Rett syndrome (RTT) is a major neurodevelopmental disorder, characterized by mental retardation and autistic behavior. Mutation of the MeCP2 gene, encoding methyl CpG-binding protein 2, causes the disease. The pathomechanism by which MeCP2 dysfunction leads to the RTT phenotype has not been elucidated. We found that MeCP2 directly regulates expression of insulin-like growth factor binding protein 3 (IGFBP3) gene in human and mouse brains. A chromatin immunoprecipitation assay showed that the IGFBP3 promoter contained an MeCP2 binding site. IGFBP3 overexpression was observed in the brains of mecp2-null mice and human RTT patients using real-time quantitative polymerase chain reaction and Western blot analyses. Moreover, mecp2-null mice showed a widely distributed and increased number of IGFBP3-positive cells in the cerebral cortex, whereas wild-type mice at the same age showed fewer IGFBP3-positive cells. These results suggest that IGFBP3 is a downstream gene regulated by MeCP2 and that the previously reported BDNF and DLX5 genes and MeCP2 may contribute directly to the transcriptional expression of IGFBP3 in the brain. Interestingly, the pathologic features of mecp2-null mice have some similarities to those of IGFBP3-transgenic mice, which show a reduction of early postnatal growth. IGFBP3 overexpression due to lack of MeCP2 may lead to delayed brain maturation.
Methyl CpG-binding protein-2 (MeCP2) serves as a major transcription repressor by forming a complex with histone deacetylase and Sin3A (1-3). Mutation of MeCP2 causes an X-linked neurodevelopmental disorder, Rett syndrome (RTT) (4), and is also found in some patients with Angelman syndrome (5, 6). Patients with these diseases have shown severe mental retardation, autistic behavior, and intractable seizures. Interestingly, most of the symptoms appear in early childhood, not at birth. Patients with RTT, MeCP2-affected girls, typically show normal features for 6 to 18 months and then develop impaired motor skills, stereotypic hand movements, abnormal breathing, microcephaly, and other symptoms. Mecp2-null mice show normal development until 4 to 5 weeks of age, and by 6 weeks they present serious neurologic symptoms and die at approximately 8 weeks after birth. These symptoms of both human RTT patients and mecp2-null mouse are thought to be due to a deficiency in the brain, although MeCP2 is ubiquitously expressed. This hypothesis is supported by some studies that show a high expression level of MeCP2 in the brain, specifically in neurons (7-9). To understand the mechanism forming these diseases, we must clarify the role direct MeCP2-downstream genes have. Recently, a few MeCP2-downstream genes have been identified, such as brain-derived neurotrophic factor (BDNF) and DLX5, which locate in neuronal cells and are thought to play an important role in early brain development (10-12). Besides these genes, altered expressions of imprinted UBE3A and GABRB3 genes in mice and RTT patients have been reported (13). However, how these genes contribute to form the RTT phenotype is still unknown.
It has recently been reported that downregulation of insulin-like growth factor binding 3 (IGFBP3) due to methylation in association with MeCP2 leads to proliferation of cancer cells in various types of cancers, such as non-small cell carcinoma of lung, hepatoma, and colorectal cancer, and that overexpression of IGFBP3 inhibits cancer cells proliferation and migration (14-16). IGFBP3 is known to be an antagonist of insulin-like growth factor (IGF) 1 and IGF-2, because its major function is binding to them and carrying them to the target tissues (17). Moreover, IGFBP3-transgenic mice show growth retardation in the brain at the early postnatal period and poor neuronal dendritic expansion (18). We can presuppose that IGFBP3-transgenic mice in the early postnatal period have poor dendritic expansion of neurons, a feature similar to RTT. Taking these facts together, we can speculate that IGFBP3 is necessary for cellular and neuronal maturation and that IGFBP3 is regulated by MeCP2, not only in malignant cells but also in normal neuronal cells. In the present study, we investigated the correlation between MeCP2 and IGFBP3 to clarify whether IGFBP3 may also be an important gene for neuronal development as a new MeCP2-downstream gene. Our findings demonstrate that overexpression of IGFBP3 directly contributes to the neurologic phenotypes associated with MeCP2 deficiency.
Materials and Methods
Mouse and Tissue Preparation
B6.129P2(C)-Mecp2tm1.1Bird/J mice lacking exons 3 and 4 were obtained from The Jackson Laboratory (Bar Harbor, ME) and subsequently bred in our institute. We used 10 or more generated mice in this study. The mecp2-null, hemizygous (mecp2-/y) male mice showed initial symptoms such as hindlimb grasping and hypoactivity at approximately 4 weeks of age and died at 8 or 9 weeks of age (19). Mecp2−/y and wild-type male littermates, as controls, were used at postconceptual days 12.5 (E12.5), and 17.5 (E17.5), as well as postnatal days 0 (P0), 28 (P28), and 60 (P60). We also used mecp2-heterozygous (mecp2+/−) mice, aged P28. Five animals were obtained from each group for histochemistry and four for RNA expression analyses. The mice were deeply anesthetized under diethyl ether and transcardially perfused with heparin-saline and 4% paraformaldehyde in 0.1 mmol/L PBS, pH 7.4, for histologic study. For expression study, unfixed brains were removed and stored at −80°C until use. This study was approved by the Ethical Committee for Animal Experiments in our institute.
Chromatin Immunoprecipitation Assay
A chromatin immunoprecipitation assay with an antibody against MeCP2 was performed as described previously (20). Cell suspensions from mice cortices were fixed in 1% formaldehyde at room temperature for 10 minutes. After being sonicated and lysated, an aliquot of each sample was removed and used in polymerase chain reaction (PCR) analysis as follows (20). The remainder of the soluble chromatin was incubated at 4°C overnight with a rabbit anti-MeCP2 polyclonal antibody (provided by Dr. S. Kudo, Hokkaido Institute of Public Health, Hokkaido, Japan) (21) as "+," and without the antibody as "−." The purified DNA was analyzed by PCR. The primer sets of mouse IGFBP3 promoter were used. A forward sequence was CAGGTGCC CGGTGAAGAC, and a reverse sequence was ATATATA GAAGCCGGGGTGG. These were amplified for 35 cycles of 94°C for 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds. The sequence of mouse IGFBP3 promoter was obtained from GenBank (accession number AL607124).
For human IGFBP3 promoter analysis we used the reported primers and reaction condition (15). The PCR primer set was CGTGAGCACGAGGAGCAGGTG (forward sequence) and CAGGAGTGGGGGTTGGGAG (reverse sequence). PCR conditions were 32 cycles of 94°C for 30 seconds, 62°C for 1 minute, and 72°C for 1 minute.
Reverse Transcriptase-PCR and Real-Time Quantitative PCR
After isolation of 1 μg of total RNA of the cerebral cortices using an RNeasy Mini Kit (QIAGEN, Valencia, CA), we carried out reverse transcription with the First-Strand cDNA Synthesis Kit (Amersham Pharmacia Biotech, Piscataway, NJ) or TaqMan Reverse Transcription Reagents (Applied Biosystems, Foster City, CA). To confirm developmental expression of mouse IGFBP3 mRNA, we performed reverse transcriptase-PCR using a primer set for IGFBP3 (forward sequence of TGAGTCCGAGGAGGAG CACAA and reverse sequence of TACGTCGTCTTT CCCCTTGGT) and for β-actin (forward sequence of TCTGGAAAGCTGTGGCGTGAT and reverse sequence of TTGG AGGCCATGTAGGCCAT) as a reference. One nanogram of each reverse-transcripted cDNA was used under the amplification condition of 35 cycles of 94°C for 30 seconds, 60°C for 30 seconds, and 72°C for 30 seconds.
For quantitative analysis we carried out PCR amplifications with Universal PCR Master Mix (Applied Biosystems) according to the manufacturer's protocol in a real-time ABI PRISM 7700 Instrument (Applied Biosystems). The samples used were wild-type and mecp2−/y male mice aged P28. Primers and probes for mouse MAP2, IGFBP3, and BDNF are available from Applied Biosystems. For each genotype, wild-type, and mecp2−/y male, four pools of RNA were analyzed: four forebrain RNA preparations per pool. Each 30 ng of reverse-transcripted cDNA was applied. The data obtained were compared by real-time PCR analysis, using MAP2 mRNA as an internal control and β-actin mRNA as a reference. Results were displayed relative to β-actin cDNA amounts and statistically compared using the Student t-test.
Immunohistochemistry and Immunoblot Analyses of Mice Brains
We made a series of cryosections of brain tissues at 10-μm thickness, stained with Nissl (Sigma, St. Quentin Fallavier, France) and immunostained with a mouse monoclonal antibody against IGFBP3 (Granzyme-Techne, Minneapolis, MN) according to the manufacturer's instructions. Additionally, we performed immunohistochemistry on a rabbit polyclonal MeCP2 antibody, as previously reported (21), at a dilution of 1:3000. These antibodies were already reported to have the epitope-specific reaction (15, 21).
Protein extraction from prepared tissues was carried out by a previous described procedure (22). After transfer to a nylon membrane, we detected IGFBP3-immunoreacted bands using an enhanced chemiluminescence system (Amersham Pharmacia Biotech), according to the manufacturer's instructions. As a reference, microtubule-associated protein 2 (MAP2) and β-actin were detected by the specific antibodies (Sigma).
Immunohistochemistry and Immunoblot Analyses of Human Brains
The human materials used (6 RTT patients and 4 controls) are summarized in the Table. All RTT patients had a clinically classical course. Four RTT patients had MeCP2 mutations in the transcription repression domain of its exon 4, although MeCP2 mutation analysis was not performed in 2 cases. As controls we used 4 brains without neuropathologic changes from individuals aged 12 to 31 years. Frozen samples of 3 RTT patients were obtained from the Harvard Brain Tissue Resource Center.
With informed consent, we performed immunohistochemistry and immunoblotting of human RTT and non-RTT brains. These 4-μm-thick paraffin-embedded specimens were immunostained with the antibodies for MeCP2 and IGFBP3. For immunoblot analysis, the primary antibodies of MeCP2 and IGFBP3 were used for all frontal cortices of RTT patients and controls.
We used StatView software (SAS Institute Inc., Cary, NC) for to analyze significance between wild-type male and mecp2−/y male, and the Student t-test was performed on the data. We considered p < 0.05 as a significant difference.
MeCP2 Binding to IGFBP3 Promoter
We obtained promoter sequences of mouse and human IGFBP3 (Fig. 1A) from the GenBank of the National Center for Biotechnology Information (accession numbers AL607124 for mouse IGFBP3 and M35878 for human IGFBP3). We found many CG-rich regions within 500 bp from exon 1. A chromatin immunoprecipitation assay showed that MeCP2 bound directly to the promoter of IGFBP3 (Fig. 1B).
IGFBP3 Expression Analyses of Mouse Brains
Reverse transcriptase-PCR showed mRNA of mouse IGFBP3 expressed in the cerebral cortices (Fig. 1C). IGFBP3 mRNA expression was higher in the prenatal period than that in the postnatal period. From the standpoint of genotype, IGFBP3 expression increased more in mecp2−/y male mice than in the wild-type male mice of the same age.
Brain cDNA from four mice at each genotype (4 pools each comprising 3 individual brain cDNA preparations) was compared with real-time PCR analysis, using MAP2 mRNA as an internal control and β-actin mRNA as a reference. As a result, both the IGFBP3 mRNA expression level and the BDNF mRNA expression level of the cortices of the mecp2−/y mice were significantly higher (approximately twice as high) than those of wild-type mice and mecp2+/− mice, whereas there was no significant difference between wild-type mice and mecp2+/− mice (Fig. 1D). Both IGFBP3 and BDNF mRNA were upregulated in P28 mecp2−/y mice, at the presymptomatic stage, compared with wild-type controls and mecp2+/− mice. The MAP2 expression ratio was 0.85 ± 0.12 (average ± SD) of wild-type male, 0.95 ± 0.15 of mecp2+/− female mice, and 0.99 ± 0.17 of mecp2−/y male mice. The IGFBP3 expression ratio was 0.85 ± 0.21 of wild-type male mice, 1.30 ± 0.27 of mecp2+/− female mice, and 1.75 ± 0.12 of mecp2−/y male mice and showed a significant difference (p < 0.001, Student t-test). The BDNF expression ratio was 0.78 ± 0.21 of wild-type male mice, 1.25 ± 0.51 of mecp2+/− female mice, and 1.88 ± 0.13 of mecp2−/y male mice and also showed a significant difference (p < 0.001). This IGFBP3 increase due to MeCP2 deficiency was thought to be at the same level as BDNF, increased expression of which due to lack of MeCP2 was consistent with the previously reported findings (10,11). Interestingly, IGFBP3 expression in mecp2+/− showed an intermediate amount between mecp2−/y and wild-type mice, and there was no significant difference. BDNF expression also had the same pattern as IGFBP3 expression.
Immunoblot analysis confirmed the higher expression of IGFBP3 as protein level in the mecp2−/y brain, especially in the period after P28 (Fig. 2A). IGFBP3 age-dependent expression may be revealed, because of faint bands at P0 and P60 (Fig. 2A). Expression of IGFBP3 protein in mecp2−/y mice increased more than that in wild-type mice. A light molecule of IGFBP3 was detected only in fetuses of both mice, whereas a heavy band was predominantly detected and the light band was faintly seen at postnatal ages. The multiple bands observed in postnatal mice were consistent with those already reported, in which the heavy band is thought to be a glycosylated and mature form (14, 17). Compared with wild-type males, mecp2−/y mice showed significant increases of IGFBP3, especially after P28.
IGFBP3 immunohistochemical analysis revealed an increase in the number of positive cells and fibers in the mutant brains. In wild-type mouse brain, IGFBP3-positive cells distributed in the cortex, and these cells were very faint (Fig. 2B). However, some IGFBP3 immunoreactivity was observed in the cytoplasm and fibers in the cortex of mecp2+/− mice (Fig. 2C). Mecp2−/y mice showed more IGFBP3-positive cells and fibers than wild-type male and mecp2+/− mice (Fig. 2B-D). MeCP2 was recognized in many nuclei in the cortex of wild-type male and mecp2+/− mice (Fig. 2E, F); however, mecp2−/y mice had no MeCP2 immunoreactivity (Fig. 2G). In the brain of E17.5 mice, IGFBP3 positivity was observed mainly in the cortical plate, hippocampus, and caudoputamen (Fig. 2H). In the cortical plate, IGFBP3 immunostaining located in the cytoplasm and/or intercellular space. In the P0 brain, IGFBP3-positive cells and fibers were limited to the cortex. The IGFBP3 immunostaining decreased with age. The mature brain of P60 mice evidenced no IGFBP3 immunoreactivity, except for striatum and fibers in the white matter (data not shown). In the white matter, IGFBP3 immunoreactivity was shown in some fibers and a small number of glial cells (data not shown).
In the brains of P28 mice, whose IGFBP3 protein concentration was the highest of the three ages examined, IGFBP3 immunohistochemical features in mecp2−/y were recognized to increase those reactivities, as determined by Western blot analysis (Fig. 2A).
Immunoexpression of IGFBP3 in Human Brains With RTT and Controls
We found similar immunostaining patterns in human brains. Immunoblot data of 5 RTT patients showed a higher level of IGFBP3 expression than that of controls, indicating that MeCP2-positive neurons express less IGFBP3 (Fig. 3A). Cortical neurons of the controls revealed no IGFBP3 immunoreactivity (Fig. 3B). However, a few IGFBP3-positive neurons and fibers were observed in a heterozygous female RTT patient (Fig. 3C). These results showed a pattern similar to that of mice brains. MeCP2-positive cells sometimes expressed low or undetectable levels of IGFBP3 (Fig. 3D, E). However, these data suggest that the presence of MeCP2 in a neuronal cell is negatively compatible with IGFBP3 expression. It is possible that the role of MeCP2 at the promoter is not only to silence the genes but also to modulate levels of expression.
Genetic Relationship Between IGFBP3 and MeCP2
We found that the promoter of IGFBP3 had a MeCP2 binding site and that IGFBP3 expression in mouse brains was regulated by MeCP2. Structural analysis of the human IGFBP3 promoter revealed a dense cluster of CpG islands spanning the region from −220 to −20 bp (14). The mouse IGFBP3 promoter showed CpG islands from −300 to −120 bp. These CG-rich sites of the IGFBP3 promoters are thought to regulate its expression. They were earlier reported to be hypermethylation in some cancer cells (15). Chromatin immunoprecipitation assays showed that the IGFBP3 promoter bound to MeCP2 in the brain tissue. MeCP2 is strongly suggested to contribute to the transcriptional expression of IGFBP3. These facts suggest that MeCP2 may be a regulator of IGFBP expression and that MeCP2 mutations may lead to IGFBP3 overexpression, as previously discussed in BDNF (10, 11).
Functions of IGFBP3 and Relationship to Phenotypes
It is known that IGFBP3 binds to IGF-1 and IGF-2, major factors for cell growth and elongation of neuronal dendrites and axons and inhibits these functions (17, 23). IGF-1 and IGF-2 are present in most biologic fluids as a complex with IGFBPs. Of the 6 IGFBPs, IGFBP3 is the most abundant in cerebrospinal fluid (23). A role for IGFBP3 in the transport and modulation of the biologic actions of the IGF has been proposed (17). Moreover, IGFBP3 appears to inhibit cell proliferation and differentiation in some cell lines (24, 25).
Overexpression of IGFBP3 shows an elevation of IGF-1 concentration and is also associated with intrauterine and postnatal growth retardation (26). These effects on growth are most likely explicable on the basis of IGF action inhibition. Although its exact physiologic function remains unclear, IGFBP3 overexpression may lead to inhibition of neuronal development.
Phenotypic Similarity of IGFBP3-Transgenic Mice and MeCP2-Null Mice
IGFBP3-transgenic mice, having 4.9 and 7.7 times the expression of wild-type mice, show retardation of early postnatal brain growth (18). MeCP2-null mice, mecp2−/y mice, showed IGFBP3 overexpression similar to that of the transgenic mice in this study, and had shown features similar to those of IGFBP3-transgenic mice (e.g. brain size reduction and thin cortex) in a previous study (18, 27). Mecp2−/y mice had delayed synaptogenesis and diminished brains (27, 28). These findings allow us to speculate that overexpression of IGFBP3 due to lack of MeCP2 leads to delayed brain maturation and plays a role in these conditions in humans as well. Interestingly, MeCP2 and IGFBP3 in mouse brain showed the highest expression at P28 before the occurrence of the initial symptoms in mecp2−/y mice, such as spastic paraplegia and hypoactivity, so these features may be associated with IGFBP3 overexpression in mice brains.
IGFBP3 Expression of MeCP2 Heterozygous Mutated Mice and RTT Patients
Genetically, mecp2+/− mice are the same genotype as human RTT. Both have heterozygous mutation of MeCP2. IGFBP3 and BDNF expression in mecp2+/− showed an intermediate amount between mecp2−/y and wild-type mice from our real-time quantitative PCR. The 1.5-fold increase of IGFBP3 may also act as one of the pathogenetic factors, whereas there were no significant differences, or this pathologic discrepancy in heterozygous mutation may be due to the difference in species. From the standpoint of IGFBP3 expression, mecp2−/y mice phenotypes may be compatible to those of human RTT. IGFBP3, which was identified as a novel MeCP2-downstream gene in the present study, exhibited an approximately 2-fold elevation of its mRNA expression in the mecp2-null brain, the same level as that of BDNF.
MeCP2 as Transcriptional Repressor
It is noteworthy that the absence of mecp2 leads to elevated levels of IGFBP3 transcripts before the onset of overt symptoms in these mice. The effect of overexpression of IGFBP3 is an early phenotype that manifests during postnatal brain development, possibly including dendritic changes in both rodents and humans. Related changes occur in brains of RTT patients (28, 29). Furthermore, seizures and heightened anxiety of mice with mutations in the mecp2 gene (30) and the glucocorticoid-related gene products, which were growth-regulated proteins, overexpressed in RTT model mice (31).
Because MeCP2 is a transcriptional repressor, the predominant current hypothesis to explain RTT is that critical genes are aberrantly expressed in its absence. Misregulation of several genes is conceivable. Aberrant expression of the MeCP2-target genes, BDNF, Dlx5, Ube3a, Gabrb3, Igfbp3, or others, might be a contributor to the phenotype. MeCP2 misregulation of the expression of these genes expression may have neurologic consequences that could give rise to disease.
It is important to clarify the gene expression changes in molecular pathways of MeCP2-downstream systems in mice and human brains with the phenotypes. Such genes, which are linked with neuronal function at critical developmental stages and play important roles in specific neuronal development, are thought to contribute to the characteristic and pathognomonic neuronal involvement in RTT.
We thank Dr. K. Suijo, Tokyo Gakugei University, for his generous assistance; Mr. S. Kumagai, NCNP, for technical assistance; and Dr. K. Endoh, University of Yamanashi, for helpful advice on genetic information. Thanks are also due to Dr. K. Ishi, a pathologist of Juntendo University Urayasu Hospital, for performing the autopsy of a Japanese case; and the Harvard Brain Tissue Resource Center for providing RTT samples.