15 July 2008. Last May, the New York Academy of Sciences (NYAS), together with the American Chemical Society's New York Section, hosted a one-day symposium entitled “DISC1 and the Neurodevelopmental Hypothesis of Schizophrenia.” The meeting, sponsored by the NYAS Biochemical Pharmacology Discussion Group, was hosted by Julia Heinrich, formerly with Wyeth Discovery Neuroscience and Robin Kleiman from Pfizer, Inc. The organizers invited four experts to present their ideas on how disrupted in schizophrenia (DISC1) and its partners relate to the etiology of the disease and to discuss potential therapeutic approaches that might be suggested by the latest data. The full symposium is currently available via webcast for NYAS members and SRF readers (Contact us for details). Here we present some of the major points of the presentations.
Heinrich started the meeting with an overview of schizophrenia, including a brief history of the research linking the disease with DISC1. There are many hypotheses on the causes of schizophrenia, including those that point to neurochemical imbalances, epigenetic factors, and neurodevelopmental problems. One of the major questions facing the field, suggested Heinrich, is how early the latter might occur ahead of the overt phenotype. There are indications, such as obstetric complications, reduced brain volume in infants, and childhood cognitive deficits, which all occur early, that schizophrenia could be grounded in early developmental events. The more recent identification of candidate susceptibility genes, including DISC1, that play a role in development, strengthens that argument.
Continuing on that theme, Akira Sawa, from Johns Hopkins University, Baltimore, Maryland, noted that there is a distinction being made in the field between pathogenesis and pathophysiology of schizophrenia (see review by Lewis et al., 2005). While pathophysiology may be concurrent with the phenotype, pathogenesis may occur much earlier. Susceptibility genes have, and will be, helpful in studying the pathogenesis, Sawa said, particularly the genes neuregulin-1, dysbindin, and DISC1.
DISC1 was discovered when University of Edinburgh researchers identified a chromosomal translocation in a large Scottish family with mental disorders (see St. Clair et al., 1990). That translocation was subsequently shown to break the DISC1 gene (see Millar et al., 2000). Researchers later showed that DISC1 associates with schizophrenia in other populations as well (see Hodgkinson et al., 2004 and Hamshere et al., 2005). Work from Tyrone Canon’s lab at UCLA then showed that DISC1 is important for gray matter morphology (see Cannon et al., 2005), suggesting variations in the gene may have an impact on brain formation or brain malformation. Together, these papers represent seminal contributions to the study of DISC1 in schizophrenia, suggested Sawa.
Sawa reviewed what is known about the biology of DISC1. The protein exists in many different isoforms, maybe even as many as 40-50, said Sawa, and it has several subcellular locations and multiple, context-dependent functions. It is found in the post-synaptic density, for example, where it binds to post-synaptic markers such as PSD95; it is also found at the centrosome, the organizing center for microtubules; and it associates with the nucleus and mitochondria. From the schizophrenia perspective, the centrosome and post-synaptic density locations may be the most important, said Sawa. He suggested that it is also possible that the gene is duplicated elsewhere in the genome. If true, that will make DISC1 knockout models even more difficult to achieve, said Sawa.
Currently there are five mouse models of abnormal DISC1, but none is a complete knockout because it is very difficult to deplete all isoforms of the protein, said Sawa. For this reason, knockdown models using RNAi have been useful, showing that the protein is important for corticogenesis, for example. But its role appears context-dependent. In the cortex, suppressing DISC1 (at least some isoforms) delays cell migration, whereas in the adult dentate gyrus it accelerates it, suggesting DISC1 plays a role in adult neurogenesis (see SRF related news story).
DISC1 is found on centrosomes in developing neurons and the post-synaptic density in mature neurons, suggesting that DISC1 variation could increase the risk for schizophrenia in two ways early during development, and later by interfering with neuronal circuitry. Sawa focused mostly on the centrosomal role.
The centrosome is the organizing center for the microtubules, which are essential for cell division, polarization, migration, and neuronal differentiation. The microtubules are critical for neurodevelopment. Sawa’s group has found that DISC1 is part of a large dynein protein complex at the centrosome, and that without DISC1, microtubule formation is compromised (see Kamiya et al., 2005). DISC1 appears to be necessary to maintain the dynein complex at the centrosome.
Genetic variations in other centrosomal proteins have also been linked to neuropsychiatric disorders. Pericentriolar material 1 (PCM1), for example, has been linked to schizophrenia (see SRF related news story), which strengthens the idea that centrosomal actions of DISC1 may be germane to the disorder, and a disease-causing protein encoded by the (BBS4) gene, linked to Bardet-Biedl syndrome, a disease with neuropsychiatric symptoms binds to PCM1 (see Kim et al., 2004). Now, in a paper in press in the Archives of General Psychiatry, Sawa and colleagues will show link of DISC1, PCM1, and BBS proteins in the pathology of schizophrenia via both genetic and biological approaches. Some of this work was already presented at the 2006 Society for Neuroscience meeting in Atlanta, Georgia (see SRF related news story). DISC1 also binds to other centrosomal proteins, such as NDE1 and NDEL1 (also known as NUDE and NUDEL, respectively) (see SRF related news story). In collaboration with Anil Malhotra’s group at Zucker Hillside Hospital, Glen Oaks, New York, Sawa and colleagues recently showed, using genetic and biological evidence, that DISC1, NDE1, and NDEL1 are linked with each other in the pathology of schizophrenia (see Burdick et al., 2008).
Nick Brandon outlined some of the ongoing DISC1 work at Wyeth Neuroscience Research, Princeton, New Jersey. Industry is interested in DISC1 because it may be useful for identifying “druggable” target molecules and because it might provide some useful animal models, suggested Brandon. On the former, Brandon and colleagues have carried out extensive yeast two-hybrid screens and mass-spec analysis of DISC1 complex to identify DISC1 partners, finding over 400 different protein molecules (see Camargo et al., 2007). From this work several protein networks that may be pertinent to schizophrenia were found, including those containing the cysteine protease NDEL1 and the cyclic AMP degrading enzyme phosphodiesterase 4 (PDE4). Analysis of the interactome also showed that it was enriched in proteins that are involved in synaptic activity and glutamate receptor function. The proteins are involved in almost every aspect of synapse biology from formation to stabilization. These networks provide a template for others in the field to build from, suggested Brandon.
Brandon said that all the indications from this work are that DISC1 is a synaptic protein. What its role is at the synapse, and how it is regulated, are key questions. Brandon and colleagues have begun to address these, and he treated the audience to some snippets of work in progress. Using highly specific antibodies that recognize different exons of mouse Disc1, the researchers have mapped expression of the protein in primary hippocampal neurons. The protein appears to be perinuclear early on, but after 14 days in vitro Disc1 is expressed widely in the neuron, including at dendritic sites, where it colocalizes with PSD95, a synaptic marker. Biochemical analysis also strongly suggests a synaptic localization for Disc1—the protein is enriched in synaptosomes and isolated PSD fractions. Brandon and colleagues are now working on identifying exactly which isoforms are present in these fractions. Preliminary analysis suggests that the isoform distribution may be quite complex, and more antibodies are needed to get a better picture of the synaptic isoforms, said Brandon. Figuring out which isoforms are present and identifying their binding partners are crucial steps to deciphering the synaptic role of DISC1.
One well-known DISC1 binding partner is NDEL1, which also binds LIS1, a protein linked to lissencephaly (see SRF related news story). As Sawa mentioned, both DISC1 and NDEL1 are found at the centrosome, as is LIS1. Brandon discussed some of the properties of NDEL1, which is still not well understood. The protein regulates the assembly of neurofilaments (Nguyen et al., 2004) and together with another protein, vimentin, is crucial for neurite outgrowth (Shim et al., 2008). It also plays important roles in cell migration (Shen et al., 2008) and neurogenesis (Shu et al., 2004). Some of these functions are probably dependent on DISC1, since it acts as a cargo receptor for NDEL1 complexes in axons (see SRF related news story). Brandon showed that DISC1 and NDEL1 are co-expressed in the hippocampus, but not necessarily in other regions of the brain—for example, the hypothalamus—indicating the interaction between the two proteins is context dependent. His group found that NDEL1 also binds to a distinct domain on DISC1 at the C-terminus (see Brandon et al., 2004) via a Ndel1 domain that includes amino acids leucine 266 and glutamate 267—when those residues are mutated, the binding to DISC1 is lost. Those amino acids are also required for neurite outgrowth, suggesting that DISC1 plays a role in that NDEL1 regulated process.
Interestingly, NDEL1 was independently identified as an endoligopeptidase that cleaves small neuropeptides, and cysteine 273 is essential for this activity. Given it binds NDEL1 very close to this cysteine residue, could DISC1 somehow regulate NDEL1 activity? Assays with purified proteins show that DISC1 is, in fact, a competitive inhibitor of NDEL1 peptidase activity (see Hayashi et al., 2005), but it was not clear if this had any physiological significance, said Brandon. To address this, in collaboration with a research group at the University of Sao Paolo in Brazil led by Mirian Hayashi, Sawa’s group developed a cell-based test of NDEL1 activity. In a paper that is in preparation, they show that NDEL1 enzymatic activity increases as PC12 cells grow in culture and that loss of cysteine 273 reduces neurite outgrowth. The groups next plan to study the relationship between peptidase activity and neurite growth in hippocampal cells. They also plan to identify which substrates are key to the function of Ndel1 in neurite outgrowth.
How do DISC1 variations relate to the pathophysiology and phenotypes seen in schizophrenia? Katherine Burdick, North Shore Long Island Jewish Health System, reviewed some of the clinical research on DISC1, and it fits with known schizophrenia endophenotypes and DISC1 molecular interactions.
Burdick reminded the audience that the impact of DISC1 variants, and in fact most genetic risk for schizophrenia, is likely to be small and transmitted in a polygenic, non-Mendelian fashion. Genetic variability alone cannot account for the disease, and there are numerous environmental influences that have been linked to schizophrenia, many associated with the critical peri- and postnatal period of development. Together, the genetic and epidemiological data suggest that schizophrenia may have a neurodevelopmental etiology.
Genetic variations in DISC1 have been linked to various psychiatric disorders, including bipolar and schizoaffective disorders, major depression, and more recently, autism. These linkages suggest that DISC1 may impact some traits that overlap among all these disorders. Increased risk for some positive symptoms of schizophrenia, such as delusions and hallucinations, have been linked to specific single nucleotide polymorphisms (SNPs) and haplotypes of the DISC1 gene (Hennah et al., 2003), and also with various cognitive domains, such as verbal and spatial working memory, subserved by hippocampal and prefrontal cortex function (see Burdick et al., 2005), but the data are quite complex, said Burdick. Different markers within the gene have been associated with cognitive deficits in different studies. This may reflect methodological differences among studies, but it could also be indicative that there are different loci within the gene that impact cognition, she said.
How do these functional affects of DISC1 variants relate to its structural effects on the brain? Burdick reviewed some of the evidence linking DISC1 to brain abnormalities. Phil Szeszko, her colleague in Anil Malhotra's group, has found that one polymorphism, a leucine to phenylalanine mutation at position 607, is linked to reduced gray matter in the anterior cingulate and superior frontal gyrus (Szeszko et al., 2007), suggesting decreased gray matter in prefrontal cortex, one area of the brain that has been consistently associated with the disease. Other groups have also found reduced hippocampal volume in leu607phe carriers and also learning and memory effects (see Cannon et al., 2005). Another single nucleotide polymorphism (SNP), at position 704 (serine/cysteine), has also been linked to reduced hippocampal volume (ser704 homozygotes) and abnormal hippocampal activation in working memory and episodic memory tests. These findings tie together cognitive effects of DISC1 variation with structural consequences, suggested Burdick.
Since schizophrenia is likely polygenic, how do genetic variations in DISC1 relate to other genetic variations, particularly in genes that interact with DISC1, such as NDEL1? Burdick and colleagues found that there are four SNPs in the NDEL1 gene that form a block that is in linkage disequilibrium with schizophrenia. Two common haplotypes form this block. The less common is associated with the disease and increases risk by about 1.3-fold. A tagging SNP in the same location also increases risk by 1.5-fold, but when viewed in the context of DISC1 genetic variation, a slightly different picture emerges. The tagging SNP risk allele (G) only confers risk (2.5-fold) in DISC1 ser704 homozygotes, suggesting there is epistasis at work. Interestingly, the DISC1 serine at position 704 is in close proximity to the Ndel1 binding site.
Other groups have found that there are similar genetic interactions between DISC1 and the NDEL1 homolog NDE1 (see SRF related news story). Burdick and colleagues have looked at NDEL1-DISC1 genetic interactions and found that they are opposite to that for NDEL1 and DISC1. While the serine at DISC1 704 is required for the NDEL1 variant to have any effect, cysteine is needed at position 704 in the case of NDEL. This is consistent with biochemical data showing that NDEL1 binds to the cys704 DISC1 preferentially, while NDE1 binds best to the ser704 DISC1 (see also Burdick et al., 2008).
How do these observations tie in with the neurodevelopmental hypothesis of schizophrenia? Burdick suggested that given DISC1’s multiple roles in embryonic development and near the age of puberty, and its role in adult neurogenesis (see SRF related news story), it is poised to be a central player in pathogenesis. “Perturbation of NDEL1/NDE1 balance might result in abnormal binding of these proteins to DISC1, disrupting the coordinated functions of the DISC1-associated protein complex to interfere with normal neuron growth and migration,” said Burdick.
The structural and functional consequences of DISC1 variation in humans is also consistent with changes seen in mice with mutations in the gene. Steven Clapcote, University of Edinburgh, Scotland, reviewed some of the data that have emerged for mouse models of schizophrenia based on DISC1 mutagenesis. In collaboration with colleagues at RIKEN Genomic Science Center, Yokohama, Japan, Clapcote and colleagues have made DISC1 mutant mice using a random mutagenesis approach. This has resulted in mice strains with five different mutations in the Disc1 gene, three in the largest exon (No. 2) and two in exon 11. Clapcote said that one of the exon 11 mutations (H738N) might interfere with Disc1 binding to Grb2, an adaptor molecule that links receptor tyrosine kinases in the ERK signal transduction pathway (see Shinoda et al., 2007), but this mutation has not been fully characterized yet. Two of the mutations in exon 2, Q31L and L100P, have been characterized, and Clapcote reviewed some of the properties of the mice.
SRF has substantially covered this work (see SRF related news story). Briefly, the two mutations have somewhat different effects in mice. Mice with the L100P mutation have slightly reduced brain volume—especially in the cortex and cerebellum. The latter was a bit surprising, said Clapcote. His group is currently following that up to get a better understanding of the relationship between Disc1 and brain size.
Behaviorally, both mutations have significant, though not always similar, effects. Both mutant lines showed reduced prepulse inhibition, latent inhibition, and impaired spatial working memory. But in an open field test (a model of psychomotor agitation—one of the positive symptoms of schizophrenia), only the L100P mice are hyperactive. This hyperactivity was exacerbated by amphetamines—psychostimulants to which schizophrenia patients are particularly sensitive. The Q31L mice, on the other hand, showed both reduced social interaction and reduced interest in sucrose solutions—a sign of anhedonia. In a forced swim test, the Q31L animals also give up and float more readily—a sign of depression.
The two strains also differ in their response to antidepressants. Rolipram, a phosphodiesterase 4B (PDE4B) inhibitor, rescued prepulse inhibition in the L100P animals but not in the Q31L mice. In the latter strain, rolipram also had no effect on depression as judged by performance in the forced swim test, but the antidepressant bupropion did work. The differential effects of the drugs may be related to Disc1 binding to Pde4b The two must dissociate for the phosphodiesterase to become active, and Clapcote noted that the Q31L animals had about half as much Pde4b activity as wild-type, which might explain their resistance to rolipram. “If PDE activity is already low, the dose may not be sufficient to reduce it any further,” he suggested. But interestingly, despite lower Pde4b activity, the Q31L mice are more depressed—not less. Though more work is needed to sort out the relationships between these mutations, their different phenotypes, and Disc1 function, their differential response to drugs might be a pharmacogenomic model for the different types of responses reported in patients with different genotypes, suggested Clapcote.—Tom Fagan.