Identification of conserved amino acids N-terminal of the PKC epsilon C1b domain crucial for protein kinase C epsilon-mediated induction of neurite outgrowth.

We have shown previously that protein kinase C (PKC) epsilon can induce neurite outgrowth independently of its catalytic activity via a region encompassing its C1 domains. In this study we aimed at identifying specific amino acids in this region crucial for induction of neurite outgrowth. Deletion studies demonstrated that only 4 amino acids N-terminal and 20 residues C-terminal of the C1 domains are necessary for neurite induction. The corresponding regions from all other novel isoforms but not from PKCalpha were also neuritogenic. Further mutation studies indicated that amino acids immediately N-terminal of the C1a domain are important for plasma membrane localization and thereby for neurite induction. Addition of phorbol ester made this construct neurite-inducing. However, mutation of amino acids flanking the C1b domain reduced the neurite-inducing capacity even in the presence of phorbol esters. Sequence alignment highlighted an 8-amino acid-long sequence N-terminal of the C1b domain that is conserved in all novel PKC isoforms. Specifically, we found that mutations of either Phe-237, Val-239, or Met-241 in PKCepsilon completely abolished the neurite-inducing capacity of PKCepsilon C1 domains. Phorbol ester treatment could not restore neurite induction but led to a plasma membrane translocation. Furthermore, if 12 amino acids were included N-terminal of the C1b domain, the C1a domain was dispensable for neurite induction. In conclusion, we have identified a highly conserved sequence N-terminal of the C1b domain that is crucial for neurite induction by PKCepsilon, indicating that this motif may be critical for some morphological effects of PKC.

The induction and elongation of neurites are cellular processes driven by cytoskeletal changes. These are under the control of different intracellular transduction pathways mediating signals from other cells or the extracellular matrix. The members of the PKC 1 family constitute one important family controlling the outgrowth of neurites. PKC isoforms have been suggested to both positively and negatively influence the outgrowth of neurites.
There are 10 different PKC isoforms that are divided into three subclasses according to their structure and requirements for activation, classical PKCs (␣, ␤I, ␤II, and ␥), novel PKCs (␦, ⑀, , and ), and atypical PKCs (/ and ). Of these isoforms, particularly PKC␦ (1-3) and PKC⑀ (3)(4)(5)(6)(7)(8) have been suggested to positively influence neurite outgrowth in several different cell types. However, there are also indications that PKC isoforms, for instance PKC⑀, can counteract outgrowth (9). Our group has shown previously that overexpression of PKC⑀ in neuroblastoma (7,8) and in immortalized neural precursor (3) cells leads to neurite outgrowth. A similar morphological effect of PKC⑀ has also been observed in fibroblasts (10,11). The neurite-inducing effect of PKC⑀ is independent of its catalytic activity, and a region from the regulatory domain of the enzyme encompassing the two C1 domains with flanking structures is necessary and sufficient for the effect (7).
Many proteins, besides PKC isoforms, contain C1 domains (12). These domains can roughly be subgrouped as typical C1 domains that bind phorbol esters and atypical C1 domains that do not bind phorbol esters (13). Classical and novel PKC isoforms contain two typical C1 domains. The structure of PKC␣ C1b (14), PKC␦ C1b (15), and PKC␥ C1b (16) domains has been determined, revealing that the C1 domain is a compact globular structure. The integrity of the domain depends on the coordinated binding of two Zn 2ϩ ions by conserved cysteine and histidine residues. On the tip of the domain is a diacylglycerol/ phorbol ester binding pocket located between two ␤ strands (15). The sides of the pocket are covered by hydrophobic residues, and binding of lipid generates a continuous hydrophobic surface that facilitates the insertion of the domain into the membrane. Besides the binding of lipids, C1 domains have been shown to mediate protein-protein interactions (17)(18)(19)(20)(21). Amino acids flanking the C1 domains have also been found to mediate interactions with other proteins. For instance, in PKC⑀ an F-actin-binding site is located C-terminal of the C1a domain (22).
One important function of C1 domains is likely the targeting of the protein to specific intracellular locations. Several C1 domain-containing proteins are localized to the Golgi apparatus, and this is mediated via a C1 domain (23)(24)(25)(26)(27)(28), and the C1a domain of PKC␤II has been shown to bind the centrosomal protein pericentrin (19). The fact that C1 domains both bind membranes and can mediate protein interaction implies that they also may serve as anchoring proteins at different membranes. We hypothesize that the PKC⑀ C1 domains induce neurites by interacting with one or several proteins and either anchor them at the membrane, perhaps as a signaling complex, or, if they are membrane-residing proteins, induce a conformational change in them.
To understand further such an interaction, this study was designed to identify a specific motif in the C1 region of PKC⑀ that is crucial for induction of neurite outgrowth. We show that a stretch of eight amino acids N-terminal of the C1b domain, which is evolutionarily conserved in all novel PKC isoforms, is critical for neurite induction. Furthermore, we identify in this amino acid sequence three hydrophobic residues, the mutation of which completely abolishes the neurite-inducing capacity of the PKC⑀ C1 domains.
To generate PKC⑀PSC1V3 with two C1a or two C1b domains, XhoI and MluI sites were inserted by PCR at both sides of the sequences encoding the C1a or C1b domains. The C1 domain-encoding sequence was thereafter inserted at the desired site in the PKC⑀PSC1V3-encoding sequence. Expression vectors encoding PKC⑀PSC1V3 fused to EGFP containing only one of the mutations, N-terminal or C-terminal of C1a or C1b, were generated by PCR-based cloning but introduced the restriction enzyme site on only one side of the C1 domain-encoding sequence.
The ExSite PCR-based site-directed mutagenesis kit (Stratagene) was used to replace the sequence QRFSVNMP in PKC⑀PSC1V3-EGFP with the scrambled sequence SMQPFNRV. The DpnI-treated PCR products were purified in 1% agarose gel and were thereafter ligated. All PCR products were sequenced to ensure that no mutations were introduced in the PCRs.
The scrambled sequence SMQPFNRV was also introduced in fulllength PKC⑀ by using the SacI and ScaI restriction enzyme sites to cleave out a PKC⑀ fragment containing the scrambled sequence. The fragment was thereafter ligated into the full-length PKC⑀-pEGFP vector.
The QuikChange site-directed mutagenesis kit (Stratagene) was used to introduce the R236A, F237A, V239G, N240G, and M241G mutations in the PKC⑀PSC1V3 sequence using the expression vector encoding PKC⑀PSC1V3-EGFP as template. The mutations were verified by sequencing. All primers are listed in Table I.
Cell Culture and Transfections-Human neuroblastoma SK-N-BE (2)C cells and SH-SY5Y cells stably transfected with cDNA encoding TrkA (30) were grown in minimum essential medium with Earle's salts and L-glutamine and supplemented with 10% fetal calf serum, 100 IU/ml penicillin, and 100 g/ml streptomycin. Rat pheochromocytoma PC12 cells were grown in RPMI 1640 medium with L-glutamine supplemented with 10% horse serum, 5% fetal calf serum, 100 IU/ml penicillin, and 100 g/ml streptomycin. All cell culture solutions were from Invitrogen. For transfections, cells were trypsinized and seeded on glass coverslips at a density of 200,000 (SK-N-BE(2)C cells) or 300,000 cells (SH-SY5Y/TrkA and PC12 cells) per 35-mm cell culture dish. The cells were transfected 24 h later using 2 l of Lipofectamine2000 (Invitrogen) and 1 g of DNA in 1 ml of serum-free medium according to the supplier's protocol. Six hours later, the transfection mixture was replaced with medium containing serum and antibiotics. Where indicated, cells were treated with 16 nM TPA (Sigma) or 50 M C 2 -ceramide (Sigma) for 17 h or 100 ng/ml NGF (Promega) for 2 days. Morphology Studies-Seventeen hours, or 2 days for experiments with NGF, after transfection, cells were fixed and mounted as described previously (7). Transfected cells, identified by the fluorescence of EGFP, were considered to have neurites if the process was longer than two cell bodies. For PC12 cells, processes longer than the length of one-half cell body were considered as neurites. 200 transfected cells per experiment were counted.
Western Blotting-Cells were seeded at a density of 2.5 ϫ 10 6 cells/ 100-mm culture dish or 1.2 ϫ 10 6 (SK-N-BE(2)C) or 2.2 ϫ 10 6 (SH-SY5Y/TrkA and PC12) cells/60-mm culture dish and transfected with 6 l of Lipofectamine2000 and 3 g of DNA in 3 ml of serum-free medium. The day after transfection cells were washed in phosphate-buffered saline and lysed in buffer (10 mM Tris, pH 7.2, 160 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 1 mM EGTA, 1 mM EDTA, Complete Protease Inhibitor Mixture (Roche Applied Science)) followed by centrifugation. Equal amounts of proteins were electrophoretically separated on a SDS-polyacrylamide gel and transferred to a polyvinylidene difluoride membrane (Millipore). Proteins were detected with primary antibodies toward GFP (1:250; Zymed Laboratories Inc.) and visualized with horseradish peroxidase-labeled secondary antibody (Amersham Biosciences) using the SuperSignal system (Pierce) as substrate. The chemiluminescence was detected with a CCD camera.
Immunofluorescence-Cells were fixed in 4% paraformaldehyde in phosphate-buffered saline for 4 min, permeabilized, and blocked with 5% normal goat serum and 0.3% Triton X-100 in Tris-buffered saline for 30 min. Syntaxin 6 was detected with a primary monoclonal antibody (Pharmingen) diluted 1:25 followed by a secondary Alexa Fluor 546conjugated antibody (Molecular Probes) diluted 1:400 in Tris-buffered saline. The coverslips were thereafter mounted as for morphology studies.
Confocal Microscopy-Live cells were examined the day after transfection. The coverslips were washed twice with buffer H (20 mM Hepes, 137 mM NaCl, 3.7 mM KCl, 1.2 mM MgSO 4 , 2.2 mM KH 2 PO 4 , 1.6 mM CaCl 2 , 10 mM glucose, pH 7.4) and mounted on the heated stage of a Nikon microscope. The localization of PKC⑀-EGFP was examined using a ϫ60 objective (NA 1.4) and a Bio-Rad Radiance 2000 confocal system with excitation wavelengths at 488 nm and emission filter 515HQ30 prior to and 1 min after addition 1 mM carbachol. Fixed cells were examined with the same settings with the modification that Alexa Fluor 546 was detected using 543 nm excitation and 600LP emission filter.

RESULTS
The Importance of the Pseudosubstrate and the V3 Region for Neurite Outgrowth Induced by the Regulatory Domain of PKC⑀-Our group has shown previously that overexpression of PKC⑀ induces neurite outgrowth in SK-N-BE (2) neuroblastoma cells via the regulatory domain and that a region of PKC⑀ encompassing the pseudosubstrate, the two C1 domains, and the V3 region (PKC⑀PSC1V3) is sufficient for this effect (7). To elucidate further which structures in PKC⑀ are necessary for neurite outgrowth, the importance of amino acids N-and Cterminal of the C1 domains was investigated. For N-terminal amino acids we studied PKC⑀ variants in which either all amino acids N-terminal of the C1a domain had been removed (PKC⑀C1V3) or in which four amino acids prior to the C1a domain were included (PKC⑀4ϩC1V3). SK-N-BE(2)C neuroblastoma cells, transfected with vectors encoding PKC⑀PSC1V3 or the truncated variants fused to EGFP, were cultured in the absence or presence of 16 nM TPA for 17 h. Thereafter the number of transfected cells with neurites was quantified (Fig. 1A). Removal of the pseudosubstrate resulted in a construct with markedly reduced capacity to stimulate neurite outgrowth, and inclusion of four amino acids N-terminal of the C1a domain generated a construct with an intermediate effect. However, TPA treatment of cells overexpressing the truncated mutants led to neurite outgrowth to a similar extent as for cells overexpressing PKC⑀PSC1V3. Thus, the necessity of the pseudosubstrate for optimal neurite induction can be compensated for by TPA treatment. Western blot on lysates from transfected cells confirmed the expression of proteins of proper size (Fig. 1B).
Next, the importance of the C-terminal part of the PKC⑀ regulatory domain for neurite outgrowth was investigated by gradually deleting parts of the V3 region. The PKC⑀PSC1V3 region contains 80 amino acids C-terminal of the C1b domain. Constructs encoding a PKC⑀ region beginning four residues N-terminal of the C1a domain and encompassing 39 (PKC⑀4ϩC1ϩ39), 20 (PKC⑀4ϩC1ϩ20), 12 (PKC⑀4ϩC1ϩ12), or 6 (PKC⑀4ϩC1ϩ6) residues C-terminal of the C1b domain were generated, and SK-N-BE(2)C cells were transfected with the vectors. Neurite outgrowth was prominent and similar in cells overexpressing PKC⑀4ϩC1V3, PKC⑀4ϩC1ϩ39, or PKC⑀4ϩC1ϩ20, whereas no neurite induction could be detected in cells overexpressing PKC⑀4ϩC1ϩ12 or PKC⑀4ϩC1ϩ6 ( Fig. 2A). Western blot on lysates of transfected cells demonstrated that non-neurite-inducing variants were not expressed at lower levels than neurite-inducing proteins (Fig. 2B), showing that the lack of neurite induction is not due to lower expression levels.
To investigate whether the PKC⑀4ϩC1ϩ12 region is capable of inducing neurite outgrowth when stimulated with TPA, cells overexpressing PKC⑀4ϩC1ϩ12, PKC⑀4ϩC1V3, or PKC⑀PSC1V3 were treated with TPA for 17 h after transfection (Fig. 2C). Still, the PKC⑀4ϩC1ϩ12 construct did not induce neurite outgrowth, suggesting that more than 12 residues Cterminal of the C1b domain are required and that 20 residues C-terminal of the C1b domain are sufficient for neurite induction.
To confirm that the PKC⑀4ϩC1V3 and PKC⑀4ϩC1ϩ12 are responsive to phorbol esters, cells expressing these EGFP fusion proteins were treated with TPA and subjected to analysis with confocal microscopy (Fig. 2D). TPA treatment led to a plasma membrane translocation of both proteins demonstrat- ing that they have retained their TPA sensitivity.
The C1 Domains of All Novel PKC Isoforms Are Capable of Inducing Neurites in Neuroblastoma Cells-We have seen pre-viously that the regulatory domains of all novel PKC isoforms to various extents have neurite-inducing capacity (3,7,8). This led us to investigate if the region corresponding to PKC⑀4ϩC1ϩ20 in other PKC isoforms also is capable of inducing neurite outgrowth. SK-N-BE(2)C cells were therefore transfected with vectors encoding EGFP fused to the C1 domains of PKC␣, -␦, -, and -flanked by 4 N-terminal and 20 C-terminal amino acids. Cells were grown for 17 h in the absence or presence of TPA and thereafter examined for neurite outgrowth (Fig. 3A). Overexpression of PKC␦4ϩC1ϩ20, but not the corresponding regions in PKC␣, -and -, led to neurite outgrowth. After TPA stimulation, overexpression of C1 domains from all novel isoforms induced neurites, whereas overexpression of the corresponding region from the classical isoform PKC␣ had no effect. The expression levels of the EGFPfused proteins were analyzed by Western blotting demonstrating that all proteins were expressed (Fig. 3C). However, the and variants were expressed at lower levels, and there is therefore a possibility that the C1 domains of these isoforms may induce neurites also in the absence of TPA, if expressed at higher levels. Thus, the C1 domains of all novel PKC isoforms, but not of PKC␣, have a capacity to induce neurite outgrowth.
Furthermore, we investigated if only one of the two PKC⑀ C1 domains is sufficient to induce neurites. For this purpose, we generated constructs encoding the isolated C1a or C1b domain including 4 residues N-terminal and 20 residues C-terminal of the C1 domain. Neuroblastoma SK-N-BE(2)C cells were transfected with the expression vectors and cultured in the absence or presence of TPA and examined for neurite outgrowth (Fig.  3B). Neither the isolated C1a nor the C1b domain could induce neurites by itself or after stimulation with TPA, indicating that overexpression of only one PKC⑀ C1 domain is not sufficient to induce neurites. This is not due to lower expression levels of the isolated C1 domains because they are expressed at higher levels than PKC⑀4ϩC1ϩ20 (Fig. 3C).
Amino Acids Flanking the C1b Domain Are Critical for Neurite Outgrowth-Our next approach was to investigate whether the C1 domains can be exchanged for each other or whether each C1 domain has unique properties of importance for the neurite-inducing effect. To test this we aimed at generating constructs encoding PKC⑀PSC1V3 with tandem C1a (PKC⑀PSC1aC1aV3) or C1b (PKC⑀PSC1bC1bV3) domains. For this purpose PCR primers were designed to introduce restriction enzyme sites on both sides of each C1 domain-encoding sequence (Fig. 4A). The cDNA encoding the C1 domain was thereafter inserted in the PKC⑀PSC1V3 sequence, from which the other C1 domain had been excised. To obtain proper controls, we also created PKC⑀PSC1aC1bV3 in which the restriction enzyme sites flanking the C1 domains were inserted. Neither the PKC⑀PSC1aC1aV3 nor the PKC⑀PSC1bC1bV3 construct had the capacity to induce neurites (not shown), but neither did control constructs containing the PKC⑀PSC1V3 region with the mutations resulting from insertion of the restriction enzyme sites flanking the C1 domains, PKC⑀PSC1V3(V166L/N167E, G221T/L222R) and PKC⑀PSC1V3-(N240L/M241E, V295T/D296R) (Fig. 4B). Thus, amino acids flanking the C1 domains seem to be critical for neurite induction.
To We have seen previously that the neurite induction of PKC⑀ C1 domains seems to depend on its plasma membrane localization (7,8), and we therefore examined the subcellular localization of the mutated PKC⑀PSC1V3 variants with confocal microscopy (Fig. 4C). PKC⑀PSC1V3 with mutations C-terminal of the C1a domain (G221T/L222R) had a similar plasma membrane localization as the wild-type. On the other hand, PKC⑀PSC1V3 mutated N-terminal of the C1a domain (V166L/ N167E) is to a large extent localized to the cytoplasm but a clear plasma membrane translocation can be induced by TPA. Thus, for these mutants localization to the plasma membrane seems to correlate to the neurite-inducing capacity of the mutant. It is therefore conceivable that the mutations suppress the neurite-inducing effect by altering the subcellular localization of the C1 domains.
In contrast to the mutations flanking the C1a domain, mutations either N-terminal or C-terminal of the C1b domain both led to constructs that completely lacked neurite-inducing activity (Fig. 4B) and for these mutants, neurite induction could not be restored by treatment with TPA. Both mutants had a perinuclear localization pattern (Fig. 4C). Treatment with TPA led to a distinct plasma membrane localization, but as mentioned above, this was not accompanied by neurite induction. Thus, it is likely that the amino acids mutated in these constructs are of importance for neurite outgrowth beyond affecting the localization of the protein.
All mutated constructs were expressed at similar levels except for PKC⑀PSC1V3 with mutations on both sides of the C1b domain (N240L/M241E and V295T/D296R) and PKC⑀PSC1V3 with mutation C-terminal of the C1b domain (V295T/D296R) that had lower expression levels (Fig. 4B). Therefore, it cannot be excluded that the lack of neurite induction by the latter construct is due to the fact that it is expressed at lower levels.
Identification of a Motif N-terminal of the C1b Domain Crucial for Neurite Outgrowth-The previous results indicate an important role for amino acids N-terminal of the PKC⑀C1b domain for PKC⑀-mediated neurite induction. A sequence alignment of amino acids N-terminal of the C1b domain of novel isoforms revealed that a stretch of residues N-terminal of the C1b domain is highly conserved among the human novel PKC isoforms and their Caenorhabditis elegans and Drosophila melanogaster counterparts (Fig. 5A). The analogous amino acids of the non-neurite-inducing PKC␣ have no similarity with this sequence (not shown). To investigate whether this amino acid sequence, which is conserved in all neurite-inducing isoforms, is crucial for neurite induction, we replaced it with a scrambled version in full-length PKC⑀ (PKC⑀-scr) and in PKC⑀PSC1V3 (PKC⑀PSC1V3-scr). This modification completely abolished the neurite-inducing capacity of both constructs (Fig. 5B), further highlighting the importance of amino acids N-terminal of the C1b domain for neurite outgrowth.
We next analyzed the role of PKC⑀ and the amino acids N-terminal of the C1b in neuronal differentiation driven by NGF. For this purpose we studied PC12 cells, in which PKC⑀ has been shown to potentiate NGF-stimulated neurite outgrowth (4) and SH-SY5Y neuroblastoma cells stably expressing TrkA (30) in which PKC⑀ has been indicated to play a role in NGF-stimulated neurite outgrowth (5,7,8). The cell lines were transfected with vectors encoding EGFP fusions of full-length PKC⑀ or PKC⑀PSC1V3 or their scrambled counterparts prior to treatment with NGF for 2 days (Fig. 5, C and D).
In PC12 cells (Fig. 5C), overexpression of either full-length PKC⑀ or PSC1V3 potentiated both basal and NGF-stimulated neurite outgrowth. In contrast, the scrambled variants did not cause such a potentiation. In SH-SY5Y/TrkA cells (Fig. 5D), only PKC⑀PSC1V3 potentiated basal and NGF-stimulated neu- rite outgrowth. In cells expressing scrambled full-length PKC⑀, the NGF-induced neurite outgrowth was actually significantly lower than in NGF-stimulated cells that expressed EGFP alone, indicating a dominant negative effect of the construct. A tendency to such suppression was also observed in cells expressing the scrambled PKC⑀PSC1V3 variant.
The scrambled PKC⑀PSC1V3 was consistently expressed at lower levels in the neuroblastoma cells (Fig. 5, B and D), which could explain its lower neurite-inducing capacity. However, its expression in PC12 cells and the expression of the scrambled full-length variant were not markedly lower than wild type variants. Furthermore, when point-mutated PKC⑀PSC1V3 variants were investigated (Fig. 7), these were expressed at the same level as wild type PKC⑀PSC1V3, and some of them lacked neurite-inducing capacity. Thus, the reduced neurite-inducing capacity of the mutants is most conceivably explained by an altered function of the constructs and not by reduced expression levels.
The findings raised the question if other properties of PKC⑀ are altered by the mutation. To examine whether this is the case, we chose to study the localization of PKC⑀-scr. Wild type PKC⑀ responds to carbachol stimulation of SK-N-BE (2)C cells with a sustained translocation to the plasma membrane (29). As shown in Fig. 6, A and B, PKC⑀-scr responds in a similar manner with a plasma membrane translocation following carbachol stimulation. Another typical property of PKC⑀ is the enrichment of the protein in perinuclear structures, presumably the Golgi network, in the presence of ceramide (31) (Fig.  6C). This localization pattern was also observed for PKC⑀-scr (Fig. 6D). We have seen that the Golgi localization is mediated by the C1b domain (28), and we therefore studied whether the PKC⑀PSC1V3-scr construct would be enriched in the Golgi complex (Fig. 6, E-G). The wild type PKC⑀PSC1V3 causes profound morphological changes of neuroblastoma cells and is primarily present in the plasma membrane (7), but the PKC⑀PSC1V3-scr shows a high degree of colocalization with syntaxin 6, a marker for the trans-Golgi network (37). Thus, scrambling of the eight amino acids N-terminal of the C1b domain does not negatively influence the localization of PKC⑀ to the Golgi complex. Taken together, these data demonstrate that several properties of PKC⑀, receptor-stimulated membrane translocation and enrichment in the Golgi complex, are not abrogated by scrambling of the motif. Thus, the motif is of specific importance for neurite outgrowth and does not lead to a general annihilation of PKC⑀ properties.
Point Mutation of PKC⑀ Phe-237, Val-239, or Met-241 Abolishes the Neurite-inducing Effect-In the conserved sequence N-terminal of C1b there are three completely conserved residues, Arg-236, Phe-237, and Pro-242 (numbering from human PKC⑀). Furthermore, for Val-239 and Met-241, there are analogous hydrophobic residues in all novel isoforms, and at the corresponding site of Asn-240 there is either an asparagine or an aspartate. Ser-238, on the other hand, is not conserved. We decided to explore the importance of each individual conserved or semi-conserved residue, but we excluded Pro-242 due to its immediate vicinity to the first cysteine of the C1b domain. By using site-directed mutagenesis, the coding sequence of PKC⑀PSC1V3 was altered to encode either alanine (for Arg-236 and Phe-237) or glycine (for Val-239, Asn-240, and Met-241). SK-N-BE(2)C cells were transfected with the plasmids and thereafter grown in the absence or presence of TPA (Fig. 7A). Overexpression of PKC⑀PSC1V3(R236A) and PKC⑀PSC1V3-(N240G) led to neurite outgrowth from the same number of cells as did overexpression of wild type PKC⑀PSC1V3. On the other hand, mutation of the Phe-237, Val-239, or Met-241 residues led to PKC⑀PSC1V3 variants that almost completely lacked neurite-inducing capacity, even after TPA treatment, suggesting that these residues are critically involved in neurite outgrowth induced by PKC⑀. All the mutated proteins were expressed at levels similar to that of the corresponding wildtype region (Fig. 7A).
All three mutants (PKC⑀PSC1V3 F237A, V239G, or M241G) were clearly enriched in the plasma membrane after TPA exposure (Fig. 7B), showing that the lack of neurite induction is FIG. 5. Identification of a motif N-ter-minal of the PKC⑀ C1b domain necessary for the capability of PKC⑀ to induce neurite outgrowth. A, sequence alignment of the region between the PKC C1a and C1b domains in Homo sapiens PKC␦, PKC⑀, PKC, and PKC and the corresponding PKC isoforms in C. elegans and D. melanogaster. The stretch of eight conserved residues immediately N-terminal of the C1b domain are marked with a red frame. B, SK-N-BE(2)C cells were transfected with vectors encoding wild type fulllength PKC⑀, PKC⑀PSC1V3, or mutated variants with a scrambled version of the stretch of eight conserved amino acids Nterminal of C1b (PKC⑀-scr and PKC⑀PSC1V3-scr) fused to EGFP. A vector encoding only EGFP was used as control (con). Cells were grown in medium in the absence or presence of 16 nM TPA for 17 h after transfection. Cells were thereafter fixed and mounted and scored for neurites. C and D, PC12 (C) and SH-SY5Y/TrkA (D) cells were transfected with the vectors encoding wild type full-length PKC⑀ (⑀FL), PKC⑀PSC1V3 (⑀C1), or mutated variants PKC⑀-scr (⑀FL-scr) and PKC⑀PSC1V3-scr (⑀C1-scr) fused to EGFP. A vector encoding only EGFP was used as control. Cells were treated with 100 ng/ml NGF for 2 days after transfection, and transfected cells were thereafter scored for neurites. Data are mean Ϯ S.E. (n ϭ 3) and expressed as percentage of transfected cells with neurites (B and C) or as the percentage of transfected cells with neurites minus the percentage of untreated EGFP-expressing cells with neurites (D). The latter calculation was done to adjust for the large variability in base-line neurite outgrowth between experiments, which was 33.0 Ϯ 11.5% for untreated SH-SY5Y/TrkA cells. *, significantly different from NGF-treated EGFP-expressing cells (p Ͻ 0.05) using analysis of variance followed by Duncan's multiple range test. Western blots using anti-GFP antibodies are included adjacent to each graph indicating expression levels of the EGFP fusion proteins. Arrows indicate the bands representing full-length PKC⑀ and PKC⑀PSC1V3 EGFP fusion proteins. not due to these mutated variants being resistant to TPAmediated plasma membrane localization.
The C1a Domain Is Dispensable for Neurite Outgrowth-We then speculated that the C1b domain alone is capable of inducing neurites if more than four amino acids N-terminal of the domain are included. To test this hypothesis, expression vectors were generated encoding the C1b domain including 22 (PKC⑀22ϩC1bϩ20) or 12 (PKC⑀12ϩC1bϩ20) amino acids Nterminal of the domain. Twenty amino acids were included C-terminal of the C1b domain. SK-N-BE(2)C cells were transfected with the expression vectors and grown in the absence or presence of TPA. The effects of these proteins on neurite outgrowth were compared with the effect of a construct containing both the C1a and the C1b domain (PKC⑀4ϩC1ϩ20) or the single C1b domain with only four amino acids N-terminal of the C1 domain (PKC⑀4ϩC1bϩ20) (Fig. 8). The neurite-inducing capacity of both PKC⑀22ϩC1bϩ20 and PKC⑀12ϩC1bϩ20 was reduced compared with the construct with both C1 domains, but following stimulation with TPA they induced neurites. The expression vectors encoding PKC⑀22ϩC1bϩ20, PKC⑀12ϩ C1bϩ20, and PKC⑀4ϩC1bϩ20 were all expressed at higher levels than PKC⑀C1ab, indicating that the reduced capacity of the constructs to induce neurites is not due to lower expression levels (Fig. 8). Thus, the C1a domain is needed for optimal induction of neurite outgrowth, but the C1b domain alone has a significant neurite-inducing effect if 12 amino acids N-terminal of the domain are included in the construct. DISCUSSION This study identifies an evolutionarily conserved motif in novel PKC isoforms that is crucial for the ability of PKC⑀ to induce neurite outgrowth. The PKC isoforms are serine/threonine kinases that presumably mediate their effects by phosphorylation of target proteins and thereby alter their functions. However, there are also several studies indicating that PKC isoforms have biochemical or cellular effects that are independent of their catalytic activity. The induction of neurites is one such kinase-independent effect of PKC⑀ (7), but PKC effects independent of catalytic activity also include suppression of sulfation of glucosaminoglycans in the Golgi apparatus by PKC⑀ (25), activation of phospholipase D by PKC␣ (32), inhi-  (2)C cells expressing PKC⑀-scr fused to EGFP were treated with 1 mM carbachol, and the localization of PKC⑀-scr-EGFP was followed with confocal microscopy. Images were taken prior to (A) and 60 s (B) after carbachol addition. C and D, SK-N-BE(2)C cells expressing wild type PKC⑀ (C) or PKC⑀-scr fused to EGFP (D) were grown in the presence of 50 M C 2 -ceramide for 17 h after transfection. Cells were fixed and mounted, and the localization of the PKC⑀ variants was examined with confocal microscopy. E-G, SK-N-BE(2)C cells expressing PKC⑀PSC1V3-scr fused to EGFP were subjected to immunofluorescence using primary monoclonal mouse antibody against the trans-Golgi marker syntaxin 6 and secondary antibody Alexa Fluor 546-conjugated goat anti-mouse and analyzed with confocal microscopy. Images demonstrate the localization of PKC⑀PSC1V3-scr-EGFP (E), syntaxin 6 (F), and a merged image (G). bition of phospholipase D by PKC⑀ (33) and PKC␦ (34), and induction of apoptosis by PKC␦ and PKC (27,35).
Several of the effects are conceivably mediated by the regulatory domain of the PKC isoform (25,27,32,33). For instance we have shown that PKC⑀-mediated neurite outgrowth is mediated by a region from the regulatory domain that encompasses the two C1 domains (7). This study provides a more detailed knowledge of particular motifs in this region that are crucial for the PKC⑀ effect. Mutagenesis analyses revealed that residues flanking the individual C1 domains are critical for the neurite-inducing effects. The mutations could be divided in two groups, those for which neurite induction could be restored by TPA and those that made the PKC⑀ construct loose its neuriteinducing effect even in the presence of TPA.
The constructs, for which the neurite-inducing effect could be restored by TPA, were those that were mutated in C1a-flanking residues. TPA treatment resulted in a plasma membrane localization of these constructs. Because wild type PKC⑀PSC1V3 induces neurites independently of TPA (7) and also spontaneously localizes to the plasma membrane, there is a clear correlation between a plasma membrane localization and neurite induction of PKC⑀ constructs. We therefore speculate that a localization to cortical areas of the cell may be a prerequisite for neurite induction by PKC⑀. This is in line with our previous results demonstrating that increasing the amount of PKC⑀ at the plasma membrane enhances its neurite-inducing effects (3,8).
It thus seems as if amino residues flanking the C1 domains may influence the localization of the domains. However, the neurite-inducing effect of some PKC⑀PSC1V3 mutants could not be restored by TPA. This was observed for constructs in which residues flanking the C1b domain had been mutated. The lack of TPA effect was not due to a resistance to TPAstimulated plasma membrane localization because the constructs readily translocated upon TPA treatment. The sequence N-terminal of the C1b domain caught our attention, and alignment demonstrated that there is a stretch of eight amino acids immediately N-terminal of the C1b domain that is conserved in all novel PKC isoforms, also in C. elegans and D. melanogaster, but it is not present in classical isoforms. We therefore propose that this previously unidentified motif is one factor that is crucial for neurite induction by PKC. We base this on the facts that it is present in all neurite-inducing C1 regions, but not in the non-neurite-inducing PKC␣, and that scrambling of the residues in the motif results in a PKC⑀ mutant without neurite-inducing capacity. The importance of the residues is further underscored by the fact that the isolated C1b domain had neurite-inducing capacity in the presence of TPA if the motif is included in the construct. Furthermore, the evolutionary conservation indicates an important biological role for the motif and that there may be a functional redundancy between the novel isoforms. The fact that the motif is conserved in novel PKC isoforms from both C. elegans and D. melanogaster indicates that the PKC effect, induction of neurites, has developed early in the evolution of multicellular organisms. The importance of the motif was also indicated by the fact that full-length PKC⑀ with the motif scrambled failed to potentiate NGF-stimulated neurite outgrowth in PC12 cells and suppressed the same effect in SH-SY5Y/TrkA cells.
The question remains as to the function of the motif. Our data clearly indicate that it does not abrogate PKC⑀ properties in general. The mutated PKC⑀ responded to receptor-stimulated phospholipase C activation and also displayed a similar Golgi localization as wild type PKC⑀. TPA treatment induces a clear plasma membrane localization of the constructs in which the motif had been mutated, but this still does not lead to neurite outgrowth. Given the fact that C1 domains also mediate interactions with other proteins (17)(18)(19)(20)(21), we hypothesize that interaction with other proteins at the plasma membrane is an important mechanism mediating the induction of neurites by the PKC⑀ C1 domains. Point mutations revealed that the hydrophobic residues Phe-237, Val-239, and Met-241 were critical. Conservation of phenylalanine or methionine on the protein surface has been suggested to imply a protein interaction site (36) supporting the hypothesis that the motif is important for protein-protein interaction.
However, the mere presence of the motif N-terminal of the C1b domain is not by itself sufficient to make a PKC isoform neurite-inducing. We made a construct encoding both C1 domains of PKC␣ in which the PKC⑀ motif had been introduced immediately N-terminal of the PKC␣ C1b domain, but this construct lacked neurite-inducing effects (data not shown). We also hypothesized that expression of a peptide, either EGFPtagged or Myc-tagged, containing the motif would break a putative protein-protein interaction and consequently block neurite outgrowth. However, this was not the case (data not shown). This may be due to aberrant folding of a short peptide, but it may also, perhaps more conceivably, be explained by a need for other components of the PSC1V3 construct in order to obtain proper binding.
In conclusion, in this study we have identified an evolutionarily conserved motif N-terminal of the C1b domain in novel PKC isoforms. The motif is necessary, but not sufficient, for neurite outgrowth by PKC⑀ and does not generally alter other properties of the enzyme.