F-Actin Binding Regions on the Androgen Receptor and Huntingtin Increase Aggregation and Alter Aggregate Characteristics

Protein aggregation is associated with neurodegeneration. Polyglutamine expansion diseases such as spinobulbar muscular atrophy and Huntington disease feature proteins that are destabilized by an expanded polyglutamine tract in their N-termini. It has previously been reported that intracellular aggregation of these target proteins, the androgen receptor (AR) and huntingtin (Htt), is modulated by actin-regulatory pathways. Sequences that flank the polyglutamine tract of AR and Htt might influence protein aggregation and toxicity through protein-protein interactions, but this has not been studied in detail. Here we have evaluated an N-terminal 127 amino acid fragment of AR and Htt exon 1. The first 50 amino acids of ARN127 and the first 14 amino acids of Htt exon 1 mediate binding to filamentous actin in vitro. Deletion of these actin-binding regions renders the polyglutamine-expanded forms of ARN127 and Htt exon 1 less aggregation-prone, and increases the SDS-solubility of aggregates that do form. These regions thus appear to alter the aggregation frequency and type of polyglutamine-induced aggregation. These findings highlight the importance of flanking sequences in determining the propensity of unstable proteins to misfold.


Introduction
Spinobulbar muscular atrophy (SBMA) and Huntington disease (HD) are devastating neurodegenerative diseases. SBMA is caused by an expanded CAG trinucleotide repeat that encodes a long polyglutamine tract in the androgen receptor (AR) [1], while HD is caused by an enlarged polyglutamine tract in the huntingtin (Htt) protein [2]. Proteolytic cleavage of AR and Htt appears to generate toxic, N-terminal fragments [3,4,5,6]. These are sufficient to recapitulate neurodegenerative phenotypes in vivo [7,8]. Nterminal fragments of expanded AR and Htt readily aggregate in vitro and in cell-culture models, thus making them useful in biochemical studies [9,10,11]. While the aggregation and toxicity of polyglutamine proteins directly correlate with the length of the polyglutamine tract [11], flanking sequences are also clearly important [12,13,14], as are intracellular signaling pathways that act via protein interactions or post-translational modifications [10,15,16]. Emerging evidence from a variety of studies of aggregation-prone proteins associated with neurodegenerative diseases suggests that there is considerable diversity among aggregates that can be formed in vitro [17,18,19], and, moreover, that some protein aggregates are likely to be more toxic than others [19,20]. Thus, protein interactions that alter aggregate conformation could play an important role in determining toxicity.
Indirect evidence implicates actin and/or actin-binding factors as an influence on polyglutamine-dependent aggregation of AR and Htt [10,21,22,23,24,25,26]. Y-27632, a rho-kinase (ROCK) inhibitor, reduces intracellular polyglutamine aggregation of Htt exon 1 and the N-terminal fragment of AR, termed ARN127 [10,25]. Y-27632 also attenuates Htt toxicity in Drosophila and improves motor function in mice [10,27]. Y-27632 blocks phosphorylation of profilin, an actin-binding protein that directly binds Htt, but not AR [28,29]. Profilin strongly inhibits aggregation of ARN127 and Htt exon 1 in cells [29], and decreases polyglutamine-mediated toxicity in Drosophila [22]. The anti-aggregation effects of profilin depend on both its polyproline binding activity (required to bind Htt), and its ability to bind Gactin (required to suppress both Htt and AR aggregation) [29]. In this study, we have identified regions of ARN127 and Htt exon 1that bind filamentous actin (F-actin) in vitro, and investigate the effect of these regions on polyglutamine-dependent aggregation.

ARN127 and Htt Exon 1 Bind F-Actin In Vitro
To test for a direct interaction between AR, Htt, and F-actin in vitro, we used an F-actin co-sedimentation assay with recombinant GST-ARN127 or GST-Htt exon 1 containing 25 glutamine repeats (Fig. 1A). Coomassie staining confirmed protein purity (Fig. S1A,E). Protein preparations were precleared by ultracentrifugation to remove any pre-existing aggregates. 0.5 mM GST-ARN127 (25) or 0.25 mM GST-Htt exon 1 (25) was incubated with F-actin (4 mM) that had been pre-polymerized in vitro for 1 hour at 25uC. As a control, proteins were incubated with an equal concentration of bovine serum albumin (BSA) instead of F-actin. After ultracentrifugation (100,000 x g), supernatant and pellet fractions were analyzed by western blot using antibody to GST. Western blot analysis was used rather than Coomassie stain because both proteins are very similar in size to G-actin (43 kDa). F-actin localized to the pellet fraction in all cases, as visualized by Coomassie stain (Fig. 1B). Both GST-ARN127(25) and GST-Htt exon 1(25) co-sedimented with F-actin while remaining soluble in its absence (Fig. 1B). GST alone did not co-sediment with F-actin (Fig. 1B). ARN127 also bound F-actin following cleavage of the GST-tag (Fig. S1B). We determined the binding affinity of GST-ARN127(25) and GST-Htt exon 1(25) for F-actin by using a constant amount of protein (16 nM for ARN127, 150 nM for Htt exon 1) mixed with increasing amounts of F-actin (Fig. 1C, D). GST-ARN127(25) bound to F-actin with an approximate dissociation constant (Kd) of 1 mM (Fig. 1C), while for GST-Htt exon1(25) the Kd was approximately 2 mM (Fig. 1D), comparable to other actin-binding proteins [30].

Sensitivity to Actin-Regulatory Pathways Requires the First 50 Amino Acids of ARN127
We have previously found that Y-27632, as well as a downstream ROCK target, profilin, inhibit ARN127 and Htt exon 1 aggregation [10,25,29]. Since both Y-27632 and profilin regulate actin assembly [31,32], we tested whether deletion of the actin-binding region of ARN127 would alter its response to their inhibitory activities. We employed fluorescence resonance energy transfer (FRET) to quantify these effects [10]. We were only able to test AR using this assay, since deletion of the N14 region of Htt prevented significant FRET, possibly due to altered orientation of the CFP and YFP moieties. Expanded ARN127 and ARQC with 65 glutamines were tagged with fluorescence donor or acceptor tags, CFP or YFP, and co-transfected into HEK293 cells as previously described [10]. Y-27632 decreased ARN127(65)-CFP/ YFP aggregation dose-dependently (Fig. 3A). Intriguingly, it increased the aggregation of ARQC(65)-CFP/YFP (Fig. 3A). Similarly, profilin 1 dose-dependently reduced ARN127(65)-CFP/ YFP aggregation, but was much less effective on ARQC(65)-CFP/ YFP aggregation (Fig. 3B). Thus, the N50 region of AR mediates effects of these actin regulators on polyglutamine-mediated aggregation.

N50 of ARN127 Affects Inclusion Type, Number, and Distribution
To better characterize the behavior of expanded AR and Htt peptides within the cell, and to determine the influence of the actin-binding regions, we transfected expanded forms of each construct into C17.2 neural precursor cells, which were used for their ease of imaging. Identical results were obtained with HEK293 cells (data not shown). We observed clear differences between inclusions formed by ARN127 (65)

N50 of ARN127 and N14 of Htt Exon 1 Promote SDS-Insoluble Aggregates
The distinct effects of actin-binding regions on AR and Htt inclusion formation suggested that they might also change aggregate characteristics. Thus, we transfected HEK293 cells with the various constructs as above. 24 h post-transfection the cells were disrupted via syringe lysis and fractionated by centrifugation (15,000 x g) to compare soluble vs. insoluble species. Supernatant and pellet fractions were resuspended in 2% SDS and resolved by SDS-PAGE and western blot. All protein in the supernatant fraction was completely solubilized by SDS (Fig. 6A). In contrast, both ARN127(65)-YFP and ARQC(65)-YFP peptides that partitioned into the insoluble pellet contained mixtures of SDS-soluble  . Immunofluorescence of HA-tagged Htt is in green, F-actin is stained with rhodamine-phalloidin (red) and DNA is stained with DAPI (blue). We did not observe significant differences in patterns of inclusion formation between the two Htt constructs. doi:10.1371/journal.pone.0009053.g004 and insoluble proteins (Fig. 6A,B). The aggregates of ARQC(65)-YFP were much more readily dissociated in 2% SDS compared to ARN127(65)-YFP (Fig. 6A,B). Parallel experiments in C17.2 cells revealed similar results (Fig. 6A). We observed similar phenomena for Htt exon 1(97)-H4 vs. HttQC(97)-H4 (Fig. 6C). The pellet fraction containing HttQC(97)-H4 was significantly more SDS-  (Fig. 6C,D). Thus flanking sequences of AR and Htt influence both the propensity for protein misfolding and the biochemical characteristics of the aggregates that result.

Discussion
In this study, we have focused on aggregation-prone fragments of AR (ARN127) and Htt (Htt exon 1), identifying amino acids 1-50 (N50) of ARN127 and 1-14 (N14) of Htt exon1 as regions that influence polyglutamine-dependent aggregation. These regions were originally identified because they mediate binding to F-actin in vitro. However, we did not observe co-localization of ARN127 or Htt exon 1 with the F-actin cytoskeleton (data not shown). This could reflect lack of significant binding, or that intracellular binding to F-actin is transient. Deletion of the N50 region of ARN127 or the N14 region of Htt decreased the total number of inclusions formed, and the aggregates that did form were more SDS-soluble. Further, the N50 region of AR regulates the characteristics of inclusions formed and mediates its responsiveness to the anti-aggregation effects of Y-27632 and profilin. Thus, the flanking sequence of the polyglutamine region does not simply alter the propensity to form a polyglutamine aggregate, but can directly affect its aggregation rate and responses, the subcellular localization of the inclusions, and the biochemical characteristics of the aggregates.
The importance of flanking sequences on Htt exon 1 has previously been demonstrated: flanking sequences have been shown to directly regulate aggregation, toxicity, and the morphology of inclusions [13,34,35]. Given the importance of primary amino acid sequence on polyglutamine aggregation in vitro [12,14], we cannot exclude that the intracellular aggregation differences observed are solely due to altered intrinsic aggregation kinetics. However, this cannot explain all observable differences noted, such as altered distribution of polyglutamine inclusions. We have found that peptides containing either the first 50 amino acids of AR or the first 14 amino acids of Htt partition to Tritoninsoluble cell fractions independent of their inherent solubility (Figs. S2, S3), suggesting that these regions could mediate protein interactions with macromolecular structures in the cytoplasm. Indeed, the first 17 amino acids of Htt have previously been implicated as a pro-aggregation-domain, a cytoplasm-targeting domain, and have been shown to bind mitochondria, the endoplasmic reticulum, and the Golgi apparatus [35,36]. This region is highly conserved across diverse species. Taken together, these studies imply that this region is likely to mediate protein interactions that play an important role in determining aggregation properties of the Htt peptide.
Although we did not detect binding of AR or Htt to F-actin in cells, the in vitro binding of the Htt N-terminus to F-actin is intriguing in light of Htt's capacity also to bind the actin remodeling factor profilin. Profilin/Htt interaction is most likely mediated by Htt's polyproline domain, immediately distal to the polyglutamine region in the Htt peptide [28,29]. Analysis of the first 17 amino acids of Htt reveal a striking similarity to another actin-binding peptide, Lifeact [37]. These observations may indicate a normal, perhaps transient, function for Htt in the regulation of the actin cytoskeleton, although this remains to be tested.
In this study, we find that sequences independent of the polyglutamine tract have profound effects on subcellular localization, detergent solubility, and inclusion formation of polyglutamine peptides. Given the apparent importance of these regions for protein interaction, this implies that modifiers of protein aggregation are likely to be found within the set of AR and Htt interacting proteins. Indeed, the recent finding that Htt-interacting proteins are enriched for genetic modifiers of toxicity is consistent with this idea [38]. Further study of AR and Htt binding proteins thus may reveal mechanisms that specify particular aggregation pathways and pathogenic features of each disease.

Constructs
Bacterial expression vectors for ARN127 Htt exon 1 and various derivatives were constructed via PCR amplification and subcloned in the pGEX4T1 backbone (Amersham Biosciences). Similarly, for mammalian expression vectors, ARQC and HttQC plasmids were constructed via PCR amplification and subcloned into pECFP-N1, pEYFP-N1 (Clontech), or pcDNA.3 backbones. GST-Htt exon 1 constructs were originally obtained from Paul Muchowski and mammalian pcDNA3-Htt-H4 constructs were originally obtained from Joan Steffan.

Cell Culture and Transfection
HEK293 cells were plated at 250,000 cells per well in a 24-well dish and transfected with .3 mg total DNA with Lipofectamine and Plus reagent (Invitrogen) according to the manufacturer's instructions. C-17.2 cells were plated at 100,000 cells per 24-well and transfected with .6 mg DNA with Lipofectamine 2000 (Invitrogen) according to the manufacturers instructions. Cells were harvested at indicated times.

FRET
All FRET measurements were carried out 48 hours after transfection of HEK293 cells, read in 96-well cell-culture plates by a fluorescence plate reader (Tecan). HEK293 cells were transfected with .075 mg total ARN127CFP/YFP DNA in a 1:3 donor:acceptor ratio. Profilin DNA was co-transfected with AR constructs at concentrations of .075 mg, .15 mg, or .225 mg. pcDNA3 backbone vector was also co-transfected for a constant final concentration of .3 mg. Y-27632 was added at the indicated concentration for 24 h prior to FRET measurements [29].

Confocal Microscopy
All images were acquired on a C1sl confocal microscope (Nikon Instruments Inc.).

Inclusion Counting
At least 100 HEK293 cells per transfection (3) were examined for total number of inclusions formed after 24 h. Approximately 100 C17.2 cells per transfection (3) were counted for inclusions. Cells were identified as having nuclear, cytoplasmic, or multiple inclusions, or a combination.

Protein Purification
GST-ARN127 plasmids were grown in E. Coli Rosetta 2 (DE3) competent cells (Novagen). Protein expression was induced with 1 mM (isopropyl b-d-thiogalactoside) IPTG for 3 h at 37uC. GST-Htt plasmids were grown in E. Coli SURE competent cells (Stratagene). Protein expression was induced with 1 mM IPTG for 3H at 30uC. Bacterial pellets were resuspended in resuspension buffer (PBS, .05% Tween, 1 mM PMSF, protease inhibitor tablet (Roche)) and lysed by sonication and 1% Triton. GSTtagged proteins were precipitated with glutathione sepharose (Amersham Biosciences) and eluted with equal volumes of elution buffer (50 mM Tris-HCl pH. 8 m 10 mM reduced glutathione). For cleavage of the GST-tag, 37.5 units of thrombin protease (Amersham Biosciences) was used for .5 ml of GST-bound glutathione sepharose 4B (Amersham Sepharose) at 4uC overnight. Thrombin was removed from cleaved AR with benzamidine sepharose 6B (Amersham Biosciences). Protein concentration was quantified via Bradford assay and Coomassie staining via Image J.

F-Actin Co-Sedimentation
Non-muscle human actin (Cytoskeleton) was polymerized according to the manufacturer's instructions. 10 mg/ml of Gactin was polymerized with polymerization buffer (10X: 500 mM KCl, 20 mM MgCl2, 10 mM ATP) in general actin buffer (5 mM Tris-HCl pH8.2 mM CaCl2, .5 mM DTT, .2 mM ATP) for 1 h at RT. Equal molar ratios of unlabeled phalloidin (Molecular Probes) were added to stabilize filaments. Recombinant, purified GST-tagged proteins were precleared via ultracentrifugation (100,000 x g for 30 min at 4uC). Proteins were added to a prepolymerized F-actin (4 uM) or a BSA control for 1 h on ice. Mixtures were ultracentrifuged for 30 min at 100,000 x g at 4uC. Supernatants and pellets were subjected to SDS-PAGE. F-actin pellets were visualized via Coomassie stain. GST-proteins were probed via western blot using GST antibody (Santa Cruz Bioscience).

Detergent Fractionation
For unexpanded AR and Htt constructs, HEK293 cells were harvested 24 h post-transfection. Cell pellets were lysed in 130 ml cold lysis buffer (PBS, 1 % Triton, 5 mM EDTA, protease inhibitor cocktail (Roche)) and subjected to high-speed ultracentrifugation (100,000 x g). For expanded AR and Htt constructs, HEK293 cells were harvested 24 h later post-transfection. Cell pellets were lysed by syringe passage in cold resuspension buffer (PBS, 5 mM EDTA, protease inhibitor cocktail (Roche)) and subjected to table-top centrifugation (15,000 x g). Supernatant and pellets fractions were denatured in 2% SDS buffer, 25 mM DTT, and boiled for 10 min. For crude cell-lysates, cell pellets were directly lysed in 200 ml of 2% SDS, 25 mM DTT and boiled for 10 minutes. All lysates were subjected to SDS-PAGE and probed via western blot using GFP antibody (Santa Cruz Biotechnology) or HA antibody (Covance). Supernatants, pellets and higher molecular weight bands were quantified using Image J.

Antibodies
Anti-rabbit N-20 antibody (1:1000, Santa Cruz) was used for the detection of cleaved AR products. Anti-rabbit GFP-antibody (Santa Cruz) or anti-mouse HA-antibody (Covance) was used for the detection of YFP-tagged or HA-tagged proteins at 1:2000 dilution for soluble proteins or 1:1000 dilution for insoluble higher-molecular weights. Anti-mouse MW7 was used to detect the C-terminus of Htt at 1:2000 dilution for soluble proteins or 1:1000 dilution for insoluble higher-molecular weights.
Anti-mouse GST-antibody (1:2000, Santa Cruz) was used for the detection of GST-tagged proteins.

In Vitro Solubility
GST-ARN127(25), GST-ARNQ(32), GST-ARQC(36), GST-Htt exon 1 (25), and GST-HttQC(25) were recombinantly purified and quantified as described above. 1.25 mg/ml of GST-AR or .75 mg/ml of GST-Htt purified proteins were ultracentrifuged (100,000 x g) and incubated at 37uC for 1 h to promote misfolding. For cleavage of the GST tag, 1.0 mg/ml of precleared GST-AR peptide were incubated with 1 NIH unit of thrombin (Invitrogen) overnight at 4uC. Proteins were ultracentrifuged and supernatant and pellet fractions were resuspended in SDS sample buffer and subjected to SDS-PAGE. Proteins were analyzed via Coomassie stain. Figure S1 Purity of Recombinantly Purified Proteins. A, Coomassie stain of GST-ARN127(25), GST-ARN127(DQ), and GST-ARN127(52). Lower molecular weight (LMW) contaminants are indicated B, Cleaved ARN127(25) co-sediments with F-actin but is soluble in the absence of F-actin. GST-tagged ARN127 (25) was treated with thrombin to cleave off the GST tag. 1 mM of thrombin-cleaved ARN127(25) was mixed with 5 mM F-actin as in Figure 1. Western blot was used to detect the N-terminus of AR (N-20 antibody) or Coomassie for actin. C, Coomassie stain of GST and GST-AR fragments, AR1-57, AR50-127, and AR78-127, D, Cleaved AR1-57 cosediments with F-actin but is soluble in the absence of F-actin. GST-tagged AR1-57 was treated with thrombin to cleave off the GST tag. 1 mM of thrombin-cleaved AR1-57 was mixed with 5 mM F-actin. Western blot was used to detect the N-terminus of AR (N-20 antibody) or Coomassie stain for actin. E, Coomassie stain of GST-Htt exon 1(25), GST-Htt1-45, and GST-Htt15-92. Found at: doi:10.1371/journal.pone.0009053.s001 (5.47 MB TIF) Figure S2 N50 of ARN127 and N14 of Htt exon 1 mediate macromolecular interactions. A, Schematic of AR and Htt peptides fused to YFP. B, The N-terminus of AR mediates macromolecular interactions. HEK293 cells were transfected with YFP fusion proteins, lysed in 1% Triton, and subjected to ultracentrifugation (100,000 x g). Supernatant and pellet fractions were analyzed via western blot with YFP antibody. ARN127(25)-YFP (top band) and ARNQ(25)-YFP are present in the Triton-insoluble pellets of HEK293 cells, while ARQC(25)-YFP is not. Tubulin indicates loading control. C, Deletion of the N14 region of Htt exon 1 (25) does not affect Triton solubility. Supernatant and pellet fractions were analyzed via western blot with MW7 antibody (detects the C-terminus of Htt). Tubulin indicates loading control. D, The first 17aa of Htt alone fused to YFP cause it to become insoluble in cell lysates. Blots are probed with YFP. Tubulin indicates loading control. Found at: doi:10.1371/journal.pone.0009053.s002 (9.60 MB TIF) Figure S3 Inherent Solubility of GST peptides. A, There are no differences in the inherent solubility of AR peptides. GST-tagged ARN127 (25), ARNQ (32), and ARQC (36) are present in equal amounts in the pellet fractions after ultracentrifugation. B, Cleavage of AR peptides from GST does not unmask drastic differences in inherent solubility. GST was cleaved from ARN127(25), ARNQ (32), and ARQC (36) with thrombin protease at 4uC overnight. Peptides were ultracentrifuged (100,000 x g) and supernatant and pellet fractions were analyzed via SDS-PAGE and Coomassie stain.