A dedicated metabolic pathway for the synthesis of UV filters from Trp is present in primate lenses.
1 16 The end product is 3OHKG, a metabolite detected only in the lens, although the immediate precursors Kyn and 3OHKyn are found throughout the body. Kyn, 3OHKyn, and 3OHKG are unstable at pH 7 and lose ammonia from the amino acid side chains.
2 Spontaneous deamination of 3OHKG leads to the formation of other major lenticular UV filters, such as AHBG,
17 3OHKG-GSH,
3 and 3OHKG-Cys,
18 that are formed in the human lens from the α,β-unsaturated ketone by reduction and thiol addition, respectively
(Fig. 1) . Because the unsaturated ketones that result from UV filter deamination are reactive, it is perhaps not surprising that, in the very high protein concentration in the lens, nucleophilic groups of proteins may also couple with these intermediates. This binding appears to be exacerbated when the lens barrier forms at middle age, because the increased time that these small molecules spend in the nucleus allows a greater percentage of the unstable molecules to decompose.
6 The barrier also restricts the flow of GSH from the cortex.
4 These two factors combine to cause an increase in modification of nuclear proteins after age 50. It is not simply a matter of the time of reaction of the crystallins with UV filters since, in an adult, before barrier formation, proteins in the lens center may have been in contact with UV filters for more than 40 years, with little or no covalent modification.
Since crystallins in the nucleus are present in the lens before birth and are not replaced, several posttranslational modifications have been reported in proteins extracted from adult lenses
(Table 2) . UV filter modification is one such modification, and it has been suggested that the brown and black coloration that is found only in human age-related nuclear cataract lenses could be the result of oxidation of these primate-specific, protein-bound UV filters.
7 9 25 26 27 If this were true, one would predict lower levels of bound UV filters in cataract lenses compared with normal lenses. These levels will be examined in a separate investigation.
In this study, we examined protein-bound UV filters in normal lenses. For each of the UV filters, the overall pattern was qualitatively similar, in that very little or none of the three UV filters could be detected bound to protein before middle age. As noted earlier, this finding is consistent with the formation of the lens barrier. Before age 40 to 50, free UV filters are present at higher concentrations than in older lenses
14 ; however, based on data for other small molecules such as H
2O,
5 and Cys,
4 they spend on average shorter times in the lens nucleus, based on signal decay for H
2O when placed in a D
2O based medium,
5 giving them less time to deaminate. In addition, nuclear GSH levels are higher in younger lenses,
14 so that α,β-unsaturated ketones that do form, will likely be intercepted before they can bind to proteins. Consistent with this view, young lenses contained either low, or undetectable, levels of the three protein-bound UV filters
(Fig. 3) . By contrast, every lens aged more than 50 years was found to have all three of the UV filters covalently attached to the nuclear proteins as Cys or Lys adducts.
Despite individual variation, after age 50, the concentrations of each of the UV filters were relatively constant in the nucleus
(Fig. 3) . The mean values for normal lenses (>age 50) were: 3OHKG, 1307 picomoles/mg protein; Kyn 37 picomoles/mg protein, and 3OHKyn 9 picomoles/mg protein, corresponding to a ratio of protein-bound 3OHKG-Kyn-3OHKyn of 145:4:1. These ratios parallel the ratios of free UV filters (178:5:1): the mean values of free UV filters were 3OHKG, 534 picomoles/mg tissue; Kyn 16, picomoles/mg tissue; and 3OHKyn, 3 picomoles/mg tissue. The data suggest that the UV filters are approximately equally able to bind to proteins in the lens, and the adducts formed may have similar stabilities. To compare the total free with the total bound levels in whole individual normal lens nuclei, we calculated the overall amount of protein-bound UV filters, based on the protein content. In lens nuclei older than 50 years, there was 9,225 ± 2,100 picomoles (
n = 5) of bound and 11,177 ± 4,280 picomoles (
n = 5) of free 3OHKG, 304 ± 173 picomoles (
n = 5) of bound and 311 ± 83 picomoles (
n = 5) of free Kyn, and 59 ± 28 picomoles (
n = 5) of bound and 50 ± 32 picomoles (
n = 5) of free 3OHKyn. Thus, in the center of older normal human lenses, UV light would be absorbed equally by protein-bound and free UV filters. This is in marked contrast to the young lens.
The amounts attached to the proteins from the lens cortex were always markedly lower than those in the nucleus. The mean cortical values for lenses aged more than 50 years were 3OHKG, 56 picomoles/mg protein, and Kyn, 2 picomoles/mg protein, corresponding to levels that are 23- and 19-fold less, respectively, than those in the nucleus. 3OHKyn attached to the cortical lens proteins was below the limits of detection.
What are the chief conclusions to be drawn from these findings? First, every normal lens examined >50 years was found to have all three UV filters bound to the nuclear proteins. Second, from a comparison of bound and free UV filters in the center of individual older lenses, it is clear that significant absorption of UVA light is mediated via modified proteins. This effect will likely be exacerbated by other posttranslational modifications that are age-related
(Table 2) . Since binding of Kyn to calf lens proteins causes them to become susceptible to photooxidation by the wavelengths of light that penetrate the cornea, and this is accompanied by the generation of reactive oxygen species,
28 these UV filter modifications may make the nuclear lens proteins prone to photooxidation after middle age. 3OHKyn is also extremely reactive under oxidative conditions and the presence of such a UV filter bound to the nuclear proteins of older normal lenses may predispose them to oxidation and therefore the sorts of changes, such as cross-linking, insolubilization and oxidation that are characteristic of age-related nuclear cataract.
It should be noted that the assay system that we used allows quantification of UV filter adducts of Lys and Cys, but not of His, which are significantly more stable.
9 According to the same assay, when the Lys, Cys, and His adducts of 3OHKyn were incubated with GSH, the yield of the GSH conjugates of Lys, Cys, and His were 73%, 61%, and 0.5%, respectively.
9 Therefore, the amounts of bound UV filters quoted herein are likely to be an underestimate. Model studies indicate however that Cys adducts are the most prevalent when proteins are incubated with UV filters. Even allowing for this underestimation, the total levels of UV filters found in this study (∼1350 picomoles/mg protein) are comparable with those of other major posttranslational modifications in the human lens
(Table 2) .
In summary, proteins isolated from all normal lenses older than 50 years contained measurable levels of each of the three kynurenine UV filters. 3OHKG was always present in the largest amounts, with lower levels of Kyn and 3OHKyn, which reflects the concentrations of the free UV filter compounds in the lens. Despite this finding, there was no clear relationship in individual lenses between free and bound UV filter concentrations, which implies that the free UV filter concentration, although giving rise to a population of reactive intermediates, does not directly correspond to the level of protein-bound UV filters. This result is consistent with the fact that mechanisms such as reduction of reactive intermediates by NADH and GSH, as well as the permeability of the barrier to small molecule diffusion, are key variables in the transport and reactivity of UV filters. The attachment of UV filters to older lenses may predispose them to oxidative damage.