SUMOylation of DISC1: A Potential Role in Neural Progenitor Proliferation in the Developing Cortex

Over the past several decades, we have used genetic and neuropathological approaches to explore compelling molecular leads to address the mechanisms of neurological and neuropsychiatric disorders [1,2,3]. For example, tau and α-synuclein proteins have been underscored as major components of inclusion bodies that occur in the brains of patients with Alzheimer's disease and Parkinson's disease, respectively [4,5]. The tau and α-synuclein proteins are pathologically phosphorylated [6,7], and the pathological roles of these posttranslational modifications have been elucidated [8]. Likewise, pathological implications of other posttranslational modifications have also been reported [9].

The DISC1 locus has received broad attention in the context of psychiatric illness, since a balanced translocation that segregated in a Scottish pedigree with schizophrenia and depression was reported [10]. Since the discovery of the gene, many neurobiologists have studied the biological function of DISC1. At the cellular level, DISC1 plays a pivotal role in neural progenitor proliferation, migration, and synaptic maturation [11,12]. Perturbation of the biological pathways involving DISC1 is known to lead to behavioral changes in rodents [13,14,15,16,17,18,19,20,21]. In contrast to biological results that support the role for DISC1 in mental conditions, genetic evidence to support the contribution of the DISC1 gene in sporadic schizophrenia has been modest [22]. The discrepancy between modest evidence in genetics and strong biological support for the role of DISC1 in mental conditions led us to propose a working hypothesis that regulation of DISC1 at the protein level, such as posttranslational modification, may play a role in its function in pathology.

DISC1 is located in multiple subcellular domains, including the centrosome, postsynaptic density, and the nucleus [11,23,24]. Like tau and α-synuclein, specific phosphorylation on DISC1 affects its localization to the centrosome and plays a crucial role in neurodevelopment [25]. Protein targeting to the nucleus is also influenced by posttranslational modifications, such as SUMOylation [26]. SUMOylation occurs on lysine residues whereby small ubiquitin-related modifiers (SUMO) and homologs are conjugated to proteins [27]. SUMOylation has emerged as a major regulator of nuclear function, including DNA replication, DNA damage response, and transcription [28,29,30]. Furthermore, an important role for SUMOylation in protein trafficking to the nucleus has also been reported [31].

In the present study, we show that human DISC1 is SUMOylated at lysine 643 (K643) with in silico, biochemical, and cell-biological approaches. This specific SUMOylation site of DISC1 regulates neural progenitor proliferation in the developing cortex.

Pregnant C57BL/6 mice were purchased from Charles River.

A K643A variant of human DISC1 or a K640A variant of mouse DISC1 (mDISC1) was introduced by PCR-based mutagenesis as mentioned previously [25]. All constructs in the present study were created as described previously [24,32]. We used a short hairpin RNA (shRNA) construct of DISC1 that was published previously (the targeting sequence is 5′-GGCAAACACTGTGAAGTGC-3′) [24]. We used a control RNAi construct with a scrambled sequence without homology to any known messenger RNA [33]. GFP-tagged SUMO1, FLAG-tagged SUMO1, and Myc-tagged Ubc9 constructs were generous gifts from Dr. Pandolfi (Beth Israel Deaconess Cancer Center, Boston, Mass., USA), Dr. Hayward (The Johns Hopkins University, Baltimore, Md., USA), and Dr. Matunis (Johns Hopkins Bloomberg School of Public Health, Baltimore, Md., USA), respectively. The following antibodies were used: mouse monoclonal anti-HA antibody (Covance); rabbit polyclonal anti-FLAG antibody (Sigma); rabbit polyclonal anti-SUMO1 antibody (Zymed); mouse monoclonal anti-BrdU antibody (BD Biosciences); anti-DISC1-C2 antibody [34], and a rabbit polyclonal anti-human DISC1 antibody (kindly provided by Dr. Akiyama, University of Tokyo, Tokyo, Japan) [35,36,37].

Prediction of DISC1 SUMOylation sites was made by using the SUMOplot™ Analysis Program from Abgent available at http://www.abgent.com/sumoplot. The score indicates the probability that the targeted lysine is SUMOylated.

To obtain C-terminal DISC1-fragmented glutathione S-transferase (GST-DISC1-C) fusion proteins, the C-terminal of DISC1 (amino acids 598-854) was inserted into the pGEX vector. The expression plasmids were introduced into BL21-competent Escherichia coli with 0.1 mM IPTG. Recombinant proteins were purified from E. coli with glutathione Sepharose or amylase beads.

In vitro SUMOylation was assayed by use of the SUMOylation assay kit (Boston Biochem). Briefly, reactions contained 10 µg GST-DISC1-C, 50 µM SUMO proteins, 100 nM SUMO-activating enzyme E1, 5 µM Ubc9, and 1 mM Mg-ATP in buffer (50 mM HEPES, pH 8.0, 100 mM NaCl, 1 mM DTT), and were incubated at 37°C for 2 h. The reactions were also run in the absence of ATP as a negative control. Proteins were separated on a NuPAGE 4-12% Bis-Tris gel (Invitrogen) and visualized with SimplyBlue SafeStain (Invitrogen). For Western blotting, an 8% Tris-Glycine gel (Invitrogen) was used.

A Ubc9 Fusion-Directed SUMOylation (UFDS) assay was performed as previously described [32], with minor modifications. Briefly, HEK293 cells were transfected with the construct of DISC1-Ubc9 or DISC1-Ubc9 (C93S) together with the SUMO1-GFP construct, and lysed 48 h after transfection. Immunoblots were conducted with the anti-DISC1 antibody described above [35].

COS7 cells were used to assay SUMOylation on DISC1. Cells were grown in Dulbecco's modified Eagle's medium with 10% FBS. The expression constructs were transfected into cells by using FuGENE® 6 (Roche Applied Sciences). U373 cells were transfected with a DISC1 or DISC1-K643A construct to examine the nuclear translocation of SUMOylated DISC1. Cells were immunostained as previously described [25]. The intensity of immunostaining of DISC1 in the nucleus versus the whole cell was quantified with Metamorph (Molecular Devices) for all experimental groups. The intensity ratio of the signal for at least 30 cells per group was analyzed in three independent experiments.

Cells were lysed in a RIPA buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM MgCl₂, 5 mM DTT, 1 mM PMSF, 1 mM EDTA, 1% Triton X-100) containing protease inhibitor mixture (Roche Applied Sciences). Supernatant fractions were obtained from lysates after centrifugation at 13,000 rpm for 15 min. Soluble fractions were used for immunoprecipitation as described previously [25].

Pregnant C57BL/6 mice at embryonic (E) day 13 were deeply anesthetized by intraperitoneal administration of 2,2,2-tribromoethanol in tert-amyl alcohol (0.4 mg/g), and intrauterine embryos were surgically manipulated as described previously [24,38]. Plasmid solutions in volumes of 1-2 μl containing RNAi plasmids (2 µg/µl) together with CAG-driven mDISC1 with mutations in the RNAi targeting site expression vector (1 µg/µl) and/or CAG-driven GFP expression vector (1 µg/µl) were injected into the lateral ventricles. To use the same amount of DNA for each condition, an empty expression vector (pCAGGS1) was used as necessary. Electronic pulses (35 V, 50 ms, 4 times) were applied using an electroporator (CUY21E, Tokiwa Science) with a forceps-type electrode (CUY650-5, Tokiwa Science).

Incorporation of BrdU was performed as previously described [25], with minor modifications. 50 mg/kg BrdU (Sigma) was injected intraperitoneally into E13 pregnant mice 48 h after electroporation. Then, 2 h after BrdU injection, brains were fixed with 4% paraformaldehyde, and 20-µm coronal sections were obtained with a cryostat (CM 1850, Leica). For BrdU immunostaining, brain slices were incubated in 2 N HCl at 37°C for 30 min before incubation with mouse anti-BrdU antibody. Images of the slices were acquired with a confocal microscope (Zeiss LSM510 and Olympus FV300).

To compare four groups, one-way ANOVA followed by the Bonferroni post hoc test for multiple comparisons was used. p values <0.05 were considered to be statistically significant.

DISC1 is found in multiple subcellular domains, and its nuclear localization and biological roles have been demonstrated [22]. This subcellular targeting may be determined in part by posttranslational modifications [25]. As a role of SUMOylation in targeting proteins to the nucleus is known [26], we hypothesized that a pool of DISC1 may be SUMOylated. We thus first predicted the SUMOylation site(s) in DISC1 by using a bioinformatics approach with the Abgent SUMOplot (http://www.abgent.com/sumoplot). By this method, DISC1 at lysine residue 643 (K643) was identified as a strong candidate SUMOylation site (fig. 1a). Since K643 is in the C-terminus of DISC1, we assayed in vitro SUMOylation with GST-tagged C-terminal DISC1 protein spanning amino acid residues 598-854 (DISC1-C). In the presence of ATP, which is crucial for SUMO-protein conjugation, DISC1-C was SUMOylated by SUMO1, SUMO2, or SUMO3 (fig. 1b). Next, to confirm whether K643 is essential for SUMOylation in DISC1-C, we performed in vitro SUMOylation assays using mutant DISC1-C where lysine residue 643 was substituted into alanine (DISC1-C-K643A). Even in the presence of ATP, DISC1-C-K643A was not SUMOylated by SUMO1 (fig. 1c). These results were consistent after normalization of the SUMOylated DISC1-C signals by the amount of the non-SUMOylated form of DISC1-C (cyan arrowhead in fig. 1c; color in online version only). Thus, these results indicate that DISC1-C is SUMOylated at lysine residue 643 in vitro.

Next, to test whether the SUMOylation of DISC1 occurs in cells, we exogenously expressed DISC1-C together with SUMO1 in cells and examined whether DISC1-C is conjugated to SUMO. DISC1-C and SUMO1 were indeed co-immunoprecipitated, and a signal for the immunoprecipitated protein complex was observed (red arrowhead in the IP panel in fig. 2a; color in online version only). Of note, a signal for SUMOylation of DISC1-C was also observed in the lysates (red arrowhead in the Input panel in fig. 2a). Then, to determine whether K643 is the SUMOylation site in DISC1-C in cells, we examined whether the mutation in which lysine residue 643 was replaced with alanine (K643A) in the C-terminal of DISC1 interfered with the SUMOylation. Immunoblotting with cells transfected with DISC1-C showed a strong signal that represents SUMOylated DISC1-C, whereas such a signal was abolished in cells expressed with DISC1-C-K643A (fig. 2b). These results suggest that the SUMOylation of DISC1-C occurs at lysine residue 643 in cells.

We next assessed whether the full-length DISC1 could also be SUMOylated in cells. A strong signal that represents SUMOylated DISC1 was detected when DISC1-Ubc9, which enhances the SUMOylation, was exogenously expressed together with SUMO1 in cells (red arrowhead). This SUMOylation on DISC1 was abolished by co-expression of SUMO1 with DISC1-Ubc9 (C93S), a dominant negative form of Ubc9 (fig. 2c).

To address the question whether K643 is the main SUMOylation site on full-length DISC1, we compared potential SUMOylation between wild-type full-length DISC1 (DISC1) and full-length DISC1-K643A. Cells were co-expressed with DISC1 or DISC1-K643A together with SUMO1 and Ubc9. DISC1 and SUMO1 were co-immunoprecipitated, and a signal for the immunoprecipitated protein complex was observed (red arrowhead in the IP panel in fig. 2d), whereas such a signal was abolished in cells expressed with DISC1-K643A. These results were consistent after normalization of the SUMOylated DISC1 signals by the amount of the non-SUMOylated form of full-length DISC1 (cyan arrowheads in fig. 2d). Of note, a signal for SUMOylation of DISC1 was also observed in the lysates (red arrowhead in the Input panel in fig. 2d). On the basis of these results, K643 is considered the main SUMOylation site on full-length DISC1.

As DISC1 is reportedly located in multiple subcellular domains [11,23,24], we tested the possible influence of SUMOylation on the subcellular distribution of DISC1. We observed decreased nuclear targeting of DISC1-K643A compared to wild-type DISC1, suggesting that SUMOylation at K643 in human DISC1 (corresponding to K640 in mDISC1) may play a role in the nuclear localization of DISC1 (online suppl. fig. 1; for all online suppl. material, see www.karger.com/doi/10.1159/000444257).

Previous studies have shown that DISC1 modulates neural progenitor cell (NPC) proliferation in vivo [25,39]. Thus, we tested how the K640 SUMOylation site in mDISC1 affected the proliferation of NPCs by in utero gene transfer combined with BrdU pulse-labeling. Consistent with our previous study [25], the proportion of cells incorporating BrdU was significantly decreased in brains with mDISC1 knockdown (mDISC1 RNAi) compared to controls, indicating that NPC proliferation was impaired. These knockdown effects were ameliorated by co-expression with wild-type mDISC1 but not by mDISC1-K640A (fig. 3), suggesting that the K640 SUMOylation site in mDISC1 is crucial for the proper proliferation of NPCs in the developing cortex.

Here, we report that DISC1 protein is SUMOylated. By combining in silico, biochemical, and cell-biological approaches, we have determined that lysine 643 in human DISC1 (which corresponds to lysine 640 in mDISC1) is a major SUMOylation site. We showed that a mutation in this residue leads to cellular deficits, including improper neural progenitor proliferation in the developing cortex.

In our biochemical analysis, the levels of SUMOylation at DISC1-K643 were relatively modest. It is known that, in general, a small pool of SUMOylated protein can have significant biological effects [40]. Therefore, the small population of SUMOylated DISC1 might have important biological effects. In cell cultures, we observed that the mutation at the lysine residue for the critical SUMOylation affected nuclear localization of DISC1. Thus far, several papers have indicated a role for nuclear DISC1 [25,39]. By examining autopsied brains from patients with neuropsychiatric conditions, Sawamura et al. [41] suggested that nuclear DISC1 in the orbitofrontal cortex may be involved in substance abuse. Furukubo-Tokunaga and coworkers [42] reported that DISC1 is involved in the regulation of sleep homeostasis, in particular in association with nuclear DISC1. More recently, Tsai and coworkers [43] have demonstrated that nuclear DISC1 is crucial for the regulation of phosphodiesterase-cyclic AMP signaling that underlies multiple psychiatric conditions. Taken together, although each study is still independent, without a cohesive or integrative story of nuclear DISC1 function as yet, assessing nuclear DISC1 will be important in studying the pathological mechanisms for mental illnesses. We hope that SUMOylation at K643 in human DISC1 may be a solid molecular probe to lead off such mechanistic studies.

In addition, we observed that the replacement of lysine with alanine at a critical SUMOylation site (mouse K640, which corresponds to human K643) disturbed neural progenitor proliferation in the developing cortex. In the experiments, we used in utero electroporation with DISC1-RNAi and expression constructs. The knockdown effect by the DISC1-RNAi construct used in the current study has been fully validated by multiple groups [11,24,25,39]. Furthermore, DISC1 knockdown-mediated aberrant neural progenitor proliferation has been shown with multiple RNAi constructs targeting different regions of DISC1 and has been rescued by co-expressing the full-length isoform of DISC1 in the present and in previous studies [25,39]. Although the significant influence of mDISC1-K640 is not a direct demonstration that SUMOylation of DISC1 is crucial for the neurodevelopmental process, we believe this is very suggestive evidence. To address this question in a more direct manner, although it is beyond the scope of the present study, we may need to develop anti-SUMOylation-specific DISC1 antibodies. Equivalent antibodies have been frequently used as tools to study protein phosphorylation [44] and have indeed been applied to a study on DISC1 in the past [25]. There is thus a successful precedence for using anti-SUMOylation-specific antibodies in biomedical studies [7,45]. Such posttranslational modification-specific antibodies may become useful probes to study the pathology of mental illness.

We thank Ms. Y. Lema for preparing the figures and organizing the manuscript, and Drs. P. Talalay and N.J. Gamo for critical reading of the manuscript. This work was supported by USPHS grants MH084018 Silvio O. Conte center (A.S.), MH094268 Silvio O. Conte center (A.S. and A.K.), MH069853 (A.S.), MH085226 (A.S.), MH088753 (A.S.), MH092443 (A.S.), MH091230 (A.K.), MH96208 (K.I.), and MH105660 (A.S. and K.I.); grants from the Stanley, RUSK, S-R, and DANA foundations (A.S.); grants from NARSAD (A.S., A.K., and K.I.); grants from MSCRF (A.S. and K.I.), and the Grand Challenge Programme grants from Pfizer (G.B.).

All animal care and use was in accordance with guidelines for the care and use of laboratory animals issued by the National Institutes of Health and The Johns Hopkins University. The authors have no ethical conflicts to disclose.

N.J.B. is a full-time employee and a shareholder at AstraZeneca. The other authors have no conflicts of interest to report.