The Hydroxyl Radical in Lens Nuclear Cataractogenesis*

Cataract is the major cause of blindness; the most common form is age-related, or senile, cataract. The reasons for the development of cataract are unknown. Here we demonstrate that nuclear cataract is associated with the extensive hydroxylation of protein-bound amino acid residues, which increases with the development of cataract by up to 15-fold in the case of DOPA. The relative abundance of the oxidized amino acids in lens protein (assessed per parent amino acid) is DOPA > o- andm-tyrosine > 3-hydroxyvaline, 5-hydroxyleucine > dityrosine. Nigrescent cataracts, in which the normally transparent lens becomes black and opaque, contain the highest level of hydroxylated amino acids yet observed in a biological tissue: for example, per 1000 parent amino acid residues, DOPA, 15; 3-hydroxyvaline, 0.3; compared with dityrosine, 0.05. The products include representatives of the hydroperoxide and DOPA pathways of protein oxidation, which can give rise to secondary reactive species, radical and otherwise. The observed relative abundance corresponds closely with that of products of hydroxyl radical or metal-dependent oxidation of isolated proteins, and not with the patterns resulting from hypochlorite or tyrosyl-radical oxidation. Although very little light in the 300–400-nm range passes the cornea and the filter compounds of the eye, we nevertheless also demonstrate that photoxidation of lens proteins with light of 310 nm, the part of the spectrum in which protein aromatic residues have residual absorbance, does not give rise to the hydroxylated aliphatic amino acids. Thus the post-translational modification of crystallins by hydroxyl radicals/Fenton systems seems to dominate their in vivo oxidation, and it could explain the known features of such nuclear cataractogenesis.

Cataract is the major cause of blindness; the most common form is age-related, or senile, cataract. The reasons for the development of cataract are unknown. Here we demonstrate that nuclear cataract is associated with the extensive hydroxylation of protein-bound amino acid residues, which increases with the development of cataract by up to 15-fold in the case of DOPA. The relative abundance of the oxidized amino acids in lens protein (assessed per parent amino acid) is DOPA > o-and m-tyrosine > 3-hydroxyvaline, 5-hydroxyleucine > dityrosine. Nigrescent cataracts, in which the normally transparent lens becomes black and opaque, contain the highest level of hydroxylated amino acids yet observed in a biological tissue: for example, per 1000 parent amino acid residues, DOPA, 15; 3-hydroxyvaline, 0.3; compared with dityrosine, 0.05. The products include representatives of the hydroperoxide and DOPA pathways of protein oxidation, which can give rise to secondary reactive species, radical and otherwise. The observed relative abundance corresponds closely with that of products of hydroxyl radical or metal-dependent oxidation of isolated proteins, and not with the patterns resulting from hypochlorite or tyrosyl-radical oxidation. Although very little light in the 300 -400-nm range passes the cornea and the filter compounds of the eye, we nevertheless also demonstrate that photoxidation of lens proteins with light of 310 nm, the part of the spectrum in which protein aromatic residues have residual absorbance, does not give rise to the hydroxylated aliphatic amino acids. Thus the post-translational modification of crystallins by hydroxyl radicals/Fenton systems seems to dominate their in vivo oxidation, and it could explain the known features of such nuclear cataractogenesis.
Cataract is a major problem worldwide. At present the etiology of the cataract is poorly understood and the only treatment involves surgical removal of the opaque lens. In the developing world, access to ophthalmic facilities is a limiting factor; it is estimated that there are 4 million newly blind each year in India alone (1). The cost of surgical procedures is now a considerable proportion of the health budget of most developed nations (2).
There are two main types of age-related cataract, nuclear and cortical. In cortical cataract the opacity, localized in the outer region of the lens, is associated with an ionic imbalance (3). Nuclear cataract is characterized by a high degree of light scattering in the center of the lens. In the vast majority of cases this opacity is colored (4). In most cases the lens in such patients is yellow or brown, but in rare cases the normally transparent and colorless lens can be transformed into a black opaque "nigrescent" cataract.
Much of what is known about the biochemical changes associated with age-related cataract has come from studies in which cataractous lenses have been classified as types I-IV on the basis of an increase in nuclear color as first proposed by Pirie (5). Type I cataract is characterized mainly by cortical changes without significant nuclear alterations. It has been shown that the increase in nuclear color is closely linked with progressive increases in the amount of protein methionine sulfoxide, urea-insoluble protein, and crystallin cross-linking (6 -10). One of the earliest changes is the loss of protein sulfhydryl groups, and these decrease progressively with the development of cataract such that in type IV nuclear cataract less than 5% of the original sulfhydryl groups remain (8).
On the basis of these observations it was postulated that the reaction of the lens proteins with H 2 O 2 derived from superoxide may be responsible for the changes associated with nuclear cataract formation (7). Peroxide is known to oxidize cysteine and methionine efficiently (11,12). H 2 O 2 may be derived from the aqueous humor which bathes the anterior segment of the lens (13) and is the primary source of nutrients and oxygen, although the lens is remarkable in its ability to detoxify external peroxide (14,15). Since the human lens grows throughout life by the addition of fiber cells, and since the crystallins in the lens nucleus undergo little turnover (16), the proteins in the central region of the lens are as old as the individual. Thus post-translational modifications to these crystallins may accumulate over a person's lifetime.
It is also possible that protein modifications linked with cataract could be the result of a reaction of lens crystallins with other oxidizing agents such as the hydroxyl radical (17)(18)(19)(20)(21), which might also partly derive from hydrogen peroxide through the transition-metal ion catalyzed Fenton reactions (22). With the recent development of HPLC 1 procedures which enable hydroxyl radical damage to proteins to be detected and quantified without any drastic treatment (e.g. derivatization or volatilization for mass spectrometry) after protein hydrolysis (23)(24)(25)(26)(27), it has become feasible to use these techniques to examine proteins from biological sources for evidence of modification by this oxidant. Furthermore, we have demonstrated that certain protein oxidation pathways give rise to reactive intermediates, such as hydroperoxides and the reducing moiety DOPA (28,29), which can generate secondary reactive species and damage other biological targets such as DNA (30), and these methods permit the assessment of the occurrence of these pathways.
In this study we analyzed the nuclei and cortices of individual cataractous lenses to assess the involvement of the hy-droxyl radical in the etiology of the nuclear cataract. We show that the development of human nuclear cataracts is associated with a pronounced increase in the levels of modified amino acids in the lens proteins, including representatives of the hydroperoxide and DOPA pathways. The observed results are consistent with hydroxyl radical damage, and not with photoxidation. We therefore propose that the hydroxyl radical may play a role in the development of the nuclear cataract in man.

MATERIALS AND METHODS
Reagents-o-Phthaldialdehyde (OPA) crystals and OPA diluent (containing 3% potassium hydroxide and 3% boric acid at pH 10.4) were from Pickering Laboratories (Mountain View, CA). 2-Mercaptoethanol was from Sigma. Mercaptoacetic acid was from Merck (VIC, Australia). Other chemicals, solvents, and chromatographic materials were of analytical reagent or HPLC grade.
Lens Dissection and Extraction-Cataractous lenses were classified on the basis of nuclear color as described by Pirie (5). Nigrescent lenses were treated as a subcategory of type IV (nigrescent type IV). Lenses were stored at Ϫ20°C and thawed immediately prior to dissection. Some nigrescent lenses, removed from patients attending eye camps in India, were shipped in alcohol and then stored at Ϫ20°C. Analyses we performed on type I and IV cataractous lenses showed that storage in alcohol does not affect the content of the hydroxylated amino acids.
Lenses were dissected using a 4-mm cork borer to obtain the central core, and the ends of the core were then removed. The average wet weight of the nucleus thus obtained was 20.3 mg. Cortices were ob-  (5) and Leu.OH2 (6) by HPLC of the prepurified (on liquid chromatography-NH 2 column) and OPA derivatized fraction (E x ϭ 340 nm; E m ϭ 440 nm). The addition was with a ␥-radiolyzed mixture of valine and leucine which contains Val.OH1 and Leu.OH2 among other oxidation products (24,27). C, electrospray mass spectrometry confirmation of identities of DOPA and o-Tyr in proteins of both cataract and normal subjects. Lens protein samples were obtained by extraction of pooled normal as well as type III lenses as described under "Materials and Methods." C i, selective ion (at m/z 198) chromatogram for DOPA fractions, with a peak area of 4273020 for a cataract type III protein sample and 115760 for a normal sample; C ii, selective ion (at m/z 182) chromatogram for o-Tyr fractions, with a peak area of 4452970 for cataract sample and 1438126 for normal sample; C iii and C iv are the corresponding mass spectra transformed from C i and C ii, respectively. The molar ratios deduced from the mass spectral data are in good agreement with those determined by HPLC of the same pooled lens protein samples: the DOPA/Tyr (mmol/mol) HPLC ratios are 0.74 (normal) and 3.88 (cataract), while the o-Tyr/ Phe (mmol/mol) HPLC ratios are 0.18 (normal) and 0.65 (cataract). These ratios are congruent with the other more detailed data presented in this report.
tained by taking a sample of the outer lens ring remaining after coring and removing the lens capsule. No reagents were added to the cortices, and no dialysis step was necessary. Lens samples were frozen individually in liquid nitrogen and powdered in an Eppendorf tube by using a Teflon hand homogenizer. After freeze-drying, the samples were hydrolyzed individually for analysis of oxidized amino acids, as follows.
The issue of artifactual formation of oxidized amino acids during tissue homogenization has been studied extensively with human plaque tissue (31). No formation of oxidized amino acids was observed when standards of tyrosine, phenylalanine, valine, and leucine were added to the sample prior to mechanical homogenization.
Amino Acid Hydrolysis-A gas phase and acid-catalyzed method described previously (23) was adopted, in which mercaptoacetic acid (5% v/v) and phenol (1% w/v) were added to the HCl (6 M) solution as reductant and antioxidant, respectively, to permit the optimal recovery of DOPA (95-100% as judged by the recovery of free DOPA). We also achieved 95-100% recovery of free o-and m-tyrosine and dityrosine. The recovery of 3-hydroxyvaline (Val.OH1) was c.80% and of 5-hydroxyleucine (Leu.OH2) c.70%, as detailed previously (24,26,27). The known quantity (approximately 4 mg) of dried lens powder was used for each hydrolysis. The hydrolysate was lyophilized and dissolved in about 200 l water (such that the final sample concentration was 20 mg/ml hydrolysate) for HPLC analysis.
Artifactual formation of m-and o-tyrosine (from free phenylalanine), and DOPA (from free tyrosine) during protein hydrolysis, in the presence or absence of protein, has also been investigated in detail in the study relating to human atherosclerotic plaques (31). The maximum artifactual formation ranges from 10 mol of m-tyrosine/mol of phenylalanine, 20 mol of o-tyrosine/mol of phenylalanine, to 100 mol of DOPA/mol of tyrosine, which are low values compared with those reported in this study for either the normal or cataractous lenses. Artifactual formation of hydroxyvaline and hydroxyleucine was not observed during such protein hydrolysis. In comparable experiments in which these four amino acids were added to lens samples prior to freezing and powdering, and then taken through the hydrolysis procedure, indistinguishable results were obtained, confirming that artifactual oxidation was very limited.
Analysis of Modified Amino Acids-All HPLC analyses were carried out on a LC-10A HPLC system (Shimadzu) consisting of two pumps, a high pressure mixer, an autosampler, and a sample cooler. System operation and peak integration was driven by Class LC-10 software (Shimadzu) operated under a PC-based Windows environment. A column oven (Waters) set at 30°C was also used for each analysis.
Detection of Val.OH1 and Leu.OH2 requires two steps of HPLC. In the first step, the hydrolysate was fractionated on a LC-NH 2 column (25 cm ϫ 4.6 mm, 5-m particle size, Supelco). The purified fraction was then analyzed by a second HPLC step on a Zorbax ODS column (25 cm ϫ 4.6 mm, 5-m particle size, Rockland Technologies) following pre-column derivatization with OPA reagent. The detailed experimental conditions have been described previously (24,27). This second HPLC procedure was also used to measure unmodified valine, leucine, and phenylalanine in the hydrolysate (without any prior fractionation).
For detection of DOPA, m-tyrosine, o-tyrosine, and dityrosine, the hydrolysate was separated on a Zorbax ODS column with a Pelliguard column (2 cm, Supelco). Elution at 1 ml/min was performed with a gradient of solvent A (100 mM sodium perchlorate in 10 mM sodium phosphate buffer, pH 2.5) and solvent B (80% methanol in water) as described previously (32). The eluent was monitored by both UV (Shimadzu, at 280 FIG. 2. Molar ratios of oxidized to parent amino acids in normal and cataractous human lens proteins. After separation of cortices (u) from nuclei (f), lens samples were hydrolysed and analyzed for oxidized and native amino acids using HPLC (see "Materials and Methods"). The data are means with standard error (S.E.), where n ϭ 4 (nigrescent (Nig) type IV); 5 (normal); 6 (for type II and for type III); and 7 (for type I and for type IV). There was a statistically significant type dependence by analysis of variance for every oxidized amino acid (p Ͻ 0.001 in every case). Unpaired Student's t tests were also performed between groups; for example, type nm) and fluorescence (Hitachi F-1080) detectors in series. The excitation wavelength was set at 280 nm for all components, while the emission wavelength was set at 320 nm for DOPA, m-tyrosine, o-tyrosine, and 410 nm for dityrosine through the built-in time program of the detector. Quantitation of (unmodified) p-tyrosine was based on UV detection due to its off-scale response in the fluorescence channel.
Confirmation of the Identity of Modified Amino Acids by Mass Spectroscopy-Pooled lens protein samples (10 mg) were hydrolyzed and then redissolved in water (500 l). The samples were then fractionated using the standard HPLC methodology described above. DOPA and o-Tyr fractions were collected and combined from 25 injections of the hydrolysate (20 l for each injection). The collected samples were freeze-dried and rechromatographed on the same HPLC column with 0.1% trifluoroacetic acid in water, replacing the standard mobile phase solvent A (which contains 100 mM sodium perchlorate and 10 mM sodium phosphate). The desalted fractions were dissolved in 80 l of 1% aqueous formic acid and acetonitrile (1:1 v/v) and subjected to mass spectroscopic analysis. The electrospray mass spectrometry was conducted on a VG Platform mass spectrometer (Fisons, Homebush, NSW, Australia). The solvent (50% acetonitrile in water) was delivered by a Phoenix (Fisons) syringe pump at a flow rate of 100 l/min; 10 l of each sample solution were injected for analysis. Dry nitrogen gas at atmospheric pressure was employed to assist evaporation of the electrospray droplets. A positive electrospray mode was used with electrospray probe tip potential at 2.19 kV, counter-electrode potential at 0.5 kV, and cone potential at 20 V.

RESULTS
Free Amino Acids Are Negligible in Human Lenses-Two human lenses (normal and nigrescent type IV) were used for determination of the proportion of free native and oxidized amino acids in lens compared with their population in lens proteins. Lens samples were powdered with the aid of liquid nitrogen and solubilized in water by sonicating. Proteins were precipitated out by means of trichloroacetic acid. Both the protein portion after amino acid hydrolysis and the supernatant portion were analyzed by HPLC for the presence of tyrosine, phenylalanine, valine, and leucine. As summarized in Table I, the free amino acids comprised less than 1% of the total amino acids in proteins in lens, in agreement with published literature (7)(8)(9). There were no detectable oxidized amino acids in the supernatant portion of the samples, confirming that any oxidized amino acids found in the lens are protein-bound.
HPLC Determination of Oxidized Amino Acids in Lens Proteins-The HPLC methods employed in this study have been used successfully in several other studies investigating protein oxidation by free radicals in biological and pathological samples, for example, native human plasma, 2 cerebral malariaaffected mice brain homogenates, 3 and rat tail collagen incubated with glucose (32). Fig. 1 shows typical HPLC traces of hydrolysates of normal as well as cataractous lens proteins. Oxidized amino acids were identified on the basis of their coelution with spiked standards. The identification of the hydroxy-aliphatic amino acids is very strong, since they undergo two successive HPLC chromatography steps, in each of which they behave characteristically. Indeed, in Fig. 1B, Leu.OH2 (peak 6) is essentially base-line resolved; similarly, Val.OH1 is well resolved from a smaller preceding peak, although the spiking material used in Fig. 1B, ii, contains very large proportions of such an adjacent molecule, which gives a slightly misleading impression. Again the identification of the individual oxidized species is very strong, with dityrosine distinguished not only by its elution position, but also by its characteristic fluorescence spectrum. o-Tyrosine (peak 3) is closely followed by a smaller peak, not present in the spiking material (Fig. 1A, ii versus i). All these individual chromatographic peaks could be readily integrated and well resolved, giving precise measurements of the oxidized species, which were quite reproducible (see below). Nevertheless, we used mass spectrometry of some such peaks isolated from pooled normal or pooled diseased lens proteins, to confirm their identity. Fig. 1C shows for the cases of DOPA and o-tyrosine that the relative abundance of the appropriate molecular ion in the two samples was similar to that observed by HPLC in the same samples, and in the later data for individual lenses. Mass spectrometry thus confirmed the identity of the species under study, although it cannot readily distinguish o-, m-, or p-tyrosine from each other.
To obtain information on interassay variation for each oxidized amino acids, repeat analyses of the same lens sample were performed on different days. As summarized in Table II, the interassay variation for oxidized amino acids was generally larger (around 10%) than that for native amino acids (around 5%), which might be due to either the modest artifactual formation of oxidized moieties during hydrolysis (in the case of DOPA, dityrosine, o-tyrosine, and m-tyrosine) or to the need for two steps of HPLC with the extra inherent variability this entails (in the case of hydroxyvaline and hydroxyleucine).
Determination of the Levels of Oxidized Amino Acids in Lens Proteins-The levels of all six amino acid derivatives have been quantified in the separated nuclear and cortical proteins of normal and cataractous human lenses (Fig. 2). Statistical analysis by analysis of variance indicated no significant variation between nuclear and cortical material in any of the lens samples. But there was a marked dependence on cataract type for every oxidized amino acid (p Ͻ 0.001 in every case). When unpaired t tests were undertaken, nigrescent type IV and type III values were significantly different from normal lens values for each oxidized amino acid (p Ͻ 0.05 in every case). Statistical differences at p Ͻ 0.05 were observed in most cases for type II and in one case for type I versus normal lens cortices and

acids as percentages of total amino acids in human lenses
Dried and powdered human lens samples (4 mg, one normal and one nigrescent type IV) were dissolved in water (200 l) by sonicating at 4°C for 5 min. Into the solutions were added 0.3% sodium desoxycholate (25 l) and 50% (v/v) trichloroacetic acid (50 l). After mixing, the mixture was centrifuged at 16,000 ϫ g for 2 min. After separation from the supernatant the protein pellet was hydrolyzed with the conditions described under "Materials and Methods." HPLC analyses for oxidized as well as native amino acids were carried out on both the supernatant and protein hydrolysate fractions. Free amino acids are expressed as percentages of total amino acids (free plus protein-bound) in the lenses and are means of duplicate measurements.  nuclei. The contents of the modified amino acids present in the normal lens are similar to those found in apoprotein B of freshly prepared (33,34) normal human plasma low density lipoprotein (Table III). Since type I lenses are characterized by cortical cataract, this finding indicates that the development of such cortical opacity is not associated with a general increase in the levels of such modified amino acids, suggesting that the hydroxyl radical is not involved in the etiology of this form of cataract. The earliest stage of nuclear cataract is the type II lens, characterized by an increase in the protein fraction which is insoluble in 8 M urea and a substantial loss in protein thiol content, on average to 30% of those seen in normal lenses (7)(8)(9). Interestingly, for DOPA, the hydroxylated amino acid produced at the highest proportional levels (modified/parent amino acid), levels in type II lenses were statistically elevated over those in the normal lenses (p Ͻ 0.004 for the groups, including both cortices and nuclei). Similarly, there were statistically significant elevations for each of the other oxidized amino acids with varying significance, but in each case, with p Ͻ 0.05. This suggests hydroxyl radical involvement even in the earliest stage of nuclear cataract.
In types III and IV and the nigrescent lenses, there were also highly elevated levels of dityrosine and the hydroxylated derivatives of Leu, Val, Tyr, and Phe. In each case, as the color of the lens increased, so did the content of the hydroxylated amino acid derivatives. In the HPLC procedure used for these analyses of aromatic derivatives, it is also possible to measure by UV detection both 3-nitrotyrosine and 3-chlorotyrosine, which are markers of modification by nitric oxide and other reactive nitrogen species, and by hypochlorite, respectively (21,29,35,36). None of the samples showed UV-detectable 3-nitrotyrosine and 3-chlorotyrosine levels (detection limits, 0.4 mmol/mol tyrosine in both cases). Thus the modifications seen in the later stages of nuclear cataract could be reasonably explained as hydroxyl radical damage to the lens crystallins, and a major role of reactive nitrogen intermediates or hypochlorite seems unlikely.
Lack of Generation of Hydroxy Aliphatic Amino Acids by 310-nm UV Photooxidation-There are strong epidemiological and physical grounds for doubting that lens protein oxidation is directly due to UV light exposure, notably because very little UV passes into the lens (see "Discussion"). Nevertheless, it was important to establish whether such UV as does reach the lens might produce the spectrum of oxidized amino acids we have observed. Virtually all light below 300 nm is absorbed by the cornea and the filter compounds of the eye (for example, see Sliney (37)). Thus we have determined the action of 310-nm light, using a well defined source. Fig. 3 shows that while DOPA and much smaller amounts of dityrosine can be generated on lens proteins in such conditions, there is no formation of either hydroxy aliphatic amino acid, valine or leucine. There is also no detectable hydroxylation of phenylalanine under these conditions, consistent with the fact that its UV extinction coefficient is much lower than that of tyrosine. Our results on aromatic residue hydroxylation are concordant with previous published data, although it is known that during exposure to lower wavelength UV phenylalanine hydroxylation can occur (38,39). Thus the data of Fig. 3 make it very unlikely that the formation of hydroxyvaline and hydroxyleucine in the cataractous lens is due to photoxidation, but rather suggest an important role for hydroxyl radicals, or a closely related species.
In other studies (24 -27, 40) we have demonstrated that the formation of the six oxidized amino acids during radiolytic hydroxyl radical exposure follows the same relative abundance as described for the lens proteins. In Fig. 3 we also extend this to metal/oxo complexes which give rise to localized Fenton chemistry, producing oxidants of very similar reactivity, probably mainly the hydroxyl radical itself (see, for example, Refs. 21 and 29). It is striking that all six oxidized species are formed, although formation of dityrosine is minimal. The relative abundance of the products is again DOPA Ͼ o-and m-tyrosine Ͼ leucine and valine hydroxides Ͼ dityrosine. In agreement with our data, it has previously been demonstrated that a Fenton-like metal catalyzed oxidation can produce substantial amounts of o-tyrosine and modest amounts of dityrosine in certain proteins (41). Finally, it is noteworthy that the relative levels of the six modified amino acids we have determined are also consistent with those we found in plasma, in apoprotein B of freshly isolated human plasma low density lipoprotein (Table III) (31), and three of these species have previously been determined in human plasma low density lipoproteins with results similar to ours (36). Thus extensive hydroxyl radical damage to the structural proteins may occur in the development of the colored cataracts. DISCUSSION These results demonstrate that damage explicable by hydroxyl radical attack on lens crystallins is associated with the development of human nuclear cataract and that the products include the reactive members of the hydroperoxide and DOPA pathways (28,29). Since many of the changes known to take place in cataract such as protein insolubilization and cross-linking can also be observed in proteins exposed to a hydroxyl radical flux (17,21,42,43), we propose that the nuclear cataract may be a condition that results in part from hydroxyl radical damage.
Prior to the onset of hydroxyl radical attack, the losses of protein sulfhydryl groups and some of the accumulation of crystallins insoluble in 8 M urea in type II lenses (7-9) may result from exposure of the crystallins to H 2 O 2 and its nucleophilic (nonradical) reaction with thiols. This would be consistent with model studies in which lens crystallins have been exposed to H 2 O 2 for extended periods (11). H 2 O 2 may be generated from biochemical reactions in the cortex, by autoxidation of sugars (44,45) or other biomolecules, or from photochemical reactions, and diffuse into the lens interior where the concentration of reduced glutathione is low.
However, evidence for hydroxyl radical modification was observed even in the proteins of type II lenses and was very pronounced in the more advanced nuclear cataract lenses. It would seem likely that this reflects the availability of transition metals (46) (reviewed in Obara (47)), which may originally have been bound to the nuclear crystallins, but become available for redox reactions, and thus hydroxyl radical formation, in the presence of H 2 O 2. Metal ions may be released or rendered accessible by the unfolding of the lens proteins following extensive sulfhydryl oxidation (46). Some oxidation of protein thiols is noticeable in the aging lens (48,49) but this process occurs to a much greater extent with the onset of nuclear cataract. At surgery, most age-related cataracts are found to be mixed (50), i.e. to involve both nucleus and cortex. The novel finding Apoprotein B of freshly isolated human plasma low density lipoprotein was prepared as previously described (33,34). 0.7 ml of low density lipoprotein (2 mg of apoB/ml) was delipidated using trichloroacetic acid/sodium desoxycholate method (26)  13 mol/mol Leu from the present studies, that nuclear cataract lenses show marked and comparable hydroxyl radical damage both to cortical and nuclear proteins, suggests that a hydroxyl radical flux could also be important in cortical cataracts in these patients. It has sometimes been assumed that (UV) photooxidation might be directly responsible for protein oxidation, and its contribution to cataract, in the lens. The epidemiological evidence however, is far from conclusive (reviewed in Harding (16)). Most incident UV light below 300 nm is absorbed by the cornea (37), and much that passes into the lens is then removed by the UV-filtering compounds of the lens, notably 3-hydroxykynurenine glucoside, which have the desirable property of being poor photosensitizers (16,51,52). If, despite this, direct photooxidation were responsible for protein oxidation in cataractogenesis, it would be expected that there would be a decreasing gradient of concentration of oxidized protein moieties from cortex to nucleus, paralleling the decreased incident UV light. This was not observed; rather nuclear amino acid modifications were as extensive and graded as those in the cortex. To clarify these issues further we presented data here showing that light of the lowest wavelengths which reach the lens (ϳ310 nm) does not generate detectable quantities of hydroxyleucine and hydroxyvaline in lens proteins in vitro, even under conditions in which formation of DOPA and dityrosine is detectable (consistent with previous reports of photooxidation of aromatic structures) (38,39). It is thus quite unlikely that direct photooxidation is the major mechanism of formation of the oxidized protein moieties we have detected; although photooxidation may contribute to the supply of hydrogen peroxide (for instance, via singlet oxygen formation), which leads to hydroxyl radicals and oxidants of similar actions through Fenton chemistry. In accord with the likely role of hydroxyl radicals and metal-oxidation systems akin to that of Fenton, we demonstrated that Fe(II)-EDTA complexes oxidized the lens proteins to form all six species under study, with the relative abundances corresponding to those we observe in the lens. The extremely limited formation of dityrosine under these conditions is consistent with much previous data (41) and is to be expected in view of the statistically unlikely requirement for the simultaneous proximate presence of pairs of tyrosyl radicals to form this moiety (53).
Some of the oxidized amino acids we have determined have been demonstrated previously in human lenses. Dityrosine was one of the first (54), although it was subsequently shown (55) that it may have originated from the lens capsule, which contains collagen, rather than from the lens itself. However, dityrosine has also been observed at very low levels in normal human lenses, increasing very slightly with age (56), and the FIG. 3. Oxidation of calf lens protein by UV irradiation and a Fenton system. Calf lens protein was prepared by extraction of calf lens in water, and subsequent lyophilization (8,9). The protein powder was redissolved in water at a concentration of 2 mg/ml. UV irradiation was done using a Luminescence Spectrometer LS 50B (Perkin-Elmer) with the excitation wavelength set at 310 nm and a slit width at 15 nm. The protein samples (2.5 ml in a fluorescence cuvette) were irradiated in the light path of the instrument for 16  reported levels are similar to those we observe in normal lenses. The present study demonstrates that levels of dityrosine increase markedly in the more advanced nuclear cataracts; thus it is a candidate component of the non-disulfide covalent cross-linking characteristic of the nuclear cataract. Modest levels of o-tyrosine were also observed in normal human lenses (56), at levels invariant with age, very similar to those observed in our normal samples, and much higher than for dityrosine, in agreement with the present data.
Nine of the lens samples we have studied were from individuals of known age (from 2 to 95 years, including both diseased and normal samples). Thus we could assess whether this cohort showed correlations between oxidized amino acids and age. In each case linear regression analysis indicated positive correlations, with modest r values (from 0.46 for dityrosine, to 0.65 for Val.OH1), and with gradients from 0.86 per year (dityrosine) to 6.23 per year (DOPA). The oldest normal sample we obtained was from a 46-year-old person. Thus clearly our more aged cohort comprised only diseased samples, and was not ideal for apportioning the influence of age and cataractogenesis. However, as mentioned above, more substantial studies of the oxidation of normal lens proteins in relation to age, measuring only o-tyrosine and dityrosine (56), demonstrated respectively, no dependence on age, and a slight dependence (r ϭ 0.5) with a gradient less than 0.006 per year. Even this latter value is more than 100-fold lower than our value, indicating that it is only in diseased lenses that the substantial enhancement of o-tyrosine and dityrosine occurs, consistent with the idea that hydroxyl radical damage is associated with cataractogenesis per se.
Extraction studies show that the color of the cataract lens is associated with the structural proteins, and part of this could reflect the modest "browning" of hydroxyl radical-damaged proteins, detectable by spectrophotometry (43,43,57,58) and in agreement with the formation of protein-bound forms of DOPA (23), a common component of polymeric pigments after further oxidation. In nigrescent lenses, characterized by a particularly high level of amino acid hydroxylation, it may be that such further oxidation of the DOPA is responsible for the unusual coloration. Other components may also contribute to the strong coloration, for example the reactive UV filter compound, 3-hydroxykynurenine, which is synthesized in the human lens from tryptophan (51,52). It has been shown that this readily binds to and colors lens proteins in the presence of oxygen (59). Other compounds present in the lens, for example glucose (60, 61) and ascorbate (62,63), may also be implicated in color formation as indicated by the elevated levels of pentosidine (64).
We conclude that drastic hydroxyl radical attack on lens proteins occurs in nuclear cataractogenesis. It is possible that the damage observed is a secondary consequence of cataractogenesis. On the other hand, the kinds of damage that hydroxyl radicals are known to inflict on proteins, including cross-linking, could explain many the features of the disease, such as the protein aggregation, and the lens coloration and opacification. In this case, selective inhibition of hydroxyl modification of proteins might offer a promising approach toward control of such cataract.