How the ends signal the end: Regulation by E3 ubiquitin ligases recognizing protein termini.

Specificity of eukaryotic protein degradation is determined by E3 ubiquitin ligases and their selective binding to protein motifs, termed "degrons," in substrates for ubiquitin-mediated proteolysis. From the discovery of the first substrate degron and the corresponding E3 to a flurry of recent studies enabled by modern systems and structural methods, it is clear that many regulatory pathways depend on E3s recognizing protein termini. Here, we review the structural basis for recognition of protein termini by E3s and how this recognition underlies biological regulation. Diverse E3s evolved to harness a substrate's N and/or C terminus (and often adjacent residues as well) in a sequence-specific manner. Regulation is achieved through selective activation of E3s and also through generation of degrons at ribosomes or by posttranslational means. Collectively, many E3 interactions with protein N and C termini enable intricate control of protein quality and responses to cellular signals.

Introduction

Cellular homeostasis relies on pathways to selectively eliminate proteins that are aberrant, superfluous, or inhibitory to biochemical processes. In eukaryotes, such protein clearance is mediated by the ubiquitin-proteasome-autophagy system: proteins requiring destruction are marked with ubiquitin, which is recognized by the 26S proteasome or autophagy machineries that mediate degradation of the ubiquitin-tagged entity. Specificity is determined by numerous E3s—more than 600 in humans—tagging proteins with ubiquitin. Some E3s are single polypeptides; others are multiprotein complexes (Figure 1). Both types of E3 typically encompass two functionalities: (1) binding to a substrate, and (2) catalyzing its covalent linkage to ubiquitin. The latter functionality is typically mediated by a hallmark E3 domain, such as RING or HECT, that either indirectly mediates ubiquitin transfer from the active site of a partner E2 enzyme or contains a catalytic Cys that carries ubiquitin's C terminus directly to a remotely bound substrate.

Figure 1

Diversity of E3 ligases targeting terminal degrons

(A) Single-chain UBR-family E3 ligases. Cartoons of RING E3 UBR1 (top) and HECT E3 UBR5 (bottom), with catalytic domains in blue and substrate-binding domains, UBR-box-1 and UBR-box-2 (aka N-domain), in pink and violet.

(B) Multi-subunit E3 ligases. Cartoons of GID (top) and CRL (bottom) RING E3s, with their catalytic subunits in blue, and interchangeable substrate receptors having indicated protein-protein interaction domain folds in red, slate, and purple.

Substrates for ubiquitylation are recruited to receptor domains or subunits within the E3s (Figure 1). Single-subunit E3s contain embedded substrate receptor domains. Meanwhile, multisubunit E3s, including cullin-RING ligase (CRL) or GID/CTLH E3s, typically employ variable and interchangeable substrate receptors that recruit different proteins for ubiquitylation by a common catalytic module. Irrespective of the type of E3, each substrate receptor recognizes "degron" motifs within substrates. Degrons are frequently short linear sequence motifs, often generated only after a posttranslational modification is added or removed.

The first identified degrons mapped to the N termini of proteins: Varshavsky and colleagues observed that the half-life of β-galactosidase in Saccharomyces cerevisiae depends on the identity of its N-terminal residue (Bachmair et al., 1986). The study discovered a set of amino acids that, when placed at the N terminus of the model substrate, confer instability. Subsequently, they employed a genetic screen to discover Ubr1 as the E3 mediating this regulation. Ubr1 has a RING E3 domain, and UBR-box-1 and UBR-box-2 (also known as the N-domain) substrate receptor domains recognizing basic and hydrophobic N-terminal residues, respectively (Baker and Varshavsky, 1991; Bartel et al., 1990). Thus, regulation by the then-called "N-end rule," subsequently renamed as "N-degron pathway," was already appreciated three decades ago as involving a protein's N-terminal sequence serving as a degron and its recognition by a specific substrate receptor domain in an E3.

Numerous N- and C-degron pathways

An early indication that many E3s may recognize protein N termini was the finding that seven mammalian proteins (UBR1 to UBR7), varying in size from 48 to 572 kDa and in the type of E3 catalytic domain, have UBR-box-1 and/or UBR-box-2 domains (Kwon et al., 1998; Tasaki et al., 2005; Tasaki et al., 2009). Most of these have been experimentally demonstrated to be N-degron E3s (Tasaki et al., 2009).

There were also early indications that N-degron pathways could be regulated, when intricate biochemical pathways were found to append an Arg at protein N termini (Gonda et al., 1989). These studies portended distinct pathways regulating the generation of different protein termini serving as degrons, designated by "the distinguishing feature"/"terminus-degron." For example, in the "Arg/N-degron" pathway, N-terminal Arg confers instability. All 20 amino acids have been demonstrated to act as destabilizing N-terminal residues (Varshavsky, 2019), with distinct regulation deciphered for Ac/N-degron, Pro/N-degron, and Gly/N-degron pathways (Chen et al., 2017; Hwang et al., 2010; Timms et al., 2019).

A great leap in our knowledge of terminal degron pathways came from prospective systems biology studies. First, the Elledge and Yen labs used global protein stability profiling to identify new degrons and match them with CRL substrate receptors (Koren et al., 2018; Lin et al., 2018). Briefly, they appended the entire collection of human open reading frames (Koren et al., 2018; Lin et al., 2018), or sequences corresponding to peptides derived from all human proteins (nested sets of 90-mers, with 45-residue overlap), to the C terminus of GFP (Koren et al., 2018). The level of the GFP-fused protein was normalized to that of a control, DsRed, expressed from the same transcript. This allowed screening for effects of genetically perturbing specific E3 ligases on the stability of individual peptides or proteins. Although the screens did not intentionally seek C-degrons, systematic interrogation of the destabilizing sequences pointed toward C-terminal degron motifs. The E3s targeting these sequences were identified by monitoring GFP stabilization upon CRL inhibition, overexpression of dominant-negative mutant cullins, and CRISPR screening revealing corresponding substrate receptors. This campaign discovered several C-degron pathways, including Gly/, ArgXXGly/, and Arg/C-degrons targeted by specific E3s containing CUL2-RBX1 as their catalytic cores (called CRL2-family E3s); GluGlu/ and ArgXX/C-degrons targeted by E3s containing CUL4-RBX1 as their catalytic cores (called CRL4-family E3s); and also Ala/ and Val/C-degrons conferring instability by unknown non-CRL-type E3s (Koren et al., 2018; Lin et al., 2018). Elledge and colleagues also used a related approach to massively expand our knowledge of N-degron pathways (Timms et al., 2019). Additional N-degron regulation was discovered by systems biology studies monitoring protein stability and roles of E3s in yeast (Kats et al., 2018; Kong et al., 2021). Meanwhile, unexpected C-degron regulation was revealed by a crystal structure of FBXO31, a CRL1-family F-box protein substrate receptor bound to a peptide from cyclin D1 (Li et al., 2018).

It is now appreciated that protein termini are particularly suited to serve as degrons as they are frequently disordered, solvent-exposed, and accessible to substrate receptors. Moreover, regulation occurs when terminal degrons are produced in an inactive or masked form and conditionally activated through posttranslational modifications or proteolytic cleavage in response to cellular signals (Lee et al., 2016; Tasaki et al., 2012; Timms and Koren, 2020; Varshavsky, 2011). Here, we review the elements of terminal degron pathways—the structurally diverse E3 substrate receptor domains recognizing terminal sequences—and how specific pathways control these elements to achieve regulation (Table 1).

Table 1

Properties of E3 ligases targeting terminal degrons

E3	Substrate receptor /recognition domain	Degron recognition fold	Terminal residue/degron	Degron generation	Catalytic domain	Pathway
Yeast Ubr1	UBR-box-1	UBR-box-1	R,H,K	Deamidation, arginylation, proteolysis	RING	N-degron
Yeast Ubr1	UBR-box-2	UBR-box-2	F,W,L,Y,I, M	Deamidation, arginylation, proteolysis	RING	N-degron
Human UBR1	UBR-box-1	UBR-box-1	R,H,K	Deamidation, arginylation, proteolysis	RING	N-degron
Human UBR1	UBR-box-2	UBR-box-2	F,W,L,Y,I, M	Deamidation, arginylation, proteolysis	RING	N-degron
Human UBR2	UBR-box-1	UBR-box-1	R,H,K	Deamidation, arginylation, proteolysis	RING	N-degron
Human UBR2	UBR-box-2	UBR-box-2	F,W,L,Y,I, M	Deamidation, arginylation, proteolysis	RING	N-degron
Human UBR4	UBR-box-1	UBR-box-1	R,H,K	Deamidation, arginylation, proteolysis	RING	N-degron
Human UBR5	UBR-box-1	UBR-box-1	R,H,K	Deamidation, arginylation, proteolysis	HECT	N-degron
Yeast GID	Gid4	Calycin-superfamily β-barrel	P, F,I, L, V	Aminopeptidase cleavage	RING	N-degron
Yeast GID	Gid10	Calycin-superfamily β-barrel	P,G,F,M,V,A	Aminopeptidase cleavage	RING	N-degron
Yeast GID	Gid11	WD40 β-propeller	T	Aminopeptidase cleavage	RING	N-degron
Human CTLH	Gid4	Calycin-superfamily β-barrel	P, F,L,V, I,W,Y	–	RING	N-degron
Yeast Psh1	–	–	Formyl-M	Formylation	–	N-degron
CRL1	FBXO31	Calycin-superfamily β-barrel	I	–	RING	C-degron
CRL2	ZER1	Armadillo repeats	G	Caspase cleavage, failure of N-myristoylation	RING	N-degron
CRL2	ZYG11B	Armadillo repeats	G	Caspase cleavage, failure of N-myristoylation	RING	N-degron
CRL2	KLHDC1/2/3/10	Kelch repeat β-propeller	GG, RG, QG, KG, WG, AG, PG	Premature protein termination, autocleavage	RING	C-degron
CRL2	FEMB1A/B/C	Ankyrin repeats	R	–	–	C-degron
CRL2	APPBP2	Tetratricopeptide repeats	RXXG	–	–	C-degron
CRL2	KLHCD10	Kelch repeat β-propeller	polyA	Ala-tailing	RING	C-degron
Pirh2	–	–	polyA	Ala-tailing	RING	C-degron
CRL4	DCAF12	WD40 β-propeller	EE	–	RING	C-degron
CRL4	TRPC4AP	Armadillo-like repeats	RXX	–	RING	C-degron
CHIP	–	Tetratricopeptide repeat	D	Caspase cleavage	U-box	C-degron
Yeast NOT4Yeast Doa10Human MARCH6	–	–	Ac-M/Ac-A/Ac-S/Ac-T/Ac-C	Acetylation	RING	N-degron

Recognition of protein termini by E3 substrate receptors

Structures—either determined experimentally or by modeling—have revealed molecular principles underlying terminal degron recognition. First, substrate-binding sites are spatially separate from E3 catalytic domains, yet they enable directing the targeted lysines toward the ubiquitylation active site (Baek et al., 2020; Horn-Ghetko et al., 2021; Pan et al., 2021; Qiao et al., 2020; Sherpa et al., 2021). Second, some particular protein-binding folds have recurrently evolved to selectively recruit terminal degrons. Third, in general, each E3 recognizes distinct protein terminal properties, either the terminal residues alone or together with subsequent or preceding residues; constellations of acidic or basic residues also harness N or C termini, respectively. Finally, and unexpectedly, some N- and C-degron-binding E3s recognize multiple different substrate motifs, either through having more than one substrate receptor domain that each can recruit different cohorts of substrates or through one substrate receptor binding divergent degrons. Below, we describe how different folds establish these principles by recognizing their cognate terminal degrons.

N-degron recognition by UBR-box domains

The founding E3 Ubr1 has the unusual property of containing not one but multiple substrate receptor domains. Ubr1 recognizes non-terminal degrons through an elusive substrate-binding mechanism and has two discrete N-degron-binding domains (Baker and Varshavsky, 1991; Xia et al., 2008b).

The UBR-box-1 domain, specific for basic N-terminal residues (Arg, His, and Lys), is conserved in all UBR E3s; however, to date, Arg/N-degron recognition has been demonstrated only for a subset of UBR orthologs in higher eukaryotes (Tasaki et al., 2005; Tasaki et al., 2009). Crystal structures showed that the UBR-box-1 domain adopts a zinc-stabilized triangular shape, with a shallow cleft distinguishing Arg/N-degron features in two pockets (Choi et al., 2010; Kim et al., 2020; Matta-Camacho et al., 2010; Muñoz-Escobar et al., 2017; Wang et al., 2008) (Figure 2A). In one, a constellation of acidic residues intimately binds the N-terminal residue's basic side chain, directly and indirectly via waters. Additionally, the N-terminal amino group is captured through three hydrogen bonds. The second pocket is hydrophobic and encapsulates the bulky side chain of the second residue.

Figure 2

N-degron recognition by UBR-family substrate-binding domains

(A) UBR-box-1 from yeast Ubr1 bound to Arg/N-degron of SCC1 (PDB ID:3NIN). Substrate-binding cleft is highlighted in dotted lines. Specificity of Arg/N-degron binding is established by an acidic pocket and network of waters binding the N terminus and Arg side chain and a hydrophobic pocket binding the subsequent residue.

(B) Overlay of UBR-box-2 from yeast Ubr1 (slate, PDB ID:7MEX) with N-degron peptide-bound bacterial ClpS (navy, PDB ID: 3DNJ). A hydrophobic pocket binds the hydrophobic N-terminal side chain. The N terminus and backbone are secured by hydrogen bonds.

(C) Cryo-EM structure representing ubiquitin transfer to an Arg/N-degron substrate by yeast Ubr1 and E2 (PDB ID: 7MEX). UBR-box-1 and UBR-box-2 are pink and slate, respectively. Active site (chemically stable mimic of thioester-linkage between E2 and ubiquitin) is marked with a black star.

The UBR-box-2 domain recognizes N-terminal bulky residues (Phe, Trp, Leu, Tyr, and Ile) (Baker and Varshavsky, 1991; Xia et al., 2008b) and an initiator Met followed by a hydrophobic residue (Kim et al., 2014; Park et al., 2015). This domain is conserved in yeast Ubr1, its mammalian orthologs UBR1 and UBR2, and the bacterial protein ClpS that is an adaptor recruiting substrates for degradation by ClpA/ClpP protease (Erbse et al., 2006). The UBR-box-2 domain without a bound degron is observed in the cryo-EM structure of Ubr1 (Pan et al., 2021), while N-degron recognition is revealed by the structures of ClpS (Kim et al., 2021; Román-Hernández et al., 2009; Schuenemann et al., 2009). The degron N terminus forms electrostatic interactions with polar side chains at the edge of a hydrophobic pocket that snugly embraces the hydrophobic side chain (Figure 2B).

Cryo-EM structures visualized Ubr1 and its cognate E2 (Ubc2) in stable complexes mimicking ubiquitin transfer to an Arg/N-degron substrate or substrate polyubiquitylation (Pan et al., 2021). The two substrate binding domains are rigidly fixed on one side (Figure 2C). The activated Ubc2∼ubiquitin intermediate, with ubiquitin's C terminus linked to Ubc2's catalytic Cys, is located on the opposite side of the structure. The structure showed the substrate sandwiched between the N-degron-binding domain and the active site.

Calycin-superfamily β-barrels recognizing terminal degrons

In calycin-superfamily β-barrels, a continuous antiparallel β-sheet forms a cylinder that is typically blocked at one end (Flower et al., 2000). The other end displays a funnel-like opening comprising long flexible loops between β-strands. Ligands bind by insertion into the funnel. Well-characterized members of this superfamily are lipocalins, which bind metabolites or other hydrophobic small-molecule ligands (Huber et al., 1987; Korndörfer et al., 2003). Recently, however, several substrate receptor subunits of multiprotein E3s have been discovered to form such β-barrel structures capturing terminal degrons.

FBXO31 is one of nearly 70 human F-box proteins, which associate via their F-box domains interchangeably with the CUL1-RBX1 cullin-RING complex. FBXO31 binds cyclin D1's C-terminal degron during genotoxic stress (Santra et al., 2009). A crystal structure revealed that the FBXO31's substrate-binding domain adopts a calycin-type β-barrel (Li et al., 2018) (Figure 3A). The degron's C terminus is captured by a lysine at the base of FBXO31's central funnel; upstream residues adopt a helical conformation contacting β-barrel strands and the surrounding loops.

Figure 3

Pliable funnel-shaped β-barrel folds recognize diverse terminal degrons

(A) C-degron-binding pocket (dotted lines) in FBXO31 CRL1 substrate receptor (PDB:5VZU). FBXO31 lysine at the base of funnel entrance to calycin-superfamily-type β-barrel anchors C terminus of helical degron from cyclin D1.

(B) GID substrate receptors, human Gid4 (red, PDB ID:6CDC) and yeast Gid10 (pink, PDB ID:7QQY), embracing peptide N termini in deep pockets (shown in dotted lines) anchored by contacts with conserved glutamate and tyrosine at the base of substrate-binding tunnels.

(C) Pliable loops of calycin-superfamily-type β-barrels from GID substrate receptors confer plasticity allowing recognition of diverse N-terminal sequences. Overlay of human Gid4 structures bound to indicated peptides initiating with Pro and Phe (PDB ID:6CDC, 7Q50, respectively). Loops at the entrance to N-degron-binding pocket are numbered.

(D) Hydrogen bonds and hydrophobic interactions enabling diverse N-terminal sequences to bind human Gid4.

Two different calycin-type β-barrel proteins—Gid4 and Gid10—are interchangeable receptors for yeast multisubunit GID E3s. The GID complex was originally identified for regulation of gluconeogenic enzymes. Gluconeogenic production of glucose is required for yeast growth on nonfermentable carbon sources; however, when glucose is restored to the media, key enzymes catalyzing gluconeogenesis are degraded in a GID E3-dependent manner (Liu and Pfirrmann, 2019; Menssen et al., 2012). "GID" refers to the glucose-induced degradation deficiency caused by mutations in E3 subunits. The subunit Gid4 recognizes N-terminal prolines of gluconeogenic substrates (Chen et al., 2017; Hämmerle et al., 1998). It is now known that yeast GID recruits distinct substrates through interchangeable receptors, including Gid10, that binds the N-terminal Pro of its only currently known substrate Art2 (Langlois et al., 2021). Moreover, either fortuitously or through systematic screening of peptides in arrays or on phage, a number of N-terminal peptides, including several initiating with non-Pro residues, were found to bind human and yeast Gid4 and Gid10 (Chrustowicz et al., 2021; Dong et al., 2020; Melnykov et al., 2019; Shin et al., 2021). Binding affinities are not only influenced by the N-terminal residues but also by conformations and interactions made by several downstream residues.

Structures showed that N-degrons protrude through the funnel-like entry of GID substrate receptor β-barrels and penetrate the central tunnel (Chrustowicz et al., 2021; Dong et al., 2020; Dong et al., 2018; Langlois et al., 2021; Shin et al., 2021) (Figure 3B). Acidic and polar side chains conserved between Gid4 and Gid10 anchor peptide N termini, with the side chain docked in a hydrophobic pocket. Downstream residues contact flexible loops that form the funnel-shaped entrance to the β-barrel tunnel. The loops adopt various conformations that maximize interactions with each individual N-terminal peptide (Figures 3C and 3D). The N-degrons adopt sequence-specific structures complementing the conformation of the binding groove (Chrustowicz et al., 2021). We propose that FBXO31 may likewise exhibit conformational pliability to bind diverse C-degrons.

Terminal degron-binding β-propellers

β-propellers are common protein-protein interaction domains, where a toroidal arrangement of β-sheets generates the propeller shape (Chen et al., 2011). Many interchangeable substrate receptors of multiprotein E3s recruit substrates to a β-propeller domain central pore. To date, six substrate receptors of multiprotein E3s have been found to recognize terminal degrons via β-propeller domains. C-degrons are recognized by Kelch β-propellers in KLHDC1, KLHDC2, KLHDC3, and KLHDC10, which function with either a CUL5-RBX2 or CUL2-RBX1 CRL catalytic core and by the WD40-repeat β-propellers in DCAF12, which functions with a CUL4-RBX1 CRL catalytic core (Koren et al., 2018; Lin et al., 2018; Okumura et al., 2020). The WD40-repeat protein Gid11 is unique among GID substrate receptors in recognizing N-terminal Thr, and not Pro, but the basis for this specificity remains unknown (Kong et al., 2021) (Figure 4A).

Figure 4

β-propeller and helical-repeat domains binding terminal degrons

(A) AlphaFold-predicted WD40-type β-propeller of yeast GID substrate receptor Gid11.

(B) Kelch-type β-propeller of CRL2 substrate receptor KLHDC2 bound to diGly/C-degron of SelK (PDB ID: 6DO3, degron-binding pocket in dotted lines). KLHDC2 contacts the the diGly/C-degron via direct and water-mediated hydrogen bonds.

(C) Armadillo repeats of CRL2 substrate receptor ZYG11B bound to Gly/N-degron (PDB ID: 7EP2). Gly/N-degron-binding pocket (dotted lines) is a narrow cavity forming hydrogen bonds with N-terminal residues.

(D) Ankyrin repeats of CRL2 substrate receptor FEM1B bound to Arg/C-degron of CDK5R1 (PDB ID:7CNG). Arg/C-degron degron-binding pocket (dotted lines) recognizes extended peptide structure: C-terminal Arg by hydrophobics flanking aliphatic portion of its side chain and acidic pocket capturing its guanidino group, and upstream hydrophobic residues by a hydrophobic pocket.

(E) The same groove in FEM1B binds Arg/C-degrons and Cys/His-rich internal degron of FNIP1 internal degron (PDB IDs: 7CNG, 7ROY). Cys/His-rich surfaces of FEM1B and Cys/His-rich internal degron form intermolecular zinc-binding site. Zinc ions (gray spheres) function as molecular glue.

How a β-propeller recognizes a terminal degron is revealed by crystal structures of KLHDC2 bound to various C-degron peptides (Rusnac et al., 2018) (Figure 4B). The β-propeller's central pore binds the diGly/C-degron and upstream residues through a network of hydrogen bonds, including via water molecules.

Helical repeats recognizing diverse degrons

Helical repeats, including armadillo, tetratricopeptide, and ankyrin repeats, form curved structures with shallow clefts that bind extended peptide-like sequences. These protein-protein interaction domains are often found in CRL substrate receptors, including Gly/N-degron-binding ZER1 and ZYG11B, ArgXXGly/C-degron-binding APPBP2, and Arg/C-degron-binding FEM1 family members that partner with the CUL2-RBX1 CRL catalytic core (Chen et al., 2021b; Koren et al., 2018; Timms et al., 2019; Yan et al., 2021a; Yan et al., 2021b; Zhao et al., 2021).

ZYG11B and ZER1 bind Gly/N-degrons similarly. A narrow cavity captures the amino group of an N-terminal Gly and interacts with the first four N-degron residues (Yan et al., 2021a) (Figure 4C).

Crystal structures showed ankyrin repeats of FEM1B and FEM1C complexes with several C-degrons (Chen et al., 2021b; Yan et al., 2021b; Zhao et al., 2021). C-degron binding is facilitated by multiple hydrogen bonds with the degron's C terminus and hydrophobic interactions with upstream residues (Figure 4D). Intriguingly, the same FEM1B cleft also binds internal degrons, albeit through a distinct mechanism. A structure with the substrate FNIP1 shows that Cys- and His-rich sequences of both FEM1B and FNIP1 co-chelate zinc atoms at the E3-substrate interface. This interaction is regulated by redox state, with FEM1B targeting FNIP1 for degradation as a part of reductive stress pathways (Manford et al., 2021; Manford et al., 2020).

Regulation of terminal degron-binding E3s

Below, we provide several examples of how terminal degron E3s are regulated through the conditional expression of substrate receptor subunits, higher-order assembly, and/or interactions with other factors that allosterically modulate substrate binding.

Regulated incorporation of substrate receptor subunits into multiprotein E3s

Substrate recognition by CRL and GID/CTLH multiprotein E3s depends on a variable substrate receptor subunit associating with core scaffold and catalytic subunits. CRLs assemble through an intricate mechanism, "on-demand," when substrates bind their receptors (Wang et al., 2020). Meanwhile, GID E3 assembly is regulated by expression of the substrate receptor subunit depending on different environmental stimuli (Figure 5A). The first GID E3 substrate receptor identified was Gid4, which is upregulated during glucose-induced recovery from growth on a nonfermentable carbon source, and it binds Pro/N-degrons of gluconeogenic enzymes to be degraded in these conditions (Chen et al., 2017; Menssen et al., 2018; Menssen et al., 2012; Santt et al., 2008). Environmental perturbations, such as heat shock, osmotic stress, and carbon, nitrogen, or amino acid starvation regulate expression of the other substrate receptors, the Gid4 homolog Gid10 (Figure 3B), and the structurally distinct Gid11 (Figure 4A) (Chrustowicz et al., 2021; Karayel et al., 2020; Kong et al., 2021; Langlois et al., 2021; Melnykov et al., 2019; Qiao et al., 2020). Although at this point, Gid4 is the only such substrate receptor subunit identified for the corresponding higher eukaryotic CTLH E3, some CTLH E3 substrates are degraded in a Gid4-independent manner suggesting multiple modes of substrate recruitment (Mohamed et al., 2021; Zavortink et al., 2020).

Figure 5

Regulation of terminal degron E3 ligases

(A) Regulated incorporation of substrate receptors activates GID E3 ligases: Gid4 that is induced during glucose recovery targets Pro/N-degron gluconeogenic enzymes Fbp1, Mdh2, Icl1, and Pck1; Gid10 that is induced during heat and osmotic shock targets Pro/N-degron substrate Art2; Gid11 that is induced in distinct environmental conditions is thought to target Thr/N-degron substrates Phm8, Gpm3, Yor283w, Cpa1, and Blm10.

(B) Supramolecular assembly of yeast GID E3 allows avid binding between two central-facing Gid4 substrate receptors and Pro/N-degrons from two protomers of oligomeric substrate Fbp1, much like how an organometallic chelate binds a central ligand. The so-called Chelator-GIDSR4 comprises the core consisting of the substrate receptor, and scaffolding and catalytic modules, joined by a supramolecular assembly module (SA). The Chelator assembly targets ubiquitin ligation to functionally relevant sites depending on Fbp1 quaternary structure.

(C) Pseudosubstrate regulation of substrate targeting by yeast E3 Ubr1. Occupancy of UBR-box-1 and UBR-box-2 sites by dipeptides mimicking N-degrons, with N-terminal Arg (RX) and hydrophobic residues (ΦX), promotes binding of Cup9 internal degron to a distinct substrate-binding site. UBR-box-1-binding to N-degron-like terminus of Roq1 generated by stress-induced Ynm3-mediated cleavage induces Ubr1 targeting of stress-related substrates.

Regulation by supramolecular assembly

A higher-order GID E3 assembly is required for degradation of its gluconeogenic substrate Fbp1 (Chrustowicz et al., 2021; Sherpa et al., 2021) (Figure 5B). Fbp1 is a homotetramer with four Pro/N-degrons, each capable of binding Gid4. However, an individual Fbp1 degron peptide binds Gid4 weakly (Chrustowicz et al., 2021). This is overcome by supramolecular assembly of two core GID E3s—connected by "supramolecular assembly modules" comprising several additional Gid subunits. In the resultant hollow oval structure, two inward-facing Gid4 substrate receptors simultaneously capture Pro/N-degrons from two Fbp1 protomers. The higher-order assembly resembles an organometallic supramolecular chelate in that multiple contacts centrally encapsulate a smaller ligand and, hence, is termed "Chelator-GID." Supramolecular assembly avidly potentiates Fbp1 binding and also targets specific lysines in functionally important regions of this substrate (Sherpa et al., 2021). Interestingly, not all substrates require the supramolecular assembly; whether a monovalent or multivalent Chelator version of the GID E3 is required depends on a substrate protomer's individual Pro/N-degron affinity for Gid4, the oligomeric architecture, and the constellation of substrate lysines available for ubiquitylation (Chrustowicz et al., 2021). Although it is clear that supramolecular assembly regulates substrate degradation, how the formation of the assembly is regulated is an open question. Interestingly, the human CTLH E3 forms at least two types of higher-order assemblies through distinct supramolecular assembly modules, but how these conform to specific substrates remains unknown (Mohamed et al., 2021; Sherpa et al., 2021).

E3 regulation by N-degron pseudosubstrates

Studies of yeast Ubr1 showed how an E3 can be regulated by substrate receptors binding to N-terminal sequences that mimic degrons, working as pseudosubstrates that promote ubiquitylation of other substrates. Varshavsky and colleagues showed that Ubr1 is autoinhibited toward the non-N-degron substrate Cup9, a transcriptional repressor of the peptide transporter Ptr2 (Du et al., 2002; Turner et al., 2000; Xia et al., 2008a). Autoinhibition is mediated by Ubr1's C-terminal domain blocking the N-terminal domain from binding Cup9 (Du et al., 2002). However, autoinhibition is released, and Cup9 can bind when pairs of dipeptides—one initiating with a basic and the other with a hydrophobic residue—occupy the UBR-box-1 and UBR-box-2 N-degron-binding sites (Figure 5C). This circuit establishes feed-forward regulation of peptide uptake, where imported peptides stimulate expression of their own transporter through ubiquitin-dependent degradation of Cup9. Notably, amino acids also activate Ubr1 orthologs from higher eukaryotes to regulate lipid droplets, suggestive of intricate metabolic regulation similar to that in yeast (Zhang et al., 2022).

A different mechanism of pseudosubstrate regulation reprograms Ubr1 during proteotoxic stress (Szoradi et al., 2018) (Figure 5C). Here, Ubr1 substrate specificity is regulated through its UBR-box-1 binding to a C-terminal fragment of Roq1 harboring an N-terminal Arg. Generation of this fragment is tightly regulated and is mediated by the protease Ynm3. Binding of the Roq1 fragment accelerates proteolysis of substrates recruited to the UBR-box-2 domain and of stress-related misfolded ER membrane and cytosolic proteins, although how Ubr1 recognizes these latter substrates remains elusive.

Regulation of terminal degron pathways by modulation of substrate

Below, we summarize how substrates undergo specific changes that generate N- and C-degrons.

Degron creation by proteolysis

Proteolysis is a common mechanism creating N- and C-degrons recognized by many E3s (Figure 6A). Exoproteolysis is a means to create a new degron only at one terminus. Indeed, the Pro/N-degrons of two GID substrates, Pck1 and Aro10 whose native sequences initiate Met-Ser and Met-Ala, respectively, are exposed by specific exopeptidases whose activities are terminated at the Pro N-degron (Chen et al., 2021a).

Figure 6

Regulation generating terminal degrons on substrates

(A) Degron generation by proteolysis: exoproteases trim protein termini to generate a new degron at one protein end, while endoproteases cleave internally to potentially simultaneously generate both C- and N-degrons.

(B) N-terminal acetylation of Met (M), Ala (A), Ser (S), Thr (T), and Cys (C) generates Ac/N-degrons.

(C) Degron generation by N-terminal amino acid addition, either by direct coupling of Arg to an N-terminal Asp (D), Glu (E), or oxidized Cys (C∗) or by sequential deamidation of Asn (N) or Gln (Q) followed by arginylation.

(D) Degron generation by C-terminal amino acid addition during ribosomal stalling. Mistranslated polypeptides are targeted for degradation by either Listerin-dependent pathway or by CRL2KLHDC10 and Pirh2 E3 ligases recognizing C-terminal polyAla tails generated by NEMF.

(E) Premature termination of selenoprotein translation under selenium-deficient conditions leads to the formation of diGly/C-degron.

(F) Defective myristoylation exposes Gly/N-degrons.

(G) In eukaryotes, formyl-Met (fMet), produced by enzyme Fmt1, is incorporated at N termini of mitochondrial proteins. Fmt1 mitochondrial import is blocked under starvation conditions by phosphorylation by Gcn2, leading to its cytosolic accumulation. Cytoplasmic proteins that are misinitiated by fMet due to Fmt1 activity are eliminated by fMet/N-degron pathway.

Endoproteolytic cleavage can endow an N-terminal product with a C-degron and/or a C-terminal product with an N-degron. Caspases, separases, calpains, cathepsins, and mitochondrial proteases all have been found to generate C-terminal fragments with N-terminal basic or hydrophobic N-degron residues recognized by Ubr1 or its mammalian orthologs (Ditzel et al., 2003; Jin et al., 2010; Kong et al., 2021; Piatkov et al., 2012; Piatkov et al., 2014; Rao et al., 2001; Yamano and Youle, 2013). A classic example is the generation of the Arg/N-degron of cohesion subunit SCC1 by its separase-mediated cleavage during cell division (Liu et al., 2016; Rao et al., 2001). Similarly, Gly/N-degrons, which are strongly underrepresented at native protein N termini, are generated by endoproteolytic caspase cleavage events and play roles in apoptosis (Timms et al., 2019).

One example of a C-degron generated by proteolysis is the caspase-generated N-terminal fragment of the microtubule-associated protein Tau. The resultant C-terminal Asp is recognized by the quality control E3 CHIP (Ravalin et al., 2021; Ravalin et al., 2019b). Also, the stability of USP1, one of the most extensively studied human deubiquitylases, is regulated by self-cleavage. In response to UV-induced DNA damage, USP1 cleaves itself after an internal ubiquitin-like fold terminating in a di-Gly motif (Huang et al., 2006). The resultant N- and C-terminal fragments are targeted for proteasomal degradation by CRL2KLHDC2 and UBR1 E3s, respectively (Lin et al., 2018; Piatkov et al., 2012; Ravalin et al., 2021; Rusnac et al., 2018). Moreover, products of other deubiquitylases that generate N-terminal fragments terminating with diGly/C-degrons are also targeted by CRL2KLHDC2 (Yeh et al., 2021).

Regulation by N-terminal modification

More than 80% of human proteins are cotranslationally modified by N-terminal acetylation (Aksnes et al., 2016). Acetylation transforms the positively charged N-terminal amino group into a hydrophobic moiety, which can influence protein folding or subcellular localization and protein-protein interactions (Aksnes et al., 2016; Hwang et al., 2010; Lee et al., 2016; Monda et al., 2013; Scott et al., 2011). Despite the prevalence of N-terminal acetylation, it is thought that most acetylated N termini are shielded in the interior of protein structures or multiprotein complexes (Shemorry et al., 2013). However, N-terminal acetylation of Met, Ala, Ser, Thr, and Cys creates specific Ac/N-degrons that would be conditionally exposed if proteins are misfolded or not properly assembled into complexes (Hwang et al., 2010; Kim and Hwang, 2014; Shemorry et al., 2013) (Figure 6B). Thus, the Ac/N-degron pathway is thought to regulate protein quality control. E3s that are implicated as recognizing Ac/N-degrons include the ER membrane-embedded yeast Doa10, its human ortholog MARCH6, and the Not4 subunit of the CCR4-NOT complex (Hwang et al., 2010; Park et al., 2015; Shemorry et al., 2013).

Regulation by terminal amino acid addition

A remarkable mechanism creating N-degrons is the appendage of Arg, by the enzyme arginyltransferase 1 (ATE1), to a protein's N-terminal Asp, Glu, and oxidized Cys (Figure 6C). These residues are often referred to as secondary destabilizing residues due to their capacity to serve as substrates for N-terminal arginylation (Gonda et al., 1989; Tasaki et al., 2012). Notably, N-terminal Asn and Gln can be converted by N-terminal Asn or Gln deamidases to Asp and Glu, which then can be arginylated by ATE1. Interestingly, in bacteria, N-terminal conjugation of bulky hydrophobic residues generates N-degrons recognizable by ClpS (Graciet et al., 2006).

C-degrons are also generated by amino acid addition during ribosomal quality control (RQC) (Figure 6D). Aberrant polypeptide chains would be cytotoxic if left to accumulate, but this is prevented by an RQC complex that recognizes a particular form of stalled ribosomes and contains the E3 Listerin (Ltn1 in yeast) (Joazeiro, 2019). Truncated nascent polypeptides are marked with carboxy-terminal Ala and Thr (CAT) tails by the RQC subunit Rqc2, which facilitates Ltn1-mediated ubiquitylation (Joazeiro, 2019; Sitron and Brandman, 2020). A recent study showed that in mammalian cells, RQC is ensured by independent actions of multiple E3s: C-terminal Ala tails synthesized by the Rqc2 homolog, NEMF, not only assist Listerin but also serve as C-degrons recognized by CRL2KLHDC10 and Pirh2 E3s (Thrun et al., 2021). Interestingly, analogous mechanisms exist in bacteria, wherein stalled ribosomes tag proteins with Ala tails that elicit destruction by the ClpXP protease (Lytvynenko et al., 2019).

Regulation by improper protein translation

C-degron pathways contribute to quality control for selenoprotein biosynthesis. Seleno-Cys is incorporated by ribosomal decoding of UGA stop codons. Failure to incorporate seleno-Cys into selenoproteins, which is aggravated by selenium deficiency, leads to premature translation termination (Lin et al., 2015; Lin et al., 2018; Rusnac et al., 2018) (Figure 6E). Truncated polypeptides terminating with Gly/ or ArgXXGly/C-degrons bind CRL2 substrate receptors, such as KLHDC2/3 or APPBP2, which direct ubiquitin-mediated clearance. Interestingly, Gly is the least abundant C-terminal residue in properly translated eukaryotic proteins, but it is very commonly produced by translation errors, suggesting a general role of the Gly/C-degron pathway in ensuring quality control for the eukaryotic proteome (Koren et al., 2018; Lin et al., 2018; Timms and Koren, 2020; Yeh et al., 2021).

Regulation by terminal degron unmasking

An N-degron pathway performs quality control for a posttranslational modification. Many eukaryotic proteins initiating with Gly are modified with a 14-carbon fatty acid myristate. N-myristoylation regulates signal transduction, protein localization at membranes, and immune signaling (Udenwobele et al., 2017; Wang et al., 2021). However, when N-myristoylation fails, exposed Gly/N-degrons are recognized by the ZYG11B and ZER1 substrate receptors of CRL2-RBX1 CRL E3s (Timms et al., 2019) (Figure 6F).

Regulation of degron-generating enzymes

The initiator Met in bacteria bears a formyl group (fMet), which is normally removed during protein translation. However, aberrant N-terminal fMet retention serves as a degradation signal in the bacterial fMet/N-degron pathway (Piatkov et al., 2015). Interestingly, eukaryotic proteins synthesized in mitochondria, which are evolutionarily related to bacteria, also initiate with fMet, which is generated pretranslationally by the formyltransferase Fmt1 imported from the cytoplasm (Figure 6G). When yeast are grown under amino acid starvation conditions, or reach stationary phase, Fmt1 is phosphorylated by the Gcn2 kinase and sequestered in the cytoplasm. However, cytosolic Fmt1 generates a pool of initiator fMet-tRNA, which can be aberrantly incorporated at protein N termini. Aberrantly formylated proteins are cleared by the fMet/N-degron pathway, wherein the N-terminal fMet is recognized by the Psh1 E3 (Kim et al., 2018).

Future perspectives

A flurry of studies—from high-throughput screening to hard-core biochemistry and cryo-EM—have collectively revealed terminal degron pathways regulating much of protein homeostasis. Although many terminal degron pathways are associated with maintaining protein quality control, it is also clear that some terminal degron pathways regulate specific proteins in response to signals, such as changes in metabolic conditions or DNA damage. We envision that future studies will reveal many other E3s also recognizing terminal degrons, biological roles of terminal degron-mediated ubiquitylation, and many new regulatory mechanisms, for example, via degrons with posttranslational modifications of residues adjacent to termini.

Terminal degron pathways also offer a potential for therapeutic targeting. Successful development of small molecules inhibiting an E3 from binding to the acetylated N terminus of its partner protein provides proof-of-principle that pockets binding to protein N termini are ligandable (Scott et al., 2017; Zhou et al., 2017). There is great pharmaceutical interest in developing small molecules that occupy substrate-binding pockets in E3s because such molecules can be elaborated to generate degrader drugs that connect E3s and disease-causing proteins. Terminal degron-binding sites in substrate receptors have pockets the size of small molecules and are attractive candidates for degrader handles (Verma et al., 2020). Thus, we anticipate that the coming years will bring many new ways through which E3s targeting protein ends will signal "the end" in natural signaling cascades and those chemically rewired for degradation of disease-causing targets.