INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The members of the protein kinase C (PKC) family are implicated in the regulation of a wide range of cellular processes. Based on structural similarities and requirement for activators, this family of serine/threonine kinases can be subgrouped into classical (, I, II, and ), novel (, , , and ), and atypical (/ and ) PKC isoforms (Nishizuka, 1992; Newton, 1995; Liu, 1996).

The outgrowth of neurites that accompanies neuronal differentiation is one cellular process that has been suggested to be regulated by PKC. Based on experiments with cell lines of various origin, both PKC (O'Driscoll et al., 1995; Corbit et al., 1999) and PKC (Hundle et al., 1995; Fagerström et al., 1996; Hundle et al., 1997; Brodie et al., 1999; Zeidman et al., 1999) have been proposed to be the PKC isoform that positively regulates neurite outgrowth. This could suggest that these isoforms have redundant functions, but in several other cell systems PKC and  have unique and sometimes opposite effects (Mischak et al., 1993; Lehel et al., 1994; Fleming et al., 1998).

We have previously demonstrated that in neuroblastoma cells, overexpression of PKC, but not PKC, II, or  leads to neurite outgrowth (Zeidman et al., 1999). The effect is mediated by the regulatory domain (RD) and independent of the catalytic activity of the kinase. We also identified a dominant negative construct that suppresses both PKC-mediated neurite induction and the outgrowth of neurites that accompanies neuronal differentiation. This suppression was observed when using two established differentiation protocols of neuroblastoma cells: treatment with retinoic acid (RA) of SK-N-BE(2) cells (Helson and Helson, 1985; Hanada et al., 1993) and with nerve growth factor (NGF) of SH-SY5Y cells stably transfected with TrkA (Lavenius et al., 1995). This provides evidence for the involvement of PKC in regulation of neurite outgrowth during differentiation of neuroblastoma cells.

The fact that increasing the levels of PKC is sufficient to induce neurites could imply that elevation of endogenous levels of PKC may be a mechanism through which neurite outgrowth is induced during neuronal differentiation. Another putative mechanism leading to PKC-mediated neurite outgrowth may be a shift toward a neurite-inducing state of PKC, which may involve either a change in localization and/or conformation of the PKC molecule. Such alterations have been shown to take place when regulators interact with the PKC molecule (reviewed in Newton, 1997). Overexpression of PKC would in this case, by increasing the total amount of molecules, lead to an increase in the absolute number of PKC molecules that spontaneously adopt the conformation and/or localization that mediates neurite outgrowth. Because PKC is the only isoform that induces neurite outgrowth in neuroblastoma cells, there are likely unique structures in PKC that would be of importance for the acquisition of a neurite-inducing state of this isoform.

Studies comparing PKC and  have shown that structures responsible for isoform-specific effects may reside both in the regulatory and the catalytic domain, depending on the effect that is elicited by PKC (Ács et al., 1997a,b; Wang et al., 1997, 1998). The C2 domain, in the RD, is crucial for the binding of PKC to its receptor for activated C-kinase (Mochly-Rosen and Gordon, 1998) and this domain has been used to specifically block the translocation and function of PKC (Johnson et al., 1996; Hundle et al., 1997). Furthermore, there is an actin-binding site between the C1 domains, unique for PKC, which is of importance for the localization of PKC and also can mediate an F-actin-induced activation of the enzyme (Prekeris et al., 1996, 1998). This is of interest because there is an increased association of PKC to the cytoskeleton during neuronal differentiation of PC12 cells (Brodie et al., 1999) and because PKC is enriched in the F-actin-rich growth cones of differentiating neuroblastoma cells (Fagerström et al., 1996). The aim of this study was to investigate whether the actin-binding site is important for PKC-mediated neurite outgrowth and to analyze whether this involves an altered conformation and/or localization of the protein during this process.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Plasmids

Plasmids containing cDNA encoding full-length PKC and PKC were generated by polymerase chain reaction (PCR) with cDNA from human placenta and SH-SY5Y cells, respectively. Other plasmids encoding full-length or RD of human PKC isoforms fused to enhanced green fluorescent protein (EGFP) cDNA have been described previously (Zeidman et al., 1999).

Plasmid encoding full-length PKC with deleted actin-binding site (ABS), i.e., amino acids (aa) 223-228, called FLABS, was generated by PCR amplifying cDNA encoding aa 1-222 and 229-737 from PKC, respectively, introducing an MluI site in the primers. The two fragments were cleaved, ligated, and subjected to a second PCR amplifying the combination of the two cDNAs (Figure 3A). A similar approach was used when constructing plasmids encoding PKC with alanine 159 changed for a glutamate when a SalI site was introduced in the primers by modification of nucleotides encoding arginines 161-163 (Figure 5A). The DNA fragments were introduced into the pEGFP-N1 vector (CLONTECH, Palo Alto, CA), thereby fusing the PKC cDNA with EGFP cDNA. Plasmid encoding the RD of PKC with deleted actin-binding site, called RDABS, was generated with PCR amplifying cDNA encoding aa 1-373 by using FLABS as template. As before, the cDNA was cloned in the pEGFP-N1 vector.

A double-stranded oligonucleotide encoding the actin-binding site from PKC fused to a linker sequence was cloned into pEGFP-C1 (CLONTECH) so that, upon expression, a fusion protein consisting of EGFP-linker-ABS, would be produced. A construct encoding the scrambled version of the actin-binding site was produced in the same way. Both constructs were also cloned into a vector containing a C-terminal myc-tag under the control of a cytomegalovirus promoter. The constructs are described in Figure 4A.

Vector encoding myristoylated PKC (myrFL) was created by insertion of a double-stranded oligonucleotide encoding the myristoylation sequence from Lyn (Resh, 1999) into the NheI and BglII sites N terminally to the start of the PKC coding sequence in the pEGFP-N1 expression vector. The construct is described in Figure 7B. To create myrFLABS, the FL cDNA was exchanged for cDNA encoding FLABS in the myrFL vector.

PCRs were performed with Pfu polymerase (Promega, Madison, WI) to minimize introduction of mutations and all PCR-generated fragments were sequenced. For all EGFP constructs expression of proteins of the anticipated size was confirmed with Western blot analysis.

Cell Culture and Transfections

Human neuroblastoma SH-SY5Y, SH-SY5Y/TrkA (Lavenius et al., 1995), and SK-N-BE(2) cells were maintained in minimal essential medium supplemented with 10% fetal bovine serum, 100 IU/ml penicillin, and 100 µg/ml streptomycin (Invitrogen, Carlsbad, CA). For transfection experiments, SH-SY5Y and SH-SY5Y/TrkA cells were trypsinized and seeded at a density of either 350,000 or 100,000 cells/35-mm cell culture dish on glass coverslips as previously described (Zeidman et al., 1999). Seeding at the lower cell density was done for differentiation of SH-SY5Y/TrkA cells by using NGF (100 ng/ml; Promega) for 4 d. SK-N-BE(2) cells were seeded on glass coverslips (300,000 cells/dish). For differentiation experiments SK-N-BE(2) cells were seeded at a density of 150,000 cells/35-mm dish and treated for 48 h with 10 µM RA (Sigma, St. Louis, MO). Ethanol (final concentration 0.25%) was added to the control to obtain the same solvent concentration. Transfections were initiated 24 h after seeding and were done with 2 µg of DNA and 4 µl of Lipofectin (Invitrogen) for SH-SY5Y cells and 4 µl of LipofectAMINE (Invitrogen) for SK-N-BE(2) cell. When indicated, 12-O-tetradecanoylphorbol-13-acetate (TPA; Sigma) was used at a concentration of 16 nM and latrunculin B (Calbiochem, San Diego, CA) at 0.6 µM.

Morphology Studies

Sixteen hours after the end of transfections, unless otherwise indicated, cells were fixed and mounted (Zeidman et al., 1999). Transfected cells were identified by the fluorescence of EGFP and examined with a fluorescence microscope. A transfected cell was considered to have long neurites if the length of the process exceeded that of two cell bodies. Two hundred transfected cells per experiment were counted.

Laser Scanning Cytometry

The amount of PKC-EGFP fusion proteins in individual cells was estimated with a laser scanning cytometer (CompuCyte) by using 488-nm excitation and a 530/30 emission filter with a 20× lens. The levels of laser strength and detection gain were set so that no pixel of PKC-EGFP-expressing cells reached maximal levels and these settings were the same for all experiments. After the scan each cell was relocated to check that only single cells were used for quantification.

Cytoskeletal Preparation

SK-N-BE(2) cells (seeded at a density of 500,000 cells/100-mm dish) were treated for 4 d with 10 µM RA (with ethanol added to the control). Crude cytoskeletal preparations were done essentially as previously described (Särndahl et al., 1993). Cells were scraped off the culture dishes in cold phosphate-buffered saline (PBS), pelleted, and lysed for 15 min on ice in a buffer containing 25 mM HEPES, pH 7.4, 2 mM MnCl₂, 4 mM iodoactetic acid, 10 µM Na₃VO₄, 1 mM EDTA, 1% Triton X-100, and Complete protease inhibitor cocktail (Roche Molecular Biochemicals, Indianapolis, IN). The lysates were centrifuged for 10 min at 500 × g to remove debris and nuclei, followed by a centrifugation for 10 min at 5000 × g. The pellet, containing the crude cytoskeleton, was washed once with the lysis buffer. The protein concentration in the supernatant was determined and used to normalize the protein amount in the cytoskeletal pellets from differently treated cells. The pellets were thereafter subjected to Western blot analysis.

Western Blot Analysis

Samples were separated with SDS-PAGE and transferred to Hybond-C extra nitrocellulose filter (Amersham Biosciences, Piscataway, NJ) as previously described (Zeidman et al., 1999). Proteins were detected with primary antibodies toward EGFP (CLONTECH), PKC (Santa Cruz Biotechnology, Santa Cruz, CA), and actin (clone C4; ICN, Costa Mesa, CA), and visualized with horseradish peroxidase-labeled secondary antibody (Amersham Biosciences) by using the SuperSignal system (Pierce Chemical, Rockford, IL) as substrate. The chemoluminescence was detected with a charge-coupled device camera (Fujifilm; Fiji Photo Film, Tokyo, Japan). Band intensities were analyzed with Lab Science software (Fujifilm).

Subcellular Fractionations

For subcellular fractionations, 3 × 10⁶ SK-N-BE(2) cells were seeded and transfected with 6.4 µg of DNA and 20 µl of LipofectAMINE 2000 (Invitrogen). Cells were washed in PBS, suspended in homogenization buffer (20 mM Tris-HCl, pH 7.5, 10 mM EGTA, 2 mM EDTA, Complete protease inhibitor cocktail), and homogenized using a Dounce homogenizer, which was followed by centrifugation for 10 min at 500 × g to remove cell debris and nuclei. Lysates were centrifuged for 1 h at 100,000 × g and the supernatant, the cytosolic fraction, was collected and the pellet was treated for 3 h with homogenization buffer containing 1% Triton X-100 followed by a 1-h centrifugation at 100,000 × g. The resulting supernatant (the Triton-soluble membrane fraction), the pellet (the Triton-insoluble cytoskeletal fraction), and the cytosolic fraction were subjected to Western blot analysis.

Immunofluorescence and Staining of F-Actin

Cells were fixed with 4% paraformaldehyde in PBS for 4 min, permeabilized, and blocked with 5% normal goat serum and 0.3% Triton X-100 in Tris-buffered saline (TBS) for 30 min. F-Actin was stained for 20 min with Alexa Fluor 546-conjugated phalloidin (Molecular Probes, Eugene, OR) diluted 1:40 in TBS. Endogenous PKC was detected with a primary polyclonal rabbit anti-PKC antibody (Santa Cruz Biotechnology) diluted 1:100 in TBS followed by the secondary antibody Alexa Fluor 488-conjugated goat anti-rabbit IgG (Molecular Probes) diluted 1:400 in TBS. Both incubations were 1 h in length. Extensive washing with TBS was done between all steps and the coverslips were mounted on object slides (Zeidman et al., 1999).

Confocal Microscopy

Cells were examined using a Bio-Rad Radiance 2000 confocal system fitted on a Nikon microscope. A 60×/numerical aperture 1.40 oil lens was used and excitation wavelengths were 488 nm (EGFP and Alexa Fluor 488) and 543 nm (Alexa Fluor 546) and the emission filters used were HQ515/30 (EGFP and Alexa Fluor 488) and 600LP (Alexa Fluor 546). Colocalization was analyzed with the LaserPix software (Bio-Rad, Hercules, CA).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Induction of Neurites by PKC

In a previous study, we demonstrated that overexpression of PKC, but not , II, or  resulted in neurite outgrowth in neuroblastoma cells. The effect was mediated by the RD and similar results were obtained by the RDs of other novel PKC isoforms (Zeidman et al., 1999). To establish whether the neurite induction is specific for full-length PKC, expression vectors encoding full-length PKC and  fused to EGFP were created. SK-N-BE(2) neuroblastoma cells were thereafter transfected with expression vectors coding for full-length or RD of all novel PKC isoforms, fused to EGFP (Figure 1A). PKC-transfected cells (38%) had long neurites, compared with 2.5-12% of cells expressing the other novel isoforms or EGFP alone, demonstrating that PKC is the only novel PKC isoform that induces neurites upon overexpression. Supporting our previous finding, neurites were induced upon overexpression of the RD from all novel PKC isoforms, except PKC, which only had a minor effect.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Induction of neurite outgrowth by overexpression of PKC. (A) SK-N-BE(2) neuroblastoma cells were transfected with expression vectors encoding full-length (FL) and RD from PKC, , , and , fused to EGFP. Empty EGFP vector () was used as control. Cells were fixed 16 h after transfection and transfected cells with neurites longer than the length of two cell bodies were counted. Data (mean ± SEM, n = 2) are presented as percentage of transfected cells with long neurites. (B) Amount of EGFP fusion proteins in individual cells was analyzed with laser scanning cytometry. For each experiment, 50-150 cells were analyzed and the average of these cells was used as the observation value from that experiment. Data (mean ± SEM) are presented as arbitrary fluorescence intensity units from two separate experiments. (C) Cells expressing full-length PKC or the PKC RD fused to EGFP with fluorescence intensity of 350,000-550,000 U were scored for the presence of neurites. Twenty cells were scored in each experiment and data are presented as percentage of cells with neurites (mean ± SEM, five separate experiments) and arbitrary fluorescence intensity units (mean ± SEM, 100 different cells). SK-N-BE(2) cells were grown with or without 10% serum (D) or in serum-free medium with or without 16 nM TPA (E) for 16 h after transfection with vectors encoding EGFP (), full-length PKC (FL), or the PKC RD (RD). Transfected cells were scored for neurites longer than two cell bodies and data, mean ± SEM, n = 3 (D) or 5 (E), are presented as percentage of transfected cells with neurites.

The fact that only full-length PKC induces neurites may be due to a higher degree of overexpression of this isoform. We therefore analyzed the concentration of the different PKC-EGFP proteins in individual cells with laser scanning cytometry (Figure 1B). The levels of PKC-EGFP were actually lower than those of full-length PKC and  EGFP fusion proteins, demonstrating that the selective neurite induction by PKC is not due to higher levels of this isoform. However, the isolated RD of PKC displayed higher expression levels than full-length PKC, which could explain why overexpression of the PKC RD leads to more cells with neurites than overexpression of the full-length protein. We therefore selected cells with similar levels of RD PKC and full-length PKC and examined whether these cells had neurites (Figure 1C). This analysis revealed that when the same amounts of EGFP fusion proteins were present in a cell, it was more probable that cells expressing the isolated RD would have neurites.

The fact that the isolated RD of PKC was more potent than the full-length protein in terms of inducing neurite outgrowth may suggest that the RD mimics a state of PKC that is favorable for neurite induction and that full-length PKC can acquire this state upon stimulation. The requirement for stimulus of the RD and the full-length PKC for efficient neurite induction was therefore investigated. SK-N-BE(2) neuroblastoma cells were transfected with expression vectors encoding RD or full-length PKC fused to EGFP and grown in the absence or presence of 10% serum or 16 nM TPA for 16 h (Figure 1, D and E). This demonstrated that the neurite-inducing effect of full-length PKC was sensitive to removal of extracellular stimulus because 37% PKC-EGFP-expressing cells grown in the presence of serum had neurites (Figure 1D), whereas the corresponding number for cells grown in the absence of serum was 29%. Potent neurite induction by full-length PKC in serum-free medium was restored by inclusion of TPA in the medium (Figure 1E). Unlike full-length PKC, neither serum nor TPA influenced the neurite-inducing capacity of PKC RD.

Increased PKC Association with Cytoskeleton during Neuronal Differentiation

The fact that overexpression of PKC leads to neurite outgrowth could imply that increased levels of PKC are required for neurite outgrowth during neuronal differentiation. SK-N-BE(2) cells were differentiated with 10 µM RA for 4 d and the content of PKC was analyzed (Figure 2A). The results demonstrate that PKC levels were unaltered during neuronal differentiation. Hence, increased amounts of PKC are not required for neurite outgrowth during neuronal differentiation, a conclusion which is further supported by the fact that PKC is not up-regulated during NGF-induced differentiation of SH-SY5Y/TrkA cells (Fagerström et al., 1996).

View larger version (27K): [in this window] [in a new window]   Figure 2.   Association of PKC with the cytoskeleton during neuronal differentiation. SK-N-BE(2) cells were induced to differentiate with 10 µM RA for 4 d. (A) Total cell lysates, normalized for total protein content, were analyzed for PKC immunoreactivity and band intensities were quantified. Data (mean ± SEM, n = 4) represent PKC levels in differentiated cells in percentage of values obtained in control cell lysates. (B) Lysed cells were divided into a Triton X-100 soluble cytosolic/membrane and a particulate cytoskeletal fraction. Cytoskeletal fractions, normalized to the Triton X-100-soluble fractions for protein content, were subjected to Western blot analysis by using anti-PKC (top) and anti-actin (bottom) antibodies. Data (mean ± SEM, n = 10) are PKC/actin intensity ratios from differentiated cells in percentage of corresponding values obtained from control cells. The increase is statistically significant (p < 0.05) using Student's t test.

It has previously been shown that there is an increased association of PKC with the cytoskeleton during NGF-induced neuronal differentiation of PC12 cells (Brodie et al., 1999) and also an enrichment of PKC in the actin-rich growth cones of differentiated neuroblastoma cells (Fagerström et al., 1996). Interaction with the cytoskeleton could be one way whereby PKC acquires a neurite-inducing capacity. We examined whether such an association takes place during RA-induced differentiation of neuroblastoma cells (Figure 2B). PKC levels in a crude cytoskeletal preparation were elevated with 75 ± 30% (n = 10) in SK-N-BE(2) cells treated with 10 µM RA for 4 d.

Actin-binding Site Is Necessary for Neurite Induction by PKC

The finding that PKC associates with the cytoskeleton during neuronal differentiation suggested that the actin-binding site, which is only found in PKC and not in other PKC isoforms, may be of importance for neurite induction. This structure is located between the C1 domains in the RD of PKC (Prekeris et al., 1996). Removal of the actin-binding site from PKC decreases its binding to F-actin in vitro but does not seem to alter other properties of the protein (Prekeris et al., 1998). To investigate the importance of this motif for the effect of PKC on neurite outgrowth, we created expression vectors encoding both EGFP-fused full-length PKC and PKC RD lacking the ABS, FLABS and RDABS (Figure 3A).

View larger version (26K): [in this window] [in a new window]   Figure 3.   Actin-binding site is of importance for neurite induction by full-length PKC. (A) PKC was modified by removing nucleotides encoding amino acids 223-228, i.e., the ABS, from wild-type PKC cDNA (wt) and replacing it with an MluI recognition sequence, coding for amino acids T and R. This mutated PKC (ABS) was also used to generate the new construct RDABS. Neurite outgrowth was examined in SK-N-BE(2) (B) and SH-SY5Y cells (C) transfected with vectors encoding wild-type FL and RD (wt) and corresponding proteins with deleted actin-binding site (ABS), all fused to EGFP. The cells were fixed and mounted 16 h after transfection and transfected cells with long neurites were counted. Data (mean ± SEM, n = 3-6) are presented as percentage transfected cells with long neurites. *, statistically significant differences with analysis of variance followed by Duncan's multiple range test.

The constructs were introduced into SK-N-BE(2) and SH-SY5Y cells and the effect on neurite outgrowth was examined (Figure 3, B and C). Neurite induction by full-length PKC was markedly reduced if the actin-binding site had been deleted. In SK-N-BE(2) cells (Figure 3B), 38% of cells overexpressing normal full-length PKC had long neurites, whereas the corresponding number for cells expressing PKC lacking the actin-binding site was 22%. In contrast, the neurite-inducing capacity of the RD was not affected by removal of the actin-binding site, because both the complete and the mutated RD caused neurite outgrowth in 50-60% of the transfected SK-N-BE(2) cells. The same effect of the removal of the actin-binding site from PKC was seen in SH-SY5Y cells (Figure 3C). The neurite induction by full-length PKC constructs were generally lower in this cell line, perhaps due to lower levels of endogenous factors that signal to PKC, and thereby render it in a neurite-inducing state.

Isolated Actin-binding Site Suppresses Neurite Outgrowth

Removal of the actin-binding site from PKC clearly reduced its ability to induce neurites. The PKC binding to and activation by F-actin have been shown to be blocked by the peptide LKKQET (Prekeris et al., 1998), which is identical to the structure that was removed in PKC without actin-binding site. We investigated whether introduction of this peptide would interfere with the PKC pathway leading to neurite outgrowth. Expression vectors encoding the isolated actin-binding site, or a scrambled actin-binding site, fused either to EGFP (EGFP-ABS and EGFP-scrambled) or to a myc-tag (myc-ABS and myc-scrambled) via a linker were created (Figure 4A).

View larger version (16K): [in this window] [in a new window]   Figure 4.   Isolated actin-binding site from PKC inhibits neurite outgrowth. (A) Expression vectors encoding the actin-binding site from PKC, or a scrambled version of this site, fused to either EGFP or a myc-tag via a linker sequence was constructed. (B) SK-N-BE(2) cells were cotransfected in a 1:7 ratio with vectors encoding EGFP, FL-EGFP, or RD-EGFP and myc-tagged ABS (ABS), scrambled ABS (Scram), or empty myc-vector (). Cells were fixed 16 h after transfection and transfected cells, identified with EGFP fluorescence, were counted and the number of cells with long neurites was determined. Data (mean ± SEM, n = 3) are presented as percentage of transfected cells with long neurites. EGFP-tagged ABS, scrambled actin-binding site (Scram), and EGFP alone (EGFP) were expressed in SK-N-BE(2) cells (C) and SH-SY5Y/TrkA cells (D). The cells were incubated in regular medium () or treated with 10 µM RA for 2 d or with 100 ng/ml NGF for 4 d (RA, NGF). The cells were fixed and the number of transfected cells bearing long neurites was assessed. Data (mean ± SEM, n = 3) are presented as percentage of transfected cells with long neurites. *, statistically significant differences with analysis of variance followed by Duncan's multiple range test compared with similar conditions by using control vector instead of ABS vector.

The expression vector coding for myc-tagged actin-binding site was cotransfected with vectors encoding FL and RD in a 7:1 ratio (Figure 4B). Because immunofluorescence staining of the myc-tag with tetramethylrhodamine B isothiocyanate-conjugated secondary antibodies was weak (our unpublished data), the EGFP fluorescence was used to identify transfected cells. When myc-ABS and myc-scrambled were expressed on their own, immunofluorescence staining by using fluorescein isothiocyanate-conjugated secondary antibody showed that the proteins are expressed in neuroblastoma cells (our unpublished data). Coexpression of the actin-binding site reduced the number of full-length PKC-expressing cells having long neurites, whereas the scrambled actin-binding site had no effect. In contrast, expression of the actin-binding site together with the PKC RD did not decrease the percentage of RD-expressing cells with long neurites, further supporting the finding presented in Figure 3, B and C, that the actin-binding site is only important for the neurite-inducing capacity of full-length PKC.

If the interaction of PKC with F-actin through the actin-binding site is part of a general mechanism involved in neurite outgrowth, it would be expected that the actin-binding site peptide would suppress outgrowth during neuronal differentiation. To explore this possibility, EGFP-tagged peptides (EGFP-ABS and EGFP-scrambled) were expressed in neuroblastoma cells that were induced to differentiate with either RA or NGF. Expression of the isolated actin-binding site led to a marked suppression of SK-N-BE(2) cells with long neurites after 48 h of RA treatment (Figure 4C). A similar effect was seen in SH-SY5Y/TrkA cells treated with NGF for 4 d (Figure 4D).

Deletion of Actin-binding Site Does not Reduce Neurite-inducing Capacity of PKC with Mutated Pseudosubstrate

The interaction with F-actin was shown to stabilize PKC in an open conformation in vitro (Prekeris et al., 1998). The decreased neurite-inducing capacity of PKC upon deletion of the actin-binding site may therefore be due to the fact that a closed conformation of PKC will be favored and the neurite-inducing domains may consequently be hidden by the catalytic domain. One way to render PKC in an open and active conformation is to mutate the alanine residue in the pseudosubstrate to a glutamate (Pears et al., 1990). To investigate whether an open conformation of PKC could compensate for the effects of the deletion of the actin-binding site, this modification of PKC and PKC with deleted actin-binding site was done (Figure 5A). When these proteins were overexpressed in SK-N-BE(2) cells, it was found that deletion of the actin-binding site of PKC with a mutated pseudosubstrate did not have an effect on neurite induction by this protein (Figure 5B). However, mutation of the pseudosubstrate did not restore the neurite-inducing capacity of PKC with deleted actin-binding site to the effect observed with wild-type PKC.

View larger version (13K): [in this window] [in a new window]   Figure 5.   Pseudosubstrate mutation does not compensate for deletion of the actin-binding site. (A) Nucleotide sequence encoding the pseudosubstrate sequence of PKC with and without actin-binding site was mutated so that alanine 159 was changed to glutamate. (B) SK-N-BE(2) cells were transfected with vectors encoding EGFP fusions of PKC, PKC E159, PKCABS, and PKCABS E159, and the number of cells with neurites longer than two cell bodies was counted. Data (mean ± SEM, n = 4) are presented as percentage of transfected cells with neurites longer than two cell bodies.

PKC Localizes Primarily to Cortical Cytoskeleton during Neurite Outgrowth

Another reported effect of deleting the actin-binding site is a decreased colocalization of PKC and F-actin in fibroblasts (Prekeris et al., 1998). The colocalization pattern of PKC and F-actin in neurites was therefore examined (Figure 6). On overexpression, PKC-EGFP was present in growth cones where a marked colocalization of PKC and cortical F-actin was seen (Figure 6, A and B). There was a similar pattern of colocalization in growth cones of SK-N-BE(2) cells, which had been induced to differentiate by RA treatment (Figure 6, C and D). Both endogenous PKC and F-actin primarily localized along the edges of the growth cone. In cells with high levels of PKC-EGFP, this fusion protein was also detected in the interior of the growth cone, perhaps due to a saturation of the binding sites in the cortical cytoskeleton. Such a distribution all over the growth cone was regularly observed for PKC without actin-binding site.

View larger version (95K): [in this window] [in a new window]   Figure 6.   Colocalization of PKC and F-actin in growth cones. SK-N-BE(2) cells were transfected with vector encoding PKC fused to EGFP (A and B) or treated with 10 µM RA for 3 d (C and D). Cells were fixed and F-actin was visualized with Alexa Fluor 546-conjugated phalloidin (B and D) and PKC was visualized using EGFP fluorescence (A) or by immunofluorescence with Alexa Fluor 488-conjugated antibodies (C).

Membrane Targeting of PKC Overcomes Requirement for Actin-binding Site

The neurite induction by PKC RD is independent of extracellular stimulus, insensitive to ablation of the actin-binding site, and is not influenced by coexpression of the actin-binding site peptide. Furthermore, the RD to a large extent localizes to the plasma membrane (Figure 7A). This, together with the fact that PKC localized to the cortical cytoskeleton in growth cones, suggested to us that targeting to the plasma membrane might overcome the requirement for the actin-binding site for optimal neurite induction by PKC.

View larger version (39K): [in this window] [in a new window]   Figure 7.   Negative effects of deletion of the actin-binding site on neurite induction are reversed by myristoylation of PKC. (A) Localization to the plasma membrane of the isolated RD from PKC fused to EGFP expressed in SK-N-BE(2) cells was shown with confocal microscopy. (B) cDNA encoding a myristoylation sequence derived from Lyn was fused to cDNA encoding FL and FLABS. In the schematic representation of the PKC-EGFP vector, the BglII site used for cloning and the PKC Kozak sequence precede the first codon, labeled 1. In the myristoylated PKC (myrFL) a PKC Kozak sequence and a sequence encoding the first 10 aa from Lyn (boxed) are inserted into the NheI and BglII sites. The original Kozak sequence and BglII site are now translated. The original starting methionine is labeled 1. (C) SK-N-BE(2) cells were transfected with expression vectors encoding FL, FLABS, and the corresponding myristoylated variants (myr), all fused to EGFP. The cells were grown for 16 h and thereafter fixed, mounted, and examined with confocal microscopy. (D) Percentage of transfected cells with long neurites was quantified in cells expressing EGFP alone, PKC RD (RD), full-length PKC (FL), myristoylated PKC (myrFL), PKC without the actin-binding site (FLABS), and myristoylated PKC without the actin-binding site (myrFLABS). After transfection, the cells were cultured for 16 h with or without 10% serum. Data (mean ± SEM, n = 5-6) are presented as percentage of transfected cells with long neurites. (E) Expression levels in single cells of the EGFP fusion proteins in D were quantified with laser scanning cytometry. Data (mean ± SEM, n = 3, 50-100 cells analyzed in each experiment) are arbitrary units of fluorescence intensity. *, statistically significant differences with analysis of variance followed by Duncan's multiple range test.

To target PKC to the plasma membrane, this isoform was tagged with a myristoylation sequence by generating expression vectors encoding myristoylated PKC (myrFL) and PKC with deleted actin-binding site (myrFLABS). These were created by addition of nucleotides encoding the myristoylation sequence from Lyn before the translation start of PKC (Figure 7B). SK-N-BE(2) cells were transfected with vectors encoding the myristoylation variants myrFL and myrFLABS along with their nonmyristoylated counterparts FL and FLABS and the subcellular localization of the proteins was analyzed with confocal microscopy (Figure 7C). The results show that myristoylation causes an increased association with the plasma membrane of both wild-type PKC and of PKC with deleted actin-binding site. The effects of these modifications on the neurite-inducing capacity of the PKC variants were thereafter examined (Figure 7D). Cells were grown in the absence or presence of serum to investigate whether the requirement of extracellular stimulation was affected by the modifications of PKC. As seen before, FL was not a potent inducer of neurite outgrowth in the absence of serum (27% cells with long neurites). Serum stimulation of FL-expressing cells enhanced this effect to around 40% cells with long neurites. The myristoylated PKC variant (myrFL), as well as the RD, was independent of serum. The percentage of transfected cells with long neurites was enhanced from 27% for nonmyristoylated to 46% for myristoylated PKC in the absence of serum.

Serum did not cause a marked potentiation of neurite outgrowth by FLABS (Figure 7D). In the absence of serum, deletion of the actin-binding site actually had no effect on the neurite-inducing capacity of PKC, which indicates that serum stimulation influences PKC through the actin-binding site. Myristoylation of PKC with deleted actin-binding site restored the neurite-inducing capacity of this protein. The effect of myrFLABS was not further enhanced by treatment with serum. The levels of neurite induction reached by myristoylation of FLABS were comparable to the effects of myrFL under serum-free conditions and of FL when cells were grown with serum. Thus, targeting of PKC to the plasma membrane overcomes the dependence on extracellular stimulus and the effect of removal of the actin-binding site.

To certify that the different effects of the various PKC constructs were not due to differences in expression levels, the amount of EGFP fluorescence in single cells was analyzed by laser scanning cytometry (Figure 7E). This demonstrated that the full-length variants were expressed at similar levels, whereas, as observed in Figure 1, the amount of the RD-EGFP protein per cell was higher. Serum did not influence the expression levels of the PKC variants.

Effects of Deletion of Actin-binding Site on Subcellular Localization of PKC

Although no apparent difference in localization between wild-type PKC and PKC without actin-binding site could be detected (Figure 7C), targeting of the latter protein to the plasma membrane overcame the attenuation of neurite-inducing capacity. To further analyze whether deletion of the actin-binding site resulted in aberrant localization of PKC, we performed a subcellular fractionation analysis comparing the distribution of full-length PKC and PKCABS to endogenous PKC (Figure 8, A and B). This demonstrated that both proteins had the same distribution pattern as endogenous PKC and there was no effect by deletion of the actin-binding site.

View larger version (80K): [in this window] [in a new window]   Figure 8.   Effects on the subcellular localization of PKC by deletion of the actin-binding site. SK-N-BE(2) cells transfected with vector encoding full-length PKC (A and C-E) and PKC without the actin-binding site (B and F-H), both fused to EGFP. Transfected cells were subjected to subcellular fractionation (A and B) and divided into a cytosolic fraction (C), a membrane fraction (M), and a cytoskeletal fraction (S), which were analyzed with Western blotting with antibodies directed toward PKC. Immunoblots from three individual experiments were quantified and data are presented as percentage of PKC in each fraction out of total PKC content. White bars represent endogenous PKC and black bars represent EGFP-tagged PKC variants. Cells expressing wild-type PKC (C-E) and PKC without actin-binding site (F-H) were fixed, and F-actin was stained with Alexa Fluor 546-conjugated phalloidin and by using confocal microscopy, a colocalization analysis was done. The images depict PKC-EGFP (C and F), F-actin (D and G), and pixels that represent colocalization of PKC and F-actin (E and H). Arrows highlight areas with cortical F-actin. The amount of phalloidin-positive pixels in the cortical cytoskeleton that was also EGFP-positive was 24% (wild-type PKC) and 14% (PKC with deleted actin-binding site) by using LaserPix software.

The same issue was also addressed by confocal microscopy and colocalization analysis (Figure 8, C-H). SK-N-BE(2) cells were transfected with vectors encoding EGFP fusion proteins of PKC with the actin-binding site intact or deleted. The actin cytoskeleton was visualized with Alexa Fluor 546-conjugated phalloidin. For wild-type PKC, a colocalization with F-actin could be detected both in the interior of the cells and, perhaps more predominantly, along the cortical cytoskeleton. For PKC with deleted actin-binding site, the colocalization was most striking in the interior of the cell. However, this PKC variant to some extent also localized to the cortical cytoskeleton, demonstrating that deletion of the actin-binding site does not abolish PKC localization to the cortical cytoskeleton. A putative role for the actin-binding site may be to strengthen the binding of PKC to this structure.

Disruption of F-Actin Results in Loss of PKC Localized at Cortical Cytoskeleton

To further explore whether the localization of PKC to cortical areas of the cell involves interaction with F-actin, SK-N-BE(2) cells were treated with latrunculin B to disrupt the microfilaments (Figure 9). In untreated cells, a substantial amount of the endogenous PKC appeared not to be present in the plasma membrane. However, a localization to the cortical cytoskeleton was also observed in several cells, invariably at cell-cell contacts, as exemplified in Figure 9, A and B. Treatment with latrunculin B (Figure 9, C and D) leads to a severe disruption of the F-actin network and upon this treatment the enrichment of PKC to cortical areas was lost. However, a few F-actin structures remained in the latrunculin B-treated cells and PKC was in several cases found to localize to these remaining structures.

View larger version (104K): [in this window] [in a new window]   Figure 9.   Disruption of F-actin leads to altered PKC localization. SK-N-BE(2) cells were treated with 0.6 µM latrunculin B for 20 min whereafter PKC was visualized with immunofluorescence (A and C) with Alexa Fluor 488-conjugated secondary antibodies, and F-actin was detected with Alexa Fluor 546-conjugated phalloidin (B and D). Cells were analyzed with a fluorescence microscope and images show control cells (A and B) and latrunculin B-treated cells (C and D). Arrows in A and B indicate cortical areas enriched in PKC. These were invariably found in cortical areas where the cell had contact with another cell.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The outgrowth of neurites is a complex process involving a number of regulatory proteins. In a recent study we found that PKC, via its RD, induces neurites in neuroblastoma cells (Zeidman et al., 1999). We also found that a dominant negative construct derived from PKC, where the second C1 domain was deleted, suppresses neurite outgrowth during neuronal differentiation, clearly indicating a crucial role for PKC in this process. This conclusion is further supported by the finding in this study that a peptide, derived from the actin-binding site of PKC, attenuates neurite outgrowth during neuronal differentiation.

The actin-binding site peptide does not suppress neurite outgrowth in general, for instance, by abolishing the interaction of several crucial proteins with F-actin, as shown by the fact that it has no effect on the induction by the isolated PKC RD. This result also demonstrates that the actin-binding site has no role for the downstream effect of PKC, which leads to neurite outgrowth. This conclusion is further supported by the findings that RDs from other novel PKC isoforms, which lack actin-binding site, and that the RD from PKC with deleted actin-binding site, efficiently induce neurites. Previously we found that a structure encompassing the C1 domains is necessary and sufficient for PKC-induced neurite outgrowth (Zeidman et al., 1999). It is therefore likely that the downstream effects of PKC are mediated by the C1 domains. These structures have in several studies been shown to exert different biological effects (Lehel et al., 1995; Pawelczyk et al., 1998; Kiss et al., 1999; Aroca et al., 2000) and to interact with other proteins (Matto-Yelin et al., 1997; Yao et al., 1997; Pawelczyk et al., 1998; Hausser et al., 1999; Johannes et al., 1999).

Instead of mediating downstream PKC effects, it is conceivable that the actin-binding site is of importance for upstream signaling to PKC in a pathway leading to neurite outgrowth. This site has been shown to be of importance both for the conformation and the subcellular localization of PKC (Prekeris et al., 1998). Mutation of the pseudosubstrate, a modification that has been demonstrated to render PKC in an open conformation and thus constitutively active (Pears et al., 1990), did not reverse the effect of the actin-binding site deletion. However, deletion of the actin-binding site did not influence the neurite-inducing capacity of PKC with mutated pseudosubstrate. This indicates that an effect on the conformation of PKC may be one mechanism through which signaling via the actin-binding site causes PKC to induce neurites.

We also found evidence supporting the importance of a proper subcellular localization of PKC during neuronal differentiation. There was an increased association of PKC with the cytoskeleton during neurite outgrowth observed both in this study using neuroblastoma cells and in PC12 cells (Brodie et al., 1999). This is in accordance with the importance of the actin-binding site that we describe herein. However, by using subcellular fractionation no differences in the proportion of PKC bound to the cytoskeleton were observed as a result of deleting the actin-binding site. The PKC actin-binding site is therefore not necessarily crucial for the interaction of PKC with F-actin in neuroblastoma cells and there are several reports demonstrating that other PKC isoforms, which lack the actin-binding site, can bind F-actin both in vivo and in vitro (Blobe et al., 1996; Nakhost et al., 1998; Slater et al., 2000). There may thus be other actin-interacting sites in the PKC molecule that could possibly be the reason why no difference in the subcellular fractionation assay was observed between wild-type PKC and PKC without actin-binding site. If the actin-binding site is not important for proper localization of PKC for neurite outgrowth, it may instead be to target PKC to specific regions of the cytoskeleton. We speculate that PKC needs to be localized to the cortical cytoskeleton and/or to the plasma membrane to exert the neurite-inducing effect. This speculation is based on the following observations: 1) PKC localized to the cortical cytoskeleton and the leading edge in the growth cone. 2) Targeting of PKC to the plasma membrane by myristoylation overcame the reduction in neurite-inducing capacity by deletion of the actin-binding site. And 3) The RD of PKC, a potent inducer of neurite outgrowth, localized to a large extent to the plasma membrane.

An enhanced concentration of PKC in the plasma membrane or the cortical cytoskeleton may thus be a trigger for neurite induction. The leading edge in the growth cone is one area where there was a prominent colocalization of PKC and F-actin. This highly dynamic area could possibly be where the C1 domains exert their effect, for instance, by inducing a conformational change of a protein already present in the leading edge or by recruiting other proteins to this site.

The dependence on extracellular stimulation for potent neurite induction by full-length PKC indicates that the effect of PKC is sensitive to signaling. Because the PKC with deleted actin-binding site was not influenced by serum, it is likely that this signaling pathway targets the actin-binding site of PKC. There are several pathways that could transduce a neuronal differentiation signal to PKC. For instance, in both differentiation protocols used in this study it is conceivable that there are increased levels of small molecules that can interact with PKC. Stimulation of growth factor receptors frequently leads to activation of phospholipase C with subsequent formation of diacylglycerol, which can induce changes in both conformation and localization of PKC. There are recent reports demonstrating that retinoids can directly bind to the PKC molecule (Hoyos et al., 2000; Radominska-Pandya et al., 2000). Alternative pathways leading to F-actin-mediated effects of PKC could include activation of the small GTPases of the Rho family that have been shown to modulate the actin cytoskeleton and to be involved in the regulation of neurite outgrowth (reviewed in Hall, 1998).

In conclusion, this study demonstrates that the PKC actin-binding site is crucial for the upstream signaling pathway acting on PKC during neurite outgrowth. We propose a model where localization to the cortical cytoskeleton and/or the plasma membrane is a necessary event for PKC to exert its neurite-inducing effect.