DISC1 regulates cell proliferation in vitro
Expression of DISC1 peaks at E14–E15 in mouse embryonic brains, a period of active neurogenesis in the cortex, and gradually declines as development proceeds (data not shown). In adult mice, DISC1 is robustly expressed in the dentate gyrus (DG) and olfactory bulb, two regions displaying active neurogenesis (Ma et al., 2002
). In the embryonic cerebral cortex, we found that DISC1 was robustly expressed in nestin and Sox2 positive neural progenitors residing in the ventricular zone (VZ)/subventricular zone (SVZ), but reduced in doublecortin (DCX) positive neurons (, S2A
). DISC1 expression was reduced in the intermediate zone (IZ) but reappeared in the cortical plate (CP), consistent with its known function in post-mitotic neurons (). The specificity of DISC1 staining was verified using embryonic mouse brains electroporated with a DISC1 short hairpin RNA sequence (shRNA) (Figure S1B
DISC1 regulates progenitor cell proliferation in vitro
Based on its expression in neural progenitors, we looked into a potential role for DISC1 in progenitor proliferation. To approach this, we generated two specific shRNAs that silenced endogenous DISC1 expression in adult hippocampal progenitors (AHPs) (shRNA1 and shRNA2; Figure S1A
). Using lentivirus expressing control or DISC1 shRNAs, we observed that DISC1 knockdown markedly decreased cell proliferation compared to control shRNA in AHPs (). Moreover, DISC1 knockdown decreased BrdU labeling (2hr pulse) by 2-fold (, S3A
) and mitotic index by 3-fold (, S3B
). Fluorescence-activated cell sorting (FACS) analysis revealed that DISC1 knockdown significantly increased the proportion of cells in G0/G1 (). The reduction of proliferation by DISC1 knockdown prompted us to pursue the reciprocal DISC1 gain-of-function experiment. Notably, overexpression of full length hDISC1 resulted in a 2-fold increase in cell proliferation, 2-fold increase in BrdU labeling, and 3-fold increase in mitotic index (, Figure S3C
, Figure S3D
). Thus, DISC1 is required for the normal proliferation of AHPs, and its overexpression promotes proliferation.
DISC1 regulates cortical progenitor proliferation in utero
We next investigated whether DISC1 regulates neural progenitor proliferation in vivo. Control or DISC1 shRNA constructs were electroporated together with a GFP-expressing vector into E13 mouse brains, and the brains were analyzed 2 days later. The positioning of GFP positive cells revealed that DISC1 knockdown resulted in a substantial reduction of cells in the VZ/SVZ (), with a corresponding increase in GFP positive cells in the CP. Therefore, DISC1 loss-of-function causes a depletion of cells from the proliferative VZ/SVZ.
DISC1 regulates progenitor cell proliferation in utero
To gain insight into the mechanism accounting for the observed difference in cell positioning caused by DISC1 knockdown, we injected BrdU into the pregnant dams 2 hours prior to collection of the electroporated brains. DISC1 knockdown resulted in a marked reduction in BrdU labeling () and mitotic index () that could be rescued by human DISC1 cDNA (not targeted by shRNA-1) (), demonstrating that DISC1 is required to maintain cortical progenitor proliferation in vivo. Conversely, DISC1 overexpression increased both the percentage of cells remaining in the VZ/SVZ and the BrdU labeling index ().
Stable β-catenin rescues DISC1-induced defects
Our observations that DISC1 knockdown reduced neural progenitor proliferation and decreased positioning in the VZ/SVZ suggest that cells may be prematurely differentiating into neurons. To test this possibility, we measured the cell cycle exit index (Sanada and Tsai, 2005
). E13 mouse brains were electroporated with DISC1 shRNA constructs and BrdU was injected 2 days later at E15 into pregnant dams. At E16, brains were collected and analyzed by immunohistochemistry using anti-GFP, -BrdU, and -Ki67 antibodies (). We observed a 2–3 fold increase in cell cycle index in DISC1 shRNA transfected embryonic brains, suggesting that the reduction of proliferating progenitors in DISC1 shRNA treated brains likely results from increased cell cycle exit. Consistent with this, a significant increase in DCX positive cells and decrease in Sox2 positive cells was observed in DISC1 shRNA transfected brains compared to control shRNA transfected brains (Figure S4A, B
). Supporting the premature neuronal differentiation, we found that DISC1 knockdown in E15 brains resulted in a significantly higher proportion of Cux2 (layer 2 and 3 marker) positive cells at postnatal day 7 compared to the control. Furthermore, more control shRNA transfected cells were found to be migrating towards layer 3 (Figure S5A
). Taken together, these results suggest that DISC1 knockdown causes premature neuronal differentiation at the expense of the progenitor pool.
Regulation of β-catenin signaling by DISC1
β-Catenin signaling is a conserved pathway implicated in maintenance of the stem cell pool (Lie et al., 2005
; Zechner et al., 2003
), neuronal differentiation (Hirabayashi et al., 2004
), and development of the central nervous system (Schuller and Rowitch, 2007
). Both DISC1 and β-catenin are highly enriched in neural progenitors in the VZ. Previous reports demonstrated that ablation of β-catenin expression in the developing brain results in the depletion of neural progenitors, whereas transgenic mice expressing stabilized β-catenin exhibit a drastically expanded progenitor pool in embryonic brains (Chenn and Walsh, 2002
; Zechner et al., 2003
). Thus, β-catenin and DISC1 share similar properties in regulating neural progenitors.
Since β-catenin is a central downstream effector of canonical Wnt signaling, we investigated the possibility that DISC1 may modulate Wnt-mediated proliferation. The addition of Wnt3a (200 ng/ml) to culture medium significantly stimulated the proliferation of AHPs, and this increase in proliferation was abolished by DISC1 knockdown (). This suggests that DISC1 function converges with downstream mediators of Wnt-dependent proliferation.
DISC1 regulates the β-catenin pathway
β-Catenin exerts its function in part through nuclear translocation to stimulate the transcription of genes containing binding sites for the lymphoid enhancer factor-T cell factor (LEF/TCF) family (Gregorieff and Clevers, 2005
). The transcription complex containing nuclear β-catenin activates the expression of target genes important for cell proliferation (Adachi et al., 2007
). Since DISC1 knockdown abrogated Wnt-dependent proliferation, we determined whether DISC1 is required for LEF/TCF activation using a luciferase reporter construct containing 8 copies of the LEF/TCF binding site (8XSuperTOPFLASH (TOP)) or mutated LEF/TCF binding sites (8XSuperFOPFLASH (FOP)) (Veeman et al., 2003
). DISC1 knockdown significantly reduced TOP, but not FOP, reporter activity in cultures treated with Wnt3a (). Importantly, TOP reporter activity could be rescued by co-expressing human DISC1 cDNA with DISC1 shRNA-1 (). The reduction in LEF/TCF activity by DISC1 knockdown could also be recapitulated in the developing brain (). Furthermore, we found that DISC1 knockdown had no effect on irrelevant CRE and C/EBP-ATF reporters, confirming the specific effect on LEF/TCF activity (Figure S6A, B
). If DISC1 loss-of-function attenuates Wnt signaling, then DISC1 gain-of-function should potentiate this signaling pathway. DISC1 overexpression increased TOP, but not FOP, reporter activity in primary neural progenitor cultures () and in embryonic brains by more than 2-fold (). Together, these results indicate that DISC1 participates in the Wnt signaling pathway and in Wnt-mediated cell proliferation.
Stable β-catenin expression overrides progenitor proliferation defects induced by DISC1 knockdown
To further decipher the mechanism by which DISC1 influences β-catenin and LEF/TCF activity, we created two β-catenin shRNAs that efficiently silenced endogenous β-catenin expression (Figure S7A
). We found that the enhancement of LEF/TCF reporter activity by DISC1 overexpression was completely abolished when the expression of endogenous β-catenin was silenced (), indicating that the effects of DISC1 on LEF/TCF reporter activity require β-catenin. We further evaluated the effects of co-expressing WT-β-catenin, or a degradation-resistant mutant of β-catenin-S33A (SA-β-catenin) with control or DISC1 shRNAs on LEF/TCF reporter activity. We found that both WT and mutant β-catenin potentiated TOP reporter activity (). However, while DISC1 knockdown significantly down-regulated reporter activity in WT-β-catenin expressing cells, it had no effect on SA-β-catenin-mediated reporter activation (). This suggests that DISC1 acts upstream of β-catenin and impacts LEF/TCF transcription by regulating β-catenin abundance.
To further examine the relationship between DISC1 and β-catenin in brain development, we determined whether SA-β-catenin can rescue progenitor proliferation in vivo when DISC1 expression was silenced. As mentioned earlier, DISC1 knockdown reduced the percentage of GFP positive cells in the VZ/SVZ, increased GFP positive cells in the CP, and reduced BrdU labeling and the mitotic index (). Remarkably, co-expression of SA-β-catenin with DISC1 shRNA-1 completely rescued these phenotypes (). This observation underscores a major role for DISC1 in regulating progenitor proliferation by modulating β-catenin levels.
DISC1 regulates β-catenin abundance
Increased β-catenin levels rescued the defects caused by DISC1 knockdown, suggesting that DISC1 may regulate β-catenin abundance. Indeed, we found that DISC1 shRNAs significantly decreased β-catenin levels in AHPs (). GSK3β regulates β-catenin stability by phosphorylating serine and threonine residues (Ser33/37 and Thr41) important for targeting β-catenin for ubiquitin-dependent proteasomal degradation (Aberle et al., 1997
). Notably, we observed that the reduction in β-catenin levels caused by DISC1 knockdown was accompanied by increases in Ser33/37 and Thr41 phosphorylation () and β-catenin ubiquitination (Figure S7B
). Thus, DISC1 loss-of-function reduces β-catenin abundance. We further evaluated the effect of DISC1 gain-of-function on β-catenin levels. Overexpression of WT-DISC1 reduced β-catenin S33/37/T41 phosphorylation, decreased ubiquitination, and increased total β-catenin levels in progenitors ( & S7C
). Collectively, these results suggest that DISC1 regulates β-catenin stability.
DISC1 regulates β-catenin stability
If DISC1 is crucial for stabilizing β-catenin, one might predict that β-catenin transcriptional targets may be elevated by DISC1 gain-of-function. Cyclin D1 (Tetsu and McCormick, 1999
) and axin2 (Leung et al., 2002
) are well-established targets of β-catenin, and cyclin D1 promotes G1 progression during the cell cycle. In AHPs, cyclin D1 levels were reduced by 50% in DISC1 shRNA-1 expressing cells and by 70% in shRNA-2 expressing cells (). A similar decrease was observed with Axin2. Conversely, cyclin D1 and axin2 expression was markedly upregulated in DISC1 overexpression cells (). These data support the notion that DISC1 controls cell proliferation by regulating β-catenin abundance, thereby fine-tuning cell cycle progression.
DISC1 regulates GSK3β activity
The increase in β-catenin phosphorylation caused by DISC1 knockdown raised the possibility that GSK3β activity may be directly or indirectly influenced by DISC1. Growth factors such as insulin activate AKT, which in turn phosphorylates GSK3β at Ser9, an inhibitory phosphorylation site (Cross et al., 1995
). Furthermore, GSK3β autophosphorylates itself at Tyr216 (Lochhead et al., 2006
), which is required for its activity. Upon transduction of AHPs with DISC1 shRNA lentiviruses, we observed a significant increase in Y216 phosphorylation, suggesting that DISC1 negatively impacts GSK3β activity (). In contrast, Ser9 phosphorylation was not affected by DISC1 knockdown (). Furthermore, DISC1 overexpression reduced Y216 phosphorylation (), further supporting the role of DISC1 in inhibition of GSK3β activity. Notably, phosphorylation of other known GSK3β substrates, Ngn2 (Ma et al., 2008
) and C/EBPα (Ross et al., 1999
), was not affected by DISC1 shRNAs or overexpression. Thus, DISC1 selectively regulates the phosphorylation of certain GSK3β substrates.
Consistent with its ability to regulate GSK3β activation, we found that DISC1 associates with GSK3β in E15 mouse embryonic brains (). To determine whether DISC1 and GSK3β directly bind each other and to map the region(s) of DISC1 required for the interaction, we generated GST-tagged DISC1 protein fragments and performed an in vitro association assay using purified His-tagged GSK3β (). Only DISC1 fragments spanning residues 1–220 and 356–595 exhibited strong interactions with GSK3β. Collectively, these results suggest that DISC1 directly interacts with GSK3β and inhibits its activity.
DISC1 regulates the GSK3β signaling pathway
To examine the effect of DISC1 fragments on GSK3β activity, GST-DISC1 fragments, GST-CASK fragments, or GST-p25 were incubated with purified active GSK3β in the presence of the GSK3β substrates β-catenin and axin (Thomas et al., 1999
) in vitro
. We then analyzed the extent of GST-β-catenin phosphorylation, GST-axin phosphorylation, and GSK3β Y216 autophosphorylation. Consistent with its ability to bind GSK3β, the DISC1 fragment spanning residues 1–220 potently inhibited β-catenin, axin, and GSK3β phosphorylation at 0.5 µM, whereas none of the other DISC1 fragments, GST-p25, or GST-CASK proteins inhibited GSK3β activity at this concentration (). However, at higher concentrations (2 µM), DISC1 fragments spanning 221–355 and 356–595 started to inhibit GSK3β autophosphorylation (Figure S7D
). This suggests that the inhibitory activity of DISC1 on GSK3β may reside on multiple domains, but that fragment 1 (1–220) possesses the most potent inhibitory activity. Furthermore, a dose response curve was established for DISC1 1–220 whereby no inhibition of GSK3β Y216 phosphorylation was observed at 0.1 µM and the inhibition of phosphorylation plateaued at 1 µM (Figure S7E
). None of the GST-DISC1 fragments inhibited AKT autophosphorylation (data not shown) indicating that DISC1 is not a general kinase inhibitor. To narrow down the domain in DISC1 that inhibits GSK3β, we synthesized two peptides from mDISC1 which are highly conserved between human and mouse (Figure S8
). The first peptide spanned amino acids 40–77 (Peptide-neg) and the second peptide spanned amino acids 195 to 238 (DISCtide-1). In vitro
kinase assays demonstrated that Peptide-neg did not inhibit GSK3β at 80 µM, whereas DISCtide-1 inhibited GSK3β at 10 µM. We further found that DISCtide-1 inhibited GSK3β more potently than L803-mts, a previously described GSK3β peptide inhibitor (Plotkin et al., 2003
Surface plasmon resonance (SPR) was used to determine whether DISCtide-1 directly binds to GSK3β. GSK3β SPR binding assays were carried out with mDISC1 Peptide-neg, DISCtide1, the GSK3β inhibitor L803-mts, and as a negative control, a peptide from the N-type calcium channel, CACNA1B (). We found that both DISCtide1 and L803-mts bound to GSK3β (percent theoretical maximal response 197% and 259%, respectively) at a concentration of 25 µM (), while both Peptide-neg and the calcium channel peptide showed much lower binding (percent theoretical maximal response 62% and 29%, respectively). Interestingly, the super-stoichiometric binding exhibited by both L803-MTS and DISCtide1 (259% and 197%, respectively) suggests that both bound to GSK3β in these assays at a peptide:protein ratio of 2:1. To identify the DISCtide1 sequences that mediate this interaction with GSK3β, we designed overlapping 15-mer peptides that covered DISCtide1 and tested their binding to GSK3β by SPR (). One 15-mer peptide that encompasses mDISC1 amino acids 211 to 225, number 43 (DISCtide2), bound to GSK3β, while all other DISCtide1 15-mers showed background levels of binding (). Similar to DISCtide1, DISCtide2 also showed super stoichiometric binding to GSK3β in these assays with a peptide:protein ratio of 2:1 (percent maximal binding 238%). Further characterization of DISCtide2 is shown in Figure S9
. Importantly, a reversed DISCtide2 sequence displayed negligible binding to GSK3β demonstrating that the binding of DISCtide2 to GSK3β is specific. Despite the specific binding to GSK3β, DISCtide2 failed to inhibit GSK3β kinase activity (data not shown), suggesting that other regions of DISCtide1 are also required for the inhibition.
DISC1 regulates progenitor proliferation by inhibiting GSK3β
If DISC1 inhibits GSK3β activity, we speculated that the negative effects of DISC1 knockdown on proliferation should be alleviated by GSK3β loss-of-function. To test this possibility, we examined the consequence of treating DISC1 knockdown AHP cells with SB-216763, a specific chemical inhibitor of GSK3. SB-216763 restored cell proliferation as evaluated by BrdU incorporation (Figure S10A
). SB-216763 also rescued neural progenitor proliferation defects in vivo. SB216763 increased BrdU labeled cells in control shRNA electroporated mouse embryonic brains (Figure S10C
) and rescued the BrdU labeling index in DISC1 shRNA electroporated embryonic brains. Consistent with these results, we further found that the GSK3β inhibitors SB-216763 and CHIR-99021 rescued TCF activity in DISC1 knockdown cells to control levels (Figure S10B
). Finally, we tested the consequence of overexpressing GSK3β in the embryonic brain. While GSK3β overexpression reduced the number of BrdU labeled cells (Figure S10D
), this was suppressed by DISC1 coexpression. Taken together, these results provide further evidence that DISC1 regulates neural progenitor proliferation by inhibiting GSK3β.
DISC1 regulates adult neural progenitor proliferation by inhibiting GSK3β
In the adult dentate gyrus, DISC1 is expressed in neural progenitors and neurons, but is absent in astrocytes (Figure S11A–C
). To determine whether DISC1 regulates hippocampal progenitor proliferation, we stereotactically injected control or DISC1 shRNA-1 lentivirus into the dentate gyrus of adult mouse brains. Five weeks later, we administered daily injections of BrdU over the course of 7 days (). Lentivirus expressing control or DISC1 shRNA showed comparable infection rates in the dentate gyrus as revealed by the GFP signal (). Consistent with our observations in embryonic brains, we observed a significant reduction in BrdU incorporation in the DISC1 shRNA-1 group compared to the control shRNA group (). This reduction was not due to increased cell death, as there was no increase in active caspase 3 labeling of GFP positive cells in the dentate gyrus (Figure S12
). To determine whether this effect was mediated through GSK3β activation, vehicle or SB-216763 (2 mg/kg) was administered to mice every other day for 2 weeks (4 weeks after the administration of virus) (). Intriguingly, SB-216763 treatment significantly increased the number of BrdU positive cells in mice infected with the control virus () and rescued the reduction in BrdU positive cells in DISC1 shRNA-1 injected mice. These results suggest that DISC1 also regulates the proliferation of adult hippocampal progenitors by inhibiting GSK3β activity. Consistent with a previous report (Duan et al., 2007
), we also observed aberrant positioning and increased complexity of dendritic morphology in DISC1 knockdown granule neurons (data not shown).
GSK3β inhibitor rescues the proliferation and behavior defect caused by DISC1 knockdown in adult mice
Behavioral consequences of DISC1 loss-of-function and GSK3β inhibition
Alterations in neurogenesis are implicated in the development of depression and other abnormal behaviors. To examine the behavioral consequences of disrupting the DISC1/GSK3β pathway, we silenced DISC1 expression in the adult dentate gyrus and evaluated behavioral consequences. Four weeks after the injection of control or DISC1 shRNA-1 lentivirus, mice were treated with vehicle or SB-216763 (2 mg/kg i.p.) every other day for 10 days. Compared to control mice, DISC1 shRNA injected mice exhibited hyperlocomotion in response to novelty, which is considered a model of positive symptoms of schizophrenia. DISC1 shRNA injected mice traveled a greater distance and spent more time moving in a novel open field than control mice (). These behaviors were normalized by SB-216763 treatment. We further tested whether DISC1 loss-of-function had consequences on depression-like behavior in the forced swim test. DISC1 shRNA infected mice displayed greater immobility (), an indicator of depressive behavior, which was also suppressed by SB-216763. Importantly, swimming velocity was unchanged (data not shown). DISC1 shRNA infected mice did not display increased anxiety, as there was no difference in time spent in the closed arms versus the aversive open arms in the elevated plus-maze (). Thus, increased GSK3β activity caused by DISC1 loss-of-function is associated with schizophrenia- and depression-like behaviors.