Phosphoinositide-dependent Kinase Phosphorylation of Protein Kinase C Apl II Increases during Intermediate Facilitation inAplysia *

Phosphorylation of protein kinase Cs (PKCs) by phosphoinositide-dependent kinase I (PDK) is critical for PKC activity. In the nervous system of the marine molluskAplysia, there are only two major PKC isoforms, the calcium-activated PKC Apl I and the calcium-independent PKC Apl II, and both PKCs are persistently activated during intermediate memory. We monitored the PDK-dependent phosphorylation of PKC Apl I and PKC Apl II using phosphopeptide antibodies. During persistent activation of PKCs in Aplysia neurons, there is a significant increase in the amount of PDK-phosphorylated PKC Apl II in the particulate fraction but no increase in the amount of PKC Apl I phosphorylated by PDK. PDK phosphorylation of PKCs was not sensitive to inhibitors of phosphatidylinositol 3-kinase, PKC, or expression of a kinase-inactive PDK. Localization of PDK-phosphorylated PKC Apl II using immunocytochemistry revealed an enrichment of phosphorylated PKC Apl II at the plasma membrane. These data suggest that increased PDK phosphorylation of PKC Apl II is important for persistent kinase activation.

In the marine mollusk, Aplysia californica, PKCs 1 are important for both short and intermediate term changes in synaptic strength between sensory and motor neurons that accompanies behavioral sensitization (1). Following behavioral sensitization, or prolonged treatment of ganglia with serotonin (5-HT), the kinase activity in the particulate fraction of both Ca 2ϩ -activated conventional PKC Apl I and Ca 2ϩ -independent novel PKC Apl II is increased (2,3). Persistent PKC activity is required for the maintenance of synaptic facilitation under some conditions (4). The mechanisms for the persistent activation of the PKCs are not well defined. Both initial PKC activation and protein translation are required for persistent activation of the PKCs (2). However, there are differences in the persistent activation of the two isoforms. The increase in PKC Apl II activity appears to be mainly because of autonomous activation of the kinase, whereas the increase in PKC Apl I appears to be mainly due to increased levels of regulated PKC Apl I on the membrane (2,3). There is increased phosphorylation of PKC Apl I at a conserved autophosphorylation site during intermediate memory; however, the phosphorylated kinase is located exclusively in the cytoplasm and presumably does not contribute to the increase in particulate kinase activity (5). Increased phosphorylation of PKC has been postulated to underlie persistent activation of PKCs during vertebrate learning models (6 -8), and PKC activity is regulated by phosphorylation (9).
All PKCs require phosphorylation at the activation loop site by phosphoinositide-dependent kinase 1 (PDK) for full catalytic activity (10). Ca 2ϩ -activated PKCs are phosphorylated by PDK soon after translation (11)(12)(13). After PDK-dependent phosphorylation, Ca 2ϩ -activated PKCs undergo two major autophosphorylations that are required for the stability and folding of the enzyme, a process termed maturation (9). Once these sites are phosphorylated, the PDK site can become de-phosphorylated without affecting the stability of the enzyme (12). Because PDK docks at a C-terminal site that is no longer easily accessible in the mature enzyme, classical PKCs are not easily re-phosphorylated (14). Thus, whereas overexpression of PDK or PDK dominant-negatives affect the proportion of PKC found in the mature fraction, it does not affect the amount of mature PKC phosphorylated at the PDK site (12), presumably due to the lack of re-phosphorylation.
PDK phosphorylation of novel PKCs has also been examined, although a detailed model for their phosphorylation has not been proposed. In contrast to Ca 2ϩ -activated PKCs (13), PDK phosphorylation of PKC␦ is regulated both by PKC autophosphorylation and by PI 3-kinase activity (15). Regulation of phosphorylation of other kinases by PDK appears to be dependent on substrate conformation and subcellular localization as opposed to activation of PDK itself (16 -18).
To examine the role of PDK in the regulation of PKC in Aplysia, we raised phosphopeptide antibodies to the PDK site in the classical PKC Apl I and the novel PKC Apl II. We then used the antibodies to characterize PDK phosphorylation of PKCs in the Aplysia nervous system. Our results demonstrate differences in PDK regulation of classical and novel PKCs in the nervous system and suggest an important role for PDK phosphorylation of PKC Apl II during intermediate facilitation.

EXPERIMENTAL PROCEDURES
Isolation of Nervous System-A. californica (50 -250 g) were obtained from Marine Specimens Unlimited at Pacific Palisades, CA, and were maintained in an aquarium for at least 3 days before experimentation. The animals were first placed in a bath of isotonic MgCl 2 /artificial seawater (1:1, v/v) and then anesthetized by injection of isotonic MgCl 2 . Pleural and pedal ganglia were isolated from the animal and pinned to silicone plastic in ice-cold dissecting medium (2,24). The ganglia were then desheathed in order to facilitate penetration of pharmacological agents and incubated in resting medium (2, 24) for 3 h at 15°C to minimize effects of dissection. In some experiments bag cell clusters were isolated from abdominal ganglia and treated in a similar fashion.
Antibodies-Peptides containing the conserved PDK site in PKC Apl I and Apl II ( Fig. 1) with the serine converted to phosphoserine were synthesized (Quality Controlled Biochemicals, Hopkington, MA). The peptides were conjugated to bovine serum albumin maleimide (Pierce) via the endogenous cysteine and injected into rabbits three times at 4-week intervals. In order to reduce the fraction of antibody that might recognize the non-phosphorylated peptide, the serum from animals was passed over an affinity column of the non-phosphorylated form of the immunizing peptide coupled to Sulfolink (Pierce). This process was repeated twice. After each passage, specifically retained antibodies were eluted from the column. The final flow through was then passed over an affinity column of the immunizing phosphopeptide coupled to Sulfolink, and retained antibodies were eluted and concentrated in a Centriplus-10. Antibodies to total PKC Apl I, total PKC Apl II, and PDK have been described (23,25).
Experiments in Sf9 Cells-High titer stocks of baculovirus encoding PDK, PDK (K-N), PKC Apl I, PKC Apl II, and PKC Apl II K-R (23,26,27) were used at a multiplicity of infection of 5 for each virus. Super-FIG. 1. Characterization of phosphopeptide antibodies. A, sequences of the PDK phosphorylation site for the Aplysia PKCs and their closest vertebrate homologue. The peptide used to raise the antibodies is in bold (pT stands for phosphothreonine). Note that although the sequence C-terminal to the PDK site is absolutely conserved, the N-terminal sequence is unique for each isoform. B, bacterially expressed GST-PKC Apl II catalytic subunit fusion protein or MBP-Apl I catalytic subunit fusion protein and nervous system extracts were separated on a 10% SDSpolyacrylamide gel, transferred to nitrocellulose, and probed with either the phosphopeptide antibody to the PDK site in PKC Apl II (anti-Thr(P)-561 (anti-Thr 561-P)) or the phosphopeptide antibody to the PDK site in PKC Apl I (anti-Thr-478 (anti-Thr 478-P)). The blots were then stripped and re-probed with either antibodies to the catalytic subunit of PKC Apl II or PKC Apl I. C, immunoprecipitation of phosphopeptide immunoreactivity. Aplysia-soluble nervous system extracts (350 g/tube) were immunoprecipitated with 20 l of serum (preimmune or the specific antibodies to PKC Apl I and PKC Apl II (25). The immunoprecipitates and voids (1:20 of total) were separated on 9% SDS-polyacrylamide gels, transferred to nitrocellulose, and probed with either the phosphopeptide antibody to the PDK site in PKC Apl II (anti-Thr(P)-561) or the phosphopeptide antibody to the PDK site in PKC Apl I (anti-Thr-478). The blots were then stripped and re-probed with either antibodies to the catalytic subunit of PKC Apl II or PKC Apl I. D, SF9 cell extracts separated into supernatant (s) and pellet (p) and immunoblotted with the phosphopeptide antibody to the PDK site (Anti-Thr 561-P), stripped and re-probed with the antibody to total PKC Apl II. E, comparison of PKC Apl II and PKC Apl II (K-R) either using normal procedures (ϩ), or homogenizing in the absence of phosphatase inhibitors (Ϫ). F, percent change in PDK phosphorylation of PKC Apl II or PKC Apl II (K-R) when homogenized in the absence of phosphatase inhibitors in the homogenization solution. (n ϭ 5, *, p Ͻ 0.01 two-tailed Student's t test between PKC Apl II and PKC Apl II K-R). Error bars are S.E. natant and membrane fractionation was as described (27).
GST Pull-downs-The GST-PDK and GST-PDK K-N constructs have been described (23). We expressed these constructs in bacteria and purified the GST fusion proteins on glutathione beads. The beads were then incubated with supernatants of Sf9 cell extracts either infected with baculovirus encoding PKC Apl I or PKC Apl II. The beads were incubated overnight and then washed three times with PBS. The beads were then eluted with sample buffer and separated on SDS gels. For experiments in ganglia, 250 g of soluble nervous system extract (5) was incubated for 4 h with beads and then washed three times with homogenization buffer before elution of the beads with sample buffer.
Constructs-The fusion protein used to generate non-phosphorylated PKC Apl I was made by inserting a BamHI fragment of PKC Apl I into pMALC (New England Biolabs, Beverly, MA) cut with BamHI. The fusion protein starts with the sequence GSLSF from the hinge region and includes the native C-terminal tail. The fusion protein used to generate non-phosphorylated PKC Apl II was made by first cutting out a fragment encoding the catalytic fragment from PKC Apl II with NheI and EcoRI. The ends of this fragment were filled in with Klenow and ligated to the pGEX3.1 vector that had been cut with EcoRI and the ends filled in with Klenow. The sequence of PKC Apl II starts with ASNEH and includes the native C-terminal tail. The PDK and PDK (K-N) constructs used to make baculovirus have been described (23). To generate an expression construct for Aplysia neurons we cut out PDK (K-N) from the BB4 vector with Xho and Kpn and inserted it into the GFP-PKC Apl II C2 pNEX-3 vector (27) cut with Xho and Kpn. This replaced the C2 domain of PKC with the PDK sequence to generate an enhanced GFP-PDK fusion protein starting with SRPGG from Aplysia PDK. This removes the first 130 amino acids of Aplysia PDK; this sequence is not conserved over evolution and does not contain any region of the kinase or PH domain.
Experiments in Ganglia-The protocol for inducing intermediate facilitation was identical to that described previously (2,3,5). Autonomous kinase activity on the membranes was measured as described (3). Immunoprecipitations were done as described (25).
Immunoblotting-Western blots were performed as described (28) with the antibody to phosphorylated PKC Apl II at 1 g/ml, the antibody to phosphorylated PKC Apl I at 1 g/ml, the antibody to total PKC Apl I at 2.5 g/ml, and the antibody to total PKC Apl II at 1 g/ml. The phosphopeptide antibodies were pre-blocked with the non-phosphorylated peptide at a molar ratio of 50:1. Goat anti-rabbit, horseradish peroxidase-conjugated secondary antibody (Pierce) was used at 0.5 g/ ml. Results were visualized by enhanced chemiluminescence (Renaissance Plus, PerkinElmer Life Sciences). Immunoblots were scanned, and analysis was performed using the public domain NIH Image program (developed at the National Institutes of Health and available on the Internet at rsb.info.nih.gov/nih-image/). We calibrated our data with the uncalibrated OD feature of NIH image that transforms the data using the formula (y ϭ log10(255/(255 Ϫ x))), where x is the pixel value (0 -254). Control experiments demonstrated that after this calibration, values were linear with respect to amount of protein over a wide range of values.
Quantitation of Immunoblots-For each blot probed with both phosphopeptide-specific and total antibodies a phospho-ratio was calculated (immunoreactivity of anti-phosphopeptide antibody compared with immunoreactivity of total antibody). This ratio, although not comparable between different blots, was useful for determining differences in phospho-ratio between supernatant and pellet or between treatments. When examining treatments (e.g. 5-HT, Bis, etc.), the percentage change was calculated (((experimental phospho-ratio Ϫ control phospho-ratio)/control phospho-ratio)⅐100). When comparing supernatant to pellet a simple ratio was calculated (pellet phospho-ratio/supernatant phosphoratio). Statistical tests involved paired two-tailed Student's t tests between control and experimental conditions or when appropriate unpaired two-tailed Student's t tests. To determine whether the phospho-

FIG. 2. PDK binds to PKC Apl II.
A, Sf9 cell cytosolic extracts from cells expressing PKC Apl II were incubated with glutathione beads and GST or GST-PDK. The beads were washed and then eluted with sample buffer. Ten percent of the initial homogenate (Start), 10% of the unbound material (Void), and the eluates from the GST-PDK beads and the GST beads were separated on a 9% SDS-polyacrylamide gel, transferred to nitrocellulose, and first Ponceau S-stained to visualize the levels of GST fusion proteins. These are the single bands seen on the blots because the GST fusion proteins are added in excess. Next the blot was probed with the antibody to the PDK site in PKC Apl II (anti-Thr(P)-561 (anti-Thr 561-P)), stripped, and re-probed with the total antibody to PKC Apl II. B, the percentage of the total PKC Apl II and the percentage of the PKC Apl II phosphorylated at the PDK site (Apl II Thr(P)-561 ((Apl II T561-P))) bound to GST-PDK beads was quantitated (n ϭ 8, p Ͻ 0.01 two-tailed Student's paired t test between PKC Apl II and PKC Apl II Thr(P)-561). Error bars are S.E. C, nervous system-soluble extracts (250 g) were incubated with glutathione beads and GST or GST-PDK. The beads were washed and then eluted with sample buffer. Five percent of the unbound material (Void) and the eluate from the GST-PDK beads and the GST beads were transferred to nitrocellulose and probed with the antibody to the PDK site in PKC Apl II (anti-Thr(P)-561), stripped, and re-probed with the total antibody to PKC Apl II. D, the percentage of the total PKC Apl II and the percentage of the PKC Apl II phosphorylated at the PDK site (Apl II Thr(P)-561 ((Apl II T-561-P))) bound to GST-PDK beads was quantitated (n ϭ 4, p Ͼ 0.5 two-tailed Student's paired t test between PKC Apl II and PKC Apl II Thr(P)-561). Error bars are S.E. ratio for pellet/supernatant was different from 1, a one-sample Student's t test was performed. To correlate phospho-ratio and autonomous activity, for each blot/kinase assay the maximum value for the phosphoratio or autonomous activity was set to 1, and other values were calculated relative to the maximum value. This allows comparison of multiple experiments on the same axis.
Injection of Sensory Neurons with GFP-PDK(K-N)-The protocol was identical to that described previously (27) with the following modification. Cells were plated on poly-L-lysine plates with 20% hemolymph to increase process outgrowth. Cells were left for 3-5 days to allow adequate expression before fixation.
Immunocytochemistry-Cultures were rinsed with artificial seawater and then fixed for 30 min in 4% paraformaldehyde in PBS with 30% sucrose. Cells were then permeabilized in 0.1% Triton X-100 in PBS with 30% sucrose for 10 min and washed 3 times with PBS. Free aldehydes were quenched with a 15-min incubation in 50 mM ammonium chloride in PBS. Nonspecific antibody binding was inhibited by incubating with a blocking solution (10% normal goat serum, 0.5% Triton X-100, in PBS) for 1 h. Cultures were incubated overnight with the primary antibody. For the chicken/rabbit PKC co-staining, 1:600 chicken anti-PKC and 1:1000 rabbit anti-PKC in blocking solution was used. The cultures were washed 4 times for 10 min with PBS and were then incubated with secondary antibody in blocking solution (1:150 CY2 anti-chicken and 1:400 CY3 anti-rabbit) for 1-2 h. For all other experiments, the rabbit phosphopeptide antibody was preabsorbed using the nonphosphorylated form of the peptide at 1 g/l for 30 min prior to use and then was added to the cultures in a 1:475 dilution in blocking solution.  Thr 478-P))). The blots were stripped and re-probed with the phosphopeptide antibody to PKC Apl II (anti-Thr(P)-561 ((Anti Thr 478-P))), stripped again, and re-probed with the antibodies to total PKC Apl I and total PKC Apl II simultaneously. Samples of GST-Apl II catalytic subunit and MBP-Apl I catalytic unit were also separated on the gel to ensure the phospho-specificity of the antibodies. Results from two separate animals are shown (a and b). B, the phospho-ratio (See "Experimental Procedures") was calculated for the supernatant, pellet, and total fractions for PKC Apl II and PKC Apl I. The percentage change in this ratio by 5-HT was then measured (n ϭ 12, *, p Ͻ 0.05, Student's paired t test between phospho-ratio in control and 5-HT-treated ganglia). C, the proportion of kinase found on the pellet was calculated for each antibody. Then the percentage change in this proportion caused by 5-HT was calculated (n ϭ 12, *, p Ͻ 0.05, Student's paired t test between proportions on pellet in control and proportions on pellet in 5-HT). D, the phospho-ratio in supernatant and pellet was compared (Pellet/Sup) for PKC Apl II and PKC Apl I before and after 5-HT (n ϭ 12, *, p Ͻ 0.05 one-sided Student's t test compared with value of 1; **, p Ͻ 0.05, Student's paired t test between control and 5-HT-treated ganglia). Error bars are S.E.

FIG. 4. Correlation of autonomous kinase activity and phosphorylation at the PDK site.
Pleural-pedal ganglia treated with control or serotonin were assayed for both autonomous activity and phosphorylation at the PDK site. Autonomous activity (pmol/min/mg) and phospho-ratio were first calculated. For each experiment, the maximum value for each measure was set to 1, and the relative values or autonomous activity and phospho-ratio were calculated. Results are from two separate experiments, n ϭ 8/experiment). The correlation between the two values is highly significant (r 2 ϭ 0.58, p Ͻ 0.01).
LSM 510 software. The mean pixel intensity of boxed plasma membrane areas (2-3/cell) and boxed cytoplasmic areas (2-3/cell) were quantitated. Aplysia neurons have highly infolded membranes due to their large size, and thus even in a confocal section (ϳ15 M) the plasma membrane is not well defined. We used a box (ϳ1-2 M wide) as a rough definition of plasma membrane-associated area. Efforts were made to use confocal images for quantitation from the mid-point of cells for standardizing between different cells. Occasionally, pigment granules could be seen in the sensory neuron cell bodies that were visible using the secondary antibody used for the total PKC Apl II antibody (CY5). Areas including these granules were not used in the quantitation. For each cell the average pixel intensity was calculated for cytosol and membrane areas for both total and anti-phosphopeptide antibodies. A phospho-ratio was then calculated for supernatant and membrane, and then the membrane/cytosol phosphate ratio for each cell was calculated.
To examine the effect of cells injected with GFP-PDK (K-N), we compared the average of the mean cytosolic intensity, the mean membrane intensity, and the phospho-ratio for the expressing cells and non-expressing cells on the same coverslip to calculate an effect of GFP-PDK (K-N) for each experiment.

RESULTS
Characterization of Phospho-specific Antibodies to the PDK Site-The threonine that is phosphorylated by PDK in vertebrate PKCs is well conserved in the Aplysia PKCs (Fig. 1A). We thus raised antibodies to phosphopeptides containing phosphothreonine in this position for both PKC Apl I and PKC Apl II. We biased the peptides used for immunization to the divergent region N-terminal of the PDK site in order to make antibodies that are specific for individual PKC isoforms and that do not recognize other PDK-phosphorylated proteins (Fig. 1A). The resulting antibodies were highly specific for PKC recognizing single bands from nervous system extracts that co-migrated with the bands recognized by our well characterized antibodies to total PKC Apl I or PKC Apl II (2, 25) (Fig. 1B). To evaluate the specificity of the antibodies for phosphorylated PKC, we blotted bacterially expressed PKC catalytic domain as a control because there is no PDK in bacteria, and thus the bacterially expressed PKC will not be phosphorylated at the PDK site. This demonstrated that the antibodies are highly specific for phosphorylated PKCs as they do not react with PKC expressed in bacteria (Fig.  1B). The bands recognized by the phosphopeptide antibodies are solely due to immunoreactivity to PKC as they can be completely immunoprecipitated by antibodies to total PKC (Fig. 1C).
A Kinase-dead PKC Has Less Stable Phosphorylation at the PDK Site-Phosphorylation at the PDK site is not due to au- FIG. 5. Effects of inhibitors on basal PDK phosphorylation of PKC site. A, ganglia were treated with either resting medium or resting medium ϩ 1 M bisindolylmaleimide (Bis) for 1 h and then homogenized and separated into supernatant (S) and pellet (P) fractions. These were separated on SDS-polyacrylamide gels and blotted with the antibody to the phosphopeptide antibody to the PDK site of PKC Apl II (anti-Thr(P)-561 (Anti Thr 561-P)). The blot was stripped and blotted with the antibody to total PKC Apl II. B, the blot was stripped and blotted with the antibody to the phosphopeptide antibody to the PDK site of PKC Apl I (anti-Thr(P)-478 (Anti-Thr 478-P)) and then stripped and blotted with the antibody to total PKC Apl I. C, the effect of Bis on the total phospho-ratio was determined. D, the proportion of kinase found on the membrane was calculated for each probed antibody, and then the percentage change in this proportion caused by 5-HT was calculated (n ϭ 4, *, p Ͻ 0.05 two-tailed Student's paired t test between control and Bis-treated ganglia). E-H, same as A-D but with 1 M LY294002 (LY) instead on Bis. Error bars are S.E. tophosphorylation, because a kinase-inactive PKC (27) was well phosphorylated at the PDK site when expressed in Sf9 cells (Fig. 1D). However, the overall level of phosphorylation at the PDK site in the kinase-inactive PKC was less than that for the wild-type kinase expressed in Sf9 cells (Fig. 1D). This decrease may be due to increased de-phosphorylation of the kinase inactive PKC Apl II. Indeed, phosphorylation at the PDK site was much more sensitive to phosphatases in the homogenization buffer for the kinase-inactive PKC Apl II K-R than for wild-type PKC Apl II (Fig. 1E; quantitated in Fig. 1F). This is consistent with previous evidence from classical PKCs that PKC autophosphorylation is required to cause a conformational change that protects the PDK site from de-phosphorylation (11).
PDK Binding to PKCs-In many cases PDK binds to its substrates. To evaluate whether PDK could bind to PKC Apl II, we first examined whether GST-PDK could pull down Sf9 cell-expressed PKCs. PKC Apl II was pulled down by the GST-PDK fusion protein but was not pulled down by GST alone (Fig.  2). Interestingly, the PKC phosphorylated at the PDK site was not bound as well as unphosphorylated PKC ( Fig. 2A, quantitated in Fig. 2B). Similar results have been observed for PKC␤II where PDK immunoprecipitated de-phosphorylated PKC better than phosphorylated PKC (14). One explanation for this result is that PDK binds to the C-terminal region of PKCs and that this binding site is not accessible in the mature folded form of the kinase (14). Thus, a higher proportion of PKC Apl II that was unphosphorylated might be in an immature, unfolded state and so more likely to have its C-terminal region accessible for binding to PDK. Interestingly, when we repeated these experiments using endogenous PKC Apl II from the nervous system, the proportion of phosphorylated and total PKC Apl II bound to GST-PDK was similar (Fig. 2C, quantitated in Fig.  2D). This may be due to the much lower proportion of immature unfolded kinase in the mature nervous system, compared with when PKC Apl II is overexpressed in the baculovirus system.
PDK Phosphorylation of PKC Apl II but Not Apl I during Intermediate Facilitation-The mechanism underlying persistent activation of PKCs is not known, but increased phosphorylation of PKCs by PDK is one possibility. We thus examined whether PDK phosphorylation of PKCs was modulated during the persistent activation of the PKCs stimulated by prolonged treatment with the facilitating transmitter, 5-HT (2). Indeed, there was a significant increase in the percentage of PKC Apl II phosphorylated at the PDK site in the pellet fraction 2 h following a 90-min treatment of Aplysia ganglia with 5-HT (Fig. 3,  A and B; p Ͻ 0.01). There was no change in the percentage of soluble PKC Apl II phosphorylated at the PDK site or in the percentage of soluble or particulate PKC Apl I phosphorylated at the PDK site (Fig. 3, A and B). As observed previously (2), there was a significant increase in the percentage of PKC Apl I in the particulate fraction (Fig. 3, A and C; p Ͻ 0.02). Although there was no significant difference in the percentage of total PKC Apl II in the particulate fraction, the percentage of PDKphosphorylated PKC Apl II in the particulate fraction did increase (Fig. 3, A and C).
In control ganglia, PKC Apl II that was not phosphorylated by PDK was mainly in the pellet fraction (i.e. the ratio of PDK phosphorylation was higher in the supernatant) (Fig. 3D). After 5-HT treatment, however, there was a shift in this ratio toward PDK phosphorylation in the pellet (Fig. 3, C and D). In contrast, PKC Apl I had approximately equal phosphorylation at the PDK site in supernatant and pellet before and after 5-HT treatment (Fig. 3, C and D).
PDK Phosphorylation Is Correlated with Autonomous Enzymatic Activity-Autonomous kinase activity is also increased by 5-HT during intermediate facilitation, and this activity derives from PKC Apl II (3). In a separate series of experiments, we determined both the ratio of phosphorylation of PKC Apl II at the PDK site and autonomous kinase activity from the pellets of ganglia treated with 5-HT (Fig. 4). There was a very good positive correlation between the relative levels of autonomous activity and the relative phospho-ratio (r 2 ϭ 0.58, p Ͻ 0.01, n ϭ 16). This result suggests a relationship between PDK phosphorylation and autonomous activity of PKC Apl II.
Neither Inhibitors of PKC nor Inhibitors of PI 3-Kinase Affect PDK Phosphorylation-To explore mechanisms that regulate the ability of PDK to phosphorylate PKC in the Aplysia nervous system, we examined whether PDK phosphorylation of PKC was regulated by either PKC activity or by PI 3-kinase activity FIG. 6. Immunocytochemistry of PKC phosphorylated at the PDK site. A, cultured Aplysia neuron double-stained with an antibody to PKC Apl II raised in rabbits; B, antibody to PKC Apl II raised in chickens. C, cultured Aplysia sensory neuron double-stained with a phosphopeptide antibody to the PDK site of PKC Apl II raised in rabbits; D, total antibody to PKC Apl II raised in chickens. E, axon and giant growth cone of an Aplysia neuron double-stained with a phosphopeptide antibody to the PDK site of PKC Apl II raised in rabbits; F, total antibody to PKC Apl II raised in chickens. N represents the nucleus, and GC represents a growth cone; scale bars are all 10 M. G, quantitation of the membrane/cytosol phospho ratio (see "Experimental Procedures"). *, p Ͻ 0.05, one-sample t test against 1, n ϭ 12. Error bars are S.E. as has been noted for Ca 2ϩ -independent PKCs in vertebrates (15). Neither inhibitors of PKC nor PI 3-kinase affected the level of phosphorylation at the PDK site (Fig. 5). Inhibitors of PKC phosphorylation did lead to an increased level of PKC on the membrane for both PKC Apl I and PKC Apl II consistent with PKC activity being required for the release of PKC from the particulate fraction (5,29,30) (Fig. 5, A and B). We also examined whether insulin, which increases the activity and amount of PKC Apl II on membranes of bag cell neurons in a wortmanninsensitive manner (31), led to increases in PDK phosphorylation of PKC Apl II. No significant changes were seen in the levels of PKC Apl II phosphorylated at the PDK site after insulin treatment (12 Ϯ 20%; change in phospho-ratio of pellet fraction by insulin, n ϭ 7, S.E., p Ͼ 0.5, two-tailed Student's t test).
Immunocytochemical Localization of Phospho-PKC Apl II-The antibodies to phosphorylated PKC also worked well in immunocytochemistry of cultured Aplysia neurons. To compare the localization of phosphorylated and non-phosphorylated PKC, we raised an antibody to total PKC Apl II in chickens. This antibody had similar specificity as the previously generated rabbit antibody raised to the same epitope and was indistinguishable in immunocytochemistry (Fig. 6, A and B). By using the chicken antibody, we were able to label PKC Apl II phosphorylated at the PDK site and total PKC Apl II in the same cells. Comparing the two staining patterns revealed a significant increase in the relative immunoreactivity of phospho-PKC Apl II on or near the plasma membrane of cell bodies (see Fig. 6, C and D; quantitated in Fig. 6G). No difference between the antibodies was seen in axons, varicosities, or growth cones (Fig. 6, E and F).
To determine whether we could control phosphorylation of PKC by PDKs using a dominant-negative strategy, we examined the ratio of staining for phospho-PKC and total PKC in neurons that were injected with a plasmid encoding a kinaseinactive PDK. The PDK was N-terminally tagged with GFP to monitor its level of expression. There was no significant change in the ratio of PDK-phosphorylated PKC Apl II to total PKC Apl II in the cytoplasm or membrane-associated regions between cells expressing or not expressing the catalytically inactive PDK (Fig. 7). Thus, at the levels of expression we can attain in neurons, kinase-inactive PDK does not appear to act as a dominant-negative to block PKC phosphorylation by PDK. DISCUSSION We generated phosphopeptide antibodies to the conserved PDK site in PKC Apl I and PKC Apl II. These antibodies are specific for PKCs phosphorylated at the PDK site as they do not recognize PKCs expressed in bacteria (where no PDK-like activity is present). Furthermore, the bands recognized by the phosphopeptide antibodies can be completely immunoprecipitated by other antibodies to PKC. The major result from this study is that phosphorylation at the PDK site of PKC Apl II is increased after prolonged 5-HT treatment and is correlated with autonomous PKC Apl II activity. In contrast, the increased activation of PKC Apl I seen at this time is not correlated with increased phosphorylation at the PDK site. We also localized PDK-phosphorylated PKC Apl II using immunocytochemistry and documented an enrichment of phosphorylated PKC Apl II at or near the plasma membrane.
Phosphorylation of Ca 2ϩ -independent or novel PKCs at the PDK site appears to be more regulated than that of conventional PKCs. In vitro phosphorylation of the novel PKC␦ by PDK requires PtdIns(3,4,5)P 3 and phorbol esters (15), whereas in vitro phosphorylation of Ca 2ϩ -activated PKCs does not (12). Initial phosphorylation of PKC␦ in vitro may resemble rephosphorylation of conventional PKCs because PKC␦ is partially active when expressed in bacteria even though the PDK site is not phosphorylated (15). Thus, PKC␦ can partially fold even in the absence of PDK phosphorylation (32). PKC Apl II resembles PKC⑀ more than PKC␦; however, there has been FIG. 7. Effect of PDK (K-N) expression in Aplysia neurons. A cultured sensory neuron was injected with a plasmid encoding PDK (K-N) tagged with GFP at its 5Ј end (arrow). This neuron was adjoined by a neuron that was not expressing PDK (K-N) (arrowhead). The neurons were double-stained with a phosphopeptide antibody to the PDK site of PKC Apl II raised in rabbits (A) or the total antibody to PKC Apl II raised in chickens (B). The GFP signal was seen only in the injected neuron (C). D, quantitation of the membrane/cytosol phospho-ratio of cells either expressing (GFP-PDK K-N) or not expressing (Control, same as Fig. 5G); *, p Ͻ 0.05, Student's one-sample t test against 100%, n ϭ 12. Both groups had a membrane/cytosol ratio significantly different from 1, but there was no effect of expressing GFP-PDK (K-N). E, in each experiment (n ϭ 3) the percentage change of the average phospho-ratio for all the GFP-expressing cells (n ϭ 1-4) to the average phospho-ratio for all the non-expressing cells (n ϭ 2-7) was calculated for both the cytosol and the membrane regions. Scale bars are 10 M. Error bars are S.E. little characterization of PKC⑀ phosphorylation by PDK. In contrast to PKC␦, there is a high basal level of PKC Apl II phosphorylation at the PDK site, and this basal level was not sensitive to PI 3-kinase inhibitors or PKC inhibitors. Thus, it is possible that phosphorylation of PKC Apl II at the PDK site may more closely resemble the constitutive phosphorylation of conventional PKCs (15). An important caveat is that we are examining PKC in primary nervous system tissue where there is probably considerable activation of most signal transduction pathways during the period of time before we gain experimental control of the ganglia. Thus, our results show that continued presence of phosphorylation of PKC at the PDK site does not require PI 3-kinase or PKC activity, but our results do not address how the initial phosphorylation of PKC Apl II at the PDK site is regulated in the nervous system. PDK phosphorylation of PKC Apl II does appear to be more regulated than PDK phosphorylation of PKC Apl I. First, PKC Apl II that is not phosphorylated at the PDK site has a distinct subcellular localization. By using biochemical techniques, it is enriched in the pellet fraction (Fig. 3D), and by using immunocytochemical techniques, it is partially excluded from the plasma membrane (Fig. 6). These results suggest that either de-phosphorylation of PKC Apl II at the PDK site affects PKC localization or that a substantial pool of endogenous PKC Apl II is not phosphorylated at the PDK site during maturation. This appears distinct from PKC Apl I where PKC Apl I is not differentially found in the pellet based on its phosphorylation at the PDK site (Fig. 3D). Second, there is increased phosphorylation of PKC Apl II but not PKC Apl I in the pellet fraction during intermediate facilitation (Fig. 3B). There was no significant change in levels of total PKC Apl II in the pellet, suggesting that this increase was due to increased phosphorylation of the PDK site. The mechanisms that regulate this phosphorylation are unclear. Notably, a percentage of particulate PKC Apl II is autonomously active at this time (3), and phosphorylation at the PDK site may be enhanced by the absence of a pseudosubstrate in the active loop. This idea is supported by the strong correlation between autonomous kinase activity and the percentage of PKC phosphorylated at the PDK site (Fig. 4). At this point it is not clear if the increased PDK phosphorylation is important in causing autonomous activation of the kinase, or a consequence of the autonomous activation. Unfortunately, because the kinase-dead PDK did not act as a dominant-negative and because there are no specific inhibitors of PDK, we cannot at this time determine whether PDK is required for activation of PKC during intermediate facilitation.
By using immunocytochemistry we found that PKC Apl II phosphorylated at the PDK site is enriched near or on the plasma membrane. This suggests that the non-phosphorylated pool of PKC is excluded from the plasma membrane. This is apparently inconsistent with our finding that there is a pool of particulate PKC that is not phosphorylated (Fig. 3D). The particulate pool of PKC not phosphorylated at the PDK site may be associated with the cytoskeleton or an internal membrane fraction and thus may be measured in our immunocytochemical measurements as cytosolic. Indeed, a large percentage of particulate PKC Apl II in Aplysia is Triton-insoluble (33).
Whereas PDK phosphorylation of PKCs has been well studied in vitro and in cell lines, this is the first characterization of PDK phosphorylation in neurons. There were important differences in the PDK phosphorylation of the classical PKC Apl I and the novel PKC Apl II in Aplysia neurons. These differences could be explained by postulating that classical PKCs are only phosphorylated at the PDK site during initial formation of the kinase, whereas novel PKCs may be regulated by re-phosphorylation of this site during the kinase life cycle. Phorbol esters stimulate de-phosphorylation of classical PKCs that directly leads to degradation of the kinase (34). In contrast, novel PKCs may be de-phosphorylated and stored as an inactive pool of enzyme that can be re-activated at a later time. Indeed, novel PKCs are less down-regulated by phorbol esters in many cell types (35)(36)(37)(38)(39), including Aplysia neurons (40) This is also the first time PDK phosphorylation has been imaged by immunocytochemistry. The relative increase in staining at membranes for PKC Apl II phosphorylated at the PDK site would be consistent with increased PDK phosphorylation of activated PKC. PDK phosphorylation of PKC Apl II may play a pivotal role in its activation during intermediate facilitation. The activation of PKC Apl II at this time requires translation (3). Understanding what regulates PDK phosphorylation should provide clues to the important proteins whose translation needs to be regulated to persistently activate PKC Apl II during intermediate facilitation.