Ubiquitin-Activating Enzyme Activity Contributes to Differential Accumulation of Mutant Huntingtin in Brain and Peripheral Tissues

Huntington's disease (HD) belongs to a family of neurodegenerative diseases caused by misfolded proteins and shares the pathological hallmark of selective accumulation of misfolded proteins in neuronal cells. Polyglutamine expansion in the HD protein, huntingtin (Htt), causes selective neurodegeneration that is more severe in the striatum and cortex than in other brain regions, but the mechanism behind this selectivity is unknown. Here we report that in HD knock-in mice, the expression levels of mutant Htt (mHtt) are higher in brain tissues than in peripheral tissues. However, the expression of N-terminal mHtt via stereotaxic injection of viral vectors in mice also results in greater accumulation of mHtt in the striatum than in muscle. We developed an in vitro assay that revealed that extracts from the striatum and cortex promote the formation of high-molecular weight (HMW) mHtt compared with the relatively unaffected cerebellar and peripheral tissue extracts. Inhibition of ubiquitin-activating enzyme E1 (Ube1) increased the levels of HMW mHtt in the relatively unaffected tissues. Importantly, the expression levels of Ube1 are lower in brain tissues than peripheral tissues and decline in the nuclear fraction with age, which is correlated with the increased accumulation of mHtt in the brain and neuronal nuclei during aging. Our findings suggest that decreased targeting of misfolded Htt to the proteasome for degradation via Ube1 may underlie the preferential accumulation of toxic forms of mHtt in the brain and its selective neurodegeneration.


Introduction
Huntington's disease (HD) is an autosomal-dominant neurodegenerative disease caused by an expansion of the polyglutamine (polyQ) tract in the huntingtin protein (Htt). This progressive neurodegenerative disease is characterized by the loss of neurons, which is most pronounced in the striatum and cerebral cortex, resulting in symptoms such as chorea, dystonia, depression, cognitive deficits, and, in the late stage, death (Landles and Bates, 2004). A pathological hallmark of the disease is the formation of cytoplasmic and intranuclear aggregates or inclusions that consist of small N-terminal fragments of mutant Htt (mHtt; DiFiglia et al., 1997;Gutekunst et al., 1999). There are transgenic mouse models that expresses N-terminal fragments of mutant proteins, but develop more severe phenotypes (Davies et al., 1997;Schilling et al., 1999;Xu et al., 2013a) than HD models expressing fulllength (FL) mHtt (Slow et al., 2003;Gray et al., 2008), suggesting that smaller N-terminal mHtt is more toxic than longer mHtt fragments. However, how N-terminal Htt fragments accumulate in the brain and cause HD pathogenesis remains to be investigated.
Although both wild-type (WT) protein and mHtt are ubiquitously expressed throughout the body, there is no correlation between the expression of Htt and the brain regions most affected by HD (Li et al., 1993;Fusco et al., 1999;Gutekunst et al., 1999;. The role of mHtt aggregates and inclusions is also controversial. mHtt can form several types of aggregate precursors, termed oligomers or protofibrils, in a polyQ length-dependent manner (Legleiter et al., 2010) and can be found in brain tissues from HD mouse models . The role, if any, of these mHtt oligomers in HD toxicity is still unclear, though evidence suggests that there are multimeric pools of oligomerized mHtt that can form aggregates with age (Marcellin et al., 2012). Since aggregate formation does not correlate well with HD toxicity, oligomerized mHtt, rather than aggregated mHtt, is suspected of being the toxic intermediary.
Despite the controversial role of mHtt aggregates, it is evident that these aggregates are ubiquitinated (Davies et al., 1997;Di-Figlia et al., 1997;Gutekunst et al., 1999). Protein ubiquitination is initiated by ubiquitin-activating enzymes E1 (UbeE1s), which then pass the ubiquitinated proteins to E2-conjugating enzymes and E3 ligases for targeting to the proteasome for degradation (Schulman and Harper, 2009). Protein ubiquitination and degradation are involved in a variety of cellular functions (Kerscher et al., 2006), as well as in removing misfolded and toxic proteins in a variety of neurodegenerative diseases (Liu et al., 2007;Tan et al., 2008;Lim and Lim, 2011). However, we know very little about how ubiquitination contributes to the preferential accumulation of mutant proteins in neuronal cells. In the current study, we observed that the differential formation of high-molecular weight (HMW) mHtt and aggregates in the brain and peripheral tissues is modulated by E1-E2-E3 targeting activity and that ubiquitin-activating enzyme expression correlates with the differential accumulation of mHtt. Our findings offer new insight into the preferential accumulation of mHtt and its selective neurodegeneration.

Materials and Methods
Animals. Animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Emory University. Full-length mHtt CAG140Q [HD knock-in (KI)] mice were provided by Dr. Michael Levine (University of California, Los Angeles, Los Angeles, Hickey et al., 2008) and were maintained at the Emory University Animal facility in accordance with IACUC guidelines. For stereotaxic injection of virus and behavioral studies, three female and two male mice per group were used. For the remainder of the experiments, both sexes were used.
Plasmids and antibodies. PRK vectors expressing 1-212 (N212) or 1-508 (N508) N-terminal Htt fragments containing 23Q (WT) or 120/ 150Q (mutant) were generated previously (Li and Li, 1998;Havel et al., 2011). The lentiviral vectors were constructed by adding a digested F2Aenhanced cyan fluorescent protein (ECFP) fragment to the C terminus of Htt in the PRK vector. The htt-F2A-ECFP DNA fragment was then excised and ligated into the pFUW viral backbone for viral production. The final virus was packaged and amplified by the Emory University Viral Vector Core, and expresses Htt and ECFP under the same promoter in the same cells in which F2A self-cleavage can separate Htt and ECFP proteins.
Western blot analysis. Whole brains from WT and HD KI mice were dissected to isolate the cortex, striatum, and cerebellum. The tissues were homogenized for 25 strokes in ice-cold homogenization buffer (10 nM Na 2 HPO 4 , 5 mM EDTA, 5 mM EGTA, 0.02% NaN3) containing a protease inhibitor mixture (Pierce Protein Biology Products, Thermo Scientific) and 100 M PMSF. For Western blotting, samples were boiled for 5 min in SDS/␤-mercaptoethanol (BME) protein loading dye and run at 50 mA on 4 -12% Tris-Glycine gels purchased from Invitrogen (catalog #EC60385). Proteins were transferred to a nitrocellulose membrane for 80 min at 105 V in Tris-Glycine buffer. After blocking, blots were probed with mEM48 or other primary antibodies overnight. The Western blots were developed using the ECL Prime Chemiluminescence kit (GE Healthcare). For subcellular fractionation, whole-brain cortex was collected from 4-, 8-, and 12-month-old HD CAG140 knock-in mice, and fractions (P1, P3, and S3) were prepared using the method of Sharp et al. (1995). Lysates were centrifuged at 1000 ϫ g for 10 min at 4°C to separate the P1 and S1. The P1 was washed again to yield the nuclear fraction. S1 was centrifuged at 17,600 ϫ g for 20 min at 4°C. The S2 was then centrifuged at 100,000 ϫ g for 1 h in a swinging bucket rotor to separate the P3 (synaptosome fraction) and the S3 (soluble cytoplasm fraction). Protein samples were prepared as described earlier.
For formic acid solubilization of mHtt aggregates, cortex, cerebellum, and striatum were dissected out from a 7-month-old heterozygous HD CAG140 knock-in mouse. Formic acid-solubilized aggregates were obtained following procedures previously outlined (Lunkes et al., 2002;Zhou et al., 2003;Landles et al., 2010). Tissues were homogenized in ice-cold buffer (50 mM Tris-HCl, pH 8.8, 100 mM NaCl, 5 mM MgCl 2 , 1 mM EDTA, pH 8.0, 0.5% NP40, plus protease inhibitors). Lysates were centrifuged at 800 ϫ g at 4°C to fractionate into crude nuclear and cytoplasmic fractions. SDS buffer (500 l, 2% SDS, 5% BME, 15% glycerol, protease inhibitors) was added to the crude nuclear pellet and boiled for 10 min. The boiled nuclear pellet was sonicated for 20 s at 5 mA, and centrifuged for 15 min at 16,200 ϫ g at room temperature. The pellet (insoluble aggregates) was incubated in 100% formic acid with shaking (350 rpm) at 37°C for 1 h. The formic acid was removed by Vacufuge (30°C, 2 h), and then the sample was reduced in 1 M Tris Base and protease inhibitors.
Immunoblot signals were quantified using either ImageJ or UN-SCAN-IT. We used the same size section on the blots for quantification. All data have been quantified as a ratio of protein to a loading control and then normalized within the experiment.
In vitro degradation assay. Htt HEK293 cells stably expressing transfected full-length Htt were grown to confluence in a 10 cm plate. Cells were washed in the plate then lysed in cold assay buffer (25 mM Tris-HCl, pH 7.6, 10 mM MgCl, 100 g/ml purified rabbit creatine kinase, 50 mM phosphocreatine, 1 mM ATP added at use). Wild-type mice were killed, and tissues (striatum, cerebellum, cortex, kidney, liver, heart, and muscle) were rapidly collected and homogenized at 1 g/1 ml in cold assay buffer using 20 strokes of a glass dounce hand homogenizer. Both cell lysates and tissues were centrifuged 500 ϫ g at 4°C for 5 min to pellet unbroken tissues and membranes. The supernatant was collected and stored on ice, while protein concentrations were determined using a BCA Protein Assay Kit (Thermo Scientific). Both tissue and cell lysates were prepared at 170 g protein/500 l assay buffer, and the appropriate number of 200 l aliquots was prepared. The samples containing both cell and tissue lysates were mixed (200 l cell lysates plus 200 l tissue lysates). The control samples contained equal volumes of tissue or cell lysates and assay buffer. The samples were incubated at 37°C with 300 rpm shaking for 0, 8, or 16 h. The reaction was stopped by the addition of protease inhibitors, PMSF, and protein loading dye.
For inhibiting the proteasome and E1 enzymes, tissues from HD CAG140 KI mice (Menalled et al., 2003) were collected and homogenized in cold assay buffer (25 mM Tris-HCl, pH 7.6, 10 mM MgCl, 100 g/ml purified rabbit creatine kinase, 50 mM phosphocreatine, and 2 mM ATP). Tissue lysates were prepared as in the in vitro degradation assay (IVDA) and diluted to 170 g protein/500 l assay buffer, and the appropriate number of aliquots was prepared. Fifty micromolar MG132 or 50 M PYR41  was added to appropriate tubes. The reaction was incubated and stopped as described above.
Stereotaxic injection. Three-month-old wild-type mice (C3H) were anesthetized using 0.39 ml/20 g 2.5% avertin. The animals were injected in either the right hemisphere or bilaterally with 1.5 l (1.08 ϫ 10 12 ) lentivirus per side with the following coordinates set from the bregma: 2 mm lateral, 0.6 mm anterior, and Ϫ3.3 mm ventral. The flow rate was 200 nanoliters/min, and there was a 3 min delay before the needle was slowly removed. The animals were monitored daily and administered the analgesic ketoprofen (Fort Dodge Animal Health) at 2 mg/ml for 3 d postsurgery.
For muscle injection, mice were anesthetized using 0.39 ml/20 g 2.5% avertin (Alfa Aesar), and 1.5 l of virus was diluted to 50 l and injected into the calf muscle using a 100 l Hamilton syringe with a 27 g disposable needle. Analgesics were administered as described.
Rotarod assay. Animals were tested on a Rotamex Rotarod (Columbus Instruments) three times per day on 3 consecutive days. After a constant slow rotation, the speed was increased gradually over the course of 10 min from 5 revolutions/s to 30 revolutions/s. Data from all 3 d were averaged and used for comparison of each group.
Statistical analyses. Quantifications of Western blots were obtained using ImageJ. The repeated-measures two-way ANOVA and two-way ANOVA were performed using Prism version 6.0 (GraphPad). An ␣ of Յ0.05 was considered to be significant.

Results
mHtt is more stable in brain than peripheral tissues mHtt is ubiquitously expressed; however, HD is characterized by neurological symptoms and degeneration in specific regions of the brain. This selective neurodegeneration has been a longstanding issue for researchers to address since it does not correlate directly with Htt expression levels or observed protein aggregation (Li et al., 1993;Fusco et al., 1999;Gutekunst et al., 1999;. We first compared mHtt expression levels in brain and peripheral tissues in a HD CAG140Q knock-in mouse model, which expresses full-length mHtt under the control of the endogenous mouse Htt promoter and shows the progressive phenotypic development and formation of mHtt aggregates (Menalled et al., 2003;Hickey et al., 2008). Western blotting revealed a different array of mHtt fragments in each tissue from a 7-month-old HD KI mouse (Fig. 1A). FL mHtt (open arrow) appears to be less stable in peripheral tissues as we were unable to detect it by immunoblotting with either the antibody to N-terminal Htt (mEM48) or antibody to expanded polyQ repeats (1C2). Fewer N-terminal mHtt fragments were seen in the peripheral tissues than in brain regions, suggesting that mHtt proteolysis may be different between brain and periph-  (Ctx), striatum (Str), cerebellum (Crb), heart (Hrt), kidney (Kid), liver (Liv), and skeletal muscle (Mus). Western blots were probed with mEM48 (1:50), 1C2 (1:10,000), and anti-vinculin. SG, Stacking gel. B, Quantitative RT-PCR assessing the expression levels of mHtt RNA in HD CAG140 KI mice. Three brain regions and four peripheral tissues (see A for abbreviations) were examined. GAPDH was used as a control. The relative Htt levels were normalized by GAPDH and calculated using r ϭ 2 (Chtt Ϫ Cgapdh) (n ϭ 6). Error bars indicate the mean Ϯ SEM. The PCR products were also analyzed by staining DNA gel (bottom). C, Immunohistochemistry with anti-Htt (mEM48) of the cortex, striatum, and kidney tissues of 6-month-old HD KI mice and age-matched WT mice controls. Scale bars, 10 m. eral tissues. We then performed qRT-PCR to quantify the levels of mHtt mRNA relative to GAPDH mRNA (Fig. 1B). There were higher levels of mHtt mRNA in all three brain regions than in the peripheral tissues, and mHtt is expressed at the lowest level in muscle. Immunohistochemistry with mEM48 revealed the accumulation and aggregation of mHtt in the cortex and striatum in 6-month-old HD KI mice, whereas there was no obvious mHtt staining signal by mEM48 in the kidney (Fig. 1C). Together, these data suggest that there are differences in mRNA expression and the post-translational modification of mHtt in the brain and in peripheral tissues, which contribute to the differential accumulation, aggregation, and subsequent toxicity of mHtt N-terminal fragments.
We also purified Htt aggregates from the cortex, striatum, and cerebellum of HD KI mice at the age of 7 months using formic acid treatment to dissolve aggregates. We found that aggregates in different brain regions are formed by multiple N-terminal Htt fragments, which are the same in different brain regions (Fig. 2). Thus, the accumulation and aggregation of mHtt in different brain regions are unlikely to be due to intrinsic proteolysis differences of Htt, but rather result from tissue-specific factors that modulate the stability of these N-terminal Htt fragments.

mHtt is more stable and forms more HMW mHtt in tissues affected in HD in vitro
To better understand the early steps in the accumulation of mHtt, we designed an IVDA that allows us to observe tissue-specific processing of mHtt by adding the same amount of mHtt to cellular lysates from different tissues. FL Htt proteins containing either 23Q or 120Q in transfected HEK293 cells  were used as a source of Htt. Lysates from individual tissues (striatum, kidney, liver, heart, and skeletal muscle) in wild-type mice were incubated with transfected Htt to test tissue-specific effects of modulation on Htt levels, stability, and aggregation. The tissue lysates in a Tris buffer contained creatine kinase, phosphocreatine, and ATP (TCP-ATP buffer), which are necessary for energy-dependent protein post-translational modifications. After incubation for different times, the transfected Htt and lysates were subjected to Western blotting (Fig. 3A). We then quantified the relative levels of mHtt on Western blots by measuring its ratio to vinculin, a cytoskeletal protein that is stable even after a long incubation time. We found that FL 120Q (open arrowhead) and N-terminal mHtt fragments are more stable in the striatal lysate than in the lysates from most peripheral tissues observed (Fig.  3 B, C). Two-way ANOVA using time and tissue type as separate variables revealed that the length of incubation time contributed the most to mHtt degradation (11.66%; p ϭ 3.29E-6), whereas tissue type contributed 6.851% to the degradation ( p ϭ 0.8137) across all groups. Multiple comparisons show that the decrease in mHtt reaches significance in the kidney ( p ϭ 0.0351) and liver (p ϭ 0.0012) lysates at 16 h, and in the muscle at both 8 and 16 h (p ϭ 0.0003 and p ϭ 5.26E-7). However, the decrease in mHtt was not significant at either 8 or 16 h in the striatal or heart lysates.
When comparing brain regions, the lysates from cortex and cerebellum yielded a significant decrease in mHtt at 16 h incubation ( p ϭ 0.0364 and p ϭ 0.0109), but there was no significant decrease with striatal lysates (Fig. 3 D, E). This analysis suggests that mHtt is more stable in vitro with the striatal lysates than with the other tissue lysates. When comparing HMW mHtt in different brain regions (Fig. 3 D, E), tissue type contributed 38.48% to the formation of HMW mHtt ( p ϭ 5.61E-5), while incubation time contributed 21.86% of the effect ( p ϭ 5.99E-7), and 28.48% of the effect on oligomer formation is caused by an interaction between the two variables ( p ϭ 1.54E-5). Multiple-comparison tests allowed us to see that more HMW mHtt was formed in the cortex and striatum lysates than in the cerebellum lysates at 16 h of incubation ( p ϭ 3.73E-9 and p ϭ 0.0301). No HMW mHtt was formed in control samples without brain tissue lysates (Fig. 3D, first two lanes). These data show that in vitro the striatal tissue is less able to clear mHtt than the other tissues tested, and therefore more HMW mHtt is formed with the striatal lysates. This difference is consistent with the preferential accumulation of mHtt in the striatum of HD KI mice (Lin et al., 2001;Menalled et al., 2003) and the selective vulnerability of striatal neurons in HD.

N-terminal mHtt is more stable in the brain than in the muscle and causes motor dysfunction
Since HD mouse models suggest that small N-terminal fragments of mHtt are more toxic than full-length mHtt (Heng et al., 2008), we chose to examine the degradation and HMW mHtt formation of N-terminal mHtt fragments in the brain and peripheral tissues. Lentiviral constructs expressing N-terminal fragments of WT mice and mHtt under the ubiquitin promoter (UBC) were injected into the striatum and skeletal muscle. Muscle was chosen for comparison with brain because it, like neurons, is also a terminally differentiated tissue. The three viral vectors (Fig. 4A) that were used express a small N-terminal Htt fragment with 23Q (N212-23Q) or 120Q (N212-120Q), or a larger N-terminal mHtt fragment (N508 -120Q). Viruses were injected into the  striatum and the right calf muscle (n ϭ 3 per group) of 3-monthold WT mice. One month postinjection, the injected tissues were isolated for Western blot analysis. All three transgenic Htt proteins were expressed clearly in the striatum of the injected mice, while we detected no or very little mHtt protein (N212-120Q and N508 -120Q) in the skeletal muscle by Western blotting, despite a low level of normal N-terminal Htt (N212-23Q) (Fig. 4B). Importantly, the shorter N-terminal mHtt (N212-120Q) formed more aggregated htt than N508 -120Q in the striatum, suggesting that the smaller N-terminal Htt fragment is more prone to misfolding (Fig. 4B). These in vivo data agree with observations from our in vitro experiments that N-terminal mHtt is more stable in the brain than in the muscle.
Next, we examined the in vivo toxicity of these mHtt fragments using the rotarod as a test of motor function. WT mice (n ϭ 5 per group) were injected bilaterally with these viral Htt vectors in the striatum and were used for longitudinal study. There was no significant difference in body weights among the three groups injected with viral vectors expressing different N-terminal Htt, though mice injected with mHtt show slightly less weight gain than mice injected with normal N-terminal Htt (Fig. 4C). One month postinjection, animals for the behavioral studies were trained for the rotarod test, and then examined at 6 weeks and once every 3 weeks afterward (Fig. 4D). Two-way ANOVA analysis revealed that Htt fragment contributed the most variance (20.09%, p ϭ 2.48E-21), and time, while accounting for only 3.76% of the effect, was still considered significant (p ϭ 1.04E-9). Mice injected with viral N212-120Q Htt performed worse than those with N508 -120Q at all time points or weeks (w) postinjection. (6 w, p ϭ 4.45E-12; 9 w, p ϭ 1.44E-7; 12 w, p ϭ 2.09E-13; 15 w, p ϭ 4.29E-8; 18 w, p ϭ 1.23E-7; 21 w, p ϭ 2.71E-6; 24 w, p ϭ 0.0002). This increased toxicity of N212-120Q is also 4 (figure legend continued.) incubation (n ϭ 4), and were normalized within the experiment to allow comparison between experiments. D, Comparison of the striatum (Str) with cortex (Ctx) and cerebellum (Crb) for their ability to process full-length mHtt by IVDA. Controls are cellular lysates only and tissue lysates only. Kid, Kidney; Liv, liver; Hrt, heart; Mus, muscle. E, All soluble mHtt fragments and full-length mHtt (top) and HMW mHtt (bottom) in D were quantified as a ratio to vinculin and normalized. Error bars indicate mean Ϯ SEM, and statistical significance was determined using repeated-measures two-way ANOVA with (Fisher's LSD test) multiple testing (*p Ͻ 0.05, **p Ͻ 0.01, ***p Ͻ 0.001; see text for exact p values). Htt is linked with ECFP via F2A, a peptide linker that can be self-cleaved in cells to separate mHtt from ECFP. Constructs are expressed under the control of the ubiquitin promoter (UBC). B, Western blot analysis of the striatum and muscle tissues that had been injected with lentiviral mHtt for 1 month. The blots were probed with mEM48 (1:50) to detect mHtt expression. C, D, Body weights (C) and rotarod performance (D) of mice injected with lentiviral mHtt into their striatum were normalized to initial weight before surgery. Error bars indicate the mean Ϯ SEM (n ϭ 5 per group). Statistical significance determined using repeated-measures two-way ANOVA with multiple testing (Bonferroni test). The p value shown (***p Ͻ 0.001) is between N212-120Q and N508 -120Q, as indicated by the bracket.  (P4D1 1:1000). D, Quantification of total ubiquitin at 0, 8, and 16 h of incubation. Ubiquitin was quantified as a ratio to the loading control (vinculin) then normalized within the experiment. Error bars indicate the mean Ϯ SEM, and statistical significance was determined using repeated-measures two-way ANOVA with multiple testing (*p Ͻ 0.05, **p Ͻ 0.01, ***p Ͻ 0.001, see text for exact p values). Str, Striatum; Ctx, cortex; Crb, cerebellum; Kid, kidney; Liv, liver; Hrt, heart; Mus, muscle.  (n ϭ 3). The ratio of HMW mHtt to vinculin was used for quantification. Error bars indicate the mean Ϯ SEM, and statistical significance was determined using repeated-measures two-way ANOVA with (Tukey's test) multiple testing (*p Ͻ 0.05, **p Ͻ 0.01). C, HD KI mouse tissue lysates were incubated in vitro in IVDA buffer and treated with the E1 inhibitor PYR41 (50 M) or the UPS inhibitor MG132 (50 M). Western blotting showing that PYR41 or MG132 increased the accumulation of oligomerized mHtt in the cerebellar tissue compared with no drug treatment (left). Striatal tissue samples were also included for comparison with peripheral tissues (right). D, Quantification of the ratio of oligomerized mHtt to vinculin after in vitro incubation at 8 h then normalized within the experiment. Error bars indicate the mean Ϯ SEM (n ϭ 3). Str, Striatum; Mus, muscle; Ctx, cortex; Crb, cerebellum.
consistent with the increased stability and aggregation of N212-120Q in the injected striatum seen by Western blotting (Fig. 4B).
Given the findings that N-terminal mHtt is more stable in the striatum and forms more HMW mHtt, we wanted to know whether there are differences in Htt ubiquitination in different brain regions and peripheral tissues using the IVDA. We compared total ubiquitin levels at 0 and 16 h after incubation of full-length mHtt (FL-120Q) with cortex, striatum, and cerebellum tissue lysates. Western blots with anti-ubiquitin (P4D1) revealed that the incubation markedly increased protein ubiquitination in striatal lysates (Fig. 5C). Although the incubation also increased protein ubiquitination in brain tissue lysates without FL-120Q, adding FL-120Q markedly increased protein ubiquitination (Fig. 5C, left). Thus, this increase reflects mHtt ubiquitination in IVDA. Importantly, striatal lysates resulted in more ubiquitination than cortex and cerebellar lysates, and brain tissue lysates yielded greater ubiquitination than peripheral tissue lysates (Fig. 5C), which was also verified by quantitative analysis of the relative levels of ubiquitin on the blots (Fig. 5D). The peripheral tissues incubated with FL-120Q also showed a rise in total ubiquitin levels at 8 h and then a decrease at 16 h of incubation, though the ubiquitin levels are lower than those with brain tissue lysates. Two-way ANOVA showed that incubation time contributed the most to ubiquitination (32.25%, p ϭ 2.01E-5), and tissue type contributed 23.76% of the variance observed in all groups ( p ϭ 0.0930, not significant; Fig. 5D). These differences can be interpreted in the following two ways: one is that ubiquitinated proteins could be cleared more easily in the peripheral tissues; and the other is that more ubiquitinated proteins are generated in the brain than in the peripheral tissues.

Inhibiting ubiquitin-activating enzyme E1 promotes aggregate formation
To determine how differences in ubiquitination could contribute to the stability of mHtt, we inhibited ubiquitination using the drug PYR41, a cell-permeable pyrazone compound that is reported to irreversibly inhibit ubiquitin-activating enzyme E1 activity Guan and Ricciardi, 2012;Su et al., 2013;Xu et al., 2013b). The IVDA was then performed using striatal and muscle lysates mixed with full-length mHtt (FL-120Q), with the addition of 50 M PYR41. For comparison, the ubiquitin- Figure 7. Differential levels of Ube1 in brain and peripheral tissues. A, Western blot analysis of Ube1 in the brain and peripheral tissues of WT and HD KI mice. The ratios of Ube1 (bottom) to the loading control are shown beneath the blot. B, Western blot analysis of subcellular fractionations of HD KI mice at different ages. The blot was probed with anti-Htt (mEM48). The HMW mHtt increases in the nuclear and synaptosomal fractions from the aged HD KI mice. C, The blot in B was probed with anti-Ube1, and immunoblot signals from short and long exposure times are presented. The ratios of Ube1 to that at 4 months are shown beneath the blot. Ctx, Cortex; Str, striatum; Crb, cerebellum; Kid, kidney; Liv, liver; Hrt, heart; Mus, muscle. proteasome system (UPS) inhibitor MG132 (50 M) was used to inhibit protein degradation by the proteasome. The blots were probed with Htt, K48 ubiquitin, K63 ubiquitin, and sumoylation antibodies. PYR41 caused greater generation of HMW mHtt in both the striatal and muscle mixtures than MG132 (Fig. 6A). Quantification of densitometry verified that PYR41 treatment of both striatal and muscle tissues increased HMW mHtt ( p ϭ 0.0015 and p ϭ 0.0454, respectively) to a greater extent than MG132 (Fig. 6B). To compare the levels of HMW mHtt formed by endogenous full-length mHtt in the striatum, cortex, and cerebellum, we used brain tissue lysates from HD KI mice to perform IVDA with MG132 or PYR41 treatment. The immunoblots were probed with anti-Htt and anti-ubiquitin to detect HMW mHtt and ubiquitinated proteins. Although total ubiquitinated proteins do not seem different between the drug-treated samples and the control without drug treatment, both MG132 and PYR41 appear to increase the level of HMW mHtt in the striatum and cortex (Fig. 6C,D, top). The lower level of HMW mHtt was seen with the cerebellar lysates than with the striatal and cortical lysates, which was also increased by PYR41. However, these increases are not statistically significance due to large variations in the limited number of experiments (n ϭ 3). Given that PYR41 blocks the degradation of ubiquitinated proteins by the ubiquitin-proteasome system , these results suggest that PYR41 may prevent the targeting of HMW mHtt to the proteasome, resulting in increased levels of HMW mHtt.
Ubiquitin-activating enzyme E1 is expressed at different levels in brain and peripheral tissues Given the above in vitro data, it was important to investigate whether Ube1 activity is correlated with the accumulation of mHtt in the brain. We therefore performed Western blot analysis of brain and peripheral tissues with known differential levels of mHtt in HD KI mice (Fig. 1). Using an antibody to Ube1 and comparing the ratio of Ube1 to eIF5␣, a translational protein that is expressed in all of the types of cells used and is involved in translation elongation, cell cycle progression, and actin dynamics (Si et al., 1996), we found that the expression levels of Ube1 are markedly lower in brain tissues than in peripheral tissues (Fig.  7A). The levels of Ube1 are also lower in the striatum (Ube1/ eIF5␣: 1.2) and cortex (1.0) than in the cerebellum (1.4). There is no difference in Ube1 expression between HD KI and WT mouse tissues (Fig. 7A), suggesting that mHtt does not impair the expression of Ube1. If Ube1 expression is indeed critical for the levels of mHtt, we should also see low levels of it in the nucleus and nerve terminals in which mHtt preferentially accumulates and forms aggregates. We next performed subcellular fractionation experiments using HD KI mice at 4, 8, and 12 months of age, and found that the aggregated mHtt is increased in the nuclear and synaptosomal fractions from the older HD KI mice (Fig. 7B). More importantly, Ube1 levels are indeed lower in the nuclear and synaptosomal fractions than in the cytosolic fraction (Fig. 7C). Compared with the level of Ube1a at 4 months, the nuclear Ube1 level significantly declines with age (from 1.0 to 0.16), while synaptosomal Ube1 levels remain unchanged. This difference suggests that Ube1 activity is also important for the age-dependent accumulation of mHtt in neuronal nuclei, whereas other synaptic modulators may alter mHtt accumulation and aggregation in synapses during aging.
Based on the above findings and the fact that E1 enzymes target ubiquitinated proteins for degradation via the proteasome, we propose that differential E1 enzyme activities in different Figure 8. A proposed model for the differential accumulation of mHtt in affected brain regions. mHtt is cleaved to form fragments, which can form HMW mHtt and be ubiquitinated. Differential targeting of the ubiquitinated mHtt to the proteasome can lead to different levels of misfolded proteins. In striatal neurons, the reduced targeting of ubiquitinated Htt to the proteasome is likely to contribute to the preferential accumulation of mHtt in these cells, affecting the cells and forming aggregates. Ube1 activity and other targeting proteins account for the differential accumulation of mHtt in different brain regions and subcellular compartments.
types of tissues and cells contribute to the differential accumulation of mHtt and its aggregation in neuronal cells. Reduced E1 enzyme activities may explain why mHtt preferentially accumulates in the brain, and forms aggregates in the nucleus and synapses (Fig. 8).

Discussion
The selective neurodegeneration seen in a variety of neurodegenerative diseases has long been a puzzle. Understanding how mutant proteins selectively accumulate in the affected brain regions would help us to understand this important and unresolved issue. In the present study, we developed a novel in vitro assay that allows us to closely monitor the accumulation of mHtt in different tissues. This IVDA demonstrates that mHtt is more stable in the brain than in peripheral tissues. Our findings here suggest that reduced targeting of misfolded mHtt to the proteasome for degradation may contribute to the preferential accumulation of mHtt in affected neurons and the selective neuronal vulnerability in individuals with HD.
We also found that mHtt mRNA expression is in general lower in the peripheral tissues than in brain regions. This difference could also help to explain the increased toxicity of mHtt in neuronal cells. However, by expressing the same viral mHtt vectors in the striatum and muscles, we still saw more abundant accumulation of mHtt in the striatum. In addition, in different brain regions, such as the cortex and striatum, more mHtt mRNA is expressed in the cortex than in the striatum, though more and earlier accumulation of mHtt and aggregates are seen in the striatum in HD KI mice (Li et al., 2000Lin et al., 2001). Thus, post-translational modification of mHtt appears to be important for the differential accumulation of misfolded forms of mHtt, at least in brain regions.
Using HD KI mice, we verified that aggregated htt is formed by small N-terminal mHtt. By injection of viral vectors expressing N-terminal mHtt into mice, we also confirmed that N-terminal mHtt fragments are toxic and can become misfolded and aggregated, causing motor dysfunction. The IVDA shows that tissues from the striatum, which is most severely affected in HD, promotes the formation of HMW mHtt by N-terminal mHtt to a greater extent than tissues from relatively unaffected regions and organs. At least two possibilities could explain this phenomenon. One is that striatal tissue favors the generation of HMW or misfolded mHtt, and the other is that HMW Htt is more stable or less efficiently cleared in the striatum.
The ubiquitin-proteasome system plays a major role in clearing misfolded proteins, a process that involves multienzyme cascades. The first step of ubiquitin activation is mediated by the E1 enzymes that bind ATP and ubiquitin, and catalyze ubiquitin C-terminal acyl-adenylation. The active ubiquitin is then passed by E1 enzymes to E2-conjugating enzymes and then to E3 ubiquitin ligases, a reaction that is necessary for targeting of ubiquitinated proteins to the proteasome for degradation (Schulman and Harper, 2009). PYR41, a commercially available cell-permeable inhibitor, was found in a high-throughput screening of an E1-E2-E3 cascade ; it does not affect ubiquitination of proteins, but blocks the degradation of ubiquitinated proteins by the proteasome , giving us a useful tool to examine the targeting of ubiquitinated proteins to the proteasome for degradation. PYR41 appears to increase the accumulation of mHtt to a greater extent than MG132 treatment, suggesting that PYR41 may also inhibit the targeting of ubiquitinated mHtt to other degradation systems, such as autophagy. In fact, autophagy can also degrade ubiquitinated proteins (Shaid et al., 2013). Although whether PYR41 affects autophagy function remains to be investigated, our findings suggest that reduced degradation of misfolded mHtt accounts for the increased accumulation of mHtt in neuronal cells.
Strong evidence in support of our hypothesis is that the expression levels of E1 enzyme Ube1 are lower in brain tissues than in peripheral tissues. The lower levels of Ube1 are consistent with the increased levels of mHtt in brain tissues and in vitro results after inhibiting Ube1 by PYR41. The correlation between the decreased levels of Ube1 and the increased amount of ubiquitinated mHtt in the brain suggests that the high levels of Ube1 may be required for targeting the ubiquitinated proteins to the proteasome for degradation more efficiently, whereas the ubiquitination of proteins can be maintained with a low level of Ube1. The reduced levels of Ube1 in the brain also suggest that this enzyme is expressed at a lower level in neuronal cells. We should point out that the clearance of misfolded proteins by the ubiquitin-proteasome system is a complex process involving many targeting proteins. For example, there are eight E1-activating enzymes to initiate ubiquitination (Schulman and Harper, 2009) and numerous E3 ligases for substrate-specific targeting (Ardley and Robinson, 2005). It remains to be investigated whether there are tissue-or cell type-specific proteins involved in targeting mHtt to the proteasome. Also, since there are different subcellular localizations of the E1 isoforms (Grenfell et al., 1994), whether different subcellular localizations of these proteins account for the differential accumulation of mHtt in the nucleus and synapses remains to be determined. Such proteins could be additional modulators to alter the degradation of mHtt in brain regions, as well as nerve terminals and different subcellular compartments. However, the intrinsic lower levels of Ube1 in brain tissues than peripheral tissues has broad implications for the selective accumulation of misfolded proteins in the brain in a variety of neurodegenerative diseases, including Alzheimer's and Parkinson's diseases. Furthermore, the IVDA assay we developed could be used to identify tissue-specific proteins that modulate the degradation of misfolded proteins in other polyQ and neurodegenerative diseases.