Molecular Psychiatry (1999) 4, 58–63  1999 Stockton Press All rights reserved 1359–4184/99 $12.00

dentatorubral-pallidoluysian atrophy, and Hunt- ORIGINAL RESEARCH ARTICLE ington’s disease (HD).1,2 CAG expansion has also been reported in schizophrenia,3–5 suggesting a role for PGE. CAG expansion in schizophrenia was difficult to repli- Detection of polyglutamine cate,6 suggesting that locus and allelic heterogeneity in expansion in a new acidic schizophrenia may lead to complex inheritance for which genetic associations are difficult to detect.7 In : a candidate for this respect, the use of nuclear families comprising a patient affected by earlier onset and/or a severe form childhood onset of a psychiatric disorder is considered an increasingly attractive method for identifying susceptibility .7 schizophrenia? We searched for PGE in childhood onset schizophrenia 8–11 1 1 1 (COS), a rare form of the disease. COS can be S Morinie`re , C Saada , S Holbert , detected using unmodified DSM-IIIR8,9 and shows the 2 3 4 E Sidransky , A Galat , El Ginns , following main features: (i) greater disease severity 2 1 JL Rapoport and C Ne´ri than adult onset schizophrenia (AOS); (ii) disruption of multiple developmental domains before the appear- 1Fondation Jean Dausset-CEPH, Paris, France; 2Child ance of psychotic symptoms, and, occasionally, pro- Neurogenetic Branch, National Institute of Mental Health gression of brain abnormalities after onset of psy- (NIMH), Bethesda, MD, USA; 3Commissariat a` l’Energie chosis;9–11 and (iii) clinical9 and neurobiological11 Atomique, De´partement D’Inge´nierie des Prote´ines continuity with AOS. Therefore, the study of COS may (CEA/DIEP), Saclay, France; 4Child Psychiatry Branch, provide new insights into the genetic susceptibility to NIMH, Bethesda, MD, USA schizophrenia in general. Lymphoblastoid cell lines (LCLs) for 32 unrelated Keywords: polyglutamine; schizophrenia; candidate COS patients recruited all over the United States (ongoing NIMH study, see Ref. 8) were tested by West- Polyglutamine expansion (PGE) encoded by a CAG ern blot analysis with the TATA-binding protein (TBP) repeat underlies eight inherited neurodegenerative dis- monoclonal mAb1C2. MAb1C2 specifically eases, among which is Huntington’s disease. CAG 12 expansion has also been reported in schizophrenia, recognizes long polyglutamines (polyglns). Above the suggesting a role for PGE. To investigate the potential detection threshold (33–35 Glns), the intensity of the 12 role of PGE as a candidate for schizophrenia, we signal increases with the number of glutamines. searched for PGE in nuclear families comprising a Sixty-kDa PGE signals were observed in eight of 11 patient affected by childhood onset schizophrenia unrelated black American COS patients (Table 1). The (COS, a rare and severe form of the disease) as a vari- intensity of the 60-kDa band was strong in two ation of the candidate approach for identifying (patients 14 and 18) of the eight positive COS patients. susceptibility genes. We tested lymphoblastoid cell In the nuclear families (NFI and NFII) of patients 14 lines from COS patients (n = 32) by analy- and 18, a weak 60-kDa band (indicative of a small PGE) sis with 1C2, a monoclonal antibody that specifically was detected in some unaffected parents and sibs, sug- recognizes long polyglutamines. Eight of 11 unrelated gesting that the weaker 60-kDa band is not indicative black American COS patients showed a 60-kDa (approximately) band indicative of PGE. A strong 60-kDa of COS (Figure 1a and b). Using high resolution, long band (suggestive of a large PGE) was detected in two SDS-PAGE (10%) gels, the weak band clearly migrated of the eight positive patients. A weaker 60-kDa band at a slightly higher position than the strong band in (suggestive of a smaller and non pathogenic PGE) was NFI (Figure 1a, bottom panel), suggesting that the weak detected in some unaffected parents or sibs of these signal does not correspond to the same protein with a two COS patients, and in six other black American COS shorter polygln tract and that the strong and weak PGE patients. The strong and weak PGE signals were found signals arise from two different . The detection to correspond to two different proteins. Unrelated black profiles for NFI and NFII were reproduced using three Americans unaffected by COS (n = 38) were negative for different batches of LCL extracts. the strong 60-kDa PGE signal. Healthy white Americans = Long polygln stretches in human proteins induce a (n 53) were negative for both the strong and weak 60- migration schift towards high molecular weights.12 For kDa PGE signals. Two-dimensional gel analysis sug- example, TBP is a 37.2-kDa protein which migrates at gested that the strong PGE signal corresponds to an acidic (pI 4 approximately) protein and resulted in a 50 kDa approximately. This suggests that the strong 60- more precise estimation (52–57 kDa) of its relative mass. kDa PGE signal corresponds to a protein with a true This protein appeared to be not represented in Gen- mass of about 47 kDa. When probed with mAb1C2, it bank, as suggested by the exclusion of several candi- was previously reported that the normal HD protein is date CAG repeats. Our data suggest that this acidic pro- barely detectable after gross overexposure if the tein might be a candidate for COS. polygln contains more than 27 units.12 Using SDS- Polyglutamine expansion (PGE) encoded by a CAG PAGE (7%, 10%) gels with prolonged migration and repeat underlies eight inherited neurodegenerative dis- long exposure, the mAb1C2-reactive 60-kDa protein in eases including spinobulbar muscular atrophy, spino- COS patient 14 was not detected as a doublet (data not cerebellar ataxia 1 (SCA1), SCA2, SCA3, SCA6, SCA7, shown), suggesting that the unexpanded allelic protein Detection of polyglutamine expansion S Morinie`re et al 59 Table 1 PGE in black American COS patients

Patient Sexa Ethnic background Age of onset Age of blood draw 60-kDab PGE signal

1 F white American 9 14 − 2 F white American 11 15 − 3 M white American 10 12 − 4 M white American 8 13 − 5 M black American 10 14 + 6 M white American 10 14 − 7 M white American 7 12 − 8 M white Iranian 11 13 − 9 M white American 11 17 − 10 F black American 10 16 − 11 M white American 11 14 − 12 F white American 9 10 − 13 M black American 11 16 + 14 F black American 12 14 ++ 15 M white American 11 15 − 16 F black American 12 16 + 17 M white American 11 16 − 18 F black American 11 13 ++ 19 F black American 12 17 − 20 M white American 7 14 − 21 F black American 5 9 + 22 M white American 6 21 − 23 F Philipino 11 12 − 24 F white Hispanic 11 16 − 25 F white American 10 13 − 26 F white Hispanic 12 14 − 27 F white Hispanic 9 17 − 28 M white American 11 12 − 29 M black American 12 19 + 30 M black American 12 16 + 31 M indian American 12 15 − 32 M black American 12 15 −

Thirty-two COS patients were tested for PGE by Western blot analysis with mAb1C2 (see Figure 1). Eight of the 11 black American COS patients are positive for PGE. aF, female; M, male. b − + ++ Approximate Mr , negative; , weak intensity (also observed in black American individuals unaffected by COS); , strong intensity (observed in COS patients only). contains less than 28 glutamines.12 In control experi- protein. The PGE signal was not detected in the PGE- ments, a 60-kDa band was not observed in three heal- negative COS patient 10 (data not shown). The PGE sig- thy white families (CEPH Utah pedigrees 1334, 1344, nal was not detected in COS patient 16 either, suggest- and 1420; 53 individuals in total; data not shown) pre- ing that the weak 60-kDa band corresponds to a protein viously reported to show CAG expansion.13 We also (pI not in the range tested) different from the acidic tested 38 unrelated black American controls with no protein detected in COS patients 14 and 18 (data not history of psychiatric diseases (see Methods). Except shown). This observation was consistent with the for 11 of them (showing a weak 60-kDa band), these observations previously made in NFI using high resol- individuals were negative for 60-kDa PGE signals as ution 1D gel (see paragraph above). illustrated in Figure 1c. To identify the gene for the polygln expanded in To further characterize the protein which carries the COS patients 14 and 18, we selected 27 proteins carry- strong PGE signal, we performed two-dimensional (2D) ing 8–34 consecutive Glns by using Genbank scanning, (pI 10–3 gradient). Using two differ- and one acidic protein, DAN2614 (52.5 kDa, pI 4.8), was ent batches of LCL extract of COS patient 14, the com- retained for PCR screening in COS family NFI (see parison of silver-stained IEF/SDS PAGE (Figure 2a) Methods). Twenty-seven CAG repeats found in with 1C2 immunoblots (Figure 2b) suggested that PGE cDNAs15–17 and CAG/CTG repeats isolated from gen- is carried by an acidic protein (isoelectrofocusing point omic DNA18 consisting of more than 10–15 consecutive [pI] of 4 approximately). Two-dimensional gel electro- units were also tested for expansion in NFI (see phoresis experiments also resulted in a more precise Methods). All the candidates tested showed unen- estimation (52–57 kDa) of the relative mass (Mr) of this larged CAG repeat alleles, and no difference in the size Detection of polyglutamine expansion S Morinie`re et al 60

Figure 1 Detection of PGE in COS. Electrophoresis was performed on 50 ␮g (4–20% polyacrylamide gradient gels) or 60 ␮g (long 10% polyacrylamide gels) of protein from LCL extracts (see Table 1). (a) A strong 60-kDa PGE signal is detected by mAb1C2 in black American COS patients 14 (nuclear family NFI, top left panel) and 18 (nuclear family NFII, top right panel; the LCL for father of patient 18 is not available, father is deceased). Parents or sibs (no history of psychiatric diseases except for the father of patient 14 affected by AOS) of these two COS patients are negative for the 60-kDa band (NFI, lane 2) or show a weak (NFI, lanes 1 and 4–6; NFII, lane 4) 60-kDa PGE signal. High resolution SDS-PAGE (10%) gel analysis (NFI, bottom panel, enlarged view of the gel) indicated that the mAb1C2-reactive weak 60-kDa band (lanes 1 and 4–6) migrates at a higher position than the strong 60-kDa band (lane 3). Sibs of patient 14 were age 13.9 (lane 4), 23.3 (lane 5), and 23.9 (lane 6) years at blow draw. NFI, SCA2 (lane 7) and SCA3 (lane 8) patients are shown as PGE positive controls. The number of Gln repeats of the TBP alleles in NFI was determined by PCR analysis as previously described15 and was comprised of 33–37 units. NFII, lanes 1 and 2 show COS patient 14 (for comparison with the strong 60-kDa PGE signal) and her mother (negative for the 60-kDa band), respectively. (b) No 60-kDa PGE signal (patients 10 and 32) or a weak 60-kDa PGE signal (patients 16, 5, 30, 29, and 13) was detected by 1C2 in the other 30 COS patients tested (see Table 1). Left end lane, COS patient 14 (shown for comparison with the strong 60-kDa PGE signal). (c) No 60-kDa PGE signal (lanes 2, 3, 4, 5, and 7) or a weak 60-kDa PGE signal (lanes 6 and 8) was detected by 1C2 in healthy black American controls. Lane 1, COS patient 14 (shown for comparison with the strong 60-kDa PGE signal). Detection of polyglutamine expansion S Morinie`re et al 61

Figure 2 Characterization of the protein which carries PGE in COS patient 14. (a) Silver-stained IEF/SDS-PAGE (right half of the gel shown) of proteins from LCL extract (500 ␮g of protein) of COS patient 14. TBP was not detected since this protein has a theoretical pI equal to greater than 10. (b) Corresponding 2D gel immunoblot with mAb1C2. Black arrow (panels a and b), mAb1C2-immunoreactive protein (Mr, 52–57 kDa; pI 4 approximately). of the CAG repeat alleles between patient 14 and her tested is likely to correspond to a different protein as parents or sibs could be observed (data not shown). indicated by high resolution 1D and 2D gel electro- Therefore, all the candidates tested were excluded. phoresis experiments. Because a strong PGE signal was This suggested that the target protein detected in not detected in age-matched controls with no history patients 14 and 18 is not represented in Genbank DNA of psychiatric disorders (in particular, one of the sisters sequences which contain long (Ͼ10–15 units) CAG of patient 14; see Figure 1, lane 4), it is unlikely that repeats. The testing of other candidates selected with there are age differences in the expression of the 52– less stringent criteria and/or the detailed amino-acid 57 kDa acidic protein observed with a peak during analysis of the target protein may allow the cloning of puberty. this gene. Altogether, our data delineate the 52–57 kDa acidic The detection of CAG expansion3–5 and the identifi- protein as a potentially interesting candidate for COS cation of susceptibility loci19 have suggested a genetic and call for three lines of additional investigation in basis for schizophrenia. What might be inherited in order to determine whether this protein might be schizophrenia is a predisposition to develop the dis- involved in schizophrenia. First, the number of LCLs ease,7,19 and the use of nuclear families comprising a currently available for COS patients is limited, and rep- patient affected by early-onset and/or severe form of lication studies in other COS patients are warranted.8 the disease is considered as an interesting variation of Second, the notion of clinical continuity from COS to the candidate gene approach to delineate genetic sus- AOS9,10 calls for the search for PGE in schizophrenia in ceptibility.7 Earlier onset and/or greater severity of the general. PGE does not appear to occur in schizophrenic disease may indeed reflect greater heritability, and patients showing enlarged CAG/CTG repeats with 50 unaffected members of nuclear families provide well- or more units.20 Recently, the longer (Ͼ19–20 units) matched controls. Using 1D and 2D gel electrophoresis CAG repeat (encoding for a polymorphic polyglutam- and mAb1C2 immunoblotting experiments, we have ine tract consisting of 12–28 units) alleles in the potass- detected a strong PGE signal which corresponds to a ium channel gene hSKCa3 have been reported to be presumably new, 52–57 kDa acidic protein and was over-represented in schizophrenic patients,21 suggest- specifically observed in black American COS patients ing a role for moderately enlarged polyglns. Finally, it 14 and 18. Because the intensity of the PGE signal in is interesting to note that, using mAb1C2, PGE signals COS patients 14 and 18 is clearly stronger than what in the 50–60 kDa range were observed in two unrelated is observed for the TBP band and for the weak 60-kDa Caucasian patients affected by schizophrenia (G Rou- band, and because of the absence of background sig- leau, personal communication), further suggesting that nals, the strong PGE signal is unlikely to result from proteins carrying polymorphic polyglns might be can- cross-reacting activities of mAb1C2 with polygluta- didates for schizophrenia in general. mines containing less than 33–35 Glns. The weak 60- While the use of the repeat expansion detection kDa band observed in some unaffected members of the (RED) technique13 has allowed the detection of families of patients 14 and 18 and in the other controls enlarged CAG/CTG repeats in schizophrenic patients,3–5 Detection of polyglutamine expansion S Morinie`re et al 62 long and unstable CAG/CTG repeats were also ident- 1D and 2D gel electrophoresis and electroblotting ified in normal individuals, including the intronic 1D gel electrophoresis was performed as follows: repeat CTG18.1 in the SEF2-1 gene and the transcribed whole cell extracts from LCLs (50 ␮g protein per lane) repeat ERDA1/Dir 1 (reviewed in Ref. 22). Expansions were analyzed by SDS-PAGE using 4–20% at these two loci appear to account for the majority of acrylamide/bisacrylamide (39/1) gradient gels (Biorad, CAG expansions detected by RED in COS.22 In parti- Hercules, CA, USA). For high resolution 1D gel electro- cular, COS patient 14 had a RED score of 90 units and phoresis, whole cell extracts from LCLs (60 ␮g protein a Dir 1 size of 95 units, and COS patient 18 had a RED per lane) were analyzed by SDS-PAGE using long 7% score of 60 units and a Dir 1 size of 57, the CTG18.1 or 10% acrylamide/bisacrylamide (39/1) gels (CBS repeat being normal (Ͻ37 units) in these two patients Scientific Co). 2D gel electrophoresis was performed as (E Sidransky and JL Rapoport, personal communi- follows: protein samples from LCLs (500 ␮g) were ana- cation). The intensity of the strong 60-kDa PGE signal lyzed by isoelectrofocusing (IEF) SDS-PAGE using was similar in these two patients, and the other COS 20 × 20 cm, 12% polyacrylamide gels (Pharmacia- patients for which elevated RED scores correlate with Biotech, Uppsala, Sweden) as recommended by the

Dir 1 sizes did not show a strong 60-kDa PGE signal, manufacturer. Relative mass (Mr) of proteins was estab- suggesting that the polygln tract detected in the candi- lished using molecular mass standards supplied by date acidic protein does not relate to Dir 1 or to other Sigma. SDS-PAGE gels were silver-stained as pre- enlarged CAG/CTG repeats detected by RED in COS. viously described.23 Electroblotting of Hybond N+ In conclusion, the small number of COS patients membranes was performed using standard procedures found positive for the strong PGE signal in our study on a Biorad system (1D gel electrophoresis) and on the might reflect the fact that any one candidate may con- Pharmacia-Biotech Multiphor II and Nova- blot sys- tribute a small portion of susceptibility to a complex tems (2D gel electrophoresis) as recommended by the disease such as schizophrenia. In this respect, the manufacturers. aforementioned additional studies in larger and/or more diverse cohorts may be expected to provide a Immunodetection of PGE deeper insight into the potential contribution of pro- Lyophilized mAb1C2 from mouse ascites fluid was teins carrying expanded polyglutamines to schizo- resuspended in sterile distilled water, aliquoted, and phrenia. stored at −20°C until use. Immunoprobing and detec- tion were performed as previously described12 with minor modifications. Membranes were blocked with Methods 5% non fat dry milk, 0.02% Tween overnight at 4°C, incubated in 0.5% non fat dry milk with mAb1C2 Origin and preparation of LCL extracts (1:2000) for 2 h at room temperature, and treated with We analyzed the following samples: (1) LCLs from 32 anti-mouse antisera conjugated to horseradish peroxi- unrelated COS patients (recruited all over the United dase (Amersham, Les Ulis, France) at 1:4000 in 0.5% States) for which a detailed clinical description has non fat dry milk for 1 h at room temperature. After been previously published;8,10,11 (2) LCLs from 38 unre- washing, membranes were processed using the ECL+TM lated black American controls including 10 healthy protein detection and revelation kit (Amersham), as individuals (NIMH; age 9–50 years; no history of psy- recommended by the manufacturer. Blots were chiatric diseases) from the Washington DC area and 28 exposed to hyperfilms MP (Amersham) for 30 s–1 h to unaffected individuals (NIGMS Cell Repository, Coriell examine background signals caused by the presence of Institute for Medical Research, NY, USA; age 2–58 normal proteins containing polyglutamine stretches. years; regional origin in the USA not identified) with Detection profiles were interpreted by comparative disease for diseases other than psychiatric analysis of band intensities for the specifically reactive diseases and including ataxia telangiectasia, cystic proteins, the TATA-binding protein (TBP) for which fibrosis, fragile site mental retardation, crypto- the major allele corresponds to 38 Gln, and the SCA2 phthalmos, hemoglobin loci mutations, and disorders (22/43 Gln) and SCA3 (23/72 Gln) proteins from of and nucleic acid and carbohydrate patients heterozygous at these loci. Experiments were metabolism (list available upon request); and (3) LCLs repeated at least three times. from three healthy, white American families (CEPH Utah pedigrees 1334, 1344, and 1420; 53 individuals Screening of candidate genes for CAG expansion in total). Whole cell extracts of LCLs were prepared The Genbank database (release No. 103) was scanned without detergent from actively growing cells. Cell for the presence of full coding sequences known or pre- extracts were obtained as previously described12 by dicted to code for proteins with polyglns. We used the

homogenization in 50 mM Tris-HCl pH 8.0, 10% (v/v) BLASTP software with a (Gln)25 query sequence as pre- glycerol, 1 mM EDTA, 5 mM KCl, 1 mM PMSF, viously described.15 We selected all subject sequences 2 ␮gml−1 Leu-peptin, and 4 ␮gml−1 A-protinin. Protein to ensure that all polyglns of at least 8–10 consecutive concentration in LCL extracts was determined using units were retrieved, and we sorted human and unique the bicinchoninic acid (BCA) reagent (Pierce, Rockford, sequences. This resulted in the selection of 27 candi- IL, USA) as recommended by the manufacturer. Ali- dates carrying 8–28 consecutive Glns (except for TBP quots were stored at −80°C until use. which shows 34 consecutive Glns): eight PGE neuro- Detection of polyglutamine expansion S Morinie`re et al 63 degenerative disease proteins, 15 normal proteins 5 O’Donovan MC, Guy C, Craddock N, Bowen T, McKeon P, Macedo (which included transcription factors such as A et al. Confirmation of association between expanded CAG/CTG repeats and both schizophrenia and bipolar disorder. Psychol Med AIB1/Rac3, ASH1, ATBF1-A, hSNF2a, N-Oct-3, 1996; 26: 1145–1153. SATB1, TBP, and the human polyhomeolytic 1 homo- 6 Petronis A, Basset AS, Honer WG, Vincent JB, Tatuch Y, Sasaki T log HPH1, the human homeobox protein HLX1, the et al. Search for unstable DNA in schizophrenia families with evi- polymerase gamma polg, and the meningioma protein dence for genetic anticipation. Am J Hum Genet 1996; 59: 905–911. MN1), and four novel proteins (DAN26, DAN15, 7 Craddock N, Owen M. Modern molecular genetic approaches to psychiatric disease. Br Med Bull 1996; 52: 434–452. AAD10, and AAD14) previously identified by 1C2 8 Gordon CT, Frazier JA, McKenna K, Giedd J, Zamethkin A, Zahn immunoscreening of a cDNA library constructed from T et al. Childhood onset schizophrenia: a NIMH study in progress. a SCA7 patient LCL.14 The theoretical molecular mass Schizophr Bull 1994; 20: 697–712. and isoelectrofocusing point (pI) were determined for 9 Asarnow RF, Caplan R, Asarnow JR. Neurobehavioural studies of each of the selected proteins using a program available schizophrenic children: a developmental perspective on schizo- phrenic disorders. In: Ha¨fner H, Gattaz WF (eds). Search for the from (URL address) http://www-biol.univ-mrs.fr. One Causes of Schizophrenia, Vol 3. Springer-Verlag: Berlin Heidel- of the proteins, DAN26, showed a theoretical mass in berg, 1995, pp 87–113. the 50–60 kDa range (52.5 kDa) and theoretical pI in 10 Rapoport JL, Giedd J, Jacobsen LK, Kumra S, Smith A, Lee P et al. the 4–6 range (pI 4.8), and was retained for PCR screen- Childhood onset schizophrenia: progressive ventricular enlarge- ing in COS family NFI. Polymorphic and/or long CAG ment during adolescence on MRI brain rescan. Arch Gen Psychiatry = 1997; 54: 897–903. repeats (n 20) consisting of more than 10 consecutive 11 Jacobsen LK, Rapoport J. Childhood onset schizophrenia: impli- units and previously found in human cDNA libraries15– cations of clinical and neurobiological research. J Child Psychol 17 were also tested including 2.116, 2.119, i.181, Psychiatry 1998; 39: 101–103. i.182,15 12501r, b01500t, and g02502r,16 and CTG1a, 12 Trottier Y, Lutz Y, Stevanin G, Imbert G, Devys D, Cancel G et al. Polyglutamine expansion as a pathological epitope in Huntington’s CTG3a, CTG4a, CTG7a, CTG20a, CAGF9, F28, H1, H16, disease and four dominant cerebellar ataxia. Nature 1995; 378: H26, H32, H38, H39, H44, H45, L69, L85, L114, L234, 403–406. and L237, some of them predicted to code for a polygln 13 Schalling M, Hudson TJ, Buetow KH, Housman DE. Direct detec- stretch.17 Finally, seven CAG/CTG repeats (GCT1C9, tion of novel expanded trinucleotide repeats in the human genome. GCT4B10, GCT5E11, GCT16E06, GCT10D04, Nature Genet 1993; 4: 135–139. 14 Imbert G, Saudou F, Yvert G, Devys D, Trottier Y, Garnier JM et al. GCT10C10, and GCT10H06) isolated from genomic Cloning of the gene for 2 reveals a locus with DNA and consisting of more than 15 consecutive units high sensitivity to expanded CAG/glutamine repeats. Nature Genet were tested as well.18 The selection and use of primers 1996; 14: 285–291. suitable for detection of CAG expansion in DNAs from 15 Ne´ri C, Albane`se V, Lebre A-S, Holbert S, Saada C, Bougueleret L COS family NFI was performed as previously et al. Survey of CAG/CTG repeats in human cDNAs representing described.15 new genes: candidates for inherited neurological diseases. Hum Mol Genet 1996; 5: 1001–1009. 16 Bulle F, Chiannilkulchai N, Pawlak A, Weissenbach J, Gyapay G, Guellaen G. Identification and chromosomal localization of human genes containing CAG/CTG repeats expressed in testis and brain. Acknowledgements Genome Res 1997; 7: 705–715. 17 Margolis RL, Abraham MR, Gatchell SB, Li S-H, Kidway AS, Bre- We thank Y Trottier and J-L Mandel (IGBMC, Illkirch, shel TS et al. cDNAs with long CAG trinucleotide repeats from France) for the 1C2 antibody and constructive dis- human brain. Hum Genet 1997; 100: 114–122. 18 Gastier JM, Brody T, Pulido JC, Businga T, Sunden S, Hu X et al. cussions, L Tora and Y Lutz (IGBMC, Illkirch, France) Development of a screening set for new (CAG/CTG)n dynamic for preparation of 1C2 ascite fluids, T Fernandez mutations. Genomics 1996; 32: 75–85. (NIMH, Bethesda, MD, USA) for help with sample and 19 Moises HW, Yang L, Kristbjarnarson H, Wiese C, Byerley W, Macci- data management, and A Brice (INSERM U289, Paris, ardi F et al. An international two-stage genome wide search for France) for lymphoblastoid cell lines of SCA2 and schizophrenia susceptibility genes. Nature Genet 1995; 11: 321– 324. SCA3 patients. 20 Jones AL, Middle F, Guy C, Spurlock G, Cairns NJ, McGuffin P et al. No evidence for expanded polyglutamine sequences in bipolar disorder and schizophrenia. Mol Psychiatry 1997; 2: 478–482. 21 Chandly KD, Fantino E, Wittekindt O, Kalman K, Tong LL, Gutman References GA et al. Isolation of a novel potassium channel gene hSKCa3 con- taining a polymorphic CAG repeat: a candidate for schizophrenia? 1 Koshy BT, Zoghbi HY. The CAG/polyglutamine tract diseases: gene Mol Psychiatry 1998; 3: 32–37. products and molecular pathogenesis. Brain Pathol 1997; 7: 927– 22 Sidransky E, Burgess C, Ikeuchi T, Lindbald K, Long RT, Philibert 942. RA et al. A triplet repeat on 17q accounts for most expansions 2 David G, Abbas N, Stevanin G, Durr A, Yvert G, Cancel G et al. detected by the repeat expansion detection technique. Am J Hum Cloning of the SCA7 gene reveals a highly unstable CAG repeat Genet 1998; 62: 1548–1551. expansion. Nature Genet 1997; 17: 65–70. 23 Gallat A, Rioux V. Convergence of amino acid composition of cer- 3 O’Donovan MC, Guy C, Craddock N, Murphy KC, Cardno AG, Jones tain groups of proteins aids in their identification on two-dimen- LA et al. Expanded CAG repeats in schizophrenia and bipolar dis- sional electrophoresis gels. Electrophoresis 1997; 18: 443–451. orders. Nature Genet 1995; 10: 380–381. 4 Morris AG, Gaitond E, McKenna PJ, Mollon JD, Hunt DM. CAG repeat expansions and schizophrenia: association with disease in Correspondence: C Neri, Fondation Jean Dausset-CEPH, 27 rue Juliette females and with early age-at-onset. Hum Mol Genet 1995; 4: Dodu, 75010 Paris, France. E-mail: neriȰcephb.fr 1957–1961. Received 1 May 1998; revised and accepted 12 June 1998