Triggering Aggresome Formation

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 283, NO. 41, pp. 27575–27584, October 10, 2008 © 2008 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

Triggering Aggresome Formation DISSECTING AGGRESOME-TARGETING AND AGGREGATION SIGNALS IN SYNPHILIN 1 *□ S

Received for publication, March 20, 2008, and in revised form, June 13, 2008 Published, JBC Papers in Press, July 17, 2008, DOI 10.1074/jbc.M802216200

Nava Zaarur, Anatoli B. Meriin, Vladimir L. Gabai, and Michael Y. Sherman1 From the Department of Biochemistry, Boston University Medical School, Boston, Massachusetts 02118

Molecular chaperones and the ubiquitin-proteasome system (UPS)2 play an important role in handling soluble abnormal polypeptides that could arise as a result of misfolding, damage, or mutations (1). However, under certain conditions these systems fail to repair or destroy abnormal species, leading to formation of small cytoplasmic aggregates. These aggregates can cause cell toxicity leading to various pathologies, including major neurodegenerative disorders (2). Recently, it was discov-

* 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. The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1–S5. 1 To whom correspondence should be addressed: Dept. of Biochemistry, Boston University Medical School, 715 Albany St., K323, Boston, MA 02118. Tel.: 617- 638-5971; Fax: 617-638-5339; E-mail: [email protected]. 2 The abbreviations used are: UPS, ubiquitin-proteasome system; 17-AAG, 17-allylamino-17-demethoxy geldanamycin; GFAP, glial fibrillary acidic protein; GFP, green fluorescent protein; si-HSF1, small interfering HSF1; UPR, unfolded protein response. □ S

OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41

ered that special machinery has evolved in cells that transport small protein aggregates in a microtubules-dependent manner to the centrosome, forming an organelle called aggresome (3–5). The aggresome serves as a storage compartment for protein aggregates and could be actively involved in their refolding and degradation. In fact, major chaperones, like Hsp70 or Hsp27, and components of UPS are recruited to aggresome (6). Furthermore, recently it was demonstrated that autophagic clearance of protein aggregates also associates with aggresome (7). There is a notion in the field that aggresome represents a protective cellular response to a buildup of aggregating abnormal polypeptides under the conditions when chaperones and UPS machineries fail to handle abnormal species (8, 9). Indeed, it was reported that there is a close correlation between aggresome formation and cell survival (10). Furthermore, toxicity of abnormal proteins is strongly enhanced by inhibition of the microtubules-dependent transport, which is required for aggresome formation (5). In line with this concept, aggresomes are usually seen in mammalian cells after inhibition of the proteasome (11–13). Beside aggresome formation, other protein aggregation pathways also appear to exist in mammalian cells. For example, mutant glial fibrillary acidic protein (GFAP) expressed in cells seems to be unable to form aggresomes and usually forms small multiple aggregates all around the cytoplasm (14). Therefore, to understand the aggresome response, it is critical to use clear mechanistic criteria of aggresomes, including the microtubule dependence of its formation and co-localization with the centrosome. Recently, a number of factors were implicated in aggresome formation. For example, a microtubule-associated histone deacetylase HDAC6 was shown to interact with aggregates of ubiquitinated proteins via its ubiquitin binding BUZ domain and facilitate their association with the dynein motor protein that drives this cargo to the aggresome (15). Other ubiquitinbinding proteins like PLIC or ataxin 3 play a role in aggresome targeting of several abnormal proteins (16 –18). These findings suggest a close connection between ubiquitination of abnormal polypeptides and their accumulation in the aggresome. A possible model is that factors with ubiquitin binding domains recognize small aggregates of ubiquitinated abnormal proteins and facilitate their association with dynein, which in turn promotes their transport to the centrosome. However, at present it is not clear whether ubiquitination of abnormal proteins is an essential prerequisite of aggresome formation. In fact, it was demonstrated that aggresome formation by a model polypeptide 250GFP does not involve ubiquitination (19). JOURNAL OF BIOLOGICAL CHEMISTRY

27575

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

Abnormal polypeptides that escape proteasome-dependent degradation and aggregate in cytosol can be transported via microtubules to an aggresome, a recently discovered organelle where aggregated proteins are stored or degraded by autophagy. We used synphilin 1, a protein implicated in Parkinson disease, as a model to study mechanisms of aggresome formation. When expressed in naı¨ve HEK293 cells, synphilin 1 forms multiple small highly mobile aggregates. However, proteasome or Hsp90 inhibition rapidly triggered their translocation into the aggresome, and surprisingly, this response was independent on the expression level of synphilin 1. Therefore, aggresome formation, but not aggregation of synphilin 1, represents a special cellular response to a failure of the proteasome/chaperone machinery. Importantly, translocation to aggresomes required a special aggresome-targeting signal within the sequence of synphilin 1, an ankyrin-like repeat domain. On the other hand, formation of multiple small aggregates required an entirely different segment within synphilin 1, indicating that aggregation and aggresome formation determinants can be separated genetically. Furthermore, substitution of the ankyrin-like repeat in synphilin 1 with an aggresome-targeting signal from huntingtin was sufficient for aggresome formation upon inhibition of the proteasome. Analogously, attachment of the ankyrin-like repeat to a huntingtin fragment lacking its aggresome-targeting signal promoted its transport to aggresomes. These findings indicate the existence of transferable signals that target aggregation-prone polypeptides to aggresomes.

Aggresome Formation by Synphilin 1

3

Wang, Y., Meriin, A. B., Zaarur, N., Romanova, N. V., Chernoff, Y. O., Costello, C. E., Sherman, M. Y., submitted for publication.

27576 JOURNAL OF BIOLOGICAL CHEMISTRY

EXPERIMENTAL PROCEDURES Reagents and Antibodies—MG132, Radicicol, nocodazole, 17-allylamino-17-demethoxy geldanamycin (17-AAG), and tunicamycin were purchased from Biomol (Plymouth Meeting, PA), actinomycin D was from Invitrogen, and emetine was from Sigma, UBEI-41 was from BioGenova (Ellicott City, MD). Antibodies against HSP72 and HSF1 were purchased from Stressgen (Victoria British Columbia, Canada), against GFP (rabbit) was from Clontech (Palo Alto, CA), and against ␥-tubulin was from Santa Cruz Biotechnology (Santa Cruz, CA). Constructs—pEGFPN1 plasmid with cloned human synphilin 1 was a gift of Dr. Mark Cookson (NIA, National Institutes of Health). For expression in the retroviral system, the gene of C-terminal-tagged synphilin 1 was subcloned into pCXbsr vector, a gift of Dr. T. Akagi (Osaka Bioscience Institute, Japan). Deletion derivates of synphilin 1 were obtained by subcloning of the PCR-generated truncated sequences into pEGFPN1. The construct ANK1-CC-ANK2-CT encompasses 349 –919 amino acids, ANK1-CC-ANK2 encompasses 349 –729 amino acids, ANK1-CC encompasses 349 –561 amino acids, CC-ANK2-CT encompasses 505–919 amino acids, NT-ANK1 encompasses 1–505 amino acids, ANK2-CT encompasses 561–919 amino acids, and CT encompasses 729 –919 amino acids. The GFP-tagged constructs 103Q and 25QP were described previously (29). To construct 25QP-CC-ANK2, 25Q-CCANK2, and P-CC-ANK2, we subcloned upstream of CC-ANK2 PCR-generated (using 25QP as a template) sequences encompassing either full huntingtin exon1 (with 25 glutamine residues) or the stretch upstream the proline-rich domain or only the proline-rich domain, respectively. ANK1-CC-103Q was constructed by subcloning of the PCR-generated sequence encoding ANK1-CC upstream of 103Q. To construct Ub-GVGFP, we subcloned a PCR-generated ubiquitin with G76V substitution into the pQCXIP (Clontech) retroviral vector. The retroviral system expressing si-HSF1 was described previously (30). Cells Cultures, Transfection, Infection, and Treatments— HEK293 (human embryonic kidney) cells were grown in Dulbecco’s modified Eagle’s medium supplemented with 10% FBS (fetal bovine serum) and MCF10A (human breast epithelial) cells in Dulbecco’s modified Eagle’s medium/F-12 50/50 medium supplemented with 5% horse serum, 10 ␮g/ml insulin, 20 ng/ml epidermal growth factor, 100 ng/ml cholera toxin, 0.5 mg/ml hydrocortisone, both supplemented with L-glutamine at 37 °C in an atmosphere of 5% CO2. Cells were transfected using Lipofectamine 2000 reagent (Invitrogen) according to the manufacture’s instructions. For production of retroviruses, HEK293T cells were co-transfected with a retroviral plasmid and the helper plasmids expressing retroviral proteins Gag-Pol, G vesicular stamatitis virus glygoprotein (VSVG pseudotype). Supernatants containing retrovirus were collected 48 h after transfection. HEK293 and MCF10A cells were infected overnight with supernatant in the presence of 10 ␮g/ml Polybrene, and selection with blasticidin was started 48 h after infection. Routinely, a treatment of transfected cells started 24 h after transfection. Unless specified otherwise, incubations with VOLUME 283 • NUMBER 41 • OCTOBER 10, 2008

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

Although several factors were implicated in the aggresome formation, basic molecular mechanisms of this process are poorly understood. One of the problems is that in many studies of protein aggregation no clear distinction is made between aggresome and other types of aggregates. For example, a p62/ sequestosome protein that binds a non-canonical K63 ubiquitin chain was reported to play a role in aggregation of several substrates (20, 21), but it remains unclear how this aggregation relates to the aggresome formation. In other words, it is not apparent whether the aggregate is formed in the centrosome and in a microtubules-dependent fashion. These clear criteria distinguishing an aggresome from other types of inclusion bodies are essential from the mechanistic point of view. A hallmark of Parkinson disease is the appearance of protein aggregates called Lewy bodies in affected dopaminergic neurons. Lewy bodies have a lot of similarities with aggresomes and probably represent the same paradigm (22, 23). The major component of Lewy bodies is ␣-synuclein (24, 25). It has been demonstrated that the recently discovered synphilin 1 associates with ␣-synuclein and is necessary for its targeting to Lewy body-like protein aggregates in cell culture (26). Another player in formation of Lewy bodies appears to be the ubiquitin ligase, parkin (6), which is also implicated in certain juvenile forms of Parkinson disease (27). In fact, in parkinsonism caused by defects in parkin, no Lewy bodies could be formed despite severe symptoms. In cell culture it was demonstrated that parkin can ubiquitinate synphilin 1 in an unusual manner, forming a K63 ubiquitin chain, which promotes aggregation of synphilin 1 (21, 28). However, distinction between aggresome and other aggregates has not been done in that work. Recently, we established a yeast model to study the mechanisms of aggresome formation and identified several novel cellular components of the aggresome machinery, including the 14-3-3 protein Bmh1, the chaperone Cdc48 (VCP/p97), and its cofactors Ufd1 and Nlp4. An important mechanistic insight into the aggresome machinery was the discovery that aggresome targeting of an aggregation-prone huntingtin fragment with expanded polyglutamine (polyQ) domain is strongly facilitated by the proline-rich region of this protein, which serves as a special aggresome-targeting signal.3 This was the first report where the aggregation domain in a polypeptide (expanded polyQ) could be dissected from the aggresome-targeting signal (proline-rich domain, P-region). Furthermore, in co-expression experiments we demonstrated that a soluble protein consisting of short polyQ and the P-region (25QP) upon forming mixed aggregates with a long polyQ polypeptide lacking the P-region (103Q) can dramatically increase aggresome targeting. Therefore, a novel paradigm has emerged from this study that special transferable aggresome-targeting signals in substrate proteins could be necessary for aggresome formation. Here, we addressed whether this paradigm has a general significance by investigating the aggresome-targeting signal in synphilin 1. In the course of this study, we demonstrate that aggresome formation is a special response to proteasome inhibition, which also can be triggered by Hsp90 inhibitors.

Aggresome Formation by Synphilin 1

RESULTS Aggregation and Aggresome Formation by Synphilin 1—According to previously established mechanism-based criteria (3), the aggresome is a large centrosome-localized conglomerate of protein aggregates formed in a microtubule-dependent manner. Here, to investigate mechanisms of aggresome formation, we used synphilin 1 as a model substrate. Synphilin 1 tagged with the green fluorescent protein GFP (synph-GFP) was transiently expressed in HEK293 cells. 24 h after transfection, about 50% of transfected cells formed multiple small aggregates (Fig. 1A), whereas in the rest of the cells synph-GFP remained soluble. Differential centrifugation confirmed that a fraction of synph-GFP forms Triton X-100-resistant aggregates (Fig. 1B). Using time-lapse microscopy, we observed that the aggregates were highly mobile (supplemental movie Fig. S1). Their mobility was apparently stochastic, and sometimes events of fusion of aggregates or breaking apart were detectable (see the movie, Fig. S1). The movement of the small synph-GFP aggregates was insensitive to microtubule poisons nocodazole or benomyl and never led to formation of a single body co-localized with the centrosome (Fig. 1C, upper panel). Very importantly, inhibition of proteasome led to a dramatic change in the aggregation pattern of synph-GFP. Indeed, incubation with 5 ␮M proteasome inhibitor MG132 for 6 h led to disappearance of multiple small aggregates and formation of a OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

MG132 and/or nocodazole were done for 6 h, and the concentration of either inhibitor was kept at 5 ␮M. Cell Lysis and Analysis—Cells were lysed with lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 5 mM phenylmethylsulfonyl fluoride) and a protease inhibitor tablet (Roche Applied Science). Samples were adjusted to have equal concentrations of total protein and subjected to SDS-PAGE electrophoresis followed by immunoblotting. For the differential centrifugation analysis, the lysates were centrifuged at 13,000 ⫻ g for 30 min at 4 °C. The pellets were resuspended in the loading buffer with 2% SDS in volumes equal to those of the supernatant. Immunostaining—Cells were briefly washed with phosphatebuffered saline, permeabilized for 2 min with 0.1% Triton X-100, fixed in 1:1 mixture of acetone-ethanol for 10 min at ⫺20 °C, and washed. The fixed cells were stained with mouse anti-␥-tubulin antibody (1:250) for 1 h at room temperature, washed, and stained for 1 h at room temperature with the secondary antibody with Alexa Fluor 594 donkey anti-mouse IgG (1:500) (Molecular Probes). Stained cells were analyzed by fluorescent microscopy. Microscopy and Aggregates Counting—Fluorescent microscopy was performed at room temperature with an Axiovert 200 (Carl Zeiss, Germany) microscope using ⫻40 or ⫻100 objectives and the manufacturer’s AxioVision 4 software. SynphGFP and other GFP-tagged polypeptides were observed through the fluorescein isothiocyanate channel. To assess the fraction of cells with either aggresomes or multiple aggregates, the cells with the respective morphology as well as all fluorescent cells were blindly counted in 10 randomly chosen fields to have more than 250 cells in total. Each experiment was repeated three times to assure reproducibility of the results.

FIGURE 1. Synph-GFP forms multiple aggregates in naı¨ve HEK293 cells and an aggresome upon treatment with the proteasome inhibitor. A, fluorescent micrograph of cells transfected with synph-GFP plasmid. Transfected cells were either left untreated or incubated with either MG132 alone or MG132 and nocodazole. B, transfected cells were incubated with or without MG132, differential centrifugation analysis was performed, and synphGFP levels in the supernatant (Sup) or pellets (Pel) were assessed by immunoblotting with anti-GFP antibody. C, in cells treated with the proteasome inhibitor, synph-GFP co-localizes with the centrosome. Cells were incubated without (upper panel) or with MG132 (lower panel) for 6 h, fixed, and immunostained with ␥-tubulin antibody. Arrows show centrosomes. Green, synphGFP; red, ␥-tubulin. D, cells infected with retrovirus expressing synph-GFP were incubated for various time periods with MG132 or MG132 and nocodazole, and fractions of cells with an aggresome and with multiple aggregates were counted.

single large synph-GFP aggregate in the perinuclear region (Fig. 1A). Noteworthy, the effect of MG132 was seen at concentrations as low as 0.5 ␮M. A similar effect was seen upon incubation of cells with another proteasome inhibitor, Velcade (not shown). Interestingly, with certain other cell lines, e.g. MCF10A JOURNAL OF BIOLOGICAL CHEMISTRY

27577

Aggresome Formation by Synphilin 1

27578 JOURNAL OF BIOLOGICAL CHEMISTRY

FIGURE 2. Aggresome formation is independent on synph-GFP levels. A, levels of synph-GFP in cells incubated with or without MG132 were assayed by immunoblotting with anti-GFP antibody. B, HEK293 cells were transfected with various amounts of synph-GFP plasmid and incubated with or without MG132. The sample with the lowest level of synph-GFP represents retroviral expression. Expression levels with different concentrations of plasmid or retroviral expression (RV) are shown below the graph. Cells with an aggresome, multiple aggregates, or soluble synph-GFP were counted. The levels of synph-GFP in these cells were quantified by immunoblotting and Quantity One (Bio-Rad) software. The error bars on the graph represent S.D. C, HEK293 cells were infected with retrovirus encoding an unstable reporter protein Ub-GV-GFP and incubated with or without 20 ␮M UBEI-41 for 16 h followed by the addition of 5 ␮M MG132 for 6 h. The levels of the Ub-GV-GFP were monitored by immunoblotting. D, fluorescent micrographs of HEK293 cells infected with the retrovirus encoding synph-GFP and treated with UBEI-41 with or without of MG132. The experimental conditions were as in C.

ment with MG132 we observed a strong aggresome formation (Fig. 2B, the lower synph-GFP level point). Therefore, triggering of aggresome formation by inhibition of proteasome is unrelated to the increase in expression levels of synph-GFP. Does suppression of the UPS by inhibiting ubiquitination similarly triggers aggresome formation? To address this question, we utilized a recently developed inhibitor of the ubiquitinactivating enzyme E1, UBEI-41, that blocks proteasomedependent degradation (31). Overnight incubation with 20 ␮M UBEI-41 of HEK293 cells expressing the short-lived reporter polypeptide Ub-GV-GFP led to a strong stabilization and accumulation of this polypeptide, as was seen with MG132 (Fig. 2C). Similarly, UBEI-41 stabilized short-lived synph-GFP deletion constructs used in experiments described in Table 1 (not VOLUME 283 • NUMBER 41 • OCTOBER 10, 2008

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

(see below), aggresomes appeared within only 2 h upon the addition of a proteasome inhibitor. Of note, GFP alone expressed from the same vector showed diffused pattern, and MG132 didn’t cause its aggregation (Fig. S2). Immunostaining with an antibody against a centrosome component, ␥-tubulin, demonstrated that the single large aggregate of synph-GFP co-localized with the centrosome (Fig. 1C, lower panel), whereas aggregates in untreated cells did not show co-localization (Fig. 1C, upper panel). Moreover, formation of this aggregate was blocked by the microtubule poisons nocodazole or benomyl, resulting in accumulation of multiple aggregates (Fig. 1, A and D). Therefore, the single synph-GFP aggregate formed in the presence of proteasome inhibitors represented bona fide aggresome. These data clearly indicate that inhibition of the proteasome leads to the switch of the aggregation patterns of synph-GFP from multiple small aggregates to aggresomes. The emergence of multiple aggregates was dependent on the cellular levels of synph-GFP (in contrast to aggresome formation, which was independent on the levels of synph-GFP, as discussed below). Indeed, in cells with low levels of synph-GFP (e.g. achieved by retrovirus-based expression) multiple aggregates were formed only in a minor fraction of the population (less than 5%, compared with almost 50% at higher expression levels), whereas in the rest of the cells synph-GFP remained soluble. To understand the relations between multiple aggregates and aggresome, we monitored the kinetics of aggresome formation after the addition of MG132 to cells expressing low levels of synph-GFP. After 2 h of incubation with the proteasome inhibitor, we observed an increase in the fraction of cells with multiple aggregates and appearance of rare cells with aggresomes (Fig. 1D). By the end of the fourth hour, the fraction of cells with multiple aggregates dropped to almost zero, whereas the fraction of cells with aggresomes increased dramatically. Therefore, it appears that multiple aggregates represent an intermediate step in aggresome formation which can proceed further only upon inhibition of the proteasome. A possible mechanism for the effect of proteasome inhibitors on aggresome formation would be that MG132 stabilizes synph-GFP, leading to its accumulation and stronger aggregation. However, incubation with MG132 for 6 h caused little increase in their levels (Fig. 2A), although efficient aggresome formation was observed at this time. Of note, overexposure of this immunoblot (to allow detection of minor bands) did not reveal any forms of synphilin 1 with molecular weights higher than the major form (Fig. S3), which indicated that little, if any, ubiquitination of synphilin 1 occurs in this system. To carefully assess whether aggresome formation by synph-GFP depends on its cellular levels, we titrated down the amounts of the synph-GFP-encoding plasmid used in transfection of HEK293 cells, which resulted in the wide range (over 20-fold difference) of synph-GFP expression levels (Fig. 2B). In none of these samples were the aggresomes seen without treatment with the proteasome inhibitor, whereas incubation with MG132 led to robust aggresome formation in all the samples (Fig. 2B). Moreover, expression of this construct placed in a retroviral system resulted in dramatically lower levels of synph-GFP (see bellow), making them undetectable by immunoblotting, but upon treat-

Aggresome Formation by Synphilin 1 TABLE 1 Percent of cells with multiple aggregate or aggresomes in HEK293 cells transfected with different constructs Construct*** Synph FL ANK1-CC-ANK2-CT ANK1-CC-ANK2 ANK1-CC CC-ANK2-CT NT-ANK1 ANK2-CT CT

ⴙMG132*

No treatment

ⴙMG132 ⴙ nocodazole**

Multiple aggregates

Aggresome

Multiple aggregates

Aggresome

Multiple aggregates

Aggresome

60.5 ⫾ 3.0 51.8 ⫾ 2.1 42 ⫾ 2.0 0 32.8 ⫾ 3.8 0 0 0

4.7 ⫾ 1.1 0 0 0 0 0 0 0

13 ⫾ 1.7 4.8 ⫾ 1.8 4.2 ⫾ 0.3 0 40.3 ⫾ 3.1 1.7 ⫾ 0.6 0 0

65.7 ⫾ 4.0 53 ⫾ 2.1 49.0 ⫾ 1.7 6.8 ⫾ 0.8 0 9.7 ⫾ 1.5 0 0

65.3 ⫾ 2.1 49.0 ⫾ 1.7 40.3 ⫾ 2.1 5.2 ⫾ 1.0 34.7 ⫾ 2.9 7.8 ⫾ 1.3 0 0

6.8 ⫾ 1.0 7.8 ⫾ 1.0 5.0 ⫾ 1.7 0 0 0 0 0

Number in the table are mean ⫾ standard deviation. * Cells were incubated with MG132 (see Materials and Methods). ** Cells were incubated with MG132 and nocodazole (see Materials and Methods). *** See Fig. 4A for a description of the constructs.

OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

shown). However, in contrast to incubation with MG132, no changes in aggregation pattern of synph-GFP were detected upon incubation with UBEI-41, and this polypeptide remained in multiple aggregates (Fig. 2D). It is noteworthy that adding MG132 even after the overnight incubation with UBEI-41 and incubation with both agents for an additional 6 h led to efficient aggresome formation (Fig. 2D), suggesting that aggresome formation is specifically triggered by proteasome inhibition. Inhibition of Hsp90 Also Triggers the Aggresome Response— Inhibition of proteasome induces transcription of a number of genes, including the heat shock protein genes (32). Therefore, we hypothesized that activation of the aggresome formation could be a part of the heat shock response. To address this possibility, we tested whether other activators of the heat shock response, like elevated temperature or inhibitors of Hsp90, can trigger formation of aggresomes by synph-GFP. In these experiments, we utilized MCF10A cells that expressed synph-GFP in a retroviral system, which accounted for very low levels of this polypeptide. Under these conditions, synph-GFP did not form detectable multiple aggregates, greatly enhancing the sensitivity of detection of the aggresomes. Of note, untreated MCF10A cells did not show aggresomes, whereas the addition of MG132 or Velcade led to formation of aggresomes in almost 100% of cells within 2 h in a nocodazole-sensitive manner (Fig. 3A). Heat treatment (43 °C for 30 min) did not affect the aggregation pattern of synph-GFP, and no aggresomes were seen up to 12 h post-heat shock (not shown) when a robust accumulation of Hsp72 was observed. However, an Hsp90 inhibitor, Radicicol, triggered formation of aggresomes (Fig. 3A). Furthermore, a similar effect was seen with a distinct Hsp90 inhibitor 17-AAG (Fig. 3A). Of note, although Radicicol triggered formation of aggresomes in almost the entire cell population, these aggresome were smaller, and fluorescence of soluble synphGFP remained higher than those triggered by MG132 (Fig. 3A). Because both proteasome and Hsp90 inhibitors triggered aggresomes, we suggested that a common signaling event initiated by these treatments activates the aggresome response. Interestingly, many signaling pathways activated by inhibitors of the proteasome are actually blocked by inhibitors of Hsp90 (e.g. extracellular signal-regulated kinase pathway), and to our knowledge only the heat shock response and the endoplasmic reticulum stress response (UPR) can be activated by both (33). To test for the role of the UPR in triggering aggresome formation, we simply incubated synph-GFP-expressing cells with a

FIGURE 3. Effects of inhibitors of transcription, translation, or HSP90 on aggresome formation. A, fluorescent micrographs (fluorescein isothiocyanate channel) of MCF10A cells infected with synph-GFP-encoding retrovirus were treated for 6 h with the indicated inhibitors. Radicicol was used at 5 ␮M, 17-AAG was at 0.5 ␮M, emetine was at 10 ␮M, and actinomycin was at 5 ␮g/ml. To deplete HSF1, cells were co-infected with si-HSF1 retrovirus. The arrows show aggresomes. B, the level of HSF1 in the MCF10A cells co-infected with synph-GFP and si-HSF1 retroviruses (seen on Fig. 3A) was assayed by immunoblotting. C, levels of Hsp72 in the MCF10A cells infected with si-HSF1 or control retrovirus after incubation with MG132.

distinct activator of the UPR, tunicamycin, and did not observe any aggresome formation (not shown), strongly suggesting that activation of UPR is not sufficient for the effects of MG132 and Radicicol. Although heat shock at 43 °C for 30 min also did not trigger formation of aggresomes, this fact did not rule out the role of the heat shock response in aggresome formation, since heat treatment could cause denaturation of synph-GFP, thus preventing it from progressing through the aggresome pathway. JOURNAL OF BIOLOGICAL CHEMISTRY

27579

Aggresome Formation by Synphilin 1

27580 JOURNAL OF BIOLOGICAL CHEMISTRY

FIGURE 4. ANK1 segment represents the aggresome-targeting region, whereas CC-ANK2 segment is responsible for aggregation. A, a scheme of the deletion constructs used in this study. B, HEK293 transfected with various constructs were incubated with 5 ␮M MG132 for 6 h, and expression levels were assayed by immunoblotting with anti-GFP antibody. The levels of the ␤-actin were used as control for equal sample loading. The bands on the immunoblot were quantified with Quantity One software. C, fluorescent micrograph of aggregation patterns of the indicated deletion mutants of synph-GFP with or without MG132 treatment. CT, C terminus.

Within the central region, ANK1 domain appears to be the aggresome-targeting signal (the ANK1 sequence is shown in Fig. S5). Indeed, deletion of ANK1 domain (construct CC-ANK2-CT, Fig. 4A) led to a dramatic change in the aggregation properties of the polypeptide. Small mobile aggregates were still formed in the naı¨ve cells, although their numbers were reduced, but no aggresome triggering was seen upon treatment with proteasome inhibitor (Fig. 4C and Table 1). Therefore, a domain(s) responsible for the formation of small aggregates (aggregation domain) is distinct from the aggresome-targeting signal and is located downstream of ANK1. Indeed, further deletion of either the CC domain (construct ANK2-CT, Fig. 4A) or the ANK2 domain (construct ANK1-CC, Fig. 4A) prevented formation of multiple aggregates (Table 1, Fig. 4C), indicating that ANK2 together with the CC domain constitute the aggregation-promoting region of synphilin 1. Importantly, the addition of MG132 increased the levels of these constructs to the level of the full-length synph-GFP, but even under these conditions no multiple aggregates were VOLUME 283 • NUMBER 41 • OCTOBER 10, 2008

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

Therefore, we decided to further test for the role of activation of the heat shock response in triggering aggresome formation by testing whether this process could be suppressed by depletion of the heat shock transcription factor Hsf1. Accordingly, Hsf1 was depleted from MCF10A cells expressing synph-GFP using Hsf1 short hairpin RNA delivered on a retrovirus. Depletion of Hsf1 (Fig. 3B) almost completely prevented induction of the heat shock protein Hsp72 in response to MG132 (Fig. 3C), but formation of aggresomes in response to the proteasome inhibitor was not affected (Fig. 3A). Therefore, the heat shock response is not relevant to the aggresome response. Because inhibition of the proteasome and Hsp90 induces a number of genes, we tested whether triggering aggresome formation by MG132 and/or Radicicol involves transcriptional response. Therefore, an inhibitor of transcription, actinomycin D, was added to MCF10A cells expressing synph-GFP simultaneously with MG132 or Radicicol. Actinomycin D completely blocked transcriptional activation of Hsp72 by either MG132 or heat shock (Fig. S4), but it did not inhibit aggresome formation (Fig. 3A), indicating that the aggresome response does not involve activation of transcription. On the other hand, inhibitor of translation emetin completely prevented aggresome formation triggered by either MG132 or Radicicol (Fig. 3A). Therefore, protein synthesis, but not RNA synthesis, is essential for the aggresome response, suggesting involvement of a shortlived protein regulator in the aggresome response. Aggresome-targeting and Aggregation-promoting Domains in Synphilin 1—To further investigate the process of aggresome targeting, we sought to identify in synphilin 1 the domains responsible for this process. Accordingly, a series of deletion mutants of synph-GFP were constructed (Fig. 4A). Of note, various constructs accumulated at different levels in naı¨ve cells because some of the deletions destabilized the polypeptides. However, upon incubation with MG132, the levels of the unstable polypeptides increased, and we were able to assess aggregation patterns of various synphilin 1 derivates accumulated at similar levels (Fig. 4B). Our main focus was on the central region of synphilin 1 that has several domains, including ankyrin-like repeat domain ANK1 and coiled-coil domain CC followed by another ankyrinlike repeat ANK2 (Fig. 4A, full-length). HEK293 cells were transiently transfected with plasmids encoding either of these constructs, and aggregation patterns in the absence and presence of MG132 were monitored. Deletion of the N-terminal region that precedes ANK1 (construct ANK1-CC-ANK2-CT, see Fig. 4A) did not change either formation of multiple aggregates or triggering the aggresome formation by MG132, indicating that this region does not affect aggregation properties of synphilin 1 (Table 1). Deletion of both the N-terminal domain and the C-terminal region downstream of ANK2 (construct ANK1-CCANK2, Fig. 4A) similarly did not change the aggregation response, although a fraction of naı¨ve cells with multiple aggregates was slightly reduced (Table 1, Fig. 4C), possibly because of its reduced stability. Therefore, the central region of synphilin 1 defines both the formation of multiple aggregates in naı¨ve cells and formation of aggresome after incubation with proteasome inhibitors.

Aggresome Formation by Synphilin 1

OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY

27581

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

research indicated that the N-terminal part of huntingtin can serve in aggresome targeting in yeast.3 An important component of this targeting is the proline-rich region (P-region), which follows immediately the polyQ domain of huntingtin, although low levels of aggresome targeting could be achieved even without the P-region. Another important component of the signal was the N17 domain that precedes the polyQ domain,3 indicating a complex nature of the aggresometargeting signal. Here we tested if the ANK1 domain of synphilin 1 can substitute this endogenous huntingtin aggresome-targeting signal to provide aggresome formation by a polypeptide with expanded polyglutamine but lacking the P-region (103Q) (29). In 5–10% of transfected HEK293 cells 103Q formed a single dense inclusion body, which differed from aggresome, as its formation was independent of microtubules (Fig. 5C). Upon incubation of cells with 5 ␮M MG132 for 16 h, aggregation of 103Q was enhanced, but under these conditions this FIGURE 5. The aggresome-targeting signals of synphilin 1 and huntingtin are transferable. A, scheme of polypeptide formed mostly multiple the deletion constructs used in this study. B and C, fluorescent micrographs and quantification of the aggreaggregates and could not form gation patterns of cells transfected with the indicated constructs and incubated with or without MG132 and nocodazole. Error bars represent S.D. The increase in the aggresome formation due to addition of either ANK1 aggresomes (Fig. 5C) due the lack of or ANK-CC is statistically significant (p value ⬍0.0005). The inhibition by nocodazole of the number of aggre- the aggresome-promoting P-region. somes formed by either ANK1–103Q or ANK1-CC-103Q is also statistically significant (p value ⬍0.0005 and ⬍0.01, respectively). D and E, fluorescent micrograph and quantification of the aggregation patterns of cells Accordingly, we constructed a transfected with the indicated constructs and incubated with or without MG132 for 6 h. Significantly higher fusion polypeptide ANK1⫺103Q fraction of cells expressing either 25Q-CC-ANK2-CT or 25QP-CC-ANK2-CT than of cells expressing CC-ANK2-CT (Fig. 5A) and monitored its aggregaformed aggresomes in response to MG132 treatment (p ⬍ 0.0005). CT, C terminus. tion properties. As mentioned above, the ANK1 construct lacks formed (Fig. 4C). Interestingly, incubation with MG132 of cells the aggregation domain and does not aggregate. Interestingly, expressing either ANK1-CC or NT-ANK1, lacking either a part the ANK1 domain completely suppressed formation of the or the entire aggregation signal, respectively, led to formation of non-aggresome inclusion bodies by 103Q (Fig. 5, B and C). aggresomes in a small fraction (about 10%) of the cells (Table 1). Incubation with MG132 led to formation of single large aggreTherefore, even without the aggregation, signal synphilin 1 gate in about 45% of cells (Fig. 5, B and C). These single aggrederivatives can form aggresomes albeit with very low efficiency, gates represented aggresome, as the addition of nocodazole which probably reflects a low efficiency aggregation in the suppressed their formation, and instead, multiple aggregates absence of the CC-ANK2 domain. were seen (Fig. 5C). Therefore, ANK1 domain of synphilin 1 can Overall, the deletion analysis indicates that ANK1 domain is provide aggresome targeting of the aggregation-prone 103Q. necessary and sufficient for aggresome targeting, whereas both In a parallel experiment, in addition to ANK1 domain we also CC and ANK2 domains are necessary, and their combination is introduced C domain, a part of the aggregate-forming region of sufficient for formation of multiple aggregates. In other words, synphilin 1. Similarly to ANK1, ANK1-CC suppressed formaANK1 represents the aggresome-targeting signal, whereas CC- tion of the non-aggresome inclusion bodies by 103Q. However, ANK2 represents the aggregation-promoting segment. with additional CC domain, multiple aggregates were seen in Aggresome Targeting Signals Can Work with Heterologous about 45% of naı¨ve cells (Fig. 5, B and C), supporting the role of Polypeptides—Furthermore, we addressed the question of CC domain in formation of small aggregates. Incubation with whether the aggresome-targeting signal is transferable and the proteasome inhibitor caused formation of the nocodazolecan target to aggresome-unrelated polypeptides. Our prior sensitive aggresome in about 18% of transfected cells (Fig. 4, B

Aggresome Formation by Synphilin 1

DISCUSSION Much evidence has accumulated to indicate that the aggresome in eukaryotic cells represents a special protective response to proteotoxic stresses. Accordingly, the aggresome is fundamentally different from other types of protein aggregates, some of which cause cellular toxicity. To understand the formation of aggresomes, it is critical to define this process using mechanism-based criteria, which include the microtubules-dependent formation and localization at the centrosome site. These criteria were essential for understanding relations between different modes of aggregation of synphilin 1. Here we demonstrate that synphilin 1 has two modes of aggregation, multiple small mobile aggregates and aggresomes, and the switch between these aggregation modes is triggered by inhibition of the proteasome. Indeed, in naı¨ve cells mostly multiple aggregates were seen, whereas in the presence of the proteasome inhibitor MG132 synphilin 1 formed aggresomes. Surprisingly, this switch was unrelated to the expression levels of synphilin 1, since proteasome inhibitors triggered aggresome

27582 JOURNAL OF BIOLOGICAL CHEMISTRY

formation in cells within a wide range of expression levels (Fig. 2B). Interestingly, with the majority of model substrates, like CFTR (cystic fibrosis transmembrane conductance regulator) or presenilin 1, aggresome formation in cell cultures required incubation with proteasome inhibitors (11, 13). Substrates that appeared to form an aggresome-like structure associated with the centrosome without treatment with proteasome inhibitors were polyglutamine-containing polypeptides (34, 35). However, as we demonstrated recently, the single centrosome-associated aggregate formed by polyQ in the absence of proteasome inhibitors is not a true aggresome since its formation does not involve microtubules-dependent transport, and thus, mechanisms of targeting of polyQ to this aggregate are fundamentally different. Therefore, aggresome formation seems to represent a cellular response to proteasome inhibition. Interestingly, the proteasome is inhibited in various neurological disorders and upon aging (36, 37). Furthermore, in vitro experiments showed that the proteasome is inhibited in cells with protein aggregates (38). Initiation of protein aggregation in various disorders could cause proteasome inhibition, which further promotes protein aggregation, and aging may facilitate this vicious cycle. An intriguing possibility is that proteasome inhibition could trigger a protective aggresome response, which may allow cells to escape the cycle and survive. An important finding here is that signaling to trigger aggresome formation upon inhibition of proteasome is unrelated to a general cessation of UPS. Indeed, suppression of ubiquitination by an inhibitor of the ubiquitin-activating enzyme E1 did not turn on the synph-GFP aggresome formation. At the same time, this treatment caused strong stabilization of the model UPS substrate UB-GV (Fig. 2C). On the other hand, the addition of MG132 in the presence of E1 inhibitor activated aggresome formation (Fig. 2D). These data suggest that there is a special sensor of the proteasome function that signals to activate aggresome formation. This putative sensor could be analogous to a sensor of the proteasome stress in yeast (39), which is distinct from a sensor of the ubiquitination stress (40). Another possibility is that aggresome formation is triggered by stabilization of an unstable regulatory polypeptide that is degraded in a proteasome-dependent and ubiquitin-independent manner. This possibility is consistent with the finding that the aggresome response requires protein synthesis but not transcription. It was previously reported that ubiquitination of synphilin 1 can lead to its aggregation and aggresome formation (4, 21). This process was associated with formation of K63 ubiquitin chain synthesized by E3 ligase parkin (41). However, these data were obtained in experiments with overexpression of parkin or other E3 ligases. It is unlikely that ubiquitination by endogenous parkin or other E3 enzymes is critical for triggering aggresome formation by synphilin 1 upon the addition of proteasome inhibitors for two reasons. First, it is unclear how proteasome inhibitors could enhance K63 ubiquitination, which normally does not target proteins for proteasome degradation. Second, MG132 can trigger aggresome formation in cells with inhibited ubiquitin-activating enzyme E1 (Fig. 2D). In another study it was suggested that aggregation of synphilin 1 is triggered by ubiquitination by an E3 ligase SIAH1 (42, 43). It VOLUME 283 • NUMBER 41 • OCTOBER 10, 2008

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

and C). Therefore, ANK1 domain of synphilin 1 can provide aggresome targeting of the aggregation-prone 103Q polypeptide, whereas CC domain facilitates formation of multiple small aggregates in this heterologous system. Next we hypothesized that, likewise, the aggresome targeting of the synphilin 1 derivatives lacking endogenous aggresometargeting signal can be promoted by the signal from huntingtin. As an aggresome-targeting signal, we used 25QP that represents the N-terminal part of normal huntingtin, with a polyQ domain not long enough to make the polypeptide aggregation prone. On the other hand, 25QP contains the P-regions, and as we demonstrated in yeast, can provide the aggresome targeting signal for 103Q in trans.3 Accordingly, we constructed a hybrid polypeptide that contains 25QP and the aggregation-promoting segment of synphilin 1 (25QP-CC-ANK2-CT, Fig. 5A). As with the CC-ANK2-CT construct (Fig. 5D), the hybrid polypeptide formed multiple aggregates in naı¨ve cells (Fig. 5D). Importantly, in contrast to CC-ANK2-CT, incubation with proteasome inhibitors led to formation of 25QP-CC-ANK2-CT aggresomes in about 75% of cells with aggregates (Fig. 5, D and E). Therefore, 25QP is able to provide the aggresome targeting signal for the aggregation domain of synphilin 1. It is noteworthy that the aggresome targeting of the related hybrid polypeptide lacking the P-region (25Q-CC-ANK2-CT, Fig. 5A) was less efficient. With this construct, aggresomes were seen in 55% of cells with aggregates, whereas in the rest of these cells multiple aggregates were seen (Fig. 5E). Therefore, as with huntingtin, the P-region appears to facilitate aggresome targeting of the synphilin 1 fragment. On the other hand, the P-region alone was insufficient for the aggresome targeting, as a hybrid construct P-CC-ANK2-CT (Fig. 5A) could not form aggresomes upon incubation with MG132 (not shown). These experiments indicate that the aggregation-prone fragment of synphilin 1 can be targeted to aggresomes by the heterologous aggresome-targeting signal of huntingtin. Overall, these data indicate that (a) inhibition of proteasome triggers aggresome formation, and (b) aggresome formation requires special transferable aggresome targeting signals in substrate polypeptides.

Aggresome Formation by Synphilin 1

OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41

ing signal of huntingtin 25QP (Fig. 5D). Furthermore, ANK1 can provide aggresome targeting of the signal-less fragment of huntingtin 103Q (Fig. 5B). These data suggest that there must be specific factors responsible for substrate recognition by the aggresome machinery. REFERENCES 1. Luo, G. R., Chen, S., and Le, W. D. (2007) Int. J. Biol. Sci. 3, 20 –26 2. Meriin, A. B., and Sherman, M. Y. (2005) Int. J. Hyperthermia 21, 403– 419 3. Corboy, M. J., Thomas, P. J., and Wigley, W. C. (2005) Methods Mol. Biol. 301, 305–327 4. Chung, K. K., Zhang, Y., Lim, K. L., Tanaka, Y., Huang, H., Gao, J., Ross, C. A., Dawson, V. L., and Dawson, T. M. (2001) Nat. Med. 7, 1144 –1150 5. Webb, J. L., Ravikumar, B., and Rubinsztein, D. C. (2004) Int. J. Biochem. Cell Biol. 36, 2541–2550 6. Junn, E., Lee, S. S., Suhr, U. T., and Mouradian, M. M. (2002) J. Biol. Chem. 277, 47870 – 47877 7. Garcia-Mata, R., Gao, Y. S., and Sztul, E. (2002) Traffic 3, 388 –396 8. Olzmann, J. A., Li, L., and Chin, L. S. (2008) Curr. Med. Chem. 15, 47– 60 9. Tanaka, M., Kim, Y. M., Lee, G., Junn, E., Iwatsubo, T., and Mouradian, M. M. (2004) J. Biol. Chem. 279, 4625– 4631 10. Taylor, J. P., Tanaka, F., Robitschek, J., Sandoval, C. M., Taye, A., Markovic-Plese, S., and Fischbeck, K. H. (2003) Hum. Mol. Genet. 12, 749 –757 11. Johnston, J. A., Ward, C. L., and Kopito, R. R. (1998) J. Cell Biol. 143, 1883–1898 12. Bandopadhyay, R., Kingsbury, A. E., Muqit, M. M., Harvey, K., Reid, A. R., Kilford, L., Engelender, S., Schlossmacher, M. G., Wood, N. W., Latchman, D. S., Harvey, R. J., and Lees, A. J. (2005) Neurobiol. Dis. 20, 401– 411 13. Kovacs, I., Lentini, K. M., Ingano, L. M., and Kovacs, D. M. (2006) J. Mol. Neurosci. 29, 9 –19 14. Quinlan, R. A., Brenner, M., Goldman, J. E., and Messing, A. (2007) Exp. Cell Res. 313, 2077–2087 15. Kawaguchi, Y., Kovacs, J. J., McLaurin, A., Vance, J. M., Ito, A., and Yao, T. P. (2003) Cell 115, 727–738 16. Heir, R., Ablasou, C., Dumontier, E., Elliott, M., Fagotto-Kaufmann, C., and Bedford, F. K. (2006) EMBO Rep. 7, 1252–1258 17. Burnett, B. G., and Pittman, R. N. (2005) Proc. Natl. Acad. Sci. U. S. A. 102, 4330 – 4335 18. Marx, F. P., Soehn, A. S., Berg, D., Melle, C., Schiesling, C., Lang, M., Kautzmann, S., Strauss, K. M., Franck, T., Engelender, S., Pahnke, J., Dawson, S., von Eggeling, F., Schulz, J. B., Riess, O., and Kruger, R. (2007) FASEB J. 21, 1759 –1767 19. Garcia-Mata, R., Bebok, Z., Sorscher, E. J., and Sztul, E. S. (1999) J. Cell Biol. 146, 1239 –1254 20. Donaldson, K. M., Li, W., Ching, K. A., Batalov, S., Tsai, C. C., and Joazeiro, C. A. (2003) Proc. Natl. Acad. Sci. U. S. A. 100, 8892– 8897 21. Lim, K. L., Chew, K. C., Tan, J. M., Wang, C., Chung, K. K., Zhang, Y., Tanaka, Y., Smith, W., Engelender, S., Ross, C. A., Dawson, V. L., and Dawson, T. M. (2005) J. Neurosci. 25, 2002–2009 22. McNaught, K. S., Shashidharan, P., Perl, D. P., Jenner, P., and Olanow, C. W. (2002) Eur. J. Neurosci. 16, 2136 –2148 23. Olanow, C. W., Perl, D. P., DeMartino, G. N., and McNaught, K. S. (2004) Lancet Neurol. 3, 496 –503 24. Spillantini, M. G., Schmidt, M. L., Lee, V. M., Trojanowski, J. Q., Jakes, R., and Goedert, M. (1997) Nature 388, 839 – 840 25. Wakabayashi, K., Tanji, K., Mori, F., and Takahashi, H. (2007) Neuropathology 27, 494 –506 26. Engelender, S., Kaminsky, Z., Guo, X., Sharp, A. H., Amaravi, R. K., Kleiderlein, J. J., Margolis, R. L., Troncoso, J. C., Lanahan, A. A., Worley, P. F., Dawson, V. L., Dawson, T. M., and Ross, C. A. (1999) Nat. Genet. 22, 110 –114 27. von Coelln, R., Dawson, V. L., and Dawson, T. M. (2004) Cell Tissue Res. 318, 175–184 28. Olzmann, J. A., and Chin, L. S. (2008) Autophagy 4, 85– 87 29. Meriin, A. B., Mabuchi, K., Gabai, V. L., Yaglom, J. A., Kazantsev, A., and Sherman, M. Y. (2001) J. Cell Biol. 153, 851– 864 30. Zaarur, N., Gabai, V. L., Porco, J. A., Jr., Calderwood, S., and Sherman,

JOURNAL OF BIOLOGICAL CHEMISTRY

27583

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

seems unlikely that SIAH1-dependent ubiquitination significantly contributes to the aggresome formation in our model, because the SIAH1 binding domain of synphilin 1 is located within the N-terminal segment (43), which is different from the aggresome domain that we found (Fig. 4A). As mentioned above, the proteasometriggered event appears to be an activation of a signaling pathway. These findings support a previous report with a model polypeptide 250-GFP that ubiquitination is not critical for aggresome targeting of some substrates (19). Interestingly, inhibitors of Hsp90 also could trigger aggresome formation (Fig. 3A). We suggest that there is a common signaling pathway initiated by inhibition of Hsp90 and proteasome that triggers the aggresome response. A potential link could be through HDAC6, a major factor in aggresome formation (15). HDAC6 directly interacts with Hsp90 and promotes its deacetylation and inactivation (44). It is possible that inhibitors of Hsp90 somehow affect function of HDAC6 in the aggresome pathway. Interestingly, Hsp90 inhibitors induce heat shock proteins and demonstrate strong protection in various models of neurodegenerative diseases (45). Our data suggest that the protective activity of Hsp90 inhibitors could, in part, be independent of the heat shock response and related to activation of the aggresome response. An important finding here is that the aggresome targeting requires a special aggresome-targeting signal in the sequence of a substrate polypeptide. In the case of synphilin 1 this signal was shown to be the ANK1 domain (Fig. 4A). Indeed, deletion of this domain prevented the aggresome targeting, but the remaining polypeptide CC-ANK2-CT still was able to form multiple aggregates (Fig. 4C). This finding is in line with our prior observation of the presence of an aggresome-targeting sequence in huntingtin. In that study we demonstrated that the N-terminal fragment of huntingtin serves as an aggresome targeting segment and that the proline-rich region of this polypeptide facilitates aggresome targeting in both mammalian cells and yeast. Therefore, it appears that the concept of aggresometargeting signals has a general significance. In other words, protein aggregation is not sufficient for the aggresome targeting, and a special aggresome-targeting signal is required. On the other hand the aggregation property of a protein is also critical, as formation of aggresomes by a polypeptide that has the entire ANK1 signal but lacks the aggregation segment (e.g. NT-ANK1 or ANK1-CC) was very inefficient. The need for the aggresome-targeting signal is probably reflected in the lack of aggresome formation by certain aggregating polypeptides, e.g. GFAP. Indeed, published data demonstrate that normal or mutant GFAP can form in cells multiple aggregates but not an aggresome (14). We also never observed an aggresome formation by GFAP either in the absence or in the presence of proteasome inhibitors (not shown). Therefore, it appears that GFAP, although able to aggregate, lacks the aggresome-targeting signal and, therefore, remains in small aggregates. Importantly, the aggresome-targeting signals are transferable. In other words, such signals can induce the formation of aggresomes while attached to various aggregating polypeptides. In fact, ANK1 signal can be replaced by the aggresome-target-

Aggresome Formation by Synphilin 1 M. Y. (2006) Cancer Res. 66, 1783–1791 31. Hershko, A., and Ciechanover, A. (1998) Annu. Rev. Biochem. 67, 425– 479 32. Yew, E. H., Cheung, N. S., Choy, M. S., Qi, R. Z., Lee, A. Y., Peng, Z. F., Melendez, A. J., Manikandan, J., Koay, E. S., Chiu, L. L., Ng, W. L., Whiteman, M., Kandiah, J., and Halliwell, B. (2005) J. Neurochem. 94, 943–956 33. Davenport, E. L., Moore, H. E., Dunlop, A. S., Sharp, S. Y., Workman, P., Morgan, G. J., and Davies, F. E. (2007) Blood 110, 2641–2649 34. Arrasate, M., Mitra, S., Schweitzer, E. S., Segal, M. R., and Finkbeiner, S. (2004) Nature 431, 805– 810 35. Boeddrich, A., Lurz, R., and Wanker, E. E. (2003) Methods Mol. Biol. 232, 217–229 36. Layfield, R., Lowe, J., and Bedford, L. (2005) Essays Biochem. 41, 157–171 37. Shah, I. M., and Di Napoli, M. (2007) Cardiovasc. Hematol. Disord. Drug Targets 7, 250 –273 38. Tseng, B. P., Green, K. N., Chan, J. L., Blurton-Jones, M., and Laferla, F. M. (2007) Neurobiol. Aging, in press 39. Elsasser, S., Gali, R. R., Schwickart, M., Larsen, C. N., Leggett, D. S., Muller,

40.

41. 42.

43.

44.

45.

B., Feng, M. T., Tubing, F., Dittmar, G. A., and Finley, D. (2002) Nat. Cell Biol. 4, 725–730 Hanna, J., Hathaway, N. A., Tone, Y., Crosas, B., Elsasser, S., Kirkpatrick, D. S., Leggett, D. S., Gygi, S. P., King, R. W., and Finley, D. (2006) Cell 127, 99 –111 Lim, K. L., Dawson, V. L., and Dawson, T. M. (2006) Neurobiol. Aging 27, 524 –529 Liani, E., Eyal, A., Avraham, E., Shemer, R., Szargel, R., Berg, D., Bornemann, A., Riess, O., Ross, C. A., Rott, R., and Engelender, S. (2004) Proc. Natl. Acad. Sci. U. S. A. 101, 5500 –5505 Nagano, Y., Yamashita, H., Takahashi, T., Kishida, S., Nakamura, T., Iseki, E., Hattori, N., Mizuno, Y., Kikuchi, A., and Matsumoto, M. (2003) J. Biol. Chem. 278, 51504 –51514 Boyault, C., Zhang, Y., Fritah, S., Caron, C., Gilquin, B., Kwon, S. H., Garrido, C., Yao, T. P., Vourc’h, C., Matthias, P., and Khochbin, S. (2007) Genes Dev. 21, 2172–2181 Waza, M., Adachi, H., Katsuno, M., Minamiyama, M., Tanaka, F., Doyu, M., and Sobue, G. (2006) J. Mol. Med. 84, 635– 646

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

27584 JOURNAL OF BIOLOGICAL CHEMISTRY

VOLUME 283 • NUMBER 41 • OCTOBER 10, 2008

Fig. S1. The time lapse analysis of movement of small aggregates. HEK293 cells were transfected with synph-GFP and observed by fluorescence microscopy. Pictures were taken every 20s. The file is played in the QuickTimeMovie. Fig. S2. GFP control. In contrast to synph-GFP, GFP alone does not form aggregates either in naïve cells or following treatment with MG132. Conditions of the experiments are described in the legend to Fig. 1. Fig. S3. Synph-GFP does not undergo significant ubiquitination. HEK293 cells were transfected with synph-GFP plasmid and at 48 hours post-transfection MG132 was added for 6 hours, during which aggresomes were formed. Cell lyzates were prepared and immunoblotted with anti-GFP antibody. Blots were heavily overexposed to show ubiquitinated forms of synph-GFP. . Fig. S4. MCF10A cells treated with 5 μM MG132 with and without 10μg/ml Actinomycin. Levels of Hsp72 were assayed by immunobloting. Fig. S5. Sequence of ANK1 domain.

Protein Synthesis, Post-Translational Modification, and Degradation: Triggering Aggresome Formation: DISSECTING AGGRESOME-TARGETING AND AGGREGATION SIGNALS IN SYNPHILIN 1 Nava Zaarur, Anatoli B. Meriin, Vladimir L. Gabai and Michael Y. Sherman

Access the most updated version of this article at doi: 10.1074/jbc.M802216200 Find articles, minireviews, Reflections and Classics on similar topics on the JBC Affinity Sites. Alerts: • When this article is cited • When a correction for this article is posted Click here to choose from all of JBC's e-mail alerts

Supplemental material: http://www.jbc.org/content/suppl/2008/07/21/M802216200.DC1.html This article cites 44 references, 15 of which can be accessed free at http://www.jbc.org/content/283/41/27575.full.html#ref-list-1

Downloaded from http://www.jbc.org/ by guest on December 27, 2015

J. Biol. Chem. 2008, 283:27575-27584. doi: 10.1074/jbc.M802216200 originally published online July 17, 2008