Systematic Uncovering of Multiple Pathways Underlying the Pathology of Huntington Disease by an Acid-cleavable Isotope-coded Affinity Tag Approach*S

Huntington disease (HD) is an autosomal dominant neurodegenerative disease that results from a CAG (glutamine) trinucleotide expansion in exon 1 of huntingtin (Htt). The aggregation of mutant Htt has been implicated in the progression of HD. The earliest degeneration occurs in the striatum. To identify proteins critical for the progression of HD, we applied acid-cleavable ICAT technology to quantitatively determine changes in protein expressions in the striatum of a transgenic HD mouse model (R6/2). The cysteine residues of striatal proteins from HD and wild-type mice were labeled, respectively, with the heavy and light forms of the ICAT reagents. Samples were trypsinized, uncovered by avidin affinity chromatography, and analyzed by nano-LC-MS/MS. Western blot analyses were used to confirm and to calibrate the ICAT ratios. Linear regression was used to uncover a group of proteins that exhibited consistent changes. In two independent ICAT experiments, we identified 427 cysteine-containing striatal proteins among which ∼66% (203 proteins) were detected in both ICAT experiments. Approximately two-thirds of proteins identified in each ICAT experiment were detected in both ICAT experiments. In total, 68 proteins with altered expressions in HD mice were identified. Elevated expressions of two down-regulated proteins (14-3-3ς and FKBP12) effectively reduced Htt aggregates in a striatal cell line, supporting the functional relevance of the above findings. Collectively by using a well defined protocol for data analysis, large scale comparisons of protein expressions by ICAT can be reliable and can provide valuable clues for identifying proteins critical for pathophysiological functions.

symptoms. The causative mutation is a CAG (glutamine) trinucleotide expansion in exon 1 of the huntingtin (Htt) gene. The normal Htt gene has 35 or fewer repeats in the N-terminal region, whereas the appearance of neurological symptoms is associated with 36 or more CAG repeats (1). The major characteristic of HD is regional degeneration of neurons in the striatum and cortex that leads to movement disorders and dementia (2,3). The toxicity of Htt in specific neurons correlates with the length of polyglutamine expansion (4). The aggregate formation of mutant Htt with poly(Q) expansion causes a wide variety of dysfunctions (5,6). For example, insufficient protein degradation has been proposed as playing a major role (7). Htt aggregates were found to recruit components of protein folding and proteolytic pathways and therefore may suppress functions of the proteasome and heat shock proteins (8 -11). In addition, transcriptional dysfunction caused by mutant Htt is critical for polyglutamine diseases (12,13). Mutant Htt with poly(Q) expansion was shown to sequester and/or interfere with proteins important for the transcriptional machinery including p53, CREB, CREB-binding protein, TAFII130, and SP1 (14 -19). These changes are specific because microarray analyses have revealed that expressions of a great number of other genes are not altered (20). It should be noted that the mechanism underlying the toxicity caused by mutant Htt remains largely controversial (21). Although aggregation is correlated with HD pathogenesis, and many beneficial treatments reduce aggregate formation (22,23), evidence from different laboratories suggests that aggregate formation might confer protective effects against the toxicity induced by soluble mutant Htt with poly(Q) expansion (24,25). Consistent with the above hypothesis, Schaffar et al. (26) demonstrated that monomers or small soluble oligomers of poly(Q)-expanded mutant Htt were suf-ficient to inactivate TATA box-binding protein, an important transcription factor. Collectively the role of Htt aggregates at different stages of HD is complex and remains to be further elucidated.
Present large scale analyses of biochemical changes in HD have mostly been focused at the level of genes. Only a handful of proteomics analyses of brain proteins using two-dimensional gel electrophoresis-based techniques have been conducted on HD (27). We thus set out to examine the global protein expression profiles in the striatum of HD mice using a quantitative proteomics approach (ICAT). This method uses two labeling reagents whose weights are only 8 Da apart for comparative studies (28,29). Recently an acid-cleavable ICAT reagent has begun to be used that avoids several shortcomings of the first generation ICAT reagent (30,31).
In the present study, 427 cysteine-containing striatal proteins were detected from two independent ICAT experiments. Western blot analyses were used to confirm and to calibrate the ICAT data followed by a linear regression analysis to remove the outliers. In total, changes in the expressions of 68 proteins were found. From this it was determined that 6% (four proteins) were up-regulated, whereas 94% (64 proteins) were down-regulated. Among them, disease stage-dependent alterations of 10 proteins in the striatum of HD mice were investigated using Western blot analysis. The functional relevance of two down-regulated proteins  was also demonstrated by their abilities to reduce Htt aggregates. Our study shows that large scale proteomics analysis using ICAT is a reliable approach for systematically identifying proteins critical for the development of HD.

EXPERIMENTAL PROCEDURES
Reagents-All reagents were purchased from Sigma except where specified. Dulbecco's modified Eagle's medium and fetal bovine serum were obtained from Invitrogen.
Animals-Male R6/2 mice and littermate controls were originally obtained from The Jackson Laboratory (Bar Harbor, ME) and were mated to female control mice (B6CBAFI/J). Offspring were identified by a PCR genotyping technique described elsewhere (23). In total, 27 R6/2 transgenic mice and 25 wild-type (WT) littermate control mice were used in this study. Animals were housed at the Institute of Biomedical Sciences Animal Care Facility under a 12-h light/dark cycle. Animal experiments were performed using protocols approved by the Academia Sinica Institutional Animal Care and Utilization Committee, Taiwan.
Nucleus-enriched Protein Preparation-In total, 27 R6/2 transgenic mice and 25 WT littermate mice were used in this study. For each preparation, striatal tissues from two to six mice were removed, resuspended in 1-2 ml of ice-cold buffer A (10 mM Hepes (pH 8), 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, 1 mM Na 3 VO 4 , 20 mM NaF, 100 nM okadaic acid, and 0.5% Nonidet P-40), and homogenized (Dounce, 10 strokes) in buffer A. Lysates were first centrifuged at 367 ϫ g for 1 min at 4°C to remove the debris and then centrifuged at 2292 ϫ g for 5 min at 4°C to collect the pellets as the nucleusenriched fractions and the supernatants as the cytosolic fractions. The pellets were resuspended in 0.5-1 ml of buffer B (20 mM Hepes (pH 8), 425 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, 80 mg/ml phenylmethylsulfonyl fluoride, 1 mM Na 3 VO 4 , 20 mM NaF, 100 nM okadaic acid, and 25% glycerol) and incubated for 60 min on ice. Protein extracts were next collected by centrifugation at 19,275 ϫ g for 5 min at 4°C to remove the pellets. The protein concentration was measured using the Bio-Rad protein assay reagent. This protocol was adopted from earlier studies using the same mouse model of HD at similar disease stages (32) or cell models with large aggregates (19). The majority of Htt aggregates were detected in the nucleus-enriched fraction, not in the cytosolic fraction or the debris (Supplemental Figs. S2 and S3). In addition, Western blot analyses using antibodies against a nuclear marker (lamin A/C) and a cytosolic marker (␣-tubulin) demonstrated the successful enrichment of the nuclear fraction. Only slight contamination of the cytosolic fraction was observed (Supplemental Fig. S1). From striatal tissues of one mouse, ϳ0.4 mg of nucleus-enriched proteins at a concentration of 0.9 -1 mg/ml was collected.
ICAT-In total, 12 R6/2 mice and 10 WT littermate control mice (at 10.5 weeks old) were used in two independent ICAT analyses. The nucleus-enriched protein fractions (1 mg) were prepared as described above and labeled with cleavable ICAT reagents (Applied Biosystems, Foster City, CA) following the manufacturer's protocol. Briefly striatal proteins collected from WT and HD mice were labeled, respectively, with isotopically light and heavy ICAT reagents. The labeled preparations were combined and digested with trypsin (Promega, Madison, WI) overnight at 37°C using an enzyme-to-protein ratio of 1:50 (w/w). The resulting peptide mixture was fractionated by three-step chromatography. Samples were first separated using a Polysulfoethyl A column (Poly LC, Columbia, MD; 2.1 mm ϫ 20 cm) at a flow rate of 200 l/min. Peptides were eluted with 100% buffer A (5 mM KH 2 PO 4 and 25% acetonitrile, pH 3.0) for 2 min followed by a linear gradient from buffer A to buffer B (5 mM KH 2 PO 4 , 350 mM KCl, and 25% acetonitrile, pH 3.0) over 48 min. Fractions were collected (one fraction/min) and subjected to affinity purification using a monomeric avidin cartridge (Applied Biosystems). The purified samples were then treated with acid to cleave the tag and separated by reverse-phase capillary liquid chromatography (Magic C18AQ, Michrom Bioresources, Auburn, CA; 75 m ϫ 11 cm) at a flow rate of 200 nl/min. The eluent was directly analyzed by ion trap mass spectrometry (LCQ Deca XP, Thermo Finnigan, San Jose, CA) under conditions described elsewhere (33). A survey scan followed by three CID events was used. The tolerances of the precursor ion and fragment ion were 2 and 1.5 amu, respectively. Peptide identification by CID was carried out in automated mode using the 3-min dynamic exclusion option.
Data Analysis of ICAT-ICAT-labeled peptides were first analyzed using SEQUEST in Bioworks 3.1 (Thermo Finnigan) (34). The tolerances of the precursor ion and fragment ion were the same as described above. The database was downloaded from the website of the National Center for Biotechnology Information (NCBI; mouse; April 12, 2004; 84,599 entries) and then analyzed by Peptide Prophet Version 1.0 (35), INTERACT_15- 10-2004 (36), and Protein Prophet.pl Version 2.0 (37). Peptide prophet was used to increase the accuracy of peptide identification, and Protein Prophet was used to reduce the redundancy of the search results. The cutoff scores of Peptide Prophet and Protein Prophet were Ն0.9. The false-positive error rate for both ICAT experiments was 0.008. Relative quantification was performed using ASAPRatio 3.0 (38). The above four software packages were kindly provided by Institute for Systems Biology Proteomics Windows Software Distribution. The resulting peptide spectra of proteins were manually checked for qualitative and quantitative results. For protein identification, the peptide sequences were first blasted through the NCBI databases to obtain the corresponding protein sequences, which were then transferred into UniProt to obtained the accession numbers using WU-Blast2 (a tool that effectively finds regions of sequence similarity; www.ebi.ac.uk/blast2). Through the process, 14 proteins were removed due to redundancy or changes in annotation. In three instances where the detected peptides were located in common domains of certain protein families, family names were assigned. The UniProt primary accession numbers of the remaining 203 proteins are listed in Table II and Supplemental  Table S2 to minimize the redundancy. Note that as observed in quite a few published studies and summarized in a recent review (88), a low number of detected peptides per protein is commonly found in quantitative proteomics approaches using chemical tagging-enrichment strategies. In our ICAT study where cysteine was used to tag the proteins, 66 of 203 proteins (ϳ33%) were identified by only one peptide (sequence). This is consistent with other studies using chemical tagging-enrichment reactions and does not reflect the quality of our mass data. Names of the proteins for which only one peptide (sequence) was detected in each ICAT experiment are labeled in green (Table II and Supplemental Table S2). Identification of these proteins needs to be validated using orthogonal methods before reaching a conclusion (Table II and Supplemental Table S2).
Six protein ratios determined by Western blot analysis with a relative standard deviation (RSD) of Յ0.06 were used to calibrate the ICAT ratios as described in detail in the Supplemental Experimental Procedures. Correlations of different batches of samples were calculated using SAS/STAT, Version 8.0 (SAS Institute, Cary, NC).
Western Blot Analysis-In total, 15 R6/2 mice and 15 WT littermates were used in the Western blot analyses. There were seven and eight mice in the groups that were 7 and 10.5 weeks old, respectively. Three to six independent preparations of nucleus-enriched proteins for each condition were prepared from two to three striatal tissues as described above and used for the Western blot analyses. Equal amounts of protein were separated by SDS-PAGE using 10% polyacrylamide gels according to the method of Laemmli (39). The resolved proteins were electroblotted onto Immobilon polyvinylidene difluoride membranes (Millipore, Bedford, MA). Membranes were blocked with 5% skim milk in PBS and incubated with an anti-␤-actin antibody (1:2000 dilution; Santa Cruz Biotechnology), anti-PKC␤ antibody (1:1000 dilution; Transduction Laboratories, Lexington, KY), anti-14-3-3 antibody (1:1000 dilution; Santa Cruz Biotechnology), anti-FKBP12 antibody (1:1000 dilution; Santa Cruz Biotechnology), anti-adducin ␣ antibody (1:1000 dilution; Santa Cruz Biotechnology), anti-casein kinase II␤ antibody (1:1000 dilution; Santa Cruz Biotechnology), anti-CSPG antibody (1:1000 dilution; Chemicon International, Temecula, CA), anti-G␤ 1 antibody (1:1000 dilution; Santa Cruz Biotechnology), anti-G␥ 2 antibody (1:1000 dilution; Santa Cruz Biotechnology), anti-PrxV antibody (1:1000 dilution; Santa Cruz Biotechnology), anti-lamin A/C antibody (1:2000 dilution; Santa Cruz Biotechnology), anti-␣-tubulin antibody (1:3000 dilution; Sigma), anti-Htt antibody (1:1000 dilution; Chemicon International), and anti-V5 antibody (1:2500; Invitrogen) at 4°C overnight followed by the corresponding secondary antibody for 1 h at room temperature. Immunoreactive bands were detected by enhanced chemiluminescence (Pierce) and recorded using Eastman Kodak Co. XAR-5 film. Values for the relative intensities were determined by normalizing the signal of the desired protein with that of the corresponding internal control (lamin A/C) and are expressed as multiples of that of R6/2 mice. Data were generated by quantitative computing densitometry from three to six independent experiments using the image analysis software package ImageQuant Version 3.15 (GE Healthcare).
Cell Culture and Transfection-The striatal progenitor cell line (ST14A) was a generous gift from Dr. E. Cattaneo (University of Milan, Milan, Italy) and was maintained in an incubation chamber gassed with 10% CO 2 and 90% air at 33°C as described previously (40). The day before transfection, cells were seeded onto a 35-mm dish at a density of 2 ϫ 10 5 cells/well. For transfection, cells were grown in wells for 24 h and then exposed to a mixture of 3 l of Lipofectamine 2000 (LF2000, Invitrogen) and 2 g of the desired DNAs for 5 h in serum-free Dulbecco's modified Eagle's medium. Next cells were washed with PBS followed by one more washing with normal growth medium, and then cells were cultured as described above for another 72 h. Filter Retardation Assay-Detection and quantification of SDSinsoluble mutant Htt aggregates were carried out as described elsewhere (41). Briefly cells or striatal tissues were suspended in ice-cold lysis buffer (5 mM Tris-HCl (pH 8.8), 1 mM EDTA, 100 mM NaCl, 5 mM MgCl 2 , 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na 3 VO 4 , 0.5 mg/ml aprotinin, 0.1 mM leupeptin, and 4 mM pepstatin) and homogenized (Dounce, 15 strokes). After centrifugation at 573 ϫ g for 1 min at 4°C, the supernatant was collected into new tubes and then centrifuged at 18,000 ϫ g for 20 min at 4°C. The pellet was resuspended in 100 l of sample buffer (10 mM Tris-HCl (pH 8), 40 mM EDTA, 4% SDS, 100 mM DTT, 5 mM MgCl 2 , and 500 mMCaCl 2 ), mixed with 100 l of 4% SDS, and boiled for 5 min. For slot-blotting, proteins were applied to OE66 membrane filters (0.2-mm pore size, Schleicher and Schuell) in triplicate through a slot-blot manifold (Bio-Rad). Blots were blocked with 5% skim milk in PBS and incubated with an anti-Htt antibody (1:500 dilution; Chemicon International) at 4°C overnight followed by the corresponding secondary antibody for 1 h at room temperature. Immunoreactive bands were detected by enhanced chemiluminescence (Pierce) and recorded using Kodak XAR-5 film.
RNA Isolation and Quantitative Real Time RT-PCR-Total RNA was isolated using the TriReagent kit (Molecular Research Center, Cincinnati, OH), treated with RNase-free DNase (RQ1, Promega) to remove the potential contamination of genomic DNA, and then transcribed into cDNA using Superscript* II reverse transcriptase. Real time quantitative PCR was performed using a TaqMan kit (PE Applied Biosystems) on a TaqMan ABI 7700 Sequence Detection System (PE Applied Biosystems) using heat-activated Taq DNA polymerase (Amplitaq Gold, PE Applied Biosystems). The sequences of primers used are as follows: for 14-3-3 (the target gene), 5Ј-GAGAGAAAC-CTTCTCTCTGTTGCTT-3Ј and 5Ј-CCAAGTAACGGTAGTAGTCAC-CCTT-3Ј; for FKBP12 (the target gene), 5Ј-GCTTGAAGATGGAAA-GAAATTTGAT-3Ј and 5Ј-TCCATAGGCATAGTCTGAGGAGATT-3Ј; and for GAPDH (the reference gene), 5Ј-TATCCGTTGTGGATCTGA-CAT-3Ј and 5Ј-ACAACCTGGTCCTCAGTGTA-3Ј. Independent reverse transcription-PCRs were performed as described elsewhere (19). The relative transcript amount of the target gene, which was calculated using standard curves of serial RNA dilutions, was normalized to that of GAPDH of the same cDNA.

RESULTS AND DISCUSSION
To systematically identify proteins important for the disease progression of HD, we applied ICAT technology to quantitatively assess the overall protein expression profiles in the striatum of 10.5-week-old HD mice, an age at which several symptoms of HD have already become manifest (e.g. deficiency in motor coordination, body wasting, and Htt aggregate formation) (23). Previous studies using two-dimensional gel electrophoresis (2DGE) also showed that the levels of oxidatively modified proteins are decreased in the brains of 10-week-old HD mice (27). To reduce the complexity and thus improve the ICAT quality as reported earlier (42), we chose to determine only the nucleus-enriched fraction of the striatum.
The nuclear preparation method used in the present study has been widely used in similar studies of HD mice at a late stage of disease progression (4) and for cell models with large aggregates (19). Western blot analyses using antibodies against a nuclear marker (lamin A/C) and a cytosolic marker (␣-tubulin) demonstrated successful enrichment of the nuclear fraction. Only slight contamination by the cytosolic fraction was observed (Supplemental Fig. S1). Western blot analyses revealed that there was no change in the protein level of lamin A/C in the striatum of R6/2 mice (98.9 Ϯ 2.6% of those in WT mice, mean Ϯ S.D. values from eight independent preparations); we therefore used lamin A/C as an internal control for the following Western blot analyses.
Two independent ICAT analyses of the nucleus-enriched, striatal proteins collected from R6/2 and wild-type mice were carried out. Approximately 380 proteins were identified in each experiment. Nearly two-thirds of the identified proteins in each ICAT experiment were found in both experiments (Fig.  1A). In total, 513 unique proteins were identified. Due to nonspecific binding of peptides in affinity chromatography, only 83% (427 of 513) of the identified peptides contained cysteine(s). The numbers of cysteine-containing proteins identified in experiments I and II were 307 and 323, respectively. Quantitation was only performed for proteins containing cysteine residues. Again nearly two-thirds (203 proteins; Supplemental Table S2) of cysteine-containing proteins were detected in both experiments (Fig. 1B) and were further analyzed using an automated algorithm (ASAPRatio (38)) to produce a ratio value (WT/HD) for each protein (designated Ratio ASAP ). The average RSD of the Ratio ASAP for experiment I and experiment II was 0.23.
Proper normalization procedures are critical for high throughput, quantitative proteomics approaches (such as ICAT). To the present, the most widely adapted algorithms (including the ASAPRatio) are designed based on the assumption that the major protein population in a proteome is unchanged, which allows normalization by aligning the center of the peptide ratio distribution peak to 1 (or log 10 Ratio ϭ 0; Supplemental Fig. S4A). However, in more severe degenera-tive diseases (such as HD) where the levels of many proteins may be altered during disease progression, the peptide ratio distribution peak is likely to be shifted in one direction (Supplemental Fig. S4B). Under such conditions, normalization by aligning the center of the ratio distribution peak to 1 as described above might be inappropriate and cause the quantitation to be less precise. To correct for the deviation due to this assumption, we performed Western blot analyses of 10 proteins whose specific antibodies were available to determine their expression ratios (designated Ratio W ; Table I and Fig. 2) as the internal calibrants. The Western blot approach was chosen as an orthogonal method to verify the fidelity of the ICAT analysis because it is commonly available and practiced in most laboratories. To ensure a tight calibration, only Ratio W values of six proteins with an RSD of Յ0.06 (i.e. 6%) were used to calibrate the Ratio ASAP value. The corrected expression ratio was designated Ratio C . As shown in Table I, when 1-fold average RSD (0.23) was used as the criteria to define a meaningful change, this calibration did not change the direction of alteration but only adjusted the magnitudes of the changes. In addition, the Ratio C values for most proteins examined were closer to the corresponding Ratio W values than their original Ratio ASAP values, indicating a moderate improvement in quantitation. To the best of our knowledge, algorithms appropriate for normalizing disease proteomes, in which the levels of a significant portion of proteins are altered, are currently unavailable and are in great demand. It should be emphasized that although the degree of changes for those markedly down-regulated proteins (i.e. ␤-actin, FKBP12, and 14-3-3) were underestimated when compared with their Ratio W , ICAT analyses of the 10 proteins examined correctly estimated the direction of the alteration (Table I). Thus, we applied this calibration to all 203 cysteine-containing proteins detected in both ICAT experiments (Supplemental Table S2) and used the calibrated Ratio C for further analyses.
Note that the ratio of each individual protein is composed of several different peptide pairs and/or the same peptide pairs with different charge states. Such variations in the protein ratio are highly dependent on the level of proteins existing in the preparation. Generally the greater the amount of a desired protein there is, the less variation there is. Careful manual inspection of the mass spectra greatly improves the data quality. Nonetheless the RSD of each individual protein usually varied between the independent ICAT experiments, and this might have interfered with the selection of candidate proteins. For example, the ratios of PrxV in experiment I and experiment II were 0.63 Ϯ 0.19 and 0.60 Ϯ 0.47, respectively (Table I). If individual RSDs were taken into consideration, PrxV might not have been considered a candidate protein.
However, because the mean values of this protein in HD mice from both ICAT experiments were well correlated, we suggest that PrxV might be regarded as a potential candidate protein for further evaluation. The Western blot analysis of PrxV validated our prediction (Fig. 2H). Our results suggest that du- plication of ICAT analyses is highly recommended and that correlation between two sets of ICAT data might provide an additional means to increase the confidence level.
We thus conducted a linear regression analysis of the logtransformed WT/HD ratios (i.e. Ratio C ) of experiment I and experiment II (Fig. 3A). The Pearson correlation coefficient was 0.63, and the p value was Ͻ0.0001. The ratios of the two independent ICAT experiments were thus significantly positively correlated. The p value from Student's t test was 0.77 (Ͼ0.05), suggesting that results from experiment I and experiment II did not significantly differ. We also performed a linear regression on the ratios of both experiments (r 2 ϭ 0.5070). When Ϯ1.96 was set as the upper and lower limits, 18 outliers were found. Among them, six proteins exhibited opposite expression ratios in the two ICAT experiments, four proteins showed alteration in only one ICAT experiment, whereas the remaining eight proteins were reproducibly regulated toward the same direction in both ICAT experiments. The reason that the latter eight proteins are considered to be outliers is because their expression ratios in the two ICAT experiments significantly differed in magnitude; this might have been due to weak signal intensities in the MS analysis. Removal of the 18 outliers improved the data quality and provided a better correlation between the duplicated ICAT experiments (Fig. 3B). To increase the chance of identifying proteins with changes, the eight outliers that showed the same direction of regulation in the two ICAT experiments were included for consideration. Note that these eight proteins have a relatively high risk of being false positives compared with the other candidate proteins that were not rejected by the linear regression process.
After removing 10 outliers as described above, we analyzed the remaining 193 proteins that showed alterations in the same direction in the two independent ICAT experiments for meaningful changes in their expression levels. Based on the results of Western blot analyses of 10 different proteins, we found that a 1-fold average RSD (0.23) was an appropriate criterion to define a meaningful change (Table I). We therefore listed proteins with a Ratio C of either Ն1.23 or Յ0.77 in Table  II. A few proteins, with little possibility of existing in the nucleus and that might have been present due to contamination from other cellular fractions, were removed from the table. Of the 68 proteins, four proteins were up-regulated, whereas the remaining 64 proteins were down-regulated in the striatum of HD mice. To the best of our knowledge, only seven of them (PKC␤II, calcium/calmodulin-dependent protein kinase type II, enolase, triose-phosphate isomerase, aconitase, creatine kinase, and cytochrome c) have been implicated in HD (27,(43)(44)(45)(46). Based on information obtained from protein knowledgebases of Swiss-Prot and TrEMBL (www.expasy.org), NCBI (www.ncbi.nlm.nih.gov/protein), European Molecular Biology Laboratory-European Bioinformatics Institute (EMBL-EBI) (www.ebi.ac.uk/InterProScan), PSORT II (89), and PubMed (www.ncbi.nlm.nih.gov/PubMed), the differentially regulated proteins fall into 11 categories, excluding six proteins whose functions are unknown at this time (Table II). In total, 64 down-regulated proteins and four up-regulated proteins were found in the striatum of 10.5-week-old HD mice. This finding is consistent with an earlier study using a microarray in which the number of down-regulated transcripts greatly outnumbered up-regulated transcripts (20). These results support the hypothesis that at the symptomatic stage of HD, alterations in protein expressions are tilted toward the side of down-regulation. The major categories in the down-regulated groups were signaling molecules and those involved in general metabolism (Table II and Fig. 4). In a recent report by Perluigi et al. (27), nine proteins from the striatal lysate of HD (R6/2) mice aged 10 weeks were identified using a combination of 2DGE and MS. Among them, six proteins were oxidized and inhibited. Interestingly levels of three suppressed proteins (enolase, aconitase, and creatine kinase) were also found to be reduced in our study (Table II). Consistent with prior reports that ICAT is capable of identifying more proteins than 2DGE (47), we found ϳ22-fold more striatal proteins in

Calibration of the ICAT ratio (WT/HD; Ratio ASAP ) using results from Western blot analyses
Expression ratios from two independent ICAT analyses and three to five Western blot analyses are shown as indicated. Ratio W , ratio detected by Western blot analysis; Ratio ASAP , ratio adjusted by the ASAPRatio C ; Ratio C , Ratio ASAP calibrated by the Ratio W ; RSD, S.D./ratio. If applicable, the abbreviation for each protein used in the text is shown in parentheses.
the present study using ICAT (203 proteins; Supplemental Table S2). In addition, the number of striatal proteins with changes herein was 68 (Table II) and was about 11-fold that reported in the study using 2DGE (six proteins (47)). Note that the age of the HD mice in the earlier study was slightly younger than that used in the present study and that both approaches (ICAT and 2DGE) have their unique limitations that prohibit them from identifying all proteins present. In the future, it would be very interesting to combine these two compensatory approaches and identify more proteins exhibiting changes during HD progression.
As described above, we performed Western blot analyses of 10 proteins to verify and calibrate the expression profiles revealed by ICAT. Among them, results from both ICAT and Western blot analyses suggested that three (CSPG, G␥ 2 , and PrxV) were up-regulated, whereas five (PKC␤, ␤-actin, FKBP12, 14-3-3, and casein kinase II␤) were down-regulated in the striatum of 10.5-week-old HD mice (Table I). Slight changes in two proteins (G␤ 1 and adducin 1␣) were detected using Western blot analyses but were not evident in the ICAT analyses when a 1-fold average RSD (0.23) was used as the criterion to define a meaningful change. It appears that the ICAT approach as analyzed and calibrated herein is not suitable for the identification of proteins exhibiting very limited changes.
We next determined the functional relevance of the above findings with ICAT. Western blot analyses revealed that, except for FKBP12, changes in none of the other proteins in the striatum HD mice were observed at the age of 7 weeks when most symptoms are not yet evident. Down-regulation of FKBP12 in HD mice was moderate at 7 weeks old and was reduced to only ϳ20% of that of WT mice at the age of 10.5 weeks. Regulation of these proteins in the striatum therefore appears to be closely associated with disease progression of HD.
Among these proteins, 14-3-3 is of the greatest interest because it is a scaffold that interacts with more than 100 proteins important for transcriptional/translational control, the cell cycle, apoptosis, intracellular trafficking, ion channel regulation, and synaptic transmission (48,49). Suppression of 14-3-3 as found in the present study therefore might jeopardize multiple functions in mutant Htt-expressing cells. Most   FIG. 3. Scatter plots demonstrating the strategy used to verify the results from the two independent ICAT experiments. A, the log-transformed WT/HD ratios of each identified protein from experiment I (Exp-I) and experiment II (Exp-II) are plotted on the y and x axes, respectively. B, the outliers that deviated from the linear regression (by Ϯ1.96 ) were removed.

TABLE II Lists of striatal proteins up-regulated or down-regulated in 10.5-week-old HD mice compared with WT mice revealed by ICAT
Expression ratios (WT/HD) from two independent ICAT analyses are shown as indicated. Only proteins with changes larger than the average RSD (i.e. Ratio C of Ն1.23 or Յ0.77) are listed. The alteration of eight proteins (in bold) was also verified by Western blot analyses (Fig. 2). The outliers whose expressions were altered in the same direction in the two ICAT experiments are included for consideration. Their accession numbers are labeled in gray. Names of the 24 proteins for which only one peptide (sequence) was detected in each ICAT experiment are labeled in green. Peptides that were not used in the quantitative analyses are labeled in orange. Note that these proteins were identified from a nucleus-enriched preparation and are likely to be located in the nucleus. Nevertheless their nuclear localizations need to be further validated. Ratio ASAP , ratio adjusted by the ASAPRatio; Ratio C , RatioASAP calibrated by the Ratio W ; Ratio W , ratio detected by Western blot analysis (Table  I); RSD, S.D./ratio; C*, cysteine labeled with heavy reagent; M#, oxidized methionine. The spectrum for each peptide is illustrated in Supplemental Table S3. CaMKII, calcium/calmodulin-dependent kinase II; SH3, Src homology 3. intriguingly, 14-3-3 was found to be associated with molecular chaperons such as heat shock proteins (50 -52). Elevation of several heat shock proteins has been found to reduce the aggregation and toxicity of mutant Htt (8,53). We thus hypothesized that the expression level of 14-3-3 might play an important role in the extent of Htt aggregate formation. Indeed results of the filter assay demonstrated that elevated expression of 14-3-3 markedly reduced the aggregation of mutant Htt with expanded poly(Q) (Fig. 5, A and B) and might lead to beneficial effects on HD progression. Indeed in a recent review by Kaneko and Hachiya (54), 14-3-3 was hypothesized as selectively recognizing and segregating misfolded proteins and therefore can be used to protect cells against toxicity caused by aggregation. Another protein that caught our attention was FKBP12 whose degeneration in the striatum of HD mice was detected as early as 7 weeks old (Fig. 2B). FKBP12 is a binding protein of FK506 that is able to protect neurons against glutamate excitotoxicity and transient oxygen-glucose deprivation via an antiexcitotoxic effect (55,56). In addition, like other FKBP members (FKBP51 and FKBP52), FKBP12 might also function as a chaperone and regulate the expressions of heat shock proteins (57). To assess its functional importance, we expressed FKBP12 in a striatal cell line harboring a mutant Htt with expanded poly(Q). Elevated FKBP12 expression reduced the formation of mutant Htt aggregates (Fig. 5, C and D). In contrast, expression of Lim (a molecule important for early development (58)) did not affect Htt aggregate formation (Supplemental Fig. S5). The mechanisms used by 14-3-3 and FKBP12 to reduce aggregates are currently unknown, and determining these was outside the scope of this study. Instead the above findings demonstrate that the list of molecules revealed by ICAT (Table II) is likely to be functionally relevant and may provide valuable information for the development of HD therapies.

ICAT Analysis of Striatum of Mice with Huntington Disease
Besides the intriguing functional role of 14-3-3 and FKBP12, the ICAT analyses also provided a list of proteins that might play critical roles in the pathology of HD. For example, the reduced expression of PKC␤II protein revealed by ICAT is in agreement with previous studies demonstrating that the transcription of PKC␤II is reduced in the striatum of both HD patients and HD mice (44,59). Low PKC levels might cause deficits in long term potentiation and impairments in corticostriatal synaptic plasticity in HD (60,61). Another very interesting observation was the reduced expression of nuclear actin, which plays a critical role in RNA polymerase II-based transcription. Down-regulation of nuclear actin might alter the nucleoskeleton and cause defective chromatin remodeling (62,63). In HD, microtubule destabilization is a primary event that precedes transcriptional dysfunction. Prevention of cytoskeletal dysfunction improves cell survival upon Htt-induced toxicity (64,65). Consistent with the proposed importance of the nuclear structure and cytoskeleton, results of our ICAT analyses suggested that the expressions of at least eight other structurally related proteins were also reduced (Table II). Functional annotation suggests that these molecules might act as cytoskeleton-associated proteins (66 -68) and that their reduction might exacerbate the microtubule disability induced by mutant Htt.
In addition, the level of nuclear cytochrome c in the striatum of HD mice was reduced (Table II). Nuclear accumulation of cytochrome c has been implicated in the remodeling of chromatin and acetylation of histone proteins (69). Reduced nuclear cytochrome c thus might contribute to the aberrant acetylation/deacetylation of histone proteins observed in HD (70,71). The marked decrease in nuclear phosphodiesterase 2A (PDE2A) is also of great interest because PDE2A catalyzes the breakdown of cAMP and cGMP. Selective suppression of other phosphodiesterase isozymes (PDE10A and PDE1B) in HD was reported earlier (72). Reduced expression of PDEs in HD is thought to cause dysfunction of cAMP-and cGMPregulated functions. Several lines of evidence also strongly imply that energy dysfunction exists in HD. Altered expressions of several proteins involved in energy metabolism were found. As shown in Table II, dysregulated energy proteins include those involved in glucose metabolism (e.g. enolase and triose-phosphate isomerase), the tricarboxylic acid cycle (aconitase), and ATP production (hexokinase and creatine kinase) (27,46,73). The impairment of energy metabolism and glycolysis appears to contribute to the neurodegenerative processes of HD. The dramatic decrease in the level of a protein similar to glyceraldehyde-3-phosphate dehydrogenase (GAPDH-like protein) was equally intriguing because GAPDH (a glycolytic enzyme) was the first Htt-interacting protein reported, and inhibition of GAPDH causes striatal lesion (74,75). Note that GAPDH seems to exhibit functions other than glycolysis. For example, translocation of GAPDH FIG. 4. Classification of down-regulated striatal proteins in HD mice. As listed in Table II, proteins with reduced levels of striatal expression in HD mice at the age of 10.5 weeks old when compared with those of wild-type mice were included. For proteins with multiple functions, the best known function is assigned.
into the nucleus occurs during apoptosis (76). In addition, overexpression of GAPDH enhances the nuclear translocation and toxicity of mHtt (77). The GAPDH-like protein (156 amino acids) is much smaller than GAPDH (333 amino acids) with which it shares high homology (83% identity in amino acids). Further experiments are required to define the function/activity of the GAPDH-like protein and understand the functional implications of its marked suppression in HD mice. Collectively our ICAT studies provide valuable information for further functional characterization.
Ample evidence suggests that formation of mutant Htt re-sults in transcriptional dysregulation by either forming aggregates or causing aberrant protein-protein interactions with a number of important transcription factors or coactivators (14 -19). Indeed results of the quantitative RT-PCR analyses demonstrated that the transcription of 14-3-3 and FKBP12 were also down-regulated in the striatum of 10.5-week-old R6/2 mice (Fig. 6), suggesting that suppression of gene expression might contribute to the reduced protein expression in the nucleus. Consistent with this hypothesis, the protein levels of 14-3-3 and FKBP12 in the cytosolic fractions were also reduced (Supplemental Fig. S3) indicating an overall reduction in their protein levels. We also compared our ICAT experiments with an earlier study in which microarray analyses of the striatum of 6-and 12-week-old R6/2 mice were conducted. Consistent with our finding of a reduced level of PKC␤ in the striatum of 10.5-week-old R6/2 mice (Fig. 2 and Table II), the transcript level of PKC␤ was also reduced in the striatum of 12-week-old HD mice. In contrast, the transcript levels of several proteins (including ␤-actin, triose-phosphate isomerase, lamin B2, calcium/calmodulin-dependent kinase II␤, casein kinase II␤, hexokinase, and creatine kinases) that showed downregulated nuclear expression in the present ICAT study were not altered (20). Additional mechanisms such as a disturbance in nuclear-cytoplasmic shuttling, a deficient protein degradative system, sequestration by Htt aggregates, and malfunctions of other post-transcriptional regulations might mediate the reduced nuclear expression of proteins reported herein.
Although the presence of one or more cysteines is a prerequisite for detection by ICAT, we found that more than 92% of the identified peptides contained only one cysteine, whereas those with three cysteines comprised less than 1% (data not shown). The low ratio of peptides with multiple cysteines might have been due to the difficulty of elution in the affinity purification step. In addition, the length distributions of peptides identified by ICAT are shown in Fig. 7A. The average lengths of peptides identified by ICAT were 13.8 and 13.2 amino acid residues/peptide in experiment I and experiment II, respectively. Approximately 80% of the identified peptides were in the range of 8 -18 residues. The peptide length profile of our study is very similar to the average length of tryptic peptides of the human proteome (78). Peptides shorter than eight amino acids are not specific enough for protein identification and can easily be lost in the separation when using multidimensional liquid chromatography. For peptides longer than 18 amino acids, the collision energy for fragmentation in MS/MS is insufficient and usually results in poor product ion spectra for subsequent identification. The observations that ICAT detects mostly peptides containing one cysteine and that are 8 -18 amino acids long might explain, at least partially, why there were so few factors involved in the transcription machinery in the list of ICAT data. In addition to the relatively low abundance of most transcription factors and cofactors, their tryptic peptides might not contain the right number of cysteines. Hypothetical tryptic profiles of four transcriptional factors/coactivators (i.e. 14-3-3, SP1, c-Fos, and STAT3) are illustrated in Fig. 7B. Based on the criteria described above, 14-3-3 was detected because it contains four peptides in the detectable range. In contrast, SP1 contains no detectable peptides, and c-Fos has only two detectable peptides that can easily be missed in a complex peptide mixture. STAT3 contains many peptides in the detectable range. However, we did not detect STAT3 using ICAT in our study, probably due to its low abundance. Because ICAT has been reported to be strongly biased toward acidic proteins (79), basic proteins, such as CREB-binding protein, c-Jun, and BTEB3, are expected to be absent from the ICAT database as reported herein. Complementary approaches (i.e. 2DGE) and/or further enrichment (80) are necessary if the goal is to globally characterize the transcription machinery.
Among the 68 proteins with altered expressions observed by ICAT, only four (CSPG, G␥ 2 , PrxV, and lamin B2) were up-regulated in the striatum of R6/2 mice, whereas the other 64 proteins were down-regulated. This distinct difference between the numbers of proteins up-regulated and down-regulated might be due to the overall degeneration at the symptomatic stage of HD, and this certainly caused difficulties in normalizing the HD proteome as described above. For upregulated proteins, CSPG has been shown to enhance neurite outgrowth of cortical neurons through the phosphatidylinosi-FIG. 7. Profiles of peptides identified by ICAT. A, size distribution of peptides identified in the two independent ICAT experiments. The sizes of most of the identified peptides ranged from 8 to 18 residues. B, predicted numbers of cysteine-containing peptides of the four transcription factors/coactivators upon trypsin cleavage. The name of the transcription factor/coactivator is listed in the upper right-hand corner of each panel. Note that ICAT detected only peptides containing cysteine(s). In addition, the greater number of cysteine-containing peptides a protein is composed of, the higher probability there is for the protein to be detected by ICAT. tol 3-kinase and PKC pathways (81). Its up-regulation in the striatum of HD mice might reflect a compensatory regulation in degenerated neurons. Another up-regulated protein, PrxV, has been shown to exhibit an antioxidant effect and is able to protect cells against elevated oxidative stress and DNA damage (82,83). Elevation of PrxV in the striatum of HD mice during disease progression therefore is likely to be part of a defense mechanism against excitotoxicity in HD mice. It was intriguing to find that G␤ 1 and G␥ 2 (from the Western blot analyses; Fig. 2, G and I) are increased in the striatum of HD mice. Both G protein subunits are primarily located in the plasma membrane and mediate signal transduction of G protein-coupled receptors. Nonetheless nuclear G proteins have been implicated in the regulation of mitosis, transcription, and nuclear signal transduction (84,85). Immunostaining of the striatum with an anti-G␤ 1 antibody clearly demonstrated localization of G␤ 1 in the nucleus of striatal cells (Supplemental Fig. S6). It will be of great interest to characterize the roles of elevated G protein subunits in the striatum of HD mice.
In summary, we performed the first ICAT analysis of the striatum of HD mice. Calibration using results obtained from Western blotting plus a linear regression analysis of two independent ICAT experiments greatly improved the data reliability. Our analyses have disclosed a list of proteins whose expression levels are altered by mutant Htt and thus might contribute to the pathology of HD (86,87). Recent reports suggest that combinations of various beneficial agents with different underlying mechanisms might prove to be promising for HD patients. The present study extends our current knowledge of the multiple pathways underlying HD pathologies and might eventually lead to the development of combination treatments based on different mechanisms.