 1, with potent UAE inhibiting activity²⁷⁸. Consistent with the mechanism of interaction between MLN4924 and the NAE complex, Compound 1 forms a covalent bond with Ub via its sulfamate group and then suppresses the formation of UAE–Ub thioester²⁷⁸. However, Compound 1 cannot be developed as a UAE inhibitor due to its nonselective effect on other E1 enzymes, including NAE and SAE. To develop highly selective UAE inhibitors, Hyer et al.¹¹ screened seven hundred small-molecular compounds and identified a potent, mechanism-based UAE inhibitor, TAK-243. Similar to Compound 1, the sulfamate moiety of TAK-243 blocks the covalent linkage between UAE and Ub. In contrast to Compound 1, TAK-243 not only inhibits UAE (IC₅₀ = 1 nmol/L), but also UBA6 (IC₅₀ = 7 nmol/L), NAE (IC₅₀ = 28 nmol/L), and SAE (IC₅₀ = 850 nmol/L)¹¹ at a lesser extent (Fig. 12). Accumulated evidence has shown that TAK-243 completely suppresses the cellular ubiquitylation and induces ER stress-mediated apoptosis¹¹. According to the data from the ClinicalTrials website, TAK-243 has now entered the clinical phase for treating patients with acute/chronic myeloid leukemia or myelodysplastic syndrome. Considering the potent SAE inhibition of Compound 1 and the high similarity between UAE and SAE, Langston et al.²⁷⁹ hypothesized that the sulfamate group of Compound 1 may also help covalently bind the SUMO–AdoS adduct. Then, several Compound 1 analogs were synthesized, and their SAE inhibitory effects were investigated using the homogenous time-resolved fluorescence assay. TAK-981 was identified to specifically bind the enzyme SAE (IC₅₀ = 1 nmol/L) more effectively than other E1 (IC₅₀, NAE = 0.96 μmol/L, UAE> 1 μmol/L). To date, TAK-981 has entered phase I clinical trials for patients with metastatic solid tumors in combination with pembrolizumab and non-Hodgkin's lymphoma in combination with rituximab (NCT03648372, NCT04074330, and NCT04381650)²⁷⁹.

Although E2s may constitute a promising target for anticancer therapeutics, comparatively fewer E2 small-molecular inhibitors have been reported than E1 inhibitors. Ceccarelli et al.²⁸⁰ screened and identified a small molecule, CC0651 that selectively inhibits the E2 Ub-conjugating enzyme Cdc34. The crystal structure of the CC0651–Cdc34A–Ub complex revealed that CC0651 stabilizes a low-affinity interaction with a composite binding pocket generated by Cdc34A and Ub²⁸¹. Similar to TAK-243, CC0651 induces the accumulation of p27 by inhibiting its degradation (IC₅₀ = 1.7 μmol/L) and suppresses the growth and proliferation of various cancer cell lines. Moreover, Tsukamoto et al.²⁸² identified leucettamol A (IC₅₀ = 105 μmol/L) from the Leucetta sponge as the first Ubc13 inhibitor, and subsequently isolated two more Ubc13 inhibitors, Manadosterols A and B, which are derived from the marine sponge Lissodendryx fibrosa (IC₅₀ = 90 nmol/L and 130 nmol/L, respectively). In contrast to several Ub- and NEDD8-related E2s, Ubc9 is the only E2 in the SUMOylation pathway. The submicromolar Ubc9 inhibitory activity of flavonoid analog 2D08 is utilized as a positive probe to identify other new Ubc9 inhibitors²⁸³.

4.2. Small-molecule ligands targeting RING E3 ligases

The Ub and Ubl pathways regulated the degradation of various substrates by establishing a PPI network involving human E3 ligase proteins as mentioned above. Over the past two decades, the development of PPI inhibitors has emerged as a viable and important technique for drug discovery targeting E3 ubiquitin ligases^8,284. However, these PPI inhibitors are still primarily based on the concept of “occupancy-driven” small-molecule modalities, also known as the inhibitor-centric approach. Unfortunately, only approximately 20% of the human proteome is druggable for an inhibitor-centric development approach. To address the critical limitations of the inhibitor-centric approach, the development of E3-associated modulators has introduced the concept of target protein degradation (TPD) as a novel pharmacologic modality²⁸⁵. Notably, proteolysis-targeting chimeras (PROTACs) have expanded the application of PPIs and provided insights into the degradation pathways for drug discovery286, 287, 288, 289. PROTACs are typically composed of an E3 small-molecule ligand, a connecting linker, and a target protein ligand²⁸⁷. The successful development of high-quality E3 small-molecule ligands is essential for PROTAC development. However, the lack of suitable E3 ligase ligands has been a major obstacle in the development of PROTAC degraders²⁹⁰. Fortunately, drug-like small-molecule ligands have been continuously to bind to the substrate receptor subunits of CRLs, such as CRL1^Skp2, CRL2^VHL, CRL3^Keap1, CRL3^SPOP, and CRL4^CRBN. Beyond that, ligands targeting other E3 ligases, such as MDM2 and cIAP, have been developed as E3 inhibitors or as part of PROTAC molecules^290,291. Among them, CRL4^CRBN and CRL2^VHL ligands have achieved notable success and are widely used in TPD studies and PROTAC design²⁹². In the following, we will highlight the recent progress in small-molecule ligands targeting RING E3 ligases.

4.2.1. Small-molecule ligands of cullin RING E3 ligases

Numerous studies have confirmed that CRL1^Skp2 inhibits apoptosis and cell cycle arrest by regulating the degradation of cell cycle inhibitors p21 and p27, suggesting that Skp2 is a promising anticancer target. Chan et al. conducted a high-throughput virtual screening to identify several potential Skp2 inhibitors²⁹³ and found that Compound 25 displayed better Skp2 inhibitory activity than the other hit compounds and suppressed the survival of cancer cells and cancer stem cells. This study was the first to demonstrate that the pharmacological inactivation of Skp2 constitutes a promising approach for cancer treatment²⁹³. Although Wu et al.²⁹⁴ had previously identified a series of Skp2 probes in silico, including thiazolidinedione derivatives C1 and C2, there was insufficient evidence to support their anticancer effect.

As a CRL2 substrate receptor, VHL regulates the degradation of HIF-1α and is involved in hypoxia adaptation. Previous studies have reported that residue hydroxy proline 564 (Hyp564) of HIF-1α is essential to bind VHL, thereby targeting Hyp of HIF-1α regulating its ubiquitin-mediated proteasomal degradation^295,296. Buckley et al.^297,298 modified the structure of Hyp in silico to rationally design Hyp analogs, discovering the first small-molecule inhibitor that suppressed the PPI of VHL and HIF-1α at submicromolar concentrations. Soares et al.²⁹⁹ further modified these Hyp analogs according to structure-guided and fragment-based drug design. Probe VH298 was identified as the first inhibitor with a double-digit nanomolar affinity for CRL2^VHL binding in vitro and in vivo²⁹⁹. Recently, a series of bifunctional small molecules containing two homologous VHL ligases, also known as Homo-PROTACs, were designed to dimerize the enzyme VHL and induce its self-degradation. The most active Homo-PROTAC, CM11, was designed as a selective and powerful VHL degrader at a concentration of 10 nmol/L³⁰⁰.

CRL3 possesses two primary E3 ubiquitin ligases, Keap1 and SPOP. CRL3^keap1 destabilizes its primary substrate Nrf2 to regulate the expression of antioxidant proteins and inflammatory regulators. Recently, CRL3^Keap1 has steadily emerged as an attractive target for cancer therapy alongside various other diseases, such as diabetes, Alzheimer's disease, and Parkinson's disease, which involve oxidative stress and inflammation³⁰¹. According to prior research, in a phase 3 trial, the clinical candidate CDDO-Me inhibited the interaction between cullin3 and Keap1. Additional research demonstrated that CDDO-Me covalently binds to the BTB domain of Keap1³⁰². Biogene conducted a high-throughput screening to identify small molecules that directly inhibited the PPI of Keap1 and Nrf2³⁰². In a homogeneous fluorescence polarization assay, more than 300 thousand compounds were screened and evaluated. As the most active compound, ML334 exhibited a high affinity for Keap1 (K_d = 1.00 μmol/L)³⁰³. Additionally, Jiang et al.³⁰⁴ modified a potential Keap1 inhibitor by screening the commercial Evotec Lead Discovery library and identified Compound 25 as an effective inhibitor of the interaction between Nrf2 and Keap1 with a half maximal effective concentration of 28.6 nmol/L. Recently, Astex Pharmaceuticals developed a series of novel oxathiazepin derivatives with potent affinity via fragment-based molecular modification and optimization. The analog KI-696 was inhibited the Keap1–Nrf2 interaction at a single nanomolar concentration (K_d = 1.30 nmol/L)³⁰⁵. SPOP with the BTB domain plays vital roles in the growth and progression of tumors by mediating the ubiquitination of several important substrates, such as PTEN and DUSP7^157,306,307 Jiang et al.^308,309 identified a series of small-molecule SPOP inhibitors by integrating virtual screening, pharmacophore modeling, and molecular docking. They further found optimal Compound 6b to inhibit SPOP-mediated PPI in ccRCC cells at a concentration of 10 μmol/L. Following optimization and SAR analyses, Compound 6lc was identified as having superior SPOP inhibition and ccRCC cytotoxicity to those of 6b^308,309.

4.2.2. Small-molecule ligands of single-subunit RING E3s ligases IAPs and MDM2 ligands

Unlike CRLs, IAPs and MDM2 are two classical single-subunit protein RING E3 ubiquitin ligases. The IAP family, which consists of cIAP1/2 (cellular inhibitor of apoptosis) and XIAP (X-linked inhibitor of apoptosis), induces apoptosis by various mechanisms, including receptor-mediated, mitochondria-mediated, and TNF factor receptor 1 (TNFR1)-mediated apotosis^310,311. The IAP family is typically composed of a ubiquitin-associated (UBA) domain, a RING domain, and multiple BIR domains (BIR1/2/3). Du et al.³¹² revealed that the second mitochondria-derived activator of caspase [SMACs, also known as direct IAP-binding protein with low pI (DIABLO)] has a promising endogenous antagonistic effect against IAPs via interacting with BIR2 and BIR3 domains. Extensive studies have shown that an N-terminal four peptide molecule (Ala1–Val2–Pro3–Ile4, AVPI) forms a similar binding model with SMAC (SMAC K_d = 420 nmol/L, AVPI K_d = 580 nmol/L)³¹³. The binding model showed that the amino group (–NH₂) and carbonyl group (–C O) of AVPI form multiple hydrogen bonds with the corresponding residues Glu314, Gln319, Trp310, Trp323, and Thr308. Based on the SARs of AVPI–BIR3, several SMAC mimetics were designed to assess the IAP inhibitory activity (Fig. 13). Genentech had advanced GDC-0152 and GDC-0917, the first-generation IAP inhibitors, through phase 1 clinical trials. Compared to pan-IAP inhibition of GDC-0152, GDC-0917 displayed a 250-fold more selectivity for cIAP over XIAP^314,315. In addition to GDC-0152 and GDC-0917, Novartis has also developed the SMAC analog LCL161, a 1st generation peptidomimetic that recently passed phase 2 clinical trials for patients with relapsed or refractory multiple myeloma³¹⁶. Subsequently, a series of peptidomimetic BIR3 IAP antagonists with bicyclic lactam skeletons were published by Sun et al.³¹⁷, which advanced the research of second-generation IAP antagonists. Cyclization of valine and proline formed [8,5]-bicyclic Debio-1143 (also known as Xevinapant), a 2nd generation IAP antagonist. Subsequent research showed that Debio-1143 possessed superior inhibitory selectivity (IC₅₀ XIAP and cIAP1/2 BIR3 = 66.4, 1.9, and 5.1 nmol/L, respectively) and pharmacokinetic properties than the majority of 1st generation inhibitors. Currently, a phase 3 clinical trial is recruiting patients with locally advanced squamous cell carcinoma of the head and neck treated with Debio-1143 in conjunction with platinum-based chemotherapy (NCT04459715).

Figure 13.

Various generations of IAP antagonists, 1st generation Smac peptidomimetic antagonist family, 2nd generation Smac peptidomimetic antagonist with bicyclic structure, 3rd generation Smac peptidomimetic antagonist family with bivalent structure, non-peptide IAP antagonists.

As described above, the BIR3 domain of IAPs binds to caspase-9 whereas the BIR2 domain binds to caspase-3/7³¹⁸. Previously reported monovalent antagonists always bind to the BIR3 domains of XIAP and c-IAP1, and the majority exhibit higher selectivity than BIR2 selectivity. Li et al.³¹⁹ proposed a 3rd generation SMAC mimetic with a distinct dimer skeleton that more effectively induced apoptosis in human cancer cells (Fig. 13). In addition, the results demonstrated that the bivalent antagonists had superior inhibitory efficacy against caspase-3/-7 and −9 by engaging both BIR2 and BIR3 domains of the IAP family. Inspired by the aforementioned discoveries, two representative bivalent IAP antagonists APG-1387³²⁰ and birinapant³²¹, were further developed by Ascentage and TetraLogic, respectively. APG-1387 had been applied to patients recruited in phase 1 clinical trials to treat advanced solid tumors or hematologic malignancies (NCT03386526). However, various druggability problems, including physicochemical features (high MW, high TPSA) and poor oral bioavailability, limited the use of either APG-1387 or birinapant in the treatment of solid tumors. Using fragment-based drug discovery, nonalanine small molecules, as a new class of IAP antagonists have recently been developed³²². Tolinapant (also known as ASTX660), a nonalanine and dual antagonist of XIAP and c-IAP1 (IC₅₀ = 2.80 and 0.22 nmol/L, respectively), was synthesized from a simple acetylamine fragment³²². Compared to previously reported antagonists, linapant exhibited reduced hERG inhibition (30% @ 30 μmol/L) and adequate oral efficacy.

MDM2 is another representative single protein RING E3 ligase³²³. MDM2 promotes p53 degradation to adversely regulate the tumor suppressor p53, thereby restraining p53-mediated cell cycle arrest and cell apoptosis (Fig. 14). In 2004, Roche reported the first selective MDM2 inhibitor, Nutlin-3a, that was identified from a series of cis-imidazoline-derived molecules³²⁴. Nutlin-3a inserts its two 4-chlorophenyl groups into the Trp23 and Leu26 pockets by mimicking the interaction of the p53 peptide, which directs its isopropoxy group toward the Phe19 pocket. Meanwhile, its cis-imidazoline scaffold replaces the helical backbone of the p53 peptide without polar hydrogen bonding interactions. According to the SARs, Roche further optimized the structure of Nutlin-3a to increase its potency and metabolic stability. As the first clinical MDM2 inhibitor, the analog molecule RG7112 had completed phase 1 clinical trials for patients with advanced solid tumors and hematologic neoplasms (NCT00623870, NCT00559533)³²⁵. According to a subsequent mechanistic investigation of the MDM2–p53 interaction, the indole ring of Trp23 residue in p53 is deeply inserted into a hydrophobic cavity on the surface of MDM2, and its NH group forms a hydrogen bond with residues in MDM2 (p53-Trp23/MDM2-Leu54)³²⁶ Inspired by this observational result, Ding et al.³²⁶ mimicked the structural interaction of Trp23 residue in p53 with MDM2 by deriving a unique spirooxindole alkaloids and subsequently identified a series of potent non-peptide MDM2 inhibitors³²⁷. Additionally, the optimal variant SAR405838 had completed phase 1 safety testing. In recent years, several analogs of SAR405838, such as RG7388103 from Roche³²⁸, APG-115 from Ascentage³²⁷, and DS-3032b from Daiichi Sankyo³²⁹, have entered clinical trials.

Figure 14.

Chemical structures of clinical MDM2 inhibitors.

Chiral drug design is important for the development of MDM2 inhibitors. In 2012, researchers at Amgen developed a series of novel piperidinone-derived MDM2 inhibitors via de novo design and chiral synthesis³³⁰. These chiral inhibitors maintain the binding model of previous molecules that occupy with the Trp23, Leu26, and Phe19 pockets of MDM2, as well as generating an electrostatic interaction between the carbonyl linker and the imidazole side chain of His96 in MDM2³³⁰. AMG232, the preferred compound, exhibits significant MDM2 inhibitory activity (surface plasmon resonance [SPR] K_D = 0.045 nmol/L, SJSA-1 EdU IC₅₀ = 9.10 nmol/L) as well as promising pharmacokinetics characteristics and antitumor activity in the SJSA-1 osteosarcoma mice-bearing xenograft model (ED₅₀ = 9.10 mg/kg)³³¹. Besides, Novartis developed a series of dihydroisoquinolinone p53-MDM2 inhibitors from the virtual screening of 50,000 compounds developed in house³³². Based on the binding pocket of p53-MDM2 by X-ray crystallography and molecular modeling, they identified a clinical candidate, NVP-CGM097³³³.

4.3. Progress and trend in the development of CRL E3 ligases-associated molecule glue degraders

Protein of interesting (POI) involves artificially inducing target protein degradation by the UPS, lysosome system, or even autophagy, which represents a R&D breakthrough of PPI to attract huge attention and funding from both academia and the pharmaceutical industry³³⁴. UPS-associated TPD involves two major approaches, PROTAC and MGD. In contrast to PROTACs, MGDs induce the proximity of ligase and POI by matching the protein surface topologies and modulating the binding attitude with ligase and POI. As a result, MGDs display comparable TPD ability while sharing the similar molecular weight and druggable structures with small-molecule agents. Although a rational approach had been reported previously to discover the MGD, largescale discovery of MGDs will require more innovations in screening assays and a deeper understanding of ligase associated PPI readout.

Over the past decades, the bioinformatics revolution has been the beneficiary of a multiomics data explosion, which contributes to improve cancer therapy strategies by extracted enormous useful information. Particularly, the multiomics data, such as genomics, transcriptomics, proteomics and metabolomics data, were generated by high-throughput technologies integrated with drug-response data, provides opportunities for identifying anticancer biomarkers and predicting drug responses. Moreover, multiple biological networks, including PPI network, drug-target network and disease-gene network, are extensively combined with multiomics data to analyze the potency of anticancer targets and facilitate PPI-based drug discovery. In the following, we want to focus on the progress in hijacking CRLs with MGDs, along with the identification of TPD via bioinformatics profiling.

4.3.1. Outline of CRL E3 ligases-associated molecule glue degraders

CRLs ubiquitinate diverse substrate proteins when brought into direct proximity to the ligase³³⁵, including CRL4^CRBN and several other substrates of the CRL4 family, such as CRL4^DCAF15 and CRL4^DDB1, attracting considerable attention from pharmaceutical companies owing to their efficacy in the discovery of MGDs (Table 1). In early 2010, CRBN was identified as an immune-regulated target of thalidomide and pomalidomide^336,337. However, the mechanism by which these thalidomide-like IMiDs regulate CRL4^CRBN, thereby inducing immunomodulatory and anti-inflammatory effects, remains unknown^336,337. Lenalidomide was not identified to selectively exploit the ubiquitination and degradation of two lymphoid transcription factors Ikaros (IKZF1) and Aiolos (IKZF3) till 2014^191,338. Celgene identified a potent CRL4^CRBN molecular glue, thalidomide-derived CC885, that selectively degraded more GSPT1 than IKZF1/3¹⁸⁹. Conversely, neither pomalidomide nor lenalidomide degraded GSPT1, indicating that specific neosubstrate recruitment characteristics are associated with structural differences across IMiDs. The aryl sulfonamide derivative indisulam is a clinical candidate exhibiting selective anticancer activity³³⁹. Han et al.³³⁹ identified the mechanism underlying the anticancer selectivity of indisulam, which recruited the premRNA splicing factor RBM39 (RNA-binding motif protein 39) to CRL4^DCAF15 for proteasomal degradation.

Table 1.

List of representative anticancer small molecules by targeting E2s, CRLs or DUBs.

Name	Structure	Target	In vitro	In cells	Cell viability, IC50	Cancer cell lines	Ref.
CC0651		Cdc34	IC50 (FP) = 2.5 μmol/L	p27 deubiquitination @ 30 μmol/L		HCT116	280,281
Leucettamol A		Ubc13	105 μmol/L				282
2D-08		Ubc9	IC50 = 6 μmol/L	100 μmol/L		BT-474	283
Compound 25		CRL1Skp2	>80% inhibition@5 μmol/L	p21, p27 deubiquitination @ 5 μmol/L	<10 μmol/L	PC-3, LNCaP	293
C1, C2, C16, C20			>80% inhibition@10 μmol/L	p21, p27 deubiquitination @ 10 μmol/L		MCF-7, TD47, LNCaP	294
VH298		CRL2VHL	IC50 (FP) = 80 nmol/L; Kd (ITC) = 90 nmol/L	HIF-1α destabilization @ 50 μmol/L	No cytotoxic	Hela	299
CM11	homo-PROTAC		Kd (ITC) = 25 nmol/L	HIF-1α destabilization @ 1 μmol/L		Hela	300
CDDO-Me		CRL3Keap1	58.9%@0.1 μmol/L		50–160 nmol/L	MCF-7	302
ML334			Kd = ∼ 1 μmol/L	EC50 = 18 μmol/L		U2OS	303
Compound 2			EC50 = 28.6 nmol/L	Nrf2–ARE induction@10 μmol/L		HepG2-ARE-C8	304
KI696			Kd (ITC) = 1.3 nmol/L				305
6lc		CRL3SPOP	KD = 30 μmol/L	PTEN deubiquitination @ 15 μmol/L	2.1–3.5 μmol/L	A498, OS-RC-2	308,309
CC885		CRL4CRBN		GSPT1 degradation @ 1 nmol/L		NB4	189
Thalidomid							336
Pomalidomide							337
Lenalidomid				IKZF1 and IKZF3 degradation @ 1 μmol/L		MM1S	191,338
Indisulam		CRL4DCAF15		RBM39 degradation @ 2 μmol/L		HCT116	339
CR8				Cyclin K degradation @ 1 μmol/L		HCT116	340
HQ461				DC50 = 0.13 μmol/L	1.5 μmol/L	A549	341
dCeMM2/3/4				Cyclin K degradation @ 2.5 μmol/L		KBM7	342
SJB2-043		USP1	0.54 μmol/L	>80% inhibition @ 1 μmol/L	1.1 μmol/L	K562	347
ML323			76 nmol/L	PCNA deubiquitination @ 20 μmol/L	3 μmol/L	H1299	348
Q29		USP2	82% inhibition @ 0.5 μmol/L	NA	4.7 μmol/L	DU145	350
ML364			1.1–1.7 μmol/L	cyclin D1 destabilization @ 0.97 μmol/L	3 μmol/L	HCT116	351
LCAHA			9.7 μmol/L	cyclin D1 destabilization @ 20 μmol/L	0.9 μmol/L	HCT116	352
XL188		USP7	90 nmol/L	p53 accumulation @5 μmol/L		MM.1S, MCF7	357
FT671			52 nmol/L	p53 accumulation @10 μmol/L	33 nmol/L	HCT116, MM.1S, MCF7	358
Compound 4			6 nmol/L	p53 accumulation @1 μmol/L	2–29 nmol/L	HCT116	359
Compound 41			0.44 nmol/L	p53 accumulation @25 nmol/L	0.09–0.45 μmol/L	MM.1S, H526	360
GNE6640			0.75 μmol/L	p53 accumulation @10 μmol/L	>50% inhibition@5 μmol/L	HCT116, MCF7, EOL-1	361
Capzimin		Rpn11	0.39 μmol/L	Deubiquitination @ 2–10 μmol/L	2.1–3.8 μmol/L	293T, A549, HCT116	364
SOP11			1.3/0.6 μmol/L	Deubiquitination @ 10 μmol/L	GI50 = 4.7 μmol/L	HCT116	365
VLX1570		USP14/UCH37	8.1/14 μmol/L	>80% inhibition @ 25 mmol/L	43–126 nmol/L	RPMI8226, KMS11, OPM-2	366

Previously reported molecular glues induce the degradation of specific substrates by binding to the CRL4 substrate receptors than the CRL4 adaptor proteins. Słabicki et al.³⁴⁰ analyzed the sensitivity of 4,518 clinical and preclinical drugs in 578 cancer cell lines via high-content screening alongside their relative E3 ligase component mRNA levels. A CDK12 inhibitor, (R)-CR8, revealed a correlation between cytotoxicity and CRL4^DDB1 mRNA levels. Further studies showed that (R)-CR8 functions as a molecular glue to regulate protein cyclin K degradation³⁴⁰. X-ray crystallography studies of the DDB1–(R)-CR8–CDK12–cyclin K complex demonstrated that (R)-CR8 binds to the ATP-binding pocket of CDK12, forming a high affinity with the DDB1 domain. Previously, the thiazolyl derivative HQ461 was identified as a potential anticancer small molecule capable of inhibiting Nrf2 activity. To reveal the functional mechanism of HQ461, Lv et al.³⁴¹ performed a pooled genome-wide CRISPR-Cas9 knockout screening in lung cancer cells A549 by targeting 19,114 genes with 4 individual sgRNAs per gene. According to the findings, HQ461 converts CDK12 into CRL4^DDB1 to initiate polyubiquitination and subsequent degradation of the partner protein cyclin K of CDK12. Mayor-Ruiz et al.³⁴² screened a library of >2000 small molecules in either wildtype or UBE2M^mut KBM7 cells and found that dCeMM1/2/3/4 displayed a correlation between cytotoxicity and UBE2M-associated cullin4 levels via scalable chemical profiling. Further studies revealed that dCeMM2/3/4 analogs regulated the ubiquitination-mediated degradation of cyclin K by prompting the interaction of CDK12 and cyclin K.

4.3.2. Multi-omics approaches empower the identification of CRL E3 ligases-associated molecule glue degraders

Interestingly, the majority of reported CRL-associated molecular glues are subordinate to CRL4 substrate receptors or adaptors, including CRBN, DCAF15, and DDB1. Notably, cell-based phenotypic assays have provided the richest source for MGD identification. It is hypothesized that numerous screening models were initially established using cell viability-based phenotypes, leading to the identification of hit compounds that interacted with CRL4 substrate receptors and adaptors. However, the traditional MDG discovery approaches face challenges in evaluating the relationship of CRL4 and cell-viability-based assay. These studies still heavily rely on the cell-viability-associated readouts to promote their MDG design, which results in the limitation that the targeted regulators are limited to those with an essential function in cell viability, such as CRL4 in possible. To overcome these limitations, it is crucial to advance phenotypic screens that can be measured, quantified, interpreted, and predicted, enabling a better understanding of MDG discovery. Modern high-throughput screening technologies, such as microarray and next-generation sequencing, have generated vast amounts of biological data that can be integrated in multiomics studies to explore the functional and mechanistic complexity of ubiquitin (Ub)-associated TPD.

Numerous existing data resources, including cancer-gene databases like Online Mendelian Inheritance in Man (OMIM) and Genetic Association Database (GAD), patient omics data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO), and cancer-associated Kyoto Encyclopedia of Genes and Genomes (KEGG), along with drug-gene databases such as DrugBank, Therapeutic Target, and Clinical Trials, serve as foundational resources for bioinformatics analyses of PPI networks and biological pathways. Based on the databases, mass spectrometry (MS)-based proteomics data analysis has emerged as a key approach for identifying the E3 ligands. Quantitative expression proteomics combined with correlative transcriptomics were employed to examine the mechanisms of action (MoA) of specific MGDs, such as CR8 and dCeMM2-4. In particular, drug-affinity enrichment and MS-based chemoproteomics, such as activity-based protein profiling (ABPP) and drug affinity-responsive target stability (DARTS), have successfully been conducted to discover potential RNF114 small-molecule ligand, natural product nimbolide³⁴³, These approaches have also facilitated the identification of specific fragment–protein interactions, such as E3 covalent engaging ligands^344,345. Moreover, advancement in imaging technology and precise visual analysis, such as cellular high-content imaging and small-molecule fluorescence probe, have provided additional insights into the MoA of ligases and ligands, when combined with the bioinformation profiling.

4.4. Deubiquitinases (DUBs) inhibitors

Most of the reported DUB inhibitors primarily target the proteins USP1, USP2, USP7, Rpn11, and USP14. Herein, we thus focused on novel DUB inhibitors developed between 2013 and 2021 and their potential for cancer therapy (Table 1). Among the human USP family of DUBs, USP1 has been implicated in the DNA damage response, which is responsible for deubiquitinating and stabilizing two crucial DNA repair proteins, PCNA and FANCD2 (fanconi anemia group D2)³⁴⁶. Chen et al.³⁴⁶ employed a self-established ubiquitin–rhodamine-based high-throughput screening method to identify a series of FDA-approved drugs and clinical candidates as potential USP1 inhibitors, including pimozide, GW7647, and flupenthixol. Utilizing a similar ubiquitin–rhodamine-based screening approach, Mistry et al.³⁴⁷ identified several USP1 inhibitors and determined their therapeutic potential for leukemia. Representative SJB2-043, derived from the hit compound C527, promotes the degradation of ID1 (a leukemic cell growth activator) and FANCD2 by inhibiting the catalytic activity of USP1 (IC₅₀ = 0.54 μmol/L), and is currently being used as a positive control to evaluate other potential USP1 inhibitors. Furthermore, Liang et al.³⁴⁸ developed the first selective and potent USP1 inhibitor, ML323, based on the discoveries of pimozide and GW7647. Phenylpyrimidin-derived ML323 inhibits USP1 with a nanomolar affinity (IC₅₀ = 76 nmol/L) and increases the monoubiquitinated PCNA (Ub-PCNA) levels (PCNA deubiquitination at 20 μmol/L in H1299)³⁴⁹. USP2 helps stabilize various tumor-associated substrates, including MDM2, fatty acid synthase, and cyclin D1^228,231,233. Hence, numerous human cancer cell lines undergo apoptosis when USP2 is silenced. Ohayon et al.³⁵⁰ screened and identified a series of UCH-L3 inhibitors and found that these compounds inhibited USP2 more effectively than UCH-L3, especially several ortho-quinones analogs. The natural product β-lapachone Q29, a clinical candidate for treating pancreatic cancer, exhibits simultaneous USP1/2 inhibition (82%–92% inhibition at 20 μmol/L)³⁵⁰. ML364, another representative small-molecule USP2 inhibitor, accelerates cyclin D1 degradation and induces cell cycle arrest in both Mino and HCT116 cancer cell lines³⁵¹. In addition, a lithocholic acid (LCA) derivative, LCAHA, destabilizes cyclin D1 in an AKT/GSK3β-independent manner, by suppressing USP2 but maintaining the expression of p27³⁵².

Recently, USP7 has emerged as a star protein of the DUB family for cancer therapy owing to its importance in regulating p53 and MDM2 levels^353,354. Several pharmaceutical companies, such as Hybrigenics, Progenra, FORMA Therapeutics, and Genentech, have contributed to the development of USP7 inhibitors³⁵⁵. Considering that a large number of small-molecule USP7 inhibitors have been developed during the past two decades, this review summarizes the progress of representative USP7 inhibitors from 2012 to the present. Early in 2012, Progenra developed several trisubstituted thiophene derivatives, such as P5091 and P22077348, as first-generation USP7 inhibitors³⁵⁶. These inhibitors covalently conjugate with the catalytic domain cysteine223 of USP7 and are applicable in cancer and immunological oncology therapies³⁵⁶; however, pan-deubiquitination and off-target defects limit their efficacy. By optimizing the structure of the lead compound pyrazolo-[3,4-d]-pyrimidin-4-one-piperidine (PyrzPPip), Lamberto et al.³⁵⁷ developed the next generation of USP7, subsequently discovering that XL188 is a highly potent and selective USP7 inhibitor (IC₅₀ = 90 nmol/L). X-ray crystallography of the USP7–XL188 complex revealed that XL188 is a noncovalent USP7 inhibitor. Additionally, the researchers of Hybrigenics have presented a series of PyrzPPip-derived USP7 inhibitors. They first identified a noncovalent USP7 inhibitor, FT671, in the series. FT671 is capable of specifically suppressing enzyme USP7 with high affinity. Unlike other reported inhibitors, FT671 is an allosteric inhibitor that targets a dynamic pocket close to the catalytic position of USP7³⁵⁸. Furthermore, Almac Discovery used the surface plasmon resonance (SPR) to identify new USP7 inhibitors through a fragment-based screen. Hydroxypiperidine-derived Compound 4, the most potent inhibitor of the hit compounds, selectively inhibits USP7 in vitro (IC₅₀ = 6 nmol/L) and in cells (IC₅₀ = 49 nmol/L) with nanomolar inhibitory IC₅₀ values³⁵⁹. Leger et al.³⁶⁰ identified benzofuran-amide as a potential scaffold for USP7 inhibition, revealing that the optimized scaffold has a large succinimide group that serves as a major potency-driving motif. The succinimide group forms two crucial hydrogen bonds in the allosteric position of USP7. The optimal Compound 41 inhibits USP7 with more potency and selectivity than other previously reported deubiquitinases³⁶⁰. In the same year, GeneTech introduced GNE6640, an additional potent allosteric USP7 inhibitor. Structural analyses revealed that GNE6640 noncovalently binds USP7 at a distance of 12 Å from catalytic cysteine and interacts with acidic residues that mediate the hydrogen bond interaction with the Lys48 side chain of ubiquitin³⁶¹.

As outlined above, the proteasome comprises a 19S regulatory particle (RP) and a 20S core particle. Although approved “omibs” drugs always target the β5 active site of the 20S core particle, resistance occurs. Rpn11 is a Zn²⁺-dependent metalloisopeptidase derived from the JAMM zinc metalloprotease family of DUBs that hydrolyzes ubiquitin from tagged proteins and is a crucial member of the proteasome 19S regulatory particle³⁶². Inactivating the enzymatic active site of Rpn11 by altering zinc-coordinating histidine residues inhibits substrate degradation in cells³⁶³. Recent research has shown that targeting Rpn11 as a potential cancer treatment target not only inhibits the proteasome but also increases the therapeutic efficacy of “omibs” drugs³⁶⁴. Li et al.³⁶⁴ screened 330,000 compounds and selected 8TQ as a promising lead (a thioester derivative). They subsequently optimized the structure of 8TQ using the FBDD approach to summarize SARs and identified the first potent and specific Rpn11 chemical probe, capzimin (IC₅₀ = ∼300 nmol/L), which inhibited the proliferation of bortezomib-resistant cancer cells by stabilizing the proteasome substrates and inducing an unfolded protein response. Utilizing the positive capzimin and their inhouse screening system, they discovered an epipolythiodioxopiperazines-derived SOP11 that inhibits proteasomal deubiquitinases, such as Rpn11 and AMSH. In addition to Rpn11, the 19S RP has two more DUBs, the cysteine proteases USP14 and UCHL5 (UCH37)³⁶⁵. VLX1570, the first DUB inhibitor containing a reactive α and β-unsaturated carbonyl group and covalently interacting with the nucleophilic residues of USP14 and UCHL5, recently entered clinical trials³⁶⁶.

5. Conclusions and perspective

Recent and ongoing research demonstrates that Ub and Ubl pathways play crucial roles in tumorigenesis by degrading or activating/deactivating key regulators of tumor growth and death. The increased knowledge supports the idea that targeting aberrant Ub or Ubl pathways is a novel therapeutic strategy for multiple types of tumors. In this regard, the fundamental advances and follow-on target-based drug discoveries have been crucial in providing vital information concerning contemporary translational efforts to develop precise treatment by targeting Ub and Ubl pathways.

However, several challenges of the Ub and Ubl field are required with solution, such as 1) numerous studies have focused on identifying the roles of E1–E2–E3 ligases in tumorigenesis; however, the oncogenic or tumor-suppressive function of these enzymes is influenced by various factors, such as the cellular context (e.g., β-TRCP) and subcellular localization (e.g., SPOP). Therefore, a comprehensive understanding of the regulatory mechanisms of Ub and Ubl enzymes may provide therapeutic alternatives to inactivate/reactivate or upregulate/downregulate these enzymes in various tumors. 2) Although our multiomics analyses decipher the potential roles of particular enzymes, there is a lack of supporting in vitro or in vivo studies for many Ub- and Ubl-associated enzymes used in anticancer therapy. 3) Targeting E1, E2, or E3 may affect multiple biological processes, resulting in different phenotypes; therefore, a deeper understanding of each enzyme is required to delineate the potential tumor-promoting effects of putative inhibitors of E1–E2–E3 cascades and develop individualized treatments. 4) Some PPI interface may lack such well-defined pocket, which makes it difficult to design selective and potent small-molecule inhibitors. Overall, the present review highlighted the tumor-suppressive or tumor-promoting roles of many Ub- and Ubl-associated enzymes. An ever-growing list of substrates and upstream regulators will help us better understand the roles of these enzymes in tumors and whether they are promising anticancer targets.

Acknowledgments

This work was supported by the following funds: National Natural Science Foundation of China (Grants 81820108022, 82003297 and 22177076), Innovation Program of Shanghai Municipal Education Commission (2019-01-07-00-10-E00056, China), Shanghai Frontiers Science Center of Disease and Syndrome Biology of Inflammatory Cancer Transformation (2021KJ03-12, China), The Scientific and Technological Innovation Action Plan of Science and Technology Commission of Shanghai (20JC1411300, China), ChenGuang project supported by Shanghai Municipal Education Commission and Shanghai Education Development Foundation (19CG49, China).

Author contributions

Yanyu Jiang and Shuaishuai Ni collected the related papers, drafted and revised the manuscript. Biying Xiao collected the related papers. Lijun Jia revised and finalized the manuscript. All authors read and approved the final manuscript.

Conflicts of interest

The authors declare that they have no conflicts of interest.