normal and expanded huntington's disease gene alleles produce

Normal and Expanded Huntington'sDisease Gene Alleles ProduceDistinguishable Proteins Due toTranslation Across the CAG Repeat

Francesca Persichetti,* Christine M. Ambrose,* Pei Ge,tSandra M. McNeil,* Jayalakshmi Srinidhi,*Mary Anne Anderson,* Barbara Jenkins,* Glenn T. Barnes,*Mabel P. Duyao,* Lisa Kanaley,* Nancy S. Wexler,§Richard H. Myers,1' Edward D. Bird,* Jean-Paul Vonsattel,tMarcy E. MacDonald,* and James F. Gusella*#*Molecular Neurogenetics Unit, Massachusetts General Hospital,Charlestown, Massachusetts, U.S.A.tLaboratory for Molecular Neuropathology, Massachusetts GeneralHospital, Charlestown, Massachusetts, U.S.A.tBrain Tissue Resource Center, McLean Hospital, Belmont,Massachusetts, U.S.A.§Hereditary Disease Foundation, Santa Monica, California, U.S.A.I'Department of Neurology, Boston University Medical School, Boston,Massachusetts, U.S.A.#Department of Genetics, Harvard Medical School, Boston,Massachusetts, U.S.A.

ABSTRACT

Background: An expanded CAG trinucleotide repeat isthe genetic trigger of neuronal degeneration in Hunting-ton's disease (HD), but its mode of action has yet to bediscovered. The sequence of the HD gene places the CAGrepeat near the 5' end in a region where it may betranslated as a variable polyglutamine segment in theprotein product, huntingtin.Materials and Methods: Antisera directed at aminoacid stretches predicted by the DNA sequence upstreamand downstream of the CAG repeat were used in West-ern blot and immunohistochemical analyses to examinehuntingtin expression from the normal and the HD allelein lymphoblastoid cells and postmortem brain tissue.

Results: CAG repeat segments of both normal and ex-panded HD alleles are indeed translated, as part of adiscrete -350-kD protein that is found primarily in thecytosol. The difference in the length of the N-terminalpolyglutamine segment is sufficient to distinguish nor-mal and HD huntingtin in a Western blot assay.Conclusions: The HD mutation does not eliminate ex-pression of the HD gene but instead produces an alteredprotein with an expanded polyglutamine stretch nearthe N terminus. Thus, HD pathogenesis is probably trig-gered by an effect at the level of huntingtin protein.

Address correspondence and reprint requests to: James F.Gusella, Molecular Neurogenetics Unit, Massachusetts Gen-eral Hospital East, Building 149, 13th Street, Charlestown,MA 02129-2060, U.S.A.

374 Copyright C 1995, Molecular Medicine, 1076-1551/95/$10.50/0Molecular Medicine, Volume 1, Number 4, May 1995 374-383

F. Persichetti et al.: Translation of CAG Repeat in HD

INTRODUCTIONHuntington's disease (HD) is an inherited neuro-degenerative disorder characterized by progres-sive choreic movements, psychiatric changes,and cognitive decline (1,2). These manifestationsresult from neuronal cell death, principally in thebasal ganglia, where medium spiny neurons arethe most vulnerable targets. Within the neostri-atum, cell loss follows a characteristic gradientproceeding along the postero-anterior, dorso-ventral, and medio-lateral axes (3). HD affectsabout 1 in 10,000 individuals in the UnitedStates and is caused by a dominant genetic defectthat was mapped to 4pl6.3 by linkage analysis(4) and recently identified by location cloning (5).

Virtually all individuals with authenticatedHD possess an unstable, expanded CAG trinucleo-tide repeat (reviewed in Refs. 6 and 7). In normalindividuals, this repeat is polymorphic, varyingfrom 6 to 34 trinucleotide units, and is inheritedin a mendelian fashion. By contrast, the CAGrepeat on disease chromosomes comprises 37 tomore than 120 units, and shows a mutation rateapproaching 1 through meiotic transmission.CAG repeat length is inversely correlated withage at onset of both neurologic and psychiatricsymptoms, and with age at death. Most individ-uals show onset in mid-life due to repeat lengthsin the range of 40-50 units, with a minoritydisplaying either juvenile onset (typically >60units) or onset very late in life (usually 37-40units).

The mechanism whereby CAG repeat expan-sion causes neuronal death has not been estab-lished, but a number of observations favor aneffect at the level of huntingtin, the HD gene'sprotein product (8). Huntingtin mRNA is notexpressed exclusively or even preferentially inthe affected brain regions, but is present in alltissues tested (9-11). RNA is produced from bothnormal and disease alleles, and contains the CAGrepeat segment. Thus, the expanded CAG repeatis most likely to affect production of huntingtinfrom the disease allele, or to alter its structure,localization, or activity (5,6,12). We have usedantisera directed against potential translated se-quences both upstream and downstream of theCAG repeat to investigate the expression of hun-tingtin from normal and HD alleles. Our dataestablish that the CAG segment is indeed trans-lated into a predicted polymorphic polyglu-tamine stretch near the N terminus of hunting-tin. The expanded repeat of the disease alleledoes not eliminate production of this -350-kD

protein. Rather, the additional glutamine resi-dues contributed by the mutation make the HDhuntingtin larger than, and therefore distin-guishable from, its normal counterpart.

MATERIALS AND METHODSPeptides, Fusion Proteins, and PolyclonalAntibody ProductionPeptidesHP1 (PAVAEEPLHRPKKELSATKKDRVNHCLTICENIV; residues 80-113 based on L27416in GenBank) and HP12 (ATLEKLMKAFESLKSF;residues 2-17) were synthesized and purified byThe Peptide Laboratory, Inc. (Berkeley, CA,U.S.A.), or Multiple Peptide Systems (San Diego,CA, U.S.A.). A 1.8-kb BamHI restriction frag-ment from the ITI 5B cDNA was subcloned intothe pGEX vector (Pharmacia, Piscataway, NJ) toexpress a 92-kD huntingtin glutathione S-trans-ferase fusion protein (huntingtin residues 1981-2580) in Escherichia coli. Polyclonal rabbit anti-sera were raised to keyhole limpet hemocyanin(KLH)-coupled peptides and the purified fusionprotein by Assay Research, Inc. (MD, U.S.A.),and East Acres Biologicals (Southbridge, MA,U.S.A.). Preimmune sera were obtained beforethe immunization protocol was initiated (13).

Cultured Lymphoblast Cell Lines andPostmortem Human Brain Tissue

Lymphoblast cell lines from HD families of vari-ous ethnic backgrounds used for genetic linkage,and disequilibrium studies (14,15) have been es-tablished (16) in the laboratory. Human post-mortem brain tissue was collected by the BrainTissue Resource Center, McLean Hospital (Bel-mont, MA, U.S.A.). At autopsy, one hemisphereof the brain was fixed and used for the evalua-tion of HD neuropathology according to thegrading system devised by Vonsattel (3). Theother hemisphere was dissected into small blocksof tissue representing various brain regions.

DNA Extraction and HD CAG RepeatLength DeterminationDNA was prepared from cultured cells and frompostmortem cerebellar brain tissue as describedpreviously (4). The length of HD CAG alleles wasassessed using polymerase chain reaction (PCR)amplification of the CAG stretch (17), excluding

375

376 Molecular Medicine, Volume 1, Number 4, May 1995

the polymorphic CCG repeat adjacent to theCAG (18).

Protein Extraction andCellular FractionationProtein extracts were made from lymphoblastcell pellets and fresh postmortem brain tissue byhomogenization in the presence of protease in-hibitors (1 mM DTT, 1 mM phenylmethanesul-fonyl fluoride [PMSF], 0.1% leupeptin, and0.1% aprotinin) in hypotonic lysis buffer con-taining 10 mM Hepes pH 8.3, 1.5 mM magne-sium chloride (19), or 50 mM Tris pH 7.5, 10%glycerol, 5 mM magnesium acetate, 0.2 mMEDTA (20), respectively. Protein concentrationswere determined with the Protein Assay Kit(Bio-RAD, Richmond, CA).

Subcellular fractions were produced by dif-ferential centrifugation of cell lysates. Nucleiwere collected by centrifugation at 2000 rpm andextracted in nuclear extraction buffer containing30 mM Hepes pH 8.3, 0.3 mM EDTA, 450 mMsodium chloride, 12 mM magnesium chloride, 1mM PMSF, 1 mM DTT, 0.1% leupeptin, and0.1% aprotinin to produce the soluble nuclearfraction (N). The postnuclear supernatant (PN),was fractionated by centrifugation at 100,000 xg to produce the 100,000 X g supernatant, SiO0(S), and the 100,000 X g pellet P100 (P).

Western BlottingExtracts (-200 jig of protein) in sample loadingbuffer (62.5 mM Tris HCI pH 6.8, 1% SDS, 0. 1 5mg/ml bromophenol blue dye, 5% glycerol,2.5% 2 mercaptoethanol) were subjected to elec-trophoresis in SDS running buffer (25 mM Trisbase, 192 mM glycine, and 0.1% SDS) for 15 hrat 70V on a 6% polyacrylamide gel or for theseparation of normal and HD huntingtin at 95 Von a 5% gel. Rainbow high molecular weightrange protein markers, 14.3-200 kD (AmershamCorp., Arlington Heights, IL, U.S.A.) were alsoloaded on each gel. Proteins were transferred tonitrocellulose membrane by electroblotting at30V for 6 hr in 25 mM Tris base, 192 mM glycine,and 20% methanol.

For immunoblot analysis, Western blotmembranes were preincubated with TBST (20mM Tris HCI pH 8.0, 150 mM NaCl, 0.1% Tween20) supplemented with dry milk powder andthen probed with the primary rabbit antibody for2 hr. The specific interactions were visualizedwith biotinylated horseradish peroxidase cou-

pled anti-rabbit immunoglobulin (VECTASTAINABC kit) using the enhanced chemilumines-cence (ECL) Western Blotting Detection SystemKit (Amersham) according to the instructions ofthe manufacturer. Blots were exposed to X-rayfilm (XAR, Kodak, Rochester, NY, U.S.A.).

ImmunohistochemistryHuman postmortem brain tissue was dissected,fixed in 4% paraformadehyde and cryoprotectedin sucrose. Frozen 7-,M cryostat sections,mounted on gelatin subbed slides, were pro-cessed for immunohistochemistry using standardprotocols (21). Immune complexes were de-tected using the avidin-biotin method employingbiotinylated horseradish peroxidase H anddiaminobenzidine substrate (VECTASTAIN ABCkit) according to the instructions of the manu-facturer (Vector Laboratories, Inc., Burlingame,CA). The immunostained sections were dehy-drated and counterstained with hematoxylin.Microphotographs were taken with Ektachrome64T slide film at a magnification of 630X.

RESULTSThe HD CAG Repeat Is TranslatedHuntingtin mRNA contains a long open readingframe with the polymorphic CAG segment nearits 5' end. Translation from a potential initiatorATG located 17 codons upstream of the CAGwould predict a large product of -3130 residueswith a mass of -350 kD. If translation wereinitiated downstream at the first potential in-frame ATG after the CAG repeat, a smaller prod-uct of 3014 amino acids with a mass of 333 kDwould be predicted. To determine whether theCAG segment is translated, we prepared rabbitpolyclonal antisera against two peptides, and onebacterial fusion protein representing differentportions of the largest predicted product. Theantipeptide antisera, HP1 and HP12, are directedagainst amino acids 80-113 and 2-17, respec-tively (based on L27416 and L12392). AntiserumHF1 was prepared using amino acids 1981-2580expressed in E. coli as a fusion with glutathione-S-transferase (GST).

Figure 1 shows the results of Western blotanalyses of normal and HD lymphoblastoid cellextracts using HP1, HP12, and HF1. In Panel A,all three immune sera detect a large -350-kDprotein that is not revealed by their correspond-


AHPl

I IPre I

hunUngmn-*O

200 kDa-

CHP1

I a

(CAG),

huntingtin I

HP12I I

HP12I

65 54

HF1 lFI IPro I

II-

HF1Il

BHPII +f

hun20n0 nk.- -

200kDa--

HP12 HFII _ + I I _ + I

FIG. 1. Antisera specific for different portions of huntingtin detect a -350 kDa protein inlymphoblast cells(A) Detection of huntingtin. Identical Western blot panels of lymphoblastoid cell protein extracts were probed withpreimmune (Pre) or immune (I) sera. The left-hand panel shows the results obtained with a 1:500 dilution of anti-peptide serum HP1; the central panel, with 1:500 dilution of anti-peptide serum HP12; and the right-hand panel, a1:10,000 dilution of anti-fusion protein HF1. The position of the 200-kD molecular weight standard (myosin) isshown. The identity of the -200-kD protein band detected by the preimmune and immune rabbit sera is notknown but also serves as a convenient marker protein in human lymphoblast protein extracts. The position of thehuntingtin band detected in both lanes of each panel by the immune sera is indicated and is estimated to be -350kD. (B) Detection of huntingtin is abolished by preincubation with antigen. Identical Western blot panels of lym-phoblast protein extract were probed with immune sera at the concentrations given in Panel A with (+) or with-out (-) preabsorption of each antiserum (HPI, HP12, and HF1) with 12 ,g/ml of its corresponding immunogen.The positions of the 200-kD marker protein (myosin) and of the specifically competed -350-kD huntingtin pro-tein are indicated. The right-hand panel was deliberately overexposed to visualize the -250-kD putative huntinig-tin breakdown product that is also competed by immunogen. This band is located between two nonspecific bandsof similar intensity that are not competed. (C) HD huntingtin is distinguishable from normal huntingtin. IdenticalWestern blot panels of lymphoblast protein extracts from HD heterozygotes (Lane 1) and HD homozygotes (Lane2) were probed with a 1:500 dilution of HP1 (left-hand panel) and HP12 (central panel), and 1:10,000 dilution ofHF1 (right-hand panel). The length of the polymorphic CAG stretch in the DNA of each individual was deter-mined by PCR analysis and is shown above each lane as the estimated number of CAG repeat units, (CAG),. Theposition of the 200-kD marker (myosin) is shown and the products of the normal and HD alleles, located close toeach other at -350 kD, are denoted by a square bracket.

ing preimmune sera. Both the immune and pre-immune HP1 and HF1 sera also detect a band at-200 kD that varies in intensity in different ex-tracts. This nonspecific band (see below) is de-tected by most other rabbit antisera that we havetested, and may be myosin.

The specificity of detection of the proteinbands was tested by preabsorption of each im-mune serum with its corresponding antigen, anexample of which is shown in Fig. lB. In all threecases, incubation with the corresponding im-munogen specifically eliminates the reactivity tothe -350-kD protein. This reactivity is not re-moved by preincubation with high concentra-

tions of bovine serum albumin (data not shown).Preabsorption with immunogen does not elimi-nate reactivity to the nonspecific -200-kD pro-tein detected by both immune and preimmunesera. Deliberate overexposure of the HF1 blots inFig. lB reveals the presence two intense bands,the specific 350-kD huntingtin band and thenonspecific -200-kD band, along with threelighter bands of intermediate size. The larger andsmaller of these intermediate bands are nonspe-cific, as they are not removed by preabsorption ofthe antiserum with immunogen. However, theremaining intermediate band, which we estimateat -250 kD, is specifically removed by preab-

377

I


sorption. The presence of this band and its rela-tive intensity is variable across experiments, sug-gesting that it is a breakdown product of the-350-kD protein. No other specific lower molec-ular weight proteins were detected in these anal-yses and use of multiple antisera against thesame blot did not reveal additional specific bands(data not shown), supporting the view that thediscrete band detected at -350 kD by all of theseantisera is huntingtin. This notion is also corrob-orated by immunoprecipitation of huntingtinfrom 35S-labeled lymphoblast extracts using HP1and HF1 (data not shown). The fact that hun-tingtin is detected using antisera directed againstpredicted epitopes both upstream and down-stream of the CAG repeat indicates that the trinu-cleotide segment is translated in a frame predictedto produce a lengthy stretch of polyglutamine.

The HD Mutation Does Not EliminateHuntingtin ExpressionHuntington's disease displays a completely dom-inant clinical phenotype, as evidenced by a num-ber of individuals who are homozygous for thedisease allele (22,23). Lymphoblast cell linesfrom such individuals make it possible to dem-onstrate directly expression of huntingtin fromthe mutant allele. For example, in Fig. 1C, HP1,HP12, and HF1 all detect huntingtin in an HDhomozygote with CAG repeat lengths of 41 and54 on the two disease chromosomes.

Interestingly, not only is huntingtin pro-duced from the HD allele, but the HD huntingtincan also be distinguished from normal hunting-tin based on its electrophoretic mobility. For ex-ample, Fig. 1C also shows typical results for anHD heterozygote (with a disease allele of 65CAGs and a normal allele of 16 CAGs). All threeantisera detect the same two protein species fromthis individual, suggesting that the allelic prod-ucts differ significantly in their migration due tothe difference in the length of the polyglutaminestretch. The predicted difference between HDand normal huntingtin is <1% in almost all HDcases and <3% in even the most severe juvenileonset cases. It has been noted previously thatchanges in the length of polyglutamine stretchescan cause disproportionate changes in the migra-tion of the proteins that contain them (H. T. M.Timmers, personal communication). We haveoptimized gel conditions to favor the separationof very large proteins and have consistentlyfound that heterozygous HD patients reveal twodistinct huntingtin bands. In all cases, detection

is prevented by preincubation with immunogen.In contrast to HD heterozygotes, neither normalindividuals nor HD homozygotes reveal two dis-tinct protein bands. The latter is not surprising,since the interallelic difference in each such in-dividual is typically < 15 CAG repeat units. Mix-ing of protein extracts from normal individualsand HD homozygotes, as well as assaying of pro-tein from adult onset HD patients, both suggestthat resolution of distinct huntingtin bands inthis gel system requires a difference of more than20 residues.

To confirm our presumption that the largerof the two huntingtin bands derives from the HDallele, while the smaller is expressed from thenormal allele, we performed the analysis shownin Figure 2. Segregation of the HD and normalalleles in a family previously reported to containHD homozygotes (22) is compared with analysisof huntingtin in corresponding lymphoblast pro-tein extracts. As expected, progeny with only thenormal alleles (16/20 CAG repeat units) display asingle huntingtin band. Progeny with two dis-ease alleles (49/43, 55/48, and 54/41 units) yielda single huntingtin band that is larger than nor-mal huntingtin. Heterozygous individuals, in-cluding both offspring (44/16 units) and parent(45/16 units), exhibit two huntingtin bands. Thesmaller band is similar in size to that seen innormals and the larger band similar to that seenin the HD homozygotes. Thus, the CAG segmentin the open reading frame of the HD gene istranslated from both normal and HD alleles con-sistent with the defect having its deleterious ef-fects at the protein level.

Northern blot and in situ hybridization stud-ies have suggested that huntingtin is expressed inmost or all tissues. To determine whether theprotein could be detected in the target tissue ofHD, we performed prepared Western blots forvarious areas of postmortem normal humanbrain. Figure 3A shows typical results for freshlycollected caudate/putamen, temporal cortex andfrontal cortex, all of which reveal the same sizeband seen in normal lymphoblasts. The two dif-ferent isoforms of huntingtin in an HD hetero-zygote are also detectable in freshly collectedpostmortem brain tissue (Fig. 3B).

Huntingtin Is a Primarily Cytosolic Protein

We have performed cell fractionation experi-ments to determine whether the structural alter-ation in HD huntingtin affects the cellular local-ization of the protein. Figure 4A shows an


A

(CAG) 44 45 49 20 44 55n20 16 43 16 16 48

20 5416 41

hunhIngOn-_

200 kDa-

huntingtin

200 kDa-

FIG. 2. Inheritance of HD and normal hun-tingtin in a sibship previously reported to con-

tain HD homozygotesA Western blot of lymphoblast protein extracts frommembers of a sibship from the Venezuelan HD pedi-gree in which both parents are affected by HD (22)(filled symbols) was probed with a 1:10,000 dilutionof HF1. Birth order and the sex of the parents andprogeny have been disguised for confidentiality. Theestimated number of CAG repeat units, (CAG)n, is

given above each lane. Protein extracts from bothparents were found to exhibit both normal and HDhuntingtin bands in separate Western blot experi-ments but only the genotyping is given here for oneof the parents. The position of the 200-kD marker(myosin) is shown, and the positions of normal andHD huntingtin at -350 kD are indicated by thesquare bracket.

example of Western blot analysis of cell fractionsproduced from normal and HD lymphoblasts bydifferential centrifugation. The cell fractionswere tested for huntingtin and for the distribu-tion of several control proteins (Fig. 4B). As ex-

pected, HD huntingtin migrated more slowlythan normal huntingtin.

Huntingtin is found primarily in the cytosol(S100) fraction when fractionation of either nor-

mal or HD homozygous lymphoblasts is per-

formed in the absence of added salt or detergent.Very little protein is detected in the nuclear frac-tions (including both nucleosol and nuclear pel-let preparations), even upon long exposures ofthe blots. A small proportion of huntingtin doespellet from the cytoplasmic fraction at 100,000 xg (P100 fraction). This minor component can beshifted into the S100 cytosolic fraction by frac-

FIG. 3. Detection of normal and HD hunting-tin in human postmortem brain extracts

(A) Normal human brain. A Western blot of proteinsextracted from dissected regions of a control humanpostmortem brain; temporal cortex (TC), frontal cor-tex (FC), and caudate and putamen (CP) wereprobed with 1:10,000 dilution of HF1. Preabsorptionof the immune antiserum with HF1 fusion proteinspecifically eliminated the huntingtin band, and theadditional bands were also detected by the preim-mune serum (data not shown). The positions of the200-kD marker (myosin) and of the huntingtin bandare shown. (B) HD brain. A Western blot of proteinsextracted from dissected regions of a grade 4 HDpostmortem brain; cerebellum (Cb) and frontal cor-tex (FC) were probed with a 1:10,000 dilution ofHF1. PCR analysis revealed that this individual car-

ried a normal CAG allele of 18 repeat units and an

expanded HD CAG allele of 41 repeats. The positionof the 200-kD marker (myosin) is shown and theproducts of the normal and HD alleles, located closeto each other at -350 kD, are denoted by a

square bracket.

tionation in the presence of increasing salt con-

centrations, suggesting that its association with a

pelletable element is relatively weak. No differ-ences were observed between normal and HDhomozygous cells, indicating that the fraction-ation characteristics of huntingtin are not alteredby the presence of the HD defect.

The cytoplasmic location of huntingtin was

confirmed by immunohistochemical analysis ofnormal and HD tissues. A preliminary survey ofour antisera indicated that the preimmune sera

did not yield significant staining and that, of theimmune sera, only HF1 exhibited reproduciblestaining. Because of the reactivity of HF1 serum

with the nonspecific 200-kD band on Westernblots (Fig. IA and B), we tested the huntingtinspecificity of the immunohistochemical stainingby preabsorption with immunogen. Figure 5Ashows a typical example of the distributionachieved in frontal cortex (in this case from an

HD heterozygote). Both neuronal cell bodies andneuropil are stained by anti-huntingtin antisera,

B

] hunthngtn

-200 kDa

I

379


A

N huntingtin -_

200 kDa-

N HD B

pRB-4

HD huntlngfln

pERKI -4-200 kDa

FIG. 4. Huntingtin is located in the cytosolic fraction of lymphoblast cells(A) Detection of huntingtin. A Western blot of proteins in subcellular fractions of lymphoblasts obtained from a

normal individual (N) and an HD homozygote (HD) was probed with a 1:10,000 dilution of HF1. Fractions are de-noted by PN, postnuclear supernatant; S, 100,000 X g supernatant (S100); P, 100,000 x g pellet (P100), and N,nucleosol. The position of the 200-kD marker (myosin) is shown, and the positions of the -350-kD normal (N)and HD (HD) huntingtin are indicated by arrows. The bands at -240 and -260 kD are also detected by the preim-mune sera. (B) Detection of control proteins. Western blots of proteins in the subcellular fractions of lymphoblastsshown in Panel A were probed with control antibodies. The top panel shows the results obtained using a mono-

clonal antibody directed at pRB, a 1 10-kD nuclear protein that is detected in the N fraction (24), and the bottompanel the results with a monoclonal antibody directed at pERKI (Transduction Laboratories), a 44-kD plasmamembrane associated protein that is detected in the P fraction (25). The band at -45 kD is probably a phosphory-lated form of ERKI that is also recognized by this monoclonal. In separate experiments (not shown), similar West-ern blots were probed with polyclonal rabbit antisera against ornithine aminotransferase (OAT), a mitochondrialprotein (26) that was detected in the P fraction and neurofibromin (27), a cytoplasmic protein that can associatewith microtubules that was found in both the P and S fractions (28).

and this signal is effectively removed by preab-sorption (Fig. SB). The location of huntingtin ismost easily seen in large neurons where the an-

tisera stain the cytoplasm but not the nuclei. Asexpected from the cell fractionation studies, no

difference in the intracellular distribution ofhuntingtin was seen between normal and HDheterozygote cells in these experiments (data notshown).

DISCUSSIONThe discovery that HD is caused by an expanded,unstable CAG trinucleotide repeat has providedan explanation for many of the genetic featuresof the disorder, but the mechanism by which thispeculiar defect leads to specific neuronal celldeath remains obscure. Initial data argue againstan effect on transcription, prompting us to ex-

plore potential mechanisms at the RNA and pro-

tein levels using huntingtin-specific antisera.From sequence analysis alone, it is not been pos-

sible to distinguish whether translation of hun-tingtin mRNA begins upstream of the CAG re-

peat, to produce a product of >3130 amino acidswith a variable polyglutamine segment near theN terminus, or downstream from the CAG

stretch to produce a smaller protein devoid ofpolyglutamine. Neither can the sequence predictwhether the expanded CAG segment on diseasechromosomes eliminates translation of themRNA or alters the start site for translation. Theimmunological studies reported here clearly es-

tablish that alleles bearing the disease-producingexpansion continue to express a protein product,since cells from HD homozygotes also possesshuntingtin. Furthermore, the detection of hun-tingtin as a -350-kD protein by antibodies di-rected against peptides predicted from both up-

stream and downstream sequences confirms thatthe CAG repeat is translated from both normaland HD alleles as a long polyglutamine tract.

The detection of a single large protein speciesin normal individuals is consistent with the ab-sence of alternatively spliced mRNAs from thislocus and suggests that huntingtin is neither pro-cessed to a smaller functional unit nor subject toextensive heterogeneous modifications that af-fect its migration. The only other specific band onWestern blots, detected occasionally at --250 kD,

probably represents a breakdown product pro-duced during preparation of protein extracts.Thus, these antisera provide no immediate evi-dence for other proteins arising from relatedmembers of a gene family.

N S P PN

- 97 kDa

-46 kDa

- 30 kDa%_W F

F. Persichetti et al.: Translation of CAG Repeat in HD 381

FIG. 5. Immunohistochemical detection ofhuntingtin in HD brain(A) Huntingtin expression. A cryostat section offrontal cortex from an HD grade 3 postmortem brainwas immunostained for huntingtin using a 1:1000dilution of HF1 immune serum (630X). The sectionis counterstained with hematoxylin. (B) Detection ofhuntingtin is abolished by preincubation with anti-gen. A section of HD frontal cortex nearly adjacentto that shown in Panel A was immunostained withHF1 immune sera that had been preabsorbed with12 gg/ml of HF1 fusion protein (as in Fig. IB)(630X). The section is counterstained withhematoxylin.

The presence of an expanded CAG segmentis associated with an HD huntingtin that is de-tectably larger than the normal isoform. The rel-ative migration of the protein varies with thelength of the CAG stretch. Normal and HD hun-tingtins cosegregate in families with normal andHD alleles, respectively, demonstrating directlythe origin of each protein band. Our gel systemdoes not allow us to resolve either individualnormal alleles or HD alleles in typical homozy-gotes, since a difference of -20 residues is re-

quired. Thus, although protein analysis couldtheoretically be used to diagnose the presence ofan HD allele, it cannot currently resolve all fouralleles in a mating and is inferior to direct DNA

analysis. The basis for the disproportionate con-tribution of additional glutamines in reducingthe electrophoretic mobility of huntingtin is notobvious, but this phenomenon has also been ob-served for smaller proteins (H. T. M. Timmers,personal communication). Consequently, addi-tional investigation might lead to a method ofseparating huntingtin products from alleles dif-fering by fewer than 20 CAG units.

Our investigations also demonstrate clearlythat both normal and HD isoforms of huntingtinare cytoplasmic proteins. The only previous in-vestigation of huntingtin's cellular location wasan immunohistochemical analysis using a singleC-terminal anti-peptide antiserum (29). Thisyielded variable results, staining either exclu-sively cytoplasm in most cells or both cytoplasmand nucleus in lymphoblasts and neurons. Bothour Western blot and immunohistochemicalanalyses failed to corroborate the suggestion ofnuclear localization, implying that the previouslyreported variable nuclear staining may havebeen due to specific conditions that allowedshuttling of the protein or at least its C-terminalportion into the nucleus in some cells, or mayhave been spurious. Our detection of huntingtinexclusively in the cytoplasm of both normal andHD cells indicates that the expanded polyglu-tamine segment does not to alter the protein'scellular compartment. However, more detailedstudies at the EM level could conceivably revealregional differences in distribution of normal andHD huntingtin within the cytoplasm.

The translation of the CAG repeat into apolyglutamine segment in huntingtin strength-ens the view that the defect acts at the proteinlevel. Possible mechanisms include inappropriatecellular expression or intracellular localization,increased activity, a dominant negative effect, ora gain of function. Any hypothesis ultimatelymust also account for the specificity of cell deathand the strong correlation between repeat lengthand disease severity.

Neither HD mRNA in situ hybridization stud-ies nor preliminary immunohistochemical anal-ysis have suggested a flagrant abnormality in thepattern of huntingtin expression in HD brain(9-11,29), although more detailed studies in avariety of cells and tissues during developmentmight still reveal significant differences. Both in-creased activity of HD huntingtin and a domi-nant negative effect of the mutation representhypotheses that cannot be tested directly, sinceno function has yet been ascribed to the protein.The gain-of-function scenario, which is particu-


larly attractive, could involve the formation of atoxic byproduct from HD huntingtin, or a novelinteraction with another cytoplasmic compo-nent. The latter would explain the cellular spec-ificity of the pathogenesis if the abnormal inter-action occurred with a protein peculiar to striatalneurons.

An important difference between the gain-of-function hypothesis and the other potentialmodels is that in the former, the pathogenicmechanism might operate without any detri-mental effects on huntingtin's normal activity.For example, this contrasts directly with thedominant negative hypothesis which presumes aloss of huntingtin's activity and places a pre-mium on discovering the normal role for thisprotein. These two opposing mechanisms may bedistinguished by manipulation of the mouse Hdhgene, either to inactivate it or to introduce anexpanded CAG segment. It is hoped that at leastone of the two will yield an accurate geneticmodel for dissecting the steps by which produc-tion of HD huntingtin leads to specific neuronaldeath.

ACKNOWLEDGMENTSWe thank Drs. J. Settleman, S. Pillai, V. Ramesh,E. Harlow, A. Bernards, S. Reeves, and H. T. M.Timmers for supplying control antisera and forhelpful discussion and the Brain Tissue ResourceCenter (McLean Hospital) for postmortem tissuesand clinical information. This work was sup-ported by National Institutes of Health GrantsNS16367, NS2203 1, and NS32765, and by grantsfrom Bristol-Myers Squibb, Inc., the HereditaryDisease Foundation, and the Huntington's Dis-ease Society of America. CA was supported bythe Andrew B. Cogan Fellowship of the Heredi-tary Disease Foundation, and SM and MD weresupported by fellowships from the Huntington'sDisease Society of America and the HereditaryDisease Foundation, respectively.

REFERENCES1. Martin JB, Gusella JF. (1986) Huntington's

disease: Pathogenesis and management. N.Engl. J. Med. 315: 1267-1276.

2. Folstein SE. (1989) Huntington's Disease. Adisorder of families. The Johns Hopkins Uni-versity Press, Baltimore.

3. Vonsattel JP, Myers RH, Stevens TJ, FerranteRJ, Bird ED, Richardson Jr EP. (1985) Neu-

ropathological classification of Huntington'sdisease. J. Neurol. Exp. Neuropath. 44: 559-577.

4. Gusella JF, Wexler NS, Conneally PM, et al.(1983) A polymorphic DNA marker geneti-cally linked to Huntington's disease. Nature306: 234-238.

5. Huntington's Disease Collaborative ResearchGroup. (1993) A novel gene containing atrinucleotide repeat that is expanded andunstable on Huntington's disease chromo-somes. Cell 72: 971-983.

6. Gusella JF, MacDonald ME, Ambrose CM,Duyao MP. (1993) Molecular genetics ofHuntington's disease. Arch. Neurol. 50: 1157-1163.

7. Gusella JF, MacDonald ME. Huntington'sdisease. Semin. Cell Biol. 1, in press.

8. Ambrose CM, Duyao MP, Barnes G, et al.(1994) Structure and expression of the Hun-tington's disease gene: Evidence against sim-ple inactivation due to an expanded CAGrepeat. Somat. Cell Mol. Genet. 20: 27-38.

9. Strong TV, Tagle DA, Valdes JM, et al. (1993)Widespread expression of the human and ratHuntington's disease gene in brain and non-neuronal tissues. Nature Genet. 5: 259-265.

10. Li SH, Schilling G, Young III WS, et al.(1993) Huntington's disease gene (IT15) iswidely expressed in human and rat tissues.Neuron 11: 985-993.

11. Landwehrmeyer BG, McNeil SM, Ge P, et al.(in press) Huntington's disease gene: Expres-sion in brain of normal and affected individ-uals. Annals Neurol.

12. MacDonald ME, Ambrose CM, Duyao MP,Gusella JF. (1993) Capturing a CAGey killer.In: Genome Analysis. Vol. 7, Genome Rear-rangement and Stability. Cold Spring HarborLaboratory Press, New York, pp. 25-41.

13. Harlow E, Lane D. Antibodies, A LaboratoryManual. (1988) Cold Spring Harbor Labora-tory Press, New York.

14. MacDonald ME, Lin C, Srinidhi L, et al.(1991) Complex patterns of linkage disequi-librium in the Huntington disease region.Am. J. Hum. Genet. 49: 723-734.

15. MacDonald ME, Novelletto A, Lin C, et al.(1992) The Huntington's disease candidateregion exhibits many different haplotypes.Nature Genet. 1: 99-103.

16. Anderson M, Gusella JF. (1984) Use of cly-closporin A in establishing Epstein-Barr vi-rus-transformed human lymphoblastoid celllines. In Vitro 20: 856-858.

F. Persichetti et al.: Translation of CAG Repeat in HD 383

17. Warner JP, Barron LH, Brock DJ. (1993) Anew polymerase chain reaction (PCR) assayfor the trinucleotide repeat that is unstableand expanded on Huntington's disease chro-mosomes. Mol. Cell. Probes 7: 235-239.

18. Rubinsztein DC, Leggo J, Barton DE, Fergu-son-Smith MA. (1993) Site of CCG polymor-phism in the HD gene. Nature Genet. 5: 214-215.

19. Miner LL, Pandalai SP, Weisberg EP, Sell SL,Kovacs DM, Kaplan BB. (1992) Cold-in-duced alterations in the binding of adreno-medullary nuclear proteins to the promoterregion of the tyrosine hydroxylase gene. J.Neurosci. Res. 33: 10-18.

20. Dignam JD. (1990) Preparation of extractsfrom higher eukaryotes. In: Deutscher MP(ed). Guide to Protein Purification. Vol. 182,Methods in Enzymology. Academic Press,New York, pp. 194-203.

21. Hockfield S, Carlson S, Evans C, Levitt P,Pintar J, Silberstein L. (1993) Selected meth-ods for antibody and nucleic acid probes. In:Molecular Probes of the Nervous System. ColdSpring Harbor Laboratory Press, New York.Vol. 1, pp. 111-226.

22. Wexler NS, Young AB, Tanzi RE, et al.(1987) Homozygotes for Huntington's dis-ease. Nature 326: 194-197.

Contributed by S. H. Orkin on March 17, 1995.

23. Myers RH, Leavitt J, Farrer LA, et al. (1989)Homozygote for Huntington's disease. Am. J.Hum. Genet. 45: 615-618.

24. Hu QJ, Bautista C, Edwards GM, Defeo-Jones D, Jones RE, Harlow E. (1991) Anti-bodies specific for the human retinoblastomaprotein identify a family of related polypep-tides. Mol. Cell Biol. 11: 5792-5799.

25. Marquardt B, Stabel S. (1992) Sequence of arat cDNA encoding the ERKI1-MAP kinase.Gene 120: 297-299.

26. Mueckler MM, Pitot HC. (1985) Sequence ofthe precursor to rat ornithine aminotransfer-ase deduced from a cDNA clone. J. Biol.Chem. 260: 12993-12997.

27. The I, Murthy AE, Hannigan GE, et al.(1993) Neurofibromatosis type 1 gene mu-tations in neuroblastoma. Nature Genet. 3:62-66.

28. Golubic M, Roudebush M, Dobrowolski S,Wolfman A, Stacey DW. (1992) Catalyticproperties, tissue and intracellular distribu-tion of neurofibromin. Oncogene 7: 2151-2159.

29. Hoogeveen AT, Willemsen R, Meyer N, et al.(1993) Characterization and localization ofthe Huntington's disease gene product. Hum.Mol. Genet. 2: 2069-2073.

normal and expanded huntington's disease gene alleles produce

Documents