Polyglutamine expansion in the N terminus of huntingtin (htt) causes selective neuronal dysfunction and cell death by unknown mechanisms. Truncated htt expressed in vitro produced htt immunoreactive cytoplasmic bodies (htt bodies). The fibrillar core of the mutant htt body resisted protease treatment and contained cathepsin D, ubiquitin, and heat shock protein (HSP) 40. The shell of the htt body was composed of globules 14-34 nm in diameter and was protease sensitive. HSP70, proteasome, dynamin, and the htt binding partners htt interacting protein 1 (HIP1), SH3-containing Grb2-like protein (SH3GL3), and 14.7K-interacting protein were reduced in their normal location and redistributed to the shell. Removal of a series of prolines adjacent to the polyglutamine region in htt blocked formation of the shell of the htt body and redistribution of dynamin, HIP1, SH3GL3, and proteasome to it. Internalization of transferrin was impaired in cells that formed htt bodies. In cortical neurons of Huntington's disease patients with early stage pathology, dynamin immunoreactivity accumulated in cytoplasmic bodies. Results suggest that accumulation of a nonfibrillar form of mutant htt in the cytoplasm contributes to neuronal dysfunction by sequestering proteins involved in vesicle trafficking.
- Huntington's disease
- vesicle trafficking
- protein aggregates
- protein interactions
Huntington's disease (HD) is an autosomal dominant disease resulting in the expansion of a polyglutamine tract in huntingtin (htt). Neuropathology of HD is characterized by selective neuronal loss in the striatum and cortex. N-terminal fragments of mutant htt form insoluble aggregates in the nucleus and cytoplasm of affected neurons in the HD brain (DiFiglia et al., 1997). The aggregates formed in HD brain have amyloid-like fibrils. N-terminal mutant htt encoded by exon 1 (aa 1-89 in normal htt) also forms amyloid-like fibrils in vitro (Scherzinger et al., 1997; Hazeki et al., 2002; Poirer et al., 2002). The role of mutant htt aggregates in cell death is controversial, because the presence of aggregates does not consistently correlate with cell death (Saudou et al., 1998; Sisodia 1998; Kim et al., 1999; Kuemmerle et al., 1999). A soluble, nonaggregated, mutant htt might contribute to neurodegeneration by engaging in abnormal protein interactions (Bao et al., 1996; Sittler et al., 1998; Passani et al., 2000; Gervais et al., 2002). Oligomers of mutant htt expressed from exon 1 form spheroids and protofibrils as intermediates (prefibrillar forms) during fibril assembly (Poirier et al., 2002). Although the toxic effects of either prefibrillar or soluble, nonfibrillar forms of mutant htt have not yet been determined in cell models of HD, it has been reported that stable, soluble oligomers of β-amyloid 1-42 have neurotoxic effects (Walsh et al., 2002).
We showed that truncated wild-type (wt) or mutant htt 1-969 aa expressed in clonal striatal cells assembled into cytoplasmic structures that contained the lysosomal enzyme cathepsin D (Kegel et al., 2000). At the EM level, some of the htt was bound to early and late autophagososomes, which are prelysosomal structures involved in the removal of cytoplasmic proteins. Immunofluorescence microscopy suggested that htt-labeled autophagosomes fuse with each other to form larger and more condensed bodies. In this study, we refer to the coalesced, htt-labeled autophagosomes as htt bodies. It is unclear whether htt bodies form in neurons of HD patients to the same extent that they do in vitro and have a pathological role. Marked accumulations of htt in the cytoplasm of HD neurons are associated with multivesicular bodies, a morphology characteristic of autophagosomes (Sapp et al., 1997). Because they accumulate htt, htt bodies might be sites to identify different forms of htt and to study the interaction of htt with other cytoplasmic proteins. The polyproline region that lies adjacent and distal to the polyglutamine tract in htt is a sequence potentially capable of different interactions with proteins that have Src homology 3 (SH3) and WW domains (Liu et al., 1997; Faber et al., 1998; Sittler et al., 1998; Passani et al., 2000). The proline stretch may also confer structural stability to the N-terminal region of htt. Therefore, we speculated that a deletion of the polyproline region in htt might affect htt structure, the formation of htt bodies, and htt interaction with binding partners.
Our results showed that distinct forms of htt are partitioned into a core and shell of the htt body. Mutant htt in the core was fibrillar and protease resistant. The htt in the shell formed oligomeric 14-34 nm globules and reduced and sequestered cytoplasmic proteins involved in vesicle trafficking, removing them from their normal locations in the cytoplasm. Polyglutamine expansion increased formation of htt bodies. Polyproline deletion reduced formation of htt bodies and prevented the redistribution of cytoplasmic proteins to the shell. The htt bodies impaired the transport of transferrin. Cortical neurons of grade 1/presymptomatic HD patients show altered distribution of dynamin, a sign that early dysfunction in endocytosis may occur. Our findings suggest that soluble oligomers of mutant htt that assemble at the periphery of htt bodies may cause cell dysfunction in HD, by sequestering proteins involved in vesicle trafficking through a polyproline-dependent mechanism.
Materials and Methods
Plasmid construction. The construction of expression plasmids of full-length htt and truncated htt has been described previously (Kim et al., 1999). A Flag tag was fused to the N terminus of htt with 18Q for wt htt and 46Q or 100Q for mutant htt. In this study, these constructs are denoted as htt with Flag tag (FH) 3144-18, FH3144-100 (for full-length htt), and FH969-18, FH969-46 and FH969-100 for truncated htt. The numbered residues in htt are based on the 23Q htt sequence. In addition, constructs were made encoding Flag fused to the N terminus of htt 1-171, 1-287, or 1-400 with different glutamine repeats (FH171-18, FH171-100, FH287-18, FH287-100, FH400-18, and FH400-100). For deletion of the region encoding prolines, htt cDNA was isolated by BamHI (314)→XhoI (823) digestion and treated with AciI, which cuts at bases encoding prolines. Two DNA bands were isolated from a gel: BamHI-AciI (bp 314-436) and AciI-XhoI (bp 571-823). A smaller band (bp 436-571) was lost during isolation. BamHI-AciI and AciI-XhoI were ligated together, and BamHI→XhoI was then cloned into BamHI NsiI of pcDNA-FLAG vectors containing 316-3221 bp of the htt gene with 18 or 46 CAG repeats. DNA sequencing confirmed that these procedures yielded a deletion of nucleotides that encode residues between aa 41-89 that include 30 prolines. The modified expression vectors were designated FH969-18 (proline-) and FH969-46 (proline-). A similar modification of truncated htt cDNA bp 316-3221 with 100 CAG repeats was unsuccessful and could not be used in this study.
Cell culture and DNA transfections. Three cells lines were used in this study. Mouse clonal striatal cells (x57) were developed by fusion of mouse day 17 embryonic striatum and mouse neuroblastoma cells (Wainwright et al., 1995). COS1 cells and MCF-7 cells were obtained from American Type Culture Collection (Manassas, VA). COS1 cells are SV40-transformed kidney fibroblasts from African green monkey. MCF-7 cells are epithelial-type cells derived from human mammary gland carcinoma. Mouse clonal striatal x57 cells (a generous gift from Dr. Alfred Heller, University of Chicago, Chicago, IL) and MCF-7 cells were cultured in 60 mm dishes in DMEM with 10% FBS, 4.5% glucose, and 50 U/ml penicillin G-streptomycin (Kim et al., 1999). MCF-7 and COS1 cells were cultured in 60 mm dishes in DMEM with 10% FBS and 50 u/ml penicillin G-streptomycin. At 80% confluence, cells were incubated in 2 ml of DMEM containing 5-10 μg of htt cDNA and 60 μl of Superfect (Qiagen, Valencia, CA). After 3 hr of incubation, cells were given 6 ml of fresh DMEM, and culturing was continued to the desired time or cells were split into 24-well plates containing poly-l-lysine-coated cover slides for use in immunohistochemistry.
Immunohistochemistry in cultured cells and brain tissue. x57, COS1, or MCF-7 cells were cultured on poly-l-lysine-coated microslips for 6-24 hr after transfection of different cDNAs encoding Flag-htt fusion proteins as described above. Cells were fixed with 4% paraformaldehyde in PBS and processed for immunohistochemistry as described previously (Qin et al., 2001). The formation of htt bodies or htt aggregates was detected with antibody against Flag (monoclonal antibody M5; Sigma, St. Louis, MO) and antibodies that recognize different htt epitopes including monoclonal antibody 2166 (aa 414-503 in htt; Chemicon, Temecula, CA), rabbit polyclonal antibodies Ab1 (aa 1-17 in htt), Ab585 (aa 585-725 in htt), and EM48 [aa 1-257 of htt without the polyglutamine and polyproline domains; a generous gift from Drs. Xiao-Jiang Li (Emory University School of Medicine, Atlanta, GA) and Stephen Hersch (Massachusetts General Hospital and Harvard Medical School, Charlestown, MA)]. Immunostained cells were examined with a confocal microscope (Radiance 2001; Bio-Rad, Hercules, CA) using a 100× oil immersion lens. Confocal images were captured with Laser Sharp 2000 software and merged in Adobe Photoshop (see more details below).
For double-labeling studies, anti-htt antibodies 2166 and EM48 were used. Anti-Flag antibody M5 was combined with rabbit polyclonal antibodies against htt interacting protein 1 (HIP1), SH3-containing Grb2-like protein (SH3GL3), heat shock protein (HSP) 40 (Santa Cruz Biotechnology, Santa Cruz, CA), HSP60 (Stress Genes, Victoria, British Columbia, Canada), HSP70 (Santa Cruz Biotechnology), cathepsin D (Calbiochem, San Diego, CA), active caspase-3 (Cell Signaling Technology, Beverly, MA), and proteasome (a generous gift from Dr Josè G. Castaño, Universidad Autónoma de Madrid, Madrid, Spain). Anti-htt polyclonal antibody Ab1 was combined with mouse monoclonal antibodies against 14.7K-interacting protein (FIP2; Transduction Laboratories, Lexington, KY), dynamin (PharMingen, San Diego, CA), ubiquitin (Santa Cruz Biotechnology), and dynamin-like protein 1 (DLP1; PharMingen). The specificity of antibodies has been validated before or tested in the present study by omission of primary or secondary antibodies. Secondary antisera included Bodipy fluorophore conjugated goat anti-mouse (Molecular Probes, Eugene, OR) and Cy3-conjugated donkey anti-mouse (Jackson Immunoresearch, West Grove, PA). The analysis of colocalization of cytoplasmic proteins in htt bodies was performed using a Krypton-Argon laser that scanned immunostained cells sequentially with single emission filters installed to detect Bodipy FL (excitation/emission maxima, 503/512) and Cy3 (excitation/emission maxima, 550/565). Single images were obtained after excitation at 488 nm to detect Bodipy FL or at 568 nm to detect Cy3, then merged using Adobe Photoshop. The sequential imaging procedure ensured that two images were taken at the same focal plane and avoided cross-excitation of the immunofluorescent labeling.
Immunoperoxidase labeling to detect dynamin or htt immunoreactivity was performed on 7 μm thick slide-mounted frozen sections of postmortem frontal cortex from human control brain (n = 3) and adult onset HD patient brain (n = 3). The method for immunoperoxidase labeling in human brain sections has been described previously (Sapp et al., 1995). The sections were pretreated with 10% methanol and 1% H2O2 for 15 min at room temperature before the immunostaining procedure. Polyclonal antibody EM48 (dilution, 1:1000) was used to detect htt. The mean age of controls and HD patients was 64 and 55 years, respectively. The postmortem interval was 17 hr for controls and 25 hr for HD cases. The control (C10, C12, and C18) and HD (A4, A13, A20) brains have been used in our previous published studies of htt localization (DiFiglia et al., 1997; Sapp et al., 1999). Two of the HD brains (A4 and A13) had grade 1 neuropathology, which signifies minimal loss of striatal neurons (Vonsattel et al., 1985). Both cases were verified to have an expanded CAG repeat in the HD gene. A4 was an asymptomatic patient. The other HD brain had grade 3 striatal pathology, which indicates significant striatal cell loss. Pyramidal cortical neurons are relatively preserved in grade 1 HD brain and depleted in grade 3 HD brain (Vonsattel and DiFiglia, 1998).
In situ proteinase K degradation of htt bodies. MCF-7 cells were cultured on poly-l-lysine-coated microslips and transfected with FH969-100. Cells were fixed in 4% paraformaldehyde in PBS 3 d after htt expression. Cells were washed in PBS and permeabilized with 0.1% Triton X-100 in PBS and then incubated with 0.25-27 μg/ml proteinase K in PBS for 10-40 min at room temperature. Cells were then washed in PBS and processed for double immunostaining using antisera 2166 and EM48 or dynamin and EM48. Degradation of htt bodies by proteinase K was analyzed with a laser confocal microscope as described above.
Quantification of htt bodies. X57 or MCF-7 cells were cultured on poly-l-lysine-coated microslips and transfected with FH969-18 or FH969-100, FH969-18 (proline-) or FH969-46, or FH969-46 (proline-). Cells were fixed and processed for immunohistochemistry using M5 and 2166. The slides were processed by an observer blind to the treatment conditions. One thousand Flag-positive cells on each microslip were examined for the expression of diffuse htt and htt bodies using a fluorescence microscope with 60× objective and an ocular grid (Eclipse TE 300; Nikon, Tokyo, Japan). The data were statistically analyzed with Student's t test or ANOVA and presented in bar figures as the number of cells with htt bodies per 1000 cells expressing exogenous htt. To quantify the number of cells colocalizing a cytoplasmic protein in htt bodies, 100 cells with htt bodies/slide were analyzed with a confocal microscope with 100× oil immersion objective.
EM examination. MCF-7 cells were cultured onto poly-l-lysine-coated microslips and transfected with FH969-18 or FH969-100, or FH287-18 or FH287-100. Cells were fixed in 4% paraformaldehyde in PBS 3 d after transfection. Cells were incubated with antibody M5 and horseradish peroxidase-conjugated anti-mouse IgG antibody. The htt bodies were visualized with DAB. Cells were postfixed in 2.5% glutaraldehyde, incubated in 1% osmium tetroxide and 1% uranyl acetate, dehydrated in increasing grades of alcohol, and embedded in an ethanol-soluble resin (LX112, LADD). Embedded cells were sectioned (Ultracut E; Reichert-Jung) and examined with a JEOL 100CX electron microscope.
Transferrin uptake assay. MCF-7 cells were cultured onto poly-l-lysine-coated microslips and transfected with FH969-18 or FH969-100. After 24 hr of htt expression, culture medium was removed, and cells were incubated in DMEM containing 2.0 mm HEPES and 0.2% BSA for 45 min at 37°C with 5.0% CO2 to remove surface-bound transferrin. The cells were washed twice with ice-cooled buffer containing 100 mm potassium acetate, 85 mm sucrose, 1 mm magnesium acetate, and 20 mm HEPES-NaOH (KSHM; pH 7.4). Microslips were incubated in 0.4 ml of KSHM with 0.2% BSA and 6.0 μg/ml human transferrin conjugated to tetramethylrhodamine at 37°C with 5% CO2 for 5-45 min. The cells were washed with cold KSHM buffer three times, fixed in 4% paraformaldehyde, and processed for immunohistochemical localization of Flag using M5 antibody and Bodipy FL-conjugated anti-mouse secondary antibody. Uptake of transferrin in nontransfected cells, in cells with diffuse htt expression and in cells with htt bodies was quantified with a fluorescence microscope using a 60× objective by an observer (E.S.) who had no knowledge of the treatment conditions. The uptake of transferrin in htt-expressing cells was rated as the same, significantly reduced (by at least 50%), or absent compared with nontransfected cells that were adjacent to the htt-expressing cells within the same microscopic field. The data were statistically analyzed with Student's t test. For illustration, confocal images were obtained as described above, and images were merged with Adobe Photoshop.
Different htt epitopes partition in htt bodies
We previously found that transient expression of Flag-htt 1-969 aa (FH969 with 18 glutamines for wild type or 46 or 100 glutamines for mutant type) in clonal striatal cells (x57 cells) produced htt immunofluorescent structures in the cytoplasm that accumulate the autophagosomal/lysosomal enzyme cathepsin D (Kim et al., 1999; Kegel et al., 2000). By electron microscopy, the htt exressed in the cytoplasm was associated with autophagosome-like bodies (Kegel et al., 2000). The formation of htt-labeled bodies coincided with the generation of N-terminal htt (N-htt) fragments. Cells with htt-labeled autophagic bodies were significantly smaller than cells without these structures. Polyglutamine expansion increased the remnants of dying cells with htt-labeled autophagic bodies (Kim et al., 1999). These findings suggested that the accumulation of mutant htt-labeled autophagic bodies contributed to cell death. A time course study of x57 cells showed that htt-labeled autophagic structures in the cytoplasm are first detected by immunofluorescence microscopy at ∼12 hr after transfection, begin to fuse and grow larger and more dense at ∼16 hr after transfection, and reach maximal sizes 24-72 hr after transfection (data not shown). Here, we refer to the larger fused htt-labeled structures as htt bodies. The htt bodies also formed in COS1 and MCF-7 cells. COS1 cells proved to have the best transfection efficiency of the three cell lines. COS1 and MCF-7 cells had a larger cytoplasmic volume than the x57 cells, which made it easier in these cells to study the cytoarchitecture of htt bodies by immunofluorescence and electron microscopy (see below). Although a greater proportion of x57 cells formed htt bodies, MCF-7 cells accumulated the highest density of htt bodies per cell. This observation may be related to the lack of the autophagy regulating protein beclin 1 in MCF-7 cells (Liang et al., 1999).
We used anti-Flag antibody (M5) and different anti-htt antisera [Ab1, EM48, monoclonal antibody (mAb) 2166, and Ab585] to examine htt bodies (Fig. 1A) in MCF-7 cells. There was strong immunoreactivity for Flag, N-htt1-17 (Ab1), htt 414-503 (mAb2166), and htt 585-725 (Ab585) along the perimeter or shell of htt bodies that produced a ring-like labeling (Fig. 1B). The center or core of the htt body was robustly labeled only by antibody EM48. The EM48 polyclonal antibody was made against aa 1-257 of htt without polyglutamine and polyproline domains and detects mutant htt aggregates in cell bodies and dendrites in the HD brain (Gutekunst et al., 1999; Li et al., 2001). To confirm the partitioning of different htt epitopes in the shell and core of htt body, double labeling was performed with antibodies mAb2166 and EM48. The results revealed that htt epitopes were segregated within the same htt body (Fig. 1C). The outermost region of the htt body was labeled by antibody 2166 (Fig. 1C, merged, red), and the core was labeled by EM48 (Fig. 1C, merged, green). An intermediate region was labeled by both antisera (Fig. 1C, merged, yellow).
Electron microscopy of htt bodies
Htt bodies were examined at the EM level in MCF-7 cells 3 d after transfection of FH969-18 or FH969-100 (Fig. 2) and immunoperoxidase labeling with Flag antibody. Consistent with the results from immunofluorescence with Flag antibody, htt bodies at the ultrastructural level exhibited Flag immunoperoxidase product heavily along the shell and weaker in the core (Fig. 2a,b). Cells with numerous htt bodies were nearly devoid of organelles (mutant htt) (Fig. 2a). The nuclei of cells enriched with htt bodies contained small foci of Flag labeling but otherwise showed a relatively normal appearance (Fig. 2a).
The core of most htt bodies (both wild type and mutant) contained a mixture of globules, and multivesicular autophagosomes (Fig. 2b,c) (wild type not shown). An additional structure clearly discernable only in htt bodies formed by mutant htt was tortuous fibrils (Fig. 2f,g). Radiating fibrils were prevalent in some of the mutant htt bodies (Fig. 2f,g). The mutant htt fibrils had an average diameter of ∼11.6 nm and a length of up to 200-250 nm.
The Flag labeling in the shell was associated with globular structures formed with both wt and mutant htt that ranged from ∼14 to 34 nm in diameter (Fig. 2h). The globules appeared singly, in random clusters, or assembled in a bead-like array that radiated from the core of the htt body (Fig. 2e,h). The globular assemblies in the shell were the same size and density in htt bodies whether formed by wt (data not shown) or mutant htt. Autophagosomes and mitochondria abutted the htt bodies (Fig. 2d,e). The membranes of these organelles appeared to fuse with the Flag-labeled globules in the shell (Fig. 1b,e). The EM results supported the findings from immunofluorescence microscopy and showed that the core and shell of htt bodies had different ultrastructures. Also, htt bodies sequestered organelles.
The core is protease resistant, and the shell is protease sensitive
Expression of the central portion of α-synuclein produces filamentous aggregates. The aggregates are resistant to proteinase K degradation; resistance to degradation serves as a sign of α-synuclein insolubility (Miake et al., 2002). We tested htt bodies for resistance to proteinase K. After 72 hr of expression of FH969-100, MCF-7 cells were fixed and treated with different concentrations of proteinase K and then double labeled using antisera mAb2166 and EM48. Proteinase K treatment (0.5 μg/ml) for 15-45 min substantially reduced diffuse htt labeling normally observed in the cytoplasm of transfected cells with either antibody. The shell in many htt bodies was also lost after proteinase K treatment, whereas the core was preserved (Fig. 3) and frequently enhanced by protease treatment. Densitometry of EM48-labeled htt bodies showed that the mean intensity of the core was significantly increased by treatment for 30 min with proteinase K (0.5 μg/ml) (p < 0.0001; Student's t test; n = 22). These results suggested that protease-sensitive and protease-resistant forms of mutant htt were present in the shell and the core of htt bodies, respectively. The core has a compact structure in which the EM48 epitope is partly hidden.
The formation of the shell domain depends on the length of htt
We looked for critical regions of htt required to form htt bodies, by using cDNAs encoding different lengths of htt. Full-length (1-3144 aa) wt and mutant htt expressed in MCF-7 cells formed fewer htt bodies compared with htt 1-969 and produced more diffuse htt or htt patches in the cytoplasm surrounding htt bodies (Fig. 4). These htt patches are not htt aggregates as formed by smaller N-htt. Expression of FH493-18 produced few cells with htt bodies (data not shown); cells expressing FH400-18 or FH400-100 (data not shown) and FH287-18 or FH287-100 formed no htt bodies (Fig. 4). Perinuclear aggregates (with no shell) appeared in some cells transfected with FH287-100, and labeling of htt occurred in the cytoplasm, nucleus, and plasma membrane of cells transfected with FH287-18Q (Fig. 4). These results suggested that in vitro assembly of htt bodies from wt or mutant htt required sequence of htt larger than or equal to aa 493.
Htt bodies alter the distribution of vesicle-associated proteins
Wild-type or mutant htt codistributes with clathrin-coated membranes in fibroblasts (Velier et al., 1998). These findings support a role for wt htt in endocytosis and suggest that mutant htt could disrupt endocytic function. Dynamin functions in endocytosis by regulating the assembly of clathrin-coated vesicles (Sever, 2002). Endogenous dynamin immunoreactivity occurred at plasma membranes and intracellular vesicles in MCF-7 cells (Fig. 3B) and clonal striatal cells (results not shown). In cells with htt bodies, dynamin labeling was markedly reduced at intracellular and plasma membranes (Fig. 3B). Some dynamin staining occurred at htt bodies and the nuclear membrane (Fig. 3B). HIP1 is an actin-binding protein involved in endocytosis (Engqvist-Goldstein et al., 1999; Metzler et al., 2001; Waelter et al., 2001). HIP1 localizes to plasma membranes (Metzler et al., 2001; Waelter et al., 2001). In cells with htt bodies, the localization of endogenous HIP1 was substantially reduced at plasma membranes and partly redistributed to htt bodies. SH3GL3 and FIP2 (a tumor necrosis factor-α-inducible coiled-coil protein) also function in membrane transport and interact with htt (Sittler et al., 1998; Hattula and Peränen 2000; So et al., 2000; Sever, 2002). In cells with htt bodies, labeling for these proteins was reduced at normal membrane locations and occurred in the shell of htt bodies (Fig. 5). SH3GL3 and FIP2 showed greater localization to htt bodies than dynamin and HIP1. DLP1 is involved in maintaining the localization and function of mitochondria (Pitts et al., 1999) and normally displays a punctate distribution throughout the cytoplasm. In cells with htt bodies, DLP1 immunoreactive structures were relocated near htt bodies but did not colocalize with htt bodies (Fig. 5). Other proteins that were partly redistributed to the shell of htt bodies included active caspase-3, which was found in ∼50% of cells with htt bodies, and proteasome and HSP70, which were present in all htt bodies (Table 1).
We questioned whether the dynamin immunoreactivity associated with htt bodies was affected by treatment with proteinase K. Double-label immunofluorescence was performed in cells transfected with FH969-100Q and treated for 15 to 30 min with 0.125-0.5 μg/ml proteinase K treatment. Antibody EM48 was used to label the core of htt bodies. Proteinase K treatment reduced, in a dose-dependent manner, dynamin immunoreactivity within the cytoplasm. Dynamin labeling associated with htt bodies was reduced by proteinase K treatment for 15 min at the lowest concentrations (Fig. 3B). In contrast, the core of htt bodies, as detected with EM48 antibody, was highly resistant to proteinase K treatment with all treatment conditions. These results suggested that the dynamin near htt bodies was protease sensitive.
The core had different and fewer detectable immunoreactive proteins than the shell. As reported previously in clonal striatal cells (Kegel et al., 2000), the lysosomal enzyme cathepsin D was concentrated in the core of htt bodies (Table 1) and accumulated to a greater extent when htt bodies were formed by mutant htt than wt htt (FH969-100 vs FH969-18). Ubiquitin and HSP40 also appeared in the core (Table 1); levels of these two proteins were not different for htt bodies formed by wt and mutant htt.
HSP60 localizes to mitochondria and was not found in htt bodies (Table 1). In cells with htt bodies, HSP60-positive mitochondria were removed from their normal location throughout the cytoplasm and concentrated around and between htt bodies.
Association between formation of htt bodies and diminished transferrin uptake
As a functional test of membrane transport, we measured internalization of transferrin by the endocytic pathway. Approximately 95% of untransfected MCF-7 cells internalized and transported rhodamine-labeled transferrin within 30 min. The fluorescence was punctate and dispersed throughout the cytoplasm (Fig. 6A, top). Within the same culture dish, there was no difference in the intracellular distribution of labeled transferrin among transfected cells expressing wt or mutant Flag-htt diffusely in the cytoplasm and cells without apparent Flag-htt expression (Fig. 6A). A significantly greater proportion (∼30%) of cells with htt bodies had little or no transferrin labeling compared with cells with no Flag-htt expression or cells with only diffuse Flag-htt labeling (∼38%; p < 0.05; Student's t test; n = 4) (Fig. 6A, B). There was no difference in the extent of transferrin incorporation between cells with htt bodies formed by wt htt and mutant htt.
Polyglutamine and polyproline regions determine the formation of htt bodies and redistribution of cytoplasmic proteins
We examined the effects of polyglutamine expansion on the frequency of htt bodies in MCF-7 cells. Results showed that the number of cells with htt bodies was significantly greater after expression of FH969-100 than FH969-18 (p < 0.001; n = 6; Student's t test) (Fig. 7A). The percentage of cells with htt bodies increased with time after transfection of FH969-100. Seventy-two hours after transfection, >40% of cells transfected with FH969-18 or FH969-100 had htt bodies. At all time points examined (6-72 hr), cells expressing mutant htt with 100Q had more htt bodies (ANOVA; F = 38.93; p < 0.05) (Fig. 7A; bar figure not shown). Expression of FH969-46 produced more htt bodies than expression of FH969-18, but the difference between mutant and wt htt was not statistically significant.
Proline-rich regions play an important role in signaling pathways that regulate membrane trafficking associated with autophagy and endosomal/lysosomal functions (Klionsky, 1999). The polyproline region in htt is required for some protein interactions including interaction with SH3GL3 (Liu et al., 1997; Sittler et al., 1998). Therefore, we speculated that removing the proline region in htt might affect the formation of htt bodies or the redistribution of cytoplasmic proteins to htt bodies. To test these possibilities, FH969-18 proline- and FH969-46 proline- were expressed in clonal striatal cells (x57) or MCF-7 cells. Subcellular fractions of cell lysates were examined by SDS-PAGE and Western blot. Removal of the proline-rich region in htt from aa 41-89 reduced the size of the Flag-htt fusion protein seen on Western blot, as expected. Polyproline deletion did not change levels of intact htt expressed in different subcellular fractions (Fig. 7B), except for a small decrease in the P1 fraction, which contains nuclei and nuclear and perinuclear membranes. The number of cells that formed htt bodies was greater in clonal striatal cells than in MCF-7 cells (Fig. 7C). In both cell lines, polyproline removal in htt significantly reduced the number of cells with htt bodies. The number of cells with htt bodies was reduced by ∼80% in cells expressing FH969-18 proline- and by 40% in cells expressing FH969-46 proline- (46Q; p < 0.05; n = 6; Student's t test) (Fig. 7C). The results suggested that the proline-rich region in htt is involved in forming htt bodies. In these studies, 46 polyglutamines were used for mutant htt because of technical difficulties in making a deletion of the proline-rich region with 100 glutamines.
Deletion of the proline-rich domain in htt changed the architecture of htt bodies. Polyproline removal abolished the partitioning of htt epitopes into a shell and core that was normally seen with antibodies mAb2166 and EM48, respectively (Fig. 7D). Instead, these anti-htt antisera produced immunofluorescent labeling that largely codistributed in patches. In some cells, the patches were smaller and more dispersed than htt bodies (Fig. 7D).
Removing the polyproline region reduced or eliminated labeling for HIP1, SH3GL3, dynamin, and proteasome in the shell of htt bodies (Fig. 8). In contrast, the colocalization of HSP40, HSP70, and ubiquitin with Flag was still found in the htt-labeled patches that formed after deletion of the proline-rich region (data not shown). These results implicate the polyproline region in htt in interactions with proteins involved in vesicle trafficking. The presence of the polyproline region facilitates sequestration of these proteins, which can lead to changes in membrane function.
Immunoreactive dynamin accumulates in cytoplasmic bodies in grade 1 HD cortical neurons
In cells with htt bodies, dynamin was absent from intracellular and plasma membranes and occurred near htt bodies and over the nucleus. We used the immunoperoxidase method to look for dynamin labeling in the cortex of control and HD brains. In control brain, most pyramidal neurons displayed low levels of diffuse cytoplasmic labeling for dynamin (Fig. 9a). In HD grade 1 (Fig. 9b,d,e) and grade 3 (Fig. 9c) cortex, the majority of neurons showed a marked increase in dynamin staining, especially in perinuclear regions where the immunoreactivity was concentrated in spherical bodies. Increasing the contrast in the digital images helped to visualize the dynamin-labeled bodies (Fig. 9d,e) and, in some cases, revealed more labeling along the periphery than in the center of these structures (Fig. 9d). The localization of dynamin in HD neurons is reminiscent of dynamin labeling in the shell of mutant htt bodies in vitro. Other sections were labeled with antibody EM48, which preferentially detects mutant htt aggregates in the HD brain (Gutekunst et al., 1999). There was no staining in control cortex with EM 48 antibody (results not shown). In the grade 1 HD brains, some cortical neurons displayed intense labeling for mutant htt in the nucleus and in perinuclear aggregates (Fig. 9f), which had a size comparable with htt bodies observed in vitro. These results suggest that, in early stages of HD, dynamin redistributes to mutant htt bodies in cortical neurons.
N-terminal fragments of mutant htt expressed from exon 1 (1-89 aa of wt) accumulate in vitro and in mice and form insoluble aggregates in nuclear and perinuclear regions of neurons (Davies et al., 1997; Scherzinger et al., 1997). The aggregates have a fibrillar, granular morphology similar to those seen in the HD brain (DiFiglia et al., 1997). The role of aggregates in HD pathogenesis is unclear; their presence in some HD cell and animal models inconsistently correlates with cell death (Saudou et al., 1998; Kuemmerle et al., 1999). A nonfibrillar soluble form of mutant htt could engage in abnormal protein interactions, thereby contributing to HD pathology (Passani et al., 2000; Gervais et al., 2002). In this study, we sought evidence of abnormal interactions by examining the htt body, which forms in cultured cells ectopically expressing a large N-terminal fragment of wt or mutant htt. Polyglutamine expansion increased the formation of htt bodies. The htt bodies develop from the fusion of htt-enriched autophagosomes (Kegel et al., 2000). We found that a protease-sensitive htt accumulated in globular assemblies at the periphery of htt bodies and colocalized with cytoplasmic proteins involved in vesicle trafficking. The presence of htt bodies hindered cellular endocytic uptake of transferin. In addition, numerous multivesicular membranes, autophagosomes, and mitochondria were incorporated into or adherent to the surface of htt bodies. The mitochondria may be damaged by htt and targeted for removal by autophagosomes. Recently, Panov et al. (2002) proposed that direct association of mutant htt with mitochondrial membranes caused calcium dysregulation in HD mitochondria. Altogether, these findings raise the possibility that an excess of soluble mutant htt that accumulates in the cytoplasm of neurons may contribute to HD pathogenesis.
The depletion of dynamin, HIP1, FIP2, and SH3GL3 in cells with htt bodies supports the possibility that there is altered function of vesicle movement in HD pathogenesis (DiFiglia et al., 1995). These proteins contribute to the assembly of large protein complexes that are involved in regulating vesicle trafficking and membrane flow (Sittler et al., 1998; Wanker et al., 1999; Hattula and Peränen, 2000; Sever, 2002). The displacement of these proteins from their normal site by htt bodies is likely to disturb functions associated with vesicle trafficking and endocytosis. We found that the uptake or distribution of transferrin in many cells with htt bodies was severely impaired, evidence that membrane transport is affected. Dynamin, which functions in clathrin-mediated endocytosis (Sever, 2002), is redistributed to perinuclear regions in HD cortical neurons of grade 1/presymptomatic HD patients. Grade 1 HD neurons also show build up of mutant htt in cytoplasmic bodies (present study; Gutekunst et al., 1999). Modregger et al. (2002) showed that PASCIN1, a protein involved in synaptic vesicle recycling, interacts with htt and is depleted from HD neuronal processes and accrues in perinuclear locations with mutant htt in presymptomatic HD brains. These findings suggest that dysfunction in membrane transport may occur at early stages of HD.
Cytoplasmic protein redistribution to htt bodies was not changed by polyglutamine expansion from 18 to 46 or 100 glutamines. This suggests that the polyglutamine region is not sufficient to regulate the sequestration of soluble proteins into htt bodies. Removal of the polyproline region in htt reduced htt bodies and their capacity to recruit HIP1, SH3GL3, FIP2, and dynamin. Our findings support an important role of the polyproline region in normal htt protein interactions. In our study, the lack of proline-rich region in wt htt reduced htt bodies much more than proline deletion in mutant htt. It may be that polyglutamine expansion interferes with the regulation or stability conferred by the proline region in htt. The polyproline region is also critical for aggregation of mutant htt. Khoshnan et al. (2002) found that an intrabody directed to the proline region in htt reduced mutant htt aggregates in vitro. Although not studied here, the HEAT (for huntingtin, elongation factor 3, regulatory A subunit of protein phosphatase 2A and TOR1) repeat sequences in the N terminus of htt may also be important in targeting htt to autophagosomes and forming htt bodies. These tandem repeats of specific amino acid residues occur in other proteins that have some role in membrane transport (Andrade and Bork, 1995; Takano and Gusella, 2002). For example, HEAT motifs have been found to mediate the membrane localization of Tor2p, a component of a signaling pathway that regulates cell growth and autophagy in yeast (Kunz et al., 2000).
Our results show that polyglutamine expansion in htt is not required for htt to sequester cytoplasmic proteins away from their normal location in the cytoplasm. High levels of wt htt are capable of recruiting cytoplasmic proteins. In other studies, we also found that N-terminal fragments of wt htt 1-287 or 1-400 form nuclear inclusions, as does mutant htt in vitro (Y. J. Kim and M. DiFiglia, unpublished observations). Transient transfection of htt in vitro, as used in the present study, is conducive to observe these characteristics of wt htt. There is precedence for toxicity of wt proteins with polyglutamine tracts. Expression of high levels of full-length wt ataxin 1, another polyglutamine-containing protein, caused neurodegeneration in a Drosphila model of spinocerebellar ataxia (Fernandez-Funez et al., 2000). Our study showed that polyglutamine expansion increased the number of cells forming htt bodies. However, relocation of interacting proteins, including HIP1, to htt bodies was polyproline dependent and polyglutamine independent. Gervais et al. (2002) showed that polyglutamine expansion in htt reduced the interaction between htt and HIP1, which increased the availability of HIP1. An elevation of HIP1 was toxic to cells in vitro, suggesting a mechanism for HD pathogenesis. The results from both studies suggest that two mechanisms, a polyproline-dependent one and a polyglutamine dependent one, differentially modulate the levels of HIP1 in HD neurons.
Mutant htt bodies have an amalgam of morphological features characteristic of cytoplasmic aggregates (fibrillar structure) and autophagosomes (multivesicular bodies and cathepsin D immunoreactivity). The size of the fibrils in mutant htt bodies (11.6 nm) is similar to the size of the fibrils identified in aggregates formed by HD exon 1 and other aggregate-prone proteins (Scherzinger et al., 1997; Bevivino and Loll, 2001; Perutz et al., 2002; Poirer et al., 2002). The presence of fibrillar structure characteristic of aggregates found in HD autopsy brains indicates that the formation of htt bodies does not depend on overexpressing htt, although it may require accumulation. Polyglutamine expansion increased the frequency of cells that formed htt bodies (present study) and caused more cathepsin D to accumulate in the core (Kegel et al., 2000). Cathepsin D is transported to maturing autophagosomes. The multivesicular bodies within the core of the htt body may be the sites of cathepsin D accumulation. The htt bodies are also regulated by autophagy. Treating cells with the autophagy inhibitor 3-methyladenine significantly reduced the frequency (Qin et al., 2003).
The formation of the shell of the htt body minimally requires an N-terminal region of htt that terminates between aa 493 and 969. N-htt fragments in this size range are present in protein extracts from the HD brain and may be partly formed from cleavage by caspases and calpains (Kim et al., 2001; Gafni and Ellerby, 2002; Wellington et al., 2002). HIP1, which was redistributed to the shell, binds to a region in htt between aa 222-480 (Michael R. Hayden, personal communication) (i.e., within the size of htt required to form htt bodies. In contrast, the core of the htt body may contain N-htt fragments that are smaller than those in the shell. Only N-terminal anti-htt antibodies Ab1 and EM48 reacted with the core.
Polyglutamine expansion in htt increases the probability that partially unfolded intermediate forms will self-associate as oligomers and evolve into a fibrillar structure (Perutz et al., 2002). The protofibril has been identified as a stable soluble prefibrillar intermediate of many aggregate prone proteins, including mutant htt (Perutz et al., 2002). Poirer et al. (2002) used transmission EM and atomic force microscopy to examine purified HD exon 1. They identified small globules of ∼4-5 nm as the oligomers of mutant htt that form protofibrils, a structure they compared in appearance to “beads on a string.” Two protofibrils were predicted to form the mature fibril. The globules we found in the shell of htt bodies in cells also assembled like beads on a string but were significantly larger (∼14-34 nm) than those identified by Poirer et al. (2002). Recently, Bitan et al. (2003) reported that β-amyloid forms large beaded super structures of 20-50 nm in vitro in addition to the 10 nm diameter protofibrils. The large beaded globules were identified by them as higher-order oligomers (Bitan et al., 2003). It was not clear from their studies whether the large super structures were involved in the formation of fibrils. We speculate that the 14-34 nm spherical bodies prevalent in the shell of htt bodies are oligomers of htt. The native full-length htt has been predicted to exist as a dimer (Bao et al., 1996); oligomers of higher order have not been reported. We cannot rule out from our analysis that htt monomers also occur in htt bodies. Biochemical analysis of htt bodies isolated from cultured cells will be necessary to determine the predominant forms of htt in htt bodies.
Wild-type and mutant htt bodies contained proteins involved in protein folding. Ubiquitin and HSP40 were present in the core and proteasome, HSP40, and HSP70 accumulated in the shell. These observations suggest that the core and shell of htt bodies contain misfolded species of htt targeted for removal or degradation. It was surprising that ubiquitin and proteasome were segregated in the core and shell, respectively, because these proteins work in concert to degrade misfolded proteins and would be expected to colocalize. The ubiquitin may be associated with htt in multivesicular bodies in the core of htt bodies. Targeting of some proteins to multivesicular bodies requires ubiquitination (Katzmann et al., 2002). An intact polyproline region in htt was required for recruiting proteasome, indicating a possible direct interaction between proteasome and the polyproline region of htt.
In summary, this study finds that exogenous expression of large truncated wt and mutant htt formed htt bodies that had structurally distinct shell and core regions. A soluble multioligomeric htt assembled in the shell and colocalized with other proteins involved in vesicle trafficking that were redistributed away from their normal site of function. The presence of htt bodies depleted and displaced organelles and impaired endocytic pathways. Because mutant htt forms more htt bodies than wt htt, these effects are more severe in the cells expressing mutant htt. We propose that, in HD neurons, a soluble oligomeric assembly of mutant htt similar to that found in the shell of the htt body contributes to HD pathogenesis by disrupting membrane trafficking.
This work was supported by National Institutes of Health Grants NS16367 (M.D.), NS13579 (M.D.), and NS38194 (N.A.) and by a grant from the Huntington's Disease Society of America (M.D.). We thank Dr. Alfred Heller from the University of Chicago (Chicago, IL) for providing us with x57 cells, Dr. Xiao-Jiang Li from Emory University (Atlanta, GA), and Dr. Stephen Hersch from Massachusetts General Hospital (Charlestown, MA) for providing us with EM48 antibody.
Correspondence should be addressed to Dr. Zheng-Hong Qin, Laboratory of Cellular Neurobiology, Massachusetts General Hospital, 114 16th Street, Room 2125, Charlestown, MA 02129-4404. E-mail:.
Copyright © 2004 Society for Neuroscience 0270-6474/04/240269-13$15.00/0