Activity and Regulation of the Centrosome-associated Proteasome*

Regulated proteolysis is important for maintaining appropriate cellular levels of many proteins. The bulk of intracellular protein degradation is catalyzed by the proteasome. Recently, the centrosome was identified as a novel site for concentration of the proteasome and associated regulatory proteins (Wigley, W. C., Fabunmi, R. P., Lee, M. G., Marino, C. R., Muallem, S., DeMartino, G. N., and Thomas, P. J. (1999) J. Cell Biol. 145, 481–490). Here we provide evidence that centrosomes contain the active 26 S proteasome that degrades ubiquitinated-protein and proteasome-specific peptide substrates. Moreover, the centrosomes contain an ubiquitin isopeptidase activity. The proteolytic activity is ATP-dependent and is inhibited by proteasome inhibitors. Notably, treatment of cells with inhibitors of proteasome activity promotes redistribution of the proteasome and associated regulatory proteins to the centrosome independent of an intact microtubule system. These data provide biochemical evidence for active proteasomal complexes at the centrosome, highlighting a novel function for this organizing structure.

Intracellular protein degradation regulates a large and diverse number of cellular processes (1). Most protein degradation within mammalian cells is catalyzed by the proteasome (2, 3), a 700,000-dalton protease, composed of 28 subunits that form a hollow cylinder. The eukaryotic proteasome is a multicatalytic protease that has six active sites within the lumen (4,5). The proteasome alone exhibits very low catalytic activity because this geometry sequesters the catalytic sites from potential substrates. Proteasome activity, however, is greatly increased by either of two regulatory proteins, PA700 (or 19 S regulatory cap) (6, 7) and PA28 (or 11 S regulator) (8,9). These proteins bind to one or both ends of the proteasome where they induce conformational changes in the proteasome, which allow access to the catalytic sites (5,6,10). PA700 is a 700,000-dalton multisubunit complex that mediates the proteasome's ATP-dependent degradation of ubiquitinated proteins. In contrast to PA700, PA28 is a simpler ring-shaped molecule that enhances the hydrolysis of only short peptides in an ATP-independent manner (8,9). PA28 probably plays a role in the generation of antigenic peptides for presentation by major histocompatibility complex class I molecules by both increasing their rate of production and altering the sites of proteasome dependent cleavage (11,12).
A fundamental issue in understanding proteasome-mediated proteolysis is the spatial relationship within the cell between the protease and its substrates. Previous studies have demonstrated that the proteasome is located in both the cytoplasm and the nucleus, as well as near the cytoplasmic face of the endoplasmic reticulum (ER) 1 (13)(14)(15) where it could degrade mutant proteins extracted from the ER or process antigenic peptides for transport to the ER. In addition to these compartments, the centrosome is a site of concentration of the 20 S proteasome, as well as PA700, PA28, and ubiquitin (16).
The centrosome is a small perinuclear structure containing a pair of centrioles surrounded by an amorphous pericentriolar matrix (PCM). The protein composition of the PCM is poorly defined but is known to contain two well characterized proteins, ␥-tubulin (17,18) and pericentrin (19), which together constitute an irregular lattice (20). The centrosome nucleates microtubule assembly and establishes their polarity with respect to fast growing (ϩ) and slow growing (Ϫ) ends (21). This nucleation activity is mediated by ␥-tubulin in the PCM (22,23). The centrosome also plays a pivotal role in the reorganization of microtubules during cell division. At interphase the centrosome is duplicated, and the resulting pair of centrosomes form the two poles of the mitotic spindle, an essential structure for proper separation of duplicated chromosomes between daughter cells (24).
In addition to the proteasome, a diverse group of proteasome substrates has been localized to the centrosome, including the tumor suppressor p53 (25), cyclins (26), presenilin 1 (27), cystic fibrosis transmembrane conductance regulator (CFTR) (16,27), and IB (28). Co-localization of the proteasome with multiple physiological substrates raises the possibility that the centrosome may be a novel site for regulated proteasome function. The goal of this work was to characterize the functional status of the proteasome at the centrosome.

EXPERIMENTAL PROCEDURES
Materials-Monoclonal anti-␥-tubulin antibodies were purchased from Sigma. Polyclonal anti-PA28 (29), anti-PA700 (16), and anti-20 S proteasome (30) were generated and characterized as described previously. Fluorescently labeled secondary antibodies were from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA). Lactacystin, MG132, and ubiquitin aldehyde (Ubal) were purchased from Calbiochem. Proteasome-specific fluorogenic substrate Suc-LLVY-7-amido-4methylcoumarin was from Bachem (Torrance, CA). Bovine 20 S proteasome (31), PA700 (32), and PA28 (8) were isolated as described previously. 26 S proteasome was assembled in vitro as described (32) * This work was supported by research Grants DK46181 (to G. N. D.) and DK49835 (to P. J. T.) from the NIDDK, National Institutes of Health, the Muscular Dystrophy Association (to G. N. D.), and Grant 9740033N from the American Heart Association (to P. J. T.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ from 20 and 80 nM purified bovine 20 S proteasome and PA700, respectively (6). Polyubiquitinated dihydrofolate reductase (Ub 5 -DHFR) containing a C-terminal hemagluttinin tag (a generous gift from Dr. Cecile M. Pickart, Johns Hopkins University, Baltimore, MD) was prepared in vitro as described. 2 Cell Culture and Preparation of Centrosomes-HEK 293 or HeLa cells (American Type Culture Collection, Rockville, MD) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 50 g/ml streptomycin, and 50 units/ml penicillin. Centrosomes were purified as described (16). Briefly, cells were exposed to 1 g/ml of cytochalasin D and 0.2 M nocodazole for 45 min to disrupt the actin cytoskeleton and microtubules, respectively. All subsequent steps were carried out at 4°C. Cells were washed sequentially in Tris-buffered saline and 0.1 ϫ Tris-buffered saline/8% sucrose and then lysed in buffer containing 1 mM Hepes, pH 7.2, 0.5% Nonidet P-40, 0.5 mM MgCl 2 , 0.1% ␤-mercaptoethanol and Complete Inhibitor Cocktail TM (Roche Molecular Biochemicals). The lysate was spun at 1500 ϫ g for 5 min to pellet unlysed nuclei and cellular debris. The resulting supernatant was incubated with 1 g/ml DNase 1 and 10 mM Hepes buffer (pH 7.2). The mixture was underlaid with 60% sucrose solution and spun at 10,000 ϫ g for 1 h. The resulting interface layer containing the sedimented centrosomes as well as the sucrose cushion was collected. At this stage, partially purified centrosomes were either directly pelleted at 25,000 ϫ g for 30 min or further purified on a discontinuous (70, 50, and 40%) sucrose gradient. The gradients were spun at 120,000 ϫ g for 1 h, and fractions were collected.

SDS-PAGE and Western
Blotting-Centrosomes were pelleted from sucrose gradient fractions at 25,000 ϫ g for 30 min and resuspended in Buffer H (20 mM Tris-HCl, pH 7.6, at 4°C, 20 mM NaCl, 1 mM EDTA, 5 mM ␤-mercaptoethanol, and 10% glycerol). Protein concentrations of cellular extracts and centrosomes were measured using the Bradford reagent (Bio-Rad) and subjected to SDS-PAGE. To quantify the amount of 20 S proteasome associated with the centrosome, purified centrosomes and known amounts of purified 20 S proteasome standards were subjected to Western blot analysis (16) and visualized by ECL (Amersham Pharmacia Biotech). A standard curve was obtained from densitometric analysis of the resulting film using Molecular Analyst software (Bio-Rad). Amounts of 20 S proteasome in the purified centrosomes and in the whole cell lysate were estimated from the standard curve.
Assays of Proteasome Activities-For studies with the synthetic peptide substrate, Suc-LLVY-7-amido-4-methylcoumarin, proteasome activities were assayed in samples of purified centrosomes and fluorescent measurements were conducted using a FL600 Microplate fluorescence reader (Biotek). For inhibitor studies of the centrosomeassociated proteolytic activity, MG132 (50 M) and lactacystin (50 or 10 M) were included in the assay mixtures 10 min prior to measurements. For studies with ubiquitinated substrate, degradation assays were carried out in buffer containing 50 mM Tris-HCl (pH 7.6), 10 mM MgCl 2 , 2 mM ATP, 10% glycerol, 5 mM dithiothreitol, 2 M Ubal, 80 nM Ub 5 -DHFR, and centrosomes. Relative concentrations of proteasome in the centrosome preparation were calculated as described above. As a control, assembled 26 S proteasome was incubated in reaction buffer with Ub 5 -DHFR substrate in the presence and absence of proteasome inhibitors. Reactions were subjected to SDS-PAGE and Western blot analysis using a monoclonal anti-hemagluttinin antibody (BabCo, Richmond, CA).

RESULTS
We previously utilized immunocytochemistry to demonstrate that the 20 S proteasome is present at the centrosome of cultured mammalian cells (16). To determine whether the centrosome-associated proteasome is catalytically active, purified centrosomes were assessed for proteasome activity using a proteasome-specific substrate. Fig. 1A shows the results of this analysis at the final step in centrosome purification after sucrose density gradient centrifugation and the subsequent pelleting of the centrosomes in each fraction. Gradient fractions contained proteasome activity with a distribution profile coincident with that of proteasome protein detected by Western blotting (Fig. 1B). The protease activity was inhibited by lactacystin, a specific inhibitor of the proteasome (3), and by MG132, a potent but less specific proteasome inhibitor (2), at concentrations that inhibit the purified proteasome (Fig. 1B). These data demonstrate that this proteolytic activity in the centrosome preparations is accounted for by the proteasome.
We performed several control experiments to verify that the presence of catalytically active proteasome in the centrosome preparations can not be accounted for by cytoplasmic contamination. Purified 20 and 26 S proteasomes were subjected to the same sucrose gradient centrifugation as the centrosome preparation. In contrast to the centrosome-associated proteasome, these complexes sediment only to the lighter sucrose density gradient fractions (data not shown), which lack detectable ␥-tubulin immunoreactivity (Fig. 1A). Furthermore, purified 20 and 26 S proteasomes do not pellet at 25,000 ϫ g, the final centrosome purification step. Thus, the proteasomal activity assayed herein must be associated with a dense particle, i.e. the centrosome. Finally, all fractions were devoid of detectable levels of marker proteins for other cellular compartments including BiP (ER), aldolase (cytosol), lamin B 1 (nucleus), and ␤-cop (Golgi) (Ref. 16 and data not shown). Thus the fact that the distribution profile of the proteasome was not identical to the distribution profile of ␥-tubulin, a centrosomal marker ( Fig.  1), may be because of heterogeneity in the size or composition of centrosomes. Alternatively, the centrosome may be destabilized during this purification and release dynamically associated proteins such as the proteasome. Taken together, these results strongly indicate that the proteasome is present as an active enzyme in purified centrosomes.
The proteasome content of the centrosomes was determined by quantitative Western blotting and compared with that of total cellular proteasome (Table I). After multiple purification steps, approximately 1% of total cellular proteasome protein, in HeLa cells under normal culture conditions, remains associated with the centrosome during the purification procedure (see "Discussion"). The specific activity of the centrosome-associated proteasome was approximately 16-fold greater than that of the purified 20 S proteasome (Table I). Moreover, PA28, which can activate the purified 20 S proteasome by over 20fold, activated the centrosome-associated proteasome by only 2-fold. These data strongly suggest that the centrosome-associated proteasome exists in an activated form, perhaps bound to activators such as PA700 and/or PA28, each of which also has been localized to the centrosome (16).
A unique characteristic of the 26 S proteasome (a complex of 20 S proteasome and PA700) is its selective degradation of ubiquitinated proteins. To test whether active 26 S proteasomes are present at the centrosome, purified centrosomes were examined for their ability to catalyze the degradation of a defined polyubiquitinated substrate (Ub 5 -DHFR) in vitro. As seen in Fig. 2A, there is a reduction in Ub 5 -DHFR after incubation with centrosomes reflecting degradation of this ubiquitinated protein substrate. This proteolysis was ATP-dependent and significantly reduced by proteasome-specific inhibitors ( Fig. 2A). These data provide further strong support for the association of active 26 S proteasomes with the centrosome. A band of unknown identity was detected at a higher molecular weight than the substrate in some but not all of the experiments ( Fig. 2A). However, this band cannot account for the loss of all of the Ub 5 -DHFR.
The 26 S proteasome catalyzes the sequential removal of ubiquitin molecules from the distal end of polyubiquitin chains. This activity is associated with at least one subunit of PA700 (33). Upon incubation of purified centrosomes with the proteasome substrate Ub 5 -DHFR in the absence of ubiquitin aldehyde, an inhibitor of isopeptidase activity, the appearance of multiple bands at positions consistent with the removal of one to three ubiquitin molecules from Ub 5 -DHFR was detectable by 30 min (Fig. 2B). Similar results were obtained for the purified PA700 complex (data not shown). In addition, the disappearance of low intensity bands at higher molecular weights was also observed. These bands may represent minor populations of Ub 5 -DHFR molecules further modified by ubiquitin molecules. Thus, in addition to the degradation of both polyubiquitinated and proteasome-specific peptide substrates, the purified centrosome preparations also contain ubiquitin isopeptidase activity.
Previously immunocytochemical studies showed that treatment of HeLa or HEK293 cells with lactacystin caused a significant increase in the diameter of staining at the centrosome commensurate with an increase in 20 S proteasome, PA28, and PA700 (16). To biochemically quantify the extent to which the proteasome and its regulatory proteins are recruited to the centrosome in response to proteasome inhibitors, centrosomes were purified from cells treated with lactacystin or MG132. Western blot analysis showed that relative to untreated cells, neither drug had a measurable effect on total cellular levels of 20 S proteasome, PA28, or PA700 (Fig. 3). By contrast, lactacystin and MG132 promoted a striking increase of 20 S proteasome, PA28, and PA700 protein at the centrosome (Fig. 3). The magnitudes of the increases with lactacystin and MG132 were: 2.4-and 4.0-fold, respectively, for the proteasome; 3.9-and 7.0-fold, respectively, for PA28; and 2.0-and 4.0-fold, respectively, for PA700 (Fig. 3). The significantly larger effect of MG132 compared with lactacystin on recruitment of proteasomal components to the centrosome was also observed by immunocytochemistry (data not shown). In contrast, neither proteasome inhibitor had any effect on ␥-tubulin levels in either the whole cell lysates or centrosome fractions (Fig. 3).
The microtubule network and its associated motor proteins (dynein, kinesin, and related proteins) are responsible for the transport of various organelles and proteins within the cell. Because the centrosome is the microtubule-organizing center of the cell, it seemed possible that the recruitment of the proteasome and related proteins to the centrosome might require an intact microtubule system. Therefore, we tested the effect of the microtubule depolymerizing agent, nocodazole, on the recruitment of the proteasome, PA28, and PA700 to the centrosome. Cells were pretreated with nocodazole for 2-3 h, at concentrations known to completely disrupt microtubules, and then treated with MG132 for 12 h. No discernable difference in levels of 20 S proteasome, PA28, or PA700 was observed at the centrosome between control and nocodazole-treated cells (Fig.  4). These results suggest that intact microtubules are not necessary for the recruitment of the proteasome, PA28, and PA700 to the centrosome in response to proteasome inhibition.

DISCUSSION
The current work demonstrates that the proteasome is present in a functional and activated form at the centrosome. Consistent with this, purified centrosomes catalyze the hydrolysis of a proteasome-specific peptide and a ubiquitinated protein substrate. This activity is inhibited by specific inhibitors of the proteasome and, in the case of the ubiquitinated-protein, is ATP-dependent. An additional associated ubiquitin isopeptidase activity is also observed in the centrosomal fraction. Both of the two known activators of the proteasome, PA700 and PA28, are present at the centrosome. PA700 is known to con-  The blot in B was exposed longer to emphasize deubiquitinated forms of DHFR, and therefore, the intensities should not be directly compared with A.
tain ubiquitin isopeptidase activity and be required for ATPdependent degradation of ubiquitinated substrates. PA28 is known to stimulate the degradation of peptide substrates by 20 S. Significantly, the addition of purified PA28 to isolated centrosomes only stimulates the proteolysis approximately 2-fold, in contrast to the 20-fold stimulation of purified 20 S proteasome. Thus, activated proteasome at the centrosome is likely accounted for by complexes consisting of 20 S and both of the regulators.
Our data demonstrate that at least 1% of the cellular proteasome remains associated with the centrosome during the multistep purification procedure. In addition to any dissociation during purification, this value is likely a significant underestimate because the recovery of centrosomes in our biochemical preparations is much less than 100%. Unfortunately, calculations of centrosome recovery are prevented by the presence of soluble, cytoplasmic forms of the centrosomal marker proteins, ␥-tubulin and pericentrin (19,34). Interestingly, about 1% of the total cellular pools of several other proteins including p85 (a regulatory subunit of phosphatidylinositol 3-kinase) and GTPase-activating protein and centrosome-associated protein have been estimated to associate with the cen-trosome (35,36). Thus, the partitioning of even low amounts of specific proteins at the centrosome may confer specialized functions to the cell.
Recently, we and others have shown that the centrosome is a site for the accumulation of proteasome substrates. Both proteins whose levels are determined by regulated proteasome-dependent degradation and misfolded proteins that are recognized and disposed of by this system have been detected at this site. Immunocytochemistry has revealed that the tumor suppressor p53 (25), presenilin 1 (27), misfolded CFTR (16,27), cyclins (26), and IB (28) are present at the centrosome; the presence of IB also has been demonstrated by biochemical analysis. 3 Moreover, the proteasome substrates cyclins A, B1, and E localize to the centrosome in a cell cycle dependent manner (37)(38)(39). This localization may be physiologically relevant, in that the proteasome plays an important role in control of the cell cycle by catalyzing the temporal degradation of cyclins and kinase inhibitors (26,40). Thus, proteasome activity at the centrosome is in a location that suggests involvement in the cell cycle. Interestingly, in fission yeast the 26 S proteasome redistributes from the nuclear periphery to a nuclearassociated spot during meiosis (14). Furthermore, mutations in at least two subunits of PA700 have phenotypes of shortened mitotic spindles and abnormal chromosome separation in yeast (41,42). The importance of the accumulation of misfolded proteins at this site remains obscure.
The current data demonstrate that levels of the proteasome and its regulators also are greatly increased in centrosomes prepared from cells in which the proteasome had been inhibited. These biochemical results support and extend previous data that demonstrated this effect using immunocytochemistry (16). The basis of this response is unclear but may represent an attempt by the cell to maintain normal proteolysis. A similar response can be evoked by the overexpression of certain abnormal proteins, which also accumulate at the centrosome (16,27). Therefore, it is possible that any deficiency in proteasome function at the centrosome caused by proteasome inhibitors or by a large excess of substrate (e.g. overexpressed abnormal proteins) promotes the recruitment of more proteasome to the site as a compensatory response. Unfortunately, inhibition of the proteasome used to promote recruitment prevents a direct ex- Centrosomes were partially purified on a 60% sucrose cushion as described under "Experimental Procedures" and subjected to SDS-PAGE and Western blot analysis with antisera against the indicated proteins followed by densitometry of the resulting films. Each bar represents the fold increase in immunoreactivity relative to untreated control cells. Results are the average of two independent experiments with a variation of Ͻ5%. The dashed line represents protein levels in control cells and is included as a visual aid to the reader. perimental demonstration of a correlation between the increased proteasome protein and increased proteasome activity at the centrosome. However, consistent with our biochemical results, Anton et al. (43) recently reported the disappearance of accumulated substrates at the centrosome upon removal of proteasome inhibitors. This observation further supports the direct measurements of protease activity at the centrosome presented here.
Additional protein systems are known to be involved in the targeting of certain proteins for degradation. For example, Hsc70 has been implicated in the conjugation of ubiquitin to certain substrates (44) targeting them for proteasomal degradation, a function which may involve the anti-apoptotic protein BAG-1 via its ubiquitin-like domain (45). Previously it has been shown that Hsc70 and other chaperones accumulate at the centrosome in response to thermal stress (25) and proteasome inhibition (16). In addition, both these treatments lead to an induction of cell stress chaperones, but not of the proteasome proteolytic machinery (Fig. 2) (46 -48) indicating distinct regulatory mechanisms for these two important classes of proteins. We have previously demonstrated that an increased load of misfolded protein at the centrosome is sufficient to induce recruitment of the proteasome (16). This further suggests the recruitment of proteasomal machinery to the centrosome in response to inhibition (Fig. 2) is because of the accumulation of substrates at this site and not simply the result of a stress-dependent induction of expression.
Intact microtubules are necessary for the formation of perinuclear substrate aggregates (27), although they are not necessary for aggregate stability (16). In contrast to substrate, our quantitative biochemical analysis indicates that microtubules are not required to achieve increased levels of centrosomeassociated proteasome, PA700, or PA28 upon proteasome inhibition. Similarly, other proteins including Nek2-protein kinase, phosphatidylinositol 3-kinase, and AKAP350 interact with the centrosome in a microtubule-independent manner (35,49,50). These results suggest that cells may utilize distinct mechanisms to move different proteins to the centrosome.
In summary, evidence presented here suggests that proteasome-catalyzed proteolysis may be regulated, in part, by its selective partitioning to the centrosome. Thus, subcellular distribution, a common mechanism for the regulation of other enzymatic systems, may also be an important feature of regulation of the proteasome pathway.