3 November 2009. There was no shortage of DISC1 chatter at the 40th annual meeting of the Society for Neuroscience, held 17-21 October 2009 in Chicago. The highlight was a mini-symposium on Monday, convened by Nicholas Brandon of Pfizer (formerly Wyeth), which covered topics ranging from the role of DISC1 in developing and mature neurons, to the question of whether DISC1 aggregation might contribute to sporadic disease. Co-organizer Akira Sawa of Johns Hopkins, Baltimore, Maryland, unfortunately had to miss the session, but a preview of the mini-symposium was published on October 14 in the Journal of Neuroscience (Brandon et al., 2009). This story covers highlights of the talks as well as posters on DISC1 presented at the meeting.
DISC1 in development
It has been nearly a decade since DISC1 was identified at the breakpoint of a 1:11 chromosomal translocation cosegregating with major mental illness in a large Scottish family (Millar et al., 2000). Since then, a picture has emerged of DISC1 as a multitasking scaffold that associates with a variety of partners to regulate neurogenesis, neuronal migration, and integration during fetal development and in adulthood. Ju Young Kim talked about her recent work from the Ming lab at Johns Hopkins University, Baltimore, Maryland, showing that DISC1 regulates the morphology, migration, and integration of new neurons in the adult hippocampus by negatively regulating the activity of the Akt/mTOR pathway via the actin-binding protein girdin, also known as KIAA1212 (see SRF related news story). This pathway is distinct from another DISC1-regulated cascade that Li-Huei Tsai and colleagues at MIT, Cambridge, Massachusetts, identified recently (see SRF related news story). That work showed that DISC1 regulates GSK3/Wnt/β-catenin signaling, which is critical for the proliferation of neuronal progenitors during development and in adult mice.
Does DISC1 tie into different signaling pathways at different stages of a cell’s life? Kim presented two pieces of data that support this idea. In Tsai’s published experiments, inhibitors of GSK3β reversed the effects of DISC1 loss on neurogenesis. However, Kim reported that GSK3β inhibitors had no effect on DISC1-related changes in morphogenesis and migration in the postnatal hippocampus. Also, expression profiling revealed little girdin in adult neuronal progenitor cells, but Kim noted that expression increased with differentiation.
Karun Singh from the Tsai lab showed new data on another DISC1 interacting protein, suggesting that it might comprise a redundant pathway for β-catenin activation in neuronal progenitor cells. The protein, DIXDC1, is a positive regulator of Wnt signaling that has mostly been studied in cancer (Wang et al., 2009), but is highly expressed in neuronal progenitors and neurons. Singh reported that DIXDC1 interacts with DISC1 in E15 mouse brain. In cell-based assays, either DIXDC1 or DISC1 activated a β-catenin-regulated reporter gene. The two appeared to compensate for each other, as overexpression of one could rescue the activity of the other on gene expression in a knockdown model.
Singh speculated that these parallel pathways could provide a biological explanation for the incomplete penetrance of DISC1 deletions, where overexpression of DIXDC1 might compensate for loss or lower expression of DISC1. In support of that idea, he found that DIXDC1 expression rescued the proliferation defect caused by DISC1 RNAi knockdown in mouse hippocampus. But the two genes were not exactly equivalent. Where DIXDC1 RNAi caused a decrease in progenitor proliferation and a neural migration defect, DISC1 could reverse the progenitor cell defect, but not the migration problem. Singh presented data that DIXDC1 affected cell migration by binding to the DISC1 partner nuclear distribution element 1 (Ndel1) protein.
More evidence for the importance of the GSK3/β-catenin pathway in DISC1 actions comes from the work of Hazel Sive of MIT, who described studies on the zebrafish homolog of DISC1. Knockdown of that gene is embryonic lethal and results in a truncated brain structure with abnormal neuronal patterning. The phenotype, which Sive showed arises from unregulated GSK/β-catenin signaling, can be substantially rescued by expression of human DISC1.
Using the zebrafish, Sive tried to introduce a C-terminal truncation analogous to the human chromosome 1:11 translocation using an antisense oligonucleotide to target the exon8/intron boundary. The treatment resulted in a 50/50 mix of wild-type and truncated mRNAs, but protein levels were not measured. The manipulation caused a severe brain defect, but surprisingly it occurred without activation of GSK3 or loss of β-catenin-driven gene expression.
This brought up the question of what is happening in people with the DISC1 translocation—are GSK3β and β-catenin involved or not? Brandon pointed out that no one has detected truncated protein in blood cells from people with the translocation, and the model of DISC1 pathology invokes haploinsufficiency. However, Brandon said, “We won’t know for sure until we have brain tissue from that family.”
Standing in for the absent Sawa, Atsushi Kamiya, also of Johns Hopkins, spoke about how neurons might balance two opposing roles of DISC1 in neurons: on the one hand, the protein regulates proliferation of neuronal progenitors through the GSK pathway, and on the other, it controls migration. Previous work from Kamiya and Sawa had shown that migration involves DISC1’s interaction with the centrosome and cytoskeletal proteins (see SRF related news story).
According to Kamiya’s new data, the switch between proliferation and migration seems to be the phosphorylation state of DISC1. He showed evidence that only DISC1 that is dephosphorylated at serine 710 (S710) inhibits GSK3β and activates β-catenin-mediated gene expression. In contrast, S710 phosphorylation increases binding of DISC1 with proteins involved in the centrosome pathway.
To look at the switch in cells, they compared the affinity of DISC1 for partner proteins in embryonic neurons. In cells from embryonic day 18, the time of peak migration, the interaction with centrosome-related proteins was stronger than with GSK3β. Conversely, in cells from embryonic day 14, when proliferation is prominent, DISC1 bound more avidly to GSK3. In cells with no DSIC1, the proliferation defect was rescued by an A710 mutant that was incapable of being phosphorylated, but not by the phosphomimic E710, while the opposite was true for the migration defect. The results suggest that phosphorylation of DISC1 may regulate neuronal migration. Kamiya says he is now testing candidate kinases.
Knowledge of how DISC1 participates in development is important, but how will that contribute to understanding mental illness and finding novel therapies? To achieve that goal, Kamiya says, it will be necessary to look past development, and understand DISC1 function in mature neurons. Along those lines, he touched on some new data that were further elaborated in two posters describing a DISC1 knockout recently generated by the Sawa lab in collaboration with Toshifumi Tomoda of the City of Hope/Beckman Research Institute in Duarte, California. The mice, which lack the first three exons of DISC1, do not show neuronal migration defects, but there is some question as to whether they express a C-terminal fragment of DISC1. What the researchers do know is that the mice have axonal and BDNF abnormalities. Cultured cortical neurons from the mice exhibit transport defects, including less movement and secretion of BDNF, which can be reversed by lithium treatment. In vivo, the mice show axonal pathology along the cortico-striatal tract, which was reversed by postnatal expression of DISC1. The mice also show some altered behaviors, such as defects in prepulse inhibition, and the researchers are planning to test whether lithium can rescue the behavioral deficits.
DISC1 in synapses
Nick Brandon echoed Kamiya’s sentiment that while understanding the role of DISC1 in development makes for interesting basic neuroscience, it may not be the shortest path to new treatments. For therapeutic insights, he is focused squarely on the synapse, and so was his talk. DISC1is found in the postsynaptic density in adult human brain (see SRF news story http://www.schizophreniaforum.org/new/detail.asp?id=1270). In cultured cells, the protein localizes around the centrosome at first, and as neurons mature it goes to synapses. Knockdown of DISC1 results in synaptic changes, including enlarged spines and increased surface expression of the GluR1 subunit of the AMPA glutamate receptor in cortical neurons. To understand the role of DISC1 in synapses, and its interaction with the cytoskeleton, Brandon revisited the DISC1 interactome (Camargo et al., 2007), and focused on a synaptic kinase they found there, Traf- and nck-interacting kinase (TNIK). TNIK has been localized to the post-synaptic density in rodent brain, and came up as a gene of interest in two genome wide association studies, one for schizophrenia (Shi et al., 2009) and the other for genes involved in prefrontal cortex function (Potkin et al., 2009). Brandon showed that TNIK is expressed throughout rodent brain, especially in the hippocampus. In the cortex and hippocampus, TNIK colocalized with synaptic DISC1, and both were enriched and associated in the postsynaptic density.
Mapping the interaction showed that DISC1binds TNIK in its kinase domain, while TNIK contacts DISC1 in the region of amino acids 335-348. Brandon then made cell-permeable peptides derived from the DISC1 binding site and showed they could inhibit TNIK activity in cells, and rapidly cause a decrease in key PSD proteins, including PSD95 and GluR1. Depleting TNIK using RNAi in hippocampal neurons resulted in a similar decrease in surface GlurR1, AMPA receptors and activity. Knockdown of DISC1 resulted in higher TNIK levels, and an increase in GluR1. PSD95 was not restored however, but instead was dramatically reduced.
Another partner of DISC1 and TNIK at the synapse is kalirin-7, an activator of the small GTPase Rac-1 that is required for both morphological and functional remodeling of synapses in mature cortical neurons (see SRF related news story). Kalirin-7 has previously been shown to be decreased in schizophrenia brain (see SRF related news story) and kalirin knockout mice show some schizophrenia-related behaviors (see SRF related news story). Brandon showed that kalirin-7 associates with DISC1 close to the TNIK site. A common endpoint for TNIK and kalirin-7 is regulation of the actin cytoskeleton, and in future work Brandon and colleagues will aimed at parsing how DISC1 regulates each of these proteins and through them, cytoskeletal dynamics.
DISC1 in aggregates?
Carsten Korth of the Heinrich Heine University in Düsseldorf, Germany, posed a different question. How might the non-mutant, full-length DISC1 be linked to sporadic illness? So far, people have looked at truncations or outright loss of the protein, but is it possible that a normal DISC1 might be somehow modified and become dysfunctional? Korth takes this idea from the filed of neurodegeneration, where misfolded or aggregated proteins cause dementing or paralyzing diseases including Alzheimer disease (AD), Huntington disease and Parkinson disease. In these disorders, protein aggregates range from large microscopically visible plagues such as in AD, to small aggregates visible only with immunostaining as in some polyglutamate disorders. Could a similar situation pertain to schizophrenia, where patients with negative symptoms and chronic progressive course might be displaying features of neurodegeneration?
To answer this question, Korth and colleagues asked if they could identify insoluble DISC1 in brains of people with schizophrenia. They fractionated homogenates of frozen brain from the Stanley Medical Research Institute Consortium, including 15 subjects with schizophrenia, 15 with bipolar, 15 with major depression and 15 normal. They find a subset of cases show detergent insoluble DISC1 aggregates, and the average level in disease tissue is 4 times higher than normal brain (Leliveld et al., 2008). The aggregates were not restricted to schizophrenia, but were seen in people with bipolar disease and depression as well. There was no correlation between aggregation and any specific clinical features in the patients, Korth said, nor did he see any signs of degeneration in the area where aggregated proteins were seen.
Self-association domains reside in the C-terminal end of DISC1 and a schizophrenia-associated polymorphism, S704C, enhances aggregation of Disc1 C-terminal fragments (Leliveld et al., 2009). In a poster presentation, Svenja Trossbach from the Korth lab showed the same results for full length DISC1.
By looking at proteins overexpressed in cells, or using purified recombinant protein, Korth and colleagues found that aggregated DISC1 loses its ability to associate with Ndel1, suggesting that aggregation might disrupt normal DISC1 function. A poster from Josef Kittler’s lab at University College London lent support to that idea. Presenter Talia Atkin showed that overexpression of DISC1 in cell lines or primary neurons in culture leads to perinuclear aggregation. The aggregates sequester Ndel1 and the cells show disrupted intracellular trafficking of mitochondria and endosomal mitochondrial trafficking.
Korth and colleagues also found a way to enhance DISC1 aggregation by treating cells with dopamine. Dopamine has been previously shown to modify the Parkin protein (LaVoie et al., 2005), and Korth showed that when DISC1-transfected cells were treated with dopamine, the amount of soluble protein was reduced and insoluble multimers increased. Along with the shift of DISC1 to insoluble fractions, adding dopamine caused of loss of Ndel1 interaction. An in vivo experiment using amphetamine-sensitization also increased insoluble dopamine, and the researcher found a correlation between insoluble protein levels and hyperactive behavior measured by open field travel. The model will be useful to look at behavior due to insoluble, full-length wild-type DISC1, Korth said.
Christopher Ross of Johns Hopkins University, Baltimore, Maryland, was at the minisymposium and told SRF that the aggregation data need to be replicated, but he found it the most exciting part of the session.
Posters of note
Two groups used DISC1 mice to look at gene-environment interactions, and in particular the relationship between immune responses and schizophrenia. Maternal infection is a risk factor for schizophrenia, and both groups used injection of poly dI:dC to mimic viral infection and activation of the innate immune system. Daisuke Ibi, from the lab of Kiyofumi Yamada at Nagoya University in Japan, injected neonatal mice expressing a dominate negative DISC1 construct and found synergistic effects on behaviors related to DISC1 expression and on loss of parvalbumin-positive neurons in adult mice (Ibi et al., 2010). In an independent study, Bagrat Abazyan of Johns Hopkins injected pregnant mice of the same dominant-negative DISC1 model with poly dI:dC, and found increases in inflammatory cytokines in the fetal brain 3-6 hours after injection. They also tested the effects of one upregulated cytokine, IL-6, on cortical neurons, and found that while the cytokine normally increased neurite complexity on long neurites in wild type cells, it affected mainly short neurites in cells from the brains of the domaint negative DISC1 mice. The results suggest mutant DISC1 can affect neuronal differentiation by altering the response to IL-6. The mice may provide a model for further analysis of the interactions between immune system activation and genetic susceptibility, the researchers say.—Pat McCaffrey.