Perturbation with Intrabodies Reveals That Calpain Cleavage Is Required for Degradation of Huntingtin Exon 1

Background Proteolytic processing of mutant huntingtin (mHtt), the protein that causes Huntington's disease (HD), is critical for mHtt toxicity and disease progression. mHtt contains several caspase and calpain cleavage sites that generate N-terminal fragments that are more toxic than full-length mHtt. Further processing is then required for the degradation of these fragments, which in turn, reduces toxicity. This unknown, secondary degradative process represents a promising therapeutic target for HD. Methodology/Principal Findings We have used intrabodies, intracellularly expressed antibody fragments, to gain insight into the mechanism of mutant huntingtin exon 1 (mHDx-1) clearance. Happ1, an intrabody recognizing the proline-rich region of mHDx-1, reduces the level of soluble mHDx-1 by increasing clearance. While proteasome and macroautophagy inhibitors reduce turnover of mHDx-1, Happ1 is still able to reduce mHDx-1 under these conditions, indicating Happ1-accelerated mHDx-1 clearance does not rely on these processes. In contrast, a calpain inhibitor or an inhibitor of lysosomal pH block Happ1-mediated acceleration of mHDx-1 clearance. These results suggest that mHDx-1 is cleaved by calpain, likely followed by lysosomal degradation and this process regulates the turnover rate of mHDx-1. Sequence analysis identifies amino acid (AA) 15 as a potential calpain cleavage site. Calpain cleavage of recombinant mHDx-1 in vitro yields fragments of sizes corresponding to this prediction. Moreover, when the site is blocked by binding of another intrabody, V L 12.3, turnover of soluble mHDx-1 in living cells is blocked. Conclusions/Significance These results indicate that calpain-mediated removal of the 15 N-terminal AAs is required for the degradation of mHDx-1, a finding that may have therapeutic implications.

We have used intrabodies, intracellularly expressed antibody fragments, to gain insight into the mechanism of mutant huntingtin exon 1 (mHDx-1) clearance. Happ1, an intrabody recognizing the proline-rich region of mHDx-1, reduces the level of soluble mHDx-1 by increasing clearance. While proteasome and macroautophagy inhibitors reduce turnover of mHDx-1, Happ1 is still able to reduce mHDx-1 under these conditions, indicating Happ1-accelerated mHDx-1 clearance does not rely on these processes. In contrast, a calpain inhibitor or an inhibitor of lysosomal pH block Happ1-mediated acceleration of mHDx-1 clearance. These results suggest that mHDx-1 is cleaved by calpain, likely followed by lysosomal degradation and this process regulates the turnover rate of mHDx-1. Sequence analysis identifies amino acid (AA) 15 as a potential calpain cleavage site. Calpain cleavage of recombinant mHDx-1 in vitro yields fragments of sizes corresponding to this prediction. Moreover, when the site is blocked by binding of another intrabody, V L 12.3, turnover of soluble mHDx-1 in living cells is blocked.

Huntington's disease (HD) is caused by the expansion of a polyglutamine (polyQ) tract in the first exon (HDx-1) of the large protein, huntingtin (Htt) [1] . Mutant Htt protein (mHtt) perturbs many cellular processes by both gain of toxic function and loss of normal function. These include axonal transport, mitochondrial metabolism, transcriptional regulation and the ubiquitin proteasome system (UPS) [2] . There is an age-dependent accumulation of mHtt protein in HD [3] , which may be partially responsible for the adult onset of symptoms despite the lifelong expression of mHtt. Increasing the clearance of mHtt could prevent this accumulation and thereby delay or prevent the onset of symptoms. Degradation of mHtt occurs through several mechanisms, suggesting a number of potential therapeutic opportunities for enhancing removal. Proteases cleave Htt, generating N-terminal fragments, some of which are more toxic than the full-length protein [4] , [5] , [6] . Increasing polyQ tract length leads to increased caspase and calpain activation and enhanced production of toxic N-terminal fragments in the HD brain [7] . These fragments are degraded by additional protease cleavage, the UPS and autophagy, which can involve isolation in an autophagosome and introduction to the lysosome by fusion, macroautophagy, or delivery to the lysosome by chaperone proteins (chaperone-mediated autophagy, CMA) [8] . Certain cleavage events generate toxic fragments, and selective prevention of these events dramatically reduces the toxicity of mHtt by the generation of other, less toxic N-terminal cleavage products [9] , [10] . Post-translational modifications such as phosphorylation also play a role in regulating Htt proteolysis [11] , [12] , and phosphorylated mHtt can be more toxic than unphosphorylated mHtt [12] . Thus, the dichotomy of mHtt processing: while some modifications increase the toxicity of the protein, these more toxic forms are intermediates in the process leading to total degradation. Since enhancing total degradation represents a powerful therapeutic strategy, a better understanding of this process is warranted. As the site of the disease-causing mutation, insight into the clearance of HDx-1 is particularly salient. We have used intrabodies (iAbs), intracellularly-expressed antibody fragments directed against various sites in HDx-1 to gain such insight. Intrabodies retain the high target specificity of antibodies but lack the immunogenic constant domains. These reagents have shown significant promise as therapeutics for proteinopathies including HD [13] . Moreover, iAbs are also powerful molecular tools for probing the functions and interactions of their targets when expressed in living cells. We have previously shown that binding of the iAb Happ1, which recognizes the proline rich region of HDx-1, results in a selective increase in the turnover of the mutant form (mHDx-1) [14] , [15] . Here we report on the mechanism of Happ1-induced turnover of mHDx-1, the study of which has revealed a new insight into mHtt cleavage.

HEK 293 cells were transfected with mHDx-1-GFP plus iAb (HDx-1:iAb = V L 12.3, 1∶1; Happ1, 1∶2). Thirty-six hours post-transfection, cells were collected for Western blotting and immunoprecipitation (IP) as previously described [14] . Briefly, cells were dislodged by pipetting, pelleted by centrifugation, rinsed with PBS, and lysed by sonication in lysis buffer. Insoluble material was removed by additional centrifugation, and the protein concentration was determined by BCA assay (Pierce). Htt protein was immunoprecipitated from the lysate by combining 400 µg lysate protein with 50 µg anti-GFP antibody (Invitrogen) conjugated to protein G sepharose beads (Sigma) and rocking for 4 hrs at RT. Beads were washed 4 times in PBS containing 0.1% Triton X100 to remove unbound protein. Seventy-five µg total lysate protein samples and bound IP samples were boiled in 6X protein loading buffer containing 20% β-mercaptoethanol (BME), separated by polyacrylamide gel electrophoresis (PAGE), transferred to nitrocellulose membrane, and immunoblotted for ubiquitin. Membranes were then stripped with Restore Western blot stripping buffer (Pierce) and re-blotted for Htt. Membranes were stripped a second time and immunoblotted for β-tubulin, used as a loading control. The ratio of immunoprecipitated ubiquitin (ubiquitinated Htt) to immunoprecipitated Htt (total Htt) was assessed using chemiluminescence densitometry. Each band was first normalized to the density of the β-tubulin band for that sample. Immunoprecipitated Htt and ubiquitin levels were compared by computing the ratio of the density of the band in the presence of Happ1 to the density of the band in the presence of V L 12.3. The ratio of immunoprecipitated ubiquitin:Htt in the presence of Happ1 vs. V L 12.3 was then compared. The experiment was repeated two additional times, giving an N of 3.

Htt turnover was assayed using the SNAP-tag fusion system for in vivo protein labeling (NEB) as previously described [14] . Briefly, ST14A cells were grown on cover slips and transfected with mHDx-1-SNAP alone or with iAb. To control for non-specific effects of Happ1, mHDx-1ΔPRR–SNAP alone or with iAb was used. Green fluorescent SNAP-substrate was added to cells 24 hrs post-transfection to label all mHDx-1. Immediately after labeling, some samples were fixed in 4% paraformaldehyde (PFA), stained for cell nuclei (topro-3-iodide, Invitrogen) and mounted for confocal microscopy. Inhibitors of proteolytic processing were then added to the medium of the remaining cultures. At 48 hrs post-transfection, these cells were fixed, stained and mounted as above. Mean intensity of green fluorescence in cells in three microscope fields per well and three wells for each condition was used to determine average Htt levels. The ratio of mean cell intensity at 24 hrs to mean cell intensity at 48 hrs was computed to determine rate of Htt turnover. The experiment was repeated twice, giving an N of 3.

HDx-1 Q46 fused to a thioredoxin tag (TRX) to promote solubility (mHDx-1-TRX) was purified as reported previously [18] and exchanged into diluent (20 mM Tris-HCl, pH 7.4 containing 150 mM NaCl and 2 mM Ca2 + ) using a disposable PD10 desalting column (GE Healthcare). 20 µg purified mHDx-1-TRX was incubated with 1 µg calpain 1 (Sigma) at 37°C for 1 hour. As a control for possible calpain cleavage in the TRX tag or linker sequence, mHDx-1-TRX was cleaved with enterokinaseMax (EKMax) (Invitrogen), a peptidase known to remove the entire TRX-linker sequence from mHDx-1-TRX according to the manufacturer's specifications. Cleavage reactions were separated by PAGE and stained with coomassie to visualize protein bands. Molecular weight of protein bands was determined by comparison to precision plus dual protein molecular weight standard (Biorad) using a FluorChem 8900 (Alpha Innotech) gel documentation system and AlphaImager software. A reaction containing only mHDx-1-TRX and diluent was used to assess uncleaved protein and reactions containing either calpain 1 or EKMax and diluent in the absence of mHDx-1-TRX were used to visualize bands corresponding to these proteins. To identify cleavage fragments, a reaction containing mHDx-1-TRX alone and one containing mHDx-1-TRX and calpain were transferred to nitrocellulose membrane after PAGE separation and blotted with an antibody recognizing the N-terminus of Htt [19] .

Brain tissue was removed from euthanized postnatal day 10 (P10) CD Sprague-Dawley rats (Charles River Laboratory, Raleigh, NC) in accordance with Duke University Medical Center Institutional Animal Care and Use Committee guidelines and approvals (approval #A248-08-09), and as described previously [14] , [20] . 250 µm thick coronal slices containing both striatum and cortex were cut using Vibratomes (Vibratome, St. Louis, MO) in ice-cold culture medium containing 15% heat-inactivated horse serum, 10 mM KCl, 10 mM HEPES, 100 U/ml penicillin/streptomycin, 1 mM MEM sodium pyruvate, and 1 mM L-glutamine in Neurobasal A (Invitrogen). Corticostriatal brain slices were then incubated at 37°C under 5% CO 2 for 1 hr before biolistic transfection with 1.6 µm gold particles coated with DNA constructs expressing yellow fluorescent protein (YFP) as a vital marker, mHDx-1 Q-73, and either V L 12.3 or the CV L non-targeting intrabody. For control transfections, gold particles were coated with YFP alone plus vector backbone DNAs. For time resolved Förster resonance energy transfer (TR-FRET) analysis, brain slices were triturated 10x through 26-gauge needles in ice-cold lysis buffer (50 mM Tris-HCl, 150 nM NaCl, 2 mM EDTA, 1% NP-40, 0.1% SDS + Roche protease inhibitor cocktail tablet) and centrifuged at 12,000 g for 10 min. at 4°C. The supernatants were collected and stored at −80°C until TR-FRET analysis. Experimental conditions were run in triplicate using a single brain slice per lysate.

TR-FRET detection of soluble mHtt was previously described [22] . In brief, samples were lysed in PBS, 0.4% TritonX100 and Complete Protease Inhibitor (Roche). 5 µl sample plus 1 µl antibody mix diluted in assay buffer (50 mM NaH 2 PO 4 , 400 mM NaF, 0.1% BSA and 0.05% Tween) were pipetted per well of a low-volume 384 microtiter plate (Sigma). Final antibody amount per well was 1 ng 2B7-terbium cryptate-labeled antibody and 10 ng MW1-D2-labeled antibody. Plates were incubated for 1 h at 4°C and measured with a Xenon-lamp Envision Reader (PerkinElmer) after excitation at 320 nm. Signal measured at 620 nm resulted from the emission of the terbium cryptate-labeled antibody and was used as a normalization signal for possible assay artifacts due to scattering, quenching, absorption or sample turbidity. Mutant Htt specific signal which resulted from the time-delayed excitation of the D2-labeled MW1 antibody after excitation by the terbium cryptate was detected at 665 nm. 665/620 nm signal ratio was calculated as “TR-FRET signal” specific for soluble mutant Htt.

We previously showed that Happ1 stimulates mHDx-1 turnover [14] . To determine which proteolytic pathway is involved, soluble lysates of HEK 293 cells co-transfected with mHDx-1 plus iAb, or mHDx-1ΔPRR plus iAb, and treated with various inhibitors of proteolytic processing were immunoblotted for Htt ( Fig. 2A ). The ratio of the Htt level in the presence of Happ1 to the Htt level in the presence of V L 12.3 was compared among the various inhibitors. In unperturbed cells, or in the presence of DMSO vehicle, the level of Htt in the presence of Happ1 is reduced compared to the level of Htt in the presence of V L 12.3. This ratio is unchanged by the addition of various inhibitors: lactacystin, a proteasome inhibitor that also affects cathepsin A of lysosomes; epoxomicin, a proteasome inhibitor; 3-MA, an inhibitor of autophagosome formation; or caspase inhibitor I, an irreversible pan-caspase inhibitor. In contrast, the addition of bafilomycin A1, an inhibitor of the vacuolar-type H(+)-ATPase that is known to inhibit autophagosome/lysosome fusion as well as lysosomal pH; or calpain inhibitor I, a pan-calpain inhibitor, significantly increases the ratio of Htt in the presence of Happ1 to Htt in the presence of V L 12.3 ( Fig. 2C ). Thus, these latter two inhibitors interfere with the mechanism by which Happ1 reduces the level of mHtt. The levels of HDx-1 in the presence of calpain inhibitor I or bafilomycin A1 are quantitatively very similar to those found using the HDx-1 construct lacking the proline-rich region to which Happ1 binds ( Fig. 2B ). Thus, it appears that these inhibitors completely abolish the effect of Happ1 on mHDx-1 clearance. The effect of bafilomycin A1 is likely due to disrupted lysosomal pH rather than inhibition of autophagosome/lysosome fusion as evidenced by the lack of effect by 3-MA, which should act upstream of bafilomycin A1 in the macro-autophagy pathway. Therefore, we infer that Happ1 reduces mHDx-1 level by a calpain-CMA-dependent mechanism. 10.1371/journal.pone.0016676.g002 Figure 2 Happ1-mediated reduction of mHDx-1 protein levels is calpain-dependent. HEK 293 cells were co-transfected with mHDx-1 and iAb in the presence of inhibitors of proteolysis or DMSO vehicle. mHDx-1 protein levels in transfected cell lysates was compared by (A, B) Western blotting and (C) densitometry. There is less mHDx-1 protein in the Happ1 transfected cells as compared to the V L 12.3 transfected cells in the presence of vehicle, Lactacystin, epoxomicin, 3-MA or caspase inhibitor 1. Happ1-mediated reduction of mHDx-1 levels is blocked by bafilomycin A1 or calpain inhibitor 1 to the same level as mHDx-1 lacking the Happ1 binding site. * = p<.05, N = 4.

To identify the site of calpain action, human HDx-1 amino acid sequence was analyzed for potential calpain 1 and 2 cleavage sites using the web application SitePrediction [16] . Using this program, AAs 12–17 with cleavage between 15 and 16, (ESLK.SF), is predicted to be the most likely site for both proteases, with greater than 99.9% specificity for calpain 1 and greater than 99% specificity for calpain 2. A secondary site at AAs 5–10, with cleavage between 8 and 9, (EKLM.KA), is predicted to have greater than 99% specificity for calpain 1 and greater than 95% specificity for calpain 2 ( Fig. S1 ).

The iAb V L 12.3 was selected for binding to an N1-20 AA fragment of HDx-1 [24] , a domain that encompasses but is not limited to the predicted calpain cleavage sites. To determine the exact location of V L 12.3 binding we used a 3 AA stepped peptide array binding assay ( Fig. 6A ). The results show that V L 12.3 binds to peptides 3, 4 and 5 which are N7-20, N10-23 and N13-26, respectively ( Fig. 6B ). This demonstrates that V L 12.3 requires AAs 15-18 at the minimum and 13–20 at the maximum for binding. Thus, V L 12.3 binding would be expected to interfere with cleavage at AA 15 and possibly sterically hinder cleavage at AA 8. 10.1371/journal.pone.0016676.g006 Figure 6 V L 12.3 recognizes AAs 13-20 of HDx-1. Three AA stepped 14-mer peptides were spotted onto nitrocellulose and binding of V L 12.3 was assessed. (A) Peptide table. (B) Dot blot showing binding of peptides 3, 4 and 5 illustrating that V L 12.3 requires AAs 15-18 at the minimum and 13-20 at the maximum for recognition of Htt.

Huntington's disease is a devastating neurodegenerative disease for which there is currently no disease modifying therapy. One of the difficulties with HD therapy development is the complex web of dysfunction resulting from the great many processes and pathways affected by mHtt protein in susceptible neurons. As a result, lowering the level of mHtt protein either by silencing expression or increased clearance remains a prime therapeutic approach for HD. For this reason, understanding the mechanism of mHtt degradation is important. Huntingtin degradation involves numerous pathways, with differential toxicity regulated by post-translational modifications. Transgenic mice expressing mHtt that is resistant to caspase-6 cleavage at AA 586 do not develop the HD-like symptoms seen in their caspase-6-sensitive counterparts, despite the presence of other caspase cleavage products in the brain [9] . Phosphorylation of serine 421 shifts processing toward these less toxic products by inhibiting cleavage at AA 586 [11] . Moreover, calpain-resistant mHtt lacking the AA 469 and AA 536 cleavage sites is less toxic and aggregation-prone than calpain cleavage-sensitive mHtt [10] . Phosphorylation of serine 536 inhibits cleavage at AA 536, which also results in reduced toxicity [25] . Modifications of mHtt can also regulate non-protease degradation events. Phosphorylation of serines 13 and 16 increases proteasomal and lysosomal degradation of Htt in turn reducing toxicity. In Drosophila , presumably due to the absence of mammalian degradative machinery and mechanisms, this modification leads to increased toxicity due to accumulation of the more toxic phosphorylated form of mHtt [12] . A better understanding of the complex process of mHtt proteolysis could eventually lead to the development of therapeutics that shift processing toward less toxic pathways and/or enhance removal. Although the generation of N-terminal mHtt fragments by caspase and calpain cleavage has been previously characterized, the subsequent degradation of the highly toxic HDx-1 fragment has remained unclear. With their high target specificity, iAbs are an ideal molecular tool for elucidating protein interactions and functions. For example, the 17 N-terminal AAs of HDx-1 are required for aggregate seeding and cytoplasmic retention [26] , [27] . Blockade of this region by the binding of the iAb V L 12.3 results in nuclear translocation and a potent inhibition of aggregation of HDx-1, illustrating the informative relationship between iAb effects and epitope function [14] , [24] . Happ1 recognizes the proline rich region of HDx-1 and increases clearance of mutant but not wild type HDx-1 [14] . We have exploited this effect of Happ1 binding to gain insight into the mechanism of HDx-1 proteolysis. Htt undergoes a variety of proteolytic processing steps including protease cleavage, proteasomal degradation, and lysosomal/autophagic degradation. In order to determine the initiating or rate-limiting step in mHDx-1 degradation, we tested inhibitors of each of these pathways in the presence and absence of Happ1 and evaluated mHDx-1 levels and turnover using the SNAP-tag method. One caveat of this method is that it requires the use of tagged HDx-1, which could alter HDx-1 conformation or stability. However, we obtained supportive evidence for the effect of V L 12.3 on unlabeled HDx-1 using the entirely independent TR-FRET technique. Moreover, the stimulatory effect of Happ1 on mHDx-1 turnover labeled with the SNAP-tag is quite consistent with the significant lowering of mHDx-1-GFP levels by Happ1. An important control in our experiments is the use of mHDx-1ΔPRR, which lacks the Happ1 binding site. Levels of this protein are not affected by Happ1, indicating that the reduced mHDx-1 levels in the presence of Happ1 do not result from non-specific iAb actions such as activation of the un-folded protein response [28] . The proteasome inhibitors epoxomicin and lactacystin do not disrupt Happ1 stimulation of HDx-1 turnover, and Happ1 does not increase ubiquitination of HDx-1. These results indicate that Happ1 does not accelerate proteasomal degradation of Htt. We also find that 3-MA, an inhibitor of autophagosome formation and the macroautophagy pathway, does not interfere with Happ1-accelerated mHDx-1 degradation. In contrast, bafilomycin A1, a vacuolar-type H(+)-ATPase inhibitor that hinders lysosome-autophagosome fusion as well as disrupting lysosomal pH, prevents Happ1-induced changes in mHDx-1 clearance rate. Due to the lack of effect of 3-MA, it is unlikely that the action of bafilomycin A1 on autophagosome/lysosome fusion is responsible for disrupting Happ1 function. It is more likely that bafilomycin A1 disrupts Happ1 function by disrupting lysosomal pH. This indicates a role for CMA, which is an autophagosome-independent lysosomal degradation process, in Happ1-enhanced mHDx-1 clearance. We next evaluated the sensitivity of the Happ1 effects to the caspase and calpain