Oxidative Stress Increases Production of (cid:98) -Amyloid Precursor Protein and (cid:98) -Amyloid (A (cid:98) ) in Mammalian Lenses, and A (cid:98) Has Toxic Effects on Lens Epithelial Cells*

Many amyloid diseases are characterized by protein aggregations linked to oxidative stress. Such diseases including those of the brain, muscle, and blood vessels exhibit plaques containing (cid:98) -amyloid (A (cid:98) ). Here we demonstrate that Alzheimer’s precursor protein ( (cid:98) APP) and A (cid:98) are present at low levels in normal lenses and increase in intact cultured monkey lenses treated with H 2 O 2 or UV radiation (known cataractogenic agents), and with phorbol 12-myristate 13-acetate. AP-1 factor binding, shown by others to up-regulate (cid:98) APP expression, increased in the monkey lenses treated with H 2 O 2 , UV radiation, or phorbol 12-myristate 13-acetate and paralleled the increase in (cid:98) APP expression. Rat lenses exposed to oxidative stress showed increased (cid:98) APP in the anterior epithelium and cortex. Incubation of cul- tured rabbit lens N/N1003A epithelial cells with A (cid:98) induced inclusions and vacuoles and was cytotoxic. A (cid:98) cross-reacting protein was readily detected in the cor- tex of a cataractous human lens. Our data show that (cid:98) APP and A (cid:98) increase in mammalian lenses as part of a response to H 2 O 2 or UV radiation and suggest that they may contribute to the mechanism by which oxidative damage leads to lens opacification.

Cataract impairs vision by opacification of the ocular lens (1). A large percentage of cataracts are of the age-related (senile) type suggesting an environmental component to this degenerative disease (1)(2)(3). Both protein aggregation (4 -8) and vacuole formation (9 -13) are associated with several types of cataracts. Alzheimer's protein is associated with amyloid diseases of the muscle (14,15), brain (16,17), and blood vessels (18), which are also characterized by protein aggregates. A pathological hallmark of Alzheimer's disease (AD) 1 is the presence of amyloid plaques, which contain ␤-amyloid protein (A␤) in brain, a proteolytic cleavage product of the Alzheimer's precursor protein (␤APP) (15,18). In muscle fiber degeneration seen in neuromuscular diseases, including inclusion body myositis, A␤-containing plaques are found (14,15). A␤ plaques have also been identified in the kidneys and lungs of patients with Alzheimer's disease (19). In the human eye, A␤ and ␤APP have been detected in the retina and are associated with aging and retinal degeneration (20).
The ␤APP gene is found on chromosome 21 (21) and almost all patients with trisomy 21 (Down's syndrome) manifest AD, other A␤ plaque formation diseases, and cataracts (22)(23)(24). A 4 -5-fold increase in ␤APP expression over normal levels has been documented in cells from Down's syndrome patients (25). Interestingly, a 2-and 2.5-fold increase in ␤APP and A␤, respectively, has been demonstrated in fibroblasts from non-␤APP gene duplication familial AD associated with chromosome 14 (26).
In addition to A␤, neuritic plaques found in AD brain tissue also contain ␣B-crystallin, a major ocular lens protein (27). ␣B-Crystallin has been associated with a host of other neurodegenerative diseases (see Refs. 28 and 29 for reviews). ␣B-Crystallin is a small heat shock protein (30), possesses a chaperone-like anti-aggregative function (31), and mediates intermediate filament assembly in vitro (32). However, the possible role of ␣B-crystallin in either promoting or inhibiting amyloid filament formation is not known.
A␤ belongs to a family of amyloidogenic proteins (16), which share a strong predisposition to form insoluble ␤-sheet structures leading to fibrillar aggregation (33). A␤ and other amyloid proteins contain a peptide consensus repeat motif: GXXX (where X is one of 11 possible residues), that has been implicated in promoting nucleation of amyloid fibrils (34,35). The seeding of protein aggregate fibrillar structures by amyloid proteins like A␤ has been proposed as part of the pathogenic mechanism in amyloid disease (35).
A␤ has also been demonstrated to be cytotoxic; A␤ added exogenously to established neuronal or cerebrovascular smooth muscle cell cultures induced vacuoles and increased cell death (15, 36 -40). However, when added very early in the establishment of primary neuronal cell cultures, A␤ can produce a neurotrophic effect (36). ␤APP and specific cleavage products are normal constituents of many cell types and appear to be involved with facilitating membrane-associated functions (37,38). For example, in fibroblasts, ␤APP is released from cells into the medium and has an autocrine function in growth regulation (43).
Oxidative stress resulting from endogenous production of reactive oxygen species is strongly linked with amyloid disease (44,45) and cataract (1-3). A␤ toxicity was recently shown to be mediated by H 2 O 2 in PC12 cells (45). The potential of antioxidants in inhibiting both cataract and AD has been indicated. For example, populations with long term consumption of vitamin C (Ն60 mg/day) from foods and/or supplements have reduced risks of cataract (42,46). Experimentally, vitamin E was demonstrated to protect nerve cells from A␤ toxicity (40).
Cellular responses to oxidative stress from UV radiation and H 2 O 2 include herpes and human immunodeficiency virus induction and increases in specific gene expression (47-50) via activation of signaling pathways which involve the Ras (50) and Src (51) proteins, and include the transcription factors AP-1 (50) and NF-B (49). A direct role for AP-1 factors in increased ␤APP expression by phorbol ester (phorbol 12-myristate 13-acetate, PMA) treatment in human glial cells and HeLa cells has been described (52). PMA stimulation leads to increased AP-1 binding via a protein kinase C-mediated signaling pathway and has been used as a positive control in experiments to test AP-1 activation during oxidative stress (48,50). We have studied the effects of oxidative stress using rat and rhesus monkey lenses in organ culture as our model system. In the present report, we demonstrate that ␤APP and A␤ normally present in mammalian lenses are increased by oxidative stress, that AP-1 binding is increased in both cultured lenses and lens epithelial cells, and that A␤ is cytotoxic to cultured lens epithelial cells.

MATERIALS AND METHODS
Lens Organ Culture-Eyes were enucleated immediately following death from Sprague-Dawley rats and rhesus monkeys (Macaca mulatta, 2-3 years of age, part of the vaccine testing program of the Center for Biologics Evaluation and Research, Food and Drug Administration). Lenses were extracted and placed into 5.0 ml (monkey) or 2.0 ml (rat) of modified TC-199 medium (53,54). Lens integrity was assessed by measuring protein leached into the medium after 30 -60 min of culture, and damaged lenses were discarded (55). After equilibration under 5% CO 2 at 37°C, groups of lenses were exposed to H 2 O 2 or PMA, or to 254 nm UV radiation from an Ultraviolet Products mineralight (2 J/m 2 ). Lenses were homogenized with disposable pestles and then sonicated in extract buffer (20 mM HEPES, 0.2 mM EDTA, 0.5 mM dithiothreitol, 450 mM NaCl, 25% glycerol, 0.5 g/ml leupeptin, 0.5 g/ml aprotinin, 0.5 mM phenylmethanesulfonyl fluoride) on ice and cleared by low speed centrifugation. Protein concentrations were estimated using a Bradford protein assay kit (Bio-Rad).
Immunohistochemistry-Rat lenses were placed in culture as described above. After 1 h H 2 O 2 was added to a final concentration of 125 M or 250 M. After a 24-h incubation at 37°C and 5% CO 2 , the lenses were removed and placed in embedding compound and immediately frozen on dry ice. Frozen sections were prepared on glass slides. Sections were fixed with 4% formalin and stained with monoclonal anti-␤APP-A4 (Boehringer Mannheim), or anti-SV40 large T-antigen monoclonal antibody (Oncogene Science). Staining was carried out with a Vectastain ABC kit (Vector) and developed using a DAB kit (Vector) according to the manufacturer.
Human eyes were obtained from the National Disease Research Interchange. Human lenses were then frozen intact in embedding compound for sectioning. Staining was carried out with a monoclonal directed against A␤ (DAKO) and developed using a Vectastain ABC and DAB kits (Vector).
Tissue Culture-N/N1003A (56) rabbit lens epithelial cells were grown in tissue culture in Dulbecco's modified Eagle's medium and 10% fetal calf serum (Life Technologies, Inc.) in 30 mm plastic tissue culture dishes under 5% CO 2 . Cells were harvested and sonicated on ice in extract buffer as described above. Tyrophostin (30 M) (Life Technologies, Inc.) was added as described elsewhere (50, 51) 1 h prior to treatment with H 2 O 2 .
Preparation of A␤ Peptides--A␤-(1-40) and A␤-(25-35) peptides were purchased from Bachem Bioscience Inc. The commercial peptides were prepared according to the manufacturer and diluted directly into the culture medium.
Immunoblotting-Proteins were resolved by SDS-PAGE on 18% acrylamide gels and electroblotted onto Immobilon P membranes (Millipore). Following transfer, the membranes were probed with a rabbit polyclonal antibody (BMB) raised against A␤ and immune complexes were visualized with 125 I-Protein A (ICN). Molecular weights were determined using SeeBlue 4 -250-kDa standards (Novex).
Electrophoretic Mobility Shift Assays-Mobility shift assays were performed as described elsewhere (57) using ϳ40 g of lens extract/ assay tube. AP-1 consensus oligonucleotides (Santa Cruz) were used. Control experiments to test for specificity of AP-1 consensus oligonucleotides using nonspecific oligonucleotide yielded no co-migrating complex. Lens protein extracts were prepared as described above.
Reverse Transcriptase (RT)-dependent Polymerase Chain Reaction (PCR) of Mouse ␤APP-RNA was isolated from 8-week FVB/N mouse lenses. 1 g of RNA was subjected to primer extension with the oligo-nucleotide primer A (2129 -2109) of the mouse ␤APP cDNA sequence (GenBank accession number M18373) with RT according to the manufacturer (Promega). The resultant cDNA was amplified by PCR according to the manufacturer (Perkin-Elmer) with primers A and B (1655-1675).

RESULTS
To begin our investigation of the lenticular expression of Alzheimer's proteins following exposure to oxidative stress, we assayed the levels of ␤APP and A␤ proteins in clear intact monkey lenses. Levels of ␤APP and A␤ increased in the monkey lenses following treatment with either UV 254 nm radiation or H 2 O 2 or PMA. Lenses were removed, incubated overnight to acclimate in organ culture, and treated the following morning with 1 mM H 2 O 2 , UV 254 nm /8 min, or 100 ng/ml PMA and incubated for an additional 6 h unless otherwise indicated. Three cross-reacting bands (ϳ92 kDa (␤APP), ϳ32 kDa, and ϳ4 kDa (A␤)) increased following treatment with 1 mM H 2 O 2 (6 or 24 h), UV, or PMA (Fig. 1). PMA has been shown elsewhere to increase ␤APP expression in HeLa and glial cells through the activation of AP-1 (52). Detectable amounts of Alzheimer's proteins in unstressed control lenses (Fig. 1, control) may reflect normal expression of this gene in the monkey lens. By contrast, an example of gene expression that is not increased by oxidative stress is the glyceraldehyde-3-phosphate dehydrogenase gene (58).
Evidence supporting expression of ␤APP in untreated mammalian lenses comes from RT-dependent PCR using RNA isolated immediately after lens removal from 8-week-old FVB/N mice (Fig. 2). The presence of ␤APP mRNA is indicated by the RT-dependent detection of a 473-base pair amplified product using oligonucleotide primers contained within separate exons of the mouse ␤APP gene. These data indicate that ␤APP is present at the RNA and protein levels in monkey and mouse lenses.
Activation of the Ras and Src pathways and increases in AP-1 binding are known to occur during oxidative stress (48,50,51), and AP-1 was shown to increase ␤APP expression (52). In this regard, we have found a significant increase in AP-1 binding to oligonucleotides containing its cognate binding site using standard electrophoretic mobility shift assays after treat- ing monkey lenses or cultured N/N1003A rabbit lens epithelial cells with 100 ng/ml PMA, 1.0 mM H 2 O 2 or UV 254 nm radiation (Fig. 3). The increase in AP-1 binding paralleled the increase in ␤APP which was demonstrated in the same monkey lens extracts in Fig. 1. Increased AP-1 binding was prevented in N/N1003A cells by pretreatment with tyrophostin, an inhibitor of tyrosine phosphorylation (49) (Fig. 3). These results are consistent with the effects of H 2 O 2 and tyrosine phosphorylation inhibitors on c-fos and c-jun expression in cultured rat lenses (58). As c-fos and c-jun expression have been shown to increase with oxidative stress in lens cells (58), our experiments do not necessarily distinguish between increased AP-1 activation and increased AP-1 factor synthesis.
We next localized the expression of ␤APP in adult rat lenses treated in organ culture for 24 h with 0, 125 M, or 250 M H 2 O 2 (Fig. 4A). The immunostaining of ␤APP in H 2 O 2 -treated lenses was predominately in the anterior epithelial layer and in cortical fiber cells. Low amounts of ␤APP were also detected in untreated rat lenses and may reflect ␤APP present in normal lenses and/or base-line oxidative stress incurred during lens removal and culturing. A control SV40 T-antigen antibody showed no staining in rat lenses either with or without H 2 O 2 treatment (data not shown). Staining of a cataractous human lens with a different monoclonal antibody directed at A␤ is shown in Fig. 4B. We detected A␤ in the cortical fiber cells below the epithelial cell surface layer (blue/black color) with this antibody in human lenses. These results differ from H 2 O 2 treated rat lenses, where ␤APP was detected in both the epithelial and cortical regions (Fig. 4A).
To assess the effect of A␤ on cultured lens cells, we cultured N/N1003A cells in the presence of A␤-(1-40) as described elsewhere (38). Extensive inclusions and vacuoles were observed with exposure to 20 g/ml A␤-(1-40) for 24 h (Fig. 5A). In addition, Ͻ50% of the epithelial cells remained attached to the surface of the culture dish after 5 days of culture in the presence of 50 g/ml A␤-(1-40) (Fig. 5B). A␤-(25-35) peptide (15, 38) (20 -50 g/ml) was cytotoxic for N/N1003A cells in culture (data not shown), producing vacuoles and decreasing cell attachment. These cytotoxic effects were produced by A␤ at similar concentrations in N/N1003A cells as for neuronal cells (36). N/N1003A cells cultured in the presence of diluent alone, cytochrome c (20 g/ml), or broad range molecular weight standards (myosin, ␤-galactosidase, phosphorylase b, serum albumin, ovalbumin, carbonic anhydrase, trypsin inhibitor, lysozyme, and aprotinin at 20 g/ml each protein; Bio-Rad) all appeared as in the control (Fig. 5A). Experiments using cells of neuronal origen have indicated that A␤-(25-35) peptide, where the amino acid order has been scrambled (40), and A␤-(1-40) peptide, where the amino acid order has been scrambled had no effect on A␤ inducing its own production in cultured muscle cells (15). DISCUSSION We have presented data showing that oxidative stress can activate AP-1 factors in monkey lenses and rabbit lens epithelial cell cultures and induce ␤APP and A␤ in monkey and rat lenses in organ culture. We have localized the increase in ␤APP in rat lenses to the epithelium and cortex. Moreover, A␤ produces vacuoles and is toxic to cultured rabbit lens epithelial cells. We have also detected A␤ cross-reacting protein in the cortex of cataractous human lenses in their opaque cortical regions. Together, these observations are consistent with the possibility that ␤-amyloids may contribute to the process of cataract formation.
Many cataracts, including age-related opacities (4 -8), the Nakano (59) (62), are associated with the accumulation of insoluble protein aggregates. Since the ability of A␤ to nucleate protein aggregation in amyloid disease is well established (33,34,43), it would seem from the present data that a search for the involvement of ␤-amyloid proteins in nucleation events associated with cataract is warranted. The formation of heavy molecular weight protein fractions, believed to be intermediates in protein insolubilization (see Ref. 7 for review), and phase separation phenomena, which are reported to occur during the early stages of cataractogenesis (63)(64)(65)(66), are processes that might be affected by A␤. The present immunohistochemical data in both rat and human lenses suggest that ␤-amyloids could contribute more to cortical than central nuclear cataracts, as both A␤ and ␤APP were found predominately in the epithelial and cortical regions rather than the central nuclear regions of the lens. It is not known if the potential deleterious effects of ␤APP proteolytic products on lens cell development or homeostasis require identical higher order A␤ structures as have been implicated in AD. ␤APP and their cleavage products are normal constituents of many cell types. However, the production of A␤-containing proteins capable of fibril formation and/or cytotoxic effects appears to be a salient feature of amyloidogenic diseases where A␤ plays a role. The up-regulation of A␤ is most dramatic in trisomy 21 individuals (4 -5-fold), where gene dosage plays a role, however, both A␤ and ␤APP are also increased more than 2-fold in some familial AD involving chromosome 14, suggesting a physiological cell signaling mechanism for A␤ up-regulation as well.
The intracellular vacuoles associated with the cytotoxicity of A␤ in cultured N/N1003A lens epithelial cells in the present study are also consistent with ␤-amyloid proteins contributing to cataract. Intracellular vacuoles appear as a pathological hallmark in diabetes-related sugar cataracts where both osmotic (67) and oxidative (68) stresses are involved. Intracellular vacuoles appear in lens epithelial cells cultured in the presence of low concentrations of glucose and galactose (69) and appear in the central epithelium as the first detectable abnormalities during galactose cataract formation in rats (70). Similar vacuolization occurs in L-buthionine sulfoximine-induced cataract in mice (71). Moreover, recent experiments on cultured rat lenses have indicated that oxidative stress caused by photochemical insult leads to vacuole formation and irreversible damage to the epithelial cells preceding and accompanying opacification (72,73).
Since oxidative stress is considered a major cause of cataract (3,67,73), our finding that H 2 O 2 and UV 254 radiation increase ␤APP and A␤ in cultured intact lenses supports the idea that these ␤-amyloid proteins play a role in cataract formation. While it has been shown that A␤ can induce H 2 O 2 in mediating toxicity, and that anti-oxidants protect cells from A␤ toxicity in primary cultures and clonal cell lines derived from the central nervous system (44), to the best of our knowledge the present study is the first report of oxidative stress increasing Alzheimer's proteins and is consistent with the role of AP-1 in ␤APP gene expression. An induction of H 2 O 2 by A␤ could provide a feedback loop mechanism in lenses allowing A␤ to increase its own expression, as has been observed in some muscle degeneration diseases (15). The ability of oxidative stress to induce A␤ suggests that the manifestation of amyloid diseases could vary in different tissues depending on the amount of exposure to oxidative stress. Moreover, the proliferative-like response to oxidative stress involving such factors as AP-1 suggests that the state of growth and/or differentiation of a given cell type could also be a factor. Since our experiments suggest that both oxidative stress and phorbol ester-mediated activation of AP-1 can increase Alzheimer's disease proteins in lens, one must consider that varied pathways and a multiplicity of signaling routes can potentially increase the ectopic expression of deleterious proteins like A␤.
The mechanism of neurotoxicity attributed to the A␤ peptide involves the generation of reactive oxygen species and destabilization of cellular calcium homeostasis (74). Indeed, recent experiments indicate that different amyloidogenic peptides, such as amylin, and ␤ 2 -microglobulin share this mechanism of affecting neurotoxicity (74). Thus it appears possible that in addition to a multiplicity of signaling pathways, several amyloidogenic peptides exist that potentially could give rise to toxic effects in tissues including the lens.
The mammalian lens is composed of anterior epithelial cells, which begin to elongate at the equatorial margin. These cells withdraw from the cell cycle and produce large amounts of crystallins. Terminal fiber cell differentiation is associated with the degradation of cellular organelles, has some characteristics of apoptosis (75), and involves p53 (76) and retinoblastoma (77-79) proteins. As cells in the lens cortex normally enter growth arrest, leading ultimately to nuclear breakdown and DNA fragmentation in the central region of the lens (75), we infer that an ectopic proliferative-like response to oxidative stress, which includes A␤ up-regulation, may be of greater consequence than a growth arrest DNA damage response (for review see Ref. 80) to the lens. This is consistent with earlier reports showing that x-irradiation-induced cataract in amphibian lenses is greatly inhibited by preventing cell proliferation and/or differentiation through hormonal manipulation (81)(82)(83). Similarly, x-irradiation-induced cataracts do not develop in ground squirrels during hibernation when there is no lens cell proliferation (82,83).
Oxidative stress at the levels employed in our experiments exceed that which would be experienced under normal in vivo conditions where such stress is chronic, producing effects over decades. However, low level oxidation-induced perturbations in cell signaling processes including the Ras/Src-mediated pathway for AP-1 activation (47-51) may help explain the observed link between environmental oxidative stress and cataract and possibly also AD. Studies that have shown the activation of herpes viral proliferation and skin cancers by sunlight (84) support this idea.
In summary, cataract is a disease that can involve protein aggregation and vacuole formation and thus shares similarities with AD and related A␤ pathologies. Oxidative stress is an important etiological factor in these diseases and up-regulates A␤ and ␤APP in lenses and lens cell cultures. In addition, the vacuole formation and cytotoxicity induced in cultured lens cells by A␤ are similar to neuronal cell cytotoxicity demon- ␤-Amyloid and Oxidative Stress in Lens strated by others. Whether A␤ can elicit a trophic response in primary lens cells as it does in neuronal cells remains to be addressed. Indeed, the normal role for these proteins in neuronal cells as well as lens cells is not known. ␤APP and A␤ proteins expressed in mammalian lenses induced by oxidative stress parallel the activation of AP-1 factor binding consistent with a role for stress-induced cell signaling in cataract formation.
Taken together, our data raise the possibility that oxidative stress, a known pathway for cataract formation, stimulates ␤APP formation or aberrant ␤APP protein cleavage in the ocular lens and imposes a strain on protein organization and cell integrity, contributing to lens opacification.