PKCγ knockout mouse lenses are more susceptible to oxidative stress damage

Cataracts, or lens opacities, are the leading cause of blindness worldwide. Cataracts increase with age, diabetes and some environmental insults, e.g. oxidative stress. Lens homeostasis and transparency depend on the function of gap junctions. Knockout and/or missense mutations of lens gap junction proteins, Cx46 or Cx50, result in cataractogenesis(Gong et al., 1997; White et al., 1998; Xia et al., 2006). However,the regulatory mechanism by which gap junctions prevent lens opacification is less understood.

Protein kinase C γ (PKCγ) is a conventional PKC, primarily found in central nervous and peripheral nervous systems, particularly abundant in cerebellar Purkinje cells and hippocampal pyramidal cells(Shutoh et al., 2003). In the lens, it is found in anterior epithelium and the fiber cells of the cortex,but not in the deeper fibers constituting the lens nucleus(Saleh et al., 2001). As with most conventional PKCs, diacylglycerol, calcium and oxidative stress signals(e.g. H₂O₂) activate this isoform. Upon activation, the soluble PKCγ translocates to the plasma membrane, and phosphorylates targets such as receptors, structural proteins, ion channels and gap junctions(Lin and Takemoto, 2005; Zampighi et al., 2005). Numerous studies indicate that the phosphorylation of such a wide spectrum of proteins appears to prevent brain ischemia and plays a protective role in cell survival (Aronowski et al.,2000).

PKCγ is the principal enzyme involved in the phosphorylation of gap junctions in the lens (Lin and Takemoto,2005; Saleh et al.,2001; Zampighi et al.,2005). Gap junctions regulate the cell-to-cell pathway responsible for the movement of nutrients and waste products from the surface into the lens interior (Zampighi et al.,2005). This pathway consists of specialized channels, named gap junctions, which link directly to the cytoplasm of adjacent cells. Gap junction channels are composed of members of the connexin family arranged as hemi-channels (hexamers) joined through their external domains in the extra-cellular space (Kistler et al., 1988). The complete channels are arranged in large aggregates called plaques(Baldo et al., 2001; Kistler et al., 1988; Konig and Zampighi,1995; Zampighi et al.,2005).

In lens epithelium, gap junctions consist of a dominant connexin 43 (Cx43)and a minor Cx50. In lens fiber cells, the gap junctions are comprised of equal amounts of Cx46 and Cx50 (Kistler et al., 1988; Konig and Zampighi, 1995). PKCγ controls gap junctions in the lens epithelium and cortex during oxidative stress. Oxidative activation of PKCγ uncouples both Cx43 and Cx50 gap junctions to the passage of fluorescent dyes, through phosphorylation of both connexins (Lin et al.,2004; Zampighi et al.,2005). Cx50 gap junction channels are disassembled into hemi-channels that remain in the plasma membrane of the fiber cells(Zampighi et al., 2005). This occurs within the lipid rafts, the microdomains in plasma membranes.

We took advantage of the observation that deletion of PKCγ produces viable mice with a slight ataxia, less memory, reduced neuropathic pain,decreased anxiety, impaired long-term potentiation, enhanced opioid responses,increased alcohol consumption and less protection against brain ischemia(Abeliovich et al., 1993; Aronowski et al., 2000; Bowers and Wehner, 2001; Malmberg et al., 1997; Narita et al., 2001; Ohsawa et al., 2001; Verbeek et al., 2005). Such mice have very small litters and short breeding lifespan (D. Lin et al. unpublished observation). We demonstrated that the loss of PKCγ control of gap junctions greatly increases opacification in the lens of PKCγknockout mice. Deletion of PKCγ altered the arrangement of gap junctions on the cortical fiber cell surface, caused failure of connexin 50 phosphorylation and the loss of gap junction uncoupling response after H₂O₂ as measured by fluorescent dyes transfer. These results demonstrate that PKCγ is essential for the protection of the lens from environmental/oxidative damage.

Monoclonal antibodies against PKCγ, caveolin 1 and Cx43 were obtained from BD Biosciences (Palo Alto, CA, USA); monoclonal mouse anti-Cx50 (amino acids 290-440) was from Zymed Laboratories (South San Francisco, CA, USA);anti-aquaporin 0 (AQP0), polyclonal rabbit anti-phosphothreonine (pT) and anti-phosphoserine (pS) were from Chemicon (Temecula, CA, USA); nonspecific rabbit IgG and protein-agarose beads (A/G PLUS) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA); anti-mouse or anti-rabbit IgG conjugated with horseradish peroxidase (HRP), and nonradioactive PKC assay system(PepTag) were obtained from Promega (Madison, WI, USA); Dulbecco's modified Eagle's medium (DMEM; low glucose), gentamycin, and penicillin-streptomycin were from Invitrogen-Life Technologies (Carlsbad, CA, USA); hydrogen peroxide(H₂O₂) and sodium fluoride (NaF) were purchased from Fisher Scientific (Pittsburgh, PA, USA); phenylmethylsulfonyl fluoride (PMSF)and protease inhibitor cocktail were from Sigma-Aldrich (St Louis, MO, USA);optimal cutting temperature (OCT) compound was obtained from Sakura Finetec(Torrance, CA, USA); Lucifer Yellow, rhodamine-dextran were from Molecular Probes (Eugene, OR, USA); and 12-O-tetradecanoylphorbol 13-acetate(TPA) was purchased from CalBiochem (San Diego, CA, USA).

Male and female mice (Mus musculus) at 2 days to 16 weeks of age were used in this study. Both the control mice (b6129pf21j100903) and PKCγ knockout mice (B6;129p-Prkcctm1St1) were initially purchased from Jackson Laboratory (Bar Harbor, MA, USA), as breeders, then maintained as a colony at Kansas State University Animal Resource Facility. This was necessary as animals younger than 1 week are not available for purchase. All experiments were performed according to an institutionally approved animal protocol.

All lenses were removed immediately and fixed in a solution of 2%paraformaldehyde, 2.5% glutaraldehyde, 0.1 mol l^-1 cacodylate. Lenses were post-fixed with osmium tetroxide, dehydrated, and embedded in Epon(LX112). Sections (1 μm thick) were stained with Toluidine Blue, and viewed and photographed under a Nikon microscope.

The eyes from 6-week-old control or PKCγ knockout mice were removed from the orbit, opened and immersed in the fixative solution composed of 3%glutaraldehyde in 0.2 mol l^-1 sodium cacodylate pH 7.3 for 2-4 h at room temperature. The lenses were washed in 0.1 mol l^-1 cacodylate and cut in halves to improve infiltration. They were postfixed in 1%OsO₄ for 90 min at room temperature, washed in 0.1 mol l^-1 sodium acetate (pH 5.0) and immersed in 0.5% uranyl acetate in 0.1 mol l^-1 sodium buffer for 12 h at 4°C. Dehydration was performed in a graded series of ethanol, passed through propylene oxide and infiltrated in Polybed 812 under vacuum(Zampighi et al., 2000). Thin sections (1 μm) were cut and were infiltrated in 20% glycerol in 0.1 mol l^-1 sodium cacodylate. The samples were frozen by immersion in liquid propane and stored in a liquid nitrogen refrigerator. They were transferred into a JEOL freeze-fracture-etch apparatus and fractured at-120°C with a knife cooled at liquid nitrogen temperature. The fracture faces were shadowed with platinum at 70° and carbon at 90°. The replicas were cleaned with sodium hypochloride and deposited on single hole,Formvar-coated grids (Zampighi et al.,2000).

Lenses were immediately removed from eyes after mice were killed, and incubated in serum-free DMEM medium at 37°C in a CO₂ incubator for 24 h. At that time all transparent lenses were used for further treatments. Six lenses per group were used. The transparent lenses were treated with H₂O₂ or phosphate-buffered saline (PBS) at 37°C in a CO₂ incubator for an additional 24 h. H₂O₂ was added into the serum-free DMEM medium every 4 h for a total of six times. After 24 h, the lenses were washed with PBS, and were examined for opacification as described previously(Behndig et al., 2001). Contrast ratios (maximum to minimum transmitted signal) were calculated from digitized images using UN-SCAN-IT 6.1.

PKCγ activity was analyzed using a PepTag Assay kit. Equal amounts of whole cell protein extracts from whole lens were immunoprecipitated with PKCγ antisera. Immunoprecipitated PKCγ/agarose bead complexes were incubated with a PKC reaction mixture (Lin and Takemoto, 2005). To stop the reactions, samples were boiled at 95°C for 5 min. Then the PKCγ reaction products (fluorescent PepTag peptides) were resolved by agarose gel electrophoresis and visualized under UV light. The phosphorylated peptide bands were excised, and fluorescence intensities were quantified by spectrophotometry, according to the manufacturer's instructions. Results are expressed as the percentage of nontreated specific PKCγ activity from an average of three experiments± s.e.m.

The whole lenses were homogenized with cell lysis buffer followed by sonication. The cell lysis buffer contained 20 mmol l^-1 Tris-HCl,pH 7.5, 0.5 mmol l^-1 EDTA, 0.5 mmol l^-1 EGTA, 0.5%Triton X-100, 0.1% protease inhibitor cocktail, 5 mmol l^-1 NaF, 5μmol l^-1 Na₃VO₄ and 2 mmol l^-1PMSF. After centrifugation at 12 000 g for 20 min, the supernatants were collected and used as whole cell extracts. Whole cell extracts were immunoprecipitated with anti-Cx50 at 4°C for 4 h, as described (Lin et al., 2003). The immunoprecipitate/agarose bead complexes were resolved by SDS-PAGE and visualized by western blot with antisera to phosphoserine (pS),phosphothreonine (pT), and/or connexin 50 (Cx50). In some cases, whole cell homogenates were blotted using antisera.

Whole-lens gap junction activity was analyzed by dye-transfer assay, as described previously (Zampighi et al.,2005). Six-week-old PKCγ knockout or control mice were killed with CO₂ and the lenses were removed immediately after the death of the animal and washed in PBS. Lenses were incubated with 50 μmol l^-1 H₂O₂ for 20 min in 2 ml serum-free DMEM(low glucose) medium. Lucifer Yellow (2.5 mg ml^-1 in PBS) was microinjected as described previously(Zampighi et al., 2005), and the lenses were subsequently incubated in serum-free DMEM at room temperature for 30 min, to allow dye transfer. Rhodamine-dextran (1%) was injected with Lucifer Yellow and used as a control for nonspecific leakage. A total 126 nl of Lucifer Yellow and rhodamine-dextran was injected into the superficial cortical fibers (around 20 μm in depth) per injection site, with a microinjection apparatus (Nanoliter 2000; World Precision Instruments, Inc.,Sarasota, FL, USA). After incubation, the lenses were fixed in 2.5%paraformaldehyde, dissected, and mounted in 3% agar. Lens gap junction dye transfer was measured using confocal microscopy. The extent of Lucifer Yellow dye transfer minus the distance of rhodamine-dextran (in μm) was expressed as gap junction dye transfer activity. Each experimental group contained six lenses and the distance of dye transfer was determined in six areas of the bow region of each lens in coded samples. Results are expressed as mean ±s.e.m., with P≤0.05.

Structural studies using light microscopy showed normal lens development at 2 days postnatal in both normal and knockout mice. A typical v-shape nuclei arrangement was observed in the lenses of both control and PKCγ knockout mice (Fig. 1). In particular,nuclei were lost in the typical organ free zone, indicating that loss of PKCγ did not alter epithelial to fiber cell differentiation. The lenses of 6-week-old PKCγ knockout mice were substantially different to those of wild-type control mice. Large vacuoles were found in the epithelia and outer cortex.

Oxidative stress is a significant damaging factor to lens homeostasis. To test if deletion of PKCγ causes intact lenses to be more susceptible to environmental insults, we treated the intact lenses in culture with H₂O₂, and lens opacification was determined as shown in Fig. 2. H₂O₂ induction of opacity in vitro has been used as a model to determine lens susceptibility to opacification(Lin et al., 2005). In each treatment, six clear lenses were initially incubated in serum-free DMEM for 24 h. Three out of six PKCγ knockout lenses were opaque after this initial incubation, whereas only two of six control lenses were opaque. For the H₂O₂-triggered cataract assay only clear lenses were incubated with H₂O₂ for 24 h (every 4 h, or six times,1-10 μmol l^-1). PBS was used as a control treatment. The surprising result is that, even without H₂O₂, after 48 h incubation only, one of six knockout lenses remained clear whereas four of six controls were clear. As previously reported, control lenses showed increased opacity with increasing H₂O₂ concentrations. Simple removal of lens from knockout eyes resulted in cataract formation. This clearly demonstrated an increased tendency to form opacities even with the stress of surgical removal.

To address whether lack of the PKCγ enzyme would account for lens opacification through improper control of gap junctions, we first determined if activation of PKCγ by activators, TPA and H₂O₂,could be demonstrated. In control lenses, endogenous PKCγ was activated by TPA and H₂O₂. Very low or no PKCγ enzyme activity was detected before or after activation in the PKCγ knockout lenses (Fig. 3A). These results demonstrate no detectable PKCγ enzyme activity in lenses from PKCγknockout mice, consistent with western blotting results(Fig. 3B).

Connexin 43 (Cx43) and Cx50 are found in lens epithelium, whereas Cx46 and Cx50 are in lens fibers. We have found that PKCγ differentially phosphorylates Cx46 and Cx50 (Lin et al.,2004). Since the activation of PKCγ disassembles only Cx50 channels in whole lenses (Zampighi et al.,2005), we measured the phosphorylation of Cx50 in the control and PKCγ knockout mice. This would allow us to determine if other PKCs could compensate. Lenses were treated with TPA or H₂O₂, both of which are strong activators of PKCα andγ, both found in lens(Lin et al., 2004; Zampighi et al., 2005). The total Cx50 from lens whole cell extracts were immunoprecipitated and resolved by SDS-PAGE. Phosphorylation of Cx50 on serines and threonines was determined by immunoblotting with antibodies against phosphoserine (pS) or phosphothreonine (pT) as shown in Fig. 4. Treatment with TPA or H₂O₂ caused phosphorylation of Cx50 on both serines and threonines in the lenses from control mice (Fig. 4). However,no detectable Cx50 phosphorylation was observed in the samples prepared in the same manner but from knockout lenses (Fig. 4). Therefore, in lenses from control animals phosphorylation of Cx50 is controlled primarily by activation of PKCγ and other PKCs do not compensate for loss of PKCγ, i.e. Cx50 is not phosphorylated.

To determine whether the phosphorylation of Cx50 results in inhibition of lens fiber gap junctions, we measured diffusion of the dye Lucifer Yellow into untreated or H₂O₂-treated lenses for 20 min(Fig. 5). In untreated control or PKCγ knockout lenses, the dye moved ∼86 μm into the lens where Cx50 is found (Zampighi et al., 2003). After H₂O₂treatment, the dye diffused only ∼32 μm into the cortex in the control lenses. In contrast, the diffusion of Lucifer Yellow into the cortex remained unchanged (∼94 μm) in the PKCγ knockout lenses after H₂O₂. The results indicate that the deletion of PKCγ abolishes the effects of H₂O₂ on inhibition of lens gap junctions in PKCγ knockout mice, however dye transfer is observed. Therefore, the gap junctions transfer dye but are not controlled properly and may be open.

Finally, we wished to demonstrate whether the deletion of PKCγ alters morphology of gap junctions in lens fiber cells. Fiber cell surface gap junctions were revealed by freeze-fracture as shown in Fig. 6. We found that a significant difference between lenses of control and knockout mice was a decrease in the frequency and arrangement of the so-called `tongue-and-groove'junctions (Simon et al., 1982; Zampighi et al., 1982). In wild-type controls, these junctions appeared as hemispheres of ∼0.5 μm in diameter that protrude within the cytoplasm of one of the fibers. In the knockout mice, these junctions were smaller and were seen with a lower frequency in the surface of the fibers. There was a similar frequency of gap junction plaques although they appeared more fragmented in the knockout animals. In controls, the gap junction plaques measured 0.7-0.8 μm in diameter (red areas, Fig. 6A,C). In the knockout mice, fiber cells also contained large plaques (Fig. 6B). In addition,there were large numbers of small plaques that were composed of up to 50 channels and were randomly distributed in the plasma membrane (small red areas, Fig. 6B,D). These small plaques were not seen in fibers from controls. The density of channels in both controls and knockout plaques was comparable (∼4,500 μm^-2). Taken together, our results suggest that open gap junctions in the PKCγknockout mice cause lenses to be more sensitive to stress.

Gap junctions provide a major regulatory network to coordinate lens growth and to maintain lens transparency throughout life. We have previously reported that PKCγ acts as an oxidative stress sensor and controls gap junctions in the lens epithelial cells and whole lens in culture. In this work we took advantage of PKCγ knockout mice, to determine effects of loss of PKCγ on subsequent lens homeostasis. Our study offers conclusive evidence that control of gap junction channels, especially Cx50 in the lens,by activated PKCγ protects lenses from stress damage. Deletion of PKCγ in the knockout mice eliminates phosphorylation of Cx50 and results in an inability to uncouple Cx50 in response to H₂O₂which, in turn, results in lens fiber cells being more susceptible to oxidative stress and subsequent lens opacification. We, thus, conclude that PKCγ is a critical regulatory enzyme which controls lens gap junctions,and its loss in the knockout mice may account for the extensive fiber damage and lens opacification through improper control of lens gap junctions by PKCγ.

PKCγ is a neural-specific, conventional PKC isoform, which is involved in multiple signaling pathways(Oancea and Meyer, 1998; Tanaka and Nishizuka, 1994; Zeidman et al., 1999). Numerous publications have demonstrated that PKCγ is also present in the lens (Berthoud et al., 2000; Saleh et al., 2001; Lin et al., 2004; Zampighi et al., 2005). Functional studies have revealed that Cx43 gap junctions are one of the major targets for activated PKCγ in the lens epithelium. PKCγ is activated by oxidative stress, e.g. H₂O₂, through oxidation of the C1B domain (Lin and Takemoto, 2005). Recently, missense mutations in the PKCγC1B domain have been identified as causing a type of autosomal dominant neurodegenerative disorder, spinocerebellar ataxia type 14 (SCA-14)(Chen et al., 2003; Chen et al., 2005; Alonso et al., 2005; Seki et al., 2005). However,it is still unclear how these C1B mutations lead to loss of Purkinje neuron and cerebellar dysfunction during ataxia. Our observations that PKCγtargets specifically the gap junction channels might also be relevant to the mechanisms involved in this disease.

Gap junctions are hydrophilic channels connecting adjacent cells. They provide a free passage of both necessary metabolites and apoptotic signals from cell to cell (Farahani et al.,2005). Thus, gap junctions play important roles in lens homeostasis when properly controlled (Gao et al., 2004; Mathias and Rae,2004; Gerido and White,2004; Yu et al.,2005; Zampighi et al.,2005). However, the passage of apoptotic signals such as high Ca²⁺ or ATP to an adjacent cell through open gap junctions could be linked to cell death (Cusato et al.,2003; Lin et al.,1998). The process is well documented in neural cells and immune cells (Cusato et al., 2003; Lin et al., 1998; Vinken et al., 2006). This work provides an in vivo model to better understand how control of gap junctions by PKCγ relates to lens growth and development. We showed previously that PKCγ differentially regulates Cx43, Cx46 and Cx50 gap junctions in the lenses. Cx50, found in the lens, is now also known to be extensively present in neural systems, for example, in the optic nerve and central nervous system projections(Schutte et al., 1998). Cx50 knockout results in a small lens with less differentiation, suggesting that Cx50 is required for lens differentiation and cell growth as well(White et al., 1998; Rong et al., 2002). Cx50 was recently found to bind to calmodulin and to be regulated by Ca²⁺(Zhang and Qi, 2005). We measured PKCγ activation in both PKCγ knockout and control mice and found that Cx50 phosphorylation and lens gap junction activity are specifically controlled by PKCγ. The total lack of Cx50 phosphorylation in the knockout mice clearly demonstrates that PKCγ phosphorylates and controls Cx50. Further, this cannot be compensated by other PKCs. We have demonstrated that the deletion of PKCγ causes lens fiber cells to have altered gap junctions, which are open and uncontrolled, and this subsequently results in fiber cell damage. A life-time accumulation of sublethal oxidative stress damage through open gap junctions would be harmful. Oxidative stress is one of the most common causative agents of disease(Cutler, 2005). Prolonged oxidative stress (such as long term H₂O₂ treatment for 24 h) causes protein oxidation, aggregation, degeneration and cell death. In our current model, sublethal H₂O₂ induced cataract formation in control mice after 48 h. In contrast, the lenses of knockout mice become opaque even without H₂O₂. This may be due to open gap junctions which cannot control entry of components from the medium(Fig. 5). It is thus possible that regulation of gap junctions via PKCγ is essential for survival of cells, not just in lenses but also in other tissues where this critical enzyme is expressed. Therefore, studies of the interplay of gap junction channels and PKCγ in control and knockout animals might provide useful information for understanding how tissues respond and deal with stress and disease.

This study is funded by grants from the National Eye Institute, R01 EY-13421 to D.J.T., and EY-04011 to G.A.Z. and a Howard Hughes Undergraduate Fellowship to S.L.