Materials and methods

Ten unrelated patients with “classical” Rett syndrome and normal controls were used in the mutation screening for six candidate genes. For mutation screening of the MECP2gene, 59 sporadic cases and nine familial RTT cases were included. Genomic DNA was extracted from blood samples using standard methods. Total RNA was isolated from lymphoblastoid cell lines using the Ultraspec^TM-II RNA kit (Biotecx). First strand cDNA was synthesised using the First-Strand cDNA Synthesis Kit (Pharmacia Biotech). PCR amplification was carried out using standard protocols. The coding region of SOX3 is GC rich; therefore, a higher temperature (98°C) was used in the denaturing step during each PCR cycle. Primer pairs were designed from published sequences to amplify the entire coding regions of the seven candidate genes (table 1). Taq PCR Core Kit (Giagen) was used for MECP2-3-2 in order to increase PCR sensitivity. PCR products were isolated from low melting temperature agarose gels and purified using Wizard^TM PCR Preps DNA Purification System (Promega). Mutation analysis was performed by direct sequencing (Thermo Sequenase radiolabelled terminator cycle sequencing Kit, Amersham Life Science) followed by electrophoresis on 6% denaturing polyacrylamide gels and autoradiography or using Applied Biosystem (ABI) 377 Sequencer. For the six deletions and one insertion detected in this study, both methods were performed to confirm the mutations.

Postmortem brain tissue samples were partly obtained from the Harvard Brain Tissue Resource Center and partly provided by Dr D Armstrong, Baylor College of Medicine, USA. Fresh frozen material from the superior frontal cortex, motor cortex, thalamus, and brain stem of six Rett girls (aged 6 to 17 years) and from seven controls (both males and females, aged 2 to 94 years) was available. Unfixed, 14 μm thick cryosections were made and stored at −20°C until use. Sections were removed from the freezer and air dried five hours before hybridisation. In situ hybridisation was carried out as described elsewhere.28 Briefly, oligonucleotides (table 2) complementary to the four genes of interest and one random nucleotide of similar length (negative control) were 3′ end labelled with³⁵S-dATP (to ensure high sensitivity and accurate cellular localisation) using terminal transferase. Forty ng of labelled oligonucleotide per slide were added to 100 μl of hybridisation cocktail containing 4 × SSC, 50% formamide, 1 × Denhardt's solution, 1% sarcosyl, 0.02 mol/l phosphate buffer (pH 7.0), 10% dextran sulphate, 50 mg sheared salmon sperm DNA, and 200 mmol/l dithiothreitol. Slides were incubated overnight at 42°C in a humidified chamber. Subsequently, the slides were rinsed five times for 45 minutes at 60°C in 1 × SSC, dehydrated through graded ethanol, air dried, dipped in photographic emulsion (Kodak NTB2 at 1:2 dilution), and exposed in the dark for six weeks. After developing, the slides were counterstained with cresyl violet and mounted.

Results

Mutations in the coding region of MECP2were found in 35% of our cases. TheMECP2 gene consists of three exons. The whole coding regions including the exon-intron junction were amplified by PCR (∼1.5 kb, Genbank accession Nos AF030876 and X99686). Primer information is shown in table 1. Altogether, we identified six deletions (from 7 bp up to 101 bp), one insertion (10 bp), six missense mutations, and three nonsense mutations (table 3) in 27 of 78 patients. Five of the six deletions and the insertion cause frameshifts, and the same mutations were not found in their parents or normal sibs. R106W, D97E, R133C, F157I, and T158M are located in MBD, R306C, R255X, and R294X are located in TRD, and R168X is in the corepressor interacting region. One patient has a 7 bp deletion at nt 1256 and a 70 bp deletion at nt 1172. Another patient has a 10 bp insertion at nt 1127 and a 55 bp deletion at nt 1215. All the deletions and the insertion detected in our patients are clustered in a 150 bp region (nt 1127-1275) which contains tandem repeats. This indicated that the region could be a hot spot for deletion owing to DNA polymerase slippage.

The UBE1 gene transcript (GenBank accession No M58028) is 3.5 kb in length and we used six pairs of primers to amplify the whole coding sequence (3.2 kb). In contrast, a single pair of primers was used to cover the entire coding region (spanning 476 bp) of UBE2I (GenBank accession No U45328). The gene structure of GdX is known (Genbank accession No J03589) and primers were therefore designed using intronic sequence to amplify two PCR fragments, each containing two complete exons and the intervening intron. In order to increase the PCR specificity and sensitivity for GdX, a two round, nested PCR was used (GDX1-2 in table 1), and in addition, theTaq PCR Core kit was used. No mutations were detected in the coding regions of UBE1,UBE2I, or GdX.

SOX3 comprises a single exon. Three pairs of primers were used to amplify and sequence the 1331 bp region containing the coding sequence. An 8 bp insertion (TGCCGCCG) in the gene was identified (between bases 1327 and 1328 of the sequence contained in the Genbank entry with accession number X71135). However, the insertion was present in all 12 subjects tested (10 patients and two normal controls) and the deletion variant may therefore represent an error in the original database sequence.

The coding region of GABRA3 spans ∼1.4 kb (Genbank accession No A28102) and three primer pairs were used to amplify the coding region. The two most 5′ fragments were amplified using a two round, nested PCR strategy to increase PCR specificity and sensitivity. Three pairs of primers were designed to amplify the entire cDNA sequence of CDR2 (∼1.3 kb, Genbank accession No M63256). No mutations were found in either gene.

From the in situ hybridisation experiments, representative photomicrographs are shown in figs 1 and 2.L1CAM mRNA was detected at relatively high levels in neurones and lower levels in glia in all brain regions investigated (prefrontal cortex shown in figs 1 and 2, motor cortex, thalamus, and brain stem). A strong ubiquitous signal was generated byMECP2 specific oligonucleotides (fig 2). Low levels of GABRA3 andGdX expression were found in sections containing tissue from these regions (fig 1). There were no detectable differences in expression levels or patterns for any of these four genes between Rett patients and controls.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Bright field microscopy of neurones in the frontal cortex showing GABRA3 and L1CAM expression. Panels show representative fields from sections from a normal control (A-D) and a Rett patient (E-H). The first (A, E) and the second (B, F) probe for GABRA3 showed similarly low levels of expression. Relatively strong expression levels were detected with the probe against L1CAM. (C, G) Hybridisation with a random oligonucleotide resulted in low background signal (D, H). No differences could be observed between patients and controls for any of the probes used.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Bright field microscopy of neurones in the frontal cortex showing MECP2 and L1CAM expression. Panels show representative fields from sections from a normal control (A-C) and a Rett patient (D-F). MECP2 (A, D) was expressed at similar levels as L1CAM (B,E). The negative control hybridisation (C, F) displays very low background signal. MECP2 was expressed ubiquitously in patients and controls in all brain tissues tested (besides the prefrontal cortex shown in this figure we also investigated expression in the motor cortex, thalamus, and brain stem, data not shown). No differences in expression between different brain regions or patients versus controls could be detected.

Discussion

Recent evidence from genetic mapping using familial Rett cases has strongly indicated the presence of a gene for RTT in Xq28, finally leading to the discovery of mutations inMECP2. Because not all patients investigated so far carry mutations in MECP2 and because of the clinical variability seen in RTT, it is possible that this disorder may be genetically heterogeneous. In this study, we therefore determined the percentage of our Rett patients withMECP2 mutations to be 35% and screened further X chromosomal and autosomal candidate genes on the basis of their genomic locations or physiological or genetic characteristics. For some of the candidate genes investigated here, we chose to sequence cDNA products, because full gene structures and sequences were not available for all the genes. In this way, we expected to be able to find mutations in the coding sequence that would be predicted to alter the amino acid sequence of the proteins. This strategy cannot identify mutations in the promoter or 5′ or 3′ untranslated regions. In addition, truncating mutations that reduce RNA stability might not be apparent. We can be sure that the latter problem was not encountered for most of the genes investigated here, because we were able to show the presence of heterozygous polymorphisms in several subjects (not shown), suggesting that effective biallelic amplification was occurring. It would clearly be desirable to extend these studies to include the remaining parts of these genes, to investigate whether other common types of mutation in these regions play a role.

By screening the MECP2 gene, 16 mutations were identified including missense mutations, nonsense mutations, an insertion, and deletions. Most of the point mutations are C→T transitions, which are located in the MBD or TRD or corepressor interacting region. The six deletions and one insertion are exclusively clustered within a 150 bp region of exon 3. Surprisingly, we detected two de novo deletions in the “hot spot” region in one patient and one deletion and one small insertion in the same region in another patient. Since it is extremely rare that two de novo mutations happen to occur in a small region and occur twice (two patients), the region is likely to be a hotspot for deletion or insertion. This region contains several tandem repeats; therefore, the region could be unstable owing to DNA polymerase slippage.

The missense mutation R306C was found in seven out of 78 patients (9%). In order to rule out the possibility of a common polymorphism, we used HhaI restriction enzyme to test 47 ethnically matched normal controls. None of the 47 normal controls carried this polymorphic site (data not shown). After completely sequencing the entire coding region, mutations were detected in only 35% of patients. There are some mutations that may be missed if they are located in introns, the promoter region, or the 3′ untranslated region or consist of huge deletions. However, we have not identified any mutation in our nine familial RTT cases. Our results indicated that there might be more than one gene for RTT.MECP2 may play a key role in the development of the disease. Other genes that may interact withMECP2 or in the same pathway may contribute to the subset of RTT patients remaining unexplained so far.

Skewed X chromosome inactivation has been found in some of our familial RTT cases (paper submitted). The XIST gene (X inactive specific transcript), located at the X inactivation centre, XIC, and expressed exclusively from the inactive X chromosome, is thought to be the major gene involved in initiating the process of X chromosome inactivation. Previous studies of two unrelated families with skewed X chromosome inactivation have shown a promoter mutation in the XIST in nine females, indicating that there may be an association between alterations in the regulation ofXIST expression and X chromosome inactivation. We then used the patients with skewed X chromosome inactivation and some other RTT patients withoutMECP2 gene mutation to test if there is any mutation in the promoter region of XIST. We could not find any mutation in the promoter region ofXIST in any of the 10 patients tested (data not shown).

We also used in situ hybridisation to investigate the expression levels of a set of candidate genes. We predicted that mutations in these genes that caused gross alterations in transcription rates or in mRNA stability or localisation would be detected by this technique. The ability to detect such changes might be of use in narrowing down the search for candidates on which to perform a mutation scan. In the case of MECP2, changes in regional or global expression levels may be important clues for an understanding of the pathogenetic mechanisms underlying the syndrome. We observed no gross alterations of expression levels in the four genes tested (GABRA3, GdX, L1CAM, andMECP2), but two of them,GABRA3 and GdX,were expressed at relatively low levels and subtle changes in expression pattern or levels in patients would therefore be difficult to detect. When using postmortem material, there is the added danger that global degrading processes acting on mRNA will have proceeded too far to be able to detect transcripts. In this study, we detected high levels of L1CAM andMECP2 transcripts, suggesting that this problem did not occur in our samples.