MECP2 mutations account for most cases of typical forms of Rett syndrome

Transcription

1 2000 Oxford University Press Human Molecular Genetics, 2000, Vol. 9, No MECP2 mutations account for most cases of typical forms of Rett syndrome Thierry Bienvenu, Alain Carrié, Nicolas de Roux 1, Marie-Claude Vinet, Philippe Jonveaux 2, Philippe Couvert, Laurent Villard 3, Alexis Arzimanoglou 4, Cherif Beldjord, Michel Fontes 3, Marc Tardieu 5 and Jamel Chelly + Laboratoire de Génétique et Physiopathologie des retards mentaux ICGM, Faculté de Médecine Cochin, 24 rue du Faubourg Saint Jacques, Paris, France, 1 Laboratoire d Hormonologie et de Biologie Moléculaire, CHU Bicêtre, 78 rue du général Leclerc, Le Kremlin Bicêtre, France, 2 Laboratoire de Génétique, Hôpitaux de Brabois, rue du Morvan, Vandoeuvre Les Nancy, France, 3 INSERM U406, Faculté de la Timone, 27 boulevard Jean Moulin, Marseille Cedex, France, 4 Service de Neurologie Pédiatrique, Hôpital Robert Debré, 48 boulevard Sérurier, Paris, France and 5 Département de Pédiatrie, Service de Neurologie, CHU Bicêtre, 78 rue du général Leclerc, Le Kremlin Bicêtre, France Received 25 January 2000; Revised and Accepted 23 March 2000 Rett syndrome (RTT) is a severe progressive neurological disorder that affects almost exclusively females, with an estimated prevalence of approximately one in female births. Most cases are sporadic, but several reports about familial recurrence support X-linked dominant inheritance with male lethality. The gene responsible for this disorder, MECP2, was recently identified by candidate gene strategy. Mutations were detected in <25% of RTT cases in this first report. To characterize the spectrum of mutations in the MECP2 gene in RTT patients, we selected 46 typical RTT patients and performed mutation screening by denaturing gradient gel electrophoresis combined with direct sequencing. We identified 30 mutations, accounting for 65% of RTT patients. They include 12 novel mutations (11 located in exon 3 and one in exon 2). Mutations, such as R270X and frameshift deletions in a (CCACC) n rich region, have been found with multiple recurrences. Most of the mutations were de novo, except in one family where the non-affected transmitter mother exhibited a bias of X inactivation. Although this study showed that MECP2 mutations account for most cases of typical forms of RTT (65%) and mutations in non-coding regions cannot be excluded for the remaining cases, an alternative hypothesis that takes into account the homogeneous phenotype and exclusive involvement of females, could be the implication in RTT of a putative second X-linked gene. INTRODUCTION Rett syndrome (RTT) (MIM ) is a progressive encephalopathy which appears to affect females only. It was first described by Rett in 1966 (1,2). After normal development up to the age of 7 18 months, developmental stagnation occurred, followed by rapid deterioration of higher brain functions. RTT is characterized by severe mental retardation, autism, gait apraxia, hypotonia, disturbance of sleep and breathing, seizures, stereotypical hand movements and deceleration of head growth. Its prevalence is estimated at 1: female births. More than 95% of cases are sporadic, but rare reports of familial recurrence have been made. Previous exclusion mapping studies using the rare RTT families mapped the locus to Xq28 (3). Xq28 is a very gene-rich region and more than one syndrome with mental handicap and neurological signs and symptoms has already been identified within it. However, using a systematic gene screening approach, Zoghbi and colleagues (4) have identified mutations in the gene MECP2 encoding X-linked methyl-cpg-binding protein 2 as the cause of some cases of RTT (5/21 sporadic patients and 1/8 familial patients). More recently, they reported further data showing that MECP2 accounts for 50% of RTT (5). MeCP2 is an abundant chromosome-binding protein that selectively binds 5-methylcytosine residues in symmetrically positioned CpG dinucleotides in mammalian genomes (6). MeCP2 is rich in the basic residues lysine and arginine (22.5%) and in proline (11%) and serine (10.5%). MeCP2 contains two functional domains, an 85 amino acid methyl- CpG-binding domain (MBD), essential for its binding to 5- methylcytosine, and a 104 amino acid transcriptional repression domain (TRD) which interacts with histone deacetylase and the transcriptional corepressor Sin3A. Interactions between this transcription repressor complex and chromatinbound MeCP2 leads to deacetylation of core histones, which in turn leads to transcriptional repression. In the present study we have analysed the entire coding sequence of the MECP2 gene in a sample of 46 typical RTT sporadic cases. We have used the denaturing gradient gel electrophoresis (DGGE) assay combined with direct DNA sequencing and characterized 12 novel mutations. The sequence differences that were found clustered in the third + Towhomcorrespondenceshouldbeaddressed.Tel: ;Fax: ; chelly@icgm.cochin.inserm.fr

2 1378 Human Molecular Genetics, 2000, Vol. 9, No. 9 exon include recurrent nonsense, frameshift and missense mutations. Although the number of RTT patients investigated is not sufficient for statistical analysis, genotype phenotype correlations suggest the presence of a different frequency of some relevant symptoms, such as epilepsy, between the group of RTT patients with a mutated allele and the group of RTT patients with no mutation. RESULTS RTT patients and mutation screening of the MECP2 gene In this study we investigated MECP2 gene in 46 patients, exclusively girls, with a uniform and typical RTT according to the international criteria (see Materials and Methods) (1,7). To carry out mutation screening by DGGE, we have designed appropriate primers to analyse the three exons of the MECP2 gene, and for each amplified segment we determined the optimal position of the chemical clamps (8). The theoretical melting analysis of each fragment was determined by the computer program developed by Lerman and colleagues (9). Exon 2 was analysed in three PCR fragments named 2.1, 2.2 and 2.3, and exon 3 was analysed in five PCR fragments named 3A, 3B, 3C, 3D and 3E (Table 1). Amplification by PCR and DGGE analysis followed by direct sequencing of fragments exhibiting abnormal migration profiles were performed in 38 typical sporadic cases of RTT and their parents. In the remaining eight patients, mutation analysis was performed by direct sequencing of PCR products. In total, this investigation of the nine fragments covering the coding part of MECP2 gene identified in 30 unrelated RTT patients the presence of 17 different mutations mainly clustered in exon 3, and some of them appeared with multiple recurrences. Most of these mutations, which account for 65% of typical RTT cases, are novel and only five mutations have been described previously. Analysis of exon 1 of the MECP2 gene revealed the presence in one RTT sporadic case of an abnormal migration pattern of the PCR fragment corresponding to this exon. The sequence of the PCR product showed a C T substitution at cdna position 15 upstream of the AUG initiation codon. This change was also identified in her unaffected mother. This RTT patient also presents a stop mutation in exon 3 of the MECP2 gene, suggesting that this change (C T at position 15) is a nonpathogenic variant. Exon 2 revealed the presence in one typical RTT case of an abnormal migration pattern of the PCR fragment 2.2, which covers part of exon 2 (Fig. 1B). The sequence of the PCR

3 Human Molecular Genetics, 2000, Vol. 9, No Figure 1. (A) DGGE results corresponding to the fragment 3A of the MECP2 gene. Lane 1, T158M; lanes 2 4, normal; lane 5, R168X. (B) Novel missense MECP2 mutations in typical sporadic RTT patients. Portions of the displayed electrophoregrams illustrate three mutations in RTT patients: P302R, R106Q and 1461A C. The underlined nucleotides and arrows indicate mutated nucleotides for each patient. (C) Novel MECP2 nonsense and frameshift mutations in typical sporadic RTT patients. Portions of the displayed electrophoregrams illustrate five mutations found in RTT patients: R198X, R270X, R294X, 677insA and 1156del17. Deletion is indicated by bold type. The underlined nucleotides and arrows indicate mutated nucleotides for each patient. product of exon 2 revealed a C T substitution at position 317. This mutation, R106Q, is the second missense mutation identified in the first part of the MBD. Another mutation R106W was found previously in the same codon (4). In contrast to these rare events occurring in exons 1 and 2, DGGE screening of exon 3 of the MECP2 gene revealed the presence of different abnormal migration patterns of PCR fragments 3A, 3B, 3C and 3D corresponding to exon 3 (Fig. 1A). The sequences of the PCR products corresponding to fragments 3A 3D revealed five nonsense mutations [R168X (n = 3), R198X (n =1),R255X(n =2), R270X (n = 5) and R294X (n = 3)] (Fig. 1C), three missense mutations [T158M (n =3),P302R(n = 1) and R306C (n = 1)] (Fig. 1B), one insertion [677insA (n = 1)], four deletions [1156del17 (n = 1), 1158del10 (n = 1), 1163del26 (n = 1) and 1164del A T

4 1380 Human Molecular Genetics, 2000, Vol. 9, No. 9 Table 1. Parameters for amplification of the MECP2 gene fragments and for DGGE conditions Fragment Sequences of primers Length (bp) Annealing temp ( C) Gradient (%) Running time (h) at 160 V Exon 1 1F: 5'-Pso-tttctttgttttaggctcca-3' R: 5'-ggccaaaccaggacatatac-3' Exon F: 5'-atgtatgatgaccccaccct-3' R: 5'-Pso-ctgtagagataggagttgct-3' Exon F: 5'-gtgatacttacatacttgtt-3' R: 5'-Pso-ggctcagcagagtggtgggc-3' Exon F: 5'-Pso-gagcccgtgcagccatcagc-3' R: 5'-cgtgtccagccttcaggcag-3' Exon 3A 3AF: 5'-Pso-tgtgtctttctgtttgtccc-3' AR: 5'-gatttgggcttcttaggtgg-3' Exon 3B 3BF: 5'-Pso-cctcccggcgagagcagaaa-3' BR: 5'-tgacctgggtggatgtggtg-3' Exon 3C 3CF: 5'-tgccttttcaaacttcgcca-3' CR: 5'-Pso-tgaggaggcgctgctgctgc-3' Exon 3D 3DF: 5'-gcagcagcagcgcctcctca-3' DR: 5'-Pso-tggcaaccgcgggctgagtca-3' Exon 3E 3EF: 5'-Pso-tgccccaaggagccagctaa-3' ER: 5'-gctttgcaatccgctccgtg-3' Pso, psoralen TA (n = 1)] (Fig. 1B and C) and one silent polymorphism (S194S). The nonsense mutations were due most frequently (four out of five cases) to C T transitions occurring in CpG dinucleotides. All the genomic deletions resulted in a shift of the translation reading frame leading to a premature termination of MeCP2 synthesis. The screening of the PCR fragment 3E revealed the presence of an abnormal migration pattern of exon 3 in only one RTT sporadic case. The sequence revealed an A C substitution changing the normal TGA stop codon of the MECP2 gene to a TGC cysteine codon (Fig. 1B). This base-pair substitution was predicted to generate a MeCP2 protein of 513 amino acids, 27 amino acids longer than the normal protein. If the mutation destroys a unique restriction site, this event was used to study the segregation of the base substitution in the family of the patient (Table 2). In addition to these mutations revealed by DGGE screening, analysis of the MECP2 gene by direct sequencing in eight patients showed five additional mutations located also in exon 3 (R168X, R255X, R306C, P322A and 1194insT). We analysed DNA samples from both parents of all individuals with a MECP2 mutation and none of the parents samples showed any abnormalities by DGGE or restriction analysis, demonstrating that these are de novo mutations, except in one case. In one family with a RTT girl bearing the T158M mutation, we also found the same mutation in the unaffected mother, but not in the normal brother. This mutation has already been described by Amir et al. (4). This change may disrupt the structure of the MBD, thereby interfering with its function. The crucial role of this domain is suggested by the fact that deletion of residues from MeCP2, which corresponds to most of the hairpin loop, resulted in a total loss of methyl-cpg-binding activity (10). To clarify the discrepancy between the phenotype and the genotype in the mother, we analysed the X-inactivation pattern in this family. We used DNA prepared from peripheral blood leukocytes and assessed the X-chromosome inactivation pattern as described by Allen et al. (11) using PCR analysis of the androgen receptor gene, which contains two methylation-sensitive sites (HpaII and HhaI) flanking a polymorphic trinucleotide repeat in the first exon. Interestingly, we found that the patient s mother presented a totally skewed pattern of X inactivation (data not shown). In the affected girl, the analysis was not conclusive because the trinucleotide repeat marker was not informative. Although these analyses did not allow unambiguous demonstration that this mutation is deleterious and assessment of whether the mutated allele lies on the inactive X chromosome, it is reasonable to propose the skewed pattern of X inactivation as the likely event involved in the rescue of the mother phenotype. Genotype phenotype correlations We first focused on the group of patients with mutations in MECP2 and looked for phenotype genotype correlation, taking into account for the genotype the type of mutation (missense, nonsense or frameshift) and its position with respect to the functional domains and the 3' end of the open reading frame (ORF). This analysis did not show any significant correlation. We next compared the 30 patients (mean age ± 1 SD: 14.6 ± 5 years) who had a mutation in the MECP2 gene, with the 16 patients (mean age ± 1 SD: 13.5 ± 5 years; mutated versus nonmutated NS) who had no detected mutation for different clinical items obtained before the onset of the genetic study, as

5 Human Molecular Genetics, 2000, Vol. 9, No Table 2. Types of MECP2 mutation detected in RTT individuals Patient no. Base change Mutation Sibling investigation methods 1 808C T R270X NlaIV ( ) 2 317G A R106Q DGGE 3 808C T R270X DGGE; NlaIV ( ) 7 502C T R168X HhpI(+) 8 473C T T158M NlaIII (+) 9 502C T R168X HphI(+) C T R294X DGGE C T R306C HhaI( ) C T T158M NlaIII (+) del26 DGGE; MnlI ( ) A T R198X MaeII (+) C T R168X HphI(+) C T R306C HhaI( ) insT BspWI (+) C T T158M NlaIII (+) C T R255X DGGE del17 DGGE; MnlI ( ) C T R168X HphI(+) C G P302R SfaNI (+) C G P322A DraII (+) C T R270X DGGE; NlaIV ( ) C T R270X DGGE; NlaIV ( ) A C X487C Fnu4HI (+) C T R270X NlaIV ( ) insA DGGE C T R294X DGGE del26 DGGE; MnlI ( ) C T R294X DGGE C T R255X DGGE del10 DGGE; MnlI ( ) Mutations either create (+) or destroy ( ) the restriction site for the enzyme shown in the right column. described in Materials and Methods (2,7). The general characteristics of the families were similar in the two groups with a mean of 2.8 (mutated patients) and 2.3 (non-mutated) children per family. The repartition of sex among brothers and sisters was identical in the two groups but the female:male ratio in families with more than one child (including the propositus) was slightly skewed towards girls in both cases (62 versus 76% girls, mutated versus non-mutated). As expected in these patients pre-selected for having a typical case of RTT, the frequency of the most characteristic symptoms of RTT (normal initial development, acquired microcephaly, stereotypic hand movement, phase of social withdrawal, breathing dysfunction) were identical in both groups. Although statistically not significant, it is worth mentioning that the patients with detected mutation lost more frequently acquired purposeful hand skills (71 versus 50%, mutated versus non-mutated; P =0.15),had more frequent peripheral vasomotor disturbances (77 versus 50%; P = 0.33) and epilepsy (41 versus 23%; P =0.25)while they were more frequently able to walk (21 versus 55%; P = 0.24). DISCUSSION In order to evaluate the prevalence of RTT related to MECP2 mutations, we have carried out a systematic analysis of the MECP2 gene in 46 typical RTT patients and screened by DGGE (n = 38) and by direct sequencing (n = 8) the whole coding sequence of this gene. Upon analysis by DGGE of exon 1 to exon 3, and sequencing of PCR fragments exhibiting abnormal DGGE migration profiles, we have identified 25/38 (66%). Direct sequencing of the whole coding sequence revealed five of eight mutations (62.5%). Altogether, these analyses allowed identification of disease-causing MECP2 mutations in 65% (30/46) of typical RTT patients. Of these mutations, 12 were novel and clustered in the third exon of the MECP2 gene. Though all patients included in this study have homogeneous clinical phenotype and fulfilled the same diagnosis criteria, and the frequency of mutations in this gene is very high, no mutation was identified in 35% of the screened patients. Although investigations reported in this study cannot exclude the presence of mutations that might lie the 3'-UTR, promoter or intronic sequences, an alternative hypothesis, which takes into account the exclusive involvement of females, could be the involvement in RTT of a putative second X-linked gene. MeCP2 acts as a molecular link by binding to 5-methylcytosine with its MBD domain and to the corepressor Sin3A via its TRD, thus recruiting histone deacetylases and other proteins to the silencing complex. Therefore, X-linked genes encoding the different components of the histone deacetylase complex could be considered as reasonable potential candidate genes for RTT. These studies identified in 30/46 unrelated families 17 different mutations with independent de novo recurrences of most of them (Fig. 2; Table 2), five of these different mutations havealreadybeenreportedbyamiret al. (4)andWanet al. (5), and 12 are novel. Although the spectrum of mutations is very heterogeneous, occurrence of mutations mainly in exon 3, and the multiple recurrences of R270X (five times) and R168X (four times), R255X (twice) and R294X (twice), points to true mutational hotspots that could influence molecular diagnosis strategies of RTT. Including the data reported by Amir et al. (4) and Wan et al. (5), the spectrum of mutations now encompasses eight missense and 15 nonsense or frameshift mutations, including four small deletions, ranging from 10 to 26 bp and localized in the region of the coding sequence of the cdna (Fig. 2). As deletion events in human genes appear to be, at least in part, related to local DNA sequence environment (11), we examined carefully the sequence environment of the short deletions and identified four CCACC direct repeat sequences distributed over the region of interest which is also a very C-rich sequence (Fig. 3). This observation could be coherent with the previously reported model of slipped mispairing (12) as molecular basis for the occurrence of deletions. Among the molecular defects reported in this work, 11 are nonsense or frameshift mutations (R168X, R198X, 677insA,

6 1382 Human Molecular Genetics, 2000, Vol. 9, No. 9 Figure 2. Distribution of the mutations in the MECP2 gene along the coding sequence. Top, mutations identified in this study. Bottom, mutations described previously (4,11). Novel mutations described in this study are underlined. Figure 3. Sequence environment of the small deletions identified in the MECP2 gene. Deleted nucleotides are indicated by bold type. The CCACC direct repeats flanking and/or overlapping the MECP2 gene deletions are underlined. R255X, R270X, R294X, 1156del17, 1158del10, 1163del26, 1165del26 and 1194insT) leading to premature polypeptide chain termination. In contrast to conventional wisdom, mrnas that contain these mutations (also referred to as nonsense mrnas and chain-termination mutations, respectively) rarely produce truncated proteins. Most nonsense mrnas are highly unstable because they are degraded by a decay pathway called nonsense-mediated mrna decay (NMD) (13,14). This process, whereby mrnas are monitored for errors that arise during gene expression, has been found in several species, including human (15,16). Typically, chaintermination mutations that reduce mrna abundance by reducing the half-life of mrna behave like loss-of-function alleles, except in some cases (i.e. mutations near the 3' end of the ORF), where the RNA surveillance system is bypassed. Although most of the mutations described here are located in the first half of the coding region and should in theory trigger the NMD process, investigation of transcripts and proteins resulting from the diverse panel of mutations is still a relevant issue that might provide additional information about MecP2 functional domains. During this study, we identified three novel mutations causing amino acid substitutions. These three missense mutations, R106Q, P302R and P322A, are drastic amino acid changes at the protein level. Moreover, all these amino acids are conserved in human, mouse and Xenopus laevis (4). The R106Q mutation is located in the MBD of the protein. A mutation identified previously in this domain is located at the same codon (R106W) (4). These two amino acid substitutions may reduce or abolish methyl-cpg binding. The P322A mutation is located in a conserved C-terminus and the other missense mutation P302R is located at the end of the TRD. Replacement of proline by arginine or alanine may cause abnormal folding of the protein. All these base substitutions were absent in >100

7 Human Molecular Genetics, 2000, Vol. 9, No normal X chromosomes. Altogether, these data suggest that these DNA variants are disease-causing mutations rather than polymorphisms. In addition, we identified an original mutation in the termination codon. The 1461A C substitution was predicted to generate a MeCP2 protein of 513 amino acids (27 amino acids longer than the normal one). It has been reported previously that the abnormal mrna translation due to a mutation in the termination codon is associated with decreased mrna stability such that no mrna or protein synthesis from the mutant allele can be detected in cells (17). Although mutations identified are heterogeneous (17 different mutations in this study), nearly 65% of typical RTT individuals were found to have changes within the coding region of exon 3. In our population, the novel R270X nonsense mutation accounts for 16% of our RTT chromosomes. This finding suggests that initial analysis of this exon would provide the most efficient approach in a mutation detection protocol. Concerning the remaining 35% of typical RTT with no mutation in the coding sequence of the MECP2 gene, our further investigation will focus on the study of the 5'- and 3'-UTR combined with quantitative studies of MECP2 mrna. In addition, we will search for X-linked candidate genes required for methyl-cpg-binding protein complexes. MATERIALS AND METHODS Patients The 46 sporadic cases of RTT investigated in this study were issued from the French register established in To validate the diagnosis, we used the international criteria adopted by the Rett Syndrome Diagnostic Criteria Work Group (1,7). Briefly, for all patients, after normal general and psychomotor development up to the age of 7 18 months, development stagnation occurred, followed by rapid deterioration of higher brain functions. This deterioration led to severe dementia, autism, loss of purposeful use of the hands, jerky truncal ataxia and acquired microcephaly. Additional insidious neurological abnormalities, such as spastic parapareses, vasomotor disturbances of the lower limbs and epilepsy were also observed. For each case, the clinical status was first reviewed by a skilled and experienced neuropaediatrician and the diagnosis was graded as established, probable or possible. Independently, a questionnaire evaluating 50 items, including the necessary, supportive and exclusion criteria of the Rett Syndrome Diagnostic Criteria Work Group was completed by two specially trained investigators. At the same time, a blood sample of the patient and her family was obtained and a lymphoblastoid cell line was established with family consent. For different genetic studies (and well before the initiation of the present study), we selected patients who: (i) had an established clinical diagnosis of RTT; (ii) fulfilled more than seven of the nine necessary criteria in the questionnaire, at least one of the supportive criteria and none of the exclusion criteria; (iii) were >8 years old. These patients were tested in the present study. The χ 2 test was used to analyse the comparative frequency of the most characteristic symptoms of RTT in patients with detected mutation and in patients with no detected mutation. Mutation analysis Denaturing gradient gel electrophoresis (DGGE). DNA was extracted from peripheral blood leukocytes or lymphoblastoid cells, and the three exons and the flanking intronic sequences of the MECP2 gene were separately PCR-amplified from genomic DNA using primers listed in Table 1, with psoralen clamps. DGGE conditions were chosen according to the Meltmap program, kindly provided by L.Lerman and colleagues (9). The denaturants were 7 M and 40% formamide, and gels were run at 60 C (8,9). PCR products were subjected to electrophoresis as described in Table 1. Formal consents were obtained from the families for mutation screening. Mutation identification. PCR products showing an abnormal migration pattern on DGGE analysis were sequenced directly on an automated sequencer (ABI 373; Perkin Elmer, Foster City, CA) using the Dye Terminator method. Every sequence variation was checked by restriction analysis of genomic DNA. In eight typical RTT patients, screening of the whole coding sequence of the MECP2 gene has been performed by direct sequencing. ACKNOWLEDGEMENTS We thank the patients and the families for their contribution in this study. We also thank Genethon bank for providing DNA samples, and colleagues from the Société Française de Neuropédiatrie who participated to the diagnosis evaluation and allowed us to study their patients. This work was supported mainly by the Association Française du syndrome de Rett (ASFR). This work was also supported by grants from Institut National de la Santé et de la Recherche Médicale (INSERM), the Association Française contre les Myopathies (AFM), the Fondation Jerome Lejeune (FJL) and the Fondation pour la Recherche Médicale. REFERENCES 1. Rett, A. (1966) Ueber ein eigenartiges hirnatrophisches syndrom bei hyperammoniamie in kindesalter. Wien. Med. Wsch., 116, Hagberg, B., Aicardi, J., Dias, K. and Ramos, O. (1983) A progressive syndrome of autism, dementia, ataxia and loss of purposeful hand use in girls: Rett s syndrome: report of 35 cases. Ann. Neurol., 14, Webb, T., Clarke, A., Hanefeld, F., Pereira, J.L., Rosenbloom, L. and Woods, C.G. (1998) Linkage analysis in Rett syndrome families suggests that there may be a critical region at Xq28. J. Med. Genet., 33, Amir, R.E., Van den Veyver, I.B., Wan, M., Tran, C.Q., Francke, U. and Zoghbi, H.Y. (1999) Rett syndrome is caused by mutations in X-linked MECP2, encoding methyl-cpg-binding protein 2. Nature Genet., 23, Wan, M., Sung Jae Lee, S., Zhang, X., Houwink-Manville, I., Song, H.R., Amir, R.E., Budden, S., Naidu, S., Pereira, J.L.P. et al. (1999). Rett syndrome and beyond: recurrent spontaneous and familial MeCP2 mutations at CpG hotspots. Am.J.Hum.Genet., 65, Lewis, J.D., Meehan, R.R., Henzen, W.J., Maurer-Fogy, I., Jeppesen, P., Klein, P. and Bird, A. (1992) Purification, sequence and cellular localization of a novel chromosomal protein that binds to methylated DNA. Cell, 69, The Rett Syndrome Diagnostic Criteria Work Group (1988) Diagnostic criteria for Rett syndrome. Ann. Neurol., 23, Bienvenu, T., Cazeneuve, C., Kaplan, J.C. and Beldjord, C. (1995) Mutation heterogeneity of cystic fibrosis in France: screening by denaturing gradient gel electrophoresis using psoralen-modified oligonucleotide. Hum. Mutat., 6,

8 1384 Human Molecular Genetics, 2000, Vol. 9, No Myers, R.M., Maniatis, T. and Lerman, L.S. (1987) Detection and localisation of single base changes by denaturing gradient gel electrophoresis. Methods Enzymol., 155, Nan, X., Meehan, R.R. and Bird, A. (1993) Dissection of the methyl-cpg binding domain from the chromosomal protein MeCP2. Nucleic Acids Res., 21, Allen, R.C., Zoghbi, H.Y., Moseley, A.B., Rosenblatt, H.M. and Belmont, J.W. (1992) Methylation of HpaII and HhaI sites near the polymorphic CAG repeat in the human androgen-receptor gene correlates with X chromosome inactivation. Am.J.Hum.Genet., 51, Cooper, D.N. and Krawczak, M. (1993) Human Gene Mutation. BIOS Scientific Publishers Limited, Oxford, UK. 13. Leeds, P., Peltz, S.W., Jacobson, A. and Culbertson, M.R. (1992) The product of the yeast UPF1 gene is required for rapid turnover of mrnas containing a premature translational termination codon. Genes Dev., 5, Leeds, P., Wood, J.M., Lee, B.S. and Culbertson, M.R. (1992) Gene products that promote mrna turnover in Saccharomyces cerevisiae. Mol. Cell. Biol., 12, Applequist, S.E., Selg, M., Raman, C. and Jack, H.M. (1996) Cloning and characterization of hupf1, a human homolog of the Saccharomyces cerevisiae non-sense mrna-reducing UPF1 protein. Nucleic Acids Res., 25, Sun, X., Perlick, H.A., Dietz, H.C. and Maquat, L.E. (1998) A mutated human homologue to yeast Upf1 protein has a dominant-negative effect on the decay of non-sense-containing mrnas in mammalian cells. Proc. Natl Acad. Sci. USA, 95, Hunt, D.M., Higgs, D.R., Winichagoon, P., Clegg, J.B. and Weatherall, D.J. (1982) Haemoglobin constant Spring has an unstable a chain messanger RNA. Br. J. Haematol., 51,