DISC1-binding proteins in neural development, signalling and schizophrenia

1A key objective of genetics and genomics research into psychiatric illness is to identify perturbed biological pathways and, as a consequence, potential targets for pharmacological intervention. The genetic entrée point need not itself explain a large fraction of the liability to schizophrenia – it is sufficient that genetic abrogation can cause schizophrenia. DISC1 and the extended DISC1 pathway illustrate this contention par excellence. DISC1 was identified through a unique family in which a chromosomal translocation event co-segregates strongly with major mental illness. This translocation event directly disrupts both a protein coding gene, DISC1, and an antisense RNA only gene, DISC2. In the intervening period, the DISC locus has been repeatedly implicated in psychiatric illness by genetic linkage, association and mutation detection (reviewed in, see Table 1 for references and summaries of recent studies). Some studies have also pointed to epistatic interactions between the DISC locus and other candidate genes. Despite these many confirmatory studies, there are also negative studies ( and references in Table 1) and, as yet, no firm basis on which to estimate the proportion of genetic liability attributable to the DISC locus. The DISC locus has appeared as a gene-wide, but not a genome-wide finding in some but not other studies. The critical issue is what we can learn from the identification of DISC1 regarding the specifics and generalities of the biological underpinning of schizophrenia and other major mental illness.

2The DISC1 protein has no known enzymatic activity; rather it exerts its effect on multiple proteins through interaction to modulate their functional states and biological activities in time and space. Many putative interacting proteins have been identified through extensive yeast-2-hybrid screening and, where these have been examined, a large proportion have been validated by downstream experimentation (reviewed in ). These multiple interactions, combined with the widespread subcellular distribution of DISC1 (reviewed in ), a complex pattern of protein isoforms and splice variants, have led to the suggestion that DISC1 acts as a protein scaffold within the cell, dynamically interacting with and affecting the function of different proteins at different locations and developmental times. DISC1-related psychiatric illness is therefore likely to arise through the simultaneous dysregulation of not just one, but more likely several, protein interactions, physiological states and activities, with a consequential complexity and pervasiveness of effect. Identifying the key DISC1 interactors is therefore of exceptional importance in coming to understand the nature of the devastating condition that is schizophrenia, and will facilitate the search for downstream elements which may be susceptible to pharmaceutical intervention.

3In this review, we will focus on what is known about the biological functions of DISC1-interacting proteins, with particular attention to aspects of their biology which potentially relate to psychiatric illness, through effects on neurodevelopment, neurotransmission or neurosignalling, along with the emerging genetic evidence implicating many of these as schizophrenia-risk factors in their own right. Whereas a neurodevelopmental component to schizophrenia is well established, and pre-morbid features are recognised, it is typically not until adolescence or early adulthood that the debilitating symptoms emerge. That DISC1 and its pathway of interacting proteins affect both neurodevelopmental pathways and also signalling pathways in the adult brain, suggest that the study of DISC1 genetics and biology may help towards a more unified understanding of schizophrenia and with it the potential to develop rational interventions in the symptomatic adult and or even earlier.

4The complex and intricate task of co-ordinating the microtubule network of neurons, which is vital for maintaining correct development, morphology and migration, is performed in large part by the microtubule organising centre, or centrosome. Multiple lines of evidence demonstrate that DISC1 is part of a protein complex at the centrosome (Fig. 1 and references in the legend) and is involved in cytoskeletal processes involved in neuronal migration, including nucleokinesis and neurite outgrowth. LIS1, NDE1, and NDEL1 are a trio of such centrosomal DISC1 interactors which play pivotal roles in the progression of the cell cycle, dynein-related transport along microtubules and nucleokinesis (reviewed in ). The localisation of NDEL1 and LIS1 in axons is known to be dependent on expression of DISC1. Knock-down of either NDEL1 or LIS1 using RNAi in culture leads to reduced neurite outgrowth, while granule neurons from heterozygous NDEL1 or LIS1 knock-out mice show impaired migration in in vitro assays. NDEL1 is also known to play a role in axon regeneration after injury and has an additional DISC1-modulated function as a cysteine endopeptidase which appears to be important for regulation of neurite outgrowth. Although highly similar in amino-acid sequence to NDEL1, the role of NDE1 in many of these processes has yet to be determined.

5DISC1 is also known to be involved in transport along microtubules to the distal parts of axons as part of a ternary complex with kinesin-1 and the adaptor protein Grb2. Cargo transported in this manner includes the DISC1 interactors LIS1, NDEL1 and 14-3-3ɛ. The interaction of DISC1 with Grb2 may also be required for neutrophin-3-related axon elongation. Of potential therapeutic relevance, expression of GRB2 mRNA is known to be upregulated following electroconvulsive seizure, an established anti-depression therapy. FEZ1, another DISC1-interacting protein, is known to be involved in the activation of the kinesin-1 motor protein, regulation of neurite outgrowth and the establishment of neuronal polarity. DISC1-FEZ1 interaction is enhanced during neurodifferentiation, and expression of the FEZ1-binding domain of DISC1 has a dominant negative effect on neurite outgrowth in a cellular model, implying co-operation of DISC1 and FEZ1 in this signalling pathway.

6At the centrosome, DISC1 also interacts with the scaffold protein pericentrin (also known as kendrin, ), a molecule that is known to play important roles in microtubule nucleation and aster formation (reviewed in ) in a seemingly DISC1-dependent manner. Intriguingly, mutations in the PCNT gene, which encodes pericentrin, are heavily implicated in a form of dwarfism associated with reduced brain size, suggesting it to be important for neurodevelopment. Recruitment of pericentrin to the centrosome is essential for correct microtubule organisation and is facilitated by another large scaffold protein PCM1. PCM1 in turn is recruited co-operatively by interacting proteins DISC1 and BBS4. Centrosomal PCM1 is known to be required for correct axon morphology and embryonic neurogenesis. Intriguingly, the level of localisation of PCM1 to the centrosome in human glial cells is altered by two common DISC1 amino-acid substitutions, Ser704Cys and Leu607Phe, one possible mechanism by which these alleles lead to DISC1 dysfunction, as measured by brain imaging and elevated risk of psychiatric illness (Table 1 and references therein).

7Other centrosomal interactors of DISC1 include the DISC1-Binding Zinc finger protein (DBZ, also known as Su48 or ZNF365). DBZ is a brain expressed protein which binds to DISC1, NDE1 and NDEL1. Co-expression of DISC1 and DBZ results in a reduction in the number of PC12 cells bearing neurites, while expression of the DISC1-binding domain of DBZ lead to reduced neurite outgrowth in mouse primary hippocampal neurons.

8Results from cell-based models such as those described above beg the question as to how binding partners of DISC1 might be involved in regulating neurodevelopment. Important insights have come from various in vivo mouse studies. In the hippocampus, knock-down of DISC1 using shRNA methods have been shown to lead to aberrant positioning and dendritic structure of adult-born neurons. Intriguingly, the defects caused by one DISC1 shRNA of mild effect were greatly enhanced by co-expression of an shRNA to knock-down levels of NDEL1, strongly implying that these two proteins co-operate together, and consistent with the role of NDEL1 established in cultured cells. Deficiencies in the migration of neurons in the developing cortex can be seen following silencing of DISC1, PCM1 or BBS4 and in NDE1 and NDEL1 knock-out mice implying that the various DISC1-containing complexes involved in microtubule regulation are critical for cortical development.

9Migration defects in cortical neurons can also be caused by silencing the DISC1-interacting protein Dixdc1 or by inhibiting DISC1-Dixdc1 interaction using an interfering peptide. Intriguingly, Dixdc1 is also an interactor of NDEL1, and a phosphorylation site key to this interaction is required to rescue migration defects caused by suppression of DIXDC1 expression. Thus, DISC1, Dixdc1 and NDEL1 appear to co-operate in regulating migration in the developing cortex. Another potential member of this pathway is the Alzheimer’s disease-related Amyloid Precursor Protein (APP, ). Knock-down of APP levels by RNAi in the developing cortex leads to migration defects reminiscent of DISC1 knock-down which can be largely reversed by DISC1 over-expression. There is also evidence to suggest that APP is involved in the localisation of DISC1 to the centrosome in the cortex.

10Both Dixdc1 and DISC1 impact on the Wnt-signalling pathway. Silencing of DISC1 or Dixdc1 reduces lymphoid enhancer factor/T cell factor (LEF/TCF) mediated transcription and thus differentiation of neural progenitors. These effects caused by down-regulation of DISC1 expression can be rescued by expression of Dixdc1 and vice versa. The key linking molecules are Glycogen Synthase Kinase 3β (GSK3β) and β-catenin. The kinase activity of GSK3β is inhibited on binding to DISC1, preventing degradation of β-catenin and allowing it to translocate to the nucleus where it stimulates transcription of neurogenesis-related genes. These effects of DISC1/Dixdc1 silencing can be rescued by expression of β-catenin or by inhibiting GSK3β. Also, and of clinical relevance, GSK3β is a well-established target for lithium chloride, widely used in the management of bipolar disorder. GSK3β-specific inhibitors can also rescue hyperlocomotion in open field tests observed in mice expressing the DISC1-L100P mutant or in which DISC1 has been silenced in the hippocampus as well as pre-pulse and latent inhibition deficits in the L100P mouse. Intriguingly, GSK3 activity is also regulated by the APP-derived β-amyloid peptide (reviewed in ).

11Girdin (also known as KIAA1212, APE, GIV and HkRP1) is another DISC1-interacting protein, over-expression of which leads to adult-born neurons of the dentate gyrus displaying enhanced dendritic growth, increased dendritic number and over-extended migration into the outer granule cell layer and molecular layer, mirroring the effects of DISC1 depletion. Incorrect neuronal localisation and impaired mossy fibre development are also seen in girdin knock-out mice. These effects of girdin appear to be mediated via its ability to bind to and increase the activity of the serine/threonine kinase Akt. DISC1 depletion increases Akt activity at least in part through binding to girdin and preventing its Akt-stimulating activity. In agreement with this, use of rapamycin to inhibit mTOR, which lies downstream of Akt signalling, can rescue the neuronal abnormalities caused by Girdin over-expression or DISC1 knock-down. Akt is also a negative modulator of GSK3β, although an inhibitor of GSK3β was not seen to rescue these girdin-related developmental defects.

12In addition to modulating the proliferation, migration and integration of neurons, it can also be postulated that proteins of the DISC1 complex might impact upon major mental illness by modulation of synaptic transmission. In support of this, DISC1 and several of its binding partners, including citron, PDE4B, LIS1, NDE1 and NDEL1, have been found to localise to the post-synaptic density (PSD, ). To date however, relatively little is understood of the synaptic functions of these proteins.

13In contrast, more is known about the role of the PSD-localised DISC1 interactor TNIK, a kinase expressed in the mouse hippocampus whose activity is involved in regulation of the cytoskeleton. DISC1-binding inhibits the kinase activity of TNIK, leading to the degradation of several key PSD proteins, including the important structural protein PSD95, and modulating the surface expression of glutamate receptor 1 (GluR1, ). More generally, knock-down of DISC1 expression leads to an increase in the formation of spines and GluR1-expessing synapses in mature rat neurons, a process dependent on its interaction with kalirin-7 (Kal7, ). Kal7-dependent regulation of spine formation occurs through its activity as a GDP/GTP exchange factor for Rac1, and DISC1 appears to inhibit its activity by binding to Kal7 and PSD95. Activation of NMDA receptors causes dissociation of DISC1, Kal7 and PSD95, making Kal7 available to modulate Rac1 and thus spine structure. Thus DISC1 appears to modulate the formation of PSD complexes and dendritic spines through regulation of TNIK, Kal7 and Rac1.

14Other DISC1 interactors of potential importance at the synapse include APP (Young-Pearce et al., 2010), given its involvement in spine formation and enhancement of NMDA receptor activity. Another DISC1 interactor, Activating Transcription Factor 4 (ATF4 or CREB2, ) is known to bind to GABAB receptors in the synapse, and its transport from the synapse to the nucleus, where it acts as a transcriptional repressor, is implicated in long-term depression and memory.

15Another important theme in DISC1 biology is its role in a wide variety of signalling pathways, including the GSK3β and Akt pathways discussed above in Section 2.2. A third such pathway involves signalling by the ubiquitous secondary messenger molecule cAMP. The phosphodiesterase 4 family of enzymes degrade cAMP (reviewed in ) and isoforms from all four PDE4 subtypes (PDE4A-D) have been demonstrated to interact with DISC1. DISC1 binds PDE4 in a low-activity conformation and PDE4 activity is diminished in mice with a mutation, Q31L, in a PDE4-binding site on DISC1. Downstream effects of DISC1–PDE4 interaction remain to be determined, but are likely to include regulation of the activity of cAMP-dependent Protein Kinase A (PKA), substrates of which include the DISC1-interactors NDE1 and ATF4. In support of this, over-expression of DISC1 exaggerates the repression of CRE-dependent gene transcription caused by ATF4 in response to PKA. PDE4 is the known target for rolipram and other small molecules which have anti-depressant and anti-psychotic activity in rodent models.

16Another family of proteins heavily implicated in major mental illness are the neuregulins and the ErbB family of receptors for which cleaved NRG domains act as ligands (reviewed in ). Intriguingly, NRG1 and NRG2 signalling is seen to increase the expression of a specific DISC1 isoform in a process dependant on the activity of BACE to cleave neuregulins, forming extracellular peptide ligands. This pathway appears to be mediated by PI3 K/Akt signalling and is transcription-dependent. Downstream effects of NRG1 signalling include inducing the expression and activity of ATF4. Expression of DISC1 also appears to be upregulated following signalling by the neuropeptide PACAP, which additionally stimulates interaction of DISC1 with DBZ. DISC1 is also implicated in dopamine signalling, which is altered in several DISC1 mouse models. Silencing of DISC1 in rat striatal neurons leads to loss of dopamine receptor-expressing cilia.

17In summary, it is increasingly apparent that DISC1 is not simply a scaffold for the formation of protein complexes, but more an active hub for regulating divergent signalling pathways, including PDE4/cAMP, Akt/mTOR and GSK3β/β-catenin that are each well known to impact upon neurodevelopmental and/or psychiatric illness. An interesting side point is the apparent convergence of the DISC1 pathway with proteins involved in Alzheimer’s disease. DISC1 is now known to interact with APP, from which the β-amyloid peptide is derived, along with the related APLP1 protein. DISC1 also inhibits the activity of GSK3β, which is modulated by and may also modulate β-amyloid peptides, and DISC1 levels are indirectly regulated by BACE, the APP-cleaving enzyme. By implication, DISC1 plays a critical role in integrating these otherwise independent pathways, elaborating the details of which represents a key future challenge.

18Several positive genetic association studies have been reported for genes encoding DISC1 interactors (summarised in Table 2), implying that multiple DISC1-related pathways need to be considered as relevant to risk of psychopathology. ATF4, CIT (encoding citron), FEZ1, NDE1, PAFAH1B1 (encoding LIS1), PCNT (encoding pericentrin), PDE4D, TNIK and YWHAE (encoding 14-3-3ɛ) are thus all implicated in schizophrenia, although some of these are single studies or report modest associations that await firm replication, with some studies failing to replicate (Table 3 and references therein). Replication of genetic association between one or more SNPs and major mental illness supports PDE4B, NDEL1 and PCM1 in their own right (Table 2 and references therein).

19It is important here also to distinguish between the strict statistical tests for significance that must be applied to gene-wide or genome-wide test of association and the insight which can be gained from specific mutations in individual cases and families. Thus, in much the same way as DISC1 was discovered at a translocation breakpoint, PDE4B was found to be directly disrupted by a t(1:16) translocation in a proband with schizophrenia, who also had an affected cousin. Similarly, both deletions and duplications at chromosomal locus 16p13.1, containing the NDE1 gene, are significantly over-represented in schizophrenia patients in Scottish and other European populations, with a similar deletion also seen in an African–American individual with the condition. Ultra-rare missense mutations in patients with schizophrenia have been reported for APP PCM1, and indeed DISC1. These rare mutations point the finger directly at these genes and associated pathways, and further demonstrate their biological relevance.

20Evidence has also been reported of transcripts encoding several of these proteins being either up- or down-regulated in brain tissue of individuals with psychiatric illness, compared to that from healthy controls (Table 4 and references therein). Such differences in mRNA expression could be the result of direct mutation in those genes, or indirectly, due to dysregulation of transcription factors or other modulatory proteins. Additionally, certain DISC1 SNPs are associated with reduced levels of FEZ1, LIS1 and NDEL1 transcripts in the hippocampus.

21Additionally, as one would predict biologically, evidence for epistatic interaction is emerging: three-way interaction between specific SNPs in CIT, DISC1 and NDEL1 has been reported for schizophrenia; there is strong statistical interplay between the HEP3 haplotype and NDE1 in the Finnish population and haplotypes of NDE1 and NDEL1 show association that is dependent on the Ser or Cys variant at position 704 in DISC1.

22Although the emphasis of genetic studies to date has been on their potential pathological impact, it is emerging that common variants of DISC1 and its interactors impact on normal variation and intermediate phenotypes, for example memory tasks and in quantitative measures of personality and mood. We have recently reported that common cis-acting variants of DISC1 modulate expression within normal subjects by up to 20%. Moreover, variants in DISC1, PDE4 and NDE1 impact on the expression of genes involved in the cytoskeleton, neurosignalling and sensory perception, and are significantly enriched for current drug development targets in psychiatry.

23In a similar vein, the recent demonstration of at least fifty different DISC1 transcripts including an abundance that are specific to foetal development and some for which expression is altered in the hippocampus of those suffering from schizophrenia or carrying DISC1 schizophrenia-risk alleles, raised a whole new series of questions about how DISC1 expression is regulated, and with what effect on neurodevelopment and signalling.

24Whereas the growing literature on DISC1 and the DISC1 pathway, as summarised here, provides multiple avenues of promising research to pursue that is relevant to neurodevelopment, signalling and psychopathology, there are gaps and limitations. For example, with respect to the DISC1 interactome derived from yeast-two-hybrid analysis, only a minority of putative DISC1 interactors have been formally tested and confirmed by co-immunoprecipitation or co-localisation in native tissue. Although multiple transcripts and protein isoforms of DISC1 have been described, the functional role of the former and the amino-acid sequence of the later remain to be determined. Evidence from the original family from which DISC1 was identified is consistent with a simple haploinsufficiency model, but in the absence of patient tissue other than lymphoblastoid cell lines, it is not possible to rule out a dominant negative effect of hypothetical truncated or fusion DISC1 protein expression during development or in the adult brain. This family is an example of an ultra rare, in this case unique, genetic event revealing a more general genetic contribution though other genetic variants at the locus. A number of amino-acid substitutions in DISC1 been described and regulatory mutations hypothecated from association evidence, but a comprehensive analysis of all possible mutations awaits the results of large scale resequencing studies. Our understanding of the biological consequences of S704C and L607F, the best studied to date, remain partial. Regarding the cell and animal models used to test for biological effects, these do not as yet model known clinical variants, nor, for obvious reasons, do the models allow testing of psychiatric phenotypes, only at best surrogates and proxies. That said, a still modest, but growing body of evidence is making links between mouse models and human studies through comparative brain histology and imaging (reviewed in ). Thus, despite the remarkable progress, much remains to be done not just in vitro or in model systems, but in clinical studies too.

25The complexities of schizophrenia and related psychiatric illness were never likely to yield to single methodologies or models, but a combined genetic and biological approach offers promise. From what might have been viewed as an unlikely start point, the molecular genetic characterisation of a single family with a high loading for psychiatric illness, the insights from the discovery of DISC1 have been manifold and far-reaching, a paradigm for future work. It is now not so much a question of the role of DISC1 per se, but much more about the DISC1 pathway in neurodevelopment and signalling, brought to light through the multiplicity of DISC1-interacting proteins. Mechanistic details remain to be filled in and potential therapeutic targets evaluated. But in the decade since DISC1 was cloned, much progress towards these goals has been made by a combination of genetics, biochemistry, neurobiology and animal models. The next decade promises further exciting prospects to enhance our understanding of the DISC1 pathway to the benefit of patients.