Comprehensive diagnosis of Rett's syndrome relying on genetic, epigenetic and expression evidence of deficiency of the methyl‐CpG‐binding protein 2 gene: study of a cohort of Israeli patients

Patients and controls

The Israeli RTT cohort included 82 patients of whom 52 were diagnosed with classic and 17 with atypical RTT, including 4 with PSV, 4 with congenital variants, 6 with early seizure onset variants, 2 with forme fruste variants and 1 male variant, using clinical criteria established by Hagberg et al.⁴ The cohort also included four females with Angelman‐like features and nine patients with diagnoses reminiscent of RTT, including seven females with autism spectrum disorder and two males with congenital severe encephalopathy. All the patients were classified by the same paediatric neurologist at the Neuropediatric Clinic at the Sheba Medical Center, Tel Hashomer, Israel, where they continue to attend for clinical follow‐up. The age of patients was in the range 2–40 years. Molecular diagnosis of RTT was carried out as a part of an ongoing study approved by the institutional review board or as an officially authorised diagnosis provided by health insurance. Patients included in the expression studies were recruited after informed consent of the parents approved by the institutional review board. This group included patients with classic (C1–C5) and atypical (A1–A7) RTT. Patients C1–C5 and A1, A3 and A4 were recruited from the Israeli cohort, patient A5 was referred from the Hospital for Sick Children, Toronto, Ontario, Canada, and patients A2, A6 and A7 were referred from the Rett Syndrome Research Foundation, Cincinnati, Ohio, USA. As regards the diagnoses of patients with atypical RTT, A1 and A2 had congenital variants, and A3 and A4 had early seizure onset variants, but A5, A6 and A7 had no specific diagnoses other than atypical RTT. In addition, a sample from an atypical patient with partial deletion of exon 4 (del exon 4a) was provided by Dr Jane Hickey via the Rett Syndrome Research Foundation. Expression studies also included 12 samples from healthy female volunteers (F1–F12) aged 17–40 years.

DNA and RNA extractions

Peripheral blood samples collected in tubes containing EDTA anticoagulant were used for the extraction of genomic DNA by the Puregene kit and total RNA by the Versagene kit (both by Gentra, Minneapolis, MN, USA). Additional blood samples were transferred from Canada and the US in Paxgene tubes (PreAnalytix, Hombrechtikan, Switzerland) and RNA was extracted by the same method. The yield and purity of the DNA and RNA extractions were determined according to the optical density ratio 260/280 nm using a spectrophotometer (GeneQuant, Cambridge, UK). RNA samples were subjected to DNAse I treatment included in the Versagene kit; 10–20 units of protector RNAse inhibitor (Roche Applied Science, Penzburg, Germany) were added and the samples were stored at −20°C.

MECP2 mutation analysis

PCR fragments spanning the MECP2 promoter region (1.2 kb upstream of exon 1) and the entire coding region (exons 1–4) were generated using seven primer pairs (sequences available on request). PCR fragments were purified on silica gel spin columns (Bioneer, Daejon, Korea) and analysed on an ABI Prism 3100 AVANT Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). The presence of mutations was confirmed by replicate sequencing of independently generated PCR products.

Multiplex ligation‐dependent probe amplification

Large deletions spanning MECP2 exons 1–4 were detected using the MLPA‐P015 probe (MRC, Amsterdam, The Netherlands) according to the manufacturer's protocol. MLPA products were separated and identified using the ABI Prism 3100 Genetic Analyzer using Genescan ROX‐500 standards and Genescan software (both by Applied Biosystems). Reproducible deviation of >30% of relative peak area as compared with the normal control sample derived from two independent experiments was considered conclusive.

X chromosome inactivation

The XCI pattern in peripheral blood was analysed using the human androgen receptor assay.¹⁶ Briefly, genomic DNA was digested with the methylation‐sensitive enzyme HpaII for preferential cleavage of the unmethylated‐active alleles. Digested and undigested DNA samples were subjected to PCR of the human androgen receptor gene, including the highly polymorphic trinucleotide repeat region. PCR products were separated using the ABI Prism 3100 Genetic Analyzer, calculating the peak areas of the undigested and digested alleles using Genescan ROX‐500 standards and Genescan software. Estimation of XCI included correction for the large digested allele peak area considering the ratio between peak areas of the small and the large undigested alleles. Skewed XCI was considered to be significant for ratios ⩾75%, whereas reported values were derived from two independent and reproducible experiments (<10% difference). The parental origin of the preferentially activated chromosome was further determined by comparing the profiles of the patient's and the parental samples.

cDNA synthesis

Double‐stranded cDNA was generated from 500 ng of total RNA in the presence of p[dN]₆ random primer hexamers (Roche Applied Science), using an Omniscript RT kit (Qiagen Hilden, Germany) according to the manufacturer's protocol. Reactions were carried out in a total volume of 40 μl.

TaqMan reactions

Relative expression levels of MECP2_e1 and MECP2_e2 transcription isoforms were determined using primers designed by Primer Express software, TaqMan probes and PCR kit (all from Applied Biosystems). The MECP2_e1 assay, designed by us, included forward primer from exon 1 (5′‐CGG AGG AGG AGG AGG A) and reverse primer from exon 3 (5′‐GGA GGT CCT GGT CTT CTG ACT T), producing a 63‐bp amplicon, and the TaqMan probe for the exon 1–3 junction (gene accession BX538060). The MECP2_e2 commercial assay included the TaqMan probe for the exon 2–3 junction. Both MECP2 probes contained 5′‐6‐carboxyfluorescein fluorophore reporter and 3′‐non‐fluorescent quencher. The commercial RNAseP kit consisting of RNAseP‐specific primers and VIC‐labelled TaqMan probe was used as an internal reference, enabling multiplexing of MECP2 and RNAseP assays. Separate pre‐runs of varying primer concentrations were carried out to obtain the highest intensity and specificity of the fluorescent reporter signal. The results were validated in separate reactions including the commercial 6‐carboxyfluorescein‐labelled ornithine decarboxylase 1 probe and primers as another reference. Quantitative PCR reactions (QtPCRs) were carried out in a volume of 20 μl in 96‐well optical plates using a common Mastermix including 2× TaqMan Universal PCR Mastermix, 900 nM MECP2_e1 or MECP2_e2 primers, 250 nM MECP2_e1 or MECP2_e2 probe and 20× RNAseP kit, and high‐performance liquid chromatography‐pure water. Aliquots of 2 μl of each cDNA species were dispensed in three wells for triplicate reactions. Reactions were carried out on ABI Prism 7000 (Applied Biosystems) under uniform conditions, including a pre‐run at 50°C for 2 min and 95°C for 10 min, and 40 cycles at 95°C for 15 s and at 60°C for 1 min. Each reaction plate included triplicate samples of a normal control pool constructed from 4–5 cDNAs generated from normal females (fig 1). Each plate also included reactions with no‐template control and RNA template (RT) for monitoring of the background signal and DNA contamination, respectively.

Quantitative analysis

Data were evaluated using the ABI Prism 7000, comparing the levels of MECP2_e1 and MECP2_e2 transcripts with those of the RNAseP internal reference gene. In most cases, the threshold of QtPCRs was automatically set at 10 standard deviations (SD) above the mean baseline emission, representing the background signal, and the specific reaction cycle threshold (Ct) number was determined relative to this threshold. Only samples with at least three QtPCR results (triplicates) were included in the data analysis. The comparative (ddCt) method was used to calculate the relative transcript number,¹⁷ previously verifying that QtPCR efficiencies of the target and reference genes were almost equal. Efficiencies of MECP2_e1, MECP2_e2 and RNAseP reactions were determined from standard curves generated by serial twofold dilutions of the cDNA sample and estimation of the cycle threshold at each dilution (fig 2). Reaction efficiencies, calculated as E (efficiency) = 10^−1/slope, yielded values of 1.84, 1.86 and 1.89 for MECP2_e1, MECP2_e2 and RNAseP, respectively, which could be considered approximately equal (standard error <0.02). Relative transcript number in the patient samples was determined according to the ddCt method comparing the patient's sample and the pooled sample of normal female controls using the following equations:

dCt=Ct_MECP2–Ct_RNAseP and ddCt=dCt_patient–dCt_{control pool}

and on inclusion of a correction for mean PCR efficiency,

MECP2_e1 transcript number=1.86^−(ddCt),

MECP2_e2 transcript number=1.84^−(ddCt)

Statistical methods

For the purpose of statistical analyses of MECP2_e1 and MECP2_e2 expression data, patients were grouped into four groups according to the MECP2 mutation type (I, splice‐site mutations; II, large deletions; III, nonsense and frame‐shift mutations; and IV, mis‐sense and in‐frame mutations). The individual mean MECP2_e1 and MECP2_e2 levels in the patient groups and those in the normal control group were compared using analysis of variance with Bonferroni correction for multiple comparisons. Correlation analysis between the levels of both MECP2 isoforms was carried out using the Spearman correlation procedure. Multinomial (polytomous) regression was applied to predict the relatedness of patients with no known MECP2 mutation to mutation groups I–IV, while combining groups I and II. All data were analysed using SAS V.9.1.3 for Unix via procedures MIXED and LOGISTIC.

Analyses of MECP2 mutations, MLPA and XCI

MECP2 mutation analysis of the Israeli cohort (n = 82) yielded 54 patients with MECP2 sequence variations, including 42 with classic RTT, 4 with PSV, 2 with congenital variants, 1 with forme fruste variant and 5 with atypical RTT. Our mutation detection rate is 80% (42/52) as regards patients with classic RTT. Table 1 summarises the clinical and mutation data of patients, including XCI regarding preferential activation of the maternal or paternal X chromosome. Apart from the known MECP2‐coding mutations and various microdeletions at the 3′‐coding end, we identified a novel mis‐sense mutation at c453C→G (p D151E) in one patient with PSV, located at the MECP2 methyl‐binding domain. In two patients with classic RTT, we detected a rare GT deletion in an intron 1 donor splice site. In one patient with classic RTT, we detected changes in both MECP2 alleles, including a 1.1‐kb deletion in the intron 3–exon 4 region and a novel maternally inherited sequence variation at c824T→C (p V275A). Using MLPA, we detected a large (⩾4.2 kb) deletion spanning parts of exon 4 and the 3′ untranslated region (UTR) in one congenital variant. We also detected four maternally inherited sequence variations (table 1, lower panel), among which c378‐19delT, c602C→T (p A201V) and c1451G→C (p R484T) have been reported in the International Rett Syndrome Association database as polymorphisms, but c753C→T (p P251P) is a novel mutation. Among the patients with MECP2 mutations, we detected significantly skewed (⩾75%) XCI in 11 of 48 (23%) informative cases, including three patients with preferential activation of the paternal X and eight patients with preferential activation of the maternal X chromosome. Ultimately, ten patients with classic RTT remained with no indication of MECP2 mutation, either by sequence analysis or by MLPA. Five of these patients were further included in quantitative analyses of MECP2 expression in peripheral blood (C1–C5), together with seven patients with atypical RTT with no mutation findings (A1–A7).

Development of quantitative assay for the analysis of MECP2 expression levels

We developed an additional strategy for the detection of MECP2 deficiency, introducing quantitative analysis of MECP2 expression levels in peripheral blood, using TaqMan probes for the target genes, MECP2_e1 and MECP2_e2, and RNaseP as a reference gene. QtPCR efficiencies of the target and reference genes were estimated from the cycle threshold values obtained from serial dilutions of a cDNA sample. Figure 2 shows that all QtPCR efficiencies were similar and within the range 92–95% of maximal efficiency. MECP2_e2 was detected at a lower cycle threshold (difference of 4Ct), suggesting that the levels of MECP2_e2 are ∼8 times as abundant in the peripheral blood, which is also consistent with other reports.⁹ The normal range of MECP2_e1 and MECP2_e2 expression levels in peripheral blood was estimated by the analysis of a series of normal female control samples (F1–F12). Figure 1 shows that in most cases, MECP2_e1 and MECP2_e2 expression levels in the normal female controls were analogous and differed by factor of 1.5 (0.09) for MECP2_e1 and 1.7 (0.14) for MECP2_e2. Thus, in the following expression assays we included a pooled sample of four cDNAs from normal female controls.

Analysis of MECP2_e1 and MECP2_e2 expression levels in patients with known mutations

To validate the consistency of quantitative expression assays with the presence of known MECP2 mutations, we examined the expression levels of MECP2_e1 and MECP2_e2 in the peripheral blood of 15 such patients. Figure 3A,B shows this analysis. Dotted lines indicate the normal range of MECP2 expression in normal female controls. Grouping the patients into four mutation groups, including splice‐site mutations, large deletions, truncating (nonsense and frame‐shift) and non‐truncating (mis‐sense and in‐frame) mutations, we compared MECP2 expression levels between these groups and the group of normal controls. We found significant differences in MECP2 expression levels between various mutation groups (F = 19.01, df = 4, p<0.001 for MECP2_e1; and F = 23.01, df = 4, p<0.001 for MECP2_e2 by analysis of variance and Bonferroni correction). The group with the splice‐site mutations had significantly lower levels of both MECP2 isoforms than the normal controls (p<0.001 for MECP2_e1 and p = 0.003 for MECP2_e2 by analysis of variance and Bonferroni correction). The large deletions were lower specifically for MECP2_e2 (p = 0.069 for MECP2_e1; p<0.001 for MECP2_e2); by contrast, the truncating mutations were lower for both isoforms (p<0.001 for MECP2_e1; p = 0.002 for MECP2_e2). The non‐truncating mutations were similar to the normal controls for both MECP2 isoforms (p = 1.000 for MECP2_e1; p = 0.161 for MECP2_e2). The overall expression levels of MECP2_e1 and MECP2_e2 were significantly correlated (Spearman r = 0.76; p<0.001).

Quantitative expression analysis of patients negative for MECP2 mutations

To obtain evidence of MECP2 mutations in patients with no previous mutation findings, we included 12 such patients in quantitative expression assays, five with classic (C1–C5) and seven with atypical (A1–A7) RTT. We took special care of the transport of blood samples from Canada and the US to ensure that the authentic RNA concentration was maintained (see Methods). Figure 3A,B shows our findings in this group. We correlated the patients with no mutations to the previous groups with mutations, while combining the groups with the splice‐site mutations (group I) and the large deletions (group II) on the basis of relative similarity of small groups (p = 0.135 for MECP2_e1 and p = 1.000 for MECP2_e2 by analysis of variance and Bonferroni correction). This analysis suggested that patients C2, C4, C5, A2, A5 and A7 may belong to mutation groups I and II (predicted probabilities to belong to these groups are p = 0.999, 0.999, 1.000, 0.919, 0.993 and 1.000, respectively, including MECP2_e1 and MECP2_e2 in polytomous regression analysis). Patients A3 and A6 were related to group III with the truncating mutations (p = 0.979 and 0.919, respectively). Patient C3 was related to the group with normal expression levels by both MECP2 isoforms (p = 1.000). Patients C1, A1 and A4 had discordant levels of MECP2_e1 and MECP2_e2; in particular one of the isoforms was overexpressed.

Effect of XCI on quantitative expression of MECP2 in peripheral blood

Among the 27 patients included in the quantitative expression assays, seven were detected with considerably skewed XCI. Preferential activation of the paternal X chromosome was detected in patients with the splice‐site mutation (donor 1a), the large deletion (del exon 4b), p R270X (XCI data in table 1) and patient C2 (80% XCI). Preferential activation of the maternal X chromosome was found in patients with p R255X and p R106W (table 1) and patient C4 (80% XCI). Patient C1 was non‐informative and other patients had random XCI. We observed interindividual differences between the patients with the same splice‐site mutation (donors 1a and 1b), as donor 1b with preferential expression of the paternal X chromosome had lower MECP2_e1 and MECP2_e2 expression levels than donor 1a with random XCI (fig 3A,B). Also, in the two patients with two similar nonsense mutations, p R270X with preferential activation of the paternal X had lower MECP2_e1 and MECP2_e2 expression levels, whereas p R255X with activation of the maternal X chromosome had normal levels of both MECP2 isoforms.