28 October 2005. A protein that underlies a severe and fairly common neurodevelopmental disease called Rett syndrome has a more sophisticated function than scientists previously suspected. Called MeCP2, the protein was thought to silence gene transcription in a process involving DNA methylation and chromatin remodeling. This process, in turn, serves as an example mechanism for the broader issue of epigenetic control of gene expression. As if this was not interesting enough, however, researchers led by Huda Zoghbi at Baylor College of Medicine in Houston report in the 26 October PNAS Early Edition that MeCP2 also regulates the RNA splicing of certain genes. The two roles of MeCP2 might even be coupled, the authors speculate. When MeCP2 comes off the promoter of a target gene to permit its expression, perhaps it promptly turns around to the pre-mRNA that can now peel off the DNA and helps control which exons go into the final mRNA, thereby influencing the sequence of the translated protein, the scientists suggest.
This study is not directly relevant to Alzheimer disease, but all the same, it is worth reading because it exemplifies a new research direction as scientists try to make deeper inroads into the complexities of the molecular pathophysiology of neurologic and psychiatric diseases. Mutations in the gene for MeCP2 have come up in the study of other neurodevelopmental diseases including autism. More broadly, MeCP2 is a neural protein known to become abundant during childhood when synapses mature, suggesting that it is important for activity-dependent plasticity.
At the basic level, interest in how proteins interact with RNA to orchestrate the splicing of pre-mRNA is being fanned by the humbling realization coming out of the Human Genome Project that humans contain fewer than 30,000 genes, probably no more than simpler creatures such as the puffer fish. The human proteome is much larger than its genome, however, implying that an expansion of variety must occur at the level of post-transcriptional processes. Part of that variety arises from regulated changes in splicing, which generate different RNA isoforms from the same pre-mRNA. By orchestrating which exons go into the final protein, alternative splicing can multiply the number of proteins that can be manufactured from a given gene. Many labs now study the proteins that oversee the cutting and pasting necessary for that. Some focus on how the splicing and transcription machineries are coupled, others focus on how alternative splicing controls formation of the synapse (Ule et al., 2005), and yet others explore how alternative splicing gives rise to tissue-specific isoforms of a given protein. One cutting edge in this area that is relevant to ARF readers concerns the question of how neuronal activity, signaling, and other aspects of the molecular context of a neuron are linked to RNA splicing (e.g., Beffert et al., 2005).
The theme of RNA processing cuts across emerging pathophysiologies of a number of neurologic and psychiatric diseases. Examples include Alzheimer disease, spinomuscular atrophy, myotonic dystrophy, schizophrenia, and fragile X mental retardation. While RNA splicing abnormalities are not root causes of these diseases, splicing shows up in various guises in all of them.
In AD research, for example, it has long been known that certain splice variants of tau are important in the development of tau pathology (D’Souza and Schellenberg, 2005). Earlier this year, researchers reported that a new candidate gene for late-onset AD might exert its influence on risk through alternative splicing (Bertram et al., 2005). Others suggested that ineffective splice variants of the human IDE gene (Farris et al., 2005) and more active splice variants of BACE might play a role (Zohar et al., 2005).
RNAs for most channels and receptor proteins are alternatively spliced. That includes metabotropic glutamate receptors, proteins that are implicated in neurologic and psychiatric diseases including mental retardation and schizophrenia (Weinberger, 2005; Niswender et al., 2005). With regard to schizophrenia, alternative splicing has been suggested for two of the susceptibility genes that draw increasing support in the field, i.e., neuregulin and ZDHHC8 (Kirov et al., 2005).
In the present study, Zoghbi's team focused on the protein underlying Rett syndrome, which affects one in every 10,000 girls and a smaller percentage of boys. While the phenotype varies, babies generally develop normally for a year or so and then regress, growing up with mental retardation, abnormal movements including a characteristic hand wringing, seizures, and without being able to speak or socialize. A break in the understanding of this baffling disease came when Zoghbi’s lab identified causal mutations in the MeCP2 gene located on the X chromosome (Amir et al., 1999). Such mutations show up on the autism spectrum, too, but favorable X chromosome inactivation patterns can lead to a milder phenotype in these cases.
Researchers immediately tried to find out exactly how this protein functions. Early studies showed that MeCP2, which stands for methyl-CpG-binding protein 2, not only binds to methylated cytosines in DNA, but also associates with a repressor complex containing histone deacetylases. This pointed toward a global role in silencing transcription, but things soon became complicated. When scientists compared transcriptional profiles of MeCP2 knockout mice with those from wild-type, they saw no clear-cut gene expression changes even though the knockout mice did have the Rett phenotype (Tudor et al., 2002). Moreover, researchers for years had difficulty identifying MeCP2 target genes. When several labs recently did manage to find some, the picture only became murkier because each seemed to be regulated by MeCP2 in a different way, not by a common mechanism of transcriptional repression. For example, BDNF turned out to be an activity-dependent target of MeCP2 that was repressed in the classic way of MeCP2 occupying its promoter (Chen et al., 2003; Martinovich et al., 2003). By contrast, repression of the target gene Dlx5 had to do with genetic imprinting and required formation of a silent chromatin loop (Horike et al., 2005). This and other data led Zoghbi to believe that no single known mechanism for MeCP2 could explain Rett pathogenesis to date and that a new, unbiased functional analysis was needed.
Toward this goal, first author Juan Young and colleagues searched for proteins that interact with MeCP2. Immunoprecipitation and mass spectrometry studies identified Y box binding protein 1 (YB-1), an evolutionarily conserved DNA and RNA binding protein that had been previously implicated in alternative splicing, regulation or transcription and translation, DNA repair, and other cellular functions. RNA was necessary to maintain the interaction between MeCP2 and YB-1. Prior work on YB-1 led them to suspect that the MeCP2-YB-1 complex serves to coordinate splicing with gene transcription by pulling YB-1 to nascent transcripts after MeCP2 is released from a gene promoter, the authors write.
The scientists first confirmed this idea by measuring the splicing of a reporter minigene in transfected cells. Next, the scientists picked a candidate neuronal gene and tested whether its splicing depends on the Rett protein. They chose the NMDA receptor subunit NR1 because of the previously reported link between neuronal activity and MeCP2 targets. An alternative splice site in exon 22 of this gene’s mRNA is known to generate different protein variants in response to activity (Mu et al., 2003). Comparing brain tissue from Rett knockout and from wild-type mice, the scientists found differences in the distribution of the splice variants in subcortical areas but not in cerebral cortex, hinting that the NR1 pre-mRNA might be a target for MeCP2 only in certain brain areas.
Finally, the researchers took a more global look at splicing changes. Using a custom-made microarray carrying probes specific to individual exons and exon-exon junctions, they performed a genomewide survey of splicing changes in cerebral cortex mRNA from a Rett MeCP2 mouse model and wild-type mice. They report changes in alternative splicing of 54 genes, most of them classic cassette exon changes. When clustered, the changes classified the genotype of the samples. (This data is published as supplemental material on the PNAS website.) Validation of the candidate transcripts identified with the array showed that 35 percent of them were spliced abnormally in cerebral cortex of other MeCP2 mutant mice. One of them was Dlx5, a gene previously identified as a MeCP2 target and known to have at least seven splice forms.
In summary, the study paints a more intricate portrait of MeCP2 as a protein with multiple functions in neurons. It raises the question whether Rett syndrome is as much a disorder of RNA splicing as of DNA expression. Given that different proteins can roll off a defective splicing machinery, studying errors in this process might lead scientists toward a better understanding of why doctors see such heterogeneous symptoms in patients who share a given genetic defect, in this case, mutations in the gene MeCP2.—Gabrielle Strobel.
Young JI, Hong EP, Castle JC, Crespo-Barreto J, Bowman AB, Rose MF, Kang D, Richman R, Johnson JM, Berget S, Zoghbi HY. Regulation of RNA splicing by the methylation-dependent transcriptional repressor methyl-CpG binding protein 2. Proc Natl Acad Sci U S A. 2005 Oct 26; [Epub ahead of print]. Abstract