| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Department of Biochemistry, Boston University Medical School, Boston, Massachusetts, USA;
School of Biology and Institute for Bioengineering and Bioscience, Georgia Institute of Technology, Atlanta, Georgia, USA; and
Cardiology Research Center, Moscow, Russia
1Correspondence: 715 Albany St., K323, Boston, MA 02118, USA. E-mail: sherma1{at}bu.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: Huntington’s disease
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
]. Protein aggregation is a hallmark and often the cause of many human disorders. For example, expansion of polyglutamine domains (polyQ) in certain proteins makes them less soluble and causes neurodegeneration in Huntington’s disease and various ataxias. Aggregates are found both in cytoplasm and in nuclei, evidently, at each of these compartments contributing to the development of a pathology. It appears that oligomers, such as protofibrils, or other small intermediates of aggregation are toxic [for a review, see (2)
], while large inclusion bodies are cytoprotective (3)
.
In contrast to experiments in a test tube, protein aggregation in cells is an elaborate tightly regulated process, which depends on various cellular components and determines function and stability of many damaged and mutant proteins in a cell. For example, polypeptides with expanded polyQ in cytosol were reported to accumulate specifically at centrosomes in association with chaperones, ubiquitin, and proteasomes, thus representing a typical aggresome (4
5
6
7)
; polyQ aggregates in the nucleus also associate with chaperones and components of ubiquitin-proteasome system and accumulate in PML domains (8
, 9)
. Aggregation of polypeptides with expanded polyQ is controlled by activities of various signaling mediators, including glucocorticoids (10)
, the protein kinases MEKK1 (11)
, Akt (12)
, and ROCK1 (13)
, Rac1-interacting protein Arfaptin2 (14)
, an SH3 domain protein SH3GL3 (15)
, a G protein-coupled receptor kinase-interacting protein GIT1 (16)
, and others.
Using a yeast model, we and others demonstrated that polyQ aggregation depends on a special prion conformation of either Rnq1, or New1, or Sup35 proteins, (17
18
19)
. Indeed, polypeptides with expanded polyQ remained soluble even at high levels in the cells deprived of these prions, in contrast to control cells where robust aggregation was seen. These data indicate that cells have developed a complicated and well-regulated machinery to bring abnormal and otherwise aggregation-predisposed polypeptides into aggregates, but the mechanism of this process is yet to be uncovered.
Here, we used a yeast model (18)
to study the mechanisms of polyQ aggregation by a genetic search for cellular factors that play a role in the aggregation process. Previously, using this model, we described involvement of various molecular chaperones at different steps of polyQ aggregation. We also demonstrated that polyQ aggregation leads to a rapid cessation of endocytosis and suggested that disruption of endocytosis is one of the causes for cellular toxicity (20)
. Here, we do not address polyQ-associated toxicity but investigate mechanisms of polyQ aggregation and its relation to endocytosis machinery.
Endocytosis is a vital cellular process essential for controlling composition of the plasma membrane, cellular signaling, and nutrient uptake. In yeast, 
50 proteins are known to play a role in formation of an endocytic vesicle and its scission; however, current understanding of their functions and multiple interactions is incomplete [for reviews, see (21
, 22)
]. Many of these proteins concentrate at the plasma membrane forming cortical patches, which are thought to be the sites for formation of endocytic vesicles. Since the composition of cortical patches varies, it was initially assumed that there are functionally different coexisting isoforms of the patches in a cell. However, recently, the dynamic nature of the cortical patches was uncovered by time lapse microscopy (23
, 24)
, and now it is widely accepted that during the internalization step of endocytosis, a multimodule complex forms and matures through precisely orchestrated, both spatially and temporally, engagements and disengagements of specific proteins. The entire process of formation, maturation, and internalization of the endocytic vesicles is very fast and takes less than a minute. In yeast, Las17 and Pan1 play a critical role in maturation of cortical patches and the formation of the endocytic vesicles. Furthermore, these proteins serve in activation of the seven-protein Arp2/3 complex responsible for F-actin polymerization, which propels endocytic vesicles into the cytosol [for reviews, see (25
, 26)
]. Here, we will refer to the dynamic structure that drives formation of endocytic vesicles as the endocytic complex (EC). Numerous components of yeast EC have structural and functional homologues in mammalian cells, and many aspects of formation of endocytic vesicle appear to be conserved in evolution [for a review, see (21)
].
In this work, using yeast strains with deletions of components of EC we demonstrate that EC is essential for polyQ aggregation. We further show that endocytosis machinery in mammalian cells is similarly involved in polyQ aggregation.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Yeast strains and plasmids
Strains with GFP-tagged endogenous proteins (27)
and HIS3 marker in a parental strain MATa his3
leu2 met15 ura3 were obtained from the Yeast GFP Clone Collection (Invitrogen, Carlsbad, CA, USA). Deletion mutants of the wild type strains BY4739 (MAT
leu2 lys2 ura3) or BY4742 (MAT his3 leu2 lys2 ura3) were obtained from the deletion library (Invitrogen) of yeast nonessential genes (28)
. Other mutant and the corresponding wild-type strains used in this work were SEY6210, MAT his3 trp1 leu3 ura3 lys2 (S. Emr); BWY1237 - isogenic with SEY6210, pan1–20 (B. Wendland); DDY0131 - ade2 his3200 leu2–3,112 lys2–801 ura3–52 (D. Drubin); DDY1166 - sla2::HIS3, isogenic with DDY0131 (D. Drubin); and las17 - las17::KanMX2 in BY4742 (H. Bussey).
Cells were routinely grown at 30°C in selective medium with 2% sugar as indicated in the text. Cells grown in the selective medium with dextrose to a midlogarithmic phase were washed twice and left in galactose medium overnight. Cells grown in raffinose medium were supplemented with galactose and induced for 1 to 7 h. Mutant strain pan1 and sla2 and the corresponding wild-type strains were grown at 25°C; for las17 and the corresponding wild-type cells, 1.2 M sorbitol was added to the medium.
The plasmid for expression of FLAG-tagged at N-terminus and GFP-tagged at C-terminus 103Q construct under control of Gal1 promoter in pYES2 vector was described previously (18)
. Gal1 promoter is inhibited by dextrose and is induced by galactose in the absence of dextrose; in raffinose, neither induction nor repression of Gal promoter takes place. FLAG-tagged at N-terminus and GFP-tagged at C-terminus 47Q construct, as described previously (11)
, was subcloned into pYES2 vector. To construct 103QP plasmid, we amplified by PCR a proline-rich domain encompassing huntingtin exon 1 downstream of polyQ region. A template for the PCR was a gift of Dr. S. Lindquist (Whitehead Institute, Cambridge). The product of the PCR was inserted in frame between polyQ stretch and EGFP in 103Q construct.
The pFA6a-mRFP-KanMX6 plasmid containing a gene encoding monomeric red fluorescent protein (29)
was a gift of Dr. Won-Ki Huh (University of California, San Francisco, CA, USA). To tag the polyQ constructs with mRFP, its gene was put in frame downstream of huntingtin-derived constructs, in place of EGFP.
Mammalian cell cultures and vectors
HEK293T (human embryonic kidney) cells and HCT-116 (human colon carcinoma) cells were grown in Dulbecco’s modified Eagle’s medium and in McCoy’s 5A medium, respectively, each supplemented with 10% fetal bovine serum and L-glutamine at 37°C in an atmosphere of 5% CO2.
A plasmid encoding the first 665 amino acid of huntingtin FLAG-tagged at N terminus, including a stretch of 97 glutamine residues, was a gift of Dr. A. Kazantsev (Massachusetts General Hospital, Boston, MA, USA); we tagged this sequence at C terminus with a PCR-amplified mRFP sequence to make 97QL construct. For constitutive expression of mRFP-tagged polyQ constructs, we subcloned them into retroviral vector pCXbsr, a gift of Dr. T. Akagi (Osaka Bioscience Institute, Japan). For the production of retroviruses, we cotransfected HEK293T cells with a retroviral plasmid and the helper plasmids expressing retroviral proteins Gag-Pol, G (VSVG pseudotype) using GenePorter (GTS). Supernatants containing the retrovirus were collected 48 h after transfection. HEK293T and HCT-116 cells were infected overnight with two-times diluted supernatant in the presence of 10 µg/ml polybrene and washed; selection with 10 µg/ml blasticidin S was started 48 h after infection.
For reduction of the cellular levels of Arp2, we used RNAi-Ready pSIREN-RetroQ vector (BD Biosciences, San Jose, CA, USA). We selected GAGAAGATTGTAGAGGTAA sequence (encompassing 361 to 379 nucleotides in ORF of human Arp2) as a target for RNAi. As a negative control, we used an empty vector. Virus production and cell infection were done as described above, but 0.5 µg/ml puromycin was used for selection. A retroviral vector expressing EGFP was used as a control for infection efficiency: usually, 70–90% of cells were fluorescent 2–3 days after infection.
For reduction of the cellular levels of N-WASP, we transfected cells with N-WASP siRNA (human, Santa Cruz Biotechnology, Santa Cruz, CA, USA) using lipofectamine 2000 (Invitrogen) in accordance with the manufacturer’s protocol. As a negative control, we used nontargeting siRNA: AACGACCAGGGCAACCGCACC. Scrambled FITC-labeled oligo DNA, a gift of Dr. P. Leavis (Boston Biomedical Research Institute, Watertown, MA, USA), was used as a control for transfection efficiency: usually 80–90% of cells were fluorescent to a various intensity the next day after transfection.
Analysis of cellular proteins
Yeast cells were disrupted at 4°C by 10 min, vortexing with 425–600 µm acid-washed glass beads in lysis buffer (50 mM Tris HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 5 mM PMSF; 1 mM benzamidine; 5 µg/ml of each: leupeptin, pepstatin A, aprotinin). Homogenates were cleared by 3 min centrifugation at 300 g. Mammalian cells were scraped in the same buffer. Samples diluted to have equal amounts of total protein were analyzed by immunoblotting.
Lysates of yeast cells with endogenous GFP- tagged endocytic proteins expressing mRFP-tagged 103Q were produced the same way and analyzed by fluorescent microscopy for association of the proteins with the polyQ aggregates.
To assay for the levels of polyQ oligomers conditionally resistant to SDS yeast cells grown in minimal media with raffinose were induced with galactose for 1 h, disrupted as described above, and supplemented with Laemmli buffer to achieve 2% SDS. The lysates were either boiled for 3 min followed by PAGE electrophoresis or incubated for 10 min at 37°C followed by 2% agarose gel electrophoresis in TBE buffer with 0.1% SDS. Both gels were subjected to immunoblotting with anti-GFP antibody to visualize EGFP-tagged 103Q.
Microscopy
Fluorescent microscopy was performed at room temperature with Axiovert 200 (Carl Zeiss, Germany) microscope using x100 objective and the manufacturer’s AxioVision 4 software.
For quantification presented in Fig. 4
, at least 200 cells with detectable fluorescence were counted on images gained from several randomly chosen fields. Then the settings for these images were adjusted to retain visible only the top 20% of the brightest cells in control culture. Aggregates in the latrunculin A-treated culture were counted using the same settings.
|
For comparison of aggregation in wild-type and mutant cells with similar levels of polyQ we used Axiovision Automatic Measurement Wizard (Zeiss) to determine the median brightness of each cell on gained images; then the presence of aggregates was assessed only for cells falling into the selected range of brightness.
As described in (20)
, the spectrum of EGFP fluorescence in live yeast cells widens to leak through Texas Red channel in cells placed between a glass slide and a coverslip. To avoid this problem, for colocalization study, cell suspension was placed on Lab-Tek* Chambered Coverglasses (NUNC) pretreated with poly-L-lysine (Sigma, #P4707).
To assess the levels of aggregation in cell cultures grown on the Chambered Coverglasses, cells with and without visible aggregates were counted in several randomly chosen fields to have not less than 400 for each sample, and a fraction of cells with detectable aggregates was calculated. Counting was conducted blindly in all of the instances where the difference between two images was not obvious enough to tell them apart (as was the case with cells treated with the inhibitors of actin polymerization or with the N-WASP inhibitor).
For staining with Texas red-conjugated phalloidin (Molecular Probes, Eugene, OR, USA), cells grown on chambered coverglasses were fixed for 10 min with 4% formaldehyde in phosphate-buffered saline (PBS), washed twice with PBS, labeled with the dye (2 U/ml in PBS) for 10 min, and, after four washes, observed under the microscope.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. In this system, 103Q, a model polypeptide derived from huntingtin with expanded polyQ is tagged with EGFP (see Fig. 1
for the description of the used constructs and respective aggregation patterns), allowing monitoring of formation of aggregates under a fluorescent microscope. While working with this system, we noticed that certain mutations associated with endocytosis cause a delay in 103Q aggregation on galactose induction (see Materials and Methods), suggesting that proteins involved in endocytosis also play a role in polyQ aggregation. Our previous observation that polyQ aggregation inhibits endocytosis (20)
also suggested a close association between the two processes. Accordingly, we hypothesized that endocytosis machinery, in turn, plays a role in controlling polyQ aggregation.
|
To address this question, we tested whether mutations associated with the endocytosis machinery affect polyQ aggregation. Robust aggregation of 103Q after prolonged induction may render moderate effects of the endocytosis mutations undetectable. Therefore, to increase the sensitivity of the test, we assessed 103Q aggregation shortly after its induction. 1–1.5 h after the addition of galactose to cultures grown with raffinose, the vast majority of the wild-type cells contained 103Q aggregates. In contrast, under these conditions, in mutant cells deficient in various components of EC, including apl1, apl3, chc1, ede1, end3, hof1 las17, rvs161, rvs167, sla1, sla2, a large fraction of cells did not contain any detectable aggregates (Fig. 2
A and Table S1 in Supplemental Data). Further galactose induction led to aggregation of 103Q in 100% of both wild-type and mutant cells, thus obscuring their aggregation defects seen at earlier time points (not shown).
|
Since there is always a significant divergence in expression levels within cell populations, and, since las17 mutant cells are generally enlarged, we compared the polyQ aggregation in individual cells with similar intracellular concentrations of this protein. To do so, for images gained with the mutant and wild-type cells (as in Fig. 2A
), we preselected a narrow range of median fluorescence using Axiovision Automatic Measurement Wizard software (see Materials and Methods). All wild-type cells falling into these range contained polyQ aggregates. In contrast, only one-third of the las17 mutant cells within the same range had visible aggregates (not shown), indicating that the discovered aggregation defects are not associated with variations in the levels of the polyQ construct. In this and many further experiments, we focused on Las17, as it plays a critical role in EC maturation and links it to actin polymerization.
A strong suppression of 103Q aggregation by mutations associated with endocytosis was also seen in uninduced cells grown in raffinose. To detect fluorescence under these conditions, exposures had to be at least 20 times longer than with galactose-induced cells (not shown). In uninduced cells, 103Q partially aggregated, forming highly mobile tiny particles, which varied in their apparent sizes and brightness (Fig. 2B
, wild-type). In rare cells, one could see brighter and larger particles, which tend to be less mobile or immobile (not shown). Both, the number of particles per cell and the fraction of cells with aggregates were strongly reduced in mutants that affect EC, including apl1, apl3, chc1, ede1, end3, hof1, las17, rvs161, rvs167, sla1, sla2 (Fig. 2B
and Table S1 in Supplemental Data). The extent of the inhibitory effects of the listed mutations varied significantly, presumably because of redundancy of many proteins involved in EC.
To assess 103Q aggregation in wild type and EC mutant cells by an alternative method, we assayed for the levels of 103Q amyloid aggregates conditionally resistant to SDS. This assay was originally employed to investigate aggregation properties of another glutamine-rich protein Sup35 and allows assessment of the formation of aggregates, which are stable in SDS if not boiled (30)
. Lysates of cells that express 103Q were incubated with 2% SDS without boiling and subjected to agarose gel electrophoresis followed by immunoblotting with anti-GFP antibodies. As seen on Fig. 2C
(bottom), SDS-resistant 103Q aggregates ranging between 700 and 4,000 kDa can be visualized as a smear band. With this assay, we compared the extent of 103Q aggregation in lysates of wild-type, las17, or chc1 deletion cells, which were induced with galactose for 1 h and accumulated similar levels of 103Q, as seen with regular SDS-PAGE (Fig. 2C
, top). Figure 2C
(bottom) shows that the levels of SDS-resistant oligomers were strongly reduced in the lysates of the mutant cells, even compared to 4 times diluted lysates of wild-type cells.
As described above, because of a strong tendency for aggregation of 103Q, we could observe the effects of the mutations in endocytosis-associated genes only at low 103Q expression levels. To investigate effects of EC mutations on polyQ aggregation at higher levels, we took advantage of two other huntingtin-derived constructs 47Q and 103QP (Fig. 1A
). The patterns of aggregation of 47Q and 103QP in yeast differ significantly from the pattern seen with 103Q (18
, 20)
: 47Q forms fibrile-like aggregates that may branch and occasionally twirl into a compressed body, while 103QP forms distinct round or granular aggregates often converging into a single spot inside a cell (Fig. 1B
). Both polypeptides exhibit decreased propensity for aggregation compared to 103Q, so a fraction of cells expressing either 47Q or 103QP did not form aggregates detectable by fluorescent microscopy. Moreover, a significant diffused fluorescent background representing soluble EGFP-tagged polyQ polypeptides was seen in many cells with aggregates (Fig. 1B
). Using these constructs, we found that mutations in several components of EC, including apl1, apl3, chc1, ede1, end3, hof1, las17, rvs161, rvs167, and sla2 strongly reduced aggregation, even after long inductions (Fig. 2D
and Table S1 in Supplemental Data).
All of these data demonstrate that various mutations in cellular machinery associated with the formation of cortical actin patches and endocytosis hinder polyQ aggregation. These data strongly suggested that ECs play a pivotal role in the aggregation process.
Endocytic complexes at later stages of maturation are recruited into 103Q aggregates
To investigate interactions between EC and polyQ in live cells, we followed association of various GFP-tagged components of EC with 103Q aggregates. For these studies, we used the Yeast GFP Collection, which contains clones with GFP-tagged endogenous proteins (27)
. The advantage of this collection is that the GFP-tagged proteins are expressed at normal physiological levels, which is critical for investigation of interactions of multicomponent complexes. Since EC is a dynamic structure, we have chosen proteins present in ECs at various maturation stages. For example, the ECs marked with Ede1-GFP or Ent1-GFP had low mobility because these proteins associate with EC at early stages and leave EC before recruitment of the actin machinery. In contrast ECs marked with Sla1-GFP, Sla2-GFP, or Pan1-GFP had much higher mobility, as these proteins stay within ECs, even after the actin machinery is recruited. In these clones with GFP-labeled components of ECs, we expressed 103Q tagged with mRFP in place of originally used EGFP. Four hours after 103Q induction, large immobile aggregates of 103Q-mRFP can be seen in a significant fraction of the cells, sometimes along with smaller mobile aggregates (Fig. 3
).
|
Formation of ECs labeled with Pan1, Sla1, and Sla2 remained unaltered in the cells without polyQ or with the low levels of polyQ aggregation (Fig. 3
, marked with "O"). In contrast, in the cells with sizeable 103Q aggregates, EC populated with Pan1, Sla1, or Sla2 were disrupted, and these GFP-labeled proteins redistributed from membrane-localized ECs to the polyQ aggregates (two stars mark the cells with almost complete relocation of the labeled proteins into polyQ; one star—the cells with partial relocation). Importantly, in most images, these proteins localized within the aggregates as clusters or discrete spots. These data strongly suggest that Pan1, Sla1, or Sla2 did not continually coaggregate with polyQ during the aggregate’s growth, but rather incorporated into an aggregate as components of a distinct structure. In contrast to Sla1, Sla2, or Pan1 that are thought to associate with EC at intermediate and late stages of EC formation, proteins that reside in EC at earlier stages, including Ede1 and Ent1 (31
32
33
34)
, were not recruited into the polyQ aggregates (Fig. 3
, bottom, and (20)
). Moreover, in cells with large 103Q aggregates early EC complexes populated with Ede1 and Ent1 were not significantly affected, i.e., most such cells contained seemingly undisturbed patches of Ede1 or Ent1 at the periphery (Fig. 3
and not shown).
To verify that the observed recruitment of Sla1 and Sla2 proteins to polyQ aggregates does not result from potential false juxtaposition of aggregates and ECs within the cell, we tested whether this association is detectable after cell disruption, leading to a dramatic dilution of cellular content. GFP-tagged Sla1 and Sla2 were undetectable by immunoblotting; however, higher sensitivity of the fluorescent microscopic detection of GFP allowed us to follow the association of GFP-tagged endocytic component with polyQ aggregates in the cell homogenates. Both Sla1-GFP and Sla2-GFP retained their physical association with cell-free polyQ aggregates (Fig. S1 in Supplemental Data). In contrast, Ede1-GFP, while seen in unbroken cells at the levels comparable to those of Sla1-GFP and Sla2-GFP (not shown), was not seen in the cell-free aggregates (Fig. S1 in Supplemental Data). These data, along with previously reported association of Pan1, but not Ent1 with polyQ aggregates (20)
, suggest that the polyQ aggregation process involves EC populated with Sla1, Sla2, and Pan1, but not Ede1 or Ent1, i.e., only EC at a late stage of its maturation.
Disruption of ECs by various mutations inhibits polyQ aggregation (see above). On the other hand, if ECs at late stages are involved in polyQ aggregation, stabilization of these complexes may increase 103Q aggregation. To test this possibility, we prevented natural disassembly of the ECs by incubation of cells with 50 µM of latrunculin A, a potent inhibitor of actin polymerization (35
, 36)
. In the latrunculin-treated cells, ECs were stabilized after recruitment of Sla1, Sla2, and Pan1, so all of these proteins were seen exclusively in immobile patches "frozen" on plasma membrane (not shown). On the other hand, latrunculin A arrested EC maturation at a step before recruitment of Arp2, the activator of actin polymerization, since all of GFP-tagged Arp2 dissociated from patches and was seen diffusely distributed throughout the cytoplasm (not shown). As mentioned above, 103Q aggregation in such cells grown in raffinose is very limited, but stabilization of ECs led to emergence of larger (brighter) aggregates of 103Q (Fig. 4
), indicating that the EC is in fact involved in polyQ aggregation.
Endocytosis machinery is involved in polyQ aggregation in mammalian cell cultures
There is a growing pool of data that internalization events in endocytosis in mammalian cells are similar to those in yeast [for a review, see (21)
]. Many yeast proteins participating in EC have structural and functional mammalian homologues involved in both assembly of endocytic complexes on the plasma membrane and subsequent polymerization of cortical actin. To test whether endocytosis machinery is critical for polyQ aggregation in mammalian cells, we employed two polyQ constructs tagged with mRFP: 103Q, identical to the construct used in yeast cells, and 97QL (long), which represents much larger N-terminal part of mutant huntingtin encompassing its first 665 amino acid, including 97 glutamine residues (Fig. 1A
). Using retroviral vectors, we established HEK293T and HCT-116 cells lines that constitutively express these polyQ constructs. In both cell lines, the 103Q and 97QL polypeptides formed multiple small cytoplasmic or nuclear aggregates. The aggregation patterns of the both polypeptides were quite similar. There were no detectable aggresomes, as seen previously with 103Q polypeptide (11
, 37)
, most likely because of lower expression levels obtained with the retroviral vectors compared to transient transfection. Generally, only a fraction of the cell population contained detectable aggregates (upper panels in Figs. S2, S4 in Supplemental Data).
In yeast, deletion of LAS17 strongly suppressed polyQ aggregation (Fig. 2A, D
). Accordingly, we tested whether WASP, the mammalian homologue of Las17, is also important for polyQ aggregation. In contrast to yeast, which has a single LAS17 gene, mammals have five genes homologous to LAS17 [for a review, see (38)
]. We used the siRNA approach to deplete the major member of this protein family N-WASP, which is expressed ubiquitously with the strongest expression in nerve tissues. HEK293T cells expressing polyQ were transfected with siRNA for N-WASP, and 2 to 3 days later were assayed for the extent of polyQ aggregation. The levels of N-WASP decreased about two-fold on the third day after the transfection (Fig. 5
A). Partial down-regulation of N-WASP resulted in suppression of polyQ aggregation by 50% compared to cells transfected with control siRNA (Fig. 5B
).
|
To test whether inhibition of WASP recruitment into EC leads to the suppression of polyQ aggregation, we employed a specific inhibitor of such recruitment wiskostatin (39)
. HCT-116 and HEK293T cells expressing polyQ constructs were incubated with 5 or 2 µM wiskostatin or DMSO overnight, and then the extent of aggregation was assayed with the fluorescent microscope. In the cells exposed to the inhibitor, polyQ aggregation was suppressed in a concentration-dependent manner, becoming practically undetectable at 5 µM wiskostatin (Fig. 5C
and Figs. S2, S5B in Supplemental Data). These data indicate that the presence of WASP in the endocytic complex is essential for polyQ aggregation.
One of the functions of WASP in endocytosis is initiation of actin polymerization by Arp2/Arp3 complex [for reviews, see (21
, 40
, 41)
]. Arp2 is recruited into EC via direct interaction with WASP proteins (44)
. Therefore, we tested whether depletion of Arp2 would also affect polyQ aggregation in mammalian cells. We constructed siRNA for human Arp2, which on delivery into the cells via a retroviral vector, strongly decreased the levels of this protein (Fig. 6
A). Depletion of Arp2 in cells expressing polyQ constructs led to a significant increase of 103Q and 97QL aggregation (Fig. 6B
), the effect opposite to one seen with N-WASP depletion. These opposing effects are in accord with the suggested role of late EC in aggregation. Indeed, Las17/WASP enters the EC at the earlier stages and is important for maturation of the complex. In contrast, Arp2 is recruited into the EC just before the initiation of actin polymerization, which promotes internalization of the endocytic vesicle followed by the dissociation of the complex. Therefore, deletion or depletion of WASP should prevent formation of late ECs, while depletion of Arp2 blocks the internalization event, thus stabilizing the complexes. As in yeast, inhibitors of actin polymerization in mammalian cells suppress the internalization step of endocytosis (42)
. Therefore, we tested if inhibition of actin polymerization can enhance polyQ aggregation, as seen with the yeast model (see above). Cells expressing 103Q or 97QL were treated with either 1 µM latrunculin A, 2 µM cytochalasin D, or DMSO as a control. The treatments caused severe disruption of the actin cytoskeleton, as seen with phalloidin staining (Fig. S3A in Supplemental Data). Importantly, neither latrunculin A, nor cytochalasin D caused cell death (not shown). Moreover, distribution and solubility of a reporter protein EGFP expressed in the treated cells were unaffected (Fig. S3B in Supplemental Data), and cellular levels of polyQ were not significantly affected by the treatments (Fig. S4B in Supplemental Data). As seen in Fig. 6C
, both inhibitors significantly increased the extent of polyQ aggregation. Not only the number of cells with detectable aggregates grew considerably in the treated cultures, but also the average number of aggregates per cell, as well as their general brightness increased (Fig. S4). Therefore, inhibition of actin polymerization has the same effect on polyQ aggregation as does Arp2 depletion, again suggesting that this process is up-regulated by stabilization of EC at the late stages of maturation.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Formation of an endocytic vesicle is mediated by EC, a dynamic complex of proteins. We found that a variety of mutations or compounds that disturbed the integrity of EC inhibited polyQ aggregation. On the other hand, treatments that stabilized ECs promoted the aggregation. Accordingly, we hypothesized that EC may be intimately involved in formation of polyQ aggregates in cytosol. Our model is strongly supported by the fact that several components of EC were found within 103Q aggregates. Interestingly, these proteins are not seen diffusely scattered over the polyQ aggregate, indicating that EC elements do not continuously coaggregate with polyQ. Rather, they were consistently seen clustered in the 103Q aggregates. A possible scenario is that polyQ monomers or soluble oligomers associate with ECs at a late stage of their maturation, which, with certain probability, may result in bringing together enough polyQ polypeptides to seed aggregation. These ECs remain entrapped in the growing aggregates. In line with this model is our finding that the components of EC found in the aggregates did not show interaction with soluble polyQ molecules in two-hybrid system or coimmunoprecipitation (not shown). Because only components of late EC were found within the aggregates, we suggested that specifically late EC are involved in polyQ aggregation (see model in Fig. 7
).
|
To examine how general is the observed mechanism, we tested our model with mammalian cultured cells taking advantage of the fact that many features of the endocytosis machinery are evolutionarily conserved. We predicted that stabilizing the short-lived EC at the later stages may promote nucleation of polyQ aggregation. Indeed, stabilization of endocytic complex by inhibition of polymerization of cortical actin, one of the latest steps in internalization, caused an increase in polyQ aggregation. Similarly, polyQ aggregation was enhanced by depletion of Arp2 (Fig. 7)
. On the other hand, according to our model, inhibition of EC progression at earlier stages might suppress the aggregation. In fact, inhibition or depletion of N-WASP, the upstream element of the complex, strongly reduced the aggregation (Fig. 7)
.
The discovery of the aggresome demonstrated that cells utilize microtubule-mediated traffic to assemble multiple small cytosolic aggregates into one body associated with a centrosome (43
44
45
46
47
48)
. Our data suggest that the endocytosis machinery can be employed by the cell to deal with aggregation-prone polyQ-containing proteins in seeding of aggregates.
Involvement of EC-related machinery in polyQ aggregation appears not to be restricted to huntingtin and its derivatives. In fact, formation of prion aggregates of a yeast QN-rich protein Sup35 also appears to involve components of EC, as deletions in Sla1, Sla2, and certain other EC-associated proteins decreased the de novo prion formation (49
, 50)
. Furthermore, aggregates of the GFP-tagged Sup35 colocalized with Sla1- and Sla2-marked patches (50)
. Aggregation of both huntingtin and Sup35 in yeast requires the Rnq1 protein to be present in a prion form (17
, 18)
. However, the suppression of this aggregation by EC-related mutations was not related to a loss of Rnq1 prion form, since Rnq1 remained in an aggregated form in the mutant strains (not shown). Rnq1 may interact with some elements of EC, as demonstrated by a high-throughput, two-hybrid analysis, and therefore Rnq1 prions and EC may cooperate in seeding polyQ aggregates in yeast. In mammals, this process may depend only on EC since no mammalian homologues of Rnq1 are known.
There are several proteins associated with endocytic machinery that interact with polyQ and may serve to recruit this polypeptide into the endocytic complex. For instance, HIP1, the mammalian structural and functional homologue of Sla2, was originally discovered as huntingtin interacting protein (51)
. Moreover, huntingtin itself associates with endocytic vesicles and has been implicated in endocytosis (52)
. This raises the possibility that normal huntingtin encounters some elements of the endocytosis machinery on the regular basis, and only upon expansion of polyQ it is committed to aggregation by the same machinery.
Endocytosis and actin cytoskeleton organization are subjects for inter- and intracellular regulation by multiple signaling pathways. For instance, members of WASP family, discussed in this work, were reported to associate and be controlled by cdc42, Rac, PIP2 among other signaling mediators [for a review, see (38)
]. It is possible that regulation of polyQ aggregation by certain signaling pathways, described in the Introduction, is mediated via changes in the actin cytoskeleton. An interesting indication that actin machinery may be involved in polyQ aggregation came from a recent finding that inhibitors of ROCK1 kinase that regulates actin machinery strongly reduce aggregation of an androgen receptor with expanded polyQ (13)
. Furthermore, depletion of certain factors that indirectly regulate actin polymerization and endocytosis, like SH3GL3 (15)
or GIT1 (16)
, also suppressed polyQ aggregation.
Association of the polyQ aggregation process with the endocytosis machinery may be relevant not only to Huntington’s disease but also to other polyQ-expansion disorders. Better understanding of the relationship between polyQ aggregation and actin cytoskeleton may suggest new therapeutic targets in neurons, accumulating proteins with expanded polyglutamine domains.
|
ACKNOWLEDGMENTS |
|---|
Received for publication July 17, 2006. Accepted for publication January 18, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Shao and M. I. Diamond Polyglutamine diseases: emerging concepts in pathogenesis and therapy Hum. Mol. Genet., October 15, 2007; 16(R2): R115 - R123. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||
