Signaling systems coordinate most cellular functions and thus provide key targets for drug discovery. Identifying appropriate targets and generating selective modulators of these targets present major challenges in the initial stages of drug discovery. Different isoenzymes are often found at key control points in signaling networks, allowing cells to tailor signaling pathways through changes in activity, regulation, spatial distribution, and compartmentalization (Wong and Scott, 2004).

The cAMP signaling pathway plays a pivotal role in many key cellular processes (Taskén and Aandahl, 2004; Smith et al., 2006). Indeed, gradients of cAMP have been identified in various cell types (Zhang et al., 2001; Zaccolo and Pozzan, 2002; Willoughby et al., 2006), leading to compartmentalized responses (Taskén and Aandahl, 2004; Wong and Scott, 2004; Smith et al., 2006). Underpinning the formation of such gradients is the degradation of cAMP by phosphodiesterases (PDEs) (Jurevicius et al., 2003; Mongillo et al., 2004; Lynch et al., 2005; McCahill et al., 2005; Willoughby et al., 2006) targeted to specific intracellular sites and signaling complexes (Houslay and Adams, 2003; Baillie and Houslay, 2005). There are many PDE gene families, eight of which code for proteins able to hydrolyze cAMP (Manganiello and Degerman, 1999; Francis et al., 2001; Beavo and Brunton, 2002; Lugnier, 2006). To date, members of the PDE3 and PDE4 families have been shown to play an important role in determining compartmentalized cAMP signaling (Jurevicius et al., 2003; Mongillo et al., 2004; Lynch et al., 2005; McCahill et al., 2005; Willoughby et al., 2006), making it important to recognize the range of isoforms that form these families as a prelude to determining their functional roles.

The cAMP-specific phosphodiesterase 4 (PDE4) family is the target of several selective inhibitors having therapeutic potential as anti-inflammatory agents, antidepressants and cognitive enhancers (Huang et al., 2001; O'Donnell and Zhang, 2004; Renau, 2004; Spina, 2004; Houslay et al., 2005). Four genes (PDE4A, PDE4B, PDE4C, and PDE4D) generate a large set of PDE4 isoforms through the use of distinct promoters and alternative pre-mRNA splicing (Conti et al., 2003; Houslay and Adams, 2003; Houslay et al., 2005). Their unique N-terminal regions, encoded by specific 5′ exons, define individual PDE4 isoforms (Fig. 1). Accordingly, PDE4 isoforms are subcategorized into long forms, which possess the regulatory upstream conserved regions, UCR1 and UCR2; short isoforms, which lack UCR1; or super-short isoforms, which lack UCR1 and have a truncated UCR2.

The PDE4B gene has been linked to schizophrenia in humans (Millar et al., 2005; Pickard et al., 2007) and knockout of the PDE4B gene in mice both generates an anti-depressant-like profile (O'Donnell and Zhang, 2004) and compromises the generation of airway hyper-reactivity and inflammatory actions mediated by macrophages (Jin et al., 2005). In addition, chronic nicotine treatment, which has an anti-depressant action, causes down-regulation of PDE4B transcripts in the nucleus accumbens, prefrontal cortex, and hippocampus of rats (Polesskaya et al., 2007).

The human PDE4B gene has been shown to encode a number of distinct isoforms, namely the long PDE4B1 (Bolger et al., 1993) and PDE4B3 (Huston et al., 1997) isoforms and the short PDE4B2 isoform (Bolger et al., 1993; Obernolte et al., 1993). Although an additional long isoform, called PDE4B4, has been identified in rodents, this isoform is not encoded by the human genome (Shepherd et al., 2003). Indeed, particular PDE4 isoforms appear to have specific functional roles as evidenced from short interfering RNA-mediated knockdown studies in cells (Lynch et al., 2005) and from physiological studies showing that nocturnal increases in PDE4B2 provide a negative feedback role in adrenergic/cAMP signaling in the pineal gland (Kim et al., 2007).

Here we identify and characterize the first super-short isoform (PDE4B5) encoded by the PDE4B gene. This isoform is highly conserved across species, is active, and responds to drug inhibition. We also show, for the first time, conservation between PDE4 subfamilies of an isoform-specific N-terminal region, with the PDE4B5 N-terminal isoform of PDE4B5 being identical to the N terminus of the super-short isoform encoded by the PDE4D gene, PDE4D6 (Wang et al., 2003).

Materials and Methods

Reagents. [³H]cAMP and ECL reagents were purchased from GE Healthcare (Little Chalfont, Buckinghamshire, UK). Dithiothreitol, N–{1-(2,3-dioleoyloxy)propyl}-N,N,N,-trimethylammonium methylsulfate and protease inhibitor tablets were obtained from Roche Diagnostics (Mannheim, Germany). Bradford reagent was purchased from Bio-Rad (Herts, UK). All other materials were from Sigma-Aldrich (Poole, UK). Protein G-Sepharose beads was purchased from GE Healthcare. Anti-FLAG M2 was supplied by Sigma-Aldrich. PDE4B antisera was as described previously (Huston et al., 1997).

Computational Prediction and Analysis of PDE4B Novel Splice Variant. Mouse transcripts were aligned to the mouse genome and the orthologous human genomic loci (Kan et al., 2005), and the resulting splice patterns were compared with those of human transcripts to identify novel patterns. Inferred novel splice variant transcript sequences were extracted from the human genome sequence. The nucleotide and translated protein sequences of the variant were searched against NCBI human transcript and protein databases using the on-line BLAST server (http://www.ncbi.nlm.nih.gov/BLAST/). Genomic annotations and conservation in the PDE4B gene locus were inspected using the University of California at Santa Cruz genome browser, hg18 human assembly (http://genome.ucsc.edu/). Conserved transcription factor binding site predictions were taken from the browser tracks named “HMR Conserved Transcription Factor Binding Sites” by Matt Weirauch and Brian Raney at the University of California at Santa Cruz.

RT-PCR. We used the OneStep RT-PCR kit (catalog no. 210212; QIAGEN, Crawley, UK). The PCR component involved 35 cycles of 94°C for 30 s, 63.5°C for 40 s, and 72°C for 50 to 120 s. Products were resolved on a 2% agarose gel run at 100 V in TAE buffer (40 mM Tris, 1 mM EDTA, pH to 8.0 using glacial acetic acid). Primer sequences and expected band sizes for PDE4B5, the PDE4B C terminus, PDE4D6, and the PDE4D N terminus (excluding PDE4D6) are shown in Table 1. RT-PCR primers were designed to be specific to their target.

TaqMan Measurements. TaqMan primer-probe reagents were obtained through the Applied Biosystems Assays-by-Design custom assay service (Applied Biosciences, Foster City, CA). Probe sequences were designed to straddle the unique splice junction characteristic of each alternative splice form. TaqMan assays were performed on an ABI 7900 real-time PCR instrument in 10-μl assays that were run in triplicate in a 384-well format optical PCR plate. The assays were calibrated with isoform-specific RT-PCR clones using the standard curve method (http://www.appliedbiosystems.com/support/tutorials/pdf/essentials_of_real_time_pcr.pdf). Standard curves generated from plasmid clones were linear across at least 6 orders of magnitude, and all reported values derived for total tissue RNA fell within the range of these standard curves. RNA was converted to cDNA for TaqMan measurements using a commercially available kit from Applied Biosystems. All assays were normalized on a tissue-to-tissue basis by adding a constant amount of input total RNA into the RT reaction. TaqMan primer locations are labeled as C in Fig. 2a.

Microarray Data. Custom oligonucleotide microarrays were purchased from Agilent Technologies (Palo Alto, CA). We designed these arrays to monitor the expression of 18,000 genes and associated alternate splicing events (Johnson et al., 2003). After alignment of 107,551 full-length human mRNA transcripts to the human genome, probes were designed to target every exon (60-mers) and every exonexon junction (36-mers on 10-nt T stilts). Poly(A)⁺ mRNA was amplified with a full-length amplification method using random-priming sequences to reproduce the entire transcript (Castle et al., 2003). Fluorescent dye-labeling, hybridization conditions, and scanning were performed as described previously (Hughes et al., 2001). Each amplified sample was hybridized twice against a common reference pool in a dye-swap experiment. The reference pool included 20 disease-free adult tissues, including peripheral leukocytes but excluding other blood fractions. Ratios shown are the means of three probes located in regions common to all isoforms.

Tissues for RT-PCR, TaqMan, and Microarray Measurements. Cell lines and human tissues were purchased as mRNA or total RNA from Clontech (Mountain View, CA). Each tissue sample was pooled from multiple donors, typically 12, by the vendor.

SDS-PAGE and Western Blotting. Acrylamide gels (4–12%) were used, and the samples boiled for 5 min after being resuspended in SDS sample buffer. Gels were run at 100 V/gel for 1 to 2 h with cooling. For detection of transfected PDE by Western blotting, 2 to 50 μg of protein samples were separated by SDS-PAGE and then transferred to nitrocellulose before being immunoblotted using the indicated specific antisera. Labeled bands were identified using peroxidase linked to anti-rabbit IgG and the GE Healthcare ECL Western Blotting Kit was used as a visualization protocol. We used polyclonal antisera able to detect all active human PDE4B isoforms as described previously (Huston et al., 1997). This polyclonal antiserum was raised against the extreme C-terminal region that is unique to the PDE4B subfamily and is found in all known active PDE4B isoforms.

Constructs. The ORF encoding human PDE4B5 (EF595686) was engineered for expression in pcDNA3. PDE4B1 (GenBank accession no. L20966) was used as a PCR template to incorporate the common super-short form region of PDE4B into the new construct. The sequence of the novel 15 N-terminal amino acids of PDE4B5 plus six amino acids (249–254) of PDE4B1, a start codon, and a NotI restriction enzyme site were incorporated into the 5′ primer. The 3′ primer incorporated amino acids from the C-terminal region of PDE4B1, a stop codon, and a KpnI enzyme restriction site. We also generated a FLAG-tagged version in which the FLAG tag (N-Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys-C) was incorporated into the 3′ primer before the stop codon. A PCR reaction using TaqDNA polymerase, the PDE4B1 DNA template, and the above primers generated a fragment of approximately 1558 base pairs, which was purified using a PCR Purification Kit (QIAGEN). This fragment was digested with NotI and KpnI before ligation (Rapid DNA Ligation Kit; Roche Diagnostics) into the multiple cloning site of pcDNA3.1 (Invitrogen, Paisley, UK) to generate either PDE4B4-pcDNA3 or PDE4B5-FLAG-pcDNA3. Generation of a plasmid encoding the N-terminally FLAG epitope-tagged version of the 100-kDa full-length DISC1 has been described previously (Millar et al., 2005).

Transient Expression of PDE4B Isoforms in COS7 Cells. Transfection was done using the COS7 simian virus 40-transformed monkey kidney cell line maintained at 37°C in an atmosphere of 5% CO₂-95% air in complete growth medium containing Dulbecco's modified Eagle's medium supplemented with 0.1% penicillin-streptomycin (10,000 U/ml), glutamine (2 mM), and 10% fetal calf serum. As described previously (Huston et al., 1997; Rena et al., 2001; Wallace et al., 2005), COS7 cells were transfected using DEAE-dextran. The DNA to be transfected (10 μg) was mixed and incubated for 15 min with 200 μl of 10 mg/ml DEAE-dextran in PBS to give a “DNA-dextran” mix. When cells reached 70% confluence in 100-mm dishes, the medium was removed, and the cells were given 10 ml of fresh Dulbecco's modified Eagle's medium containing 0.1 mM chloroquine and the DNA-dextran mix (450 μl). The cells were then incubated for 4 h at 37°C. After this period the medium was removed and the cells were shocked with 10% DMSO in PBS. After PBS washing, the cells were returned to normal growth medium and left for a further 2 days before use. For determination of PDE activity the cells were homogenized in KHEM buffer (50 mM KCl, 10 mM EGTA, 1.92 mM MgCl₂, 1 mM dithiothreitol, and 50 mM HEPES, final pH 7.2,) containing “complete” protease inhibitors (Roche Diagnostics) of final concentrations of 40 μg/ml phenylmethylsulfonyl fluoride, 156 μg/ml benzamine, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin A, and 1 μg/ml antipain. In such transfected cells, >98% of the total PDE activity is due to the recombinant PDE4 isoform (Huston et al., 1997). In some instances the transfected COS7 cells were plated onto six-well plates for use in experiments and then serum-starved overnight before being treated with the indicated ligands for the stated lengths of time.

Subcellular Fractions. Disruption of COS7 cells was done as described previously (McPhee et al., 1995; Bolger et al., 1996; Huston et al., 1997). Cell homogenization was performed in KHEM buffer containing 1 mM dithiothreitol and a mixture of protease inhibitors at final concentrations of 40 μg/ml phenylmethylsulfonyl fluoride, 156 μg/ml benzamidine, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin A, and 1 μg/ml antipain. Pellet fractions were also resuspended in this mixture. We then generated a low-speed, P1 pellet (1000g_av for 10 min) and a high-speed, P2 pellet (60 min at 100,000g_av), which left a high-speed supernatant (S2) fraction. The homogenization procedure was complete in that no detectable latent lactate dehydrogenase activity was present in the P1 pellet, indicating an absence of cytosolic proteins. Equal volumes of samples were applied such that detection indicates relative distribution across these three cellular subfractions.

Confocal Analyses. Confocal imaging for analyzing PDE4 isoforms was performed as described previously (Rena et al., 2001; Shepherd et al., 2003; Wallace et al., 2005). Here, PDE4B5 was transiently overexpressed and visualized in COS cells using PDE4B-specific antisera (Huston et al., 1997). Cells were transfected using dithiothreitol and N–{1-(2,3-dioleoyloxy)propyl}-N,N,N,-trimethylammonium methylsulfate (Roche Diagnostics) with the PDE4A11-pcDNA3 plasmid. Protein was expressed for 48 h and cells were fixed in 4% paraformaldehyde containing 5% sucrose. After permeabilization in 0.2% Triton X-100, proteins were blocked using 10% goat serum and 2% bovine serum albumin before PDE4A11 was detected using an antibody raised against the C terminus of human PDE4A and stained using Alexa 594 (Invitrogen). Cells were observed using a Zeiss Pascal laser scanning microscope.

Immunoprecipitation. HEK293 cells expressing PDE4B5 and/or FlagDISC1 were washed with ice-cold PBS and lysed in PBS containing 1% Triton X-100, 1 mM dithiothreitol, 10 mM NaF, and 5 mM NaPP_i with a protease inhibitor cocktail added (Roche Diagnostics). Lysates were solubilized by rotation on a rotary wheel for 30 min at 4°C. Insoluble material was removed by a 15-min centrifugation at 14,000g_av at 4°C followed by preclearing of lysed supernatants by incubation with protein G-Sepharose beads for 30 min at 4°C. Equalized amounts of precleared lysates were incubated with the PDE4B antibody for a minimum of 3 h at 4°C, and immunocomplexes were captured after incubation with protein G-Sepharose beads for a further 1 to 2 h. The immunoprecipitates were washed three times in lysis buffer and eluted from the beads by the addition of Laemmli buffer (Laemmli, 1970).

Assay of cAMP PDE Activity. PDE activity using 1 μM cAMP as substrate was assayed by a modification of the procedure of Thompson and Appleman (1971) and Londesborough (1976) as described previously (Marchmont and Houslay, 1980; Sullivan et al., 1998; Rena et al., 2001). All assays were conducted at 30°C, and, in all experiments, a freshly prepared slurry of Dowex/H₂O/ethanol (1:1:1) was used. In all experiments, initial rates were taken from linear time courses of activity. Dose-dependent inhibition by rolipram was determined in the presence of 1 μM cAMP concentrations of cAMP as substrate over the indicated range of rolipram concentrations. The IC₅₀ was then determined from these values. Rolipram was dissolved in 100% DMSO as a 1 mM stock solution and diluted in 20 mM Tris-Cl, pH 7.4, and 10 mM MgCl₂ to provide a range of concentrations in the assay. The residual levels of DMSO were shown not to affect PDE activity over the ranges used in this study.

Protein Analysis. Protein concentration was determined using bovine serum albumin as standard (Bradford, 1976).

Results

Genomic Identification of PDE4B5. Mouse expressed sequence tags (ESTs) (e.g., BQ769324) aligned to the human genome indicate a previously unknown 5′ exon of the PDE4B gene (see Materials and Methods; Fig. 2a). From the novel first exon splice site, the 54 nt upstream (in the 5′ direction) and the 6 nt downstream (in the 3′ direction) show very high conservation across 15 species. At the 5′ end of this highly conserved sequence exists an in-frame ATG putative start codon (Fig. 2b). The sequence beyond 6 nt upstream of the ATG contains many insertions and deletions across species, as are also present beyond 6 nt downstream of the splice site. However, between the putative start codon and the splice site, no insertions, deletions, or stop codons exist in any of the species examined, including vertebrates, chicken, zebrafish, and fugu. This high rate of protein reading frame evolutionary conservation suggests the existence of a functional protein-coding region.

If translated, the alternate first exon would produce a protein of 484 amino acids with a unique N-terminal region of 16 amino acids (Fig. 2c). This protein would preserve the phosphodiesterase catalytic domain but eliminate the 250 N-terminal residues found in long isoforms, replacing them with the novel 16 amino acids. Thus, the putative novel PDE4B isoform, which we label PDE4B5, would fall into the category of a super-short isoform (Houslay, 2001), lacking UCR1 and having a truncated UCR2 (Fig. 1).

Intriguingly, with a BLAST search against the human protein database we found that the PDE4B novel variant is homologous to PDE4D6, a recently discovered PDE4D splice variant. The putative proteins coding PDE4B5 and PDE4D6 nucleotide sequences show 81% identity (40 of 48 identical nucleotides). Furthermore, the residues of the two unique N-terminal proteins perfectly match over the entire 16-amino acid sequence (Fig. 2c). This, to our knowledge, is the first description of conservation of an amino acid sequence over a PDE4 isoform-specific N-terminal region.

Validation and Expression of PDE4B5. To validate human transcription of the PDE4B5 isoform identified by mouse transcripts, we designed RT-PCR primers targeting the prediction. The forward primer was placed in the unique, novel exon of PDE4B5 and the reverse primer in the PDE4B catalytic region (Fig. 2a, A; see Materials and Methods), and both primers were designed to be specific to PDE4B. The gel image shows a bright band at the predicted size (442 nt) in brain sections (fetal brain, cerebellum, frontal lobe, pons, putamen, thalamus, and hippocampus) (Fig. 3A). Weaker signals were observed in retina, spinal cord, pituitary, fetal kidney, jejunum, ileum, lung carcinoma A549 cells, testis, HeLa cells, and G361 melanoma cells, and no band of another size was observed. The RT-PCR gel for total PDE4B expression, using PDE4B-specifc probes targeting the shared catalytic unit, shows fairly ubiquitous expression (Fig. 3B). We then quantitatively measured PDE4B5 and PDE4B N terminus (short and long, but not super-short) transcript levels using TaqMan in four tissues (Fig. 3C). These measurements show expression of the PDE4B long and short forms in all four tissues, whereas PDE4B5 (super-short) expression is largely constrained to fetal brain, with low levels in cell line A549, in agreement with the RT-PCR gel images.

Subsequent analysis shows only limited homology matches between PDE4B primers and PDE4D. The PDE4B5 forward primer, which lies outside of the coding region, and the PDE4B5 reverse primer, in the PDE4B common catalytic domain, have a maximal continuous homology to the PDE4D locus at a level of 4 and 6 nt, respectively. However, even if these matches were to produce products, their size would be significantly different (>50 nt) than the expected 442-nt band.

Given the homology to PDE4D6, we measured PDE4D6 expression using RT-PCR, as per PDE4B5. Again, we designed the forward primer to be specific to the first exon of PDE4D6, unique to the super-short isoform, and placed the reverse primer in the common catalytic region. We also designed primers to amplify the longer isoforms of PDE4D but not the super-short form, placing the forward primer in the region common to the long and short isoforms but not present in the super-short isoform. The PDE4D6 gel image (Fig. 4A) shows expression primarily in brain tissues. The PDE4D long and short gel image (Fig. 4B) shows expression in many tissues, but very low expression in liver, kidney, and K-562 cells.

Given the high expression similarity of PDE4B5 and PDE4D6 isoforms at the sequence and expression level, we examined whether they are similarly regulated. Conservation of the 2000 nt upstream of the common PDE4B5/PDE4D6 ATG codon is low but several cross-species conserved transcription factor binding sites are predicted for each isoform. Upstream of PDE4B5, conserved POU1F1, RFX1, RFX, BRN2, and OCT1 binding sites are predicted, and upstream of PDE4D6 only SRF is predicted. Thus, these predictions find no common transcription factor binding sites between these two super-short isoforms, suggesting that the expression of these two super-short species is distinctly controlled.

Expression of PDE4 Genes in Human Tissues. We measured the expression of PDE4A, PDE4B, PDE4C, and PDE4D using microarray experiments and show relative expression to a pool of tissues (Fig. 5). The microarray probes were designed to monitor constitutively transcribed exons and junctions and thus monitor overall gene expression. PDE4A shows slightly higher expression in monocytes, skeletal muscle, testis, and pons. PDE4B shows high enrichment in blood fractions and nervous system tissues. PDE4C shows ubiquitous expression. PDE4D is enriched in blood fractions and skeletal muscle, in agreement with the PDE4D gel image.

Size of PDE4B5 on SDS-PAGE. The ORF of PDE4B5 was engineered for expression in mammalian cells by cloning into pcDNA3. Transfection of COS1 cells with PDE4B5-pcDNA3 allowed for the detection of a single immunoreactive species of 58 ± 2 kDa, detected using a PDE4B-specific antiserum (Fig. 6a). Untransfected cells express PDE4B2 at levels not evident at the exposure level used here, which detects only the recombinant species, that is overexpressed so as to provide >98% of total PDE activity in these cells. This size agreed well with a predicted size of 57.7 kDa, as derived from the primary amino acid sequence. The short PDE4B2 isoform migrated as a 68-kDa species, as shown previously by us (Huston et al., 1997; Shepherd et al., 2003; Lynch et al., 2005).

Activity of PDE4B5. We transfected cells to express PDE4B5 and treated them with the archetypal PDE4-selective inhibitor, rolipram. More than 97% of the total cAMP PDE activity was inhibited by 1 μM cAMP rolipram as a substrate. Assayed with 1 μM cAMP, such transfected cells had a cAMP PDE activity of 2 to 4 nmol of cAMP hydrolyzed/min/mg of cell protein, whereas empty vector transfected cells had an activity of 4 to 6 pmol of cAMP hydrolyzed/min/mg of cell protein (n = 3). Thus, in PDE4B5-transfected cells, PDE4B5 accounts for >98% of the total cAMP activity.

Analysis of PDE4B5 activity showed that it had a K_m for cAMP of 5.8 ± 0.4 μM(n = 3). By analyzing equal immunoreactive amounts of PDE4B5 and PDE4B2, expressed in COS1 cells lysates, we were able to determine the V_max of PDE4B5 to be 18 ± 3% PDE4B2 in these cells.

We then determined the sensitivity of PDE4B5 to inhibition by rolipram and cilomilast, a compound that has been in phase 3 clinical trials for chronic obstructive pulmonary disease (Fig. 7). This analysis gave IC₅₀ values of 380 ± 63 nM (n = 4) and 114 ± 17 nM (n = 3), for rolipram and cilomilast, respectively. Rolipram binds to the catalytic site of PDE4B, thus providing competitive inhibition. Using the Cheng-Prusoff equation (Cheng and Prusoff, 1973) (K_I = IC₅₀/[1 + (S/K_m)]); K_I values for inhibition of PDE4B5 by rolipram and cilomilast are 324 and 97 nM, respectively.

Intracellular Distribution of PDE4B5. COS1 cells transfected to express PDE4B5 were disrupted and separated out in low-speed membrane (P1), high-speed membrane (P2), and high-speed supernatant (S2) fractions (Fig. 6a). These were then analyzed on a volume-for-volume basis for both PDE4 activity and for PDE4B5 immunoreactivity to determine the relative distribution of PDE4B5 among these three fractions (Table 2). This analysis showed that PDE4B5 was found predominantly in the high-speed supernatant, cytosolic fraction but was also evidently associated with membrane fractions, which accounted for approximately 30% of the total PDE4B5 (Table 2).

We also analyzed the distribution of PDE4B5 in transfected COS1 cells (Fig. 6b). As can be seen, although PDE4B5 is excluded from the nucleus, it is distributed throughout the cytosol with small amounts associated with the plasma membrane. Distribution through the cell interior is uneven, which may indicate its association with cytosolic vesicles/complexes in addition to soluble forms. This observation would be consistent with biochemical fractionation. A similar distribution was seen using a FLAG epitope-tagged form of PDE4B5 (Fig. 6b).

PDE4B5 Interacts with DISC1. DISC1 has previously been shown to interact with the long PDE4B1 and PDE4B3 isoforms as well as with the short PDE4B2 isoform (Millar et al., 2005). Here we show that the novel, super-short PDE4B5 isoform is able to interact with the full-length 100-kDa DISC1 (fl-DISC1) isoform. This is evident from their coimmunoprecipitation as a complex from HEK293 cells that were cotransfected to express both PDE4B5 and fl-DISC1 isoforms (Fig. 8).

Discussion

In determining the human transcriptome, the set of human mRNA genes has been largely established, and the current need is to establish alternative isoforms, their expression, and their function. To this end, the availability of genomic sequences and RNA transcripts from nonhuman species is a powerful, complementary resource to human RNA transcripts (Kan et al., 2005).

The human PDE4B gene has been shown to encode the long PDE4B1 and PDE4B3 isoforms and the short PDE4B2 isoform (Fig. 1; Table 3). These three isoforms are also seen in rodents, which additionally express the short PDE4B4 isoform, which is not found in humans (Shepherd et al., 2003) (Table 3). Here, we show that mouse EST transcripts predict a novel PDE4B isoform, which we label PDE4B5. This transcript replaces the seven 5′ exons of transcript NM_002600 with a single novel 5′ exon. Using RT-PCR, we confirmed human transcription of the variant and established that it is expressed specifically in brain sections. The 3′ region of this novel exon contains a putative in-frame ATG start codon, followed by a novel 16-residue ORF. That this ORF is conserved in vertebrae, chicken, frog, zebrafish, and fugu strongly suggests a functional role. The protein encoded by this isoform would be 484 residues and would contain the PDE catalytic domain and PDE4B C terminus but would lack the UCR1 domain and encode a naked, truncated UCR2 domain. Because the protein would be shorter than the previously identified long and short PDE4B forms, we classify it as a “super-short” isoform of PDE4B (Houslay, 2001) (Table 3).

By transfecting a PDE4B5 cDNA construct into COS1 cells, we were able to demonstrate that the cDNA generates a protein of 58 ± 2 kDa. Immunoreactivity measurements show that although this protein is found in membranes, it is found dominantly in the cytosol (high speed supernatant fraction). Confocal microscopy immunolocalization observations of recombinant PDE4B5 using an anti-PDE4B anti-serum similarly show a distribution throughout the cytosol with small amounts associated with the plasma membrane. PDE4B5 shows an uneven distribution throughout the cytosol, which may indicate association with cytosolic vesicles/complexes in addition to soluble forms.

The transfected PDE4B5 cDNA generates an active protein with a K_m for cAMP of 5.8 ± 0.4 μM. Rolipram binds in the PDE4 catalytic domain and indeed inhibits activity of this super-short isoform (containing the catalytic domain). Measuring inhibition by rolipram and cilomilast, we find IC₅₀ values of 380 ± 63 nM (n = 4) and 114 ± 17 nM (n = 3), respectively.

Comparing PDE4B5 to other PDE4 genes, we found that the novel 16 residues at the N terminus of PDE4B5 perfectly match the N terminus of the recently discovered super-short isoform of PDE4D, called PDE4D6 (Wang et al., 2003). Expanding on previous assessments (Wang et al., 2003), we found that the PDE4D6 transcript, like PDE4B5, is brain-specific. However, we identified no common transcription factor binding sites upstream of each isoform, suggesting that these isoforms are independently regulated.

Microarray gene expression profiling shows PDE4B and PDE4D enriched in white blood cell fractions. PDE4D is also overexpressed in skeletal muscle, whereas PDE4B is high in several nervous tissues, including dorsal raphe and hypothalamus, and, to a lesser degree, in muscle. However, what is invisible to these microarray “gene monitoring” experiments is the high specificity of individual isoforms to specific tissues, such as the high specificity of PDE4B5 and PDE4D6 isoforms to brain sections.

Pharmaceutical manipulation of PDE4 genes may aid in treatment of diseases, such as stroke and inflammation, including chronic obstructive pulmonary disease and asthma. Indeed, PDE4B and PDE4D gene expression is enriched in white-blood cell fractions, possibly suggesting an inflammatory role. PDE4B activity has also been implicated in psychiatric disorders via its interaction with DISC1 (Millar et al., 2005; Pickard et al., 2007), one of the most validated genetic risk factors for schizophrenia (Porteous and Millar, 2006). Indeed, very recently it has been shown that missense mutations in mouse DISC1 that confer depression-like and schizophrenia-like phenotypes interfere with PDE4B binding to DISC1 (Clapcote et al., 2007). Importantly, knockout of the PDE4B gene in mice yields an antidepressant-like phenotype (O'Donnell and Zhang, 2004) as does chemical ablation of PDE4 activity using the selective inhibitor rolipram (Wachtel, 1983; Zhang et al., 2002, 2006; Kanes et al., 2007). Indeed, the antidepressant phenotype observed upon chronic nicotinic treatment of rats leads to a specific down-regulation of PDE4B transcripts in brain (Polesskaya et al., 2007), consistent with a key role of PDE4B in regulating depression and psychosis. Both long (PDE4B1 and PDE4B3) and short (PDE4B2) isoforms have been shown to interact with DISC1 and UCR2 identified as a binding site (Millar et al., 2005). Intriguingly, here we show that the super-short PDE4B5 isoform can also interact with DISC1 (Fig. 8). This indicates that either the interaction site in UCR2 lies within the residual portion of the truncated UCR2 found in PDE4B2 or that there is an additional site for interaction with DISC1 that lies within the PDE4B catalytic unit. Indeed, recent analyses of various scaffold proteins that serve to sequester PDE4 have identified multiple binding sites (Bolger et al., 2006; Baillie et al., 2007; Sachs et al., 2007; Stefan et al., 2007), and our recent observations (H. Murdoch and M. D. Houslay, unpublished observations) indicate that the PDE4B catalytic unit does indeed contain a further binding site for full-length DISC1.

Despite the pharmaceutical significance of PDE4, its repertoire of isoforms available to the cell, how these isoforms are regulated, their distinct biological function, and their relationship to disease state are still unclear. Here, we have used mouse transcripts and isoform-specific expression monitoring to discover an intriguing, novel, active, brain-specific PDE4B isoform.