Exp. Eye Res. (2000) 71, 599±607doi:10.1006/exer.2000.0912, available online at http://www.idealibrary.com on

Distribution of Ferritin and Redox-active Transition Metals in Normal

and Cataractous Human Lenses

BRETT GARNERab*, KARIN ROBERGb, MINGWEI QIANbc, JOHN W. EATONbc

AND ROGER J. W. TRUSCOTTa

aAustralian Cataract Research Foundation, Department of Chemistry, University of Wollongong, NSW 2522,Australia, bDepartment of Pathology II, Faculty of Health Sciences, University of LinkoÈping, S-581 85LinkoÈping, Sweden and cDepartment of Pediatrics, Baylor College of Medicine, Houston TX, U.S.A.

(Received Oxford 25 May 2000, accepted in revised form 23 August 2000 and published electronically

(Pirie, 19cataract,

0014-483

* AddressGlycobiologOxford, Souglycob.ox.ac

9 October 2000)

Previous studies have shown that lenticular levels of Fe and Cu are elevated in age-related cataract.However, it is not known if these metals are present in a state that is permissive for redox reactions thatmay lead to the formation of free radicals. In addition, there is little data available concerning theconcentration and lenticular distribution of ferritin, the major intracellular Fe-sequestering protein, inthe lens. The aim of the present work was therefore to determine the distribution of ferritin and theredox-availability of Fe and Cu in healthy and cataractous lenses. Lens ferritin distribution was assessedby ELISA and immunohistochemistry. A modi®ed ELISA detected ferritin in an `insoluble' lens proteinfraction. Ferritin levels were not signi®cantly different in the cortex vs nucleus of healthy lenses. Incontrast, ferritin levels in the cataractous lens nuclei appeared to be 70 % lower compared to the cortex.This was at least partially due to the presence of ferritin within an insoluble protein fraction of thehomogenized lenses. In normal lenses, ferritin staining was most intense in the epithelium, with diffusestaining observed throughout the cortex and nucleus. The redox-availability of lenticular metals wasdetermined using: (1) autometallography; (2) Ferene-S as a chromogenic Fe chelator; and (3) NO�release from nitrosocysteine to probe for redox-active Cu. The autometallography studies showed thatthe cataractous lenses stained more heavily for redox-active metals in both the nucleus and cortex whencompared to age-matched control lenses. Chelatable Fe was detected in homogenized control lenses afterincubation with Ferene-S, with almost three-fold higher levels detected in the cataractous lenses onaverage. The Cu-catalysed liberation of NO� from added nitrosocysteine was not demonstrated in anylens sample. When exogenous Cu (50 nM) was added to the lenses, it was rapidly chelated. Thecataractous samples were approximately twice as effective at redox-inactivation of added Cu. Thesestudies provide evidence that a chelatable pool of potentially redox-active Fe is present at increasedconcentrations in human cataractous lenses. In contrast, it seems that lenticular Cu may not be readily

available for participation in redox reactions. # 2000 Academic Press

oxyl

Key words: cataract; iron; copper; ferritin; hydr

1. Introduction

Cataract is the world's largest single cause of blindness

(Kupfer, Underwood and Gillen, 1994). The causes ofcataract can be multifactorial, and are often associated

with other speci®c disease states. For example, a higher

prevalence of cataract is noted in subjects with dia-betes, chronic renal failure, certain hereditary syn-

dromes and, in the general population, with ageing

(Harding, 1991; Levi et al., 1998). One feature thatthese different states of cataract appear to have in

common is that the structural order of the major pro-

teins of the lens, the crystallins, is altered, leading tochanged protein conformation, aggregation, precipi-

tation and in some cases a yellow to brown colouration

68; Harding, 1991). In age-related nuclearan important cause of post-translational

5/00/120599�09 $35.00/0

correspondence to: Brett Garner, The Oxfordy Institute, Department of Biochemistry, University ofth Parks Road, Oxford OX1 3QU, U.K. E-mail: brett@.uk

-radical.

protein modi®cation results from oxidative damage(Truscott and Augusteyn, 1977; Garner and Spector,1980). The recent identi®cation of speci®c hydroxy-lated amino acids in age-related cataractous lensproteins provides strong evidence for HO� as acataractogenic species (Fu et al., 1998). In addition,in an ex vivo model system, human cataractous lenshomogenates have the capacity to generate higherlevels of HO� when compared to non-cataractouslenses (Garner, Davies and Truscott, 2000).

Previous studies suggested that metal ion-catalysedformation of HO� could contribute to the proteinmodi®cations observed in cataract (Garland, 1990). Ingeneral, there appear to be higher levels of Fe and Cupresent in human cataractous lenses compared tohealthy controls (Lakomaa and Eklund, 1978; Eckhert,1983; Cook and McGahan, 1986; Garland, 1990;Cekic, 1998; Cekic et al., 1999; Garner et al., 2000).However, it is not certain where these metals are

located within the lens nor to what degree they areavailable for participation in redox reactions that could

# 2000 Academic Press

bance at 594 nm. The relative amounts of lenticular

supplemented with 50 nM CuSO4 prior to addition of

generate free radicals; HO� in particular (Garner et al.,2000).

It is clear that in other cell types the levels andintracellular location of the major Fe-binding protein,ferritin, can modulate Fe-catalysed radical formation(Balla et al., 1992; Garner, Roberg and Brunk, 1998).However, there are only limited data available on thedistribution of ferritin in the normal and cataractoushuman lens. The aims of the present study weretherefore to use techniques that could reveal thepossible redox-availability and location of Fe and Cuin the lens and also to study the relative distribution oflenticular ferritin.

2. Materials and Methods

Materials

Ascorbic acid, 3-(2-pyridyl)-5,6-bis(2-[5-furylsulfo-nic acid])-1,2,4-triazene (Ferene-S), Triton X-100,Chelex-100, CuSO4 and bovine serum albumin (BSA)were from Sigma Chemical Co. (St. Louis, MO, U.S.A.).All other materials were obtained through commercialsuppliers or as described previously (Garner et al.,1997, 1998).

Lens Collection

Cataractous lenses were imported from Cambodiaafter extracapsular extraction to alleviate blindness.The lenses were all classi®ed as Type II±III based oncolour intensity (Pirie, 1968) and were stored in either70 % (v/v) HPLC-grade ethanol in Milli-Q H2O or 10 %neutral buffered formalin (Sigma, Cat. no. HT50-1-4)and kept on ice during transit. The ethanol-®xed andformalin-®xed lenses were subsequently stored at ÿ20and 48C, respectively, until use. Non-cataractouslenses were obtained in Australia from donor eyes usedfor corneal grafting and stored as for the cataracts. Atotal of 27 cataractous lenses were collected[age � 63 + 9 years (mean + S.D.), of which 18were from females]. Twelve non-cataractous lenseswere obtained [age � 69 + 6 years (mean + S.D.), ofwhich three were from females]. Ethical approval wasfrom the Eastern Sydney Area Health Service,Research Ethics Committee (Ref. 90/057) and theUniversity of Wollongong Human Ethics Committee(Ref. HE96/145).

Lens Dissection and Sectioning

Frozen (ethanol-®xed) lenses were cored with a5 mm i.d. chrome-plated brass trephine and approxi-mately 1 mm of both the anterior and posteriorsurface of the nuclear core was removed. Theresulting sample was de®ned as the lens nucleus. Allremaining material was de®ned as the cortex. Notethat the lens capsule and super®cial cortical layerswere not available in the case of the cataractous

600

samples. Paraf®n-embedded formalin-®xed lenseswere sectioned using routine methods.

Ferene-S Assay for Chelatable Fe

Chelatable Fe, which represents a potentially redox-active pool, was analysed using the chromogenicchelator, Ferene-S (Artiss, Vinogradov and Zak, 1981;Eskelinen, Haikonen and RaÈisaÈnen, 1983; Qian, Liuand Eaton, 1998). This assay does not detect tightlybound Fe, e.g. in heme or in the ferritin `core'. Lenseswere homogenized (Garner et al., 2000) in `lysisbuffer' (1 % Triton X-100, 0.5 % Nonidet P40, 0.15 M

NaCl, 10 mM Tris±HCl, pH 7.2 at 228C) that had beenpreviously treated with Chelex resin to removeadventitious transition metals (van Reyk et al.,1995). A 50 ml aliquot of the lens homogenates wasadded to 200 ml of 120 mM sodium ascorbate solutionand incubated at 228C for 5 min in the dark. Onehundred ml of Ferene-S stock solution [prepared byadding 35 mg Ferene-S to 10 ml of 40 % (w/v)ammonium acetate in H2O] was then added, andthe mixture made up to 1 ml with H2O (®nal pH 5.5).After incubation for a further 15 min at 228C in thedark, the reaction mixtures were centrifuged for10 min at 12 000 rpm in an Eppendorf microfugeand the presence of the Ferene-S-Fe2� complex in thesupernatants was determined by measuring absor-

B. GARNER ET AL.

Fe were estimated by comparing A594 values.

Nitrosocysteine/NO� Assay for Redox-Active Cu

The ability of Cu� to destabilize S-nitrosothiolsyielding NO� is well known and the mechanism forthis reaction has been studied in detail (Goren et al.,1996; Stubauer, Giuffre and Sarti, 1999). This facilereaction was used in order to assess the levels ofredox-active Cu in lens homogenates. The rate of NO�release from nitrosocysteine is directly proportional tothe concentration of redox-active Cu in the sampleand the limit of detection for this assay is in the pmolrange (MQ and JWE, unpublished data). For theestimation of lenticular redox-active Cu, lenses werehomogenized in Chelex treated lysis buffer and 40 mladded to 2 ml Chelex treated 100 mM phosphatebuffer (pH 7.4). Freshly prepared nitrosocysteine(Saville, 1958) was then added to give a ®nalconcentration of 250 mM. The release of NO� fromthe mixtures was monitored continuously for3.5 min, with constant stirring at room temperature,using a Clark-type NO� electrode (ISO-NO WorldPrecision Instruments, Sarasota, FL, U.S.A.). Whereindicated, the lens homogenates and buffers were

nitrosocysteine.

Measurement of Lenticular Ferritin by ELISA

Lenses were separated into nucleus and cortex,

lyophilized and subsequently homogenized inChelex treated lysis buffer at a 1 : 2 (w : v) ratio.

sections (too large for microscopy) were recordedusing a high resolution ¯at bed scanner.

T

Homogenates were then diluted in phosphate-bufferedsaline (PBS) containing 0.1 % (v/v) Tween 20 (PBST)and 0.5 % (w/v) BSA and assayed for ferritin contentusing a double antibody sandwich ELISA (Garneret al., 1997). Brie¯y, 96 well microtitre plates (NUNCMaxiSorp, Kamstrup, Denmark) were coated with100 ml rabbit anti-human ferritin polyclonal antibody(DAKO, Glostrup, Denmark) diluted 1/700 in0.01 MM phosphate buffer, 0.145 M NaCl, pH 7.2and incubated at 48C for 18 hr. After washingwith PBST, plates were blocked with 200 ml PBST0.5 % BSA for 30 min at 228C. Lens samples andhuman ferritin standards (Calibrator Cat. no.1355279, Roche Diagnostics) were then added tothe plates in a volume of 100 ml. The range of theferritin standards was from 2 to 200 ng mlÿ1. Afterincubation of samples and standards for 2 hr at 228C,the plates were washed with PBST and incubated for1 hr with peroxidase-conjugated rabbit anti-humanferritin polyclonal antibody (DAKO) diluted 1/10 000in PBST. Finally the plates were washed in PBST and100 ml 3,30,5,50-tetramethylbenzidine (TMB, DAKO)was used as a peroxidase substrate. After acidi®cation,the absorbance of the reaction product was measuredat 450 nm using an Anthos HT automated platereader (Laboratorie Design AB, LidingoÈ, Sweden).

Ferritin that was associated with a crude prep-aration of an insoluble protein fraction of lenses wasalso assessed by direct addition of the peroxidase-conjugated rabbit anti-human ferritin polyclonalantibody to the precipitated proteins. The assay wasperformed in Eppendorf tubes that were previouslyblocked by incubation with 200 ml of 0.5 % BSA/PBSTfor 25 min on a rotary mixing device. Lens homogen-ates were added to the PBST-rinsed tubes and theprecipitated insoluble fraction isolated by centrifu-gation for 10 min at 12 200 g with subsequentaspiration of the supernatant. The conjugated anti-body was added to the resuspended pellet as for theconventional ELISA method except that the tubes wereincubated on a rotary mixing device for 1 hr. Non-bound antibody was removed after centrifugation ofthe mixtures as above and further washing theinsoluble fraction with 1.2 ml PBST in the rotarymixer. The TMB peroxidase substrate was added as forthe ELISA and the absorbance of the acidi®edsupernatants determined after a ®nal centrifugation

REDOX-ACTIVE METALS, FERRITIN AND CATARAC

step. Ferritin concentration was expressed relative to

Cat. no. X903), used as a negative control, did notresult in detectable staining.

the original protein content of the lens homogenates.

Demonstration of Lenticular Redox-active Heavy Metalsby Autometallography and Light Microscopy

Metal localization was demonstrated by lightmicroscopic cytochemistry using a modi®ed sul-phide-silver technique (Zdolsek, Roberg and Brunk,1993). Although this technique does not discriminate

between different heavy metals, since Fe and Cu arethe most abundant redox-active intralenticular metal

ions (Cekic, 1998; Cekic et al., 1999), it can be usedto estimate both redox availability and localization ofthese metals at the cellular level. Formalin-®xed lenseswere sectioned and incubated at 228C in 2 % (w/v)glutaraldehyde in 0.1 M Na-cacodylate±HCl buffer,pH 7.2, with 0.1 M sucrose for 3 hr. Following a shortrinse in double-distilled water, the lenses weresulphidated in 1 % (v/v) ammonium sulphide in70 % (v/v) ethanol, pH 9, at 228C for 15 min (toconvert the metals into their sulphide form), rinsedwith distilled water and developed in a physical,colloid protected, photographic-type developer at268C for 60 min (110 mg Ag-lactate and 850 mghydroquinone were separately prepared in 15 mldouble distilled water, mixed with 60 ml 25 % gumarabic and 10 ml Na-citrate buffer, pH 3.8). Negativecontrols were developed without previous sulphida-tion. After developing, the lenses were quickly rinsedwith distilled water and mounted for light microscopy.Images were photographed directly using a Nikonmicroscope, and Kodak Tri-X pro 400 ASA ®lm orcaptured by video and staining intensity quanti®edusing NIH-Image software. Images of whole lens

601

Demonstration of Lenticular Ferritin Distribution byImmuno¯uorescence Microscopy

Formalin-®xed, paraf®n-embedded lens sectionswere rinsed sequentially in xylol, ethanol, water,PBS, and ®nally exposed to 0.1 % saponin and 0.8 %BSA in PBS for 20 min at 228C. The sections werethen placed in a humidi®er, 30 ml of the same rabbitanti-human ferritin polyclonal antibody as used forthe ELISA studies (1 : 100 in PBS containing 0.1 %saponin and 0.8 % BSA) was added to each sampleand incubated at 48C overnight, rinsed for 2 � 5 minin PBS with 0.1 % saponin and 0.8 % BSA, andincubated for 60 min at room temperature with 30 mlanti-rabbit IgG Texas Red conjugate (1 : 200 in PBSwith 0.1 % saponin and 0.8 % BSA). Following a rinsein PBS and then distilled water, the sections weremounted in VectorShield and examined and photo-graphed in a Nikon photomicroscope using greenexciting light and a red barrier ®lter (G-1B, DM 580Nikon ®lter cube). Non-immune rabbit IgG (DAKO,

Protein Determinations

The protein content of lens homogenates (preparedin lysis buffer as above) was determined using the

bicinchoninic acid method (Sigma, Cat. no. TPRO562)with BSA as a standard.

FIG. 1. Staining for lenticular heavy metals using autometallography. Cataractous (A, B, D and E) and healthy (C and F)control lenses were sectioned and stained using the sulphide-silver method for cytochemical demonstration of Fe and Cu. Directdigital images of the whole lens sections (A±C) and digital microscopic (�25 original magni®cation) images (D±F) are shown.

.

602 B. GARNER ET AL.

Statistical Analysis

Statistical signi®cance was determined by using the

The cataractous lenses show increased heavy metal staining

two-tailed Student's t-test. A P-value 50.05 was

considered signi®cant.

3. Results

Detection of Lenticular Fe by Ferene-S Chelation

Ferene-S is a chromogenic chelator that has beenpreviously used to measure `loosely' bound andtherefore potentially redox-active Fe in biologicalsamples (Artiss et al., 1981; Eskelinen et al., 1983;Qian et al., 1998). We found that on average,homogenized cataractous human lenses contained2.9-fold higher levels of chelatable Fe compared to

(age-matched) non-cataractous control lenses FIG. 2. Quantitative analysis of redox-active metals

determined by autometallography. Digitized microscopic(�25) images of control and cataractous lenses werequantitated using the NIH Image software and the valuesexpressed relative to the average staining intensity observedin the cataractous samples. The `negative control' was acataractous lens that did not undergo the sulphidation step.

(P � 0.002, n � 8 for each lens category).

Demonstration of Lenticular Redox-active Metals byAutometallography

The cytochemical sulphide-silver technique, whencombined with either light or electron microscopy,can be used to localize redox-active metals at acellular and subcellular level, respectively (Zdolseket al., 1993). Only metals that are redox-available(i.e. available for sulphidation) are detected by thistechnique. Staining of lens sections using this methodrevealed that cataractous samples contained morereactive metal than control lenses. This is illustratedin Fig. 1 which shows digital images of whole lenssections [Fig. 1(A)±(C)] as well as images obtained vialow power (�25) microscopic magni®cation[Fig. 1(D)±(F)]. It should be noted that the cataractswere derived from extracapsular surgical extractionand capsules as well as the super®cial layers of cortex

are therefore not present. Fig. 2 illustrates the relativestaining intensity after integration of digitized images

of lens sections. The cataractous samples containedsigni®cantly higher levels of redox-available metals(P � 0.002). Although autometallography is not anFe-speci®c technique, based on the data above thatdemonstrated Ferene-S-detectable Fe was increased incataracts, we suspected that a proportion of the heavystaining in the cataracts was due to higher levels ofredox-active Fe (Cu is another relevant metal ion andits level in lenses was also examined, see below).

The control lenses did not stain strongly usingautometallography. However, a light positive stainingwas noted around the lens capsule. This is clearly seenat higher magni®cation (Fig. 3). The morphology of

Data are mean + S.E.; cataract, n � 5; control, n � 4.

the lens at the anterior surface is also shown[Fig. 3(A)]. The lens capsule, epithelial layer and

FIG. 3. Demonstration of redox-active metals in the lenscapsule. Normal lenses were stained as explained in thelegend to Fig. 1 and examined under �60 originalmagni®cation. A phase contrast image of an unstainedsection of the anterior portion of the lens is shown in (A).Note the distinct layers (from bottom right) are capsule(comprised largely of type IV collagen), the epithelial layer(note the presence of nuclei), and several layers ofdifferentiated ®bre cells (anuclear). A similar lens sectionis shown in (B) after staining by the sulphide-silver method.

FIG. 4. ELISA analysis of lenticular ferritin. Ferritin levelswere quanti®ed in lens homogenates (A) by ELISA and inthe insoluble fraction of lens homogenates (B) by directaddition of HRP-conjugated anti-ferritin to washed proteinprecipitates. Data are means + S.E. for n � 8 (A) and n � 5(B) samples and are expressed relative to total proteincontent of the sample. Black bars represent nucleus and

REDOX-ACTIVE METALS, FERRITIN AND CATARACT 603

®bre cell layers are de®ned by phase contrastmicroscopy [Fig. 3(A) and (B)]. In non-cataractoussamples, stained using the sulphide-silver technique,positivity was localized at and subjacent to the lenscapsule [Fig. 3(B)]. The autometallography resultstogether with the data from the Ferene-S studiesabove, suggest that levels of redox-active Fe are higherin cataractous lenses. We next investigated whetherthe level of the major intracellular Fe-binding protein,

Staining was present only at and subjacent to the capsule(dark regions in top left hand side).

ferritin, was also elevated at sites where Fe was

increased.

Analysis of Lenticular Ferritin Distribution by ELISA andImmunohistochemistry

Ferritin levels were ®rst assessed using ELISAanalysis of the homogenized nuclear and corticalregions of lenses. Control lenses contained similarlevels of ferritin in the nucleus and cortex [Fig. 4(A)].The levels detected, approximately 120 ng mgÿ1

protein, are in agreement with previously publishedvalues (112±310 ng mgÿ1 protein) for human lenses

(Levi et al., 1998). In the cataractous samples, whilesimilar ferritin levels were found in the cortex(approximately 90 ng mgÿ1 protein), signi®cantlylower levels (P � 0.001) were detected in the nucleus(approximately 20 ng mgÿ1 protein). This dramaticdifference was not easily explained. Since an increasedproportion of lens proteins are aggregated andcrosslinked in the nucleus of age-related cataractouslenses, it was possible that a proportion of the ferritinin the detergent-solubilized samples was still `trapped'within the insoluble fraction. To address this, theinsoluble fraction of lens homogenates was washedfurther in lysis buffer to remove ferritin which wouldbe immunodetectable by the conventional ELISAmethod, and the HRP-conjugated anti human ferritinantibody was added directly to the insoluble proteinsuspension. After further washing the insolubleprecipitate to remove unbound antibody, the amountof immunodetectable ferritin present was measured(see Materials and Methods section for further details).The amounts of ferritin detected in the insolublefraction of the cataractous nuclei and cortices werefour-fold (P � 0.031) and two-fold (P � 0.044)higher than in the controls, respectively [Fig. 4(B)].This indicates that part of the explanation for thelower levels of ferritin detected in the cataractous

white bars cortex.

nuclei by the conventional ELISA is due to theassociation of ferritin with an insoluble fraction

FIG. 5. Immunohistochemical demonstration of ferritin inthe human lens. Phase contrast (A and C) and ¯uorescence(B and D) images of the anterior (A and B) and posterior (Cand D) regions of a non-cataractous lens are shown. Ferritinis indicated by strong ¯uorescence in the anterior epithelium(B) and by diffuse and scattered staining proximal to theposterior surface and throughout the deeper regions. �60original magni®cation. Diffuse staining for ferritin was alsopresent in the lens nuclei (data not shown).

604

(i.e. a fraction which would normally be washed outof the 96 well plate after the initial sample incubationstep).

Immunohistochemical studies of ferritin distri-bution in normal lenses showed that ferritin waspresent at the highest levels in the epithelial layer atthe anterior surface of the lens with a weaker positivestaining fairly evenly distributed throughout thecortex and to the posterior (Fig. 5). Weak positivestaining for ferritin was also evident throughoutthe nuclear region of the control and cataractouslenses (not shown), consistent with the ELISA datapresented above.

Lack of Evidence for Redox-active Cu in Lenses

Since reduced Cu effectively destabilizes nitro-sothiols to yield NO� (Goren et al., 1996), we usedthis reaction to probe for redox-active Cu in humanlens homogenates. In these experiments, nitrosocys-teine was added to lens homogenates and the rate ofNO� release was quanti®ed and used as an index of Cuavailability. We found no evidence that lens hom-ogenates from either cataractous or control subjectscatalyse signi®cant NO� production from nitrosocys-teine (i.e. NO� release was at or below the levelsobserved in buffer alone). In fact, when 50 nM Cu wasadded to the lens homogenates, we found that the rate

B. GARNER ET AL.

of NO� release was signi®cantly inhibited compared tosamples that contained only buffer with added Cu. As

FIG. 6. Redox inactivation of exogenously added Cu bylens homogenates. Cataractous (n � 8, line iii) and control(n � 2, line ii) lens homogenates were incubated with50 nM Cu for 5 min and the ability of the Cu/lens mixture todegrade nitrosocysteine (to yield NO�) was assessed asdescribed in the Materials and Methods section. The uppertrace (line i) represents Cu-catalysed NO� release in buffercontaining 50 nM Cu, the bottom trace (line iv) representsspontaneous NO� release from buffer alone. After approxi-mately 30 sec, the ability of Cu to degrade nitrosocysteine issigni®cantly inhibited by control lens homogenates. Thecataractous samples were more effective in their capacity toinhibit Cu-catalysed NO� release (i.e. lower the rate of NO�release). Data are representative of two independentexperiments.

T

shown in Fig. 6, both cataractous and control lenshomogenates inhibited the NO� release caused by theadded Cu. In addition, the cataractous samplesappeared to be approximately twice as ef®cient inthis activity (assessed at the 2 min time point). Thesedata suggest that Cu present within either normal orcataractous lenses (Nath, Srivastava and Singh, 1969;Cook and McGahan, 1986; Cekic, 1998), may not be

REDOX-ACTIVE METALS, FERRITIN AND CATARAC

freely available to participate in aqueous phase redox

reactions.

4. Discussion

Previous studies indicate that total lenticular Feand Cu concentrations can be elevated in age-relatedcataract (Lakomaa and Eklund, 1978; Cook andMcGahan, 1986; Garland, 1990; Cekic, 1998; Cekicet al., 1999; Garner et al., 2000). In three of thesestudies (Lakomaa and Eklund, 1978; Cekic et al.,1999; Garner et al., 2000) that used very similarmethodology [depending on the method used and thespecies studied, lenticular Fe contents have beenreported to vary by orders of magnitude (McGahan,1992)], Fe was measured by atomic absorptionspectrometry after the lenses were dried and dissolvedin nitric acid and found to be present in normal lensesat levels of 23, 17 and 7 mg gÿ1 dry weight,respectively. These studies also found that Fe levelswere, on average, 1.7±3.1-fold higher in cataractouslenses. Although a high degree of intersamplevariation was observed in the levels of Fe detected inthe cataractous samples (standard deviations werereported at 55 (n � 14), 58 (n � 37) and 77 %(n � 7) of the total Fe values, respectively), the datado indicate that Fe levels are increased in thecataractous lens. This does not, however, necessarilyindicate that a higher content of `redox-available'metal is present. Using electron paramagnetic reson-ance spectroscopy, cataractous lenses were shown tohave a greater capacity to stimulate HO� productionwhen exposed to H2O2 as compared to non-catar-actous lenses (Garner et al., 2000). This raised thepossibility that higher levels of redox-active metalscould indeed be present in the cataractous lenses andthis was further investigated in the present work.

Using Ferene-S and autometallography asindicators, potentially redox-available Fe was foundto be present at higher levels in cataractous samples.Under the conditions employed in our study, theFerene-S-available Fe is not realistically expected torepresent a `free' Fe pool. In fact, the free Fe pool ofmost cells is extremely low and probably onlypresent in compartments where Fe binding proteinsare releasing Fe or undergoing degradation, i.e. trans-ferrin in endosomes (Breuer, Greenberg andCabantchick, 1997) and ferritin in lysosomes (Brunand Brunk, 1970). However, the source of chelatable

Fe in the lens, where most of the cells do not containthis acidic vacuolar apparatus, remains to be

determined. The major intracellular Fe pools wouldnormally be expected to include: (1) heme Fe; (2) Fe inthe active sites of non-heme enzymes (e.g. Fe : Sclusters); (3) Fe in the ferritin core; and (4)`delocalized' Fe which is anomalously bound toproteins [in either a native or glycated state (Qianet al., 1998)] or present in degraded Fe-storageproteins (ferritin in particular). The Ferene-S-chela-table Fe in the lens is most likely derived from pools (2)and (4). Our suggestion that this may re¯ect a redox-active pool is supported by the fact that the ascorbateused in the assay procedure is able to reduce the Fepresent thus generating a ligand for Ferene-S. It ispossible that a proportion of the Fe detected is derivedfrom transferrin as this protein has been shown to betransported into the lens under certain conditions(McGahan, 1992). The chelatable Fe pool may also bederived from modi®ed ferritin (see below). Regardlessof the source, the data clearly show that cataractouslenses contain higher levels of a chelatable andtherefore potentially redox-available Fe pool. Thispool is predicted to result in higher levels of HO�production when H2O2 is formed in the lens andthereby likely explains the increased generation ofHO� when lenses were exposed to H2O2 (Garner et al.,2000). Interestingly, cigarette smoke-induced cataract(in rats) is associated with increased lens Fe (Avunduket al., 1999). Furthermore, this cataract is inhibitedby the Fe-chelator, deferoxamine (Avunduk et al.,1999). It is possible that a similar therapeutic strategymay retard the progress of age-related cataract inhumans, especially where cigarette smoke exposure isalso implicated.

We found no evidence for the presence of redox-active Cu in the lens. In fact, both control andcataractous lenses effectively chelated and redox-inactivated this metal ion. It is possible that thecrystallins can bind Cu (perhaps via interaction withhistidine residues) and thereby decrease its redox-availability towards the nitrosocysteine which wasused in our study. The chelation of Cu by lens proteinsis consistent with previous proposals (Cook andBentley, 1986; Ortwerth and James, 1999). Theincreased ability of the cataractous samples to bindCu might be due to Cu binding to newly exposed siteson modi®ed crystallins or to advanced glycation endproducts (AGEs) which are formed on crystallins (afterreaction of lysine residues with ascorbate oxidationproducts for example) during ageing and cataractogen-esis (Dunn et al., 1989, 1990; Baynes, 1994; Tessier,Obrenovich and Monnier, 1999). In agreement withthis, others have found increased Cu binding to AGE-modi®ed crystallins derived from cataractous lenses(Saxena, Saxena and Monnier, 1998). It appears,though, that ascorbate autoxidation may still bestimulated by such complexes (Saxena et al., 1998).

Age-related glycation of the collagen-rich lens

605

capsule (Cohen and Yu-Wu, 1983; Bailey et al.,1993) may also result in Cu binding to the resulting

or UV-A exposure but these issues will require furtherstudy.

Cohn, Sr Barbara Roberts and The Sydney Eye HospitalLions Eye Bank are thanked for collection of lenses.

protein-AGEs (Qian et al., 1998). This would beconsistent with our present data demonstrating faintstaining for metal localization at the lens capsule(using autometallography). Given that H2O2 can beproduced in the aqueous humour (Eaton, 1997;Spector, Ma and Wang, 1998), the capsule may beparticularly vulnerable to metal-catalysed HO�-mediated damage. We are not aware of any studythat has assessed this directly.

Ferritin was distributed throughout the entire lens,with the highest concentrations in the epithelium. Inthe cataractous lenses, nuclear ferritin levels werelower, possibly due to an increased pool of aggregatedferritin within the insoluble crystallin fraction.However, while the ferritin levels of the cortex weresimilar in both healthy and cataractous lenses, theadditional ferritin associated with the insolublefraction of the cataractous nuclei was still belowthat detected in the healthy lens nuclei (approxi-mately 45 vs 125 ng mgÿ1 protein). It is possible thatsome ferritin remained trapped in regions of theinsoluble fraction that are not accessible to theantibody. An additional factor contributing to thelower levels of ferritin detected could be that oxidativedamage, derivitization with AGEs or crosslinking withother proteins decreased the immunogenicity of thecomplexed nuclear ferritin. Given that general proteinturnover in the lens nucleus is very low or absent(Bloemendal, 1977), we speculate that accumulatedpost-translational modi®cation(s) of a proportion ofthe ferritin present in the cataractous nuclei maycompromise Fe binding. In addition, it is known thatUV-A radiation can release Fe2� from ferritin withoutdamaging the protein (Aubailly, Santus and Salmon,1991). In particular, 365 nm light was found torelease at least 80 Fe2� ions per ferritin molecule in aprocess that was oxygen-independent (Aubailly et al.,1991). Since light of this wavelength is thought toreach the lens (Weale, 1988), we propose that UV-Aradiation could be cataractogenic due to its ability torelease Fe from lenticular ferritin and therebyfacilitate HO� generation.

A cataract-hyperferritinaemia syndrome hasrecently been identi®ed in which patients havemarked over production of L-ferritin (which lacksferroxidase activity, Levi et al., 1994) and a highincidence of premature cataract (Cazzola et al., 1997).It is unclear how the speci®c over production of L-ferritin causes cataract. Perhaps the resulting ferritincomplex (24 mer) is de®cient in ferroxidase activity,leading to an inability to convert Fe2� to the less toxicFe3� form (and less effectively form an Fe `core').Alternatively, there is evidence that ferritin reachessuch high local concentrations in the lens that itbecomes crystallized (Mumford et al., 2000), possiblyforming a nidus for crystallin precipitation.

In conclusion, the present studies provide evidence

606

for the presence of increased levels of redox-availabletransition metals in age-related cataractous lenses.

Because Cu can be redox-inactivated by lens proteins,it seems more likely that Fe is potentially cataracto-genic. The source of the increased redox-activelenticular Fe may be related to ferritin degradation

B. GARNER ET AL.

Acknowledgements

This work was supported by LinkoÈping University HospitalGuest Researcher Grants (BG, MQ and JWE), the AustralianNational Health and Medical Research Council (RJWT,grant no. 980495) and the Swedish Medical ResearchCouncil (Prof. Ulf T. Brunk, grant no. 4481). Prof. Brunk isthanked for comments on the manuscript. Dr Geoffrey

References

Artiss, J. D., Vinogradov, S. and Zak, B. (1981). Spectro-photometric study of several sensitive reagents forserum iron. Clin. Biochem. 14, 311±5.

Aubailly, M., Santus, R. and Salmon, S. (1991). Ferrous ionrelease from ferritin by ultraviolet-A radiations. Photo-chem. Photobiol. 54, 769±73.

Avunduk, A. M., Yardimci, S., Avunduk, M. C. and Kurnaz,L. (1999). Cataractous changes in rat lens followingcigarette smoke exposure is prevented by parenteraldeferoxamine therapy. Arch. Ophthalmol. 117,1368±72.

Bailey, A. J., Sims, T. J., Avery, N. C. and Miles, C. A. (1993).Chemistry of collagen cross-links: glucose-mediatedcovalent cross-linking of type-IV collagen in lenscapsules. Biochem. J. 296, 489±96.

Balla, G., Jacob, H. S., Balla, J., Rosenberg, M., Nath, K.,Apple, F., Eaton, J. W. and Vercellotti, G. M. (1992).Ferritin: a cytoprotective antioxidant stratagem ofendothelium. J. Biol. Chem. 267, 18148±53.

Baynes, J. W. (1994). Role of metal-catalyzed autoxidationin Maillard reaction damage to proteins in vivo. RedoxReport 1, 31±5.

Bloemendal, H. (1977). The vertebrate eye lens. Science 197,127±38.

Breuer, W., Greenberg, E. and Cabantchick, Z. I. (1997).Newly delivered transferrin iron and oxidative cellinjury. FEBS Letters 403, 213±9.

Brun, A. and Brunk, U. T. (1970). Histochemical indicationsfor lysosomal localization of heavy metals in normal ratbrain and liver. J. Histochem. Cytochem. 18, 820±7.

Cazzola, M., Bergamaschi, G., Tonon, L., Arbustini, E.,Grasso, M., Vercesi, E., Barosi, G., Bianchi, P. E., Cairo,G. and Arosio, P. (1997). Hereditary hyperferritinemia-cataract syndrome: relationship between phenotypesand speci®c mutations in the iron-responsive element offerritin light-chain mRNA. Blood 90, 814±21.

Cekic, O. (1998). Copper, lead, cadmium and calcium incataractous lenses. Ophthalmic Res. 30, 49±53.

Cekic, O., Bardak, Y., Totan, Y., Kavakli, S., Akyol, O.,Ozdemir, O. and Karel, F. (1999). Nickel, chromium,manganese, iron and aluminium levels in humancataractous and normal lenses. Ophthalmic Res. 31,332±6.

Cohen, M. P. and Yu-Wu, V. (1983). Age-related changes in

non-enzymatic glycosylation of human basement mem-branes. Exp. Gerontol. 18, 461±9.

T

Cook, C. S. and Bentley, P. J. (1986). Copper exchanges andtoxicity in the rabbit lens in vitro. Exp. Eye Res. 42,107±16.

Cook, C. S. and McGahan, M. C. (1986). Copper concen-tration in cornea, iris, normal, and cataractous lensesand intraocular ¯uid of vertebrates. Curr. Eye Res. 5,69±77.

Dunn, J. A., Ahmed, M. U., Murtiashaw, M. H., Richardson,J. M., Walla, M. D., Thorpe, S. R. and Baynes, J. W.(1990). Reaction of ascorbate with lysine and proteinunder autoxidizing conditions: formation of N-epsilon-(carboxymethyl)lysine by reaction between lysine andproducts of autoxidation of ascorbate. Biochemistry 29,10964±70.

Dunn, J. A., Patrick, J. S., Thorpe, S. R. and Baynes, J. W.(1989). Oxidation of glycated proteins: age-dependentaccumulation of N-epsilon-(carboxymethyl)lysine inlens proteins. Biochemistry 28, 9464±946.

Eaton, J. W. (1997). Is the lens canned?. Free Radic. Biol. Med.11, 201±13.

Eckhert, C. D. (1983). Elemental concentrations in oculartissues of various species. Exp. Eye Res. 37, 639±47.

Eskelinen, S., Haikonen, M. and RaÈisaÈnen, S. (1983).Ferene-S as the chromogen for serum iron determi-nations. Scand. J. Clin. Lab. Invest. 43, 453±5.

Fu, S., Dean, R. T., Southan, M. and Truscott, R. J. W.(1998). The hydroxyl radical in lens nuclear catar-actogenesis. J. Biol. Chem. 273, 28603±9.

Garland, D. (1990). Role of site-speci®c, metal-catalyzedoxidation in lens aging and cataract: a hypothesis. Exp.Eye Res. 50, 677±82.

Garner, B., Davies, M. J. and Truscott, R. J. W. (2000).Formation of hydroxyl radicals in the human lens isrelated to the severity of nuclear cataract. Exp. Eye Res.70, 81±8.

Garner, B., Li, W., Roberg, K. and Brunk, U. T. (1997). Onthe cytoprotective role of ferritin in macrophages and itsability to enhance lysosomal stability. Free Rad. Res. 27,487±500.

Garner, B., Roberg, K. and Brunk, U. T. (1998). Endogenousferritin protects cells with iron-laden lysosomes againstoxidative stress. Free Rad. Res. 29, 103±14.

Garner, M. H. and Spector, A. (1980). Selective oxidation ofcysteine and methionine in normal and senile catar-actous lenses. Proc. Natl. Acad. Sci. U.S.A. 77, 1274±7.

Goren, A. C. F., Schrammel, A., Schmidt, K. and Mayer, B.(1996). Decomposition of S-nitrosoglutathione in thepresence of copper ions and glutathione. Arch. Biochem.Biophys. 330, 219±28.

Harding, J. J. (1991). Cataract. Biochemistry, Epidemiologyand Pharmacology. Chapman and Hall: London, U.K.

Kupfer, C., Underwood, B. and Gillen, T. (1994). Leadingcauses of visual impairment world wide. In Principlesand Practice of Ophthalmology Basic Science. (Albert, D. M.and Jakobiec, F. A., Eds.). W. B. Saunders and Co.:Philadelphia, U.S.A.

Lakomaa, E. L. and Eklund, P. (1978). Trace elementanalysis of human cataractous lenses by neutronactivation analysis and atomic absorbtion spectrometrywith special reference to pseudo-exfoliation of the lenscapsule. Ophthalmic Res. 10, 302±6.

REDOX-ACTIVE METALS, FERRITIN AND CATARAC

Analysis of ferritins in lymphoblastoid cell lines and inthe lens of subjects with hereditary hyperferritinemia-cataract syndrome. Blood 39, 4180±7.

Levi, S., Santambrogio, P., Cozzi, A., Rovida, E., Corsi, B.,Tamborini, E. and Spada, S. (1994). The role of the L-chain in ferritin iron incorporation. Studies of homoand heteropolymers. J. Mol. Biol. 328, 649±54.

McGahan, M. C. (1992). Does the lens serve as a `sink' foriron during ocular in¯ammation. Exp. Eye Res. 54,525±30.

Mumford, A. D., Cree, I. A., Arnold, J. D., Hagan, M. C.,Rixon, K. C. and Harding, J. J. (. (2000). The lens inhereditary hyperferritinaemia cataract syndrome con-tains crystalline deposits of L-ferritin. Brit. J. Ophthalmol.84, 697±700.

Nath, R., Srivastava, S. K. and Singh, K. (1969). Accumu-lation of copper and inhibition of lactate dehydrogenaseactivity in human senile cataractous lens. Exp. Biology7, 25±6.

Ortwerth, B. J. and James, H. L. (1999). Lens proteins blockthe copper-mediated formation of reactive oxygenspecies during glycation reactions in vitro. Biochem.Biophys. Res. Commun. 259, 706±10.

Pirie, A. (1968). Color and solubility of the proteins ofhuman cataract. Invest. Ophthalmol. 7, 634±42.

Qian, M., Liu, M. and Eaton, J. W. (1998). Transition metalsbind to glycated proteins forming redox active `glyco-chelates': implications for the pathogenesis of certaindiabetic complications. Biochem. Biophys. Res. Comm.250, 385±9.

Saville, B. (1958). A scheme for colorimetric determinationof microgram amounts of thiols. Analyst 83, 670±2.

Saxena, P., Saxena, A. K. and Monnier, V. M. (1998).Evidence of transition metal-catalysed oxidation ofascorbate by advanced glycation end products in thecataractous human lens [abstract]. Invest. Ophthalmol.Vis. Sci. 39, S-891.

Spector, A., Ma, W. and Wang, R.-R. (1998). The aqueoushumour is capable of generating and degrading H2O2.Invest. Ophthalmol. Vis. Sci. 39, 1188±97.

Stubauer, G., GiuffreÂ, A. and Sarti, P. (1999). Mechanism ofS-nitrosothiol formation and degradation mediated bycopper ions. J. Biol. Chem. 274, 28128±33.

Tessier, F., Obrenovich, M. and Monnier, V. M. (1999).Structure and mechanism of formation of human lens¯uorophore LM-1. J. Biol. Chem. 274, 20796±804.

Truscott, R. J. W. and Augusteyn, R. C. (1977). Oxidativechanges in human lens proteins during senile nuclearcataract formation. Biochim. Biophys. Acta 492, 43±52.

van Reyk, D. M., Brown, A. J., Jessup, W. and Dean, R. T.(1995). Batch-to-batch variation of Chelex-100 con-founds metal-catalysed oxidation. Leaching of inhibi-tory compounds from a batch of Chelex-100 and theirremoval by a pre-washing procedure. Free Rad. Res. 23,533±5.

Weale, R. A. (1988). Age and transmittance of the humancrystalline lens. J. Physiol. 395, 577±87.

Zdolsek, J., Roberg, K. and Brunk, U. T. (1993). Visualisationof iron in cultured macrophages: a cytochemical light

607

Levi, S., Girelli, D., Perrone, F., Pasti, M., Beaumont, C.,Corrocher, R., Albertini, A. and Arosio, P. (1998).

and electron microscopic study. Free Radic. Biol. Med.15, 1±11.