The cAMP-dependent signaling pathway is critically involved in memory-related synaptic plasticity. cAMP-specific type 4 phosphodiesterases (PDE4) play a role in this process by regulating the cAMP concentration. However, it is unclear how PDE4 is involved in regulating synaptic plasticity. To address this issue in Aplysia sensory-to-motor synapses, we identified a long isoform of the PDE4 homolog in Aplysia kurodai (apPDE), with genetic and biochemical properties similar to those of mammalian PDE4s. Furthermore, apPDE is localized to the membrane and presynaptic region. Both apPDE overexpression and knock-down impaired short- and long-term facilitation, indicating that an appropriate expression level of apPDE in synaptic regions is required for normal synaptic facilitation. By using fluorescence resonance energy transfer-based measurement of in vivo protein kinase A (PKA) activation, we found that the PKA activation by 5-hydroxytryptamine (5-HT) was impaired in both apPDE-overexpressed and knock-down synapses. Analogous to the inhibition of apPDE by RNA interference, chronic rolipram treatment before 5-HT stimulation also impaired the PKA activation by 5-HT, suggesting that regulation of the synaptic cAMP level by PDE4 is critical for normal synaptic facilitation. Together, we suggest that PDE4s localized in the synapses play a critical role in regulating the optimum cAMP level required for normal synaptic plasticity.
cAMP is a ubiquitous second-messenger molecule and activates several downstream effectors such as the protein kinase A (PKA) holoenzyme. In the PKA pathway, a cAMP-specific phosphodiesterase (PDE) acts as a negative regulator by reducing the intracellular cAMP level (Soderling and Beavo, 2000). PDEs have been classified into the 11 families based on their amino acid sequence, substrate specificity, and sensitivity for biochemical activators or inhibitors (Soderling et al., 1998a,b, 1999; Fawcett et al., 2000; Francis et al., 2001; Mehats et al., 2002). Among these, the type 4 PDE family contains four subtype genes (PDE4A, PDE4B, PDE4C, and PDE4D) (Verghese et al., 1995). According to their structures, each subtype is classified as super-short, short, and long isoforms (Houslay and Adams, 2003). Long-form PDE4 is known to be targeted to membrane fractions or interact with its partner molecules, such as anchoring proteins (Shakur et al., 1993; Beard et al., 2002; Bolger et al., 2003; Conti et al., 2003). Therefore, this isoform is believed to regulate the spatial and temporal characteristics of cAMP gradients in synapses.
cAMP-dependent signaling has come into the spotlight as a memory regulator. Previous reports with type 1 adenylyl cyclase transgenic mice (Wang et al., 2004) and rolipram, a PDE4-specific inhibitor (Barad et al., 1998), suggest that the elevation of cAMP level enhances synaptic plasticity and memory. In contrast, several conflicting studies, such as those on the dunce mutant in Drosophila (Dudai et al., 1976; Tully and Quinn, 1985; Chen et al., 1986; Dubnau and Tully, 1998) and Giα knock-out mice (Pineda et al., 2004), reported the impaired learning and memory phenotype caused by the elimination of the cAMP-negative regulator. These results suggest that the optimum range of cAMP level and PKA activity are required for normal memory formation. However, the exact role of PDE4 in this process is not clear.
In Aplysia, the sensitization of gill and siphon withdrawal reflex and related short- and long-term synaptic facilitation represent a simple form of learning and memory. These processes are dependent on the increase of cAMP and activation of PKA in sensory neurons by neurotransmitter 5-hydroxytryptamine (5-HT) (Castellucci et al., 1980, 1982; Bernier et al., 1982; Greenberg et al., 1987; Schacher et al., 1988; Byrne and Kandel, 1996). cAMP-specific PDEs are thought to be key elements in this process; however, little is known about the role of PDE4 in synaptic facilitation. In the present study, to investigate the role of PDE4 in synaptic plasticity between Aplysia sensory and motor neuron, we cloned a novel long isoform of the Aplysia PDE4 homolog (apPDE), which showed genetic structure, biochemical properties, and subcellular localization similar to those of mammalian long-isoform PDE4. Our results suggest that cAMP level in synaptic regions is regulated by apPDE and that this regulation is critical for short- and long-term synaptic facilitation in Aplysia.
Materials and Methods
Cloning. We obtained the apPDE fragment of Aplysia kurodai from the expressed sequence tagged (EST) database (pYESTrp2-based, 5′CAP and R5′CAP library) by using the basic local alignment search tool search. The super-short and short isoforms of apPDEs were identified before the discovery of the long isoform of PDE. In brief, we used a 1006 bp apPDE fragment from EST as a probe to screen ∼6 × 105 clones of Aplysia kurodai cDNA libraries and isolated several clones encoding parts of apPDE. Based on the sequences of these clones, we obtained the full length of super-short and short isoforms of apPDE by performing PCR with specific primers. For the super-short form of apPDE, we used the following primers: PDE-XhoI-s, 5′-CA CTCGAG GCC ACC ACC ATG CCT TCT GT G TGC GCG-3′; PDE-BamHI-a, 5′-CG GGATCC TCA TTT CTT GTC CTG T CC-3′. For the short form of apPDE, we performed PCR in pYESTrp2 library with designed primer sets: BcoIII-s, 5′-CGCACTTACATTCGCACTG-3′; PDE-2A, 5′-GTGGTCGACGTCATAAAT-3′. PCR products were cloned into a pGEM-T vector (Promega, Madison, WI) and confirmed by sequencing.
For the cloning of the long isoform of apPDE, we used a 418 bp region between the upstream conserved region 1 (UCR1) and UCR2 domain as a probe. We screened the R5′CAP library using this probe and found two different clones partially containing the long-form-specific N-terminal domain and the catalytic domains. We combined these partial clones to obtain full length clones. To confirm the full sequence of these clones, we designed two primers based on our full sequences: apPDE_RT-s, 5′-CTCTGCTCCGCGTCTTCTAT-3′; new PDE-a, 5′-TCATTTCTTGTCCTGTCCTCT-3′. Using cDNA from the sensory cluster in the pleural ganglion as a template, we performed reverse transcription (RT)-PCR with the above primer set. We obtained specific full-length PCR products and confirmed their sequence (see Fig. 2 A).
Plasmid construction. For pcDNA3.1-apPDE-FLAG constructs, we performed PCR with PDE-BamHI-s and PDE-BamHI-a: 5′-CG GGA TCC TTA TTT TGT TTC ACC AG-3′. We inserted BamHI-digested PCR product into a BamHI-digested pcDNA3.1 vector (Invitrogen, San Diego, CA).
For the FLAG-tagged expression vector (pNEXδ-apPDE-FLAG), we generated PCR products of apPDE by 30 cycles of PCR performed with the following primers: PDE-XhoI-s, 5′-GA CTC GAG GCC ACC ACC ATG CCT TCT GTG TGC GC-3′; PDE-BamHI-a, 5′-CG GGA TCC TCA TTT CTT GTC CTG TCC-3′. We inserted SalI-BamHI-digested PCR product into the pNEXδ vector, which was digested with SalI and BamHI.
For pNEXδ-apPDEΔN-FLAG, we generated the PCR product from pNEXδ-apPDE-FLAG by using the following primers: BamHI-apPDE-deltN-s, 5′-CG GGA TCC ATG CCC CAA AGG AGG GAG TCC-3′; BamHI-LacZ-a, 5′-CG GGA TCC GAC ACC AGA CCA ACT GG.-3′. We ligated the BamHI-digested PCR product and pNEXδ.
PDE activity assay. We assayed PDE activity by a modified procedure of Thompson and Appleman (1971) and Bauer and Schwabe (1980). The principle of this method is that 3H-labeled cAMP is converted to 3H-labeled 5′-AMP by phosphodiesterase activity. Degraded 3H-labeled cAMP is purified with QAE-Sephadex (Sigma, St. Louis, MO) columns. We prepared the apPDE protein by overexpressing pcDNA3.1-apPDE-FLAG in COS7 cells. We confirmed that the overexpressed apPDE protein mainly existed in the soluble part by Western blotting using monoclonal mouse anti-FLAG (Sigma) (data not shown). The assay consists of a two-step procedure. First, we performed reactions at pH 8.0 in 40 mm Tris-HCl buffer containing 0.5 mm MgCl2, 1 μm [3H]cAMP (80,000 cpm), and diluted apPDE-overexpressed COS7 cytosolic extract into a final volume of 0.2 ml. We initiated the reaction by adding the extract to the assay mixture, performed in a 1.5 ml tube at 30°C for 10 min, and terminated by heating the tubes to 95°C for 3 min, followed by immediate cooling on ice. In the second step, 50 μlof Crotalus atrox snake venom (1 mg/ml) (Sigma) was added for [3H]adenosine formation. After 10 min of incubation at 30°C, we terminated the reaction by applying an aliquot of 0.2 ml to the QAE A-25 Sephadex column previously equilibrated with 3 ml of 30 mm ammonium formate, pH 6.0. We collected elutes directly into scintillation vials and counted radioactivity in 8 ml of a scintillation fluid. We determined dose-dependent inhibition in the presence of 1 μm cAMP and over a range of 10 nm to 100 μm rolipram (Sigma) and 1 μm to 1 mm 3-isobutyl-1-methylxanthine (IBMX) (Sigma). PDE inhibitors (rolipram and IBMX) were dissolved in 100% dimethylsulfoxide (DMSO) and diluted with the enzyme diluents. The final concentration of DMSO in the assay mixture was 1%. We determined the IC50 from these values, using a nonlinear least-squares fitting algorithm in Origin 6 software (Microcal, Northampton, MA). We determined Km values over a substrate range of 0.1–20 μm cAMP (eight different concentrations) by fitting to the hyperbolic form of the Michaelis-Menten equation by using a least-square procedure in the Origin 6 software.
Immunocytochemistry and image analysis. We performed immunocytochemistry as described by Martin et al. (1997). The monoclonal mouse anti-FLAG antibody (Sigma) and the secondary cyanine 3 (Cy3)-conjugated anti-mouse antibody (Amersham Biosciences, Piscataway, NJ) were used in dilution ratios of 1:250 and 1:1000, respectively. We obtained and analyzed fluorescence images by a confocal laser-scanning microscope (Radiance 2000; Zeiss, Jena, Germany) and NIH Image J software (National Institutes of Health, Bethesda, MD), respectively.
In vivo measurement of PKA activity. Plasmid constructs for in vivo cAMP level imaging were gifts from Manuela Zaccolo (University of Padua, Padua, Italy), and the method used for analysis was described previously (Zaccolo and Pozzan, 2002). In brief, because cAMP causes dissociation of the PKA regulatory and catalytic subunit by binding to the regulatory subunit, the elevated cAMP level can be detected by decreased fluorescence resonance energy transfer (FRET) between these subunits (Zaccolo and Pozzan, 2002). This reduced FRET also indicates the degree of PKA activation, because the dissociation of PKA regulatory subunit by cAMP binding produces free catalytic subunits that could phosphorylate their target proteins. Decreased FRET can be converted into an elevated cyan fluorescent protein (CFP)/yellow fluorescent protein (YFP) emission ratio. We performed FRET imaging on a Zeiss LSM510 confocal laser scanning microscope. RII (type II regulatory subunit)-CFP- and CAT (catalytic subunit)-YFP-expressed Aplysia sensory neurons were excited at 458 nm, and fluorescence emission was collected with BP475–525 (for CFP) and LP530 (for YFP) emission filters. To test whether FRET between RII-CFP and CAT-YFP probes occurred in Aplysia neuron, we used the acceptor bleaching method and calculated FRET efficiency as described previously (Karpova et al., 2003). After background correction, we measured both CFP and YFP emission values using the NIH Image J software, and the CFP/YFP ratios were subsequently calculated. 5-HT (Sigma) was freshly prepared by dissolving it in L15 (Sigma)/artificial seawater (ASW). A 200 mm stock solution of 8-(4-chlorophenylthio)-cAMP (8-CPT-cAMP; Sigma) in DMSO was diluted freshly by dissolving in L15/ASW and used. In the chronic rolipram experiment, we applied diluted rolipram into bath 24 and 48 h before FRET measurement. The final concentration of DMSO was below 0.1%.
Statistics. Student's t test and two-way ANOVA with Bonferroni's post hoc test was performed using Prism version 4.00 (GraphPad Software, San Diego, CA).
Cell cultures, electrophysiology, and the induction of long-term facilitation by 5-HT. We prepared both culture dishes and medium as described previously (Schacher and Proshansky, 1983; Montarolo et al., 1986). We isolated sensory cells from the pleural ganglia of adult animals (100–150 g) and cultured these for sensory neuron culture or cocultured these for sensory-to-motor culture with a left F cluster innervating to the siphon (LFS) motor neuron from the abdominal ganglia of adult animals, as described previously (Montarolo et al., 1986). Cultures were maintained for 3–4 d at 18°C and then used.
For the electrophysiological recording, the motor cell was then impaled intracellularly with a glass microelectrode filled with 2 m K-acetate, 0.5 m KCl, and 10 mm K-HEPES (10–15 MΩ), and the membrane potential was maintained at –30 mV below its resting value. The EPSP was evoked in the LFS by stimulating the sensory neurons with a brief depolarizing stimulus with an extracellular electrode. To examine basal synaptic transmission, we measured the EPSP before and 48 h after microinjection of pNEXδ-apPDE-FLAG or specific double-stranded RNA (dsRNA). To investigate the effect of apPDE on synaptic plasticity, we measured the initial EPSP value 24 h after microinjection. We then gave one pulse of 5-HT to induce short-term facilitation or five pulses of 5-HT for 5 min at 15 min intervals to induce long-term facilitation. The amount of synaptic facilitation was calculated as the percentage change in EPSP amplitude recorded after the 5-HT treatment versus its initial value before treatment. All data are presented as the mean ± SEM percentage change in EPSP amplitude.
RNA synthesis by in vitro transcription. To prepare template DNAs for in vitro transcription, we inserted the cDNA for apPDE-specific fragment into a pLITMUS28i vector (New England Biolabs, Beverly, MA). We synthesized dsRNA by in vitro transcription of linearized template DNAs by using T7 RNA polymerase according to the instructions of the manufacturer (Ambion, Austin, TX).
apPDE is similar to mammalian long-form PDE4
We obtained an EST clone homologous to the catalytic domain of the Drosophila dunce gene from the Aplysia EST database. By using this EST clone as a probe, we cloned full-length cDNA by library screening and nested PCR (see Materials and Methods).
The long isoform of mammalian PDE4 contains a unique N-terminal domain and complete UCR1 and UCR2 domains as well as catalytic domains; however, the short and super-short isoforms contain partial UCR1 and UCR2 domains and only the UCR2 domain, respectively (Conti et al., 2003). Based on this, we found that the structure of the 792 aa candidate (apPDE) is most similar to that of the mammalian long isoform of PDE4 (supplemental Fig. 1A,B, available at www.jneurosci.org as supplemental material). We also found a high degree of similarity between the overall amino acid sequences of apPDE and mammalian PDE4D or dunce (dunce, 60% identity and 73% similarity; human PDE4D7, PDE4D8, and rat PDE4D7, 64% identity and 77% similarity) (supplemental Fig. 1A, available at www.jneurosci.org as supplemental material). Among PDE4 isoforms, the overall amino acid sequence homology was highest with PDE4D7 and PDE4D8 (long isoform of PDE4D), suggesting that apPDE is a PDE4D homolog.
apPDE also contained the domain for cAMP hydrolysis and canonical PKA phosphorylation sites that are conserved in the mammalian PDE4 and dunce (Sette and Conti, 1996) (supplemental Fig. 1A, available at www.jneurosci.org as supplemental material). These similar structural properties suggest that this clone is most probably similar to a long isoform of PDE4.
apPDE showed cAMP-specific activity and rolipram-sensitive inhibition
PDE4 is known to have substrate specificity for cAMP and to have lower IC50 values for rolipram, a PDE4-specific inhibitor, than for IBMX, a general PDE inhibitor (Houslay and Adams, 2003). To confirm whether apPDE has a PDE4-like phosphodiesterase activity, we performed the biochemical PDE activity assay to measure these features.
First, the phosphodiesterase activity of apPDE was only cAMP specific and showed no activity for cGMP (Fig. 1A). We found that the Km value for cAMP was 5.7 ± 0.7 μm (mean ± SEM; n = 3). Second, the inhibitory effect of rolipram on cAMP-specific PDE activity was higher than that of IBMX (Fig. 1B). We found that the IC50 values for rolipram and IBMX were 4.0 ± 1.8 μm (mean ± SEM; n = 3) and 188.6 ± 13.6 μm (mean ± SEM, n = 3), respectively. Considered together, we conclude that apPDE has biochemical features similar to those of mammalian PDE4s.
apPDE was targeted to presynaptic regions
Previous studies suggest that long-form PDE4 is mainly localized to the membrane fraction in non-neuronal cells, and its unique N-terminal domain directs this localization as well as interaction with its partners (Shakur et al., 1993; Beard et al., 2002; Bolger et al., 2003). However, little is known about the cellular localization of PDE4 in neurons. Therefore, we examined the cellular distribution of apPDE in sensory neurons.
First, the existence of apPDE mRNA in Aplysia sensory neurons was revealed by RT-PCR performed with a specific primer set spanning the full length of apPDE (Fig. 2A). Next, we overexpressed FLAG-tagged apPDE to examine the cellular localization of PDE4. Immunocytochemistry data indicated that apPDE-FLAG is mainly localized to the membrane region in soma and neurites (Fig. 2B). In particular, overexpressed apPDE was highly enriched in neurite ends and varicosity-like regions (Fig. 2B), suggesting that apPDE is targeted to synaptic regions in distal neurite regions. To further examine the synaptic localization of apPDE more clearly, we used a presynaptic marker, synaptophysin-enhanced green fluorescent protein (EGFP) (Kim et al., 2003). When apPDE-FLAG was coinjected with synaptophysin-EGFP into sensory neurons, most apPDE-FLAG puncta were colocalized with the synaptophysin-EGFP puncta (Fig. 2C). We also confirmed the synaptic localization of apPDE by using Aplysia sensory-to-motor synapse cultures and by analyzing the colocalization patterns (Fig. 2D). As found in the sensory neurons, the synaptophysin-EGFP puncta were colocalized with the apPDE-FLAG signals in the sensory-to-motor synapses, indicating that apPDE was targeted to presynaptic sites (Fig. 2D).
apPDE-specific N-terminal domain was required for synaptic targeting
We tested the role of the N-terminal domain of apPDE in targeting to the synaptic site. By using a mutant construct from which 1–115 aa residues were deleted from the full length of apPDE (apPDEΔN) (Fig. 3A), we compared its localization with that of the wild-type construct. Based on immunocytochemistry data, we analyzed colocalization between apPDE or apPDEΔN and synaptophysin by calculating the ratio between colocalized apPDE or apPDEΔN puncta and synaptophysin puncta (colocalization ratio; see Materials and Methods).
The N-terminal domain of PDE4 is known to be involved in membrane targeting (Shakur et al., 1993; Beard et al., 2002; Bolger et al., 2003). In accordance with this, apPDEΔN showed disrupted membrane localization and lower colocalization ratio than apPDE with synaptophysin puncta in distal neurites of sensory neurons (Fig. 3B). The average colocalization ratio of apPDEΔN was less than that of wild-type apPDE by ∼50% (apPDE, 0.99 ± 0.01, n = 7 vs apPDEΔN, 0.53 ± 0.08, n = 14; **p < 0.01, unpaired t test), indicating that the apPDE-specific N-terminal domain is critical for colocalization with synaptophysin puncta or synaptic targeting.
apPDE overexpression blocked synaptic facilitation in Aplysia sensory-to-motor synapses
Because the elevation of cAMP level and activation of the PKA pathway are critical for 5-HT-induced synaptic facilitation in Aplysia (Castellucci et al., 1980, 1982; Bernier et al., 1982; Greenberg et al., 1987; Schacher et al., 1988; Byrne and Kandel, 1996), we investigated the role of apPDE in synaptic facilitation by overexpressing apPDE in Aplysia sensory-to-motor culture. Because apPDE had cAMP-specific phosphodiesterase activity (Fig. 1A) and was localized to synaptic regions (Fig. 2D), we speculated that manipulated expression of apPDE would alter the cAMP level in synaptic regions.
We microinjected the apPDE or control (EGFP) plasmid construct into the sensory neurons of the sensory-to-motor culture. Both short- and long-term synaptic facilitation induced by 5-HT were significantly impaired when apPDE was overexpressed in the presynaptic sensory neurons (Fig. 4A,B). Short-term synaptic facilitation induced by 1 pulse of 5-HT (5 min) was considerably reduced by overexpression of apPDE (17.9 ± 4.9%; n = 6; *p < 0.05, **p < 0.01, unpaired t test) (Fig. 5A,B) when compared with that in control cells expressing EGFP (51.6 ± 8.0%; n = 5). Long-term facilitation induced by five pulses of 5-HT was also significantly blocked by apPDE overexpression (18.1 ± 4.1%; n = 6) (Fig. 4A,B) when compared with that in control cells expressing EGFP (64.4 ± 7.0%; n = 5; **p < 0.01, ***p < 0.001, unpaired t test) (Fig. 4A,B). However, apPDE overexpression did not affect basal synaptic transmission (percentage change, 8.4 ± 1.7%; n = 6), which was comparable with that of the control cells (EGFP-expressed cells, 10.5 ± 6.6%; n = 5). These results suggest that apPDE overexpression in synaptic regions reduces 5-HT-induced long-term and short-term facilitation in Aplysia sensory-to-motor synapses probably by reducing the cAMP concentration.
Blockade of apPDE by RNA interference reduced synaptic plasticity in Aplysia sensory-to-motor synapses
To further examine the effect of cAMP regulation by apPDE on synaptic plasticity, we attempted to lower the expression level of apPDE by microinjecting apPDE-specific dsRNA into sensory neurons of the sensory-to-motor synapse. In our previous report, this method was found to inhibit gene expression specifically and effectively in Aplysia neurons (Lee et al., 2001, 2003). We performed immunocytochemistry to examine the inhibitory effect of apPDE dsRNA on the expression of apPDE. apPDE FLAG expression, as determined by anti-FLAG antibody staining, was significantly reduced in sensory neurons injected with apPDE dsRNA compared with that in cells injected with control luciferase dsRNA (dsLuci) (normalized fluorescence intensity, dsLuci injected, 0.78 ± 0.16; apPDE dsRNA injected, 0.34 ± 0.06) (supplemental Fig. 2, available at www.jneurosci.org as supplemental material).
It was observed that the introduction of apPDE dsRNA had no significant effect on the basal synaptic strength (percentage change, –2.0 ± 5.9%; n = 5) compared with that of control synapses injected with luciferase dsRNA (percentage change, 4.2 ± 2.1%; n = 8). On the contrary, cells injected with apPDE dsRNA produced neither short-term (percentage change, 7.7 ± 6.6%; n = 6; ***p < 0.001, unpaired t test) nor long-term (percentage change, 22.4 ± 9.8%; n = 7; *p < 0.05, unpaired t test) facilitation (Fig. 4C,D). However, control cells injected with luciferase dsRNA showed normal synaptic facilitation (percentage change, short-term facilitation, 70.3 ± 7.7%, n = 4; long-term facilitation, 70.4 ± 15.1%, n = 7) (Fig. 4C,D). Together, these results indicate that reduced apPDE expression in synaptic regions blocks the 5-HT-induced short-term and long-term synaptic facilitation by disrupting the regulation of cAMP in Aplysia sensory-to-motor synapses.
Overexpression of apPDE reduced presynaptic PKA activation
To examine the effects of apPDE overexpression or knock-down on the kinetics of the in vivo cAMP level attributable to 5-HT stimulation, we performed in vivo measurement of PKA activation in living Aplysia sensory neurons using a modified FRET measurement between the regulatory and catalytic subunits of PKA (Bacskai et al., 1993; Zaccolo and Pozzan, 2002) (see Materials and Methods) (supplemental Fig. 3, available at www.jneurosci.org as supplemental material).
First, to determine the effect of apPDE overexpression on the PKA activation in synaptic region, apPDE-FLAG was coexpressed with RII-CFP and CAT-YFP in sensory neurons of the sensory-to-motor synapses. Because apPDE is highly localized to the presynaptic varicosities (Fig. 2D), the CFP/YFP ratio was measured in these regions. As shown in Figure 5A, the PKA activation and the stable maintenance of its elevated activity was observed in the control synapses after treatment with 10 μm 5-HT (microinjected with empty vector). However, overexpression of apPDE resulted in a lowered PKA activity in the presynaptic varicosities than in control varicosities (Fig. 5A). This reduction in PKA activity attributable to apPDE overexpression was statistically significant (two-way ANOVA; F = 43.26; df = 1; p < 0.01). Individual comparisons with post hoc tests revealed that synaptic regions of EGFP (control)-expressed and apPDE-overexpressed cells showed that PKA activity at the all of the time points after 5-HT treatment was significantly different (*p < 0.01) (Fig. 5A). This result indicates that apPDE could reduce the cAMP level in vivo. Moreover, this result suggests that impaired synaptic facilitation is attributable to the impaired PKA activation produced by the overexpression of apPDE in the synaptic regions.
Next, we investigated the effect of apPDE knock-down on cAMP kinetics. After microinjection of apPDE dsRNA with FRET sensor constructs (RII-CFP and CAT-YFP) into sensory neurons of the sensory-to-motor synapses, the FRET values were measured before and after treatment of 10 μm 5-HT at the presynaptic varicosities. In the control synapses, which were microinjected with luciferase dsRNA, treatment with 10 μm 5-HT caused normal PKA activation (Fig. 5B). In the synapses injected with apPDE dsRNA, the initial PKA activation caused by 5-HT was not considerably different from that in the control synapses (Fig. 5B); however, the PKA activity dropped to its basal level 90 s after 5-HT stimulation (Fig. 5B). A two-way ANOVA test indicated a significant effect of apPDE knockdown (F = 4.70; df = 1; p < 0.05) on the PKA activity after 5-HT treatment. Individual comparisons with post hoc tests revealed that the synaptic region of luciferase dsRNA- and apPDE dsRNA-injected cells showed significantly different PKA activity 90 s after 5-HT stimulation (*p < 0.05) (Fig. 5B). We also performed PKA activation measurement using FRET to test whether exogenous cAMP treatment to apPDE knock-down synapses could rescue the impaired PKA activation. We found that the application of 200 μm 8-CPT-cAMP with 5-HT could activate PKA in apPDE knock-down synapses (supplemental Fig. 4, available at www.jneurosci.org as supplemental material), indicating that lowered PKA activity by injection of apPDE dsRNA (Fig. 5B) was attributable to neither the nonspecific effect of dsRNA on the FRET measurement nor a ceiling effect of the saturated cAMP level. Together, the knock-down of apPDE impairs the elevation of cAMP level induced by 5-HT in synaptic regions, which might contribute to the impaired synaptic facilitation in the Aplysia sensory-to-motor synapses.
Chronic rolipram treatment had an inhibitory effect similar to that of apPDE knock-down on cAMP kinetics
Our results showed that inhibition of apPDE expression impaired the 5-HT-induced synaptic facilitation and PKA activation in the Aplysia sensory-to-motor synapses (Figs. 4D, 5B). However, treatment with rolipram, a PDE4-specific inhibitor, is known to enhance the hippocampal synaptic plasticity and memory by enhancing the cAMP levels (Barad et al., 1998). We hypothesized that these conflicting results might be attributable to the difference in the degree of PDE4 inhibition.
To test this hypothesis, we imitated the inhibitory effect of apPDE dsRNA by chronic treatment with rolipram. In Aplysia sensory neurons, 0.1 μm rolipram was demonstrated to produce an increase in membrane excitability, which was regulated by cAMP-dependent inhibition of K+ channels (Goldsmith and Abrams, 1992). To measure the effect of chronic inhibition of PDE4 on PKA activation, the sensory-to-motor cultures were treated with 0.1 μm rolipram for 24 and 48 h before 5-HT treatment, and FRET values were measured after 5-HT stimulation. Overall, the PKA activation kinetics of 24 h-treated synapses decreased slightly, but there were no significant differences in the effect of 5-HT treatment between DMSO and rolipram (24 h treated) groups at all time points (two-way ANOVA, p = 0.39) (Fig. 5C). In the case of the 48 h-treated synapses, the PKA activity was significantly lower than that in the control synapses 60 s after the 5-HT treatment (two-way ANOVA; F = 8.70; df = 1.0; p < 0.05) (Fig. 5C). Individual comparisons with post hoc tests revealed that the DMSO and 48 h (rolipram) treated synaptic regions showed significantly different PKA activity from 90 to 120 s after the 5-HT treatment (*p < 0.05) (Fig. 5C). These results suggest that chronic inhibition of PDE4 with rolipram impairs the increase in cAMP caused by the 5-HT in the synaptic regions; this is analogous to the effect of apPDE knock-down.
apPDE is most probably close to a long isoform of mammalian PDE4D and targeted to presynaptic regions
Aplysia PDE showed high overall sequence homology with mammalian PDE4Ds (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). This gene also showed a genetic structure, cAMP-specific substrate activity, and rolipram-sensitive inhibition of PDE activity similar to those of mammalian PDE4s (supplemental Fig. 1 A, B, available at www.jneurosci.org as supplemental material) (Fig. 1 A, B).
Our immunocytochemistry data showed that apPDE is mainly targeted to the membrane region in soma and proximal neurites (Fig. 2B). This is consistent with other in vitro experiments with mammalian long-form PDE4, which reported that long-isoform PDE4s are detected in the membrane fraction (Bolger et al., 1997; Jin et al., 1998; Shepherd et al., 2003). apPDE was also highly distributed at synaptic regions in distal neurites colocalized with a synaptic marker, synaptophysin-EGFP (Fig. 2C,D), similar to mammalian long-form PDE4A in the cerebellum (Shakur et al., 1995).
PKA subunits in Aplysia neuron also have the compartmentalized function and localization. For example, it has been known that type II regulatory subunit of PKA in Aplysia sensory neuron is also enriched in neurites in contrast to type I regulatory subunit in soma, and the expression level of each regulatory subunit is differentially affected by long-term facilitation (Liu et al., 2004). In a similar manner, short isoform of apPDE, which is found to be in Aplysia (see Materials and Methods), may contribute to the cAMP regulation in nonsynaptic region and have a distinct role for synaptic facilitation. Furthermore, we suggest that both apPDE and type II regulatory subunit of PKA may cooperate to regulate the PKA signaling cascade in the synaptic region because our results showed that synaptic PKA activity is tightly regulated by apPDE (Fig. 5A–C).
We also found that the long-isoform-specific N-terminal domain in apPDE is critical for synaptic targeting (Fig. 3B). However, we cannot exclude the possibility that other long-form-specific domains (UCR1 and UCR2) might be related with this localization because the colocalization ratio was not completely diminished by the deletion of this N-terminal domain (see Results). In concurrence with this, related mammalian PDE4 studies using chimeric constructs reported that UCR1 or UCR2 is also required for membrane targeting in non-neuronal cells (Shakur et al., 1993, 1995; Smith et al., 1996; Bolger et al., 1997; Jin et al., 1998; Beard et al., 2002; Shepherd et al., 2003). Future studies should be performed to find other molecular requirements for synaptic targeting of PDE4.
apPDE regulates synaptic facilitation by regulating cAMP level in synaptic regions
We analyzed the functional role of apPDE in synaptic facilitation through the overexpression or RNA interference (RNAi) of apPDE. Overexpressed apPDE did not affect the basal synaptic transmission; however, it did impair short-term and long-term synaptic facilitation (Fig. 4B). This might be the result of increased phosphodiesterase activity in the synaptic regions where apPDE was localized (Fig. 2D). Furthermore, knock-down of apPDE in the synaptic regions resulted in decreased synaptic facilitation without affecting the basal synaptic transmission (Fig. 4D). This result is consistent with the impaired learning and memory phenotype of the Drosophila dunce mutant (Dudai et al., 1976), Giα knock-out mice (Pineda et al., 2004), and Gsα-overexpressed Drosophila (Connolly et al., 1996). These data indicate that a balanced and optimized cAMP level regulated by PDE4s in synaptic regions is presumably important for synaptic plasticity and memory formation.
To test this possibility, we determined the relative change in the PKA activity caused by 5-HT in both apPDE-overexpressed and knock-down synapses. Because our in vivo imaging using FRET sensors represents the dissociation between regulatory and catalytic subunit (Bacskai et al., 1993; Zaccolo and Pozzan, 2002), indicating the degree of PKA activation, this technique allowed us to estimate the relative change in the in vivo PKA activity. We found that not only the overexpression but also the knock-down of apPDE impairs the PKA activation by 5-HT stimulation (Fig. 5A,B). We also found that the treatment of a membrane-permeable cAMP analog activated the PKA in the apPDE knockdown synapses, indicating that reduced PKA activity in the apPDE knock-down synapses was not simply attributable to a ceiling effect of the saturated cAMP level in the apPDE knock-down synapses (supplemental Fig. 4, available at www.jneurosci.org as supplemental material). Therefore, we suggest two possible molecular mechanisms for this observation. First, chronic PDE inhibition may impair cAMP generation in response to external stimulation. A previous report has shown that the persistent (>24 h) activation of adenylyl cyclase in non-neuronal cells enhances the inhibitory pathway of adenylyl cyclase by the increase and decrease of the Giα and Gsα expressions, respectively (Hadcock et al., 1990). Thus, PDE inhibition may cause the downregulation and/or desensitization of adenylyl cyclase in apPDE-inhibited sensory neurons. Second, it is possible that the inhibition of one isoform of PDE induces the expression of other PDE isoforms as a compensatory mechanism. Our data showed the rapid decay of PKA activation (<90 s) in response to 5-HT stimulation in apPDE-inhibited neurons, suggesting that there could be robust activities of other PDE isoforms. In the mammalian system, the activity of the short-isoform PDE4 is controlled by change in its expression level attributable to extracellular signals, such as elevated cAMP levels, although the activity of the long isoform of PDE4 is mainly modified by phosphorylation, ligand binding, and domain–domain interaction (D'Sa et al., 2002; Conti et al., 2003; Richter et al., 2005). Because the short isoforms of PDE4 were also found in Aplysia (data not shown; see Materials and Methods), these short isoforms are possibly involved in this compensatory mechanism in the PDE-inhibited cells. More studies will be required to clarify this question.
Effects of chronic administration of rolipram on cAMP dynamics and synaptic facilitation
The use of PDE4 inhibitors as promising drugs for ameliorating memory problems and several mental disorders, such as depression and anxiety, has been deliberated (Zhang et al., 2000, 2005; O'Donnell and Zhang, 2004). In contrast, other reports have also suggested that PKA overactivation by genetically manipulating animals or with the use of PKA activating agents would lead to impaired memory formation or instrumental learning (Baldwin et al., 2002; Pineda et al., 2004). Analogous to the results of PKA overactivation, inhibition of apPDE expression also impaired the synaptic facilitation in the Aplysia sensory-to-motor synapse (Fig. 4D). Our hypothesis was that these conflicting results arose from the different effects of the acute and chronic inhibitions of PDE4 on the intracellular cAMP levels. Because dsRNA effectively inhibits gene expression in Aplysia neurons over an extended period (Lee et al., 2001, 2003) (supplemental Fig. 2, available at www.jneurosci.org as supplemental material), the inhibitory effect of apPDE dsRNA is expected to be chronic when compared with that of acute PDE4 inhibitor treatment.
Chronic administration of rolipram, a PDE4 inhibitor, blocked the PKA activation caused by 5-HT, indicating impaired elevation of the cAMP level. This inhibition was more effective when the treatment with rolipram was prolonged (Fig. 5C). Therefore, chronic inhibition of the PDE4 activity or PDE4 downregulation in the synaptic sites could impair the cAMP homeostasis required for synaptic plasticity. These results suggest that the acute treatment with a PDE4 inhibitor would be more effective than chronic treatment to cure memory diseases or dementia.
In summary, we characterized the critical role of apPDE in cAMP kinetics during the synaptic facilitation process. apPDE downregulated the PKA activity in the synaptic region by reducing the cAMP level and showed the impaired synaptic facilitation when its normal expression level was disrupted. Moreover, our results suggest that the acute and chronic PDE4 inhibitions could lead to conflicting clinical effects in the treatment of memory-related diseases. To determine the precise molecular mechanisms governing cAMP kinetics and PDE4, future studies should be performed to characterize how each PDE4 isoform can differentially regulate the cAMP-dependent signaling and modulate the synaptic plasticity and memory.
This work was supported by the Korea Ministry of Science and Technology under National Research Laboratory Program Grant M1-0104-00-0140 and Brain Research Center–Frontier Program Grant M103KV010012 03K2201 01210 and by the Marine and Extreme Genome Research Center Program, Ministry of Marine Affairs and Fisheries, Republic of Korea.
Correspondence should be addressed to Dr. Bong-Kiun Kaang, National Research Laboratory of Neurobiology, Institute of Molecular Biology and Genetics, School of Biological Sciences, College of Natural Sciences, Seoul National University, San 56-1 Silim-dong Kwanak-gu, Seoul 151-742, Korea. E-mail:.
Copyright © 2005 Society for Neuroscience 0270-6474/05/259037-09$15.00/0