Non-oxidative modification of lens crystallins by kynurenine: a novel post-translational protein modification with possible relevance to ageing and cataract

Non-oxidative modi¢cation of lens crystallins by kynurenine:a novel post-translational protein modi¢cation with possible relevance to

ageing and cataract

Brett Garner a;b;1;*, Denis C. Shaw c, Robyn A. Lindner b, John A. Carver b,Roger J.W. Truscott a;b

a Australian Cataract Research Foundation, University of Wollongong, Wollongong, NSW 2522, Australiab Department of Chemistry, University of Wollongong, Wollongong, NSW 2522, Australia

c Division of Biochemistry and Molecular Biology, John Curtin School of Medical Research, Australian National University,Canberra, ACT 2601, Australia

Received 2 July 1999; received in revised form 26 October 1999; accepted 28 October 1999

Abstract

In humans, the crystallin proteins of the ocular lens become yellow-coloured and fluorescent with ageing. With thedevelopment of senile nuclear cataract, the crystallins become brown and additional fluorophores are formed. Themechanism underlying crystallin colouration is not known but may involve interaction with kynurenine-derived UV filtercompounds. We have recently identified a sulphur-linked glutathionyl-3-hydroxykynurenine glucoside adduct in the lens andspeculated that kynurenine may also form adducts with GSH and possibly with nucleophilic amino acids of the crystallins(e.g. Cys). Here we show that kynurenine modifies calf lens crystallins non-oxidatively to yield coloured (365 nm absorbing),fluorescent (Ex 380 nm/Em 450^490 nm) protein adducts. Carboxymethylation and succinylation of crystallins inhibitedkynurenine-mediated modification by approx. 90%, suggesting that Cys, Lys and possibly His residues may be involved. Thiswas confirmed by showing that kynurenine formed adducts with GSH as well as with poly-His and poly-Lys. NMR studiesrevealed that the novel poly-Lys-kynurenine covalent linkage was via the O-amino group of the Lys side chain and the LC ofthe kynurenine side chain. Analysis of tryptic peptides of kynurenine-modified crystallins revealed that all of the colouredpeptides contained either His, Cys or an internal Lys residue. We propose a novel mechanism of kynurenine-mediatedcrystallin modification which does not require UV light or oxidative conditions as catalysts. Rather, we suggest that the sidechain of kynurenine-derived lens UV filters becomes deaminated to yield an K,L-unsaturated carbonyl which is highlysusceptible to attack by nucleophilic amino acid residues of the crystallins. The inability of the lens fibre cells to metabolise

0167-4838 / 00 / $ ^ see front matter ß 2000 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 7 - 4 8 3 8 ( 9 9 ) 0 0 2 3 4 - 4

Abbreviations: AHBG, 4-(2-amino-3-hydroxyphenyl)-4-oxobutanoic acid O-L-D-glucoside; CLP, calf lens protein; CM, carboxymeth-ylated; CMS, carboxymethylated and succinylated; COSY, correlation spectroscopy; DTNB, dithiobis-2-nitrobenzoic acid; GSH, glu-tathione; GSH-Kyn, glutathione-kynurenine adduct; GSH-3-OHKG, glutathione-3-hydroxykynurenine-O-L-D-glucoside adduct; HPLC,high performance liquid chromatography; HPLC-ESIMS, HPLC with in-line electrospray ionisation mass spectrometry; Kyn, kynur-enine; NOESY, nuclear Overhauser e¡ect spectroscopy; 3-OHKyn, 3-hydroxykynurenine; 3-OHKG, 3-hydroxykynurenine-O-L-D-gluco-side; PMSF, phenylmethylsulphonyl £uoride; PVDF, polyvinylidene di£uoride; TFA, tri£uoroacetic acid; TNBS, 2,4,6-trinitrobenzene-sulphonic acid; TOCSY, total correlation spectroscopy; SDS, sodium dodecyl sulphate

* Corresponding author, at address a. Fax: +44 (0) 1865 275 216; E-mail : [email protected] Present address: Oxford Glycobiology Institute, Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1

3QU, UK.

BBAPRO 36037 27-1-00

Biochimica et Biophysica Acta 1476 (2000) 265^278www.elsevier.com/locate/bba

their constituent proteins results in the accumulation of coloured/fluorescent crystallins with age. ß 2000 Elsevier ScienceB.V. All rights reserved.

Keywords: Crystallin; Kynurenine; Protein modi¢cation; Lens; Ageing; Cataract

1. Introduction

The crystallins form a family of proteins which arepresent at high concentrations in the ¢bre cells of thelens [1]. Lens transparency is dependent on the main-tenance of the crystallins in a native state, forming ahighly ordered transparent medium through whichlight passes to permit vision. Modi¢cation of thecrystallins, for example by glycation or oxidation,can result in altered crystallin conformation, precip-itation, and lens opacity and may induce cataract[2,3]. In primate lenses, the ¢bre cells also containUV ¢lter compounds which are derived from theTrp oxidation product, kynurenine (Kyn). TheseUV ¢lters are thought to protect the retina fromthe radiation in the 295^400 nm range which passesthrough the cornea and aqueous humour and mayalso play a role in reduction of chromatic aberration.In order of decreasing concentration in the lens, themajor UV ¢lters are 3-hydroxykynurenine O-L-D-glu-coside (3-OHKG), 4-(2-amino-3-hydroxyphenyl)-4-oxobutanoic acid O-L-D-glucoside (AHBG), Kynand 3-hydroxykynurenine (3-OHKyn) [4,5]. In addi-tion, we have recently discovered a novel lenticular£uorophore which is formed via the nucleophilic ad-dition of the glutathionyl Cys to a deaminated formof 3-OHKG [6]. Unlike the other UV ¢lters, the con-centration of the GSH-3-OHKG appears to increasewith age [6].

As the human lens ages, and with the onset ofsenile nuclear cataract, the crystallins become yellowand £uorescent [7^10]. In cataract, the lens may be-come coloured dark brown to black in the most se-vere cases [11]. It has been proposed that the cova-lent attachment of the Kyn-derived UV ¢ltercompounds may, at least in part, underlie the mod-i¢cations of the crystallins which result in colouredand £uorescent products [12^16]. However, thechemical means by which modi¢cation occurs isnot fully understood. Our group has recently shownthat, under oxidative conditions, 3-OHKyn forms

cross-links between the K-amino groups of speci¢camino acid residues [17]. This reactivity is dependenton the redox active hydroxyl moiety of 3-OHKyn.Others have suggested that it is necessary for3-OHKG to become deglucosylated in order for itto react with crystallins, and that this may be in-duced by photoexcitation at 365 nm [16]. Our recentwork, which identi¢ed the GSH-3-OHKG adduct inhuman lenses and utilised mass spectrometry andtwo-dimensional nuclear magnetic resonance(NMR) studies to de¢ne the structure of the covalentlinkage, indicates that the side chain of Kyn-derivedUV ¢lters plays a crucial role in the formation ofcovalent adducts with amino acids under non-oxida-tive conditions [6].

The aim of the present studies was to examine thepossible covalent modi¢cation of calf lens crystallinsby Kyn and characterise the products for colour and£uorescence development. We further sought to iden-tify which amino acids were involved in any observedmodi¢cations.

2. Materials and methods

2.1. Materials

All organic solvents were HPLC grade (Ajax, Uni-chrom, Auburn, NSW, Australia). Reduced gluta-thione (GSH), L-kynurenine, poly-Lys, poly-His,5,5P-dithiobis-2-nitrobenzoic acid (DTNB), 2,4,6-trinitrobenzenesulphonic acid (TNBS), iodoaceticacid (99% pure), succinic anhydride and phenylmeth-ylsulphonyl £uoride (PMSF) were from Sigma(St. Louis, MO, USA). Tri£uoroacetic acid (s 99%pure, TFA) was from Aldrich (Milwaukee, MI,USA) and acetic acid (s 99.8% pure) from BDH(Poole, UK). Other reagents were obtained throughcommercial suppliers. Milli-Q water (puri¢ed to18 M6 cm32) was used in the preparation of all so-lutions.

BBAPRO 36037 27-1-00

B. Garner et al. / Biochimica et Biophysica Acta 1476 (2000) 265^278266

2.2. Preparation of bovine calf lens crystallins

Calf eyes (up to 2 years old) were obtained freshfrom a local abattoir and the lenses removed andstored at 320³C. A crude crystallin preparationwas obtained after lenses were homogenised in 80%ethanol, placed on ice for 30 min and centrifuged at12 000 rpm in an Eppendorf microfuge at 4³C for15 min. The supernatant was discarded and the pelletre-extracted and subsequently dried under vacuum.The major crystallin classes were puri¢ed as de-scribed previously [18] with minor modi¢cations.Brie£y, lenses were homogenised in 50 mM Tris(pH 7.2) which contained 5 mM ethylenediaminete-traacetic acid (EDTA), 1 mM dithiothreitol (DTT),0.04% (w/v) NaN3 and 0.1 mM PMSF. After centri-fugation for 15 min at 12 000 rpm and 4³C in anEppendorf microfuge, the supernatant (2 ml) wasloaded onto a Sephacryl S-300 (Pharmacia, Uppsala,Sweden) gel ¢ltration column (90U2.5 cm) and thecrystallins eluted with 50 mM Tris, 5 mM EDTA,0.04% NaN3 at a £ow rate of 20 ml/h and collectedby monitoring 280 nm absorbance. Crystallin frac-tions were subsequently concentrated by ultra¢ltra-tion in an Amicon cell using a YM10 membrane.

2.3. Modi¢cation of calf lens crystallins by kynurenine

Crystallins prepared by either ethanol precipitationor by gel ¢ltration were dissolved in 25 mM carbo-nate bu¡er (pH 9). In some cases the crystallins were¢rst solubilised in 5% (w/v) sodium dodecyl sulphate(SDS) to prevent aggregation and precipitation. Pre-treatment of the crystallins with SDS did not a¡ectsubsequent reactivity with any of the reactants em-ployed. Kynurenine (3 mg/ml) was added to the crys-tallins (approx. 5 mg/ml) and the solutions wereplaced under argon (to limit thiol autoxidation),sealed with plastic ¢lm and incubated in the darkfor 48 h at 37³C. After incubation, Kyn-treated crys-tallins were either re-isolated by precipitating twicewith 80% ethanol, followed by dissolving in 6 Mguanidine HCl, dialysis against 1000 vols. of H2Oand re-solubilisation in guanidine HCl, or by passagethrough two consecutive columns of Sephadex G25and eluting with H2O to remove low-molecularweight compounds (unreacted Kyn in particular).In the initial studies, the Sephadex G25-isolated

Kyn-modi¢ed crystallins were also lyophilised thendissolved in 6 M guanidine HCl and dialysed against1000 vols. of H2O to ensure that the protein colour-ation was due to covalent modi¢cation. The addi-tional dialysis step did not reduce the 365 nm absor-bance of the puri¢ed Kyn-modi¢ed crystallins andwas therefore not routinely included. Later studiesof tryptic peptides derived from Kyn-modi¢ed crys-tallins con¢rmed that modi¢cations were due to co-valent attachment of the UV ¢lter to the protein (seebelow).

2.4. Carboxymethylation and succinylation ofcrystallins

Where indicated, SDS-solubilised crystallins werecarboxymethylated or carboxymethylated and succi-nylated according to established methods [19].Brie£y, carboxymethylation was achieved by reacting50 mg of crystallin in pH 9 carbonate bu¡er contain-ing 0.1 M iodoacetic acid for 30 min at 22³C. Excessiodoacetic acid was removed by Sephadex G25 chro-matography. In some experiments, the carboxyme-thylated crystallins were subsequently succinylatedby additions of 10 mg of succinic anhydride at ¢veintervals over 30 min at 22³C. The pH was main-tained at pH 8^9 throughout the succinylation byaddition of NaOH. Carboxymethylation and succi-nylation were conducted primarily to derivatise Cys(-SH) and Lys (O-NH2) residues, respectively, in or-der to assess their relative contribution towards Kynadduct formation (note: `Cys' refers only to reducedcysteine throughout this article). It should be notedthat these reagents are not totally selective; however,e.g. His residues are also susceptible to carboxy-methylation and succinylation [20,21]. In one seriesof experiments, GSH (33 mM) was added to thecrystallin-Kyn incubation mixture in order to com-petitively inhibit crystallin modi¢cation.

Modi¢cation of crystallin Cys and Lys residues (byiodoacetic acid, succinic anhydride or Kyn) was as-sessed by subsequent reactivity toward DTNB andTNBS as assessed by absorbance at 412 nm and340 nm, respectively [22,23], and compared tosham-treated crystallins (i.e. exposed to the samepH and incubation conditions employed for carbox-ymethylation and succinylation). The absorbance ofcontrol samples, which did not contain the colori-

BBAPRO 36037 27-1-00

B. Garner et al. / Biochimica et Biophysica Acta 1476 (2000) 265^278 267

metric reagents, was subtracted from the treatedcrystallins.

2.5. Modi¢cation of poly-His and poly-Lys

Poly-His and poly-Lys were incubated under thesame conditions used to generate crystallin-Kyn ad-ducts (described above). After 72 h at 37³C, themodi¢ed polypeptides were puri¢ed by two sequen-tial passes through Sephadex G25 and the absor-bance and £uorescence characteristics assessed. Inthe case of poly-His, the puri¢ed product was onlysparingly soluble in aqueous solution, and was there-fore dissolved in 6 M guanidine HCl, dialysedagainst 1000 vols. of H2O and ¢nally redissolved in6 M guanidine HCl before absorbance/£uorescenceanalysis.

2.6. Fluorescence and UV-visible absorbancespectrophotometric measurements

The modi¢ed crystallins (devoid of low-molecularweight compounds) were dissolved in either 6 M gua-nidine HCl or H2O and used to obtain UV-visibleabsorbance spectra, using a Shimadzu UV-265 spec-trophotometer (Kyoto, Japan), and £uorescencespectra, recorded on a Hitachi F-4500 £uorescencespectrometer (Tokyo, Japan) in three-dimensionalscan mode. Slit widths were routinely 2.5 nm forexcitation and 5 nm for emission.

2.7. Structural characterisation of poly-Lys-Kyn byNMR spectroscopy

One- and two-dimensional 1H-NMR spectra wereacquired at 400 MHz and 25³C using a VarianUnity-400 spectrometer (Varian, Palo Alto, CA,USA) as previously described [6]. The poly-Lys-Kyn adduct was dissolved in D2O to give a concen-tration of approx. 4 mM. TOCSY (spin lock times of30 and 80 ms) and NOESY (mixing time of 250 ms)NMR experiments were acquired in the phase-sensi-tive mode using time-proportional phase incrementa-tion. A coherence-selected gradient COSY experi-ment was acquired in the absolute value mode.Typically, 512 t1 increments, with up to 96 scansper increment, were acquired over 2048 data pointswhich were zero-¢lled to 2048 data points in both

dimensions and multiplied by a Gaussian windowfunction prior to Fourier transformation. All chem-ical shift values (N) are given in ppm. Spectra werereferenced to residual HDO at N 4.81.

2.8. Preparation and amino acid sequence analysis ofKyn-modi¢ed peptides

Kynurenine-modi¢ed crystallins (routinely 5^7 mgof protein) were dried under vacuum and resus-pended in 1 ml of 0.1 M NH4CO3, 0.1 mM CaCl2bu¡er (pH 8) for 24 h at 37³C in the presence oftrypsin (Sigma, cat. No. T8642) at a ¢nal ratio(w/w) of 1:25. Trypsin was added in two aliquots,one at t = 0 h and the other at t = 12 h. The trypticdigests were centrifuged for 2 min at 12 000 rpm andthe supernatant injected onto a C8, 250U4.6 mm,5 Wm particle size, 100 Aî pore size, HPLC column(Microsorb, Rainin, Walnut Creek, CA, USA) forseparation of the peptides. The HPLC system con-sisted of an LC 1150 HPLC pump (ICI instruments),a Rheodyne 7125 sample injector (equipped with a100 Wl sample loop) and a SD 2100 UV-Vis variablewavelength detector (ICI Instruments), set at either229 or 365 nm. Peptides were eluted using a linearH2O/acetonitrile gradient (0.05% v/v TFA) of 0^80%acetonitrile over 40 min and at a £ow rate of 1.0 ml/min. Isolated peptides were then dried under vac-uum, redissolved in 100 Wl H2O/0.05% TFA, re-pu-ri¢ed using a more shallow gradient (0^40% acetoni-trile/0.05% TFA over 40 min) and the fractionsadded directly to polyvinylidene di£uoride (PVDF)discs (obtained from either Bio-Rad or PE AppliedBiosystems). Amino acid sequencing was performedon an Applied Biosystems Procise 494 protein se-quencer using a standard pulsed liquid blot sequenc-ing procedure [24]. The major amino acid released ateach step was used to determine the peptide se-quence.

2.9. HPLC-mass spectrometry

The supernatant from the GSH-Kyn-crystallin in-cubation mixture was collected and further analysedby microbore HPLC (Applied Biosystems, Model172 Separation System, Foster City, CA, USA) usingan Alltima 250U2.1 mm, C18 column (Alltech, cat.No. 88371, Deer¢eld, IL, USA). Samples were eluted

BBAPRO 36037 27-1-00


using an acetonitrile gradient in aqueous 1% (v/v)acetic acid and a £ow rate of 200 Wl/min. The gra-dient was run linearly from 0 to 40% acetonitrileover 40 min. Samples were detected by monitoringabsorbance at 360 nm and by in-line electrosprayionisation mass spectrometry (HPLC-ESIMS).Mass spectra were obtained on a VG Quattro quad-rupole mass spectrometer (VG Biotech, Altrincham,UK) equipped with an upgraded ESI source. Massspectra were acquired in positive ion mode with ascan rate of 100 m/z per second. The source wasmaintained at 150³C.

3. Results

3.1. Modi¢cation of crystallins by Kyn

Our initial experiments aimed to study whetherKyn was reactive towards crystallins and, if so,which were the main amino acids involved. A crudecrystallin preparation was isolated by ethanol precip-itation of whole calf lens homogenates. This puri¢-cation procedure excludes low-molecular weight len-ticular compounds such as GSH and ascorbic acid.The isolated crystallin preparation was then incu-bated for 48 h at 37³C at pH 9 under argon. Theseconditions were chosen based on our previous workwhich optimised conditions for synthesis of a noveladduct of GSH and the Kyn-derived UV ¢lter com-pound, 3-OHKG [6]. The rate-limiting step in thissynthetic reaction is the deamination of the Kynside chain. The kinetics and yields of adduct forma-tion were similar for both 3-OHKG and Kyn (i.e.yields of approx. 20% with respect to initial3-OHKG or Kyn concentration after 48 h at pH9). We previously demonstrated an approx. 10-foldincrease in the reactivity of the Kyn side chain whenthe pH was adjusted from 7 to 9 (see also Section 4).Fig. 1 demonstrates that after separation from theKyn reaction mixture, the crystallins exhibited anabsorbance peak at 365 nm. The control incubationdid not absorb signi¢cantly at this wavelength. Theabsorbance peak at 279 nm is due mostly to Trpresidues in the proteins. The Kyn-treated crystallinsalso became £uorescent, with maximum £uorescenceintensity observed at 380 nm excitation and at 450and 490 nm emission (Fig. 1 inset). This is the ¢rst

demonstration that Kyn can modify crystallins in theabsence of (photo)oxidation.

We next considered the possibility that a speci¢cfraction of the crystallins was preferentially targettedfor Kyn-mediated modi¢cation. In order to test this,the major crystallin classes were separated using gel-¢ltration chromatography and each of the crystallinsisolated were incubated with Kyn at pH 9 as above.Gel ¢ltration allowed the separation of the crystallinsinto ¢ve fractions. These were: K, LH, LL, QS and QT

(Fig. 2). Incubation of each of the fractions (as in-dicated in Fig. 2) with Kyn resulted in the generationof some degree of 365 nm absorbance in the crystal-lins (Table 1). We predicted that possible `nucleo-philic' amino acid side chains (such as are presentin Cys, Lys or His) might be involved in the forma-tion of the Kyn adducts. Interestingly, the relativeproportion of Cys present in the major crystallinfamilies (K, L or Q) was correlated with the colourintensity of each of the crystallin classes (Table 1).This raised the possibility that either the availabilityor relative content of Cys could in£uence the crystal-lin's reactivity towards Kyn. It also seemed likely,given the substantial level of colour produced inK-crystallin, that other residues such as Lys andHis were involved. We next used chemical modi¢ca-

Fig. 1. Absorbance and £uorescence pro¢les of Kyn-modi¢edcalf lens proteins (CLP). CLP were incubated at pH 9 for 48 hin the presence of Kyn (see Section 2 for details). The Kyn-modi¢ed proteins were then re-isolated using Sephadex G25chromatography and dialysis of the guanidine HCl-solubilisedproteins against water. The maximum (non-Trp) absorbance ofthe Kyn-modi¢ed CLP was at 365 nm (upper trace). A controlsample which was incubated in the absence of Kyn is alsoshown (lower trace). The inset shows a contour plot of the £uo-rophores present in Kyn-modi¢ed CLP. The maximum intensityof the £uorophores was attained with 380 nm Ex/450 and490 nm Em.

BBAPRO 36037 27-1-00


tion of the crystallins to examine which were themain residues involved in Kyn-mediated modi¢ca-tion.

3.2. Inhibition of Kyn-mediated crystallin modi¢cationby selective amino acid derivatisation

Calf lens crystallins were carboxymethylated (CM)or carboxymethylated and succinylated (CMS) in or-der to derivatise reactive thiol (Cys) and amino (Lys)residues. Under our experimental conditions His

modi¢cation is also likely to occur [20,21]. Approx.90% of Cys and 50% of the reactive Lys residueswere modi¢ed by CM and CMS as de¢ned by reac-tivity with DTNB and TNBS, respectively. WhenCM-crystallins or CMS-crystallins were subsequentlytreated with Kyn, the formation of coloured, £uores-cent adducts was inhibited. Carboxymethylationalone inhibited the development of 365 nm absor-bance by 71% while 365 nm absorbance in CMS-crystallins was inhibited by 91%. Fig. 3 shows thatthe formation of £uorescence (using the major £uo-rophore at Ex 380 nm/Em 450 nm to quantify) wasalso inhibited by 68% and 96% in the CM- andCMS-crystallins, respectively. This indicates thatthe crystallin residues which were susceptible to car-

Fig. 3. Inhibition of Kyn-mediated modi¢cation of calf lensproteins by carboxymethylation alone, carboxymethylation andsuccinylation, or GSH addition. Isolated CLP (5 mg/ml) wereincubated for 48 h at pH 9 with Kyn (14.4 mM). The CLPwere re-puri¢ed by passing through Sephadex G25 twice andthe £uorescence measured. Where indicated, CLP were carboxy-methylated (CM) or carboxymethylated and succinylated(CMS) prior to incubation. A control sample incubated withoutKyn was also included as was a sample which was incubated inthe presence of Kyn and GSH (33 mM). (A) Control CLP; (B)Kyn-modi¢ed CLP; (C) Kyn-modi¢ed CM-CLP; (D) Kyn-modi¢ed CMS-CLP; (E) Kyn-modi¢ed CLP with GSH present.The histogram shows relative £uorescence intensity of A^E,where condition (B) is de¢ned as 100%.

Fig. 2. Separation of the major crystallin classes by gel-¢ltrationchromatography. Whole calf lenses were homogenised, appliedto a Sephacryl S-300 column and eluted with 50 mM Tris bu¡-er. The fractions indicated were collected, concentrated and in-cubated individually with Kyn to give the Kyn-modi¢ed pro-teins which are compared in Table 1.

Table 1Comparison of Kyn-mediated 365 nm absorbance in the majorcrystallin classes versus their content of the nucleophilic aminoacids: Cys, Lys and Hisa

Crystallin A365 nm % Cys % Lys % His

K 0.56 þ 0.18 0.3 4.3 4.1LH 0.90 þ 0.33LL 1.44 þ 0.12 1.7 3.7 3.4QS 2.15 þ 0.30QT 1.18 þ 0.14 3.3 2.7 3.0aThe major classes of crystallins were separated by gel-¢ltrationchromatography and individually assessed for their reactivitytowards Kyn at pH 9. The Kyn-modi¢ed crystallins were thenassessed for development of colour (365 nm absorbance). Cor-relations between the levels of 365 nm absorbance detected andthe levels of Cys were signi¢cant when the crystallin `families'(i.e. K, LH�L and QS�T) were compared. For absorbance at 365nm vs. % Cys, R2 = 0.99, P = 0.05. Data are mean þ range, n = 2.

BBAPRO 36037 27-1-00


Fig. 4. Characterisation of the GSH-Kyn adduct by microbore HPLC/electrospray ionisation mass spectrometry. A sample of theGSH/Kyn mixture resulting from the incubation condition described in `E' of Fig. 3 was collected after removal of CLP and analysedby HPLC-ESIMS (see Section 2 for details). (A) Chromatogram for absorbance at 360 nm; (B) chromatogram for the +ve ion at m/z209; (C) chromatogram for the +ve ion at m/z 500; (D) mass spectrum of m/z 500 +ve ion and proposed structure of GSH-Kyn. Thefragmentation ions shown at m/z 308 and m/z 371 are consistent with the loss of Glu (129 Da) and deaminated Kyn (192 Da) fromthe proposed structure. The y ordinates show relative signal intensity.

BBAPRO 36037 27-1-00


boxymethylation and succinylation also played animportant role in the protein's reactivity towardKyn. It is most likely that these residues includeCys, Lys and His. In agreement with this, the reac-tivity of Kyn-modi¢ed crystallins towards DTNB orTNBS was decreased by 70 þ 6% and 83 þ 5%, re-spectively (mean þ range, n = 2) when compared tocrystallins incubated in the absence of Kyn.

We also assessed whether Kyn-mediated crystallinmodi¢cation could be competitively inhibited byadding an excess of GSH, i.e. a competing nucleo-phile. Fig. 3 shows that in the presence of GSH, atapprox. 100-fold molar excess (with respect to pro-tein thiol concentration), the formation of the majorcrystallin £uorophore was inhibited by 93%. The de-velopment of 365 nm absorbance was also inhibitedby 61%. If our prediction that GSH protected thecrystallins by a competitive reaction was correct,then a GSH-Kyn adduct should also be detectablein the reaction mixtures. This was directly assessedby HPLC-ESIMS analysis of the crystallin-free frac-tion of the reaction mixtures. This fraction of thereaction mixture was obtained by washing the Sepha-dex G25 column (which was initially used to re-iso-late the `modi¢ed' crystallin) with a further 5 columnvolumes of H2O. Fig. 4A shows that the recoveredfraction contained two 360 nm-absorbing com-pounds which were separated by microbore HPLC.The positively charged molecular ions correspondingto these compounds had m/z values of 209 and 500,i.e. molecular masses of 208 and 499 Da, respectively(Fig. 4B,C). The 208 Da compound was due to re-sidual Kyn and the 499 Da compound was readilyexplained by the formation of an adduct of deami-nated Kyn with GSH (Fig. 4D inset). The split peakcorresponding to the GSH-Kyn adduct (Fig. 4A) isexplicable by the non-stereoselective Michael addi-tion at the L-carbon of the Kyn side chain. The re-sulting diastereoisomers both have a mass of 499 Dawhich is indicated by the positive ion value atm/z = 500 (Fig. 4C). Fragment ions which coelutedwith the GSH-Kyn adduct were 129 and 192 Daless than 500 and can be explained by the loss ofGlu and deaminated Kyn (respectively) from the pro-posed structure (Fig. 4D inset). This proposed struc-ture and fragmentation pattern is entirely consistentwith our previous observations concerning the for-mation of GSH-3-OHKG [6]. The present data con-

¢rm that GSH protected the crystallins from modi¢-cation by competing for reaction with deaminatedKyn. The other major positive molecular ions de-tected were for GSH (m/z = 308) and GSSG (m/z =613) which eluted at 9.5 and 14 min, respectively, butdo not appear on the chromatogram (Fig. 4A) asthey do not absorb signi¢cantly at 360 nm. The rel-ative abundance of the GSH ion was almost an orderof magnitude higher than GSSG, consistent with themaintenance of a reducing environment.

To further assess the possible reactivity of Kyntowards Lys and His residues, incubations of poly-Lys and poly-His were also conducted at pH 9. There-isolated, modi¢ed polypeptides (see Section 2 fordetails) became yellow coloured (365 nm absorbancemaximum). Both of the Kyn-modi¢ed polypeptideswere also £uorescent, with the major £uorescencepeaks at Ex 380 nm/Em 450 nm for poly-His andEx 380 nm/Em 490 nm for poly-Lys.

Fig. 5. (A) Structure of the poly-Lys-Kyn adduct as determinedby NMR spectroscopy. The arrow indicates an observed nu-clear Overhauser e¡ect. (B) Proposed structure of the poly-His-Kyn adduct. Note that adduct formation at N-1 of the imida-zole ring is also theoretically possible.

BBAPRO 36037 27-1-00


3.3. Structural characterisation of poly-Lys-Kyn byNMR spectroscopy

In order to elucidate the structure of the covalentlinkage of Kyn to Lys, the synthetic poly-Lys-Kynadduct was analysed by one- and two-dimensionalNMR spectroscopy (see Fig. 5A for numberingscheme of the Lys-Kyn adduct). The aromatic regionof the one-dimensional 1H-NMR spectrum revealedfour resonances at N 6.93 (1H), 7.46 (1H), 6.84 (1H)and 7.87 (1H) which were assigned from the COSYand TOCSY spectra to protons at C-3, C-4, C-5 andC-6, respectively. Their presence indicates that Lysaddition did not occur at the aromatic ring. Thealiphatic side chain of Kyn contains the CH2-CH(C-8 and C-9) moiety which displayed resonancesat N 3.76 (2H) and 4.10 (1H), respectively. The pres-ence of two protons at C-8 excludes the possibility ofLys addition at this site. The chemical shifts of theprotons in the poly-Lys side chains were observed attheir expected values (N 4.34, 1.82^1.76, 1.50, 1.76and 3.06 for K-CH to O-CH2, respectively) andwere much more intense than the Kyn resonancesindicating that only a small amount of the polymerhad been modi¢ed by Kyn. An isolated Lys O-CH2

resonance from Kyn-modi¢ed poly-Lys was observedat N 3.20 (2H) in the one-dimensional 1H-NMR spec-trum. Integration of this resonance and the unmodi-¢ed poly-Lys O-CH2 resonance at N 3.06 (2H), re-vealed an intensity ratio of 1:25, i.e. suggesting 4%modi¢cation of the poly-Lys side chains by Kyn. Inaddition, the NOESY spectrum revealed a strong

nuclear Overhauser e¡ect across the O-amino groupbetween the O-CH2 protons (N 3.20) of Lys and theC-9 proton (N 4.10) of the Kyn side chain. Overall,these NMR data indicate that the covalent attach-ment of Kyn to poly-Lys was via an N-linked nucle-ophilic addition of the Lys O-amino group to the C-9of the deaminated Kyn side chain, and are analogousto the previously reported structure of the GSH-3-OHKG adduct detected in aged human lenses [6].

3.4. Characterisation of tryptic peptides ofKyn-modi¢ed crystallins

Further information on the likely sites of modi¢-cation in the crystallins was sought by amino acidsequence analysis of coloured/£uorescent peptides.Tryptic peptides were generated using both the crudecrystallin preparation and the puri¢ed crystallin sub-fractions (see Section 2 for details). Results of a typ-ical tryptic digest are illustrated in Fig. 6 using the QT

fraction as an example. At least 68 peptides wereresolved when assessed by 229 nm absorbance (Fig.6A). Of these, seven coloured (365 nm absorbing)fractions were collected (Fig. 6B). The coloured pep-tides were also £uorescent as indicated by Ex360 nm/Em 500 nm monitoring (not shown). The280 nm:365 nm absorbance ratio for the whole tryp-tic digest shown in Fig. 6 was 1:0.17. The isolatedcoloured peptides were re-puri¢ed on a more shallowacetonitrile gradient to give homogeneous samples asassessed by 365 nm absorbance (see inset to Fig. 6Bfor example).

Table 2Tryptic peptide sequence data for Kyn-modi¢ed crystallinsa

Peptide detected Identi¢cation

1. HFSPEDLTVK KA 79^882. HFSPEELK KB 83^903. VLGDVIEVHG K KB 93^1034. VLGDVIEVHG KHEER KB 93^1075. IIIFEQENFQ GHSHELNGPC (PNLK) LB2 18^376. AGSVLVQAGP WVGYEQANCK LB2 58^777. KMEVIDDDVP SFHAH (GYQEK) LB2 120^1348. SCR QA=B=D 77^799. GQMSEITDDC PSLQDR QA=B 101^116

10. FHLTEVHSLN VLEGSWVLYE (MPSYR) QB 116^135aTryptic peptides of Kyn-modi¢ed crystallins were puri¢ed by reversed phase HPLC (monitoring absorbance at 365 nm) and se-quenced as described in Section 2. The residues in parentheses were not detected but are predicted to be at the C-terminus of the pep-tides indicated. The underlined residues indicate possible sites of Kyn addition.

BBAPRO 36037 27-1-00


The sequence data for a panel of puri¢ed, colouredpeptides from all of the crystallin classes are given inTable 2. In all cases, at least one of the candidatenucleophilic amino acids (i.e. Cys, His or Lys) waspresent in the peptide, with His representing a com-mon potential site for modi¢cation. In most cases,the relative abundance of the His residues was re-duced in the sequencing cycles, consistent with theirmodi¢cation by Kyn. Speci¢cally, the yield of His inpeptides 1 and 2 was decreased by approx. 50% (i.e.3.7^4.0 pmol of His versus 7.1^8.0 pmol of Phe), andin peptide 3 by approx. 80% (i.e. 0.8 pmol of Hisversus 4.5 pmol of Val). In peptide 4, His-101 andHis-104 levels were decreased by approx. 85%. Inpeptide 11, His-117 and His-122 levels were de-creased by 70 and 85%, respectively. In the other

His-containing peptides (5 and 7) there was no evi-dence for His loss. However, alternative sites of mod-i¢cation are present at Cys-37 and Lys-120 in pep-tides 5 and 7, respectively. While the technique is notquantitative for Cys residues, the yield of Lys-120was decreased approx. 30%. The data could also in-dicate that a proportion of His (or Lys in the case ofpeptide 7) may not be modi¢ed in these colouredpeptides. The variable recovery of the modi¢ed resi-dues could be due to loss of the Kyn moiety duringcyclisation in the Edman reaction (5 min in 100%TFA at 48³C) or possibly due to the presence ofminor contaminating peptides. These peptide studies,in addition to the studies assessing the reactivity ofKyn-modi¢ed CLP with DTNB and TNBS (de-scribed above), indicate that His, Cys and Lys arethe residues that are most likely to be modi¢ed incrystallins in vitro. We are currently searching foranalogous Kyn- (and 3-OHKG-) modi¢ed trypticpeptides derived from aged/cataractous humanlenses.

4. Discussion

The present work has shown that Kyn has thecapacity to form covalent adducts with crystallinsunder non-oxidative conditions. Previous work hasshown that 3-OHKyn and its autoxidation products[25] can also covalently modify crystallins, but in anoxygen-dependent manner [14,17]. Similarly, it hasbeen proposed that 3-OHKG undergoes deglucosy-lation upon photo-oxidation, leading to the forma-tion of a phenoxyl radical capable of modifying crys-tallins to form £uorescent adducts [16]. Under non-oxidative conditions, Kyn would be considered non-reactive, and for this reason its ability to modifycrystallins has not been studied in detail. However,the recent discovery of a GSH-3-OHKG adduct inthe human lens has revealed a novel reaction path-way for molecules which contain the Kyn amino acidside chain [6]. Using mass spectrometry and two-di-mensional NMR studies, we determined the structureof the GSH-3-OHKG adduct and proposed that itwas formed via deamination of the 3-OHKG aminoacid side chain to form an K,L-unsaturated carbonylwhich was highly susceptible to nucleophilic attackby the Cys of GSH [6]. With this mechanism in

Fig. 6. Puri¢cation of tryptic peptides of Kyn-modi¢ed Q-crys-tallins by reversed phase HPLC. Q-Crystallins (QT) were isolatedusing Sephacryl S-300 gel ¢ltration, incubated with Kyn for 48h at pH 9, re-puri¢ed and digested by trypsin. The resultingpeptides were puri¢ed in two steps using reversed phase HPLC.(A) Chromatogram for absorbance of peptide mixture at 229nm; (B) chromatogram for absorbance at 365 nm (fractionscollected for second HPLC puri¢cation step are labelled 1^7).(B, inset) Example of peptide (No. 2) purity after secondHPLC step. Volumes injected were 20 Wl and 100 Wl for A andB, respectively.

BBAPRO 36037 27-1-00


mind, we were able to synthesise the GSH-3-OHKGadduct and speculated that similar Kyn-related ad-ducts may be formed with GSH and nucleophilicamino acids in proteins. This was con¢rmed in thepresent work.

In the present studies, the reaction of Kyn withcrystallins was routinely performed at pH 9. In otherwork, we have also shown that the Kyn side chain of3-OHKG has the ability to form adducts with GSH[6] and proteins [26] at physiological pH and thatadduct formation is accelerated at pH 9. The in-creased rate of Kyn adduct formation at the higherpH is explicable by two factors. First, deamination ofthe Kyn side chain is favoured at the higher pH[27,28], and second, the nucleophilicity of aminoacids is closely related to their pKa values, the depro-tonated forms being more reactive [29,30]. Since pro-tein turnover is either extremely slow or absent in thecentre of the lens [1], we propose that throughout thelifetime of an individual, a pool of the Kyn-derivedUV ¢lters (including Kyn, 3-OHKG and 3-OHKyn)becomes deaminated and subsequently forms adductswith the nucleophilic amino acids of the crystallinswhich we have identi¢ed here (namely His, Cys andLys). This proposal is consistent with the higher lev-els of both GSH- and protein-3-OHKG adducts de-tected in aged human lenses [6,26]. The reactionmechanism for Kyn deamination and addition toprotein (e.g. with Cys as the nucleophile) is summa-rised in Scheme 1. The products which result fromanalogous reactions of the Kyn side chain with Hisor Lys are given in Fig. 5. The pKa values for theside chains of His, Cys and Lys are 6.0, 8.33 and10.53, respectively, which indicates that His may bemore likely to be involved in protein modi¢cationsin the lens (pHW7). Of possible signi¢cance, the nu-cleophilic addition of His residues of insulin andapolipoprotein-B100 to 4-hydroxynonenal (i.e. anK,L-unsaturated carbonyl) also occurs under physio-logical conditions [31,32].

The overall structure of the protein could also in-£uence a speci¢c amino acid's reactivity. Because Lysis charged, it is more likely to be exposed on theprotein surface and may therefore be preferentiallyinvolved in Kyn-mediated crystallin modi¢cations invivo. Crystallin Cys residues are often present in hy-drophobic domains (of the Q-crystallins in particular)and this would be predicted to limit Kyn-mediated

modi¢cation. Interestingly, the Cys content of bovineLA3=A4-crystallins is higher than that of LB2, yet wehave not detected coloured Cys-Kyn-containing tryp-tic peptides from the former groups. A possible ex-planation for this is that the surface-exposed Cys

Scheme 1. Deamination of Kyn and its covalent attachment tothe Cys residue of a protein. The carbonyl function in the Kynside chain underlies an inherent instability with respect to lossof the amino group from the L-carbon atom. At pH 7, the car-bonyl group acidi¢es the adjacent H atom (1), resulting in theelimination of ammonia from the L-carbon and bond rearrange-ment to form an K,L-unsaturated carbonyl (2). Note that atphysiological pH (7.4), this reaction is slow and therefore islikely to be the rate-limiting step in Kyn adduct formation invivo. The L-carbon atom of this type of structure is well-knownto be highly susceptible to non-stereoselective nucleophilic at-tack, e.g. by the thiolate anion of a Cys residue of a protein(2). This results in the formation of an enolate intermediate(not shown) that becomes protonated to give the Kyn-proteinadduct (3). Adducts of GSH-benzoyl acrylic acid, GSH-3-OHKG and GSH-Kyn all yield equal proportions of each ofthe diastereoisomers predicted to be formed at the L-carbon. Inproteins, however, the relative orientation of neighbouring ami-no acids could in£uence isomer formation. R = H (Kyn), OH(3-OHKyn), O-glucose (3-OHKG). X and Y indicate other ami-no acids in the polypeptide.

BBAPRO 36037 27-1-00


residues are suggested to be involved in interactionswith aromatic side chains of other crystallin subunits[33]. This is thought to `protect' the sulphydryl resi-dues from forming disulphides and, we speculate,may also limit their interaction with Kyn. In con-trast, it appears that while Cys residues of Q-crystal-lins are largely buried in hydrophobic domains, cer-tain surface-exposed Cys residues of the Q-crystallinN-terminal domains react readily with GSH to formmixed disulphides [34]. The predicted `vulnerability'of such surface exposed Cys is consistent with thecoloured Cys-containing Q-crystallin peptides de-tected in the present work (esp. SCR). Factors inaddition to the pKa value of an amino acid side chaintherefore are also likely to in£uence reactivity to-wards Kyn. The observed decrease in levels of boththe cysteinyl-SH (which includes oxidative loss) andlysyl O-NH2 reported in crystallins isolated from thenuclear region of human cataracts may be partly dueto modi¢cation with UV ¢lters as we have describedhere [35,36].

Deamination of the side chain of Kyn (and struc-turally related molecules) to yield an unsaturated ke-tone is predicted to give a product that will reactwith several competing nucleophiles. For example,the amino group at the 2 position of the Kyn ringcan act as a nucleophile and thereby form Kyn yel-low-like structures [27]. A compound which may bederived by such a route (xanthurenic acid glucoside)has been isolated from brunescent human lenses [37].In the presence of a competing nucleophile, such asthe Cys residue of GSH, GSH-Kyn and GSH-3-OHKG [6] adducts are also formed. In addition, ifthe unsaturated ketone derivative of 3-OHKG is re-duced, the second most abundant UV ¢lter, AHBG,will be formed [38].

Our results showing that GSH is a competing nu-cleophile suggest that such a mechanism may alsooperate in the lens to prevent crystallin modi¢cation.As we age, however, there is a diminished availabilityof reduced GSH in the nuclear region of the lens [39].This feature may be due to the development of abarrier to the di¡usion of GSH from its site of syn-thesis (or reduction) in the lens cortex, to the interiorof the lens [40]. Lack of crystallin protection by GSHmay therefore contribute to non-oxidative lens col-ouration over time.

Another question which remains to be addressed

concerns the e¡ect that Kyn modi¢cation of crystal-lins may have on the structure and function of theprotein. Oxidative modi¢cation of crystallins by3-OHKyn leads to protein cross-linking and aggre-gation [13,14,17]. In the present work, SDS-PAGEanalysis of the isolated Kyn-modi¢ed crystallinsshowed that no high-molecular weight adducts wereformed (data not shown), arguing against the occur-rence of inter-molecular cross-links. This is consis-tent with our proposed reaction mechanism (Scheme1). It is known that glycation or oxidation of K-crys-tallin can a¡ect its chaperone activity [41,42]. Of in-terest, the putative peptide-binding sites of KA- andKB-crystallins which are believed to be involved inchaperone function [42] are the same as those ofthe Kyn-modi¢ed tryptic peptides we have detectedhere (i.e. peptides 1^4, Table 2).

The possibility that the covalent attachment ofKyn to crystallins leads to increased photosensitivityis another plausible means by which function couldbe altered. Preliminary results from our group show(using EPR spectrometry) that the Kyn-modi¢edcrystallins generate higher levels of protein-boundradicals when excited at 300 nm and compared tonon-modi¢ed crystallins. We speculate that the grad-ual accumulation of Kyn-modi¢ed crystallins in thelens nucleus may sensitise the lens to subsequent(photo)oxidative damage. It is also possible that oth-er post-translational modi¢cations of lens crystallinsmay in£uence their reactivity towards Kyn. Oxida-tion of His to its 2-imidazolone introduces an adja-cent carbonyl (electron withdrawing) which coulddecrease the pKa value of the N-3 proton (for exam-ple) to yield a stronger nucleophile [43]. Oxidation ofCys residues to form disulphides would render theseside chains inactive as the formation of the mercap-tide (i.e. the nucleophile) would not be possible [44].Glycation of the O-NH2 group of Lys side chainswould similarly prevent Kyn addition [45].

We estimate, based on the molar extinction coef-¢cient of GSH-3-OHKG [6], that the modi¢cation ofonly a single amino acid residue would be su¤cientto detectably increase a crystallin's £uorescent prop-erties. We are currently using HPLC-ESIMS tech-niques in an attempt to identify speci¢c modi¢edtryptic peptides from aged human lens crystallins,and thereby determine the relative contribution thatspeci¢c types of modi¢cations make to age-related

BBAPRO 36037 27-1-00


lens colouration. However, this issue remains to beresolved accurately because several post-translationalprotein modi¢cations may occur (e.g. oxidation,glycation) and the resulting tryptic peptide maps(and their MS fragmentation patterns) are very com-plex.

In conclusion, the present studies identify a novelmechanism of post-translational protein modi¢cationwhich results from the deamination of the side chainof Kyn and subsequent nucleophilic addition of ami-no acids. We suggest that in humans this process(and analogous reactions of other lens UV ¢lters)contributes to the known age-related increase inlens colouration and the associated decreased percep-tion of colours in the violet and blue regions. Crys-tallins which are modi¢ed by UV ¢lters via the mech-anism proposed here, may become functionallyimpaired or be subsequently involved in oxidativereactions that occur in cataract.

Acknowledgements

The assistance of Ms Anna Palenkas in peptidecollection, Mrs Jill McGovern in peptide sequencing,and Ms Teresa Murphy in puri¢cation of crystallinclasses is greatly appreciated. Prof. Stephen Pyne isthanked for valuable discussions. This work was sup-ported by the Australian National Health and Med-ical Research Council (JAC, Grant No. 980497;RJWT, Grant No. 980495).

References

[1] E.R. Berman, Biochemistry of the Eye, Plenum Press, NewYork, 1991.

[2] B.J. Ortwerth, P.R. Olesen, Ascorbic acid-induced crosslink-ing of lens proteins: evidence supporting a Maillard reac-tion, Biochim. Biophys. Acta 956 (1988) 10^22.

[3] S. Fu, R.T. Dean, M. Southan, R.J.W. Truscott, The hy-droxyl radical in lens nuclear cataractogenesis, J. Biol.Chem. 273 (1998) 28603^28609.

[4] R. van Heyningen, Fluorescent glucoside in the human lens,Nature 230 (1971) 393^394.

[5] A.M. Wood, R.J.W. Truscott, UV ¢lters in human lenses:tryptophan catabolism, Exp. Eye Res. 56 (1993) 317^325.

[6] B. Garner, S. Vazquez, R. Gri¤th, R.A. Lindner, J.A.Carver, R.J.W. Truscott, Identi¢cation of glutathionyl-3-hy-droxykynurenine glucoside as a novel £uorophore associated

with aging of the human lens, J. Biol. Chem. 274 (1999)20847^20854.

[7] S. Lerman, R. Borkman, Spectroscopic evaluation and clas-si¢cation of the normal, aging, and cataractous lens, Oph-thalmic Res. 8 (1976) 335^353.

[8] G.J.H. Bessems, E. Keizer, J. Wollensak, H.J. Hoenders,Non-tryptophan £uorescence of crystallins from normaland cataractous human lenses, Invest. Ophthalmol. Vis.Sci. 28 (1987) 1157^1163.

[9] M.C. Yappert, L. Sundeep, D. Borchman, Age dependenceand distribution of green and blue £uorophores in humanlens homogenates, Invest. Ophthalmol. Vis. Sci. 33 (1992)3555^3560.

[10] E.L. Finley, J. Dillon, K.K. Crouch, K.L. Schey, Identi¢ca-tion of tryptophan oxidation products in bovine alpha-crys-tallin, Protein Sci. 7 (1998) 2391^2397.

[11] A. Pirie, Color and solubility of the proteins of human cat-aract, Invest. Ophthalmol. 7 (1968) 634^642.

[12] R.J.W. Truscott, S.G. Pyne, M. Manthey, Reactive metab-olite hypothesis for human senile cataract, Lens Eye Toxic.Res. 8 (1991) 251^257.

[13] G.M. Stutchbury, R.J.W. Truscott, The modi¢cation of pro-teins by 3-hydroxykynurenine, Exp. Eye Res. 57 (1993) 149^155.

[14] J.A. Aquilina, J.A. Carver, R.J.W. Truscott, Oxidationproducts of 3-hydroxykynurenine bind to lens proteins: rele-vance for nuclear cataract, Exp. Eye Res. 64 (1997) 727^735.

[15] A.R. Ellozy, R.H. Wang, J. Dillon, Model studies on thephotochemical production of lenticular £uorophores, Photo-chem. Photobiol. 59 (1994) 479^484.

[16] J. Dillon, M. Skonieczna, K. Mandal, D. Paik, The photo-chemical attachment of the O-glucoside of 3-hydroxykynur-enine to K-crystallin: a model for lenticular aging, Photo-chem. Photobiol. 69 (1999) 248^253.

[17] J.A. Aquilina, J.A. Carver, R.J.W. Truscott, Elucidation ofa novel polypeptide cross-link involving 3-hydroxykynure-nine, Biochemistry 38 (1999) 11455^11464.

[18] C. Slingsby, O.A. Bateman, Rapid separation of bovineL-crystallin subunits LB1, LB2, LB3, LA3 and LA4, Exp.Eye Res. 51 (1990) 21^26.

[19] M. Hollecker, Counting integral numbers of residues bychemical modi¢cation, in: T.E. Creighton (Ed.), ProteinStructure, IRL Press, Oxford, 1989, pp. 145^153.

[20] G.E. Means, R.E. Feeney, Chemical Modi¢cation of Pro-teins, Holden-Day, San Francisco, CA, 1971.

[21] A.M. Crest¢eld, W.H. Stein, S. Moore, Alkylation and iden-ti¢cation of the histidine residues at the active site of ribo-nuclease, J. Biol. Chem. 238 (1963) 2413^2420.

[22] P.C. Jocelyn, Spectrophotometric assay of thiols, MethodsEnzymol. 143 (1987) 44^67.

[23] U.P. Steinbrecher, Oxidation of human low density lipopro-tein results in derivatization of lysine residues of apolipopro-tein B by lipid peroxide decomposition products, J. Biol.Chem. 262 (1987) 3603^3608.

[24] G.W. Kilby, M.M. Sheil, D.C. Shaw, J.J. Harding, R.J.W.

BBAPRO 36037 27-1-00


Truscott, Amino acid sequence of bovine gamma E (IVa)lens crystallin, Protein Sci. 6 (1997) 909^912.

[25] S. Vazquez, B. Garner, M.M. Sheil, R.J.W. Truscott, Char-acterisation of the major autoxidation products of 3-hydrox-ykynurenine under physiological conditions, Free Radic.Res. (1999) in press.

[26] B.D. Hood, B. Garner, R.J.W. Truscott, Human lens colo-ration and aging. Evidence for crystallin modi¢cation by themajor ultraviolet ¢lter, 3-hydroxy-kynurenine O-L-D-gluco-side, J. Biol. Chem. 274 (1999) 32547^32550.

[27] T. Tokuyama, S. Senoh, T. Sakan, K.S. Brown, B. Witkop,The photoreduction of kynurenic acid to kynurenine yellowand the occurrence of 3-hydroxy-L-kynurenine in butter£ies,J. Am. Chem. Soc. 89 (1967) 1017^1021.

[28] A. Butenandt, U. Schiedt, E. Biekert, R.J.T. Cromartie,Uë ber Ommochrome, IV. Mitteilung: Konstitution des Xan-thommatins, Justus Liebigs Ann. Chem. 590 (1955) 75^90.

[29] M. Friedman, J.S. Wall, Application of a Hammett-Taftrelation to kinetics of alkylation of amino acid and peptidemodel compounds with acrylonitrile, J. Am. Chem. Soc. 86(1964) 3735^3741.

[30] M. Friedman, J.S. Wall, Additive linear free-energy relation-ships in reaction kinetics of amino groups with K,L-unsatu-rated compounds, J. Org. Chem. 31 (1966) 2888^2894.

[31] K. Uchida, E.R. Stadtman, Modi¢cation of histidine resi-dues in proteins by reaction with 4-hydroxynonenal, Proc.Natl. Acad. Sci. USA 89 (1992) 4544^4548.

[32] G. Ju«rgens, J. Lang, H. Esterbauer, Modi¢cation of humanlow-density lipoprotein by the lipid peroxidation product 4-hydroxynonenal, Biochim. Biophys. Acta 875 (1986) 103^114.

[33] C. Slingsby, H.P.C. Driessen, D. Mahadevan, B. Bax, T.L.Blundell, Evolutionary and functional relationships betweenbasic and acidic L-crystallins, Exp. Eye Res. 46 (1988) 375^403.

[34] C. Slingsby, B. Norledge, A. Simpson, O.A. Bateman, G.Wright, H.P.C. Driessen, P.F. Lindley, D.S. Moss, B. Bax,X-ray di¡raction and structure of crystallins, Prog. RetinalEye Res. 18 (1996) 3^29.

[35] R.J.W. Truscott, R.C. Augusteyn, Changes in human lensproteins during nuclear cataract formation, Exp. Eye Res. 24(1977) 159^170.

[36] S. Garcia-Castineiras, M.N. Miranda-Rivera, Loss of freeamino groups in the water-insoluble fraction of nuclear se-nile cataracts, Invest. Ophthalmol. Vis. Sci. 24 (1983) 1181^1187.

[37] Y. Shirao, E. Shirao, K. Ando, Y. Iwakuchi, D. Balasubra-manian, Glucoside of xanthurenic acid accumulates in bru-nescent but not in non-brunescent lens nuclei in humans,Invest. Ophthalmol. Vis. Sci. 40 (1999) S522.

[38] R.J.W. Truscott, A.M. Wood, J.A. Carver, M.M. Sheil,G.M. Stutchbury, J. Zhu, G.W. Kilby, A new UV-¢lter com-pound in human lenses, FEBS Lett. 384 (1994) 173^176.

[39] J.E. Dickerson, M.F. Lou, Free cysteine levels in normalhuman lenses, Exp. Eye Res. 65 (1997) 451^454.

[40] M.H.J. Sweeney, R.J.W. Truscott, An impediment to gluta-thione di¡usion in older normal human lenses: a possibleprecondition for nuclear cataract, Exp. Eye Res. 67 (1998)587^595.

[41] B.K. Derham, J.J. Harding, alpha-Crystallin as a molecularchaperone, Prog. Retinal Eye Res. 18 (1999) 463^509.

[42] K.K. Sharma, G.S. Kumar, A.S. Murphy, K. Kester, Iden-ti¢cation of 1,1P-bi(4-anilo)naphthalene-5,5P-disulfonic acidbinding sequences in K-crystallin, J. Biol. Chem. 273 (1998)15474^15478.

[43] E.L. Finley, J. Dillon, R.K. Crouch, K.L. Schey, Sites andextent of oxidation in soluble human K-crystallin with age,Invest. Ophthalmol. Vis. Sci. 40 (1999) S301.

[44] S.R.A. Hanson, D.L. Smith, J.B. Smith, Deamidation anddisul¢de bonding in human lens Q-crystallins, Exp. Eye Res.67 (1998) 301^312.

[45] J.A. Dunn, J.S. Patrick, S.R. Thorpe, J.W. Baynes, Oxida-tion of glycated proteins: age-dependent accumulation ofNE-(carboxymethyl)lysine in lens protein, Biochemistry 28(1989) 9464^9468.

BBAPRO 36037 27-1-00


Documents

Non-oxidative modification of lens crystallins by kynurenine: a novel post-translational protein modification with possible relevance to ageing and cataract