DISC1–NDEL1/NUDEL protein interaction, an essential component for neurite outgrowth, is modulated by genetic variations of DISC1

Abstract

Disrupted-In-Schizophrenia-1 (DISC1) is a unique susceptibility gene for major mental conditions, because of the segregation of its genetic variant with hereditary psychosis in a Scottish pedigree. Genetic association studies reproducibly suggest involvement of DISC1 in both schizophrenia and bipolar disorder in several ethnic groups. The DISC1 protein is multifunctional, and a pool of DISC1 in the dynein motor complex is required for neurite outgrowth in PC12 cells as well as proper neuronal migration and dendritic arborization in the developing cerebral cortex in vivo. Here, we show that a specific interaction between DISC1 and nuclear distribution element-like (NDEL1/NUDEL) is required for neurite outgrowth in differentiating PC12 cells. Among several components of the dynein motor complex, DISC1 and NDEL1 are selectively upregulated during neurite outgrowth upon differentiation in PC12 cells. The NDEL1 binding site of DISC1 was narrowed down to a small portion of exon 13, corresponding to amino acids 802–835 of DISC1. We demonstrate that genetic variants of DISC1, proximal to the NDEL1 binding site, affect the interaction between DISC1 and NDEL1.

INTRODUCTION

Disrupted-In-Schizophrenia-1 (DISC1) is a promising risk gene for major mental conditions, including schizophrenia and bipolar disorder, in various ethnic populations (1–10). This gene is unique among several susceptibility genes for major mental conditions, in that disease-associated genetic variations have been shown to segregate with hereditary psychosis in a pedigree (11,12). In this family, disruption of this gene by a chromosomal-balanced translocation acts as a major factor for the familial psychosis with incomplete penetrance (11,12). The chromosomal abnormality occurs in the middle of the open reading frame of the DISC1 gene, potentially resulting in a C-terminal truncated form of DISC1. Functional analysis in cell models and use of patient lymphoblasts suggest that loss of DISC1 function, either by haploinsufficiency, dominant-negative effects or both underlies the pathophysiology of mental disorders in the Scottish family (5,13–16). It is also known that DISC1 has several common genetic variations that affect changes at the protein level, such as a common non-synonymous polymorphism that consists of a serine to cysteine substitution at codon 704 (Ser704Cys), and deletion of 22 amino acids from exon 11 (7,12,17–19). Functional influences of the Ser704Cys polymorphism on cognitive function have been reported (7,20). In addition, a frameshift variation that results in the loss of exon 13 from the DISC1 gene has been reported (21,22). In an American pedigree, this variation occurs in a hereditary manner and might play a role in familial psychosis, including schizophrenia and schizoaffective disorder (21). However, it remains elusive how these genetic variations impact the molecular and cellular function of DISC1.

DISC1 is multifunctional with several interacting proteins, including components of the dynein motor protein complex, such as nuclear distribution element-like (NDEL1/NUDEL) and lissencephaly 1 protein (LIS1) (5,13,16,23–29). Using overexpression and RNA interference (RNAi), our group as well as others (14,23,25) have demonstrated that DISC1 is required for regulation of microtubule dynamics, neurite outgrowth and neuronal migration in vivo.

In the present study, we show that the interaction of DISC1 and NDEL1 is required for neurite outgrowth in differentiating PC12 cells. We identified a critical domain of DISC1 for binding to NDEL1 is located in exon 13, between amino acids 802 and 835. We find that variations of DISC1 influence the interaction of DISC1 and NDEL1.

RESULTS

DISC1 and NDEL1 interaction domains

Thus far, two independent groups have proposed different domains of DISC1 to be critical for the interaction of DISC1 with NDEL1 (24,28). Morris et al. (24) have reported that NDEL1 is binding amino acids 598–696 of DISC1. In contrast, Brandon et al. (28) have concluded that the coiled-coil/leucine zipper domain which comprises amino acids 807–828 of DISC1 is important for the interaction with NDEL1. To address this question, a series of DISC1 mutants with HA-tag were exogenously expressed together with myc-tagged NDEL1 in HEK293 cells. Cell lysates were then immunoprecipitated using an antibody to HA and analyzed by western blotting (Fig. 1A). NDEL1 was co-precipitated with wild-type DISC1 (wt) and a truncated DISC1 containing amino acids 1–835 [(1–835)N]. Further C-terminal deletion of DISC1, including (1–801)N, dramatically weakened the interaction with NDEL1 in co-immunoprecipitation. Deletion of DISC1 self-interacting region (403–504 amino acids) completely abolished the interaction of exogenous DISC1 [(1–597)N-Δ(403–504)] with NDEL1. Weak association of NDEL1 and any truncated DISC1 shorter than [(1–835)N] may be via a bridge of endogenous DISC1 between NDEL1 and exogenous DISC1. Thus, we conclude that amino acids 802–835 of DISC1 are necessary for its interaction with NDEL1, which is consistent with in vitro binding previously reported by Brandon et al. (28).

Figure 1.

Protein interaction of DISC1 and NDEL1. (A) Domain mapping of DISC1 for DISC1–NDEL1 binding. A series of DISC1 mutants with HA-tag depicted in the right scheme were co-transfected with myc-tagged NDEL1 in HEK293 cells. Protein extracts were immunoprecipitated with an anti-HA antibody. Immunoprecipitates were analyzed by western blotting with an anti-myc antibody. NDEL1 was co-precipitated with wild-type DISC1 (wt) and a truncated DISC1 containing amino acids 1–835 [(1–835)N]. Further C-terminal deletion of DISC1, including (1–801)N, dramatically weakened the interaction with NDEL1 in co-immunoprecipitation. Deletion of DISC1 self-interacting region (403–504 amino acids) completely abolished the interaction of exogenous DISC1 [(1–597)N-Δ(403–504)] with NDEL1. Thus, the crucial region for DISC1–NDEL1 interaction comprises amino acids 802–835 of DISC1 (top panel). The inputs of each protein are also shown (middle and bottom panels). (B) Leucine 266 and glutamate 267 of NDEL1 are critical for DISC1–NDEL1 binding. HEK 293 cells were transfected with HA-tagged DISC1 and myc-tagged NDEL1 or NDEL1^LE266/267AA. Lysates were then immunoprecipitated with an anti-HA antibody and western-blotted with an antibody to myc to analyze the DISC1–NDEL1 interaction. Substitution of alanines for leucine 266 and glutamate 267 abolishes the NDEL1–DISC1 interaction. The inputs of each protein are shown (bottom panels). A schematic diagram of NDEL1 in the bottom scheme shows that the residues leucine 266 and glutamate 267, which are crucial for DISC1 binding, overlap the Dyn HC binding site (256–291). The LIS1 binding region (114–133) is quite distinct from the DISC1 binding region. (C) Interaction of NDEL1^LE266/267AA with Dynein and LIS1 as assessed by co-immunoprecipitation using an anti-HA antibody in HEK293 cells. NDEL1^LE266/267AA displays weaker binding with Dyn HC and Dyn IC when compared with wild-type NDEL1, whereas binding affinity to LIS1 remains unchanged (top panels). The inputs of each protein are shown (bottom panels). The graph represents densitometry analysis of western blotting (*P<0.005). Error bars indicate standard deviations.

Brandon et al. (28) also reported that residues leucine 266 and glutamate 267 of NDEL1 were important for the interaction between DISC1 and NDEL1. We observed almost no binding of NDEL1 to DISC1 when leucine 266 and glutamate 267 were each substituted with alanine residues (NDEL1^LE266/267AA) (Fig. 1B). LIS1 reportedly binds to NDEL1 and DISC1 (28,30,31). The substitutions of leucine 266 and glutamate 267 of NDEL1 did not affect the interaction of NDEL1 and LIS1. In contrast, we observed a slight decrease in the binding of NDEL1 to dynein heavy chain (Dyn HC) and dynein intermediate chain (Dyn IC), respectively (Fig. 1C).

Importance of the DISC1 and NDEL1 interaction for neurite outgrowth in PC12 cells

Although DISC1 is known to mediate neurite outgrowth in differentiating PC12 cells (14,23,25), it remains elusive how other proteins of the dynein motor complex contribute to this process. To address this question, we examined protein levels of components in the dynein motor complex, including DISC1, NDEL1, LIS1, Dyn IC and a subunit of dynactin (p150^glued) treated daily with nerve growth factor (NGF). Of interest, only DISC1 and NDEL1 are selectively and dramatically increased during neurite outgrowth upon differentiation with NGF, whereas expression of other components of the dynein motor complex remain unchanged (Fig. 2A). Thus, DISC1 and NDEL1 may have a unique regulatory role in neurite outgrowth.

Figure 2.

DISC1 and NDEL1 in neurite outgrowth of differentiating PC12 cells. (A) DISC1 and NDEL1 are selectively upregulated during differentiation of PC12 cells by adding NGF. In contrast, no changes were observed in the expression of β-tubulin or other components of dynein motor complex such as Dyn IC, LIS1 and dynactin (p150^glued). Two independent anti-DISC1 antibodies (D27 and mExon3) consistently detected the increase in DISC1 protein [100 kDa isoform(s)]. The graph represents densitometry analysis of western blotting in the time course. (B) Influence of plasmids expressing interfering shRNA (NDEL1 RNAi-1,-2), on expression of exogenous (exo) and endogenous (endo) rat NDEL1 protein. NDEL1 RNAi-1 and -2 suppress 90 and 70% of exogenous rat NDEL1 expression, respectively (top panels). NDEL1 RNAi-1 and -2 also suppress endogenous rat NDEL1 expression in differentiated PC12 cells but control RNAi has no effect (bottom panels). IB: antibodies used for western blotting. (C) Inhibition of neurite outgrowth in differentiated PC12 cells is observed following transfection of both RNAi-1 and RNAi-2 (*P<0.001). RNAi-1 has a stronger effect than RNAi-2. Co-expression with human NDEL1 partially rescues the suppression of neurite outgrowth by RNAi-1 (**P<0.005). In contrast, co-expression of NDEL1^LE266/267AA with RNAi-1 does not rescue suppression of neurite outgrowth at all. The length of the longest extension in each cell was measured after 3 days of differentiation. Bars represent averages of 150 cells for each group in three independent experiments. Error bars represent standard deviations. Representative images of PC12 cells are shown. Blue, nucleus; Red, β-tubulin; Green, GFP. Scale bar, 50 µm.

We previously reported that either suppression of DISC1 protein expression by RNAi, or overexpression of a dominant-negative C-terminal truncated DISC1, inhibited neurite outgrowth in differentiated PC12 cells (14,23). Another study also reported that stable expression of DISC1-induced enhanced neurite extension in neuronal PC12 cells (25). Based on these results as well as reported observation that DISC1 and NDEL1 interact in cells (23,24,28), we hypothesize that NDEL1 might also play a role in neurite outgrowth. Therefore, we used two independent short-hairpin RNAs (shRNAs) to knockdown NDEL1 protein expression and assayed neurite outgrowth. Knockdown of NDEL1 expression consistently results in dramatic inhibition of neurite outgrowth (Fig. 2B and C). Co-expression of wild-type human NDEL1 can partially rescue neurite outgrowth. In contrast, mutant human NDEL1 deficient in binding to DISC1 (NDEL1^LE266/267AA) has no effect, suggesting the DISC1–NDEL1 protein interaction is required for neurite outgrowth.

Inhibition of the DISC1 and NDEL1 interaction disturbs neurite outgrowth, through redistribution of NDEL1

To confirm the requirement of DISC1–NDEL1 protein interaction in neurite outgrowth, we expressed a short protein fragment that closely corresponds to the minimal binding domain of DISC1 to NDEL1. We used a myc-tagged short fragment including amino acids 802–835 of DISC1 [DISC1(788–849)] in the present study, because of instability of any shorter fragment including amino acids 802–835 of DISC1. Expression of DISC1 (788–849) fragment, but not unrelated fragment at the similar size, almost completely blocks the protein interaction of DISC1 and NDEL1 in differentiated PC12 cells, as demonstrated in co-immunoprecipitation (Fig. 3A). In differentiated PC12 cells, endogenous DISC1 and NDEL1 are co-localized in the perinuclear region, including the centrosome (Fig. 3B). Overexpression of DISC1 (788–849) fragment redistributes both NDEL1 and DISC1 from the perinuclear region (Fig. 3B). These results suggest that DISC1 (788–849) fragment competes endogenous DISC1 for binding to NDEL1, possibly blocking a role of DISC1 to tag NDEL1 at the centrosome. Furthermore, NDEL1 at the centrosome may also contribute to the proper localization of DISC1. Disturbance of DISC1–NDEL1 interaction by DISC1 (788–849) fragment leads to an inhibition of neurite outgrowth (Fig. 3C).

Figure 3.

Requirement of the DISC1–NDEL1 protein interaction in neurite outgrowth of PC12 cells. (A) The short fragment of DISC1 containing the DISC1–NDEL1 interacting region [DISC1(788–849)], but a control fragment (see the information in the Materials and Methods section), reduces the DISC1–NDEL1 interaction. The interaction of myc-tagged DISC1 with endogenous NDEL1 is reduced when HA-tagged DISC1 (788–849) is co-expressed, as shown by co-immunoprecipitation with an anti-myc antibody in HEK293 cells (top panel). The inputs of each immunoprecipitation are shown (bottom panels). (B) Endogenous DISC1 (red) and NDEL1 (green) are co-localized in the perinuclear region, including the centrosome, in differentiated PC12 cells with mock transfection (arrowheads in left panels). Overexpression of DISC1 (788–849) (green) redistributes endogenous DISC1 (red) and NDEL1 (red) from the perinuclear regions (arrows in middle and right panels). Scale bar, 10 µm. (C) Neurite outgrowth is inhibited by overexpression of DISC1 (788–849) in differentiated PC12 cells when compared with cells with transfection of mock or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (*P<0.001). The length of the longest extension in each cell was measured 4 days after differentiation. Bars represent averages of 150 cells for each group in three independent experiments. Error bars represent standard deviations. Representative images of PC12 cells with mock, GAPDH (blue) or DISC1 (788–849) (blue) are shown (right panels). Red, β-tubulin; Green, GFP. Scale bar, 30 µm.

Significant influence of exon 11 DISC1 splice variants on the DISC1 and NDEL1 interaction

DISC1 has two major splice variants, referred to as L (DISC1) and Lv (DISC1^Δ22aa) in humans (17). DISC1^Δ22aa utilizes an alternative splice site in exon 11, which results in the deletion of 22 amino acids (748–769) (Fig. 4A). Both isoforms are substantially expressed in human, primate and rodent neuronal and peripheral tissues (17,19). Because the 22 amino acid change occurs adjacent to the binding domain for NDEL1, we explored how the protein interaction between DISC1 and NDEL1 might be affected.

Figure 4.

Influence of the exon 11 DISC1 splice variation on the interaction of DISC1 with NDEL1. (A) A schematic diagram of DISC1 and DISC1^Δ22aa. DISC1^Δ22aa, a common genetic variation of DISC1, utilizes an alternative splice site in exon 11. Consequently, DISC1^Δ22aa lacks amino acids 748–770 of DISC1, which is adjacent to the NDEL1 binding site (802–835). (B) The NDEL1 binding affinity of DISC1 is higher than that of DISC1^Δ22aa in a quantitative yeast two-hybrid assay (*P<0.005, **P<0.05). Bars represent averages of β-Gal unit in three independent experiments. Error bars represent standard deviations. (C) HEK 293 cells were transfected with HA-tagged DISC1 or DISC1^Δ22aa and myc-tagged NDEL1. Lysates were then immunoprecipitated with an anti-HA antibody and western-blotted with an antibody to myc to analyze the DISC1–NDEL1 interaction. DISC1^Δ22aa displays diminished binding to NDEL1 when compared with DISC1 as shown by co-immunopecipitation with an anti-HA antibody (top panel) (*P<0.005). The inputs of each protein are also shown (bottom panels).

In yeast two-hybrid assays, the interaction between a DISC1 fragment (597–854 amino acids) and a NDEL1 fragment (201–280 amino acids) is significantly diminished with the deletion of the 22 amino acids (Fig. 4B). The combination of the DISC1^Δ22aa fragment (597–854 amino acids) and full-length NDEL1 in a subsequent yeast two-hybrid assay reveals a similar decrease in binding affinity. To confirm these results by co-immunoprecipitation, DISC1 or DISC1^Δ22aa were co-expressed with NDEL1 in HEK293 cells. Indeed, binding of DISC1^Δ22aa to NDEL1 is weaker than that of DISC1 (Fig. 4C). Taken together, we conclude that the 22 amino acid region in exon 11 greatly influences the DISC1 and NDEL1 protein interaction.

Ser704cys variation exerts a mild influence on the DISC1 and NDEL1 interaction

We addressed whether the Ser704Cys variation of DISC1 (resulting in the proteins DISC1^Ser and DISC1^Cys, respectively) might influence the protein interaction of DISC1 and NDEL1, using the same strategies as above (Fig. 5A). In yeast two-hybrid assays, the DISC1^Cys fragment (597–854 amino acids) displays a slightly stronger interaction with NDEL1 when compared with that of the DISC1^Ser fragment, although the differences remain in a suggestive range (Fig. 5B). A similar tendency was observed in co-immunoprecipitation, where the binding of NDEL1 to DISC1 was slightly increased for DISC1^Cys as compared with DISC1^Ser (Fig. 5C). Thus, the Ser704Cys genetic variation might influence the DISC1–NDEL1 protein interaction, but the effect is very mild.

Figure 5.

Influence of the Ser704Cys variation of DISC1 on the DISC1–NDEL1 protein interaction. (A) A schematic diagram of DISC1^Ser and DISC1^Cys, which is the Ser704Cys variation of DISC1. (B) The NDEL1 binding affinity of DISC1^Cys appears to be slightly higher than that of DISC1^Ser in a quantitative yeast two-hybrid assay, but is not statistically significant. Bars represent averages of β-Gal unit in three independent experiments. Error bars represent standard deviations. (C) HEK 293 cells were transfected with HA-tagged DISC1^Ser or DISC1^Cys and myc-tagged NDEL1. Lysates were then immunoprecipitated with an anti-HA antibody and western-blotted with an antibody to myc to analyze the DISC1–NDEL1 interaction. DISC1^Cys displays slightly stronger binding with NDEL1 when compared with DISC1^Ser as shown by co-immunoprecipitation with an anti-HA antibody (top panel). The inputs of each protein are also shown (bottom panels). (D) Influence of the Ser704Cys variation of DISC1 on the cognitive function of patients with schizophrenia and bipolar disorder. Statistical analysis of the SNP rs821616 genotype versus measures of cognitive functioning obtained by RBANS within schizophrenia and bipolar disorder subjects. A and T correspond to alanine and thymine, which are the first codons of serine (AGC) and cysteine (TGC) residues. A/A and T/T represent individuals with homozygous serine (Ser) allele and cysteine (Cys) allele, respectively. A/T represent individuals with heterozygous allele. Among the five indices of quantitative assessment, individuals with homozygous Cys allele (T/T) performed slightly lower than individuals with homozygous Ser allele (A/A) or heterozygous allele (A/T) in the delayed memory task (P=0.078). When considering the total RBANS score, performance was also lower in patients homozygous for the minor allele (P=0.088).

As described earlier, the Ser704Cys variation has been suggested to associate with cognitive dysfunction (7,20). We explored a possible association of the Ser704Cys variation with cognitive function using the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) (32). We observed a mild association between the delayed memory score of RBANS and the Cys 704 variant (genotype T/T) (P=0.078) within a group that contained schizophrenia and bipolar disorder subjects (Fig. 5D). Across all diagnoses, subjects homozygous for the Cys variant (genotype T/T) had slightly reduced the total score of RBANS (P=0.088) as compared with subjects homozygous for Ser variant (genotype A/A) subjects and heterozygous (genotype A/T) subjects (Fig. 5D).

DISCUSSION

This study presents two major findings. First, we demonstrate that the DISC1–NDEL1 protein interaction is critical for neurite outgrowth. Among several components of the dynein motor complex, DISC1 and NDEL1 are selectively upregulated during neurite outgrowth upon differentiation, and interference of their interaction suppresses outgrowth. Secondly, we show that genetic variations of DISC1 near the NDEL1 binding site affect the interaction between DISC1 and NDEL1.

Callicott et al. (7) have reported that a common, non-synonymous polymorphism (Ser704Cys) is associated with schizophrenia, where the Ser allele confers an increased risk of schizophrenia. Furthermore, subjects that are homozygous for the Ser allele display deficits along four axes: cognitive impairment as assessed by lower Wisconsin Card Sorting Test category scores, failure of appropriate modulation of hippocampal function during the N-back working memory task, lower N-acetylaspartate value and smaller hippocampal volume (7). In contrast, the Cys allele of this polymorphism was reportedly associated with impaired cognitive ability as tested by Moray House Test scores in aged women (20). Hashimoto et al. (33) has recently reported that the Cys allele is associated with an increased risk for major depressive disorder and is also associated with a reduced gray matter volume in the cingulate cortex of healthy subjects. Here we observe that DISC1^Cys has a slightly stronger binding affinity to NDEL1 compared with DISC1^Ser. Subjects who are homozygous for the Cys allele display poorer executions than those who are heterozygous or homozygous for the Ser allele. This genetic data is similar to those reported by Thomson et al. (20), especially in that the Cys allele looks recessive. Exact mechanisms whereby this polymorphism displays these complex phenotypic impacts are still unclear. Conceivably, this polymorphism may affect more than one cellular cascade, such as this NDEL1 pathway and ERK signaling (33). There could be changes in binding to other partners as well and NDEL1 may not be the only player in the pathology. The difference in NDEL1 binding affinity of the Ser704Cys variation of DISC1 is not particularly robust. Nevertheless, such subtle but consistent changes may increase a risk of major mental disorders by providing a chronic disturbance of DISC1–NDEL1 function. Alternatively, we do not exclude the possibility that this Ser704Cys variation is a mere genetic marker for unknown functional SNPs adjacent to this variation.

In the present study, we also focus on a deletion/insertion variation in exon 11, because the region is located adjacent to the NDEL1 binding site. DISC1 binds to NDEL1 stronger than DISC1^Δ22aa does in both yeast two-hybrid assays and co-immunoprecipitation. Because this variation occurs at the mRNA level but not at the genomic level, no study has ever addressed whether or not some alteration might occur in this variation in association with major mental conditions. In normal subjects of several species, including humans, primates and rodents, both isoforms are expressed (17–19). In our experiments, we observed an approximate 1:3 expression of DISC1 to DISC1^Δ22aa in primate brains (19). A minor disturbance in the ratio of insertion to deletion allele might lead to subtle but significant changes in the DISC1 and NDEL1 interaction which should be relevant to the pathology of major mental conditions. We predict that several intronic polymorphisms adjacent to exon 11 may affect the alternative use of this splice site, and alter the ratio of DISC1 to DISC1^Δ22aa. Thus, it may be worthwhile to compare the ratio of DISC1 to DISC1^Δ22aa between controls and patient populations in relation to haplotypes associated with schizophrenia and bipolar disorder (7). Further studies may include a systematic study of polymorphisms proximal to exon 11, in a large number of subjects.

There were two contradictory reports for minimal binding region of DISC1 for NDEL1 (24,28). Morris et al. (24) used four kinds of DISC1 constructs [DISC1 (amino acids 1–854), DISC1 (amino acids 1–597), DISC1 (amino acids 293–854) and DISC1 (amino acids 697–854)] for co-immunoprecipitation, and conclude that NDEL1 is binding amino acids 598–696 of DISC1. In contrast, Brandon et al. (28) initially performed co-immunoprecipitation with DISC1 (amino acids 1–854) or DISC1 (amino acids 1–598), and narrowed down the binding domain of DISC1 for NDEL1 into the C-terminal region (amino acids 599–854). Following the co-immunoprecipitation, they used in vitro binding assay and concluded that the importance of the coiled coil/leucine zipper in the C-terminus of DISC1 (amino acids 807–828) for the interaction of NDEL1. Our data supports the conclusion by Brandon et al., indicating that amino acids 802–835 of DISC1 is necessary and crucial for binding to NDEL1. However, considering the above experimental designs in comparison, we do not exclude the possibility that amino acids 598–696 of DISC1 may have an influence on the protein interaction.

The role of DISC1 in neurite outgrowth and neuronal migration has also been demonstrated (14). Using the DISC1 (788–849) fragment as well as a mutant form of NDEL1 (NDEL1^LE266/267AA), we now demonstrate that the DISC1–NDEL1 interaction is required for neurite outgrowth. Expression of the blocking peptide against this protein interaction alters endogenous localization of both DISC1 and NDEL1 proteins at the centrosome. This result may imply that DISC1 and NDEL1 work together in a bidirectional manner and act as adaptor-like molecules to locate their binding partners to the proper subcellular domains, such as the centrosome. NDEL1 has been reported to associate with cytoskeletal components, playing a role in cytoskeletal stabilization, cell mitosis, membrane trafficking and neuronal migration (30,31,34–42). Because DISC1 interacts with NDEL1 in a bidirectional manner, DISC1 may also contribute to these biological events. NDEL1 is also reported as endo-oligopeptidase A (EOPA) that hydrolyzes numerous bioactive peptides (43,44). Since the catalytic center (cysteine-273) is close to the critical residues for DISC1 interaction, it is possible that disturbance of this enzymatic activity may underlie the inhibition of neurite outgrowth observed in PC12 cells following expression of the blocking peptide or NDEL1^LE266/267AA.

In this study, we present the significance of the DISC1–NDEL1 interaction in neurite outgrowth in PC12 cells. Future studies will assess the significance of the DISC1 and NDEL1 interaction in vivo.

MATERIALS AND METHODS

Plasmids and antibodies

All the deletion and point-mutated DISC1 and NDEL1 expression constructs were made by PCR-based mutagenesis protocol (45). Rat monoclonal anti-NDEL1 antibody were prepared as described (46). The rabbit polyclonal anti-DISC1 antibody (D27) was a gift from Dr. Nicholas. J. Brandon (Merck Sharp and Dohme, UK). The anti-mouse DISC1 antibody (mExon3) was raised in rabbits against amino acids 360–374 of mouse DISC1 and affinity-purified. The following antibodies were also used: mouse monoclonal anti-p150^glued antibody (Transduction Laboratories, San Jose, CA, USA); mouse monoclonal antibodies against β-tubulin, LIS1 and Dyn IC (Sigma–Aldrich, St Louis, MO, USA); rabbit polyclonal anti-Dyn HC (Santa Cruz, Santa Cruz, CA, USA); mouse monoclonal antibodies against HA-tag and myc-tag (BAbCO, Berkeley, CA, USA). An affinity-purified rabbit antiserum against GFP (Molecular Probes, Eugene, OR, USA) and a mouse monoclonal antibody against GFP (Molecular Probes) were used to measure the length of neurites of GFP-transfected PC12 cells in the neurite outgrowth assay. Plasmids expressing interfering shRNA (47) were generated to suppress endogenous NDEL1 protein expression. We produced six shRNA plasmids and selected two representative ones. The sequences of these shRNA plasmids are as follows: These shRNAs target sequences are common in mouse and rat NDEL1. The effect of shRNA was initially tested by the extent of suppression in co-transfected rat NDEL1 in HEK293 cells. Once suppression against exogenous NDEL1 was confirmed, selected shRNAs were further tested against endogenous NDEL1 in rat PC12 cells.

• RNAi-1 with strong suppression, 5′-GGAGAAACTAGAGCA TCAG-3′.

• RNAi-2 with milder suppression, 5′-CAAAGAAATAGAGA CCTGC-3′.

Human NDEL1 expression construct was used to test functional complementation in the neurite outgrowth assay in PC12 cells with RNAi-1 to rat NDEL1, after it was confirmed that RNAi-1 has no effects on human NDEL1. Human NDEL1 sequences have two mismatches to rodent NDEL1 as follows: 5′-GGAGAAGCTAGAGCATCAA-3′ (bold denotes mismatched nucleotides).

An unrelated fragment to DISC1 (GenBank accession no. XM_222150) was used for a control to DISC1 (788–849) fragment.

Cell culture and fluorescence staining

HEK293 cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) with 10% fetal bovine serum (FBS). PC12 cells were maintained in DMEM with 10% FBS and 5% horse serum (HS). Differentiation was initiated by adding 50 ng/ml of NGF with culture medium changed to DMEM with 1% FBS and 1% HS. NGF was supplemented daily after differentiation.

Transfection of expression constructs or RNAi constructs was carried out with Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) for PC12 cells and with PolyFect Transfection Reagent (Qiagen, Valencia, CA, USA) for HEK293 cells.

Cells were fixed with 3.7% paraformaldehyde in PBS, and permeabilized with 0.1% Triton X-100. For some staining, ice-cold methanol at −20°C was used as fixative. After blocking with 1% BSA and 0.5% normal goat serum in PBS, cells were treated with primary antibodies for 1 h followed by secondary antibodies conjugated to AMCA, Rhodamine Red-X or Cy2 (Jackson ImmunoResearch, West Grove, PA, USA) for 1 h. Hoechst 33258 (Molecular Probes) was used at 1:500 dilution for 3 min to visualize nuclei.

Neurite outgrowth assay

Neurite outgrowth in PC12 cells was assayed as described (23). In this study, we added a minor modification: to obtain clearer images of cell morphology, cells were co-transfected with Mock, GAPDH, DISC1 (788–849) or RNAi constructs together with the GFP construct. Cells were stained with an anti-GFP antibody. The length of the longest process of each neurite-harboring cell stained green was measured using Image J (http://rsb.info.nih.gov/ij/). Confocal microscopy (Zeiss LSM 510 Meta, Grottingen, Germany) was used for epifluorescent image collection. A Zeiss Axiovert 135 microscope mounted with a charge-coupled device camera (Roper Scientific CoolSnap HQ cooled 12 bit, Roper Scientific, Trenton, NJ, USA) was used to obtain PC12 cell images in the neurite outgrowth assay. Cell morphology of 150 cells per group was analyzed in three independent experiments in a blinded manner. Statistical analyses were conducted using a one-way ANOVA followed by post-hoc testing.

Co-immunoprecipitation and cell extraction

Immunoprecipitation: cells were lysed in a RIPA buffer [50 mm HEPES, pH 7.4, 150 mm NaCl, 5 mm MgCl₂, 5 mm dithiothreitol (DTT), 1 mm phenylmethane sulfonylfluoride (PMSF), 1 mm EDTA, 1% Triton X-100, protease inhibitor mixture (Roche, Basel, Switzerland)]. Supernatant fractions obtained after centrifugation at 12 000g for 15 min were incubated with primary antibodies and protein G Plus/Protein A agarose (Calbiochem, Darmstadt, Germany). The immunoprecipitates were analyzed with SDS–PAGE followed by western blotting after extensive washing. ProFound™ Mammalian HA Tag IP/Co-IP Kit and ProFound™ c-Myc Tag IP/Co-IP Kit (Pierce, Rockford, IL, USA) were also used for immunoprecipitation. Endogenous NDEL1 binding to exogenous DISC1^Δ22aa, DISC1^Cys and DISC1^Ser was analyzed by densitometry.

Cell extraction: cells were homogenized in ice-cold lysis buffer [50 mm Tris–HCl, pH 7.4, 150 mm NaCl, 5 mm 1,4-DTT, 1 mm PMSF, 1% NP-40, 0.1% SDS and a protease inhibitor mixture]. Cells extracted were mixed with SDS–PAGE loading buffer after measured protein concentrations and analyzed with SDS–PAGE followed by western blotting.

Densitometry analysis and statistical analysis

Quantitative densitometric measurement of western blotting was performed using EagleSight program (Stratagene, La Jolla, CA, USA). Statistical analyses were conducted using a one-way ANOVA followed by post-hoc testing. Values depicted are mean±standard deviations.

Quantitative yeast two-hybrid assay

Quantitative yeast two-hybrid assay for β-galactosidase activity in yeast cultures were performed according to ProQuest™ Two-Hybrid System protocol (Invitrogen). Chlorophenol red-β-d-galactosidase (CPRG) was used as a substrate. Statistical analyses were conducted using a one-way ANOVA followed by post-hoc testing.

Cognitive measurement (RBANS)

A total of 272 subjects meeting diagnostic criteria for schizophrenia (n=104), bipolar disorder (n=98) or no mental illness (n=70) were recruited from the Sheppard Pratt Mental Health System. The RBANS scores was used to measure cognitive functioning in all subjects (32). Five index scores that assessed immediate memory, visuo-spatial function, language, attention and delayed memory were obtained for each subject through clinical interviews as described (48,49). Statistical analyses were performed using Stata 9™. Statistical analyses were conducted using a one-way ANOVA followed by post-hoc testing.

SNP genotyping

A total of 272 subjects meeting diagnostic criteria for schizophrenia (n=104), bipolar disorder (n=98) or no mental illness (n=70) were recruited from the Sheppard Pratt Mental Health System. Genomic DNA of 20 ng was extracted from the blood serum of each subject. Taqman Assay-on-Demand (Applied Biosystems, Foster City, CA, USA) was used to genotype all samples at SNP locus rs821616, responsible for the amino acid change from serine to cysteine at residue 704 of human DISC1. ABI Prism^® 7900 (Applied Biosystems) was used for the detection of fluorescent probes, and data was subsequently obtained using SDS 2.1 software (Applied Biosystems) for allelic determination.

ACKNOWLEDGEMENTS

We thank Y. Lema for preparation of the figures. We thank C. Engelhard for critical reading of the manuscript. This work was supported by grants from U.S. Public Health Service Grant MH-69853 (A.S.) as well as foundation grants from Stanley, NARSAD and S-R (A.S.).

Conflict of Interest statement. None declared.