DISC1 and PDE4B Are Interacting Genetic Factors in Schizophrenia That Regulate cAMP Signaling

See allHide authors and affiliations

Science  18 Nov 2005:
Vol. 310, Issue 5751, pp. 1187-1191
DOI: 10.1126/science.1112915


The disrupted in schizophrenia 1 (DISC1) gene is a candidate susceptibility factor for schizophrenia, but its mechanistic role in the disorder is unknown. Here we report that the gene encoding phosphodiesterase 4B (PDE4B) is disrupted by a balanced translocation in a subject diagnosed with schizophrenia and a relative with chronic psychiatric illness. The PDEs inactivate adenosine 3′,5′-monophosphate (cAMP), a second messenger implicated in learning, memory, and mood. We show that DISC1 interacts with the UCR2 domain of PDE4B and that elevation of cellular cAMP leads to dissociation of PDE4B from DISC1 and an increase in PDE4B activity. We propose a mechanistic model whereby DISC1 sequesters PDE4B in resting cells and releases it in an activated state in response to elevated cAMP.

Schizophrenia and bipolar affective disorder are common, debilitating conditions that are determined in part by genetic factors. The DISC1 gene (for disrupted in schizophrenia 1) is a candidate susceptibility factor for psychiatric illness with both genetic and biological plausibility (1). DISC1 is disrupted by a balanced chromosomal translocation t(1;11)(q42;q14) cosegregating with schizophrenia (7 cases), bipolar affective disorder (1 case), and related affective disorders (10 cases) in a large Scottish family [maximum logarithm of the odds ratio for linkage (lod score) = 7.1] (2, 3). Translocation carriers, even without a major psychiatric diagnosis, have reduced amplitudes of the P300 event-related potential, consistent with an underlying cognitive defect (3). A number of independent genetic linkage and association studies provide confirmatory evidence for involvement of the DISC1 locus in schizophrenia, schizoaffective disorder, and bipolar affective disorder (14). DISC1 is expressed in regions of the brain implicated in the genesis of psychiatric symptoms (57) and binds diverse proteins in the central nervous system, including the neurodevelopmental proteins FEZ1 (8) and NUDEL (9, 10).

In ongoing studies of psychiatric patients with chromosomal abnormalities, we identified a proband with schizophrenia who carried a balanced t(1;16)(p31.2;q21) translocation. The subject had a history of repeated psychotic episodes with auditory hallucinations and delusions and was treated with standard anti-psychotic medications. A cousin of the proband who also carried the translocation had a psychotic illness with prolonged hospital admission but was not available for interview (11). Using fluorescence in situ hybridization (FISH) (11), we showed that the translocation breakpoint on chromosome 16 lies within an intron of the Cadherin 8 (CDH8) gene (fig. S1) and that the 1p31.2 translocation breakpoint disrupts the B1 isoform of the phosphodiesterase 4B (PDE4B) gene (Fig. 1, A to C). CDH8 is a cell adhesion molecule that potentially has a neuronal function (11), but we focused on PDE4B for the following reasons: (i) PDE action is the sole means of inactivating intracellular adenosine 3′,5′-monophosphate (cAMP), a key second messenger involved in learning, memory, and mood (1214). (ii) Mutations in the fruit fly Drosophila dunce gene, which encodes an ortholog of mammalian PDE4, cause learning and memory deficits (13). (iii) Mice deficient in PDE4D behave as if they are taking antidepressants (15). (iv) Rolipram, an antidepressant, is a selective inhibitor of PDE4 (16). And (v), most intriguingly, we found PDE4B (but not CDH8) to interact robustly with DISC1 in a large-scale yeast two-hybrid screen using full-length DISC1 as bait (9).

Fig. 1.

Structure and expression of the PDE4B gene in psychiatric patients. (A) Genomic organization of PDE4B, showing positions of exons (vertical lines), alternative transcript forms, a scale bar, and bacterial artificial chromosomes (BACs; closed boxes) used as FISH probes to define the translocation breakpoint (dashed vertical line). Only the PDE4B1 form is directly disrupted. (B and C) FISH with probes synthesized from BACs RP11-433N2 and RP11-442I11, respectively, on t(1;16) patient metaphase spreads. Arrows mark the normal and derived chromosomes. (D) PDE4B1 expression in lymphoblastoid cells, with (T) or without (N) the t(1;16), and with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) loading controls. (E) DISC1 expression in lymphoblastoid cells, with (T) or without (N) the t(1;11), and with GAPDH loading controls. Markers are indicated in kilodaltons.

To determine the effect of the t(1;16) and t(1;11) translocations on PDE4B and DISC1 expression respectively, we analyzed patient-derived lymphoblastoid cell lines. Cell lines derived from family members with or without the t(1;16) translocation revealed that PDE4B1 protein expression was reduced by ∼50% in the presence of the t(1;16) translocation (Fig. 1D; fig. S2). In the case of the t(1;11) translocation, we demonstrated transcription of DISC1 from the derived chromosome 1 (fig. S3A). However, DISC1 transcript levels are reduced overall, and when we used antibody R47, which is specific for an epitope within the N terminus of DISC1 (7), we found no evidence [fig. S3, B to D, and (11)] for a putative C-terminally truncated protein (8, 9). Expression of all DISC1 species was consistently reduced to about half normal levels in each of five t(1;11) cell lines from different individuals (Fig. 1E; fig. S2). These are the same DISC1 species that are detectable in human brain tissue (7), which indicates that proportionate reduction of DISC1 expression in brain is the most likely consequence of inheriting the translocation. The reduction in DISC1 is specific, as we see no change in the relative level of TRAX, which maps immediately 5′ of DISC1 (fig. S4, A and B). Haploinsufficiency for DISC1 in the t(1;11) translocation cases and for PDE4B in the t(1;16) translocation cases is therefore the most likely mechanistic explanation for susceptibility to schizophrenia in these individuals.

PDE4 isoforms are classified as long, short, or supershort (17), depending on the presence of conserved regulatory regions (UCR1 and UCR2) linked to the catalytic unit (Fig. 2A). Long isoforms uniquely express UCR1, whose phosphorylation by protein kinase A (PKA) triggers a conformational change in the UCR1 and 2 module, leading to increased catalytic activity (17). To determine which of the many known PDE4 isoforms could interact with DISC1, we transfected expression constructs into HEK293 cells (11). We analyzed the long PDE4B1 and PDE4B3 isoforms, which each contain UCR1, as well as the short PDE4B2 isoform, which does not. All three of these human isoforms coimmunoprecipitate with DISC1 (Fig. 2B), which shows that binding is not specific to one particular isoform. Indeed, DISC1 binds representative long isoforms from all four PDE4 genes (fig. S5). Removal of UCR2 abolished PDE4B1 binding to DISC1 (Fig. 2C; fig. S6). Moreover, the UCR2 domain alone was capable of binding DISC1 (Fig. 2D), which suggests that UCR2 is a specific DISC1 interaction domain.

Fig. 2.

Human DISC1 binds the PDE4B UCR2 domain in transfected cells. (A) Domain organization of type 4 phosphodiesterases. The common structure of core regions forming the UCR1/UCR2 module and the catalytic domain are indicated. Each PDE4 isoform contains a unique N terminus. (B) HEK293 cells were cotransfected with DISC1 and PDE4 expression vectors. FLAG epitope–tagged immunoprecipitates from lysates prepared from HEK293 cells cotransfected with either empty vector or FLAG-DISC1 and PDE4B1-B3 plasmids, probed by Western blot analysis with antibodies against FLAG and pan-PDE4B. Markers are indicated in kilodaltons. (C) FLAG-tagged immunoprecipitates from HEK293 cells cotransfected with DISC1 and the indicated FLAG-PDE4B1 constructs (top) probed by Western analysis with the R47 antibody to detect total DISC1 and PDE4B1-associated DISC1. For a schematic representation of PDE4B1 constructs and expression of the corresponding FLAG-tagged PDE4B1 proteins, see fig. S6. (D) Glutathione S-transferase (GST) pull-downs from lysates of HEK293 cells cotransfected with empty FLAG vector or FLAG-DISC1 and GST or GST-UCR2 plasmids, probed with antibody against FLAG to detect total DISC1 and DISC1 that specifically interacts with the UCR2 domain. (E) Mapping the PDE4B binding site on DISC1. 35S-labeled NΔ DISC1 fragments (depicted in fig. S8) and FLAG-tagged, unlabeled PDE4B3 were synthesized by in vitro transcription-translation, mixed, and incubated for 3 hours in a binding reaction before immunoprecipitation with antibody against FLAG. Input DISC1 and PDE4B3-bound DISC1 were detected by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) followed by autoradiography. Equal capture of FLAG-tagged PDE4B3 was determined by Western analysis with antibody against FLAG (lower). Input signal represents 1/50 of total binding reaction. Markers are indicated in kilodaltons.

To identify the site of PDE4 binding on DISC1, we performed binding assays after expression of DISC1 and PDE4B by in vitro transcription-translation. Full-length DISC1 (1 to 854), a C-terminal truncated DISC1 (1 to 597), and the N-terminal “head” domain of DISC1 (1 to 358) were tested for their ability to bind 35S-labeled PDE4B3 in vitro. All three V5-tagged DISC1 protein fragments efficiently coimmunoprecipitated PDE4B3 (fig. S7), which indicated that PDE4B binding to DISC1 is direct and that the DISC1 N-terminal domain is required for the interaction. We next generated a series of constructs that progressively delete amino acids from the N terminus of DISC1 (Fig. 2E; fig. S8) and showed that amino acids 219 to 283 are required for binding to PDE4B. DISC1 binding was restored by amino acids 180 to 250 (fig. S9).

To determine whether and where endogenous DISC1 and PDE4B interact, we performed coimmunoprecipitation and colocalization studies using conventional and confocal microscopy in the human neuroblastoma–derived cell lines SH-SY5Y and LAN5 and in primary rat hippocampal cells. DISC1 antiserum (R47) coprecipitates PDE4B1 from SH-SY5Y extracts (Fig. 3A, top), whereas pan-PDE4B, an antibody that recognizes all PDE4B isoforms, specifically coimmunoprecipitates the 71-kD DISC1 isoform (Fig. 3A, bottom). Thus, isoform PDE4B1 associates with the 71-kD isoform of DISC1. In cell fractionation studies, the PDE4B1 isoform was most abundant in the mitochondria-enriched fraction P2 (Fig. 3B), as was the 71-kD isoform of DISC1, which is also predominantly mitochondrial (7). PDE4B partially colocalizes with DISC1 in the mitochondria of SH-SY5Y (Fig. 3, C and D; fig. S10 and S11) and LAN5 (fig. S10) cells. In rat hippocampal primary cultures, DISC1 and PDE4B expression overlap substantially in both neurons and proliferating nonneuronal cells (Fig. 3E; fig. S10 and S11).

Fig. 3.

Interaction of endogenous DISC1 and PDE4B in human SH-SY5Y cells. (A) Control immunoglobulin IgG and DISC1 immunoprecipitates (R47) from lysates of SH-SY5Y cell were probed by Western blot analysis to detect DISC1 (R47) and PDE4B1 (top), and PDE4B immunoprecipitates (pan-PDE4B) from SH-SY5Y cell extracts were probed for PDE4B (pan-PDE4B) and DISC1 (bottom). The DISC1 71-kD isoform is indicated with an arrow. (B) Protein extracts from SH-SY5Y subcellular fractions were probed by Western analysis with antibodies against DISC1, PDE4B1, and cytochrome c (as labeled). P1, nuclear fraction plus whole-cell debris; P2, predominantly mitochondrial; P3, other membrane elements; C, cytosolic. The DISC1 71-kD isoform is indicated with an arrow. (C) Immunofluorescence in SH-SY5Y human neuroblastoma cells. (Left) DISC1 (R47 antibody); middle, PDE4B (antibody against pan-PDE4B); (right) merged signal plus DAPI (nuclear DNA) stain (See also fig. S11). (D) Confocal immunofluorescence in differentiated SH-SY5Y human neuroblastoma cells, left to right as in (C). (Inset) Highlighted region enlarged 2.5 times. (E) Immunofluorescence in hippocampal primary cells, neuron and undifferentiated cell, left to right as in (C), except (left), DISC1 (C2 antibody) (See also fig. S11). Neurons were identified by using an antibody against NeuN.

To further investigate DISC1-PDE4 binding, we determined the effect of elevated cAMP. Overexpression studies show that long PDE4 isoforms are activated by PKA in response to increased intracellular cAMP and, thereby, provide part of the cellular desensitization system for cAMP signaling (18, 19). PKA phosphorylation of UCR1 induces conformational changes that concomitantly affect enzymatic activity and the interaction between UCR1 and UCR2 (20). To increase cAMP production and simultaneously to block its hydrolysis, SH-SY5Y cells were treated with forskolin plus the nonspecific phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX). This dramatically decreased the amount of DISC1 coprecipitating with PDE4B (Fig. 4, A and D). When cAMP levels were raised by using the cell permeable cAMP analog 8-bromo-cAMP, we similarly observed a reduction in PDE4B-DISC1 binding (Fig. 4, B and D). The action of forskolin plus IBMX was ablated when we added the PKA-specific inhibitor H89 (Fig. 4, C and D). These data indicate that the interaction between PDE4B and DISC1 is dynamic and directly influenced by altered cAMP levels through the action of PKA. Our data also indicate that DISC1 binds predominantly to the dephosphorylated, low-activity form of PDE4B, consistent with a model whereby PKA phosphorylation leads to release, from DISC1, of an activated population of PDE4B in response to elevated cAMP levels (fig. S12). To test this, we measured PDE4B cAMP hydrolyzing activity in SH-SY5Y cells and observed a 48% increase over controls, in the presence of IBMX and forskolin, that was inhibited by H89, consistent with PKA mediation of this increase (Fig. 4E). This increase in PDE4B activity is due to activation of UCR1-containing long PDE4B isoforms, because the short isoform PDE4B2 was not immunoprecipitated from SH-SY5Y cells (Fig. 4F). Using an antibody specific for the PKA-phosphorylated forms of PDE4 (18, 21), we identified a 104-kD PDE4B immunoreactive species in PDE4B immunoprecipitates from cells treated with IBMX and forskolin (Fig. 4F) that is not seen in control cells, consistent with PKA phosphorylation of UCR1 as the mechanism of PDE4B activation.

Fig. 4.

Effects of cAMP on DISC1 and PDE4B. SH-SY5Y cells were mock-treated or treated with (A) IBMX and forskolin, (B) 8-bromo-cAMP, (C) IBMX and forskolin with or without H89 for 30 min. Immunoprecipitates from a PDE4B-specific antibody were prepared and analyzed by Western blot analysis with the antibodies against PDE4B or the R47 antibody for DISC1. The DISC1 71-kD isoform is indicated with an arrow. (D) Relative coimmunoprecipitation of DISC1 in treated versus untreated cells (plotted as means + SEM of at least three independent tests). (E and F) SH-SY5Y cells were either untreated (Ctr) or challenged for 30 min with a mixture of either IBMX and forskolin or IBMX and forskolin plus H89. Immunoprecipitates from PDE4B-specific antibody were prepared from lysates for either (E) PDE4 activity assays by measurement of cAMP hydrolyzing activity, or (F) Western blot analysis with a PDE4B-specific antibody or with an antiserum specific for the single PKA phosphorylation site in UCR1. Methods for obtaining these data are available online [refs. S8, S9, S13, S14, and S21 in (11)].

Our study provides primary evidence for PDE4B as a genetic susceptibility factor for schizophrenia. We show that DISC1, an established candidate, interacts with PDE4B in a compartmentalized fashion. This interaction is dynamic and is cAMP- and PKA-dependent. We speculate that functional variation in DISC1 and/or PDE4 will modulate their interaction and affect mitochondrial cAMP catabolism with a concomitant physiological and psychiatric outcome. A unifying link between schizophrenia and bipolar affective disorder, and between DISC1 and PDE4, may occur at the cognitive level of learning and memory and at the molecular level of cAMP signaling.

Supporting Online Material

Materials and Methods

Figs. S1 to S12

References and Notes

References and Notes

View Abstract

Navigate This Article