Department of Molecular and Cell Biology, University of California at Berkeley, Berkeley, CA, USAHoward Hughes Medical Institute, University of California at Berkeley, Berkeley, CA, USA
Department of Molecular and Cell Biology, University of California at Berkeley, Berkeley, CA, USAHoward Hughes Medical Institute, University of California at Berkeley, Berkeley, CA, USA
Department of Molecular and Cell Biology, University of California at Berkeley, Berkeley, CA, USAHoward Hughes Medical Institute, University of California at Berkeley, Berkeley, CA, USA
Induced protein degradation accomplishes elimination, rather than inhibition, of pathological proteins. Key to the success of this novel therapeutic modality is the modification of proteins with ubiquitin chains, which is brought about by molecular glues or bivalent compounds that induce proximity between the target protein and an E3 ligase. The human genome encodes ∼600 E3 ligases that differ widely in their structures, catalytic mechanisms, modes of regulation, and physiological roles. While many of these enzymes hold great promise for drug discovery, few have been successfully engaged by small-molecule degraders. Here, we review E3 ligases that are being used for induced protein degradation. Based on these prior successes and our growing understanding of the biology and biochemistry of E3 ligases, we propose new ubiquitylation enzymes that can be harnessed for drug discovery to firmly establish induced protein degradation as a specific and efficient therapeutic approach. 诱导蛋白降解实现了对病理蛋白的消除，而不是抑制。这种新型治疗模式的成功关键在于通过分子胶或双价化合物使目标蛋白与E3连接酶之间产生接近，从而将蛋白质修饰为泛素链。人类基因组编码了大约600种结构、催化机制、调控方式和生理作用各异的E3连接酶。虽然其中许多酶在药物发现方面具有巨大潜力，但只有少数能够成功地与小分子降解剂相互作用。在这里，我们回顾了正在用于诱导蛋白降解的E3连接酶。基于这些先前的成功以及我们对E3连接酶生物学和生物化学的日益了解，我们提出了可以用于药物发现的新型泛素化酶，以确立诱导蛋白降解作为一种特异而高效的治疗方法。
Induced protein degradation has emerged as a new therapeutic modality against targets that were once deemed undruggable. Key to eliminating pathological proteins from cells is their modification with ubiquitin chains that are recognized by receptors of the 26S proteasome (
). E2 enzymes can then charge a Cys residue in E3 ligases of the HECT (homologous to E6-AP C terminus), RBR (RING-in between-RING), and RCR (RING Cys relay) families, thereby enabling these enzymes to ubiquitylate a target (
) (Figure 1A). Alternatively, E2s can be activated by E3 ligases with signature RING (really interesting new gene) domains to transfer ubiquitin from their own active site to the substrate (
) (Figure 1B). By recruiting both substrates and ubiquitin-charged E2 enzymes, E3 ligases occupy a central place in the enzymatic cascade leading to protein ubiquitylation. 诱导蛋白降解已成为一种针对曾被认为无法药物靶点的新的治疗模式。从细胞中清除病理蛋白的关键是将其修饰为被26S蛋白酶体受体识别的泛素链。这种泛素聚合物的组装由E1泛素激活酶启动，它们将泛素转移到约40个E2泛素连接酶的活性位点半胱氨酸上。然后，E2酶可以将泛素充电到HECT（与E6-AP C端同源）、RBR（环中环）和RCR（环半胱氨酸中继）家族的E3连接酶的半胱氨酸残基上，从而使这些酶能够泛素化目标。或者，E2酶可以被具有特征RING（真正有趣的新基因）结构域的E3连接酶激活，将泛素从其自身的活性位点转移到底物上。通过招募底物和带电的E2酶，E3连接酶在导致蛋白质泛素化的酶级级联中占据着核心位置。
Figure 1E3 ligase families and modes of substrate recognition in cells 图1E3中的连接酶家族和细胞中底物识别的方式
(A) The HECT (homologous to E6-AP C terminus), RBR (RING-in between-RING), and RCR (RING Cys relay) E3 ligases possess a catalytic Cys residue that is charged with ubiquitin by an E2 enzyme. The E3 ligase then catalyzes substrate modification.
(B) E3 ligases of the RING (really interesting new gene) family bind both a charged E2 enzyme and a substrate and promote transfer of ubiquitin from the active site of the E2 directly onto the substrate. The Cullin-RING ligases are modular RING-family E3 ligases that utilize exchangeable adaptor proteins to recruit specific substrates.
(C) E3 ligases recognize their physiological targets through short sequence motifs referred to as degrons. Degrons can be constitutively accessible to an E3 ligase or formed only upon posttranslational modification. While many degrons are linear sequence motifs, structural degrons require a folded protein domain. The structural depictions were derived from the following pdb files: PDB: 6DO5 (KLHDC2), 1P22 (βTrCP), and 6W66 (FBXW7FBXL17 (not FBXW7)).
The human genome encodes ∼600 E3 ligases, which differ widely in how many proteins they modify: while genetic studies identified a single essential substrate for the E3 ligases MDM2, CUL3KEAP1, and SCFFBXL5 (
). E3 ligases also vary in the products of their enzymatic activity: rather than tagging their substrates with a single possible modification, they can assemble different types of ubiquitin conjugates that elicit distinct outcomes in cells (
), while enzymes synthesizing short chains of mixed topology or K63-linked polymers help eliminate proteins, aggregates, or even organelles through the autophagy-lysosome system (
). Notably, E3 ligases do not have to act alone: collaboration between enzymes results in more complex heterotypic polymers, such as K11/K48- or K29/K48-branched chains (
). Having that many E3 ligases at their disposal, cells use a wide range of approaches to fine-tune their activities, which include inhibitors of substrate binding or catalytic activity, changes in their intracellular localization, or expression in specific tissues or at particular stages of development or disease (
To ensure that ubiquitylation occurs at the right time and place, E3 ligases recognize their targets through specific motifs referred to as degrons (Figure 1C). Some degrons, such as the D box present in APC/C substrates or the carboxy terminus of C-end-rule substrates (
), are linear sequence stretches that are constitutively accessible to the E3 ligase. Other degrons require posttranslational modifications, such as phosphorylation (
The SCFbeta-TRCP-ubiquitin ligase complex associates specifically with phosphorylated destruction motifs in IkappaBalpha and beta-catenin and stimulates IkappaBalpha ubiquitination in vitro.
). While short linear motifs have been studied most extensively, structural degrons also exist: providing a recent example, SCFFBXL17 recognizes the shape, rather than the sequence, of its target broad-complex, tramtrack, and bric-à-brac (BTB) domain (
The SCFbeta-TRCP-ubiquitin ligase complex associates specifically with phosphorylated destruction motifs in IkappaBalpha and beta-catenin and stimulates IkappaBalpha ubiquitination in vitro.
As shown for the APC/C, fusing a degron to heterologous proteins can deliver an unnatural target, or neo-substrate, to an E3 ligase for ubiquitylation (
). Genetic fusions were soon superseded by a small-molecule inhibitor of a cellular protein that was connected to a peptide degron for the E3 ligase SCFβTrCP (
). This first proteolysis-targeting chimera, or PROTAC, triggered the ubiquitin- and proteasome-dependent degradation of its recombinant drug target in Xenopus laevis extract. With breathtaking speed, we have learned to replace the degron with small molecules, resulting in compounds that can recruit specific proteins to E3 ligases to elicit their proteolytic ubiquitylation (
). Compared with traditional inhibitors, small-molecule degraders act catalytically and, hence, at lower concentrations, interfere with all functions of a protein—rather than shutting down only one activity—and prevent compensatory upregulation of the drug target. In this review, we will focus on the E3 ligases as the key machineries of induced protein degradation. In particular, we will build on our increasing understanding of the biology and biochemistry of ubiquitylation to propose new enzymes that could expand our toolbox and firmly establish this exciting therapeutic modality. 如APC/C所示，将降解标记融合到异源蛋白质上可以将非自然靶点或新型底物传递给E3连接酶进行泛素化。遗传融合很快被连接到肽降解标记的细胞蛋白小分子抑制剂所取代。这种第一个蛋白质降解靶向嵌合物（PROTAC）在Xenopus laevis提取物中触发了其重组药物靶点的泛素和蛋白酶体依赖性降解。我们以惊人的速度学会了用小分子取代降解标记，从而产生能够招募特定蛋白质到E3连接酶以引发其蛋白质降解泛素化的化合物。与传统抑制剂相比，小分子降解剂具有催化作用，因此在较低浓度下干扰蛋白质的所有功能，而不仅仅是关闭一个活性，并防止药物靶点的补偿性上调。在本综述中，我们将重点关注E3连接酶作为诱导蛋白质降解的关键机制。 特别是，我们将在对泛素化的生物学和生物化学的日益理解的基础上，提出新的酶，以扩展我们的工具箱，并牢固地确立这种令人兴奋的治疗模式。
Current E3 ligase toolbox for induced protein degradation 目前用于诱导蛋白质降解的E3连接酶工具箱
Small molecules that target proteins for ubiquitylation fall into two major classes: molecular glues and bivalent compounds that we refer to as PROTACs. While molecular glues sandwich between two proteins (
) (Figure 2B). Molecular glues are expected to have better pharmacological properties, yet PROTACs allow for modular design to rapidly connect one enzyme with many targets. Very few of the 600 human E3 ligases have been engaged by small-molecule degraders, and we have yet to learn how to predict which enzyme would work best against a given target. It is therefore likely that we are missing out on opportunities to improve the efficiency of induced protein degradation. Importantly, cancer cells are able to evolve resistance against degraders by downregulating the E3 ligase (
), expanding our E3 ligase toolbox is imperative to increase the impact of induced protein degradation as a therapeutic strategy. Learning from prior successes provides one means to accomplish this goal. 靶向蛋白质泛素化的小分子可分为两大类：分子粘合剂和我们称之为PROTACs的双价化合物。分子粘合剂夹在两个蛋白质之间（
(A) Molecular glues are small compounds that sandwich between an E3 ligase and a protein. (Right) Structural view of the E3 ligase subunit CRBN bound to the molecular glue lenalidomide and its target, the CK1α kinase.
(B) Bivalent molecules, also referred to as PROTACs, combine independent warheads against an E3 ligase and a target through a short linker. (Right) Structural view of the E3 ligase subunit VHL bound to the PROTAC MZ1 and its target, BRD4.
(C) Molecular glues can indirectly induce ubiquitylation by triggering the oligomerization of a protein, which likely allows for multivalent recognition of a low-affinity degron by the E3 ligase. (Bottom) Structural view of BTB domains of BCL6 bound to its oligomerization-inducing molecular glue, BI-3802. The structural depictions were derived from the following pdb files: PDB: 5FQD (CRBN), 5T35 (VHL), and 6XMX (BCL6).
E3 ligases targeted by natural molecular glues E3酶受天然分子粘合物靶向
Providing a blueprint for drugging E3 ligases, plants use the hormone auxin, or indoleacetic acid, as a molecular glue to deliver transcriptional repressors to the E3 ligase SCFTIR1 (
). By tethering transcriptional regulators to SCFTIR1, auxin induces their proteasomal degradation and thereby molds the genetic expression landscape of the plant. Variations of this theme are employed to sense other hormones, such as gibberellic acid and jasmonate (
), which points to the versatility and robustness of induced protein degradation as a signaling modality. Indeed, expression of the auxin receptor TIR1 allows for hormone-inducible degradation of proteins that had been fused to the short plant degron in many heterologous cell types (
Auxin signals through an E3 ligase of the SCF family, which are composed of a CUL1 scaffold, the RING-domain subunit RBX1, and SKP1 as a shared factor that connects the catalytic CUL1-RBX1 module to exchangeable substrate adaptors with F-box domains (
). Among the 22 human FBXL proteins with similar domain architecture, several have interesting links to disease or small-molecule regulation: FBXL2 activity toward a phosphatidylinositol 3-kinase subunit or inositol 1,4,5-triphosphate receptor requires prenylation (
). These observations imply that FBXL E3 ligases could be promising candidates for developing molecular glues to instigate protein degradation. 生长素通过SCF家族的E3连接酶进行信号传导，该家族由CUL1支架、RING结构域亚单位RBX1和作为共享因子的SKP1组成，将催化的CUL1-RBX1模块与具有F-box结构域的可交换底物适配体连接起来（
). Nimbolide covalently modifies a Cys residue in a domain of RNF114 that had been implicated in substrate recruitment, allowing researchers to convert nimbolide into an RNF114-dependent PROTAC (
). Although the physiological roles of UBR7 and RNF114 remain ill defined, these results highlight that not only receptors of natural molecular glues, but also natural products themselves, could point to novel E3 ligases for small-molecule-induced protein degradation. 回忆起生长素，天然产物阿斯卡霉素作为一种分子胶水，将E3连接酶UBR7与TP53肿瘤抑制因子连接起来。此外，印楝树的一种抗增殖产物尼姆波酮抑制了E3连接酶RNF114。尼姆波酮共价修饰了RNF114中一个被认为与底物招募有关的结构域中的一个半胱氨酸残基，使研究人员能够将尼姆波酮转化为依赖于RNF114的PROTAC。虽然UBR7和RNF114的生理作用尚未明确，但这些结果强调，不仅天然分子胶水的受体，而且天然产物本身也可能指向新的E3连接酶，用于小分子诱导的蛋白质降解。
E3 ligases targeted by synthetic molecular glues 合成分子胶水靶向的E3连接酶
While nature might have taken the lead, synthetic compounds are rapidly catching up and will allow us to harness the many human ubiquitylation enzymes in a more prospective fashion. This field was jump-started by a molecular glue with a simple structure and dark past, thalidomide. Prescribed as a sleep medication and treatment against morning sickness, thalidomide disrupted limb development in newborns and caused peripheral neuropathy in older patients (
). Thalidomide and its derivatives, collectively referred to as IMIDs, were later found to be efficient against multiple myeloma or del5q myelodysplastic syndrome. These compounds all bind cereblon (CRBN), a substrate adaptor of CUL4-RING E3 ligases (
), and a natural product that takes thalidomide's place in cellular regulation is not known. 尽管自然可能占据了先导地位，但合成化合物正在迅速赶上，并将使我们能够更有前景地利用许多人类泛素化酶。这个领域是由一种结构简单但过去不为人知的分子胶水——沙利度胺（thalidomide）推动起来的。沙利度胺曾被作为睡眠药物和对抗晨吐的治疗药物，但它破坏了新生儿的肢体发育，并导致老年患者出现周围神经病变。后来发现，沙利度胺及其衍生物（统称为IMIDs）对多发性骨髓瘤或del5q骨髓增生异常综合征具有高效作用。这些化合物都与CUL4-RING E3连接酶的底物适配器cereblon（CRBN）结合，并招募锌指转录因子或激酶CK1α参与CUL4依赖的泛素化和蛋白酶体降解。沙利度胺还将转录因子SALL4传递给CUL4，这可能解释了肢体发育的异常。至于其神经系统副作用的靶点尚待确定。 CRBN的生理底物仍然非常稀缺（
CRBN is one of ∼100 exchangeable substrate adaptors of modular E3 ligases that contain CUL4A or the related CUL4B as a scaffold, RBX1 as the catalytic center, and DDB1 as a shared factor to recruit the substrate adaptor (
). Substrate adaptors of the CUL4 E3 ligases often show structural flexibility that allows them to adapt to diverse surfaces presented by physiological or unnatural binding partners (
). This feature is exploited by viral factors that recruit CUL4 to host proteins to induce their degradation and thereby blunt an anti-infectious immune response (
). CUL4 E3 ligases therefore appear to be well suited to accommodate unnatural targets, a prerequisite for use in induced protein degradation. In line with this notion, the sulfonamide indisulam exploits an enzyme built around the adaptor DCAF15, CUL4DCAF15, to eliminate RNA-binding proteins (
). The catalytic module composed of CUL4, RBX1, and DDB1 can even be recruited to proteins independent of adaptors, as shown by compounds that cause degradation of a cyclin-dependent kinase regulating transcriptional elongation (
It is a potential caveat of harnessing CUL4 E3 ligases that these enzymes often regulate processes essential for development. DDB1 is critical for stem cell division and differentiation (
Mutations in CUL4B, which encodes a ubiquitin E3 ligase subunit, cause an X-linked mental retardation syndrome associated with aggressive outbursts, seizures, relative macrocephaly, central obesity, hypogonadism, pes cavus, and tremor.
). To prevent repeating thalidomide's history, it is important that the physiological functions of CUL4 E3 ligases and organismal consequences of engaging them with small molecules are understood in depth before these enzymes are exploited for therapeutic applications in patients. 利用CUL4 E3连接酶的潜在注意事项是，这些酶通常调节发育过程中的关键环节。DDB1对干细胞的分裂和分化至关重要（
Mutations in CUL4B, which encodes a ubiquitin E3 ligase subunit, cause an X-linked mental retardation syndrome associated with aggressive outbursts, seizures, relative macrocephaly, central obesity, hypogonadism, pes cavus, and tremor.
While IMIDs were introduced into the clinic without understanding their mechanism of action, recent work taught us that it is possible to prospectively select an E3 ligase for generating synthetic molecular glues. Proof-of-concept compounds brought hypophosphorylated β-catenin, a driver of colon cancer (
). Although these molecules effectively induced β-catenin ubiquitylation in vitro, they were less potent in driving protein degradation in vivo, which in part might be due to the large number of endogenous substrates that compete for access to this E3 ligase (
). Thus, as we are beginning to choose new E3 ligase handles, there is an urgent need to decipher the physiological regulation of these enzymes in cellular or organismal models. 虽然在不了解其作用机制的情况下，IMIDs已被引入临床，但最近的研究告诉我们，可以预先选择一个E3连接酶来生成合成的分子胶。概念验证化合物将低磷酸化的β-连环蛋白（结肠癌的驱动因子）与E3连接酶SCF（）接近。尽管这些分子在体外有效地诱导了β-连环蛋白的泛素化，但在体内促进蛋白质降解的效果较弱，部分原因可能是由于大量内源性底物竞争访问该E3连接酶。因此，当我们开始选择新的E3连接酶处理物时，迫切需要解密这些酶在细胞或有机体模型中的生理调节。
E3 ligases recruited by protein oligomerization 蛋白质寡聚化招募的E3连接酶
While tethering a neo-substrate to an E3 ligase is one way for compounds to induce degradation, molecular glues can also trigger turnover by promoting protein oligomerization (Figure 2C). This paradigm was recently established with compounds targeting BCL6, a transcriptional repressor and driver of B cell malignancies (
). Molecules that induce further oligomerization of BCL6 dimers allow for recruitment of the E3 ligase SIAH1 and thereby trigger BCL6 ubiquitylation and degradation of the transcriptional repressor (
), induced oligomerization likely establishes multivalent recognition of a low-affinity degron by the E3 ligase. Compounds that displace a tRNA synthetase variant from heat shock protein 70 (HSP70) also cause SIAH1-dependent ubiquitylation and degradation (
), and it is tempting to speculate that release from HSP70 similarly results in protein oligomerization and multivalent detection by the E3 ligase. As SIAH1 expression increases upon accumulation of aggregation-prone proteins (
), these compounds likely exploit a physiological mechanism controlling substrate recognition by this E3 ligase. 将新型底物与E3连接酶相连是化合物诱导降解的一种方式，分子胶也可以通过促进蛋白质寡聚化来触发蛋白质的降解（图2C）。最近，这种范式已经在针对B细胞恶性肿瘤的转录抑制因子和驱动因子BCL6的化合物中得到了证实。BCL6含有一个BTB结构域，与约100个CUL3适配体和约100个转录因子中的结构域相同。大多数BTB结构域介导同源二聚化，这对其功能是必需的。诱导BCL6二聚体进一步寡聚化的分子可以招募E3连接酶SIAH1，从而触发BCL6的泛素化和转录抑制因子的降解。由于SIAH1本身是二聚体并且识别一个短的VxP基序，诱导的寡聚化可能通过E3连接酶对低亲和力降解子的多价识别来建立。 使tRNA合成酶变体从热休克蛋白70（HSP70）中释放的化合物也会引起SIAH1依赖的泛素化和降解（
Reminiscent of SIAH1, RIPLET and TRIM65 recognize their ubiquitylation targets RIG-I and MDA5 only after they polymerize upon binding to exogenous RNA molecules (
). Molecular glues that oligomerize substrates of these enzymes might therefore emerge as indirect, yet powerful, means to instigate drug-induced protein ubiquitylation. However, programs aimed at developing such compounds need to carefully control against potentially toxic protein aggregation, a common cause of neurodegenerative disease (
E3 ligases targeted by inhibitors-turned-PROTACs E3酶受抑制剂转化为PROTACs所针对的酶
With so many E3 ligases encoded in our genome, it is not surprising that dysregulation of some enzymes causes or sustains disease. While this led to an interest in developing E3 ligase inhibitors, ubiquitylation enzymes do not possess readily accessible pockets at their active sites, such as the eponymous ATP-binding clefts in kinases. Early programs thus wished to block interactions between E3 ligases and their key substrates, which resulted in compounds that can now be re-programmed into PROTACs. A case in point is made by MDM2, which ubiquitylates the tumor suppressor TP53 (
). Inhibition of MDM2 is considered beneficial in cancers that maintain wild-type TP53 alleles but downregulate the tumor suppressor by MDM2 overexpression. Among the first programs to successfully target an E3 ligase, molecules referred to as “Nutlins” thus competitively inhibited TP53-binding to MDM2, yet these molecules stalled in the clinic due to toxicity based on TP53-dependent neutropenia and thrombocytopenia (
). Nutlins were recently converted into PROTACs that trigger neo-substrate degradation at concentrations at which they only partially inhibit the turnover of TP53 (
). These inhibitors-turned-PROTACs might thus provide therapeutic benefit without the drastic side effects of complete MDM2 loss. 在我们的基因组中编码了如此多的E3连接酶，因此一些酶的失调导致或维持疾病并不令人意外。尽管如此，这引起了对开发E3连接酶抑制剂的兴趣，但泛素化酶在其活性位点上并没有容易访问的口袋，例如激酶中的著名ATP结合裂口。因此，早期的项目希望阻断E3连接酶与其关键底物之间的相互作用，这导致了现在可以重新编程为PROTACs的化合物。一个例子是MDM2，它泛素化肿瘤抑制因子TP53。抑制MDM2在保持野生型TP53等位基因但通过MDM2过表达下调肿瘤抑制因子的癌症中被认为是有益的。因此，被称为“Nutlins”的分子竞争性地抑制了TP53与MDM2的结合，然而这些分子由于基于TP53依赖性中性粒细胞减少症和血小板减少症的毒性而在临床中停滞不前。 Nutlins最近被转化为PROTACs，以在仅部分抑制TP53的转化率的浓度下触发新底物的降解。这些转化的抑制剂-PROTACs可能因此在不引起完全MDM2丧失的剧烈副作用的情况下提供治疗效益。
A similar logic motivated the development of inhibitors against cIAP1, as overexpression of this E3 ligase is known to drive tumorigenesis and help cancer cells survive cytotoxic treatments (
). Interestingly, small-molecule cIAP1 antagonists not only prevent substrate recognition, but also cause autoubiquitylation and proteasomal degradation of the E3 ligase (
), suggesting that molecules originally designed to inhibit substrate recognition might provide a robust path forward toward new PROTAC E3 ligases. 类似的逻辑推动了针对cIAP1的抑制剂的开发，因为已知这种E3连接酶的过度表达会促进肿瘤发生，并帮助癌细胞在细胞毒性治疗中存活（
). A similar strategy, focusing on the interaction between E3 ligase and degron, provided a starting point for potent PROTACs using CUL2VHL. As a key enzyme of the hypoxic stress response (
). To ensure that this reaction occurs only when sufficient oxygen is available, recognition of HIF1α by CUL2VHL requires hydroxylation of a Pro residue in the HIF1α degron (
). As oxygen levels drop, prolyl hydroxylation of HIF1α decreases, its ubiquitylation is prevented, and the transcription factor accumulates to drive a compensatory gene expression program. Starting with peptide mimetics of the HIF1α degron, compounds that compete with HIF1α for access to CUL2VHL were found (
Is the success of CUL2VHL-directed PROTACs due to a property of the HIF1α degron and its unique posttranslational modification, or could other members of the CUL2 family of E3 ligases yield similarly efficient PROTACs or molecular glues? We argue that the latter is an exciting possibility. Using an intermediary Elongin B/Elongin C dimer, CUL2 binds exchangeable substrate adaptors characterized by a VHL-box motif (
). This architecture is highly similar to CUL5 E3 ligases that use Elongin B/Elongin C to connect CUL5 with substrate adaptors containing a SOCS-box motif (
). Both CUL2 and CUL5 E3 ligases are hijacked by viruses that remodel the host proteome, showing that these enzymes can be re-programmed to target new substrates (
). Akin to CUL2VHL, several CUL2 and CUL5 enzymes play key roles in stress responses and thus might be kept in a state poised for rapid activation. This includes CUL2FEM1B, a core component of the reductive stress response (
), an observation that indicates that E3 ligases of this family might be able to accept small molecules for PROTAC development. CUL2导向的PROTACs的成功是由于HIF1α降解物及其独特的翻译后修饰性质，还是CUL2家族的其他成员的E3连接酶能够产生同样高效的PROTACs或分子胶水？我们认为后者是一个令人兴奋的可能性。通过中介的Elongin B/Elongin C二聚体，CUL2与可交换的底物适配体结合，这些适配体具有VHL-box基序（
Using biology as a guide for novel degrader E3 ligases 以生物学为指导，开发新型降解酶E3连接酶
E3 ligases that have already been engaged by small molecules provide guideposts for finding new enzymes for PROTAC or molecular glue development, yet prior success as a starting point might resemble looking for keys under the streetlight. What if a better key could be found outside of the zone of light? As an alternative strategy, we thus propose to build on our increasing understanding of the rich biology and biochemistry of E3 ligases to prospectively select an optimal enzyme for a target at hand. To choose an E3 ligase tailored to the task, features such as mode of substrate selection, catalytic specificity, regulation, dynamics of enzyme expression and assembly, and physiological function should be taken into account. As emerging technologies, including DNA-encoded libraries (
), are allowing us to rapidly isolate small-molecular binders against many targets, generating PROTAC handles for E3 ligases chosen due to their biological or catalytic properties will provide many opportunities to expand the reach of induced protein degradation. 已经被小分子激活的E3连接酶为寻找新的PROTAC或分子胶的酶提供了指引，然而，以此为起点的先前成功可能类似于在路灯下寻找钥匙。如果在光线区域之外找到更好的钥匙会怎样呢？因此，作为一种替代策略，我们建议在对E3连接酶的丰富生物学和生物化学的理解不断增加的基础上，有针对性地选择一个最佳酶来处理手头的目标。在选择适合任务的E3连接酶时，应考虑到底物选择方式、催化特异性、调控、酶表达和组装的动力学以及生理功能等特征。随着包括DNA编码文库在内的新兴技术的出现，我们能够快速分离出针对许多目标的小分子结合物，选择具有生物学或催化特性的E3连接酶生成PROTAC手柄将提供许多扩展诱导蛋白降解的机会。
E3 ligases poised to be active E3连接酶准备活跃起来
The mainstay E3 ligases for PROTAC development, CUL4CRBN and CUL2VHL, recognize a large number of structurally related proteins and an oxygen-responsive transcription factor, respectively. As CRBN was also suggested to act as an HSP90 co-chaperone specific for transmembrane proteins (
), it appears that both enzymes are embedded in stress response or quality control pathways that rapidly jump into action when needed. In fact, quality control E3 ligases fulfill many requirements for efficient PROTAC handles: they are expressed at high levels to cope with a potentially large number of substrates during stress (Figure 3A), and they must be flexible enough to ubiquitylate many proteins that differ in primary sequence and tertiary structure. Furthermore, quality control E3 ligases frequently modify their targets with highly effective ubiquitin conjugates, such as the proteasomal priority signals K11/K48-branched chains (
), as only rapid degradation can prevent aggregation of their natural targets. 主要用于PROTAC开发的E3连接酶，CUL4和CUL2，分别识别大量结构相关的蛋白质和氧感应转录因子。由于CRBN也被认为是HSP90的共分子伴侣，专门用于跨膜蛋白（） ，因此看起来这两种酶都嵌入在应激反应或质量控制途径中，当需要时迅速行动。事实上，质量控制E3连接酶满足了有效的PROTAC手柄的许多要求：它们在应激期间表达水平较高，以应对潜在的大量底物（图3A），并且它们必须足够灵活，能够泛素化许多在主要序列和三级结构上有所不同的蛋白质。此外，质量控制E3连接酶经常使用高效的泛素连接物修饰其靶标，例如蛋白酶体优先信号K11/K48-分支链（；；），因为只有快速降解才能防止其天然靶标的聚集。
Figure 3Expression patterns as guidance to select E3 ligases for degrader development 图3 表达模式作为选择E3连接酶进行降解剂开发的指导
). E3 ligases that have been successfully targeted are shown in green, while quality control E3 ligases as new candidate PROTAC targeted are shown in orange.
(B) Examples of tissue-specific E3 ligases that show moderate to high expression in fewer than 6 tissues were isolated from expression data averaged for each tissue (
(C) E3 ligases showing high expression (>30 FPKM in at least one cancer type) and at least a 10-fold enrichment over their median expression across 25 cancer types were isolated from the dataset E-MTAB-2706. Expression data from cancer cell lines of a given cancer type were averaged and fold change over the median is represented in the heatmap.
Quality control E3 ligases poised to modify proteins with potent ubiquitin chains therefore provide unique opportunities to generate molecular glues or PROTACs. Importantly, as cells can experience a wide range of adverse conditions, they possess many types of quality control E3 ligases that the degrader community could choose from. In addition to the CUL2 enzymes of the reductive stress or C-end-rule pathways, which were described above, an interesting family of E3 ligases is the N-recognins (
). These enzymes are best known for determining the stability of a protein based on its amino-terminal residue, a pathway aptly named N-end rule. The specificity of these E3 ligases is imposed by their namesake UBR box, which detects a protein's amino terminus, but can bind amino acids with micromolar affinity (
), which suggests that we should be able to engage them with small molecules. In addition to their role in the N-end rule, the best-understood members UBR1, UBR2, UBR4, and UBR5 act in quality control: UBR1 and UBR2 collaborate with the chaperone HSP70 to degrade misfolded cytoplasmic proteins (
); and UBR4 and the HECT-E3 UBR5 assemble K11/K48-branched ubiquitin chains on newly synthesized, yet misfolded, proteins that are presented by the chaperones BAG6, HSP70, or HSP90 (
). However, as with CUL4, molecules targeting UBR-family E3 ligases need to be carefully assessed for effects on the natural targets of these enzymes, as mutations in the genes encoding these enzymes have been associated with both developmental and neurodegenerative diseases (
E3 ligases at the right time and place 在正确的时间和地点的E3连接酶
Quality control E3 ligases can be considered as housekeeping enzymes, and they are accordingly expressed in most tissues at high levels (Figure 3A). By contrast, in line with the crucial role of ubiquitylation in cell fate specification (
), some E3 ligases are present predominantly in specific tissues or at particular stages of development (Figure 3B). This includes E3 ligases that control stem cell behavior (
), while other enzymes are induced upon differentiation and remain expressed in the resulting cell type. Under the right conditions, either a tissue-specific or a housekeeping E3 ligase might be the enzyme of choice for degrader development: tissue-specific enzymes could allow us to access the ubiquitylation machinery in only a subset of cells, thereby reducing the potential for on-target toxicity in organs that are not affected by a disease. Conversely, the generalists among E3 ligases might be particularly useful to build modular PROTAC handles that could be employed against many targets and across multiple tissues. Notably, many of the frequently engaged PROTAC-E3 ligases to date, including CRBN, are expressed at high levels across most tissues (Figure 3A). Other enzymes discussed above, such as the C-end-rule adaptor KLHDC2, the quality control E3 ligases UBR4 and UBR5, and the oxidative stress sensor KEAP1, also fall into the category of broadly expressed E3 ligases. 质量控制的E3连接酶可以被视为日常维护的酶，因此它们在大多数组织中以高水平表达（图3A）。相比之下，与泛素化在细胞命运规定中的关键作用相一致（
It is likely a reflection of our choices of laboratory model systems that few tissue-specific E3 ligases are understood in mechanistic detail. Some enzymes, however, have gained attention as their mutation causes developmental defects, an observation that could independently point to E3 ligases that act predominantly in a particular tissue. Mutations in the CUL3-adaptors KLHL31 and KLHL41, for example, trigger myopathies, and these proteins are accordingly expressed at high levels in heart and muscle (
). However, CUL3 family E3 ligases often monoubiquitylate their targets and may not be the best enzymes for drug discovery programs aimed at inducing protein degradation. It is thus interesting that E3 ligases of the MURF family, also known as TRIM54, TRIM55, and TRIM63, are similarly expressed, particularly in muscle and heart. These E3 ligases can drive proteasomal degradation, potentially in collaboration with CUL4 E3 ligases that have been implemented in PROTAC approaches (
), might provide a unique handle for inducing tissue-specific protein degradation. 很可能是由于我们选择的实验室模型系统有限，导致对于少数组织特异性的E3连接酶的机制细节了解不多。然而，一些酶因突变而引起发育缺陷，这一观察结果可能独立地指向在特定组织中起主导作用的E3连接酶。例如，CUL3适配蛋白KLHL31和KLHL41的突变会引发肌肉病，因此这些蛋白在心脏和肌肉中的表达水平较高。然而，CUL3家族的E3连接酶通常会对其目标进行单一泛素化，可能不是诱导蛋白降解的药物发现项目中最好的酶。因此，有趣的是，MURF家族的E3连接酶（也称为TRIM54、TRIM55和TRIM63）在肌肉和心脏中也有类似的表达。这些E3连接酶可以与已经应用于PROTAC方法的CUL4家族的E3连接酶合作，推动蛋白质酶解降解（
In addition to select expression in healthy tissues, E3 ligases that are induced in specific diseases could provide starting points for the development of molecular glues or PROTACs with reduced side effects. E3 ligases of the MAGE family are, for example, expressed only in germ cells, but become re-activated in cancer (
). Following this example, a survey of ∼600 human E3 ligases revealed several enzymes that are upregulated in small subsets of cancers and hence could be exploited to develop degradation approaches against the respective malignancy (Figure 3C). These include the plasma membrane E3 ligase RNF43, which restricts the accumulation of receptors of the WNT morphogen and can be mutated in colon cancer (
). RNF43 was recently engaged by bispecific antibodies, referred to as AbTACs, which bring the E3 ligase in proximity to a membrane protein to trigger ubiquitin-dependent endocytosis (
). We speculate that such AbTACs should work best in tissues that show the highest levels of RNF43. 除了在健康组织中选择表达外，特定疾病中诱导的E3连接酶可能为开发具有减少副作用的分子胶或PROTAC提供起点。例如，MAGE家族的E3连接酶仅在生殖细胞中表达，但在癌症中重新激活。根据这个例子，对大约600种人类E3连接酶的调查发现了几种在小部分癌症中上调的酶，因此可以利用它们来开发针对相应恶性肿瘤的降解方法。其中包括负调控WNT形态发生素受体积累的细胞膜E3连接酶RNF43，它在结肠癌中可能发生突变。最近，RNF43被称为AbTACs的双特异性抗体所利用，这些抗体将E3连接酶与细胞膜蛋白接近，触发泛素依赖的内吞作用。我们推测，这样的AbTACs在RNF43水平最高的组织中效果最好。
It is important to note that gene expression analyses are only an approximation of protein levels and must be complemented with quantitative proteomics before choosing E3 ligases based on abundance. In addition, even constitutively expressed E3 ligases might be activated only under specific conditions. Most subunits of the APC/C, for example, are present throughout the cell cycle, but a combination of inhibitors and posttranslational modifications ensures that this E3 ligase ubiquitylates its targets during mitosis and in the subsequent G1 phase (
). The same applies to RBR family E3 ligases that often exist in autoinhibited conformations that are overcome by specific partners, including other E3 ligases or phosphorylated ubiquitin (
). In addition, as shown for Cullin RING ligases, even expressed substrate adaptors do not form an active E3 ligase unless they have recognized their target and are allowed to engage the cullin scaffold in a process promoted by the exchange factor CAND1 (
). In many cases, we know little about additional factors that are required for E3 ligase function: indeed, many enzymes of the CUL3 family depend on proteins referred to as co-adaptors that provide the means for localized E3 ligase activation beyond mere adaptor expression (
). Finally, regulatory circuits that alter the intracellular location of an E3 ligase might also have an impact on the efficiency of PROTAC approaches using this enzyme (
). Choosing the best E3 ligase for drug discovery will therefore greatly benefit from mechanistic analyses geared toward revealing physiological mechanisms of regulation. 需要注意的是，基因表达分析只是蛋白质水平的近似值，在选择丰度为基础的E3连接酶之前，必须与定量蛋白质组学相结合。此外，即使是持续表达的E3连接酶也可能只在特定条件下被激活。例如，大多数APC/C的亚单位在细胞周期中始终存在，但是抑制剂和翻译后修饰的组合确保了这种E3连接酶在有丝分裂和随后的G1期间泛素化其靶标。同样，RBR家族的E3连接酶通常存在于被自我抑制的构象中，这些构象可以被特定的伴侣（包括其他E3连接酶或磷酸化的泛素）克服。此外，正如Cullin RING连接酶所示，即使是表达的底物适配体也不会形成活性的E3连接酶，除非它们已经识别到其靶标并被允许与CAND1交换因子促进的Cullin支架发生作用。 在许多情况下，我们对于E3连接酶功能所需的其他因素了解甚少：事实上，CUL3家族的许多酶依赖于被称为辅助蛋白的蛋白质，这些蛋白质提供了定位的E3连接酶激活方式，超越了仅仅适配器的表达（
Most degrader programs do not wish to saturate the E3 ligase with the PROTAC handle, so that the physiological function of the ubiquitylation enzyme is not affected. However, such an effect is not necessarily deleterious: an E3 ligase inhibitor used as a handle could produce a PROTAC that both stabilizes the natural E3 target and degrades its neo-substrate. In cancer therapy, such polypharmacology could result in synthetic lethality and yield therapeutic effects at lower compound concentrations. Indeed, the aforementioned PROTACs built around an MDM2 inhibitor not only elicited degradation of their intended neo-substrate, but also stabilized TP53, and accordingly killed cancer cells more efficiently than PROTACs relying on an innocuous E3 handle (
). In a similar manner, PROTACs engaging cIAP1 could gain efficiency against cancer cells by simultaneously degrading their neo-substrate and promoting caspase activation (
). Polypharmacology does not need to be limited to oncology: PROTACs based on inhibitors of the oxidative stress sensor KEAP1 could both stabilize the physiological substrate of CUL3KEAP1, the antioxidant transcription factor NRF2, and induce degradation of the target protein of the second PROTAC warhead (
), KEAP1-targeting PROTACs might be of particular use in neurodegenerative disease. 大多数降解程序不希望用PROTAC手柄饱和E3连接酶，以免影响泛素化酶的生理功能。然而，这种效果不一定是有害的：作为手柄使用的E3连接酶抑制剂可以产生既稳定天然E3靶点又降解其新底物的PROTAC。在癌症治疗中，这种多药作用可能导致合成致死性，并在较低的化合物浓度下产生治疗效果。事实上，围绕MDM2抑制剂构建的上述PROTAC不仅引发了其预期的新底物的降解，还稳定了TP53，并且比依赖无害的E3手柄的PROTAC更有效地杀死了癌细胞。以类似的方式，与cIAP1相互作用的PROTAC可以通过同时降解其新底物和促进半胱天冬酶活化来提高对癌细胞的效果。 多功能药理学不仅限于肿瘤学：基于氧化应激传感器KEAP1抑制剂的PROTACs可以同时稳定CUL3的生理底物抗氧化转录因子NRF2，并降解第二个PROTAC战斗头的目标蛋白。由于NRF2促进蛋白酶体亚单位的表达并且对神经毒性蛋白聚集具有高度保护作用，针对KEAP1的PROTACs在神经退行性疾病中可能特别有用。
In a variation on the theme of polypharmacology, many PROTACs use an IMID as their starting point and, thus, a molecular glue that by itself can trigger protein degradation (
). If the initial molecular glue retains its activity within the PROTAC, such compounds could lead to degradation of multiple proteins at a time to produce a more sustained therapeutic effect. Indeed, the aforementioned PROTACs built on IMIDs as E3 handles typically retain their ability to trigger the turnover of Ikaros, Aiolos, and likely other targets, which could add to the effects caused by the degradation of the intended target. Synthetic lethality screens will provide a powerful means to determine degradation pairs in a rational manner that would be effective in a specific disease background (
As this review has illustrated, progress in the field of induced protein degradation has been stunning. A mere 10 years ago, E3 ligases were deemed undruggable, yet we are now discussing how to engage many of the ∼600 human ubiquitylation enzymes for innovative drug discovery. While we focused this review on enzymes that elicit proteasomal degradation, the universe of induced ubiquitylation extends far beyond this application. It includes E3 ligases of the CUL3 family that can monoubiquitylate intrinsically disordered proteins to keep them soluble (
). Induced monoubiquitylation might therefore be useful against proteins, such as FUS or TDP-43, that are essential and cannot be degraded, yet cause a toxic gain-of-function when aggregating (
). We can also envision scenarios in which it is attractive to decorate pathological proteins with K63-linked ubiquitin chains that establish reversible protein interactions (
). As an example, compounds that deliver the catalytic domain of the K63-specific NEDD4 to the mitochondrial outer membrane trigger organellar clustering in perinuclear regions (
). In fact, we can imagine molecules that pair proteins with enzymes installing ubiquitin-like modifications: compounds that link pathological variants of Huntingtin to the ubiquitin-like LC3, an inducer of autophagosome formation (
The galaxy of E3 ligases that is available for innovative drug discovery therefore seems endless. But, in the words of Douglas Adams, “don't panic.” Since the days when E3 ligases were thought to be undruggable, we have learned much about their structure, function, and regulation, which revealed deep biology that can now be exploited for tailored drug discovery. We are beginning to dissect gene expression programs at the single-cell level throughout metazoan development, and even proteomics has reached the age of single-cell detection (
). As we will learn when and where E3 ligases are expressed, we will be able to predict ever more precisely which ubiquitylation enzyme should be engaged by a PROTAC. This progress is complemented by sophisticated genetic screening platforms that point to E3 ligases essential for cell function or survival, providing us with entry points that might prevent a cancer from becoming resistant against treatment. Exciting developments in small-molecule discovery, highlighted by the integration of DNA-encoded libraries with artificial intelligence (
), will greatly facilitate endeavors to find chemical matter as E3 binders. The biology of ubiquitylation is therefore an exciting launching pad for developing new therapeutic paradigms that will allow us to tackle diseases—rather than proteins—that previously were deemed to be undruggable. E3连接酶的星系似乎是无穷无尽的，因此对于创新药物发现来说，选择很多。但是，用道格拉斯·亚当斯的话来说，“不要惊慌”。自从E3连接酶被认为是无法药物化的那一天起，我们对它们的结构、功能和调控有了很多了解，揭示了深层次的生物学，现在可以利用这些知识进行定制药物发现。我们开始在多细胞动物发育过程中剖析基因表达程序，甚至蛋白质组学也已经进入了单细胞检测的时代。当我们了解E3连接酶在何时何地表达时，我们将能够更加准确地预测哪种泛素化酶应该与PROTAC结合。这一进展得到了复杂的基因筛选平台的支持，这些平台指出了对细胞功能或生存至关重要的E3连接酶，为我们提供了可能防止癌症对治疗产生抗药性的入口点。小分子发现方面的令人兴奋的进展，通过DNA编码文库与人工智能的整合（）突出了这一点，将极大地促进寻找作为E3连接酶结合物的化学物质的努力。 泛素化的生物学因此成为开发新的治疗范式的激动人心的起点，这将使我们能够解决以前被认为无法治疗的疾病，而不是蛋白质。
Significance 重要性
Induced protein degradation has emerged as a new therapeutic modality that enables elimination, rather than mere inhibition, of previously undruggable pathological proteins. Small-molecule degraders induce proximity between an E3 ligase and the therapeutic target and thereby elicit protein ubiquitylation and proteasomal degradation. Although the human genome encodes ∼600 E3 ligases, very few are currently used to trigger small-molecule-dependent degradation. By carefully selecting new E3 ligases as targets for drug discovery, we can impose tissue- or disease-specific degradation, overcome resistance to degraders used in the clinic, and eliminate proteins that so far have resisted this therapeutic modality. Novel E3 ligases will therefore greatly expand the therapeutic reach of induced protein degradation. 诱导蛋白降解已成为一种新的治疗模式，能够消除先前无法药物治疗的病理蛋白，而不仅仅是抑制它们。小分子降解剂能够促使E3连接酶与治疗靶点接近，从而引发蛋白泛素化和蛋白酶体降解。虽然人类基因组编码了大约600种E3连接酶，但目前只有很少数被用于触发小分子依赖性降解。通过精选新的E3连接酶作为药物发现的目标，我们可以实现组织或疾病特异性的降解，克服临床使用的降解剂的耐药性，并消除迄今为止抵抗这种治疗模式的蛋白质。因此，新型E3连接酶将极大地扩展诱导蛋白降解的治疗范围。
Acknowledgments 致谢
We apologize to those whose work we could not include due to space constraints. We are grateful to all members of the Rapé lab and our collaborators for stimulating discussions and for testing the boundaries of knowledge every day. P.J. is a recipient of a Siebel Institute postdoctoral fellowship; D.L.H. is a recipient of an HHMI/Helen Hay Whitney postdoctoral fellowship; M.R. is an Investigator of the Howard Hughes Medical Institute. 我们对由于空间限制而无法包含的工作表示歉意。我们感谢Rapé实验室的所有成员和合作者每天进行的激发性讨论和对知识边界的测试。P.J.是Siebel研究所博士后奖学金的获得者；D.L.H.是HHMI/Helen Hay Whitney博士后奖学金的获得者；M.R.是Howard Hughes医学研究所的研究员。
Author contributions 作者贡献
P.J. and M.R. planned the first draft of the manuscript; P.J. and D.H. provided figures; P.J., D.H., and M.R. wrote the manuscript. P.J.和M.R.计划了手稿的初稿；P.J.和D.H.提供了图表；P.J.，D.H.和M.R.撰写了手稿。
Declaration of interests 利益声明
M.R. is a founder of Nurix Therapeutics and member of its scientific advisory board. M.R. is a member of the scientific advisory of Monte Rosa Therapeutics. M.R. holds patents in the induced protein degradation space. M.R.是Nurix Therapeutics的创始人和科学顾问委员会成员。M.R.是Monte Rosa Therapeutics的科学顾问委员会成员。M.R.在诱导蛋白质降解领域拥有专利。
References 参考资料
Alazami A.M. 阿拉扎米 A.M.
Al-Saif A. 阿尔萨伊夫A.
Al-Semari A. 阿尔塞马里A.
Bohlega S. 波赫莱加S.
Zlitni S.
Alzahrani F. 阿尔扎赫拉尼F.
Bavi P. 巴维P.
Kaya N. 卡亚N.
Colak D.
Khalak H. 哈拉克H.
et al. 等人
Mutations in C2orf37, encoding a nucleolar protein, cause hypogonadism, alopecia, diabetes mellitus, mental retardation, and extrapyramidal syndrome. C2orf37基因突变导致性腺发育不全、脱发、糖尿病、智力障碍和锥体外系综合征。
Am. J. Hum. Genet.2008; 83 美国人类遗传学杂志，2008年；83: 684-691
Mutations in CUL4B, which encodes a ubiquitin E3 ligase subunit, cause an X-linked mental retardation syndrome associated with aggressive outbursts, seizures, relative macrocephaly, central obesity, hypogonadism, pes cavus, and tremor.
The SCFbeta-TRCP-ubiquitin ligase complex associates specifically with phosphorylated destruction motifs in IkappaBalpha and beta-catenin and stimulates IkappaBalpha ubiquitination in vitro.
00157-4&id=fx1.jpg){kind=link}
00157-4&id=gr1.jpg){kind=link}
00157-4&id=gr2.jpg){kind=link}
00157-4&id=gr3.jpg){kind=link}