Magical Moments in Protease Biology: Proteasome Autocatalytic Activation and PI31-Mediated Inhibition

Abstract

Proteases regulate nearly every aspect of cellular function. Spatial and temporal control of protease activity contributes to this functional versatility and includes both zymogen activation mechanisms and the existence of dedicated protease inhibitors. Unlike conventional proteases, the proteasome harbors six individual proteases within a single macromolecular complex. This unique arrangement poses formidable challenges by requiring simultaneous activation or inhibition of all six active sites. Recent work from multiple labs has uncovered key aspects of assembly-coupled autocatalytic activation as well as the mechanism of proteasome inhibition by the endogenous inhibitor PI31. These elegant mechanisms highlight and expand the remarkable complexity and beauty of protease biology.

Protease Activation and Inhibition

Protease activity is subject to regulatory control through activating mechanisms and dedicated inhibitors. Zymogen (see Glossary) activation typically occurs through cleavage of inhibitory propeptides [1,2]. In many cases, this proceeds through straightforward cleavage of one protease by another. In other cases, activation proceeds spontaneously and autocatalytically through protease-specific mechanisms. Conversely, some mature proteases are subject to regulation by inhibitory proteins, many of which appear exquisitely specific and may have no other function within the cell [3]. The spatial and temporal control provided by these activating and inhibitory mechanisms contribute to the immense complexity and sophistication of protease biology.

Like other proteases, the proteasome is subject to regulated activation and inhibition. However, the proteasome is unique in harboring six individual protease subunits within a single macromolecular complex (Fig. 1A). Furthermore, the protease active sites are not exposed but rather sequestered within its central barrel-shaped core particle (CP), and these active sites are accessible only through gates at either end of the CP barrel which are too narrow to admit most folded proteins (Fig. 1B) [4–6].

Figure 1. The proteasome core particle is a six-protease molecular machine.

A) Structure of wild-type yeast CP (left panel). Coloring reflects individual subunits. The right panel shows a schematic representation of the CP interior with the six protease subunits indicated. Both panels were generated using PDB: 5CZ4. B) Comparison of closed and open gate CP conformations. The open gate is due to binding of the regulatory particle, and is a consequence of reorientation of the α-subunit N-termini. Left panel, PDB: 5CZ4; right panel, PDB: 6FVY.

The initial observations suggesting that proteasome activation was autocatalytic in nature occurred during the 1990’s [7,8]; similarly, the discovery of PI31 (proteasome inhibitor 31 kDa) as a dedicated proteasome inhibitor occurred in 1992 [9]. The proteasome’s unique status as a six-protease complex creates significant challenges for both activation and inhibition, and the detailed mechanisms for both remained poorly understood for several decades. Here we will review recent advances in understanding how these remarkable regulatory pathways occur. These advances reflect the efforts of multiple laboratories on three continents, working in different model systems, and utilizing different experimental approaches and techniques.

A Multi-Protease Molecular Machine

The proteasome is capable of destroying a broad range of substrates and yet is highly selective in substrate recognition. These two properties are reflected in its overall architecture, which consists of a central CP bound at one or both ends by a second subcomplex known as the regulatory particle (RP). Substrate selectivity is generated through the proteasome’s preferential degradation of ubiquitinated proteins, and this is accounted for by the RP which recognizes and removes the ubiquitin signal [10,11]. Given the fact that the CP gate is narrow and constitutively closed, the RP also opens the gate, unfolds the substrate, and injects it into the CP interior.

Whereas the RP governs selectivity, the CP accounts for the breadth of the proteasome’s substrate repertoire. The CP is a 700 kDa 28-subunit cylindrical structure composed of four stacked heptameric rings (Fig. 1A) [12]. The outer α-rings harbor the narrow gates that allow substrate access to the CP interior (Fig. 1B), while the inner β-rings harbor the proteasome active sites. Unlike conventional proteases which are typically monofunctional enzymes, the proteasome harbors three different protease subunits that have different substrate specificities: β5 cleaves after hydrophobic residues, β2 after basic residues, and β1 after acidic residues. Because there are two β-rings, the proteasome harbors two copies of each protease subunit, resulting in a total of six individual proteases.

Autocatalytic Activation of the Proteasome

Given the immense degradative capacity of the proteasome, its protease subunits must remain inactive during proteasome biogenesis until they have been safely sequestered within the CP interior. This is achieved by the provision of N-terminal propeptides for each of the active site subunits. Propeptides also exist for two non-catalytic subunits (β6 and β7; [12]) although their functions during proteasome assembly remain mostly unknown. The capacity of the active site propeptides to ensure proteasome inactivity is a consequence of the unique active site chemistry of the proteasome, which is a member of the threonine N-terminal nucleophile family of proteases [13,14]. As the name indicates, these enzymes require cleavage of their propeptides to expose the active site threonine at the enzyme’s amino-terminus. However, sequestration of the active sites within the CP interior precludes access of an external protease that could cleave the propeptides, explaining why proteasome activation must be autocatalytic. The critical event that links proteasome assembly and autocatalytic activation is the joining of two half-CP (α₇β₇) precursors to create the full four-ring CP barrel (“CP fusion”; Fig. 2A). Indeed, this was first recognized in the 1990’s [7,8], but the detailed mechanisms that could explain assembly-coupled autocatalytic activation of six individual proteases remained poorly understood. Recent advances in this area have unfolded in four stages, as described below.

Figure 2. Structural basis for proteasome autocatalytic activation.

A) Schematic of CP assembly. Individual CP subunits are assembled into half-CP particles with the assistance of chaperones Pba1-4 and Ump1. Fusion of two half-CP intermediates creates the preholoproteasome which then undergoes autocatalytic activation to yield mature CP. In this process, Pba1/2 are released while Ump1 is degraded. The fate of the cleaved propeptides (shown in gray) is unknown, but they are likely degraded as well. B) Complementary β-subunit midline segments are unresolved (i.e. flexible) prior to CP fusion, resulting in misalignment of the catalytic centers which maintains the proteasome in an “off” state until the moment of half-CP fusion. In both panels, the unresolved regions have been displayed on the mature CP structures (yeast, PDB: 5CZ4; human, PDB: 8QYO). A similar arrangement is seen for the β2 and β1 protease subunits [18,21]. B) A critical arginine-aspartate midline salt bridge stabilizes the catalytic center residues through an intricate hydrogen bonding network (dashed blue lines). The arginine residue stabilizes both the canonical Lys/Asp triad residues, and the proposed “second triad” Ser/Asp residues. A similar arrangement was seen for the β2/β6 pair [18]. C) The proposed catalytic pentad with both triads being arranged in a mirror-image conformation around the active site threonine. Both sets of residues are hydrogen bonded to the active site threonine. (Panels B-C: PDB: 5CZ4.)

Misalignment of the catalytic triads prevents premature activation

A high-resolution crystal structure of mature yeast CP was obtained in 1997 [12], but there were no structures of CP assembly intermediates, a fact that greatly limited our understanding of proteasome biogenesis. An initial advance occurred in yeast in 2015 through the recognition that defined CP mutants might be enriched for assembly intermediates [15]. In this case, a mutant of β4 (S142F; also known as pre1-1) combined with overexpression of the Pba1/2 chaperones allowed for acquisition of both pre- and post-fusion assembly intermediates. However, analysis was limited by the low resolution (~20 Å) of these structures [15]. The first high-resolution cryo-electron microscopy (cryo-EM) structures of CP assembly intermediates were reported six years later using a mutant of β1 (G34D; also known as pre3-1) [16,17]. These structures included the 13S assembly intermediate (3.6 Å) which contained a complete α-ring, partial β-ring (β2–4), and three of the five CP assembly chaperones (Pba1/2 and Ump1). A second assembly intermediate, known as pre-15S, contained two additional β-subunits (β5–6; 3.2 Å).

Most cryo-EM structures contain unresolved regions, which typically reflect flexible protein segments that are lost upon computational averaging of individual particles. Although not appreciated at the time of initial publication [16], it was later recognized that all β-subunits present in the pre-15S complex showed multiple unresolved regions, mainly clustered at the site of the future β/β-ring interface (Fig. 2B) [18]. Furthermore, these unresolved regions were complementary between opposing subunits at the CP midline. For example, the two unresolved regions in the active site subunit β5 precisely matched the three unresolved regions in its cross-midline partner β3 (Fig. 2B). A similar arrangement was seen for the active site subunit β2 and its cross-midline partner β6. β1 was not present in the pre-15S structure, so it was unclear if β1 was similarly regulated [18].

Like other proteases, the proteasome utilizes a catalytic triad mechanism whereby two additional residues (Lys and Asp) contribute to proteolysis via protonation/deprotonation of the active site threonine’s side chain hydroxyl [14]. Critically, for both β5 and β2, one of these unresolved loops appeared to directly control the triad residues, suggesting that the catalytic triads are not aligned prior to half-CP fusion and thus providing a straightforward mechanism to explain how the active site subunits remain “off” until the appropriate moment in CP assembly [18]. Even more tellingly, β2’s triad Lys residue was rotated ~180 degrees away from the active site threonine, further supporting the notion that misalignment of the catalytic triads prevents premature proteasome activation until half-CP fusion [18]. Separately from this work, a renewed analysis of the pre1-1 mutant combined with Pba1/2 overexpression allowed for a new high-resolution cryo-EM structure of a post-fusion assembly intermediate referred to as “late-PC” that independently supported this model [19]. Here, the β4-S142F mutation prevented proper β5/β3 alignment, resulting in unresolved midline segments and preventing β5 propeptide cleavage. Finally, it should be noted that an earlier study of a bacterial proteasome had proposed that β/β-ring midline contacts are critical for the overall arrangement of the catalytic triads [20]. However, prokaryotic proteasomes differ from their eukaryotic counterparts since they harbor only a single α-subunit and a single β-subunit; furthermore, the critical advance indicating that the catalytic triads were misaligned prior to half-CP fusion owing to flexibility in these midline segments was not yet apparent since structures of pre-fusion assembly intermediates had not yet been obtained.

Concordant results were obtained independently for the human proteasome and reported in parallel with the yeast findings. Here, the experimental approach was different: all 14 CP subunits were expressed simultaneously in insect cells along with the 5 CP chaperones (Pba1-4 and Ump1) [21]. An affinity tag on β2 allowed for the isolation of a variety of assembly intermediates. As in yeast, the different β-subunits from pre-fusion structures showed similar unresolved segments that were complementary at the future CP midline (Fig. 2B), further supporting the notion that autocatalytic activation proceeds through an unstructured-to-structured transition that occurs at the exact moment of CP fusion. This work also extended the model to the β1 subunit and illuminated several other proteasome assembly intermediates for the first time [21]. Importantly, two other studies independently reached similar conclusions for endogenous human CP. In both cases, proteasome intermediates were enriched by affinity purification of one of the CP chaperones, Pba2 (known as PAC2 or PSMG2 in humans) [22,23], a clever approach which also allowed for other important insights into CP biogenesis, including aspects of α/β-ring configuration and specific functions of the chaperones.

Enforced alignment of the catalytic triads upon half-CP fusion drives autocatalytic activation

A key feature of the mechanism promoting fusion-induced triad alignment is the presence of an aspartate-arginine salt bridge that spans the CP midline between opposing subunits (β2/β6 and β5/β3) [18] (Fig. 2C). The highly conserved arginine residue in the β2 and β5 protease subunits is hydrogen bonded to the lysine and aspartate members of their respective catalytic triads (Fig. 2C). Thus, establishment of these salt bridges should rigidify the catalytic triads and, once stably aligned, autocatalytic activation should proceed through spontaneous chemistry.

A prediction of this structural model is that disruption of the salt bridge should uncouple the normally coincident processes of half-CP fusion and autocatalytic activation. This prediction was experimentally confirmed for both β2 and β5 and in the case of β5 allowed for determination of a high-resolution structure of the preholoproteasome, a post-fusion but pre-activation intermediate that retains assembly chaperones (Pba1/2, Ump1) and active site propeptides [18]. While these salt bridge interactions appear critical, it is likely that still other midline contacts contribute to fusion-coupled active site remodeling, as indicated by mutational analyses in the bacterial and human systems [20,23]. A corresponding arginine residue is also present in β1 and is similarly hydrogen bonded to its triad residues. However, there is no corresponding midline salt bridge, suggesting that other mechanisms may help to stabilize the triad, including the long C-terminal tail of β7 which reaches across the CP midline [24].

Autocatalytic activation requires a so-called catalytic pentad

A curious aspect of this emerging model was the observation that only some of the unstructured β-subunit segments appeared to control the catalytic triad residues. In particular, there was a second unstructured loop on the opposite side of the active site threonines in β5 and β2, respectively [25] (Fig. 2B). Of potential interest to this structural arrangement was a prior proposal by Michael Groll and colleagues that the proteasome might use a unique catalytic mechanism [26]. Owing to its status as an N-terminal nucleophile, its active site threonines actually have two nucleophiles: the side chain hydroxyl and the free amino-terminus. While the canonical triad controls threonine’s side chain hydroxyl group, the authors proposed that a “second triad” of residues might be necessary to regulate the protonation state of the free aminoterminus [26]. Indeed, the very aspartate and serine residues they identified are controlled by this second flexible loop in an analogous manner to the canonical triad residues (Fig. 2C–D) [25]. Furthermore, the critical midline salt bridge arginine residue was also hydrogen bonded to this second loop containing these “second triad” residues (Fig. 2C), creating a striking mirror image arrangement of the canonical Lys/Asp triad residues and the proposed Ser/Asp triad residues (Fig. 2D).

Mutation of these “second triad” residues again stabilized the preholoproteasome, reflecting an uncoupling of half-CP fusion and autocatalytic activation for each of the three protease subunits [25]. These mutants were also specifically defective in the proteolytic activity of their associated protease subunit, indicating that these residues also function during the normal enzymatic cycle [25]. The overall constellation of residues (Lys/Asp-Thr-Ser/Asp) has been referred to as a catalytic pentad.

Failure of the catalytic pentad mechanism causes some human proteasomopathies

An emerging family of human diseases is caused by inherited mutations in the proteasome [27–29]. These have been documented for both CP chaperones (Ump1/POMP and Pba2/PSMG2) and CP subunits, including the variant active site subunits (PSMB8/β5i, PSMB10/β2i, PSMB9/β1i) utilized by the immunoproteasome which is specifically adapted for the generation of peptides for MHC-I presentation [30]. These diseases are characterized by varying degrees of immune dysregulation and developmental defects. How individual mutations compromise proteasome function, and how proteasome dysfunction leads to pathologic states represent important questions.

Recent studies implicate the catalytic pentad model for two separate human mutations: β5i-G201V and β2i-G201R [25,31]. Both studies modeled these mutations in the more tractable yeast system. The corresponding yeast β5-Gly205 residue is the immediate neighbor of the serine residue from the second catalytic triad; the corresponding yeast β2-Gly191 residue is three residues away from the second triad aspartate and a G191R mutation is predicted to sterically interfere with the second triad [25].

While the β5-G205V mutation was lethal in yeast, a more conservative β5-G205A mutation caused specific accumulation of preholoproteasome, consistent with a defect in autocatalytic activation, as well as a defect in substrate proteolysis by β5, but not β2 or β1 [25]. Meanwhile, the lethality of the β5-G205V mutant could be suppressed by a secondary mutation in the β5-propeptide, and proteasomes from this strain similarly showed a major defect in autocatalytic activation [31]. The β2-G191R mutation was also lethal in yeast, but a G191A mutation was viable and caused a specific defect in substrate proteolysis by β2 [25]; an activation defect was not seen although the alanine substitution was considerably more conservative than the original arginine substitution. Still another patient-derived mutation, β2i-D56H [32], involves the canonical catalytic triad aspartate, suggesting that disruption of either triad can result in human disease.

Multisite Proteasome Inhibition By PI31

While this emerging mechanism of autocatalytic activation addresses the proteasome’s unique status as a six-protease complex, a similar challenge applies to endogenous inhibitors of the proteasome which, in order to achieve complete inhibition, must simultaneously target six active sites. PI31 (also known as PSMF1 or Fub1) was discovered by the Demartino laboratory in 1992 through a biochemical screen to identify regulators of CP activity [9]. There are also FDA-approved small molecule inhibitors of the proteasome (e.g. bortezomib, carfilzomib) that have been widely used in multiple myeloma, an aggressive cancer of plasma cells [33,34]. But unlike these drugs which preferentially target one (β5) of the three types of proteasome active site [35], PI31 was found to simultaneously inhibit all three active site types [36–38]. This was a surprising observation and one whose molecular basis remained unknown for 30 years.

A mechanistically simple approach would have been for PI31 to obstruct the CP gate, either by capping it or plugging it, thereby preventing all substrate access. However, it turns out that the relatively unstructured C-terminal domain of PI31 actually enters into the CP interior where it simultaneously interacts with all six active site pockets (Fig. 3A–B) [39]. A similar arrangement has now been seen for PI31 in three different organisms: yeast [39], a microsporidian parasite (Vairimorpha necatrix) [40], and a human(PI31)/bovine (CP) hybrid [41] (Fig. 3A). The sheer molecular gymnastics required for this function are worth contemplating: not only must PI31 traverse the narrow CP gate in a largely unfolded state, it must then simultaneously inhibit all six active sites since any cleavage of PI31 would likely alter its conformation, leading to additional proteolysis and cascading failure. Extensive biochemical validation of the structural models has been provided in the yeast and human systems. The key prediction of this model was that specific mutations in one part of PI31 should abrogate inhibition at one type of active site without affecting the other two types of active site, and this is indeed the case [39,41].

Figure 3. PI31 enters the CP interior and simultaneously inhibits all six protease subunits.

A) High-resolution structures of PI31 bound to the proteasome for yeast, microsporidian, and human. In each case, the cryo-EM density is shown with the modeled PI31 homodimer within. B) The PI31 dimer winds around the CP interior through an elaborate arrangement that allows PI31 to bind and fill all of the active site pockets. Each active site residue is shown in red. For clarity, the second copy of each active site subunit has been omitted. Images generated using PDB/EMDB entries: 7TEO/25848 (yeast), 8ADN/15365 (microsporidian), and 8FZ6/29604 (human).

An additional challenge facing PI31 is that because it sits in close proximity to the catalytic active sites, it must evade its own degradation to function as a durable inhibitor. PI31 uses different mechanisms at each site. Substrate residues on either side of a proteolytic cleavage site can be referred to as P3-P2-P1-P1’-P2’-P3’, where the scissile bond is between P1 and P1’. PI31 exists as a homodimer within the CP interior and yeast PI31 dimerizes in front of the β5 active site such that there is actually no polypeptide bond in between PI31’s P1 and P1’ residues, preventing PI31 cleavage (Fig. 4A) [39]. The human approach is conceptually similar but operationally distinct: here the P1 (Met208) and P1’ (Asp197) residues are within the same polypeptide chain, but are part of a loop that tracks away from the active site nucleophile, again resulting in a protein conformation that lacks a scissile bond (Fig. 4A) [41].

Figure 4. Detailed aspects of PI31-mediated proteasome inhibition.

A) Yeast and human PI31 utilize conceptually similar but operationally distinct mechanisms to evade degradation at the β5 active site. Yeast PI31’s P1 and P1’ residues derive from different polypeptides, ensuring that there is no scissile bond in front of the active site. Human PI31 achieves the same outcome through a looping mechanism that positions the P1 and P1’ residues in front of the active site, but again with no scissile bond between them. B) Inhibition of human β2 requires PI31’s ARFD motif, where Arg242 is multifocally bound at an acidic pocket located at the β2/β3 interface, with additional hydrogen bonds (blue dashed lines) to both subunits. C) A potent and specific β2 inhibitor was developed based on PI31’s inhibitory mechanism [42]. Schematic and structural panels generated using PDB entries 7TEO (for yeast) and 8FZ6 (for human).

At the β2 site, a scissile bond is present in front of the active site threonine but PI31’s conformation ensures that the scissile bond is far from the active site (5.7 Å in yeast; 6 Å in humans), which would tend to thwart proteolysis (Fig. 4B) [39,41]. A similar approach is employed at β1 by yeast PI31 [39]. However, the human sequences which mediate β1 inhibition remain unknown and reflect another important evolutionary difference between the proteins.

PI31-Inspired Chemical Proteasome Inhibitors

These fundamental advances in understanding PI31 have important clinical translational implications inasmuch as PI31’s structure provides a blueprint to design new proteasome inhibitors based on its evolutionarily optimized mechanisms. Currently available drugs preferentially target β5: bortezomib, for example, shows a half-maximal inhibitory concentration (IC₅₀) of 8.2 nM at β5 versus 140 nM at β1 and 1500 nM at β2 [35]. There are currently no approved inhibitors that preferentially target β2 or β1.

To try to develop a β2-specific inhibitor, the P4-P1 residues that target the human β2 active site (Ala-Arg-Phe-Asp; Fig. 4B) were combined with a C-terminal boronic acid (the same electrophile used in bortezomib) along with substitution of the P1 residue to leucine due to chemical incompatibility of the aspartate [42]. This molecule, ARFL-Boro, proved to be a highly potent and specific β2 inhibitor: its IC50 at β2 was 45 nM, versus 3,980 nM at β5 and 5,700 nM at β1 (Fig. 4C) [42]. Arginine at the P3 position is likely a major contributor to β2 specificity since it is situated at the β2/β3 interface where it is cradled within a highly acidic pocket and also hydrogen bonded to both CP subunits (Fig. 4B) [41–44].

ARFL-Boro was highly effective in killing multiple myeloma cells in culture (IC₅₀=181 nM), indicating that the molecule shows good cell permeability and establishing β2 as a therapeutic target in myeloma [42]. Early genetic findings indicated that inactivation of two proteasome active sites was not additive, but highly synergistic [45,46]. This synergy could be recapitulated chemically by combining ARFL-Boro and bortezomib [42]. Indeed, there is evidence that therapeutic efficacy with existing proteasome inhibitors occurs primarily at doses where there is also inhibition of a second active site (β1 for bortezomib; β2 for carfilzomib) [47]. Thus, potent and specific β2 inhibitors might be useful as single agents or as synergistic β2/β5 combination therapies.

PI31 and Neurodegenerative Disease

The recent structural data begins to explain how PI31 inhibits the proteasome, but why PI31 exists in the first place remains largely mysterious. PI31 is widely expressed across tissue types [48] although its abundance appears to be significantly lower than that of CP subunits [38,39]. Two lines of evidence suggest that PI31 may have a role in neurodegenerative disease. First, a neuronal knock-out of PI31 resulted in a neurodegenerative-like phenotype in mice [49]. More recently, PI31 has been genetically linked to both Alzheimer’s disease and Parkinson’s disease [50,51]. Interestingly, many of the human mutations mapped to PI31’s C-terminal proteasome-inhibitory domain rather than its less conserved and poorly understood N-terminal domain. There was a mutational hot-spot at Arg242 which corresponds to the P3 residue from the β2-inhibitory domain (Fig. 4B). R242H was identified in the Alzheimer’s disease study [50] while R242H, R242C, and R242G were identified in the Parkinson’s study [51]. The R242H mutant was experimentally shown to be defective at inhibiting β2 in vitro [42], as anticipated from the structural analysis [39,41]. The R242C and R242G mutants have not been experimentally validated but are expected to be more severe mutants. Two other mutations within PI31’s C-terminus have been identified (L228V, R231*; [51]), but their functional significance is unknown.

The patient-derived C-terminal variants suggest that PI31’s proteasome-inhibitory function is critical in these diseases. However, it is worth noting that alternate models view PI31 as a positive regulator of proteasome function by either promoting RP-CP assembly or proteasome transport within cells [52–54]. In these models, PI31’s inhibitory activity is viewed as an in vitro phenomenon and not its primary function in cells. While the human/bovine structure was obtained through in vitro reconstitution, it should be noted that the yeast and microsporidian structures represent endogenous PI31 purified from cells [39,40], suggesting that CP inhibition is PI31’s physiologic role. Nevertheless, it is possible that CP inhibition could somehow serve an overall positive function in proteasome regulation. Still another model proposes that PI31 functions to specifically promote assembly of the immunoproteasome, although binding of PI31 to immature CP, which would be expected of a chaperone, was not seen [55]. This recently recognized role of PI31 in neurodegenerative disease adds a new urgency to understanding its precise cellular function and rationale.

Concluding Remarks

Given the impressive complexity that exists within the ubiquitin system and within the proteasome’s ubiquitin recognition and processing capacities, it can be easy to forget that the proteasome is still a protease and subject to many of the same regulatory mechanisms that govern other proteases, including activating and inhibitory mechanisms (see Outstanding Questions). However, the proteasome’s unique status as a six-protease molecular machine has necessitated sophisticated biochemical mechanisms that allow for simultaneous multi-protease regulation. The advances discussed here regarding autocatalytic activation and PI31-mediated inhibition build on decades of foundational investigation from both the proteasome and broader protease fields, uncovering elegant and conceptually innovative mechanisms that attest to, and even broaden, the remarkably complex and fascinating field of protease biology.

Outstanding Questions.

• In addition to the standard proteasome, there are cell-type specific proteasome variants including the immunoproteasome, thymoproteasome, and spermatoproteasome. Do these specialized proteasome types utilize similar mechanisms of autocatalytic activation?

• The specific cellular function provide by PI31 remains poorly understood. Why does PI31 exist and what is its specific role in cells? How does this function relate to its newly recognized role in neurodegenerative disease?

• While proteasome inhibition is mediated by its C-terminal domain, PI31’s N-terminal domain remains mysterious and was poorly resolved in the cryo-EM structures. What is the function of this domain?

• Other endogenous proteasome inhibitors exist, including catalytic core regulators (CCRS) [56], which merit further study.

Highlights.

• Like other proteases, the proteasome is subject to regulation via activating and inhibitory mechanisms.

• However, unlike conventional proteases, the proteasome harbors six individual protease subunits within a single macromolecular complex, which has necessitated novel and elegant solutions for simultaneous six-protease activation or inhibition.

• Proteasome autocatalytic activation proceeds through a switch-like mechanism that simultaneously aligns each protease subunit’s catalytic center residues. Defects in this process are the cause of some forms of a family of inherited diseases known as proteasomopathies.

• PI31’s recently determined cryo-EM structure revealed an unprecedented inhibitory mechanism.

• PI31’s clinical relevance has emerged through its links to neurodegenerative disease and through recent efforts to develop a new class of proteasome inhibitors based on PI31’s structure.

Glossary

Catalytic Pentad

A proposed cluster of five residues that regulate the proteasome active sites. In addition to the threonine nucleophile, two of the residues regulate threonine’s side chain hydroxyl group while the other two residues are proposed to regulate threonine’s free aminoterminus.

Catalytic Triad

A cluster of three residues required for catalysis by proteases and other enzymes. The first residue is the active site nucleophile whose protonation state is regulated by the other two residues. For example, cysteine, histidine, and aspartate constitute the catalytic triad in many cysteine proteases.

Core Particle (CP)

The barrel-shaped subcomplex of the proteasome that harbors its six protease subunits.

Immunoproteasome

A specialized form of the proteasome that harbors different proteolytically active subunits (β5i, β2i, and β1i) and which is specifically adapted to the generation of peptides which are presented by MHC (major histocompatibility complex) class I molecules.

N-terminal nucleophile enzymes

A class of proteases and transferases whose defining feature is the exposure of their active site nucleophile residue as the first residue of the protein. This is typically achieved through post-translational cleavage of propeptide sequences.

Preholoproteasome

A proteasome CP assembly intermediate which has undergone the fusion of two half-CP precursors, but which has not yet undergone autocatalytic activation.

Propeptide

A protein sequence which is present upon synthesis of the protein, but which is later cleaved to generate the mature form of the protein.

Proteasome

The multisubunit molecular machine responsible for most non-lysosomal protein degradation. The proteasome preferentially degrades proteins that have been previously modified by the small protein ubiquitin. The canonical proteasome consists of a complex containing both CP and RP, although other proteasome variants can be generated through binding of CP to other regulatory complexes, including PA200/Blm10 and PA28.

Regulatory Particle (RP)

The multisubunit subcomplex of the proteasome that binds the CP and that endows the proteasome with the capacity to specifically recognize and process ubiquitinated proteins.

Spermatoproteasome

A testis-specific form of the proteasome that harbors a unique α4 subunit.

Thymoproteasome

A thymus-specific form of the proteasome that harbors a unique β5 subunit (β5t) along with two immunoproteasome-type subunits (β2i and β1i).

Zymogen

An inactive form of an enzyme. For proteases, activation typically occurs through the cleavage of propeptides. Such cleavage may be performed by another protease or may occur spontaneously through autocatalytic mechanisms, as in the proteasome.