Voltar
Artigo Acesso aberto Revisado por pares
BIBTEX RIS

Cytosolic Targeting Domains of γ and δ Calmodulin-dependent Protein Kinase II

2001; Elsevier BV; Volume: 276; Issue: 45 Linguagem: Inglês

10.1074/jbc.m103013200

ISSN

1083-351X

Autores

Nicole Caran, Lesley D. Johnson, Kimberley J. Jenkins, Robert M. Tombes,

Tópico(s)

14-3-3 protein interactions

Resumo

Ca2+/calmodulin-dependent protein kinase II (CaMK-II) isozyme variability is the result of alternative usage of variable domain sequences. Isozyme expression is cell type-specific to transduce the appropriate Ca2+signals. We have determined the subcellular targeting domain of δE CaMK-II, an isozyme that induces neurite outgrowth, and of a structurally similar isozyme, γC CaMK-II, which does not induce neurite outgrowth. δE CaMK-II co-localizes with filamentous actin in the perinuclear region and in cellular extensions. In contrast, γC CaMK-II is uniformly cytosolic. Constitutively active δE CaMK-II induces F-actin-rich extensions, thereby supporting a functional role for its localization. C-terminal constructs, which lack central variable domain sequences, can oligomerize and localize like full-length δE and γC CaMK-II. Central variable domains themselves are monomeric and have no targeting capability. The C-terminal 95 residues of δ CaMK-II also has no targeting capability but can efficiently oligomerize. These findings define a targeting domain for γ and δ CaMK-IIs that is in between the central variable and association domains. This domain is responsible for the subcellular targeting differences between γ and δ CaMK-IIs. Ca2+/calmodulin-dependent protein kinase II (CaMK-II) isozyme variability is the result of alternative usage of variable domain sequences. Isozyme expression is cell type-specific to transduce the appropriate Ca2+signals. We have determined the subcellular targeting domain of δE CaMK-II, an isozyme that induces neurite outgrowth, and of a structurally similar isozyme, γC CaMK-II, which does not induce neurite outgrowth. δE CaMK-II co-localizes with filamentous actin in the perinuclear region and in cellular extensions. In contrast, γC CaMK-II is uniformly cytosolic. Constitutively active δE CaMK-II induces F-actin-rich extensions, thereby supporting a functional role for its localization. C-terminal constructs, which lack central variable domain sequences, can oligomerize and localize like full-length δE and γC CaMK-II. Central variable domains themselves are monomeric and have no targeting capability. The C-terminal 95 residues of δ CaMK-II also has no targeting capability but can efficiently oligomerize. These findings define a targeting domain for γ and δ CaMK-IIs that is in between the central variable and association domains. This domain is responsible for the subcellular targeting differences between γ and δ CaMK-IIs. Ca2+/CaM-dependent protein kinase type II green fluorescent protein enhanced green fluorescent protein phosphate-buffered saline Tris-buffered saline, pH 7.4, 0.05% Tween 20, 0.05% sodium azide The multifunctionality of the type II Ca2+/calmodulin-dependent protein kinase (CaMK-II)1 family is reflected in the diversity of its gene products. CaMK-II isozyme variability is the result of alternative mRNA splicing from four genes (α, β, γ, and δ) found on separate chromosomes in humans (1Tobimatsu T. Fujisawa H. J. Biol. Chem. 1989; 264: 17907-17912Abstract Full Text PDF PubMed Google Scholar, 2Nghiem P. Saati S.M. Martens C.L. Gardner P. Schulman H. J. Biol. Chem. 1993; 268: 5471-5479Abstract Full Text PDF PubMed Google Scholar, 3Edman C.F. Schulman H. Biochim. Biophys. Acta. 1994; 1221: 89-101Crossref PubMed Scopus (158) Google Scholar, 4Urquidi V. Ashcroft S.J.H. FEBS Lett. 1995; 358: 23-26Crossref PubMed Scopus (49) Google Scholar, 5Singer H.A. Benscoter H.A. Schworer C.M. J. Biol. Chem. 1997; 272: 9393-9400Abstract Full Text Full Text PDF PubMed Scopus (70) Google Scholar, 6Bayer K.-U. Löhler J. Schulman H. Harbers K. Mol. Brain Res. 1999; 70: 147-154Crossref PubMed Scopus (131) Google Scholar, 7Soderling T.R. Chang B. Brickey D. J. Biol. Chem. 2001; 276: 3719-3722Abstract Full Text Full Text PDF PubMed Scopus (226) Google Scholar). Splicing occurs primarily in the central variable domain where up to seven alternative exons can be used in different combinations to yield over two dozen unique isozymes (8Tombes R.M. Krystal G.W. Biochim. Biophys. Acta. 1997; 1355: 281-292Crossref PubMed Scopus (63) Google Scholar, 9Bayer K.U. Harbers K. Schulman H. EMBO J. 1998; 17: 5598-5605Crossref PubMed Scopus (110) Google Scholar, 10Takeuchi Y. Yamamoto H. Fukunaga K. Miyakawa T. Miyamoto E. J. Neurochem. 2000; 74: 2557-2567Crossref PubMed Scopus (49) Google Scholar). An additional alternative splice domain is found only in δ CaMK-II at the C-terminal tail (11Mayer P. Mohlig M. Idlibe D. Pfeiffer A. Basic Res. Cardiol. 1995; 90: 372-379Crossref PubMed Scopus (48) Google Scholar,12Tombes R.M. Mikkelsen R.B. Jarvis W.D. Grant S. Biochim. Biophys. Acta. 1999; 1452: 1-11Crossref PubMed Scopus (24) Google Scholar). Similar to other multifunctional protein kinase families, CaMK-II isozymes gain specificity by subcellular targeting to locations, such as the nucleus, the plasma membrane, the cytoskeleton, and specialized structures, such as post-synaptic densities or the sarcoplasmic reticulum (4Urquidi V. Ashcroft S.J.H. FEBS Lett. 1995; 358: 23-26Crossref PubMed Scopus (49) Google Scholar, 10Takeuchi Y. Yamamoto H. Fukunaga K. Miyakawa T. Miyamoto E. J. Neurochem. 2000; 74: 2557-2567Crossref PubMed Scopus (49) Google Scholar, 13Srinivasan M. Edman C.F. Schulman H. J. Cell Biol. 1994; 126: 839-852Crossref PubMed Scopus (237) Google Scholar, 14Heist E.K. Srinivasan M. Schulman H. J. Biol. Chem. 1998; 273: 19763-19771Abstract Full Text Full Text PDF PubMed Scopus (98) Google Scholar, 15Shen K. Meyer T. Science. 1999; 284: 162-166Crossref PubMed Scopus (541) Google Scholar, 16Braun A.P. Schulman H. Annu. Rev. Physiol. 1995; 57: 417-445Crossref PubMed Scopus (738) Google Scholar). CaMK-II holoenzyme is normally dodecameric, and targeting can be influenced by the heterooligomerization of CaMK-II monomers (16Braun A.P. Schulman H. Annu. Rev. Physiol. 1995; 57: 417-445Crossref PubMed Scopus (738) Google Scholar, 17Kolb S.J. Hudmon A. Ginsberg T.R. Waxham M.N. J. Biol. Chem. 1998; 273: 31555-31564Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar, 18Shen K. Meyer T. J. Neurochem. 1998; 70: 96-104Crossref PubMed Scopus (39) Google Scholar). For example, cytosolic CaMK-IIs can redirect nuclear-targeted isozymes to the cytosol by heterooligomerization (13Srinivasan M. Edman C.F. Schulman H. J. Cell Biol. 1994; 126: 839-852Crossref PubMed Scopus (237) Google Scholar,19Shen K. Teruel M.N. Subramanian K. Meyer T. Neuron. 1998; 21: 593-606Abstract Full Text Full Text PDF PubMed Scopus (286) Google Scholar). Properly targeted CaMK-IIs can influence cellular events unlike catalytically identical but mistargeted isozymes (20Ramirez M.T. Zhao X.L. Schulman H. Brown J.H. J. Biol. Chem. 1997; 272: 31203-31208Abstract Full Text Full Text PDF PubMed Scopus (175) Google Scholar, 21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). CaMK-II localization can respond to the activation state of CaMK-II (15Shen K. Meyer T. Science. 1999; 284: 162-166Crossref PubMed Scopus (541) Google Scholar, 22Koh Y.H. Popova E. Thomas U. Griffith L.C. Budnik V. Cell. 1999; 98: 353-363Abstract Full Text Full Text PDF PubMed Scopus (205) Google Scholar) or to phosphorylation by other kinases (14Heist E.K. Srinivasan M. Schulman H. J. Biol. Chem. 1998; 273: 19763-19771Abstract Full Text Full Text PDF PubMed Scopus (98) Google Scholar). Other than nuclear localization sequences, the targeting domains have not been defined for CaMK-II isozymes, particularly those encoded by the γ and δ genes. Although α and β are the predominant CaMK-II genes transcribed in the central nervous system, δ isozymes are the most common CaMK-II gene product in embryonic cells (6Bayer K.-U. Löhler J. Schulman H. Harbers K. Mol. Brain Res. 1999; 70: 147-154Crossref PubMed Scopus (131) Google Scholar, 8Tombes R.M. Krystal G.W. Biochim. Biophys. Acta. 1997; 1355: 281-292Crossref PubMed Scopus (63) Google Scholar, 10Takeuchi Y. Yamamoto H. Fukunaga K. Miyakawa T. Miyamoto E. J. Neurochem. 2000; 74: 2557-2567Crossref PubMed Scopus (49) Google Scholar, 12Tombes R.M. Mikkelsen R.B. Jarvis W.D. Grant S. Biochim. Biophys. Acta. 1999; 1452: 1-11Crossref PubMed Scopus (24) Google Scholar). Mouse embryonic cells express some βe, γC, and γBgene products but primarily express δC CaMK-II (12Tombes R.M. Mikkelsen R.B. Jarvis W.D. Grant S. Biochim. Biophys. Acta. 1999; 1452: 1-11Crossref PubMed Scopus (24) Google Scholar, 21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar,23Brocke L. Srinivasan M. Schulman H. J. Neurosci. 1995; 15: 6797-6808Crossref PubMed Google Scholar). During early mouse development, γ and δ CaMK-II gene products are the primary gene products expressed throughout the embryo including the developing nervous system (6Bayer K.-U. Löhler J. Schulman H. Harbers K. Mol. Brain Res. 1999; 70: 147-154Crossref PubMed Scopus (131) Google Scholar). δ CaMK-II isozymes have been implicated in rodent neuronal and muscle differentiation (12Tombes R.M. Mikkelsen R.B. Jarvis W.D. Grant S. Biochim. Biophys. Acta. 1999; 1452: 1-11Crossref PubMed Scopus (24) Google Scholar, 21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar,24VanBerkum M.F. Goodman C.S. Neuron. 1995; 14: 43-56Abstract Full Text PDF PubMed Scopus (62) Google Scholar, 25Donai H. Murakami T. Amano T. Sogawa Y. Yamauchi T. Mol. Brain Res. 2000; 85: 189-199Crossref PubMed Scopus (24) Google Scholar, 26Hoch B. Wobus A.M. Krause E.G. Karczewski P. J. Cell. Biochem. 2000; 79: 293-300Crossref PubMed Scopus (21) Google Scholar, 27Hoch B. Haase H. Schulze W. Hagemann D. Morano I. Krause E.G. Karczewski P. J. Cell. Biochem. 1998; 68: 259-268Crossref PubMed Scopus (47) Google Scholar). Human δE CaMK-II was originally cloned from neuroblastoma cells (8Tombes R.M. Krystal G.W. Biochim. Biophys. Acta. 1997; 1355: 281-292Crossref PubMed Scopus (63) Google Scholar). δE is also known as δ10 or δ9 depending on the presence or absence of the C-terminal tail (11Mayer P. Mohlig M. Idlibe D. Pfeiffer A. Basic Res. Cardiol. 1995; 90: 372-379Crossref PubMed Scopus (48) Google Scholar). γC CaMK-II was originally cloned from human T lymphocytes where it is particularly enriched (2Nghiem P. Saati S.M. Martens C.L. Gardner P. Schulman H. J. Biol. Chem. 1993; 268: 5471-5479Abstract Full Text PDF PubMed Google Scholar), but its function is not known. Both γC and δE CaMK-II encode 57-kDa proteins with one alternative exon in their central variable domain. Despite this structural similarity and the extranuclear location of both isozymes, only δE induces neurite outgrowth (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). Their catalytic properties are similar, suggesting that subtle targeting differences are responsible for their disparate functional roles. It is not known whether sequence differences in their one alternative exon are responsible for their targeting differences. The intent of this work has been to define the targeting domain for both γC and δE CaMK-II. Green fluorescent protein (GFP)-linked deletion constructs of γC and δE CaMK-II were prepared, and their oligomeric nature and localization were evaluated. Our findings define a targeting domain that is located in between the central variable domain and the C-terminal association (oligomerization) domain. The localization of δE CaMK-II suggests that it promotes neurite outgrowth through the stabilization of the actin cytoskeleton. NIH/3T3 mouse embryo fibroblasts were cultured on polystyrene dishes in Dulbecco's modified Eagle's medium (BioWhitaker, Walkersville, MD) with 10% fetal bovine serum (Life Technologies, Inc.), supplemented with penicillin and/or streptomycin in a 5% CO2 humidified chamber at 37 °C. The first 301 and the last 22 amino acids encoded by full-length wild type or constitutively active CaMK-II cDNAs used in this study were identical as described previously (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). Constitutively active mutants were prepared by point mutagenesis of Thr287 to Asp287, which renders CaMK-II active in the absence of Ca2+/CaM. Using polymerase chain reaction-mediated directional cloning, CaMK-II cDNAs were linked to sequences encoding enhanced green fluorescent protein (EGFP) by using the vector pEGFP-C1 (CLONTECH, Palo Alto, CA). EGFP is placed at the NH2 terminus of CaMK-II with only one additional codon (encoding a glycine residue) separating the two sequences. Primers were synthesized containing restriction enzyme sites (BspE1 at the 5′ end and BamHI at the 3′ end) to amplify the desired cDNA fragment. They enabled the in-frame introduction of CaMK-II domains on the C-terminal side of EGFP. Proper clone construction was confirmed by DNA sequencing in both directions as described previously (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar) by evaluation of enzymatic activity (for full-length CaMK-IIs) and by anti-GFP immunoblotting. Sequence analysis was performed using Gene Jockey II (Biosoft, Inc, Cambridge, United Kingdom). cDNAs encoding full-length CaMK-II isozymes were transfected into NIH/3T3 cells using LipofectAMINE PLUS (Life Technologies, Inc.) for 3 h followed by culture for at least an additional 18 h. Transfection efficiencies routinely exceeded 50%. Transfected cells were grown for 1–2 days, harvested with trypsin-EDTA, and then washed with phosphate buffered saline (PBS) containing 2.5 mm EGTA. Pellets were resuspended in 3 volumes of ice-cold homogenization buffer, which consisted of 30 mm Hepes, pH 7.4, 2.6 mm EGTA, 20 mm MgCl2, 80 mm β-glycerol phosphate, 0.1 μm okadaic acid (Life Technologies, Inc.), 0.01 mg/ml each chymostatin, leupeptin, aprotinin, pepstatin, and soybean trypsin inhibitor (Sigma). Samples were then sonicated (three 5-s bursts on ice), centrifuged at 10,000 × g for 15 min at 4 °C and either assayed immediately or frozen and stored at −80 °C. Lysates prepared by sonication solubilized over 90% of the total CaMK-II activity as measured by solution assays and immunoblots (data not shown). Cytosolic fractions were diluted to 0.1–0.2 mg/ml protein in homogenization buffer, and 10 μl was assayed for CaMK-II activity as described previously (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). Whole cell lysates were separated on 10% polyacrylamide gels using the Mini-Protean II gel electrophoresis system (Bio-Rad). Proteins were transferred to 0.45 μm of nitrocellulose sheets for 1 h at 100V and blocked with TBSTA containing 2.5% nonfat dry milk, 2.5% bovine serum albumin, and 2% normal goat serum for 1 h. The anti-GFP antibody was a mouse monoclonal IgG (CLONTECH). Primary antibodies were typically diluted to 0.5 μg/ml in 2% bovine serum albumin/TBSTA and incubated between 1 and 12 h with the nitrocellulose blot. Blots were washed three times with TBSTA and incubated for 1 h with 0.5 μg/ml alkaline phosphatase-coupled goat anti-mouse IgG (Kierkegaard Perry Labs, Gaithersburg, MD) in 2% bovine serum albumin/TBSTA. Blots were developed with 0.25 mg/ml 5-bromo-4-chloro-3-indolyl phosphate and 0.25 mg/ml nitro blue tetrazolium (Sigma) in 0.1 m Tris, 0.1 m NaCl, 5 mm MgCl2, pH 9.4. Monomeric molecular weights of bands were interpolated from a linear plot of log Mr of standards versus RF. Whole cell lysates were filtered through 0.45-μm syringe tip filters and then loaded onto a 40 × 1.0-cm Superose-12 gel filtration column (Amersham Pharmacia Biotech, Inc., Piscataway, NJ) in 50 mm Tris, pH 7.4, 150 mm NaCl, 0.1 mm dithiothreitol, 5% glycerol, and 0.001 mg/ml each chymostatin, leupeptin, aprotinin, pepstatin, and soybean trypsin inhibitor. Superose-12 was the matrix chosen, because it spanned predicted monomeric and oligomeric CaMK-II sizes. The reported exclusion limit for Superose-12 is 2 × 106 (Amersham Pharmacia Biotech). Retention times of standards, which included thyroglobulin (Mr = 669,000), β-amylase (Mr = 200,000), bovine serum albumin (Mr = 66,000), and carbonic anhydrase (Mr = 29,000), were determined from in-line absorbance at 280 nm. Sample fractions were assessed for GFP fluorescence using the Fluorstar fluorescence microtiter plate reader (B&L Systems, Maarsen, Holland) or with anti-GFP immunoblotting. Elution volumes (Ve) were plotted as a function of the void volume (V0) and the total volume (VT). V0 was determined using blue dextran, and VT was determined using glycine. Native molecular weights of samples were interpolated from a linear plot of log Mr of standards versus Kav (Ve −V0/VT −V0) as described previously (28Scopes R.K. Cantor C.R. Protein Purification: Principles and Practice. 2nd Ed. Springer-Verlag New York Inc., New York1987: 188Google Scholar). Transfected fibroblasts were grown on polystyrene dishes or glass coverslips for 1–2 days after transfection. Cells were imaged either directly on a heated stage or after fixation. Fixation was routinely performed as follows. Coverslips were first rinsed in PBS, incubated with fresh 4% formaldehyde in PBS for 15 min, permeabilized with 0.05% Nonidet P-40 (Pierce) in PBS, and then post-fixed with 4% formaldehyde in PBS for 5 min. Cells were stained with 100 nm rhodamine-phalloidin (Cytoskeleton, Inc., Denver, CO) for actin, the AA2-purified mouse monoclonal antibody for total tubulin (courtesy of Dr. Anthony Frankfurter, University of Virginia) at 1 μg/ml, or the V9 mouse monoclonal antibody (Sigma) for vimentin. For the latter two, 1 μg/ml rhodamine-coupled goat anti-mouse IgG (Kierkegaard Perry Labs) in 2% bovine serum albumin/TBSTA was used as the secondary antibody. Cells were imaged using the Olympus Fastscan 2000 12-bit digital camera mounted on an Olympus IX70 fluorescent microscope (Olympus America, Melville, NY). Images were compiled using Photoshop 5.5 (Adobe Systems, Inc, San Jose, CA). To characterize the minimal targeting domain of γC and δE CaMK-II, deletion constructs (Fig. 1, A–F) were linked at their NH2 termini to EGFP. This and other labs have shown that the catalytic properties of GFP-linked full-length constructs (A) are not affected by an NH2-terminal GFP domain (18Shen K. Meyer T. J. Neurochem. 1998; 70: 96-104Crossref PubMed Scopus (39) Google Scholar, 21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). C-terminal constructs (B) encode the variable and association domains over the last 185 (γC) or 182 (δE) residues. Variable domain constructs (C and D) begin with Ser311(γC) or Thr311 (δE) and end 108–111 (C) or 51–54 (D) residues downstream. Association domain constructs (E and F) have a normal C terminus but begin with Thr354 (γC) or Thr351 (δE) for construct E or with Ala398 for construct F as shown in Fig. 1. The sequence comprising these last 195 (γC) or 192 (δE) residues is shown with differences highlighted (Fig. 1). These 11-GFP-CaMK-II constructs were transfected into NIH/3T3 cells. After 2 days, cells were harvested and evaluated by anti-GFP immunoblots (Fig. 2). As expected, full-length γC and δE migrated at 84 kDa, which was 27 kDa larger than full-length CaMK-II alone (A). All other constructs were proportionally smaller as predicted. Lysates containing transfected full-length CaMK-II exhibited Ca2+/CaM-dependent catalytic activity, which exceeded total endogenous CaMK-II activity levels by at least 4-fold (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). Predicted and observed monomeric molecular weights (Mr), as determined by SDS-polyacrylamide gel electrophoresis, are summarized in TableI.Table IMolecular weight summary of constructsConstructResiduesPredicted WeightMonomeric WeightNative WeightRatioMrMrMrγC A1–49583,00084,000750,0009 B311–49548,00050,000650,00013 C311–42139,00042,00065,0001.5 D311–36433,00034,00045,0001.3 E354–49543,00045,000500,00011δE A1–49283,00084,000750,0009 B311–49247,00050,000600,00012 C311–41839,00042,00060,0001.4 D311–36133,00034,00045,0001.3 E351–49243,00045,000600,00013 F398–49237,00040,000450,00011Summary includes predicted monomeric size of constructs, actual monomeric mass (Mr) as determined by polyacrylamide gel electrophoresis, native mass (Mr) of the major peak as interpolated by Superose-12 gel filtration, and the ratio of native to monomeric molecular weight. Open table in a new tab Summary includes predicted monomeric size of constructs, actual monomeric mass (Mr) as determined by polyacrylamide gel electrophoresis, native mass (Mr) of the major peak as interpolated by Superose-12 gel filtration, and the ratio of native to monomeric molecular weight. CaMK-II targeting may depend upon oligomerization. Therefore, we evaluated the native molecular weights of all expressed constructs. Previous studies have indicated that the minimal oligomerization domain of α CaMK-II begins approximately at Ala384 (17Kolb S.J. Hudmon A. Ginsberg T.R. Waxham M.N. J. Biol. Chem. 1998; 273: 31555-31564Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar), which corresponds to Ala398 in δE CaMK-II. Other studies indicate that the α CaMK-II association domain comprises the last 135 amino acids but is most dependent upon the last 110 residues (18Shen K. Meyer T. J. Neurochem. 1998; 70: 96-104Crossref PubMed Scopus (39) Google Scholar). The boundaries of the association domain had not yet been determined for γ or δ CaMK-IIs, but we expected constructs A, B, E, and possibly F to oligomerize based on homology to α CaMK-II. To assess oligomerization, lysates of cells transiently transfected with γC and δE CaMK-II constructs were applied to a Superose-12 gel filtration column (see under "Experimental Procedures"). Anti-GFP immunoblots demonstrated that protein constructs remained intact through gel filtration and peaked at different elution volumes (Fig. 3,δE only). Quantitative profiles for all constructs were obtained by analyzing the GFP fluorescence of each fraction (Fig. 3, bottom panels). Elution profiles were plotted as a function of Kav, and native sizes were interpolated (Table I). For both γC and δE, constructs A, B, E, and F exhibited major peaks corresponding to oligomers of at least 9 subunits. Construct A had a substantial secondary peak corresponding to monomer and additional shoulders on both of these peaks. Constructs B, E, and F all had less relative contribution from their "monomeric" peaks and were, therefore, considered to have oligomerized more efficiently. Constructs C and D did not oligomerize whatsoever as they showed single peaks corresponding to their monomeric size. These results indicate that the C-terminal 142 residues of γ and δ CaMK-II can efficiently oligomerize, and that the C-terminal 95 amino acids of δ CaMK-II are sufficient for oligomerization. The localization patterns of full-length GFP-linked γC and δE CaMK-IIs were determined using conventional fluorescence microscopy of living cells 2 days after transfection. The populations of cells are shown at low magnification, and individual cells are shown at higher magnification using both phase-contrast and fluorescence illumination (Fig. 4). Note that these images demonstrate the high transfection efficiencies typically obtained in these experiments. Both GFP-linked CaMK-IIs demonstrated localization patterns that were similar to both the indirect immunofluorescence pattern of transfected non-GFP-linked CaMK-IIs (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar) and to endogenous CaMK-II determined by using gene-specific antibodies (12Tombes R.M. Mikkelsen R.B. Jarvis W.D. Grant S. Biochim. Biophys. Acta. 1999; 1452: 1-11Crossref PubMed Scopus (24) Google Scholar). γC CaMK-II exhibited a dispersed and uniform distribution throughout the cytoplasm. In contrast, δE CaMK-II exhibited a distinctive perinuclear and cortical cytoplasmic staining as previously reported for endogenous δ CaMK-II in rodent fibroblasts, astrocytes, and myoblasts (10Takeuchi Y. Yamamoto H. Fukunaga K. Miyakawa T. Miyamoto E. J. Neurochem. 2000; 74: 2557-2567Crossref PubMed Scopus (49) Google Scholar, 12Tombes R.M. Mikkelsen R.B. Jarvis W.D. Grant S. Biochim. Biophys. Acta. 1999; 1452: 1-11Crossref PubMed Scopus (24) Google Scholar, 27Hoch B. Haase H. Schulze W. Hagemann D. Morano I. Krause E.G. Karczewski P. J. Cell. Biochem. 1998; 68: 259-268Crossref PubMed Scopus (47) Google Scholar). C-terminal construct B of γC (Fig. 5,top row) and δE (Fig. 5, bottom row) CaMK-II localized similar to full-length construct A. γC was uniformly dispersed throughout the cytoplasm, whereas δE showed cortical and perinuclear localization. High intensity fluorescent particles were occasionally seen with some constructs (see Fig. 5, γC constructs B and E and δE constructs E and F). Constructs C and D span the entire variable domain. Construct C ends within the first third of the association domain, whereas construct D encodes a smaller segment that ends in a region that links the variable and association domains (see Fig. 1). Neither construct exhibited any targeting, as they were found throughout the cytoplasm and nucleus in an identical pattern to GFP alone (Fig. 1, GFP panel,top row). GFP is slightly more enriched in the nucleus but is not exclusively found in the nucleus. Construct E has a normal C terminus but begins at either Thr354 (γ) or Thr351 (δ) and thus lacks any variable domain sequences. Construct E is oligomeric (see Table I). Although it was lacking variable domain sequences, construct E localized much like constructs A or B, i.e. it exhibited the striking perinuclear distribution for δ and the dispersed cytoplasmic localization for γ CaMK-II. Construct F was prepared for δ CaMK-II only. This 95 amino acid domain begins at Ala399 and continues to the C terminus. This construct is synthesized at its predicted size and is efficiently oligomeric at 11 times its monomeric size (Fig. 3 and Table I). δ Construct F, however, was not targeted like full-length δE constructs A, B, or E. Like GFP alone, δ construct F was found throughout the cell in both the nucleus and the cytoplasm, but unlike GFP, it was oligomeric and often seen as small cytoplasmic fluorescent particles (Fig. 5). These findings indicate that alternative exons in the central variable domain are not necessary for cytoplasmic targeting of γ and δ CaMK-IIs. Rather, the sequence that comprises the difference between constructs E and F is necessary for targeting. Further support for this conclusion comes from our observations that GFP-linked δCCaMK-II oligomerizes and localizes in an identical perinuclear fashion to δE. The δC gene product is a δ CaMK-II isozyme, which naturally lacks alternative variable domain sequences (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). Living cells transfected with full-length δE construct A or C-terminal constructs B and E often exhibited filaments in the perinuclear region. These filaments were more easily seen when cells were permeabilized and fixed (Fig. 6). γC CaMK-II constructs did not exhibit these filaments. The δE filamentous pattern was observed even when cells were detergent washed or extracted, which supports a co-localization with the cytoskeleton and not endomembranes. Therefore, fixed cells were counterstained with antibodies against tubulin or vimentin and were counterstained with phalloidin for actin. δE CaMK-II co-localized with F-actin around the nucleus and at the cell cortex. The arrows point out some of these co-localizing CaMK-II and actin fibers (Fig. 6). Vimentin and tubulin fibers were enriched in the perinuclear region but not in the same bundles or fibers as CaMK-II and actin. CaMK-II did not co-localize with actin stress fibers, the predominant actin structure in these cells. A particularly striking-merged color image of a δE CaMK-II pattern (green) with F-actin (red) reveals co-localizing yellow filaments passing around and above the nucleus and into a short actin-rich extension (Fig. 7). Actin stress fibers can be seen as fibers that run along the base of the cell and do not co-localize with CaMK-II.Figure 7Wild-type δE CaMK-II and actin co-localization. NIH/3T3 cells transfected with full-length δE CaMK-II were fixed with formaldehyde after 2 days and counterstained with rhodamine-phalloidin. Images were combined into the color view with CaMK-II in green, actin in red, and coincident fluorescence in yellow. Scale bar corresponds to 25 μm.View Large Image Figure ViewerDownload Hi-res image Download (PPT) From this actin co-localization and our previous finding that constitutively active GFP-linked δE CaMK-II can induce neurite outgrowth (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar), we predicted that transfected constitutively active δE CaMK-II might induce actin polymerization and demonstrate a more pronounced co-localization with actin, and this is what we observed. δE CaMK-II co-localized with enhanced levels of F-actin along the entire length of the cellular extensions that it induced, particularly at the tip (Fig. 8, middle panel). Cells that were not transfected or were transfected with constitutively active γC CaMK-II exhibited normal actin patterns and showed no extensions (Fig. 8, top panel). Both of these constitutively active constructs were previously shown to exhibit high levels of autonomous (Ca2+/CaM-independent) activity (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). At a higher magnification, constitutively active δE CaMK-II could be seen at the tips of these extensions in a pattern that co-localized with bundled and cortical F-actin (Fig. 8, bottom panel). We interpret these results as indicating that cellular extensions are formed in the presence of constitutively active δE CaMK-II through either the induction or stabilization of actin fibers. Cytosolic CaMK-II activity is involved in the differentiation of adipocytes, myocytes, and pre-neuronal cells (24VanBerkum M.F. Goodman C.S. Neuron. 1995; 14: 43-56Abstract Full Text PDF PubMed Scopus (62) Google Scholar, 29Solem M. McMahon T. Messing R.O. J. Neurosci. 1995; 15: 5966-5975Crossref PubMed Google Scholar, 30Massé T. Kelly P.T. J. Neurosci. 1997; 17: 924-931Crossref PubMed Google Scholar, 31Nomura T. Kumatoriya K. Yoshimura Y. Yamauchi T. Brain Res. 1997; 766: 129-141Crossref PubMed Scopus (27) Google Scholar, 32Wu G.Y. Cline H.T. Science. 1998; 279: 222-226Crossref PubMed Scopus (325) Google Scholar, 33Goshima Y. Ohsako S. Yamauchi T. J. Neurosci. 1993; 13: 559-567Crossref PubMed Google Scholar, 34Wang H.-y. Goligorsky M.S. Malbon C.C. J. Biol. Chem. 1997; 272: 1817-1821Abstract Full Text Full Text PDF PubMed Scopus (22) Google Scholar). Of the four CaMK-II genes, δ CaMK-II gene products are the most highly expressed in these cells and have been shown to directly induce neuritogenesis (12Tombes R.M. Mikkelsen R.B. Jarvis W.D. Grant S. Biochim. Biophys. Acta. 1999; 1452: 1-11Crossref PubMed Scopus (24) Google Scholar, 21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar, 25Donai H. Murakami T. Amano T. Sogawa Y. Yamauchi T. Mol. Brain Res. 2000; 85: 189-199Crossref PubMed Scopus (24) Google Scholar). Most CaMK-II gene products are strikingly similar in structural and kinetic features, leading to the conclusion that δ CaMK-IIs, like other protein kinases, must be selectively targeted to substrates to influence events such as neurite outgrowth. In this study, we have demonstrated that δE CaMK-II, an isozyme independently discovered in heart and pre-neuronal cells (8Tombes R.M. Krystal G.W. Biochim. Biophys. Acta. 1997; 1355: 281-292Crossref PubMed Scopus (63) Google Scholar, 11Mayer P. Mohlig M. Idlibe D. Pfeiffer A. Basic Res. Cardiol. 1995; 90: 372-379Crossref PubMed Scopus (48) Google Scholar, 21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar), co-localizes with F-actin in the perinuclear region and at the cell periphery. This is the identical localization for endogenous δ CaMK-II in mouse embryonic fibroblasts (12Tombes R.M. Mikkelsen R.B. Jarvis W.D. Grant S. Biochim. Biophys. Acta. 1999; 1452: 1-11Crossref PubMed Scopus (24) Google Scholar), rat myoblasts (27Hoch B. Haase H. Schulze W. Hagemann D. Morano I. Krause E.G. Karczewski P. J. Cell. Biochem. 1998; 68: 259-268Crossref PubMed Scopus (47) Google Scholar), and rat astrocytes (10Takeuchi Y. Yamamoto H. Fukunaga K. Miyakawa T. Miyamoto E. J. Neurochem. 2000; 74: 2557-2567Crossref PubMed Scopus (49) Google Scholar). When δE CaMK-II is made constitutively active and is transfected, it induces cellular extensions (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar), localizes along and at the tips of these cellular extensions, and increases filamentous F-actin. γCCaMK-II, a catalytically and structurally similar isozyme originally found at high levels in T lymphocytes (35Nghiem P. Ollick T. Gardner P. Schulman H. Nature. 1994; 371: 347-350Crossref PubMed Scopus (100) Google Scholar), is found in the cytosol but does not induce neurite outgrowth and does not co-localize with F-actin. We have also shown that γC and δE CaMK-II targeting is not dependent on the central variable domain. Constructs C and D target no differently than GFP alone, whereas C-terminal construct E, which lacks central variable domain sequences, targets like full-length CaMK-II. This finding was somewhat surprising, because there are γ and δ CaMK-II isozymes that have central variable domain (nuclear)-targeting sequences (10Takeuchi Y. Yamamoto H. Fukunaga K. Miyakawa T. Miyamoto E. J. Neurochem. 2000; 74: 2557-2567Crossref PubMed Scopus (49) Google Scholar, 13Srinivasan M. Edman C.F. Schulman H. J. Cell Biol. 1994; 126: 839-852Crossref PubMed Scopus (237) Google Scholar, 14Heist E.K. Srinivasan M. Schulman H. J. Biol. Chem. 1998; 273: 19763-19771Abstract Full Text Full Text PDF PubMed Scopus (98) Google Scholar). Because δ targeting does not require alternative exons in the variable domain, the δE CaMK-II targeting shown here represents the "default" targeting pattern for δ gene products such as δC CaMK-II, which is the principal δ CaMK-II isozyme expressed in embryonic cells (6Bayer K.-U. Löhler J. Schulman H. Harbers K. Mol. Brain Res. 1999; 70: 147-154Crossref PubMed Scopus (131) Google Scholar, 8Tombes R.M. Krystal G.W. Biochim. Biophys. Acta. 1997; 1355: 281-292Crossref PubMed Scopus (63) Google Scholar, 10Takeuchi Y. Yamamoto H. Fukunaga K. Miyakawa T. Miyamoto E. J. Neurochem. 2000; 74: 2557-2567Crossref PubMed Scopus (49) Google Scholar, 12Tombes R.M. Mikkelsen R.B. Jarvis W.D. Grant S. Biochim. Biophys. Acta. 1999; 1452: 1-11Crossref PubMed Scopus (24) Google Scholar, 21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). This conclusion is consistent with our finding that δC and δECaMK-II show identical localization and are equally capable of inducing neurite outgrowth (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). The variable domains of both δC and α CaMK-II are structurally similar (8Tombes R.M. Krystal G.W. Biochim. Biophys. Acta. 1997; 1355: 281-292Crossref PubMed Scopus (63) Google Scholar), i.e. they lack alternative exons and are therefore the simplest products of their respective genes. Whereas δC CaMK-II has targeting sequences in the C-terminal domain, α CaMK-II can be targeted to the post-synaptic density by heterooligomerization with β CaMK-II (19Shen K. Teruel M.N. Subramanian K. Meyer T. Neuron. 1998; 21: 593-606Abstract Full Text Full Text PDF PubMed Scopus (286) Google Scholar) or to the sarcoplasmic reticulum by heterooligomerization with αKAP (9Bayer K.U. Harbers K. Schulman H. EMBO J. 1998; 17: 5598-5605Crossref PubMed Scopus (110) Google Scholar). Although we have now shown that central variable domain sequences are not necessarily required for CaMK-II targeting, it is undoubtedly clear that proper targeting is necessary for CaMK-II function. γC CaMK-II and δE CaMK-II targeting requires only the last 150 residues, which does not include the central variable domain. Our findings indicate that the first 50 residues of this C-terminal domain are necessary for targeting, and the last 95 residues are minimally necessary for oligomerization. Therefore, targeting is dependent upon oligomerization, but oligomerization alone does not result in targeting. The 50 amino acid domain corresponds to δE Thr351-Ala399 (Fig. 1), which is analogous to α Thr337-Ala384. In a model of full-length oligomeric α CaMK-II, this domain is part of a linker between the NH2-terminal peripheral catalytic domain and the C-terminal association domain, which form a central oligomeric core (36Dosemeci A. Reese T.S. Petersen J.D. Choi C. Beushausen S. Biochem. Biophys. Res. Commun. 1999; 263: 657-662Crossref PubMed Scopus (10) Google Scholar). This creates the structural potential for the interaction of this domain with other proteins. Oligomerization has also been reported as necessary for CaMK-II targeting to the NR2B subunit of the N-methyl, d-aspartate receptor, although the targeting sequences are found in the catalytic domain (37Strack S. McNeill R.B. Colbran R.J. J. Biol. Chem. 2000; 275: 23798-23806Abstract Full Text Full Text PDF PubMed Scopus (275) Google Scholar). Regardless of where the targeting domain resides, its juxtaposition as an oligomer may present a unique three-dimensional targeting site that is not present in monomeric CaMK-II. Both γC and δE CaMK-II are composed of three separate exons in their variable region (8Tombes R.M. Krystal G.W. Biochim. Biophys. Acta. 1997; 1355: 281-292Crossref PubMed Scopus (63) Google Scholar). This has been confirmed through the examination of the human δ CaMK-II gene on chromosome four (4q25) and the human γ CaMK-II gene on chromosome 10 (10q22) via sequence analysis through the National Center for Biotechnology Information. The second of these three exons (δE329–342, EPQTTVIHNPDGNK) contains some unique and repeated sequence elements that are absent from γC (21Johnson L.D. Willoughby C.A. Burke S.H. Paik D.S. Jenkins K.J. Tombes R.M. J. Neurochem. 2000; 75: 2380-2391Crossref PubMed Scopus (19) Google Scholar). This exon, however, is not responsible for δ CaMK-II targeting to the perinuclear region as described here. Although the function of this domain is not yet known in δ CaMK-II, the analogous γ domain (EPQTTVVHNATDGIK) is used in γACaMK-II to regulate the targeting of a preceding nuclear targeting domain (10Takeuchi Y. Yamamoto H. Fukunaga K. Miyakawa T. Miyamoto E. J. Neurochem. 2000; 74: 2557-2567Crossref PubMed Scopus (49) Google Scholar). However, no known δ CaMK-II expresses this domain in combination with the nuclear targeting domain (8Tombes R.M. Krystal G.W. Biochim. Biophys. Acta. 1997; 1355: 281-292Crossref PubMed Scopus (63) Google Scholar). CaMK-II has been reported to form 100-nm "clusters" in cultured hippocampal neurons (38Dosemeci A. Reese T.S. Petersen J. Tao-Cheng J.H. J. Neurosci. 2000; 20: 3076-3084Crossref PubMed Google Scholar). It is not clear what constitutes the molecular basis of these clusters, but we do not believe that they are related to the much larger fluorescent particles as described here. The particles observed here were seen only with oligomeric constructs A, B, E, and F and, therefore, were not an artifact of EGFP by itself. They were most prominent with δ construct F, which interestingly was the only oligomeric construct that lacked any subcellular targeting capability. We also observed these particles more often with γC than with δE oligomeric constructs. We suspect that when overexpressed oligomeric CaMK-IIs exceed the level of endogenous binding partners, they are more prone to cellular disposal pathways, appearing as accumulations of fluorescence. In this study, we have shown that differentially effective γ and δ CaMK-II isozymes are differentially targeted. δE CaMK-II co-localizes with F-actin in the perinuclear region and the cellular cortex in a manner that is consistent with its role in neuritogenesis. Our evidence indicates that the targeting of δE CaMK-II is dependent on sequences residing between Thr351 and Ala398. Targeting is also dependent upon oligomerization but not on the central variable domain. δ CaMK-II oligomerization requires no more than the last 95 residues. Further identification of binding targets and substrates of δ CaMK-II isozymes will help identify their precise locus of action and may account for known effects of Ca2+-dependent protein phosphorylation on neurite stabilization, outgrowth, and turning (39Lau P.-M. Zucker R.S. Bentley D. J. Cell Biol. 1999; 145: 1265-1276Crossref PubMed Scopus (79) Google Scholar, 40Gallo G. Letourneau P.C. Curr. Biol. 1999; 9: R490-R492Abstract Full Text Full Text PDF PubMed Scopus (41) Google Scholar, 41Meberg P.J. Ono S. Minamide L.S. Takahashi M. Bamburg J.R. Cell Motil. Cytoskeleton. 1998; 39: 172-190Crossref PubMed Scopus (217) Google Scholar). This work is dedicated to Dr. G. Watson James, III. We are extremely grateful to Amanda Itnyre, H. Helen Han, and Jessica Myers for technical assistance and to Drs. Helen Fillmore, Ann Kwiatkowski, Richard Moran, Donald Porter, and Shirley Taylor for technical and editorial advice.

Referência(s)