The Type and the Localization of cAMP-dependent Protein Kinase Regulate Transmission of cAMP Signals to the Nucleus in Cortical and Cerebellar Granule Cells*

cAMP signals are received and transmitted by multiple isoforms of cAMP-dependent protein kinases, typically determined by their specific regulatory subunits. In the brain the major regulatory isoform RIIβ and the RII-anchor protein, AKAP150 (rat) or 75 (bovine), are differentially expressed. Cortical neurons express RIIβ and AKAP75; conversely, granule cerebellar cells express predominantly RIα and RIIα. Cortical neurons accumulate PKA catalytic subunit and phosphorylated cAMP responsive element binding protein very efficiently into nuclei upon cAMP induction, whereas granule cerebellar cells fail to do so. Down-regulation of RIIβ synthesis by antisense oligonucleotides inhibited cAMP-induced nuclear signaling in cortical neurons. Expression in cerebellar granule cells of RIIβ and AKAP75 genes by microinjection of specific expression vectors, markedly stimulated cAMP-induced transcription of the lacZ gene driven by a cAMP-responsive element promoter. These data indicate that the composition of PKA in cortical and granule cells underlies the differential ability of these cells to transmit cAMP signals to the nucleus.

expression (1)(2)(3). Multiple isoforms of PKA are determined by their specific regulatory subunits. Four regulatory subunits (RI␣, RI␤, RII␣, and RII␤) have been cloned. PKA containing RII␤ is the predominant PKA isoform in the brain and is expressed in the cortex, whereas in the brainstem and cerebellum (except Purkinje cells) RII␤ has not been found (4 -6). In mammalian brain, signals triggered by cAMP are targeted to specific effector sites by the tethering of cAMP-dependent protein kinases to intracellular compartments (4,7,8). PKAII is bound to membranes via specific anchor proteins (AKAPs), which bind R subunits. Bovine brain AKAP75 has been studied as prototype of kinase A anchor protein and shares high homology with human AKAP79 and rat AKAP150 (9,10). These proteins have similar properties, related sequences, and are recognized by the same antibodies (9 -11). AKAP150/75 and RII␤ are co-localized in the dendritic cytoskeleton and perikarya of forebrain neurons; both proteins have not been found in cerebellar granule cells (6).
Although the structure and expression pattern of the PKA regulatory subunits and AKAPs are well documented, the functional role of these proteins in the transduction of cAMP signals is still poorly understood. It is not known how the different PKA isoforms in different districts of the central nervous system receive and transmit cAMP signals.
We have chosen primary cortical and granule cerebellar neurons as prototype cells with different PKA composition and localization. PKA in cortical neurons is mainly of II␤ type and is membrane-anchored by AKAP150/75. Conversely, granule cerebellar cells do not express AKAP75/150 and RII␤. The R subunits expressed by these cells are RI␣ and RII␣ (6,12).
We have studied the activation by cAMP of these enzymes and the transmission of the signals to the nucleus by measuring the accumulation of C-PKA in the nucleus, CREB phosphorylation, and the transcription of a cAMP-induced promoter following cAMP stimulation. Also, we have manipulated the composition of PKA in granule and cortical cells by downregulating RII␤ in cortical cells or by expressing AKAP75 and RII␤ in granule cells, respectively.
The results presented here indicate that RII␤ and the PKAanchor protein, AKAP75, amplify the transmission of cAMP signals to the nucleus and suggest that the composition of PKA might influence the ability of the cell to receive and transmit cAMP signals to the nucleus. MATERIALS (13). Briefly, cortices of 16-day-old rat embryos were dissected and incubated with papain. Tissue fragments were mechanically dissociated and the cells plated in polylysine-coated dishes in 1:1 minimum Eagle's medium/F12 medium containing 2 mM glutamine and 10% fetal calf serum. Cerebellar granule cells were obtained from 7-day-old rat pups. Tissue fragments were digested with trypsin and mechanically dissociated. The cells were plated in polylysine-coated dishes in 25 mM K ϩ BME containing 10% fetal calf serum. 24 h after plating, 10 M Cytosine C Arabinoside was added to the cultures to prevent the growth of non-neuronal cells. Under these conditions the contamination of glial cells, measured by staining with glial fibrillary protein, was less than 10%. The cells were grown for 7 days and stimulated with 10 M forskolin and 0.5 mM IBMX (Sigma) or increasing concentrations of dibutyryl-cAMP (Calbiochem). PKA inhibition was achieved by treating the cells with 10 M of the PKA inhibitor H89 (Biomol, Plymouth Meeting, PA) (14). We have also found that a 2-h treatment with PKI-amide (50 g/ml) (Life Technologies, Inc.) inhibits 50% of PKA activity in vivo.
PKA Assay-PKA was measured by phosphorylation of the synthetic peptide, kemptide, in the presence of [ 32 P]ATP (DuPont) (3,000 Ci/ mmol, final specific activity 125-150 cpm/pmol) with 5 g of cytoplasmic extracts in the presence or absence of 0.5 mM cAMP for 10 min of incubation at 30°C. PKA activity was expressed in picomoles of phosphorylated kemptide/min/g of cytoplasmic or nuclear proteins (15,16). C-PKA/total PKA ratio is a measure of PKA activation. The activation of cytoplasmic PKA is shown as the ratio between C-PKA (ϪcAMP) and total PKA (ϩ0.5 mM cAMP). Nuclear PKA catalytic activity was measured in sucrose gradient-fractionated nuclei in the presence or absence of 10 M of the PKA-specific inhibitor pseudosubstrate (PKI) (16).
Treatment of the Cells with Specific Antisense Oligonucleotides-Antisense oligonucleotides specific to the NH 2 terminus of rat RII␤ protein (5Ј-CCGCGGGATCTCGATGCTCA-3Ј) (GenTech, France) or mismatched phosphorotioate oligonucleotides were directly added to the culture medium for 72 h (addition every 24 h, 6 M final concentration). Each experimental condition was performed in triplicate. Cell viability was determined by trypan blue staining. The cells showed 90 -95% viability after 72 h of treatment with oligonucleotides.
Immunoblot Analysis-50 g of cytoplasmic or 30 g of nuclear proteins were resolved in 10% SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose filters. The filters were rinsed in Tris-buffered saline with Tween 20 (10 mM Tris-HCl, pH 8, 150 mM NaCl, and 0.05% Tween 20) and with 10% nonfat dry milk in the same buffer with Tween 20. The filters were incubated with specific antibodies in 5% nonfat dry milk for 1 h (RII␣, RII␤) or 6 h (CREB, PCREB). After washing, the filters were incubated with peroxidase-conjugated monoclonal anti-rabbit IgG (Sigma) in 5% nonfat dry milk in Trisbuffered saline with Tween 20 for 1 h and developed using the Renaissance chemiluminescence kit (NEN Life Science Products). Anti-rat PCREB antibodies (17) were purchased from Upstate Biotechnology, Lake Placid, NY. Specific anti-RII␤ or anti-RII␣ antibodies were generated by immunizing rabbits with a synthetic RII␤ peptide (peptide 31-57 from the AUG of the rat sequence) or RII␣ (peptide containing the residues 53-73 from the start codon of the rat protein) cross-linked to soybean trypsin inhibitor. The total IgGs were purified, and the specificity of each preparation was tested by immunoprecipitation, immunofluorescence, and immunoblot by preadsorbing the antibodies to the specific peptides or control peptides (12,18).
Microinjection of DNA Expression Vectors-Cerebellar granule cells were grown for 7 days and injected with the DNA solutions in phosphate buffer (25 ng/ml of each plasmid). The plasmid vectors used were: CMV-GFP (CLONTECH); RSV-lacZ; CRE-lacZ containing five CRE elements and the vasoactive intestinal peptide promoter driving lacZ gene (19,20). In some experiments plasmid vectors expressing C-PKA were used (21,22). AKAP plasmids contained the sequence of AKAP75 or AKAP45 driven by the cytomegalovirus promoter (23). All plasmids preparations were tested in the rat thyroid cell line, FRTL-5, and in the PC12 cell line by stable transfection. The expression of the specific proteins was measured by immunoanalyses (immunoblot, immunofluorescence, and immunoprecipitation) and Northern blot (18,24).
The injection apparatus consisted of a phase contrast microscope connected to a computer-aided image analyzer (AIS automatic injection system, Zeiss, Germany) and a computer-operated microinjector (Eppendorf, Germany). In each experiment 150 cells/dish and 2 dishes/ DNA were injected. 18 h after the injection the cells were stimulated with 0.1 mM forskolin in the presence of 100 mM IBMX for 4 h, washed in phosphate-buffered saline solution and fixed with 4% paraformaldehyde for 30 min. Cells were permeabilized with 0.1% Triton X-100 and after extensive washes were incubated first with the monoclonal anti-␤-galactosidase antibody (Sigma) and then with fluorescein-tagged goat anti-mouse IgG antibody (Sigma) in phosphate-buffered saline containing 0.2% skin porcine gelatin for 30 min at room temperature. In all experiments, the fluorescein-tagged anti-mouse IgG antibodies were injected alone as control of the microinjection procedures. The specificity of anti-␤-galactosidase antibodies was tested by omitting the first antibody. Under our conditions the efficiency of injection was about 20%.

Differential Response to cAMP of PKA in Granule Cells and
Cortical Neurons-The binding of cAMP to the inactive PKA tetrameric holoenzyme induces its dissociation, thereby releasing active catalytic subunits in the cytoplasm. We have studied cAMP signaling in primary cultures of cortical neurons and cerebellar granule cells by measuring the cytoplasmic and nuclear PKA catalytic activity following cAMP stimulation for various periods of time. We have used forskolin, a stimulator of adenylyl cyclase, and IBMX, the inhibitor of cAMP phosphodiesterases, to maintain constant cAMP levels. Fig. 1A shows the time course of cytoplasmic PKA activity in cortical and granule cells stimulated by 10 M forskolin in the presence of 0.5 mM IBMX. In granule cells, cAMP-induced PKA activity was marked and persistent. In cortical cells PKA activation was lower, peaking at 15 min of cAMP stimulation and returning to the basal value within 20 min. The activation of PKA in granule cerebellar cells reached the plateau between 1 and 5 min and remained constant up to 30 min after the initial cAMP stimulation. The cAMP dose-response curve indicated that the cerebellar granule cells PKA was activated at lower cAMP concentration relative to PKA in cortical neurons (Fig. 1B).
Nuclear C-PKA accumulation, a sensitive marker of cAMP stimulation, was very efficient in cortical cells compared with granule cerebellar cells (Fig. 1C), although the cytoplasmic PKA activation was similar in the two cell types (Fig. 1D). The cortical enzyme, albeit dissociated poorly, contributed significantly to the total mass of nuclear PKA because the absolute amount of cytoplasmic PKA was higher in cortical than in granule cerebellar cells (see the legend of Fig. 1).
These data indicate that the granule cells PKA, stimulated by cAMP, dissociates efficiently, but does not transmit C-PKA to the nucleus. In contrast, cortical cells respond very efficiently to cAMP with a significant increase of nuclear C-PKA activity.
We suggest that the composition of PKA in the two cell types might be the cause of the different responses to cAMP. Indeed, the regulatory RII␤ type is abundantly expressed in cortical cells, is membrane-bound, binds cAMP with lower affinity and has a longer half-life compared with RI␣ and RII␣ (8,12,15,18,25). These features suggest that PKAII␤ might be the sensor of continuous and persistent cAMP signals.
CREB Phosphorylation in Granule and Cortical Cells-The biologically relevant effect of PKA translocation to the nucleus is the regulation of gene expression mediated by the phosphorylation of the nuclear transcription factor CREB. CREB is bound to a specific DNA sequence motif (CRE) and, upon phosphorylation by PKA catalytic subunit (26) or Ca 2ϩ /calmodulindependent protein kinase IV (27), binds the adaptor proteins (CBP and p300). The association of CREB with these proteins facilitates the assembly of the transcriptional machinery and leads to the activation of cAMP-induced genes (28). We assayed CREB phosphorylation in cortical and granule cells stimulated with forskolin by immunoblot of nuclear proteins with antibodies that recognize the phosphorylated form of CREB (17). Fig. 2 shows that the phosphorylation of CREB (PCREB) was strongly induced by forskolin (10 M, 10 min) in cortical neurons. In granule cells the PCREB signal was detectable in basal conditions but was only slightly induced by forskolin. Forskolin did not induce the de novo synthesis of CREB, because the total amount of the protein was not modified by forskolin treatment (Fig. 2). The phosphorylation of CREB was induced by a cAMPdependent pathway, because it was mimicked by the cAMP analogue, 8-bromo-cAMP and was inhibited by 10 M of the cell-permeable PKA inhibitor, H89 (14) (data not shown).
Because the PCREB signal detected by Western blot might originate from contaminating glial cells present in our cultures, we have investigated the effect of cAMP-elevating agents in pure type I astrocyte cultures. Under the conditions described in Fig. 2, stimuli-dependent cAMP accumulation did not induce CREB phosphorylation (data not shown).
Inhibition of RII␤ Expression Impairs Nuclear Response to cAMP in Cortical Neurons-To test whether the expression of PKAII␤ affects CREB phosphorylation in cortical cells, we treated cortical neurons with specific anti-RII␤ antisense oligonucleotides for 72 h (see "Materials and Methods"). This treatment specifically reduced RII␤ protein levels (over 50% reduction, Fig. 3 lane 2), but did not affect the other membranebound PKA regulatory subunit RII␣, as shown by the immunoblot with specific anti-RII␤ or RII␣ antibodies (Fig. 3). The reduction of RII␤ protein was also detected by the ligand binding assay or overlay (23,29) in RII␤ antisense-treated cell extracts. Also, total PKA activity was not affected by anti-RII␤ antisense treatment (data not shown). RII␤ levels did not change in cells treated with mismatched oligonucleotides (Fig.  3, lane 1 and legend).
Cortical cells treated with the RII␤ antisense oligonucleotide were stimulated with forskolin and CREB phosphorylation as-  sayed as described in Fig. 2. In these cells, forskolin-induced CREB phosphorylation was significantly reduced, whereas total CREB concentration was not affected (Fig. 4, upper panel). The decreased cAMP-induced CREB phosphorylation in anti-RII␤ oligonucleotide-treated cells was accompanied by a parallel reduced accumulation of C-PKA in the nucleus (Fig. 4, lower  inset).
These data indicate that RII␤ levels influence cAMP-induced nuclear C-PKA accumulation and CREB phosphorylation and suggest that the efficient transmission of cAMP signals to the nucleus in cortical cells might be dependent on the abundance of PKAII␤. This observation is consistent with the finding that neural gene expression and c-fos induction by cAMP are defective in specific areas of the central nervous system in RII␤Ϫ/Ϫ mice (30,31).

Expression of RII␤ and AKAP75 Enhances cAMP-induced Transcription in Granule
Cells-To test the hypothesis that membrane-bound PKAII␤ plays a crucial role in the cAMPinduced transcription, we microinjected granule cells with expression vectors containing the cDNAs encoding RII␤, the specific brain RII-anchor protein, AKAP75, and the green fluorescent protein, GFP. Because RII␤ and AKAP75 are not expressed in granule cells (8,12), their synthesis should result in the formation of new PKAII␤ bound to AKAP75 as in cortical cells. We first assayed the expression of RII␤ and AKAP75 in microinjected cells. Fig. 5 shows the staining of microinjected cells with specific antibodies to AKAP75 (1) and to RII␤ (2 and 3). Both AKAP75 and RII␤ were efficiently expressed in microinjected granule cells. The cells injected with RII␤ expression vector only showed a diffuse cytoplasmic staining (Fig. 6A), whereas the cells microinjected with both AKAP75 and RII␤ expression vectors showed a single RII␤ spot-shaped signal in the cell body (Fig. 6), likely corresponding to the cytoskeletal and perinuclear localizations of RII␤ in neural cells (6).
To test the effects of the expression of RII␤ and AKAP75 on cAMP-induced transcription, a reporter gene (lacZ) driven by five tandemly linked cAMP-responsive elements, (5ϫ CRE-lacZ) was co-injected with the plasmids indicated above. Cells were stimulated for 4 h with 1 M forskolin, and the expression of the lacZ gene under control of the CRE promoter was monitored by immunofluorescence with anti-␤-galactosidase antibodies. Fig. 6 shows a representative picture of granule cells microinjected with the combination AKAP75-RII␤ and CRE-lacZ genes. Treatment with forskolin for 4 h significantly increased the number of lacZ positive cells. Table I shows that cells microinjected with a control plasmid carrying the lacZ fused to a constitutive non-cAMP-dependent promoter (the long terminal repeats of Rous sarcoma virus, RSV-lacZ) efficiently synthesized ␤-galactosidase. Cells microinjected with the CRE-lacZ construct did not show ␤-galactosidase signal following forskolin stimulation. Injection of RII␤ or AKAP75 expression vectors stimulated CRE-lacZ expression (3 to 6 and 2.8 to 5 positive cells). Co-injection of RII␤ and AKAP75 expression vectors resulted in a marked increase in the number of lacZ-expressing cells (4 to 14 positive cells, Ϫ or ϩ forskolin, respectively). The number of positive cells was dramatically reduced if a mutant version of AKAP75 was coinjected with RII␤, AKAP45, which binds RII but fails to localize it to the membranes (18,32,33) or with a vector expressing PKI, the specific PKA inhibitor (34). Moreover pretreatment of microinjected cells with 10 M H89, a PKA inhibitor (14), resulted in a significant reduction of ␤-galactosidase-expressing cells (data not shown). The data, derived from six independent FIG. 3. Treatment with RII␤ antisense oligonucleotides reduces RII␤ content in cortical neurons. Representative immunoblots of total cellular extracts from cortical neurons treated with mismatched (1) or anti-RII␤ antisense oligonucleotides (2) using specific anti-RII␣ (upper panel) or anti-RII␤ (lower panel) specific antibodies (see "Materials and Methods"). The lower graph shows the relative amount of RII␣ and RII␤ derived from the densitometric analysis of the immunoblots in three experiments. The treatments with specific oligonucleotides are described under "Materials and Methods." RII␤ and RII␣ content was not influenced by treatment with mismatched oligonucleotides. In some experiments we noticed a 25% reduction of both RII proteins in cells treated with nonspecific oligonucleotides.

FIG. 4. Down-regulation of RII␤ reduces phosphorylated CREB and C-PKA in the nuclei of cortical neurons. Top panels,
immunoblot of phosphorylated CREB in nuclear extracts of cortical neurons treated with mismatched oligonucleotides (1) or RII␤ specific antisense (2) and stimulated with 10 M forskolin (FSK) as described in Fig. 2. Total CREB is shown in the lower blot obtained from the same cell extracts. Lower panels, nuclear C-PKA accumulation in cortical cells treated with mismatched oligonucleotides (1) or RII␤ specific antisense (2) and stimulated with (ϩ) or without (Ϫ) 10 M forskolin (FSK) as described in Fig. 2. The basal nuclear C-PKA in mismatched or RII␤ oligonucleotide-treated cells was 2.8 Ϯ 0.5 pmol/min/g in cortical cells and 2.7 Ϯ 0.4 pmol/min/g, respectively.
experiments, indicate that the combination of RII␤-AKAP expression vectors significantly increased the number of ␤-galactosidase positive cells in the presence of cAMP. DISCUSSION Targeting protein kinases and phosphatases in proximity of their substrates represents an important mechanism to convey intracellular signals to specific cellular sites (4,7). Two different protein families bind PKA to the cell compartments: the Akinase anchor proteins AKAPs and the microtubule-associated proteins MAP2. AKAPs bind the regulatory subunit RII␤ with nanomolar affinity and localize PKA to the dendritic cytoskeleton and other cellular compartments (Golgi apparatus, primary branches of dendrites and perikarya), whereas MAP2 predominantly binds the RII␣ isoform (5). AKAP 150/75, the rat or the bovine PKA anchor protein, can also bind RII␣ but with lower affinity (35)(36)(37). The levels of RII␤ and AKAP 75 differ in cell populations of brain areas and their expression pattern is strictly correlated. Accordingly, in cortical neurons, which express AKAP 75, RII␤ is the most abundant PKAregulatory subunit; conversely, in cerebellar granule cells AKAP 75 and RII␤ has not been found (5,8,12). Because the amounts of RII␣ and MAP2 are comparable in both cell populations (8,12), we decided to test if the expression of RII␤-AKAP might interfere with the cytosolic and nuclear responses to cAMP in cortical neurons and cerebellar granule cells. PKA activity and CREB phosphorylation were measured as functional correlates of cAMP stimulation. Although it has been shown that CREB is also a substrate for other protein kinases (27,38), we have analyzed only cAMP-induced events.
The response to cAMP signals was different in granule cerebellar and cortical cells. In granule cerebellar cells, cytoplasmic PKA dissociated very efficiently, but nuclear C-PKA accumulation induced by cAMP was rather weak. Cortical cells, on the other hand, did not activate efficiently cytoplasmic PKA but accumulated PKA in the nuclei in response to cAMP. The different PKA composition and localization in the two cell types might account for these different responses. Granule cells contain mainly type I and type II␣ PKA, which bind cAMP with higher affinity compared with type PKAII␤ (15,18,25). These subunits are soluble (RI␣) or partly soluble (RII␣) in the cytosol (10,12). PKAI and PKAII␣ dissociate efficiently at low cAMP levels (Fig. 1B). PKA present in cortical cells, composed mainly of type II␤ isoenzyme, dissociates at high cAMP levels and rapidly reassociates (Fig. 1, A and B). Membrane-bound PKAII␤ might be the sensor of persistent cAMP signals and the preferential source of nuclear C-PKA in cortical cells. This interpretation is strengthened by the observation that the cAMPdependent nuclear signaling is inhibited in cortical cells with lower RII␤ levels (Fig. 4) or in the RII␤-defective mouse (30,31). We have reconstituted the "cortical" PKA by expressing RII␤ and AKAP75 in the granule cells by microinjection of specific expression vectors. AKAP75 and RII␤ plasmids were efficiently expressed alone or in combination in microinjected granule cerebellar cells (Fig. 5). The expression of AKAP75 localized RII␤ in discrete spots inside the cell body close to the nucleus (perikaryon) (Fig. 6B). It is worth noting that expression of AKAP75 in non-neuronal cells localizes RII under the plasma membrane (33,39). Although, we did not identify conclusively the specific compartment in microinjected granule cells, the data shown indicate that different partners interacting with AKAP75 are present in neuronal and non-neuronal cells.
cAMP-induced transcription was stimulated by RII␤ or AKAP75, but the simultaneous injection of both genes markedly stimulated cAMP-induced lacZ expression (Table I). The stimulatory effect on transcription was inhibited by PKA inhibitors (Table I and Fig. 7) and dependent on AKAP75 membrane-binding domain, because a mutant lacking the segment of the protein, which localizes AKAP75 to the membrane, inhibited the RII␤ stimulatory effect (Table I). The expression of both RII␤ and AKAP generates high levels of membrane-bound PKAII␤, that ultimately might be responsible for the stimulation of cAMP-dependent transcription. There is evidence indicating that membrane-bound PKAII is more stable than cytosolic PKAII, which can be rapidly activated and degraded. Overexpression of AKAP75 increased RII␤ levels in kidney cells (33). Conversely, cytosolic translocation of PKAII, attained by expression of AKAP45 decreased RII␤ levels in thyroid, kidney, and PC12 cells (18,23,24). 2 Reposition of PKAII in the membranes might protect the enzyme from degradation and might reduce the basal PKA activation (33). 3 We suggest that membrane-bound PKA, releasing high levels of C-PKA in proximity of the nuclear envelope, can promote the entry of C-PKA into the nucleus. We do not exclude, however, that membrane-bound PKA could also modulate a Ca 2ϩ -dependent component of CREdriven gene expression. In cerebellar granule cells the low expression of RII␤ and AKAP 150/75 probably underlies the weak cAMP-stimulated transcription and suggests that other mechanisms could be responsible for CRE-driven gene expression in these cells (40).
Increasing evidence indicate that CREB is a multifunctional transcription factor that can be activated by cAMP and Ca 2ϩdependent transduction pathways (41,42), and its activation is critical for long-term memory formation in different biological systems (43)(44)(45)(46). Stimuli that generate long-lasting long term potentiation have been shown to induce CRE-mediated gene expression that was reduced by L-type Ca 2ϩ channel blockers (14), but the cAMP pathway appears to be necessary for the cellular processes related to long-term memory (20,47).
As to the biological correlates of our findings, we suggest that the differential composition and localization of PKA in discrete populations of neurons might explain the different activation of cAMP-induced transcription in these cells by the same type of signal. Consistent with our data, it has been found that in RII␤ defective mouse gene induction by cAMP in striatal neurons is significantly inhibited (31).