The benefits of local depletion: The centrosome as a scaffold for ubiquitin-proteasome-mediated degradation

ABSTRACT

The centrosome is the major microtubule-organizing center in animal cells but is dispensable for proper microtubule spindle formation in many biological contexts and is thus thought to fulfill additional functions. Recent observations suggest that the centrosome acts as a scaffold for proteasomal degradation in the cell to regulate a variety of biological processes including cell fate acquisition, cell cycle control, stress response, and cell morphogenesis. Here, we review the body of studies indicating a role for the centrosome in promoting proteasomal degradation of ubiquitin-proteasome substrates and explore the functional relevance of this system in different biological contexts. We discuss a potential role for the centrosome in coordinating local degradation of proteasomal substrates, allowing cells to achieve stringent spatiotemporal control over various signaling processes.

Introduction

Biological processes rely on a vast array of signaling and information-processing networks operating at the cellular and molecular level. Though a multitude of signaling strategies have been identified and well-characterized over the past half-century, it remains unclear how they unfold in space and time Centrosomes are well-known for their role in promoting timely separation of mitotic chromosomes, but are also emerging as potential hubs for cell signaling components, particularly those that rely on the ubiquitin-proteasome system (UPS).^1,2 Consistent with as yet unidentified roles in signaling and proteostasis, we recently found that a C. elegans transcriptional co-activator, the β-catenin SYS-1, is negatively regulated in a cell-cycle dependent manner by localizing to centrosomes during mitosis.³ Here, we review additional studies that suggest an intriguing role for the centrosome as a physical scaffold for concentrating protein degradation and associated regulatory events during cell signaling. The centrosome is the major microtubule-organizing center (MTOC) in animal cells. It is composed of 2 centrioles –bilaterally symmetric cylindrical appendages made of short microtubule triplets - surrounded by a pericentriolar material (PCM) – a proteinaceous matrix composed of microtubule-anchoring γ-tubulin ring complexes (γ-TuRCs), filamentous scaffolds, and various mitotic regulators.^1,4 Centrosomes are highly dynamic non-membrane-bound organelles whose composition, activity, and behavior can be ultimately controlled by a number of different upstream intracellular cues such as cell cycle status and cell stress.^5,6 Given the centrosome's long-standing, canonical role as a microtubule-organizing center, proteomic analysis of purified centrosomes identify a surprisingly broad list of centrosomal proteins whose functional characterization resist – with even the most generous speculation - placement into any category even tangentially related to microtubule behavior.^7,8 These data, if assumed biologically relevant, suggest centrosomes may confer additional functional roles in biology.

Centrosome activity and behavior changes with the cell cycle

The composition and activity of the centrosome is dynamic and ultimately controlled by cell cycle regulatory factors. A newly divided cell inherits one centrosome. During G1 to S-phase transition, new daughter centrioles are formed orthogonally to each mother centriole to produce 2 centrosomes, each with its own PCM and hybrid mother/daughter centriole pair. Throughout interphase, the centrosomes remain tethered via proteinaceous linkers.⁹ Cell cycle regulators that trigger mitotic entry promote cleavage of these linkers, allowing microtubule-based motors to drive the 2 centrosomes to opposite poles of the cell.^10,11 Mitotic entry also results in a dynamic elaboration of centrosome structure during a process called centrosome maturation in which a physical expansion of the PCM and associated factors triggers an increase in microtubule-nucleating activity.^12,13 Surrounding the centrioles, the interphase PCM is supported by a structured series of concentrically-arranged toroidal rings rich in coiled-coil domain scaffold proteins such as Pericentrin, Cep192, and CDK5RAP2.¹⁴ While the substructure of the mitotic PCM is less clear, it has been demonstrated that its maturation is cued by phosphorylation of Pericentrin by the mitotic kinase Plk1 to allow further recruitment of Cep192, γ-tubulin, and other core components.¹² Properly separated mature centrosomes organize the bipolar mitotic spindle to ensure proper chromosome segregation and successful cell division.² In post-mitotic cells, the centrosome can be targeted to the cell cortex, and components of the mother centriole may dock at the plasma membrane to form the primary cilium.¹⁵ As centrosome activity dynamically changes according to the cell cycle status of the cell, it seems plausible that proteasomes or substrates associated with the centrosome may also retain this property, a possibility we address in this manuscript.

The ubiquitin-proteasome system in eukaryotes

The ubiquitin-proteasome system is a major protein degradation pathway in eukaryotic cells and is largely responsible for clearance of undesired protein molecules. For a protein fated for degradation, passage through the ubiquitin-proteasome pathway begins with covalent attachment of the small protein tag ubiquitin. A monomer of ubiquitin enters into the UPS pathway upon forming an ATP-dependent isopeptide bond with an E1 activating enzyme before being covalently attached to an E2 conjugating enzyme. An E3 ligase complex then catalyzes the transfer of ubiquitin from the E2 enzyme to a lysine residue on the target substrate. In this manner, additional ubiquitin monomers can be linked to the substrate's original conjugate to from polyubiquitin chains.¹⁶ Polyubiquitinated proteins are subject to proteolysis by the 26S proteasome, a large 2.5Mda macromolecular assembly.¹⁷ The 26S proteasome is composed of 2 complexes: (1) the 20S catalytic subunit, a cylindrical complex with an internal core equipped with various forms of peptidase activity and (2) the 19S regulatory subunit, a ringed amalgamation of scaffolds, ubiquitin-receptors, and ATPases that cooperate to recognize, unfold, and ultimately translocate substrates into the 20S core.¹⁸

Early reports of centrosome-associated degradation

Preliminary evidence of centrosome-associated degradation emerged when indirect immunofluorescence experiments using antibodies against proteasomal antigens stained spindle poles in dividing cells and detergent-insoluble perinuclear structures in interphase cells.¹⁹ This pattern was further characterized when Wigley et al first showed in 1999 that the 20S and 19S proteasomal complexes both displayed punctate enrichment at γ-tubulin-labeled centrosomes in HEK293 and HeLa cells.²⁰ The fact that a number of well-characterized proteasomal substrates shared this localization pattern suggested that degradation of substrates by active proteasomes is concentrated at the centrosomes.^21-23 Consistent with this idea, Fabunmi et al, 2000 showed that purified centrosomes from HeLa cells were able to degrade ubiquitinated substrates and proteasome-specific peptides in a manner that was ATP-dependent yet sensitive to proteasome inhibitors. Purified centrosomes also retained the ability to cleave single ubiquitins from a polyubiquitin chain linked to a substrate, a general feature of 19S proteasome subunits.²⁴ Thus, centrosome fractions display the same enzymatic profile as intact, active proteasomes. Intriguingly, many of these early studies showed that inhibition of proteasome activity led to redistribution of cytosolic proteasomes to the centrosome, suggesting a mechanism to concentrate degradation specifically at the centrosome when proteasome substrates were in excess.^{20,24, 25}

Concurrent with these key discoveries were reports demonstrating that large centrosome-associated protein aggregates could form in response to either proteasome inhibition or overexpression of misfolded proteins. Overexpression of inefficiently-folded alleles of cystic fibrosis transmembrane conductance regulator (CFTR) or presenilin-1 in HEK293 cells leads to CFTR accumulation in large inclusion bodies at the centrosome, an effect exacerbated upon proteasomal inhibition.²² These structures, dubbed 'aggresomes', are composed of electron-dense protein aggregates that depend on minus-end-directed microtubule trafficking for accumulation at centrosomes.²⁶ Aggresome formation depends on the histone deacetylase HDAC6 which simultaneously binds polyubiquitinated unfolded proteins and dynein motors to participate in active trafficking of aggregate particles to the centrosome.²⁷ Aggresomes are rich in ubiquitin, 19S and 20S proteasome subunits, and the chaperones Hsp70 and Hsp90.²⁸ Additionally, aggresome formation triggers redistribution of vimentin intermediate filaments from cytoplasmic networks to a cage-like structure surrounding the aggresome core.²² Thus, when the capacity of the proteasome is exceeded, as in cases where toxic proteins are overexpressed,²⁹ a dramatic cellular response involving cytoskeletal reorganization and trafficking events ensures that misfolded/damaged proteins are sequestered along with ubiquitin-proteasome and other quality-control components specifically at the centrosome.

Aggregated proteins that bind to the 19S regulatory complex but resist entry into the 20S core cannot be subject to proteolysis and may inhibit the proteasome.³⁰ Therefore, aggresomes are largely thought to be cleared by autophagy rather than proteasome-mediated degradation.³¹ Since proteasomes are unable to process aggregates, the fact that they are actively targeted to aggresomes presents a paradox. It was recently demonstrated that proteasomes play a role in aggresome clearance not by catalyzing substrate degradation but by removing ubiquitin chains from polyubiquitinated protein aggregates.³² Freed from their aggregated substrates, accumulation of unconjugated polyubiquitin chains triggers activation of the autophagy pathway. Thus, centrosomal proteasomes may act as substrate sensors, using their non-degradative enzymatic activities to influence the ubiquitin profile of the cell, provoking an alternative degradation strategy. Since many neurodegenerative disorders involve aggregation of cytotoxic proteins, it is not surprising that aggresomes and aggresome-like inclusions appear in several disease states such as Parkinson,²⁷ Alzheimer,³³ Huntington's,³⁴ and multiple myeloma.³⁵ Future studies elucidating the explicit role of the centrosome in coordinating cellular response to unfolded protein stress should provide important insights into the pathophysiology of diseases vulnerable to defects in protein degradation.

Less clear is the capacity of centrosome-associated proteasomes to degrade substrates in the absence of proteotoxic stress, although several reviews suggest such a mechanism may be operational during basal physiological conditions.^28,36,37 Indeed, isolated examples of centrosomally-degraded substrates involved in a host of biological processes continue to emerge, which we discuss below (Individual examples covered in this review are shown in Fig. 1).

Figure 1.

Components of the ubiquitin-proteasome system and substrates localize to the centrosome (composed of the marked PCM and centrioles) in a conserved manner. Representative examples. of ubiquitin-proteasome components (shown above the dashed line) and substrates (shown below the dashed line) are color-coded according to the organism in which centrosomal localization was observed. Citations for individual reports are provided in brackets.

Centrosome function depends on the proteasome

Virtually all aspects of centrosome behavior and organization are under stringent control by the proteasome (Fig. 1). In HeLa cells, inhibition of the proteasome leads to overaccumulation of a number of different centrosomal proteins such as γ-tubulin, ninein, pericentrin, dynactin and PCM1.³⁸ As a consequence of misregulation of core centrosomal factors, cells lacking proteasome activity are severely deficient in microtubule growth and organization, harboring ectopic sites of microtubule nucleation and shrunken microtubule asters around centrosomes. Transmission electron micrographs of wild type cells exhibit a characteristic anchoring of microtubules at the PCM; however, discernable sites of microtubule anchorage within the PCM are lost with proteasome inhibition, suggesting that the proteasome limits the dosage of critical centrosomal factors necessary for normal MTOC function.³⁸

The cell-cycle regulated centrosomal kinase Nek2 is responsible for regulating centrosome separation at G2/M transition. Nek2 exhibits a dynamic pattern of recruitment to the centrosome with upwards of 70% of centrosome-bound Nek2 subject to rapid turnover in fluorescence recovery after photobleaching experiments. Interestingly, dynamic recruitment of new Nek2 molecules depends on its degradation since inhibition of the proteasome or observing a non-degradable Nek2 allele both lead to an attenuated pattern of recovery.³⁹ This experiment strongly suggests that the centrosomal pool of Nek2 is subject to active degradation by the proteasome. Since Nek2 activity at the centrosome is primarily exerted at G2/M transition, it is possible that degradation of excess Nek2 at centrosomes at other cell cycle stages limits this activity to prevent untimely centrosome splitting.

More extensively characterized is the highly conserved role of proteasomal degradation in limiting centriole duplication to a single event at G1/S transition in eukaryotic cells. Centriole duplication is initiated by kinase cascades under the regime of cell cycle regulators.⁴⁰ Polo-like-kinase 4 (Plk4) is a serine/threonine kinase that promotes centriole duplication during S-phase.^40,41 Plk4 is rapidly turned over and its expression is tightly regulated by a Skip/Cullin/F-box (SCF)-class E3 ubiquitin ligase and the 26S proteasome.^42-44 The substrate adaptor component of SCF ligase in flies, Slimb, binds to a Plk4 phosodegron to target the protein for degradation. Slimb localizes to centrosomes throughout the cell cycle and restricts levels of centrosomal Plk4 at S-phase. This pattern of precise spatiotemporal regulation of Plk4 dosage prevents any residual Plk4 activity from triggering an undesired reduplication of centrioles.^43,44 Given that Plk4 is overexpressed in over a quarter of human tumors, it is rapidly emerging as a potential cancer target, and suppression of Plk4 reduces cell proliferation in a number of breast cancer cell lines.⁴⁵ A more comprehensive understanding of centrosome-specific degradation pathways may allow for novel therapeutic strategies that modulate degradation of substrates specifically at the centrosome without disrupting the bulk of total cellular ubiquitin-proteasomal degradation.

Consistent with regulation of Plk4 by centrosome-bound Slimb in flies, localization of core SCF components to centrosomes has also been reported in multiple human cell lines, revealing the possibility that centrosomal targeting of this essential ubiquitin ligase complex is a conserved and functionally relevant feature of cycling cells.⁴⁶ Given the aforementioned reports of active, centrosome-bound 26S proteasomes, it seems likely that the centrosome harbors its own “in-house” degradation system equipped to limit its own protein content, especially in the case of proteins that may perturb centrosome functionality when in excess. An “in-house” centrosomal degradation system would presumably allow cells to specify an optimal centrosome composition without having to fine tune overall cellular levels of every individual centrosomal protein to fulfill a stoichiometric demand. Since centrosomal factors may be multifunctional, locally-acting mechanisms restrict their accumulation specifically at centrosomes without affecting their abundance in other subcellular locales. This phenomenon has been demonstrated in principle by observations of the centrosomal kinase monopolar spindle 1 (Mps1). Mps1, a target of the proteasome, localizes to centrosomes, nuclear pores and to kinetochores, where, in the latter, it mediates the spindle assembly checkpoint during mitosis.^47-49 During G1/S transition, centrosomal Cyclin A-Cdk2 stabilizes Mps1 to promote centriole duplication. Overexpression of Cyclin A or inhibition of the proteasome result in over-accumulation of Mps1 specifically at centrosomes. Conversely, depletion of Cyclin A results in loss of centrosomal Mps1 while having very little effect on the nuclear pore-associated pool or whole cell levels.⁴⁸ Thus, centrosome-associated proteasomes may allow cells to confine degradation to the centrosomal subpopulation of a multifunctional protein, leaving other subcellular pools and associated processes intact. It is also worth noting that centrosomal Mps1 is degraded by an ubiquitin-independent proteasomal degradation mechanism mediated by Antizyme, a highly conserved protein that localizes to the centrosome, binds non-ubiquitinated substrates and targets them to the proteasome for degradation.^50,51 Loss of Antizyme-mediated Mps1 degradation results in centrosome amplification and contributes to aberrant MTOC activity and subsequent aneuploidy in human melanomas.⁵² Antizyme regulation of centrosome duplication via Mps1 degradation suggests diverse modes of proteasome-mediated degradation, including ubiquitin-independent proteolysis, are utilized by centrosome-associated proteasomes.⁵³

Centrosome-association of rapidly-degraded cell cycle regulatory factors

Progression through mitosis requires thorough spatiotemporal regulation of mitotic cyclins by the ubiquitin-proteasome system. A well-characterized prerequisite for metaphase to anaphase transition (MAT) is the ubiquitination of cyclin B by the multi-subunit E3 ligase anaphase promoting complex/cyclosome (APC/C) and degradation by the proteasome.^54,55 Cyclin B has been reported to localize to mitotic centrosomes in multiple systems, raising the possibility that centrosome-associated proteasomal degradation may play a role in regulating cyclin B levels during MAT.^56,57 This notion is well-supported by several experiments. One preliminary hint emerged from elegant cell fusion experiments that generated cells containing 2 mitotic spindles. Upon fusion, spindles at different mitotic stages initiated anaphase asynchronously, invoking the possibility that each spindle, despite sharing a common cytoplasm, exerts local, autonomous control over cyclin B degradation to cue its own mitotic exit.⁵⁸ This idea was reinforced by subsequent in vivo studies in flies and human cells, revealing that cyclin B follows a dynamic spatiotemporal pattern of regulation during mitosis. Cyclin B::GFP is concentrated at the centrosomes and spindle until late metaphase. Prior to anaphase entry, cyclin B degradation progresses in a stepwise pattern, disappearing along the spindle in a pattern that begins at the centrosomes and ends at the metaphase plate.^23,59 In this way, cyclin B initially disappears from centrosomes and is gradually restricted to the spindle equator (Fig. 2A, top panel). Consistent with degradation of cyclin B at centrosomes is the observation that key components of APC/C also localize to the centrosomes and spindle throughout mitosis. A non-degradable allele of cyclin B obtained by mutating its APC/C binding sites persists on centrosomes throughout mitosis, suggesting cyclin B removal from metaphase centrosomes involves its polyubiquitination and subsequent degradation.⁶⁰

Figure 2.

Centrosome-associated degradation pathways. A) Centrosome and spindle-associated degradation of Cyclin B. Top panel: At metaphase, degradation of cyclin B begins at centrosomes in a step-wise pattern that propagates to the mitotic spindle in a direction toward the spindle equator and is effectively cleared prior to anaphase entry. Bottom panel: Centrosome-bound cyclin B is polyubiquitinated (small red circles represent ubiquitin tags) by the E3 ubiquitin ligase APC/C and the E2 ubiquitin conjugating enzyme Vihar, marking it for degradation by the proteasome. Vihar is polyubiquitinated and marked for degradation by APC/C, resulting in a negative feedback loop on APC/C ubiquitination of cyclin B. (B) Centrosome-associated degradation of SYS-1 during C. elegans asymmetric cell division. Top panel: SYS-1 localizes to centrosomes, where it is subject to dynamic turnover via the proteasome during division and is asymmetrically distributed between daughter cell nuclei after division. Middle panel: When SYS-1 centrosomal localization is disrupted by rsa-2 (RNAi), its degradation is attenuated during division, resulting in elevated SYS-1 levels in both daughter cell nuclei. Bottom panel: When the proteasome is inhibited, SYS-1 accumulation at centrosomes is increased and its turnover is decreased. In the absence of the proteasome, SYS-1 protein avoids negative regulation during asymmetric division and is severely overaccumulated in both daughter cells. (C) Centrosomal degradation pathways control dendrite growth in neurons. CaMKIIβ¹⁰⁴ binds centrosomes by interaction with PCM1 and phosphorylates the APC/C component Cdc20, causing it to dissociate from centrosomes. Unphosphorylated Cdc20 localizes to centrosomes and polyubiquitinates the helix-loop-helix protein Id1 targeting it for degradation by the centrosomal proteasome. In the absence of negative regulation by Cdc20 and the proteasome, Id1 retracts dendrites. Id1 negatively regulates proteasomes by preventing association of S5a with the 19S complex, potentially autoregulating its own degradation.

APC/C-mediated cyclin B degradation requires Vihar, an E2 ubiquitin-conjugating enzyme that also localizes to centrosomes during metaphase before disappearing as mitosis progresses (Fig. 2A, bottom panel). Loss of Vihar results in increased levels of cyclin B and severe overaccumulation at mitotic centrosomes. Overexpression of Vihar, on the other hand, leads to loss of centrosomal cyclin B. These two lines of evidence indicate that this E2 ligase is responsible for concentrating initial cyclin B degradation at centrosomes.⁶¹ Vihar, itself, is targeted for degradation via polyubiquitination by APC/C and disappears from centrosomes upon entry into anaphase. APC/C, therefore, autoregulates its ability to polyubiquitinate centrosomal cyclin B by promoting degradation of its cofactor Vihar. It is thought that this feedback mechanism terminates the first ‘wave’ of cyclin B degradation at the centrosomes before degradation of cyclin B progresses to the spindle, presumably mediated by other E2 ligases.²³ That cyclin B and several of its negative regulators transiently associate with the centrosome immediately preceding initiation of cyclin B degradation strongly suggests that both ubiquitination and degradation are actively concentrated at centrosomes. It is tempting to speculate that localization of cyclin B and its regulatory factors to the mitotic apparatus brings them in close physical proximity to spindle components to allow cells to rapidly pause mitosis in response to spindle defects by quickly inhibiting cyclin B degradation. In fact, the centrosome may play a role in coordinating degradation of spindle-associated cyclin B, a possibility that was raised by observations of an intriguing mutant isolated in Drosophila called centrosome fall off (cfo). Centrosomes in cfo mutants become physically detached from the mitotic spindle. In cfo mutants, initial cyclin B degradation at centrosomes proceeds as usual but fails to propagate to the mitotic spindle.⁶² These data confirm that there are 2 discreet ‘waves’ of cyclin B degradation, and that progression from the first wave to the second requires physical contact between the spindle and centrosomes.

Another function of this centrosome-associated degradation pathway may be to ensure timely and efficient clearance of cyclin B. Indeed, thorough clearance of cyclin B is absolutely required for irreversible commitment to MAT.⁶³ To ensure all chromosomes are tightly bound to kinetochore microtubules before allowing chromatid separation, cyclin B degradation can be inhibited by a spindle assembly checkpoint (SAC) pathway that monitors kinetochore-microtubule attachment and tension.⁶⁴ This pathway must be disabled upon entry into anaphase, so as not to mistake loosely-secured kinetochores in metaphase chromosomes for those in actively-separating anaphase chromatids, in which tension has been released due to chromatid separation.⁶⁵ Desensitizing the cell to this surveillance mechanism is accomplished, at least in part, through complete removal of cyclin B by late anaphase. This was shown by a recent study that relied on small cyclin B deletions to experimentally reduce its degradation efficiency. The results showed that even a modest amount of residual cyclin B in anaphase can trigger SAC reactivation, in turn, stabilizing any leftover cyclin B and shutting down mitotic progression.⁶⁶ Clearly, mitotic cells must guarantee thorough and timely cyclin B clearance to irreversibly shut down SAC, a task that may be ensured by actively concentrating cytoplasmic cyclin B to the centrosomes just prior to the initiation of degradation. Probing the mechanisms by which cyclin B is targeted to the centrosome would allow further studies to test the functional relevance of cyclin B centrosomal enrichment on both its rate of degradation and its dynamic spatiotemporal regulation during mitosis.

Centrosome-associated degradation of cell fate determinants

Cell differentiation during development relies on extensive transcriptional reprogramming, in which the activity of downstream transcription factors is controlled by networks of cell signaling pathways. Under default conditions, terminal effectors of many cell signaling pathways experience both high production and degradation rates when at steady state, such that even slight modulation of either production or degradation can trigger rapid increase or decrease in intracellular concentration, respectively. Key transcriptional regulators of cell fate exhibit half-life values on the scale of minutes to a few hours. These include p53⁶⁷ (<20 minutes), β-catenin⁶⁸ (∼1hr), and the Hippo pathway transcription factor Yap (1hr-2.5hr).^69,70 This pattern of constitutive production and degradation may represent a strategy to optimize signaling pathway response time by allowing for maximum responsiveness at the time of signal acquisition, but not long periods after signaling events have ceased.⁷¹ Timely and efficient degradation of signaling effectors is critical for cell fate specification, and the regulation of cell fate-determining factors by centrosomal degradation pathways has been reported in several studies.

Smad1 is a transcription factor that mediates TGFβ signaling events utilized during key vertebrate cell fate specification processes such as dorsal-ventral axis determination by BMP. Phosphorylation by the kinase GSK3 targets Smad1 for polyubiquitination and degradation by the proteasome. In Cos7 cells, phosphospecific antibodies against GSK3-phosphorylated Smad1 show enriched staining at centrosomes, a pattern dramatically enhanced by inhibition of the proteasome. This visualization of the degradation-fated pool of Smad1 indicates that its degradation is rate-limiting and likely concentrated at centrosomes. The latter notion is further confirmed by localization of the proteasome subunit S20α5 and polyubiquitin to the centrosomes in this cell line.⁷² Subsequent experiments revealed that phospho-Smad1 localized asymmetrically to mitotic centrosomes in human embryonic stem cells (hESCs). Phospho-Smad1 also colocalized with centrosomal microtubules. Interestingly, depolymerization of microtubules by nocodazole treatment led to increased phospho-Smad1 levels, a preliminary indication that proper degradation of pSmad1 requires microtubule-dependent transport to the centrosome. GSK3-phosphorylated β-catenin (another substrate marked for degradation⁷³) and polyubiquitinated proteins showed the same localization pattern during hESC division, suggesting that association with mitotic centrosomes may be a general feature of various proteasome-fated substrates.⁷⁴

We recently showed that C. elegans embryos use such a strategy to limit levels of SYS-1/β-catenin during asymmetric cell division.³ SYS-1/β-catenin is a transcription factor acting downstream of the Wnt signaling pathway to activate target genes necessary for cell fate specification.^75-82 During asymmetric division, SYS-1 is targeted to mitotic centrosomes by its interaction with the centrosomal protein RSA-2.⁸³ Disrupting SYS-1 centrosomal localization by depletion of RSA-2 leads to increased SYS-1 retention by daughter cells after division, indicating the presence of a negative regulatory mechanism involving centrosomal targeting. Photobleaching experiments show that the entire centrosomal SYS-1 population turns over on the scale of few minutes, a common feature of centrosomal proteins.^39,84 Interestingly, inhibition of the proteasome not only causes a severe increase in SYS-1 specifically at centrosomes but also attenuates its turnover rate, suggesting that centrosomal SYS-1 is actively degraded (Fig. 2B). This centrosome-associated degradation mechanism may represent a strategy to efficiently degrade SYS-1 during division, thereby limiting its retention by daughter cells.³ Thus, regulation of SYS-1 may be coupled to the cell cycle by localization to mitotic centrosomes and subsequent degradation.

Can centrosome-associated degradation broadly link cellular processes such as the cell cycle and cell signaling? We have found that, in addition to enriched centrosomal polyubiquitin, Rpt4/S10b, an ATPase subunit of the proteasome lid, also localizes to mitotic centrosomes during C. elegans embryogenesis, indicating conservation of proteasome targeting to centrosomes from humans to nematodes (unpublished data). This temporally and spatially restricted Rpt4 pattern suggests that a local increase of proteasome function at the mitotic centrosome may directly link the cell cycle to enhanced substrate degradation through PCM accumulation, and that additional targets of this mechanism await discovery. Like vertebrate β-catenin,⁷³ SYS-1 protein is likely subject to high turnover rate and is tightly regulated by Wnt signaling to allow precise control over cellular concentration. Recent in vivo observations demonstrate that a range of relative concentrations of various Wnt signaling components including SYS-1 and its cofactor TCF can produce a continuum of signaling strengths throughout C. elegans embryogenesis.⁸⁵ The explicit role of centrosome-associated degradation in modulating the dosage of SYS-1 and other clients in various developmental contexts remains a challenge for future studies.

Since the size of the centrosome increases dramatically during PCM maturation, it is possible that centrosome-associated degradation mechanisms can be used as a strategy to synchronize protein degradation with the cell division. A recent mass spectrometry-based characterization of S-phase and M-phase centrosomes showed centrosomal accumulation of polyubiquitinated proteins specifically during mitosis. Depletion of the endogenous proteasome inhibitor Ecm29⁸⁶ led to a significant reduction in polyubiquitinated proteins at the centrosome, supporting the view that these substrates may be actively degraded by centrosomal proteasomes.⁸⁷ While proteasomal activity is roughly the same between S-phase and M-phase centrosomes, centrosomal targeting of polyubiquitinated proteins increases during maturation, demonstrating that other substrates may share this mode of division-coupled regulation. Together, these studies suggest that centrosome-associated degradation strategies control various signaling factors during mitotic divisions, linking pathway activity to the cell cycle.

Local degradation by centrosomal proteasomes controls neuronal morphology

Neurons are morphologically elaborate post-mitotic cells that can dynamically control the behavior of cellular appendages to establish and reorganize cell-cell contacts. This remarkable property is thought to be achieved by confining basic biological processes like translation, vesicular trafficking, and protein degradation to local subcellular domains.^88,89 Several studies have revealed a role for centrosome-associated degradation in neuronal dendrite elaboration. In rat cerebellar granule neurons, the APC/C subunit Cdc20 promotes dendrite growth and arborization by marking the centrosomal helix-loop-helix protein Id1 for degradation.⁹⁰ Cdc20, itself, binds centrosomes through a 19-amino acid centrosomal localization domain, a region that is necessary for its function in dendrite arborization. Importantly, Cdc20 lacking its native centrosomal localization domain can be driven to centrosomes upon fusion with the conserved centrosome-targeting domain PACT, which rescues its function. These experiments elegantly demonstrate that Cdc20s functional contribution to promoting proper dendrite elaboration depends explicitly on its local activity at the centrosome.^90,91 More recently, it was shown that the proteasomal subunit S5a/Rpn10, an ubiquitin receptor that binds substrates to the 19S proteasome complex, stimulates local degradation by centrosomal proteasomes in order to promote dendrite growth. Id1 negatively regulates centrosomal proteasomes by preventing the association of S5a to the 19S complex, demonstrating that centrosome-associated degradation mechanisms in cerebellar neurons can be subject to negative regulation to modulate dendrite extension/retraction.⁹² That Id1, itself, is both a target and a regulator of the centrosomal proteasome suggests that dendrite growth may be controlled by an autoregulatory feedback loop that modulates local proteasome activity. These studies compellingly demonstrate that local proteasomal degradation imposed on as-yet-unidentified substrates contributes to growth dynamics of certain subcellular appendages (Fig. 2C). These studies also raise implications about the potential of centrosome-associated degradation in controlling subcellular organization in other contexts if additional, well-established observations of centrosome positioning are considered. Given that centrosomes have been shown to act as polarity cues in multiple systems^93,94 and tend to assume stereotyped subcellular positions in a number of differentiated cell types (epithelial cells, migrating fibroblasts),⁹⁵ it could be imagined that centrosome-associated degradation may be used as a strategy to locally deplete polarity-determining factors from certain subcellular domains in order to contribute to intracellular organization.

Centrosome-associated degradation in cells of the immune system

In addition to neurons, there are reports of centrosome-associated degradation mechanisms in various cells of the immune system. Centrosomes are known to participate in molecular mechanisms underlying immune system function, most notably in the formation of cytotoxic immunological synapses. Activation of cytotoxic T lymphocytes upon binding of an antigen presenting cell is followed by centrosome migration and physical docking at the site of cell-cell contact.⁹⁶ This dynamic polarization event is required for the formation of the immunological synapse (IS), an integrin-rich cell-cell adhesion that allows the T cell to secrete lytic granules exclusively to its target. Thus, docking of the centrosome to the contact site allows not only specification of the IS site but also active, microtubule-dependent trafficking of secretory materials.⁹⁴

It was recently shown that leukocytes from patients with elevated body temperature harbor damaged centrosomes with decreased γ-tubulin levels.⁹⁷ Cells in culture also display this feature after heat stress.^97,98 Centrosomes in these cells are deficient in microtubule nucleation, cilia assembly, and fail to retain proper levels of core components such as γ-tubulin, pericentrin, PCM1, and centrin. Heat-stress induced centrosome inactivation is dependent on degradation of centrosomal proteins by the centrosomal proteasome since a centrosomally-targeted dominant negative Rpn10 allele rescues centrosomes upon expression in heat-stressed cells. This mechanism of centrosome inactivation is likely to be physiologically relevant to the immune system because both centrosome docking and immunological synapse formation are compromised Jurkat cells after heat-stress.⁹⁷

Earlier studies using in vitro models of immune response suggest that centrosomes may act as sites of antigen processing. Proteins from invading pathogens are processed by modified proteasomes into short peptides to be displayed on the cell surface for presentation to cytotoxic T cells via MHC class I molecules.⁹⁹ During immune response, signaling via the cytokine IFNγ induces expression of PA28, an alternative regulatory complex that associates with a modified 20S proteasome core to form the immunoproteasome.¹⁰⁰ Interestingly, components of the PA28 complex have previously been localized to the centrosome.²⁰ A few examples of centrosome-associated degradation have been elucidated in the context of antigen processing. The influenza virus nucleoprotein, from which antigens are derived following infection, is a highly unstable, rapidly-degraded substrate that localizes to centrosomes upon proteasome inhibition in human osteosarcoma cells. Interestingly, restoring proteasome activity while inhibiting protein synthesis in these cells results in the rapid clearance of centrosome-associated nucleoprotein, strongly suggesting that nucleoprotein proteolysis is concentrated at centrosomes.¹⁰¹

Expression of an unstable form of the HIV protein Nef in antigen-presenting cells results in rapid degradation and evokes a cytotoxic T lymphocyte response. This form of Nef localizes to centrosomes in a microtubule-dependent manner in Cos7 cells. Depolymerizing microtubules by nocodazole disrupts Nef centrosomal localization and also leads to decreased antigen presentation of an epitope tagged form of Nef, suggesting that microtubule-based trafficking of Nef to the centrosome plays a role in its antigen processing.¹⁰²

The most compelling evidence of centrosome-associated antigen processing comes from a study using the human papilloma virus protein E7, a model tumor antigen. When E7 is driven to centrosomes by fusion with γ-tubulin, dendritic cells display increased MHC class I antigen presentation. Furthermore, vaccination of mice with γ-tubulin::E7 or E7 alone demonstrated that fusion to γ tubulin enhances cytotoxic T-cell activation during immune response. Using an E7-expressing tumor model, mice vaccinated with the γ-tubulin::E7 fusion display significantly less pulmonary tumor metastases than mice vaccinated with E7 alone. Together, these studies reveal the centrosome as an essential site for proteasome-dependent antigen processing events using a variety of cell types and models and raise therapeutic implications.¹⁰³

Concluding remarks

That centrosomal proteasomes regulate the localization and expression of clients involved in a broad range of processes such as centrosome organization, cell cycle progression, cell fate assignment, subcellular morphology, and immune response suggests that centrosome-associated degradation may represent a fundamental biological mechanism key to cell function. The majority of tumor cells display centrosome abnormalities with respect to either copy number or morphology and therefore may misregulate protein stability in addition to protein inheritance after cell division. Thus, it would be of importance to gain an understanding of the functional relevance of centrosomes in regulating the activity of the ubiquitin-proteasome system. In particular, future studies should test whether presence or absence of functional centrosomes influences the activity of the ubiquitin-proteasome system and its clients. Further exploration into this phenomenon will allow movement beyond characterization of the participants and interactions involved in signaling networks toward fostering an understanding of how individual signaling events are coordinated in space and time.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This research was supported by grants from the American Cancer Society (RSG-11-140-01-DC), the Roy J. Carver Charitable Trust (13-4131), and the National Science Foundation (IOS-1456941).