Protein Kinase Cγ Regulation of Gap Junction Activity through Caveolin-1–Containing Lipid Rafts

Abstract

purpose. To demonstrate the interactions of PKCγ with caveolin (Cav)-1 and connexin(Cx)43 in lipid rafts and its regulation of gap junctions.

methods. N/N1003A lens epithelial cells, bovine primary lens epithelial cells, and stably transfected N/N1003A lens epithelial cells were used. Coimmunoprecipitation and Western blot analysis were used to detect protein expression and their interactions. Cav-1–containing lipid rafts and redistribution of Cav-1, PKCγ, and Cx43 were analyzed by sucrose gradients and by consequent Western blot analysis. Cell surface gap junction Cx43 plaques were detected by confocal microscopy. PKCγ activity was measured with a PKC assay kit.

results. Cav-1 and -2 were found in N/N1003A and bovine primary lens epithelial cells. Cx43 was associated with Cav-1 in lipid rafts. Phorbol ester (TPA) and insulin-like growth factor (IGF)-1 recruited PKCγ into Cav-1–containing lipid rafts and stimulated the interactions of PKCγ with Cav-1 and Cx43. TPA and IGF-1 induced redistribution of Cav-1 and Cx43 from light-density fractions to higher density fractions on sucrose gradients. PKCγ redistributed with Cav-1– and Cx43-containing fractions on stimulation with TPA or IGF-1. Overexpression of PKCγ-enhanced green fluorescent protein (EGFP) increased the interaction of PKCγ-EGFP with Cav-1 and Cx43 and decreased gap junction Cx43 plaques without exogenous growth factors. Overexpression of a loss-of-function PKCγ mutant did not decrease gap junction Cx43 plaques or cause redistribution in lipid rafts, even though the PKCγ mutant still interacted with Cav-1 and Cx43.

conclusions. Activation of PKCγ by TPA or IGF-1 stimulated the interaction of PKCγ with Cav-1 and Cx43 in lipid rafts, causing Cx43, Cav-1, and PKCγ to redistribute within the lipid rafts, and this resulted in a decrease in gap junction plaques.

Gap junctions are somewhat nonspecific cell-to-cell communication channels that allow small molecules, metabolites, and ions, to pass between adjacent cells. A connexon is composed of a hexamer of connexin proteins that form a hemichannel. A gap junction channel is formed by two connexons, contributed equally by adjacent cells. There have been at least 13 genes encoding gap junction protein connexins identified in vertebrates so far. ¹ Connexins are assembled into connexons in the endoplasmic reticulum (ER) and trafficked to the plasma membrane through ER–Golgi intermediate compartments and/or trans-Golgi. At this time the gap junction proteins become detergent resistant. ^{1 2 3} Hemichannels move to the plasma membrane and dock head-to-head with other hemichannels in the adjacent cells to form whole gap junction channels. ⁴ However, clustering of gap junctions may be needed to form functional gap junction channels. ^{4 5 6} This kind of plasma membrane structure is termed a gap junction plaque, which may vary from 100 nm to several micrometers in diameter and can contain up to 10,000 connexons. Newly synthesized gap junctions always merge into the outside of existing gap junction plaques, and the old gap junctions in the central area of plaques are internalized in approximately 1 to 5 hours. ⁶ Different types of connexin channels segregate into the different plaques forming both hetero- and homoconnexons. ⁷ This dynamic process accurately regulates gap junction assembly and disassembly in living cells. 

Gap junction assembly and disassembly are regulated when the cells are exposed to environmental stimuli, such as growth factors or stress factors. For example, insulin-like growth factor (IGF)-1 causes disassembly of surface connexin (Cx) 43 plaques, ⁸ as does exposure of cells to hydrogen peroxide. ⁹ This is brought about by activation of protein kinase C (PKC) and phosphorylation of connexin proteins. ^{8 9} Activation of diverse kinases can downregulate the formation of gap junction plaques by phosphorylation on serine, threonine, and/or tyrosine at the C terminus of connexin proteins. Cx43 can be phosphorylated on serines 255, 279, 282, 325, 328, 330, 364, 368, or 372 by PKC, PKA, casein kinase 1, or mitogen-activated protein kinase (MAPK). ^{10 11 12 13 14} How and where these gap junction plaques redistribute is not known. However, a loss of membrane connexin proteins is not observed at early periods, ¹² and so internalization and degradation of connexin proteins does not appear necessary for a decrease in the number of plaques to occur. 

The plasma membrane is not uniform in structure. Instead, it is composed of microdomains referred to as lipid rafts, which are particular lipid domain structures characterized by insolubility in nonionic detergents and sodium carbonate at alkaline pH. Lipid rafts are in the well-organized plasma membrane regions that are enriched in cholesterol and sphingolipid, have a light, buoyant density and are rich in proteins that function in cellular signaling and endocytosis. ^{15 16 17} Caveolins (Cavs) are integral membrane proteins found in some types of lipid rafts and in a specific type of flask-shaped vesicle called a caveolae. ¹⁵ Cav-1 and -2 cofractionate on sucrose density gradients, which contain both membrane lipid rafts and caveolae. ¹⁸  

Two decades ago, Alcala et al. ¹⁹ showed that the lipid composition of gap junctions is enriched in cholesterol and sphingomyelin, suggesting that high rigidity of membrane regions exert significant constraints on the movement of gap junctions. Most recently, it has been determined, by using overexpression of connexins in fibroblast cells, that connexin proteins target lipid rafts and at least partially colocalize with Cavs. ²⁰ The Cx43 targeting to lipid rafts does not require the SH2/SH3 domains or PDZ domain of the C terminus, and the binding sites for Cx43 on Cav-1 include both the Cav-scaffolding domain (residues 82-101) and the C-terminal domain (residues 135-178). ²⁰  

We have reported that PKCα and -γ are predominant isoforms of PKC in lens epithelial cells. ²¹ Activation of PKCγ by lens epithelial-derived growth factor (LEDGF), phorbol ester (TPA), or IGF-1 causes phosphorylation of Cx43 at serine and a decrease in gap junction dye transfer and Cx43 cell surface plaques. ^{8 22 23} It is known that the transient redistribution of Cx43 plaques does not result in a loss from the cell surface at early time periods in casein kinase I inhibition. ¹² Rather, a redistribution may occur within lipid raft domains. In this study, we used N/N1003A lens epithelial cells and bovine primary lens epithelial cells as our experimental model systems. Using endogenous PKCγ, overexpression of PKCγ-enhanced green fluorescent protein (EGFP), and a loss-of-function mutant we determined that PKCγ can regulate the distribution of gap junctions within Cav-1–containing lipid rafts. The results demonstrate that PKCγ coimmunoprecipitated with Cav-1 and Cx43 in lipid rafts when the cells were exposed to TPA or IGF-1. However, Cx43 always cofractionated on sucrose gradients with Cav-1 in lipid rafts. TPA and IGF-1 induced a redistribution of Cav-1 and Cx43 within lipid rafts and codistribution of some endogenous PKCγ. Overexpression of PKCγ-EGFP increased the interaction between the PKCγ-GFP fusion protein and both Cav-1 and Cx43, and this caused a decrease in gap junction Cx43 plaques in the absence of exogenous growth factors. The loss-of-function mutant had no effect. Therefore, PKCγ activity is necessary for gap junction redistribution out of plaque clusters, and this may result in redistribution to lipid rafts at higher sucrose density. 

Material and Methods

Monoclonal antibodies against PKCγ, Cx43, and Cav-1, -2, and -3 were purchased from BD Biosciences-Transduction Laboratories (San Diego, CA). Monoclonal anti-EGFP was purchased from BD Biosciences-Clontech (Palo Alto, CA). Polyclonal anti-Cx43 was purchased from Chemicon (Temecula, CA). Anti-mouse or anti-rabbit IgG conjugated with horseradish peroxidase (HRP) was purchased from Promega (Madison, WI). Dulbecco’s modified Eagle’s medium (DMEM; low glucose), trypsin-EDTA, and transfection reagent (Lipofectamine) were from Invitrogen Life Technologies (Carlsbad, CA). Protein agarose beads (Protein A/G Plus) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). G418 and PKC assay kits were from Calbiochem EMD Biosciences, Inc. (San Diego, CA). All protease inhibitors and other chemicals were from Sigma-Aldrich (St. Louis, MO). A site-directed mutagenesis kit (Quikchange) was from Stratagene Corp. (La Jolla, CA). Cell culture dishes (Delta T) were purchased from Bioptechs Inc. (Butler, PA). Alexa Fluor 568 and antifade medium (SlowFade) were purchased from Molecular Probes (Eugene, OR). 

Cell Culture

N/N1003A rabbit lens epithelial cells (a gift from John Reddan, Oakland University, Rochester, MI) were cultured in 75-cm² flasks in DMEM (low glucose) supplemented with 10% fetal bovine serum and 50 μg/mL gentamicin, 0.05 U/mL penicillin, and 50 μg/mL streptomycin (pH 7.4) at 37°C in an atmosphere of 90% air and 10% CO₂. When they reached 95% confluence, the cells were used for experiments after 2 hours of serum starvation. 

Bovine lenses were isolated from fresh eyeballs and were incubated in trypsin-EDTA for 30 minutes at 37°C to release epithelial cells. ²¹ The isolated bovine primary lens epithelial cells were collected and cultured in the same medium under the same conditions as N/N1003A cells. The cells were used at passage 4 when they reached 95% confluence. 

Immunoprecipitation and Western Blot Analysis

Confluent cells were collected and lysed on ice with 0.5 mL of lysis buffer followed by homogenization and sonication. The cell lysis buffer contained 20 mM Tris-HCl (pH 7.5), 0.5 mM EDTA, 0.5 mM EGTA, 0.5% Triton X-100, 25 μg/mL aprotinin, and 25 μg/mL leupeptin. After centrifugation at 12,000g (13,000 rpm; μSpeedFuge SFR13K; Savant, Farmingdale, NY) for 20 minutes, the supernatants were collected and used as whole-cell extracts. Two micrograms per milliliter immunoprecipitation antibodies were incubated with precleared whole-cell extracts at 4°C for 4 hours with constant rotation. After this, 20 μL of the protein agarose beads were added to the mixture and further incubated for 2 hours. The beads were collected by centrifugation and washed with phosphate-buffered saline (PBS) four times, and the proteins were extracted with 20 μL of 2× sample loading buffer (containing 50 mM Tris-HCl [pH 6.8], 100 mM dithiothreitol [DTT], 2% sodium dodecyl sulfate (SDS), 10% glycerol, and 0.1% bromophenol blue) and boiled for 3 minutes. Western blot analysis were used to visualize immunoreactive bands, as described previously. ^{8 23}  

Preparation of Detergent-Resistant Membrane Fractions

Confluent cells were washed twice with PBS, homogenized with Tris-MgCl₂ buffer containing 50 mM Tris-HCl [pH 7.5], 20 mM MgCl₂, 25 μg/mL aprotinin, 25 μg/mL leupeptin, and 1 mM phenylmethylsulfonyl fluoride (PMSF). The membrane fractions were separated from soluble proteins by ultracentrifugation at 100,000g for 1 hour at 4°C, and they were consequently resuspended on ice for 15 minutes in cell lysis buffer containing 1% (not 0.5%) Triton X-100 as indicated before for sonication. To get rid of cell debris, the suspensions were subjected to centrifugation at 12,000g for 20 minutes. Furthermore, the supernatants were subjected to ultracentrifugation at 100,000g for 1 hour at 4°C to separate detergent-soluble and detergent-resistant membrane fractions. Equal amounts of proteins (20 μg/well) from detergent-resistant membrane fractions were separated by SDS-PAGE and were subjected to immunoblot analysis, as described earlier. 

Sucrose Gradient Centrifugation and Isolation of Lipid Rafts

Lipid rafts and caveolae-enriched membrane fractions were prepared as described by Schubert et al. ²⁰ with some modifications. Briefly, cells from three 75-cm² flasks were collected with cell lysis buffer containing 1% Triton X-100, and incubated on ice for 15 minutes. After this, the cells were sonicated three times, 10 seconds each. Whole-cell lysates were mixed with equal volume of 80% sucrose in Mes-NaCl buffer containing 25 mM Mes (pH 6.5), and 150 mM NaCl without Triton X-100, loaded at the bottoms of 12-mL ultracentrifuge tubes, and then overlaid with 8 mL of a 5% to 35% continuous sucrose gradient in Mes-NaCl buffer. The samples were ultracentrifuged at 245,000g for 22 hours at 4°C with a swinging bucket rotor (SW41 Ti; Beckman Instruments, Fullerton, CA). Twelve 1-mL fractions were collected from the top of each gradient. Protein samples were precipitated with 10% trichloroacetic acid (TCA), incubated on ice for more than 30 minutes, pelleted by centrifugation at 20,000g for 15 minutes at 4°C. The pellets were washed three times with 100 mM Tris-HCl (pH 7.0)/acetone (1:5), solubilized by boiling the samples in 2× sample loading buffer for 5 minutes, and then subjected to immunoblot analysis, as described earlier. 

To investigate the interaction of PKCγ with Cav-1 and Cx43 in lipid rafts, the sucrose gradient fractions containing these three proteins were collected and combined ²⁴ and further solubilized by sonication with addition of 0.1% SDS. The mixtures were subjected to coimmunoprecipitation with Cav-1 antibody, as described earlier. 

Site-Directed Mutagenesis and Cloning

The preparation of construct with double mutation at autophosphorylation sites of C terminus of PKCγ-EGFP was performed with the site-directed mutagenesis kit (Quikchange; Stratagene), according to the protocol provided. The mutated bases were underlined, as shown in the oligonucleotide sequences: for T655A, 5′-G GCA GCG CCA GCA CTG GCC CCG CCA GAC CGC TTG GT-3′ (15 cycles, 95°C, 30 seconds; 50°C, 30 seconds; 65°C, 14 minutes), and for T674A, 5′-GCT GAT TTC CAG GGC TTT GCT TAT GTG AAC CCG GAC TTC-3′ (15 cycles, 95°C, 30 seconds; 50°C, 30 seconds; 65°C, 14 minutes). For introducing the double mutation T655/674A (Dm) in the C terminus, mutant T655A plasmid DNA was used as a template for synthesis of the T674A mutation. Point mutations were confirmed by sequencing. 

Stable Transfection of PKCγ-EGFP and Its Double Mutation in N/N1003A Cells

Transient transfection into N/N1003A cells was performed by transfection reagent according to manufacturer’s protocol (Lipofectamine; Invitrogen). Clones were selected with 500 μg/mL G418 for 6 weeks and the EGFP fluorescence was checked under a fluorescence microscope. 

Translocation of PKCγ-EGFP and Its Double Mutation in Living Cells

The stably transfected cells were seeded in culture dishes (Delta T; Bioptechs Inc.) for overnight culture. The cells were cultured in serum-free medium for 2 hours before experiments. EGFP fluorescence for PKCγ-EGFP and a mutant PKCγ were measured by laser scanning confocal microscope (Carl Zeiss Meditec, Oberkochen, Germany). 

PKCγ Activity Assay for PKCγ-EGFP and Its Double Mutation

The stably transfected cells were scraped into 0.5 mL of whole-cell lysis buffer, and incubated on ice for 15 minutes. The cells were sonicated and the supernatants were collected by centrifugation at 20,000g for 15 minutes at 4°C. Equal amounts of proteins were incubated with anti-EGFP monoclonal antibodies or PKCγ antibodies for 4 hours with constant rotation at 4°C. After this, 20 μL of protein-agarose beads were added to the mixture and incubated for another 2 hours. The beads were collected by centrifugation and washed four times with cold PBS. The coimmunoprecipitation complexes were used for detection of PKC activity according to the protocol provided in the PKC assay kit (Calbiochem). PKC specific activity is expressed as picomoles of phosphate incorporated into substrate per minute per microgram protein. 

Measurement of Cell Surface Gap Junction Cx43 Plaques

The experiments were performed as described previously with some modifications. ^{8 23} Briefly, the transfected cells were fixed with 2.5% paraformaldehyde for 5 minutes and labeled with anti-Cx 43 for 2 hours at room temperature. The fixed cells were then washed three times with blocking buffer and incubated with the secondary antisera, which is attached to a fluorochrome and has specific excitation and emission wavelength. Alexa Fluor 568 (Molecular Probes), a goat anti-rabbit antibody, has excitation and emission wavelengths of 578 and 603 nm, respectively, and emits red. The cells were then washed and mounted on slides with antifade medium (SlowFade; Molecular Probes). Slides were examined by laser scanning confocal microscope (Carl Zeiss Meditec). The cell surface plaques larger than 1 μm in diameter from single transfected cells in each set (30 cells per set from triplicate experiments) were counted. Values of P < 0.05 were considered to be statistically significant. 

Statistical Analysis

All analyses represent triplicate experiments. Thirty transfected cells per set from triplicate experiments were used for PKCγ translocation and gap junction assays. Statistics were analyzed with Student’s t-test. The level of significance was considered at P < 0.05. All data are the mean ± SEM. 

Results

Enhancement of the Interactions of PKCγ with Cav-1 and Cx43 by TPA and IGF-1

Conventional PKC isoforms, such as PKCγ, are cytoplasmic when inactive. When cells are exposed to environmental stimuli, such as calcium, diacylglycerol (DAG), phorbol ester (TPA), or IGF-1, PKC γ can translocate from the cytosol to the membrane and interact with and phosphorylate connexin 43 (Cx43). ^{8 23 25} To determine whether Cavs were involved in Cx43/PKCγ interactions, these proteins were identified in Western blot and coimmunoprecipitation experiments (Fig. 1) . Cav-1 and -2, but not -3 (not shown), were present in N/N1003A cells and bovine primary lens epithelial cells (Fig. 1A) . To optimize coimmunoprecipitation methods, N/N cells from a 95% confluent flask were extracted with 500-μL cell lysis buffer, and the whole-cell extracts were separated by centrifugation at 2000g for 20 minutes (Fig. 1B , lane 1, pellets). The supernatants were further centrifuged at 12,000g for 20 minutes (Fig. 1B , lane 2, pellets), and the consequent supernatants were finally centrifuged at 100,000g for 1 hour at 4°C (Fig. 1B ; lane 3, pellets; lane 4, supernatants). All the pellet samples were dissolved in 500 μL protein loading buffer. Twenty microliters of protein samples were resolved by SDS-PAGE and were immunoblotted with the desired antibodies shown in Figure 1B . The pellets collected at 2,000g (lane 1) or 12,000g (lane 2) contained low levels of Cx43, Cav-1, and PKCγ. PKCγ was mostly in the supernatants (lane 4) and pellets (lane 3) after 100,000g centrifugation. Cx43 and Cav-1, however, were abundant in both supernatants and pellets centrifuged at 100,000g (lanes 3, 4). Thus, centrifugation at 12,000g did not result in a loss of Cx43, Cav-1, or PKCγ. 

To examine the interactions of Cav-1 with Cx43 and with PKCγ, N/N1003A cells and bovine lens primary epithelial cells were treated with TPA (300 nM, 30 minutes) or IGF-1 (25 ng/mL, 30 minutes), whole-cell lysates (precleared by 12,000g centrifugation) were immunoprecipitated with anti-Cav-1 or anti-Cx43, and Western blot analysis was performed to detect the interacting proteins. As we have reported, both TPA and IGF enhance the interactions between PKCγ and Cx43, ^{8 23} and the current data also indicate that the interaction of PKCγ with Cav-1 was stimulated or enhanced by treatments with TPA or IGF-1 (Figs. 1C 1D) . Of note, Cav-1 always coseparated with Cx43, with or without treatments (Fig. 1C 1D) . The results indicate that Cav-1 may colocalize with Cx43 endogenously, perhaps in a similar lipid domain. PKCγ can interact with these proteins after stimulation and translocation to membranes. 

Localization of Cx43, Cav-1, and PKCγ in Lipid Rafts

Cav-1 was localized in detergent-resistant membrane fractions based on Triton X-100 solubility. To determine whether Cx43 and Cav-1 cofractionate, N/N1003A cells were treated with 300 nM TPA or 25 ng/mL IGF-1 for 30 minutes, and cell fractions were separated by centrifugation and subjected to Western blot analysis (Fig. 2A) . The data show that Cav-1 and Cx43 colocalized in Triton-resistant membrane fractions. TPA or IGF-1 treatments (30 minutes) did not induce a change in localization of either Cav-1 or Cx 43 out of detergent-resistant membrane fractions. 

In confluent cells, Cx43 moves as a hexamer connexon to cell surfaces to form gap junctions that connect with the adjacent cells to form functional gap junction cell-to-cell channels that can be viewed as large clusters called plaques. Therefore, we further determined whether Cx43 localized in Cav-1–containing membrane lipid rafts. The whole-cell lysates were fractionated by 5% to 35% sucrose gradient centrifugation, as described in the Methods section and shown in Figure 3 . Lipid raft–rich fractions were obtained through the combination of fractions 3 to 7 or 3 to 5. ²⁴ Coimmunoprecipitation data are shown in Figure 2B . Cx43 was associated with Cav-1 in lipid rafts in the absence or presence of IGF-1 or TPA stimuli. PKCγ was translocated to lipid rafts and interacted with Cx43 and Cav-1 in the presence of stimuli. The results demonstrate that both Cav-1 and Cx43 colocalize in lipid rafts before and after stimulation and that the colocalization of PKCγ with both of these proteins is stimulated by either TPA or IGF-1. No difference in interaction was observed in the fractions between 3 and 7 (Fig. 2B , upper) or 3 and 5 (Fig. 2B , lower). The results of coimmunoprecipitation experiments (Figs. 1 2) suggest that these proteins may directly interact with each other and are not just cosedimented on sucrose gradients. 

Movement of Cx43, PKCγ, and Cav-1 into Higher Sucrose Density Lipid Rafts by TPA or IGF-1

TPA and IGF-1 treatments activate PKCγ and decrease cell surface Cx43 plaques and gap junction activity, as has been reported. ^{8 22 23} However, reports suggest that, at early periods after treatment most of the disassembly of plaques is not due to either degradation or internalization of Cx43 protein. ¹² Because Cx43 has been found to associate with lipid rafts in an overexpression system, ²⁰ we wanted to determine whether movement in lipid rafts occurs as endogenous Cx43 gap junction plaques decrease. The whole-cell lysates from N/N003A cells or bovine lens primary epithelial cells were subjected to 5% to 35% continuous sucrose gradient centrifugation, and were separated into 12 1-mL fractions. Localization of a particular protein was investigated by Western blot as shown in Figure 3 . Cav-1 was largely enriched in light sucrose density fraction 3 and to a lesser extent in 4 in untreated cells. Cx 43 was found in fraction 3, whereas PKCγ was largely found in the higher sucrose density fractions 6, 7, and 8. The results indicate that Cx43 colocalized in lipid rafts with Cav-1, whereas PKCγ was not localized in lipid raft microdomains from unstimulated control cells (Fig. 3A , Cont). The activation of PKCγ by TPA or IGF-1 promoted PKCγ redistribution into lipid rafts of slightly higher density containing both Cx43 and Cav-1 (Fig. 3A , fractions 4, 5). Cav-1, Cx43, and PKCγ were thus redistributed into slightly higher density fractions (Fig. 3A) after TPA or IGF-1 stimulation. However, PKCγ was also distributed throughout the fractions. This was confirmed by results of translocation studies that suggest that only a fraction of total cell PKCγ translocates to membranes. These studies, along with the results of coimmunoprecipitation (Fig. 2) , suggest that PKCγ associates with Cav-1 and Cx43 after stimulation and that Cx43 always associates with Cav-1, but that both of them are colocalized into higher density lipid rafts after stimulation, and that PKCγ moves to these same higher density Cav-1–containing fractions. 

Similar analysis using the whole-cell extracts from bovine primary lens epithelial cells showed that initially Cav-1 and Cx43 were in fractions 3, 4, and 5. TPA and IGF-1 stimulated their redistribution to higher density fractions, similar to the profiles of N/N1003A cells (Fig. 3A) . Primary lens epithelial cells have high levels of endogenous PKCγ activity, ²¹ and PKCγ is translocated to some degree, even in unstimulated cells (Fig. 3B) . However, when the cells were exposed to TPA, there was an increase in PKCγ in fractions 4 to 6, indicating a possible movement of PKCγ into lipid rafts containing Cx43 and Cav-1. 

Localization of Overexpressed PKCγ-EGFP Fusion Proteins

To determine further whether PKCγ activity is essential for movement of Cav-1 and Cx43 and for disassembly of gap junction plaques, we fused PKCγ or its double mutant T655/674A to EGFP at the C terminus ²⁶ and established stable cell lines expressing these fusion proteins. Stability of fusion protein expression was confirmed by checking EGFP fluorescence and by Western blot (Fig. 4A) . The confluent cells were harvested and extracted with Tris-MgCl₂ buffer without Triton X-100. The supernatants and membrane fractions were separated by ultracentrifugation at 35,000 rpm (100,000g) for 1 hour at 4°C. Ten-microgram protein samples were subjected to SDS-PAGE and immunoblot analysis. PKCγ-EGFP fusion proteins were overexpressed in N/N cells and localized to both cytosolic and membrane fractions, even if not stimulated (Fig. 4A) . Overexpression of PKCγ may override the ability of other cytosolic proteins to maintain PKCγ in the cytosolic fractions. Endogenous PKCγ was located in cytosolic fractions and had to be stimulated to translocate (Fig. 4A) . This suggests that some of the overexpressed PKCγ may associate with membrane fractions without growth factor addition. 

We next examined the effect of TPA and IGF-1 on fusion protein translocation, using confocal imaging of EGFP to follow the movements of PKCγ-EGFP and its mutant in living cells. As in the previous experiments, the cells were seeded in culture dishes (Delta T; Bioptechs Inc.) and treated with either 300 nM TPA or 25 ng/mL IGF-1, and the images were collected at 30 minutes. In total, 30 live cells were imaged, and sample images were shown in Figure 4B . EGFP protein was located throughout the cells, and was not sensitive to TPA or IGF-1 stimulation. In contrast, much of the PKCγ-EGFP fusion protein was mainly translocated from cytosol to plasma membrane when cells were treated with TPA or IGF-1 in agreement with the results of endogenous PKCγ translocation. The PKCγ double mutant T655/674A (PKCγDm-EGFP) has no PKC activity (Fig. 4C) , but still has an intact C1 domain. Translocation still occurred, to some extent, for PKCγDm-EGFP fusion proteins in response to TPA or IGF-1. Therefore, the double mutant, although not enzymatically active, still translocated to membranes. However, it was not enzymatically activated by TPA or IGF-1 (Fig. 4C) and had no PKC activity. These results are consistent with the observation on PKCβ translocation. ²⁷  

Effects of Stable Overexpression of PKCγ-EGFP and Loss-of-Function PKCγDm-EGFP on Cell Surface Connexin 43 Gap Junction Plaques

We have reported that overexpression of PKCγ, but not PKCα, causes gap junction disassembly in N/N cells. ²³ In this report we show that overexpression of nonactive PKCγ is not sufficient to cause a decrease in Cx43 gap junction plaques. Cx43 in transfected and nontransfected cells was labeled with Cx43 antisera. Representative Cx43 plaques are shown in Figure 5A in conjunction with the quantitative data from triplicate experiments shown in Figure 5B . Because larger plaques may be more functional, plaques larger than 1 μm were counted. Overexpression of EGFP protein did not affect the cell surface plaques, and no abnormal phenotype was found in cells with overexpression. Overexpression of the PKCγ-EGFP fusion protein decreased gap junction plaques significantly. However, the loss-of-function PKCγ mutant (PKCγDm-EGFP) did not affect cell surface gap junction plaques. The cells overexpressing the PKCγ double mutant increased in number, as did the nontransfected cells. However, overexpression of functional PKCγ slowed the cell growth rate by approximately half, but cell morphology appeared the same (data not shown). 

Coimmunoprecipitation of PKCγ-EGFP and PKCγDm-EGFP with Cav-1 and Cx 43

PKCγ has two cysteine-rich domains, C1A and C1B, both of which bind DAG, but the second C1B domain can also interact with proteins. ²⁸ Thus, protein interaction would occur in the absence of PKCγ enzyme activity in mutants with an intact C1 domain. To determine whether PKCγ enzyme activity is necessary for association with Cav-1 and Cx43, PKCγ-EGFP or PKCγDm-EGFP stably transfected cells were used. The whole-cell lysates were immunoprecipitated with monoclonal anti-EGFP (Fig. 6A) . In normal growing cells, endogenous PKCγ did not immunoprecipitate with either Cx43 or Cav-1 (Figs. 1C 1D) . For a positive control, the whole-cell lysates from nontransfected N/N1003A cells were incubated with anti-PKCγ and were subjected to coimmunoprecipitation and Western blot (lane 2) along with the samples from PKCγ-EGFP (lane 3) and PKCγDm-EGFP (lane 4) stably transfected cells. The films were overexposed, and Cx43 and Cav-1 bands were observed (lane 2), even though they were very faint compared with lanes 3 and 4. This indicates that only low levels of PKCγ immunoprecipitated with Cx43 or Cav-1 in untreated conditions (i.e., without TPA or IGF-1). The results in Figure 6 demonstrate that both PKCγ-EGFP and PKCγDm-EGFP fusion proteins interacted strongly with Cav-1 or Cx43 in the absence of stimuli (lanes 3 and 4), and these associations were stronger than that of endogenous unstimulated PKCγ (lane 2; greater than 10 times; also compare with Figs. 1C 1D ) based on quantitative analysis shown in Figure 6B . As a negative control, lane 1 revealed that EGFP did not interact with Cx43 or Cav-1. Overall, overexpression of wild-type PKCγ or its double mutant both located partially in membranes (Fig. 4A) , and they were associated with Cav-1 and Cx43, even though the mutant was without enzymatic activity. 

Association of PKCγ-EGFP and PKCγDm-EGFP with Cav-1 and Cx 43 in Lipid Rafts and Redistribution to Higher Density Fractions

Overexpressed PKCγ and its mutant were both associated with membrane fractions (Fig. 4A) and interacted with Cav-1 and Cx43 from unstimulated cells (Fig. 6) , but the loss-of-function PKCγ mutant did not decrease Cx43 plaques (Fig. 5) . To determine whether PKCγ enzyme activity is necessary for redistribution in lipid rafts, sucrose gradients were used to detect changes in lipid raft migration patterns in transfected cells (Fig. 7) . Most of the Cx43 and Cav-1 were in fractions 3 and 4 in control cells (transfected EGFP only) similar to results shown in Figures 3A and 3B . Overexpression of PKCγ-EGFP caused a shift of Cx43 and Cav-1 out of fraction 3 into higher density fractions. This is similar to the results using TPA or IGF-1 on nontransfected cells to stimulate endogenous PKCγ (Fig. 3) . Therefore, overexpression of active PKCγ caused a shift of Cx43 and Cav-1 to higher density fractions. Overexpression of loss-of-function PKCγ did not change the detergent-resistant membrane patterns of Cx43 and Cav-1 when compared with transfected EGFP cells (control). 

Endogenous PKCγ and exogenous fusion proteins of PKCγ-EGFP or PKCγDm-EGFP were also visualized in Fig. 7 . Endogenous PKCγ was localized in fractions throughout and especially in high-density fractions in EGFP cells (control). Both overexpressed and mutant PKCγ was mainly located in lower density fractions 4 to 8, even though PKCγDm-EGFP had no enzyme activity, indicating that exogenous PKCγ fusion proteins associated with Cav-1 and Cx43 in lipid rafts. However, the mutant PKCγ did not cause the shift in migration of Cx43 and Cav-1 into higher density fractions. The mutant PKCγ also localized to slightly lower density fraction 3 as did Cx43 and Cav-1 from unstimulated cells. These results indicate that active PKCγ is required for both loss of Cx43 plaques and for redistribution in lipid rafts. 

Discussion

The present findings indicated Cx43 to be colocalized with Cav-1 in lipid rafts in stimulated and unstimulated cells. PKCγ was activated and translocated to Cav-containing lipid rafts by TPA and IGF-1. Phosphorylation of Cx43 by PKCγ may have contributed to the segregation of gap junctions from large plaques into single or oligomeric connexons, and this could make them redistribute within lipid rafts. Moreover, overexpression of functional PKCγ mimics this effect on decreases of gap junction plaques similar to that observed with endogenous PKCγ activated by TPA or IGF-1. PKCγ activity is required for gap junction redistribution in Cav-containing lipid rafts, and this could be cause a reduction in gap junction plaques. 

Like Cav-1, Cx43 was found to be Triton-insoluble in lipid rafts in N/N1003A and bovine primary lens epithelial cells, and both proteins were not significantly changed in detergent-resistant membrane localization in the presence or absence of TPA or IGF-1. When Cav-1 is synthesized, it is oligomerized in the ER and trafficked to plasma membrane in a Triton-insoluble form. ^{15 16} The results presented herein are mainly consistent with those in previous reports showing that at least mature Cx43 is Triton-insoluble when it reaches the membrane. ^{1 2} Cx43 is initially trafficked to plasma membranes as a connexon oligomer, and then this complex moves into gap junction plaques. 

The interaction of Cav-1 with Cx43 has been documented recently. Physical interaction studies show that both the scaffolding domain (amino acids [aa] 82-101) and C-terminal domain (aa 13-178) of Cav-1 recognized Cx43. ²⁰ The consensus sequence for binding to a Cav-1 scaffolding domain can be ΦXΦXXXXΦ, where Φ is an aromatic amino acid (Phe, Tyr, or Trp). ¹⁵ Cx43 has a perfect consensus sequence in residues 25-32 of the first transmembrane sequence. This may suggest that Cav-1 binding could occur during Cx43 gap junction assembly. ²⁰ Our results based on sucrose gradient velocity assays indicate that endogenous Cx43, like Cav-1, localizes in cholesterol-sphingolipid raft domains. Lipid composition analysis reported by separate groups showed that gap junctions are always coupled with highly rigid regions rich in cholesterol and sphingolipids. ¹⁹ Although the C-terminal domain of Cx43 is not necessary for targeting of Cx43 to lipid rafts, ²⁰ gap junction assembly and disassembly are regulated through phosphorylation status at the C terminus of Cx43. ^{12 29 30} Direct interaction and association of Cav-1 with Cx 43 before and after disassembly of gap junction plaques indicates that Cav-1 may always be associated with Cx43 in lipid rafts during this process, because both proteins shift together within the membrane fractions. 

The interaction of PKCγ with Cav-1 had not been demonstrated until now. Previous reports demonstrated that PKCγ can interact with and phosphorylate Cx43, and therefore decrease gap junction activities. ^{8 23 25} Our present study indicates that both PKCγ and the loss-of-function mutant PKCγ (T655/674A) interact with Cav-1 and Cx43 (Figs. 2 6) . PKCγ may interact with Cav-1 through two possible consensus sequences (ΦXΦXXXXΦ) at the carboxyl-terminal domain, the first one is in residues 539-546 (wsfgvlly), and the second in residues 673-680 (ftyvnpdf). Mutation of the second site (T674A) showed no effect on coimmunoprecipitation assay, suggesting that PKCγ activity is not necessary for the interaction between PKCγ and Cav-1. However, this particular mutation would not have altered the Cav-1 consensus binding motif of this region. At this time, the mechanism by which Cav-1–Cx43 redistribution in lipid rafts or the disassembly of gap junction plaques occurs after PKCγ phosphorylation is not known. However, our results strongly suggest that this occurs within Cav-containing lipid rafts. 

Gap junction assembly is a dynamic process in living cells. In the cytoplasm of oocyte cells, Cx50 hemichannels are coated in 0.1- to 0.5-μm diameter vesicles of 5 to 40 Cx50 hemichannels. These vesicles traffic to and fuse with plasma membranes, and thousands of vesicles are inserted every second. ³¹ Segregation and aggregation of gap junctions in lipid rafts of plasma membranes could provide a membrane microdomain for plaque assembly. Most reports have indicated that phosphorylation of the C terminus of connexins at serine and threonine or Cavs at serine, threonine, and tyrosine promote gap junction disassembly or caveolae translocation, respectively. ^{10 11 12 13 15 16 32 33} Our present study shows that stimulation by TPA or IGF-1 caused activation of PKCγ, which enhanced the interaction of PKCγ with Cx43 and Cav-1, caused redistribution in lipid rafts, and caused gap junction plaque disassembly. Furthermore, overexpression of functional PKCγ-EGFP decreased gap junctional Cx43 plaques, although nonactive overexpressed PKCγ did not. These data suggest that active PKCγ might be translocated to lipid rafts after stimulation, which, in turn, enhances the interactions with and phosphorylation of Cx43 and Cav-1, reduces cell surface Cx43 gap junction plaques, and redistributes Cx43 with Cav-1 in lipid rafts. 

According to our experimental results and previous reports, ^{29 34} we hypothesize that the dynamic equilibrium between gap junction plaques and single or oligomeric gap junction channels at plasma membrane cell surfaces may be achieved through redistribution in lipid rafts. Stimulation by growth factors, such as IGF-1, would quickly disassemble functional gap junction plaques through activation of PKCγ. This would be transient, as growth factors provide such signals. Earlier formed plaques would be disaggregated into single or oligomeric gap junction channels and move within lipid rafts out of gap junction plaques. Thus, Cav-1 may associate with Cx43 and may work like a molecular chaperone or scaffold. To verify this, more experiments should be performed to demonstrate the phosphorylation of Cav-1 and Cx43 and regulation of both the interaction between Cav-1 and Cx 43 and gap junction activity, the changes of cell cytoskeletal proteins and their contribution to movement of Cx43 within lipid rafts, and the functional effects in situ using growth factors. 

 This is publication 03-279-J of the Kansas Agricultural Experiment Station.

 Supported by National Eye Institute Grant EY13421 (DJT).

 Submitted for publication March 21, 2003; revised June 24, 2003; accepted July 14, 2003.

 Disclosure: D. Lin, None; J. Zhou, None; P.S. Zelenka, None; D.J. Takemoto, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

TPA and IGF-1 enhanced the interactions of PKCγ with Cav-1 and Cx43 in rabbit lens epithelial N/N1003A cells and bovine primary lens epithelial cells. (A) Presence of Cavs in N/N1003A cells and bovine primary lens epithelial cells. Twenty micrograms of proteins from whole-cell lysates were loaded in each lane, separated on 10% SDS-PAGE, and Western blotted with antibodies against Cav-1 (1:1000) or Cav-2 (1:250). Cav-1 and -2, but not -3 (not shown), were present in N/N1003A and bovine primary lens epithelial cells. (B) N/N1003A cells from a 95% confluent flask (75 cm²) were extracted with 500 μL cell lysis buffer, and the whole-cell extracts were separated by centrifugation at 2000g for 20 minutes. Lane 1: pellets dissolved as samples. The supernatants were separated by 12,000g for 20 minutes. Lane 2: pellets were dissolved as samples, and the consequent supernatants were finally separated by 100,000g for 1 hour at 4°C. Lanes 3 and 4: pellets and supernatants, respectively, collected by this final spin. All the pellet samples collected by continuation of centrifugation at different speeds were dissolved in 500 μL protein loading buffer. Twenty microliters of protein samples were resolved by SDS-PAGE and were immunoblotted with the desired antibodies, as indicated. There were small amounts of PKCγ, Cx43, and Cav-1 in the pellets precipitated by centrifugation at 2000g (lane 1). Lane 2: very low levels or none of PKCγ, Cx43, or Cav-1 were detected in the fractions, which were pelleted from the supernatants of lane 1 by further centrifugation of 12,000g. PKCγ was mostly in the supernatants (lane 4) and pellets (lane 3) after 100,000g centrifugation. Cx43 and Cav-1, however, were largely abundant in the pellets (lane 3) and also some in the supernatants (lane 4) centrifuged by 100,000g. (C) Coimmunoprecipitation of Cav-1, PKCγ, and Cx43 in confluent N/N1003A cells that were serum starved for 2 hours and treated with 300 nM TPA or 25 ng/mL IGF-1 for 30 minutes. Cont: control cells with no treatments. The whole-cell lysates containing 2 mg total proteins were immunoprecipitated with 2 μg/mL monoclonal antibodies against Cav-1, Cx43, or PKCγ at 4°C for 4 hours or overnight. Nonspecific rabbit IgG (NS) was used as the negative control. Coimmunoprecipitated complexes were subjected to Western blot by use of anti-Cx43, anti-PKCγ, or anti-Cav-1 as probes. IGF-1 and TPA enhanced the interaction of Cav-1 with PKCγ; however, Cx43 interacted with Cav-1 in the presence or absence of TPA or IGF-1. IP, immunoprecipitation antibody; IB, immunoblot antibody. (D) Coimmunoprecipitation of Cav-1, PKCγ, and Cx43 in bovine primary lens epithelial cells. Bovine primary lens epithelial cells at passage 4 were serum starved for 2 hours, and then treated with 300 nM TPA or 25 ng/mL IGF-1 for 30 minutes. Whole-cell lysates (2 mg of total proteins) were subjected to coimmunoprecipitation and Western blot as for (C). All samples were observed in triplicate, and the examples are shown. The interactions between Cav-1 or Cx43 and PKCγ were enhanced by IGF-1 or TPA. Cx43 interacted with Cav-1 in the presence or absence of TPA or IGF-1. IP, immunoprecipitation antisera; IB, antisera used in immunoblotting.

Figure 2.

 

Cx43 colocalized with Cav-1 in Triton-resistant membrane fractions in N/N1003A cells. (A) Cx43 localization in Triton-resistant membrane fractions. Serum-starved N/N1003A cells were treated with 300 nM TPA or 25 ng/mL IGF-1 for 30 minutes, and Triton X-100–resistant fractions were prepared. Proteins (20 μg/well) were separated by 8% SDS-PAGE and subjected to immunoblotting using Cx43 or Cav-1 antisera as probes. Both Cx43 and Cav-1 were localized in Triton-resistant membrane fractions. Neither IGF-1 nor TPA induced significant localization of Cx43 or Cav-1 out of detergent-resistant fractions compared with the control. (B) Coimmunoprecipitation of Cav-1 with Cx43 and PKCγ in fractions obtained by sucrose gradient sedimentation. The sucrose gradient fractions 3 to 7 (upper) and 3 to 5 (lower), from a 5% to 35% sucrose gradient, as shown in Figure 3 , were collected, combined, and resolubilized together by sonication with addition of 0.1% SDS. The mixtures were considered to be the Cav-containing lipid raft fractions. These pooled fractions were subjected to coimmunoprecipitation with Cav-1 antibody and immunoblotted as described in (A). The experiments were run in triplicate. Cx43 was always immunoprecipitated with Cav-1 in lipid rafts in the absence or presence of stimulation by IGF-1 or TPA. PKCγ was translocated to lipid rafts and was coimmunoprecipitated with Cx43 and Cav-1 after stimulation of cells by TPA or IGF-1. This demonstrates that the proteins in the lipid raft fractions—PKCγ, Cx43, and Cav-1—can be physically coimmunoprecipitated. IP: immunoprecipitation antisera. IB: antisera used in immunoblotting.

Figure 3.

 

Movement of Cx43, Cav-1, and PKCγ in lipid raft domains. Serum-starved N/N1003A or bovine primary lens epithelial cells were treated with 25 ng/mL IGF-1 or 300 nM TPA for 30 minutes. The whole-cell lysates were extracted with cell lysis buffer containing 1% Triton X-100 and separated on 5% to 35% sucrose continuous-density gradients. Five milligrams of total proteins were loaded per gradient. Twelve 1-mL fractions were collected from the top of each gradient. Proteins were precipitated by 10% TCA, separated by SDS-PAGE, and immunoblotted with particular antisera as indicated. IB: antisera used in immunoblotting. (A) Cx43 colocalized with Cav-1 in lipid rafts in N/N1003A cells. IGF-1 or TPA stimulated Cx43 to shift with Cav-1 to higher density fractions within lipid rafts. Activation of PKCγ by TPA or IGF-1 promoted PKCγ redistribution from out-of-lipid rafts into lipid rafts at slightly higher density fractions. (B) Cx43 colocalized with Cav-1 in lipid rafts in bovine primary lens epithelial cells. Movement of Cx43 with Cav-1 to higher density fractions was induced by IGF-1 or TPA. Activation of PKCγ by TPA promoted movement of PKCγ into lipid rafts containing Cx43 and Cav-1.

Figure 4.

 

PKCγ-EGFP fusion proteins are localized in both soluble and membrane fractions. (A) Overexpressed PKCγ-EGFP or PKCγDm-EGFP were localized in both cytosol and membranes of N/N1003A stably transfected cells. The confluent cells were harvested and extracted with Tris-MgCl₂ buffer without Triton X-100. The supernatants and membrane fractions were separated by ultracentrifugation at 35,000 rpm (100,000g) for 1 hour at 4°C. Ten micrograms of protein samples were subjected to SDS-PAGE and immunoblotted with monoclonal anti-PKCγ. (B) Exogenous PKCγ-EGFP or PKCγDm-EGFP translocated partially to the membranes after stimulation by IGF-1 or TPA. Stably transfected cells overexpressing PKCγ-EGFP, PKCγDm-EGFP, or EGFP were seeded in culture dishes for overnight culture. TPA (300 nM) or IGF-1 (25 ng/mL) was applied to 2-hour serum-starved cells. EGFP fluorescence was measured, and the images were collected at 30 minutes by laser scanning microscope. Bars, 5 μm. (C) IGF-1 and TPA enhanced endogenous and exogenous PKCγ-specific activity in N/N1003A cells. N/N1003A cells and stably transfected N/N1003A cells overexpressing PKCγ-EGFP, PKCγDm-EGFP, or EGFP were serum starved for 2 hours, then stimulated with IGF-1 (25 ng/mL) or TPA (300 nM) for 30 minutes. Whole-cell lysates (2 mg proteins) were immunoprecipitated with 2 μg anti-PKCγ (for lysates from N/N1003A cells) or 2 μg anti-EGFP (for lysates from transfected cells) at 4°C for 4 hours. The coimmunoprecipitation complexes were used as enzyme samples, and their PKCγ–specific activity was analyzed. Endogenous PKCγ activity was significantly enhanced by IGF-1 or TPA, as was overexpressed PKCγ-EGFP. Double mutant T655/674A (PKCγDm-EGFP) completely abolished PKCγ enzyme activity. *P < 0.05 was considered to be statistically significant. Results are the mean ± SEM of triplicate samples.

Figure 5.

 

Overexpression of PKCγ-EGFP, but not PKCγDm-EGFP, caused disassembly of cell surface Cx43 plaques. The transfected cells were fixed with 2.5% paraformaldehyde for 5 minutes and labeled with rabbit anti-Cx43 for 2 hours at room temperature. They were then washed and incubated with the secondary antisera Alexa Fluor 568, washed, and mounted on slides. Slides were examined by laser scanning confocal microscope. The images were collected by use of FITC (green) and TRITC (red) filters, and the merged images were shown as examples (A). Transfected cells were visualized in green (EGFP) and red (Cx43). The single red areas showed the nontransfected cells labeled only with Alexa Fluor 568. (B) The plaques larger than 1 μm from single transfected cells in each set (total 30 cells per set) were counted (white arrow) in triplicate experiments. P < 0.05 was considered to be statistically significant. Overexpression of either EGFP or PKCγDm-EGFP did not affect cell surface Cx43 plaques. Active exogenous PKCγ-EGFP significantly decreased Cx43 plaques. Scale bar, 10 μm.

Figure 6.

 

(A) PKCγ-EGFP and PKCγDm-EGFP proteins coimmunoprecipitated with Cav-1 and Cx43 in stably transfected N/N1003A cells. Stably transfected cells were harvested, and whole-cell lysates were extracted. The whole-cell lysates (2 mg) were coimmunoprecipitated with 2 μg monoclonal anti-EGFP (overexpression samples) or anti-PKCγ (N/N1003A nontransfected cells). The immunoprecipitation complexes were subjected to Western blot as shown. The band densities were digitized as described previously, ⁸ and PKCγ bands were set as 1 unit (row 3) and the blank (lane 1) as 0. The relevant pixel densities were calibrated based on each single-lane PKCγ band. Lane 1: cells transfected with EGFP only; lane 2: nontransfected cells; lane 3: cells transfected with PKCγ-EGFP; lane 4: cells transfected with PKCγDm-EGFP. IP: immunoprecipitation antisera; IB: antisera used in Western blot. (B) Quantitation of pixel density. Both PKCγ-EGFP (lane 3) and PKCγDm-EGFP (lane 4) fusion proteins interacted with Cav-1 or Cx43 in the absence of stimuli. These associations were much stronger than that of endogenous PKCγ (lane 2, and compare with Figs. 1C 1D ). As the negative control, lane 1 revealed that EGFP did not interact with Cx43 or Cav-1.

Figure 7.

 

Redistribution of Cx43 and Cav-1 was induced by overexpression of PKCγ-EGFP, but not of PKCγDm-EGFP. Serum-starved stably transfected cells were collected and extracted with cell lysis buffer containing 1% Triton X-100 and separated on 5% to 35% sucrose continuous density gradients, as described in the text. Five milligrams total proteins were loaded per gradient. Twelve 1-mL fractions were collected from the top of each gradient. Proteins were precipitated by 10% TCA, separated by SDS-PAGE, and immunoblotted with particular antisera, as indicated. Most of Cx43 and Cav-1 were in fractions 3 and 4 in control cells (transfected EGFP only). Overexpression of PKCγ-EGFP caused a shift of Cx43 and Cav-1 out of fraction 3 into higher density fractions. However, overexpression of loss-of-function PKCγ did not change the detergent-resistant membrane patterns when compared with transfected EGFP cells (control). Endogenous PKCγ from untreated cells was localized in fractions throughout and especially in higher density fractions. PKCγ-EGFP was especially enriched in fractions 5 and 6, similar to Cav-1 and Cx43, but PKCγDm-EGFP was mostly enriched in fraction 4 similar to Cav-1 and Cx43 localization in untransfected cells. Active PKCγ was essential for the overexpression of PKCγ-induced movement of both Cav-1 and Cx43 within lipid rafts. IB: antisera used in Western blot.

The authors thank Guido Zampighi of University of California at Los Angles and Larry J. Takemoto of Kansas State University for helpful discussions.