You are on page 1of 14

Exp. Eye Res. (2002) 74, 141154 doi:10.1006/exer.2002.1112, available online at http://www.idealibrary.

com on

Drusen are Cold Spots for Proteolysis: Expression of Matrix Metalloproteinases and Their Tissue Inhibitor Proteins in Age-related Macular Degeneration
SER G IU T. LEU a, S U CH I TRA BATN I a, M O N T E J . R A D E K E b, LI N CO L N V. J O H N S O N b, D O N H . A N D E R S O N b A N D D E N N I S O . C L E G G ab* Department of Molecular, Cellular and Development Biology, University of California, Santa Barbara, CA 93106, U.S.A. and bCenter for the Study of Macular Degeneration, Neuroscience Research Institute, University of California, Santa Barbara, CA 93106, U.S.A. (Received St. Louis 13 July 2001 and accepted in revised form 5 October 2001)
Drusen are abnormal extracellular matrix deposits characteristic of age-related macular degeneration (AMD), a leading cause of blindness in the aging human population. The mechanisms underlying drusen formation are not well characterized. The purpose of this study was to examine the expression of matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) in drusen, and in the surrounding cells and tissue. To assess the extent of MMP and TIMP expression by retinal pigment epithelial (RPE) cells, cDNA arrays were screened with probes generated from cultured human RPE cells. The distribution of MMP-1, -2 and -3 and TIMP-1, -2, -3 and -4 was determined using immunohistochemistry in human RPE choroid from donor eyes with and without a clinical history of AMD. Gelatinase activity was assessed in unxed frozen sections using in situ zymography. In cultured RPE cells, expression of 10 MMP and all four known TIMP mRNAs was detected. MMP immunoreactivity was widespread in the RPE choroid, but was absent from the interior of drusen. TIMP-3, but not other TIMPs, was detected in the drusen interior. Likewise, metal ion dependent gelatinase activity could be detected in RPE choroid, but not in drusen. These results show that, while metalloproteinase activity is widespread throughout the RPE choroid, drusen are cold spots for proteolysis. The data lead to the speculation that high TIMP-3 concentrations within drusen could inhibit MMPs and as a result slow the proteolytic # 2002 Elsevier Science Ltd degradation of these deposits. Key words: matrix metalloproteinase; tissue inhibitor of metalloproteinase; retinal pigment epithelium; drusen; age-related macular degeneration; collagenase; gelatinase; in situ zymography; neurodegeneration; retina; choroids; Bruch's membrane.
a

1. Introduction Age-related molecular degeneration (AMD) is the leading cause of blindness in the United States (Klein, Klein and Linton, 1992; Egan and Seddon, 1994; Penfold et al., 2001). Loss of vision in AMD is a result of degeneration of rods and cones in the macular region of the central retina, which is responsible for high acuity vision. Death of the photoreceptors appears to be a consequence of degeneration of neighboring retinal pigment epithelial (RPE) cells. The loss in vision may be local (`dry' type or non-exudative AMD) or may occur over a larger area (`wet' type or exudative AMD) and involve neovascularization of the choroid and of Bruch's membrane. This form of the disease accounts for about 8090 % of the blindness observed in patients with AMD, although `dry' type AMD is more common. AMD is a multifactorial disease with both environmental (Egan and Seddon, 1994; Seddon
* Address correspondence to: Dennis O. Clegg, Center for the Study of Macular Degeneration, Neuroscience Research Institute and Department of Molecular, Cellular and Developmental Biology, University of California, Santa Barbara, CA 93106, U.S.A. E-mail: clegg@lifesci.ucsb.edu

et al., 1996) and genetic factors (Heiba et al., 1994; Seddon, Ajani and Mitchell, et al., 1997) contributing to the disease. However, what factor(s) actually causes the transition from normal aging to disease is not understood. The genetic analysis of the disorder is further compounded by the commonly observed clinical heterogeneity. Although mutations in the photoreceptor specic gene ABCR (ATP binding Cassette Transporter; ABCA4) have been recently detected in patients with AMD (Allikmets et al., 1997), this nding is controversial, and at best accounts for only a small percent of AMD cases studied (Allikmets, 2000; Rivera et al., 2000; Webster et al., 2001). The formation of drusen, abnormal deposits in the extracellular matrix (ECM), is an important hallmark of AMD (Abdelsalam, Del Priore and Zarbin, 1999; Penfold et al., 2001). Typically, drusen lie between the RPE basement membrane and the inner collagenous layer of Bruch's membrane. Drusen have been reported to be immunoreactive to antibodies specic for a variety of ECM molecules such as collagen, laminin, bronectin, vitronectin and heparan sulfate
# 2002 Elsevier Science Ltd

0014-4835/02/01014114 $35.00/0

142

S . T. L E U E T A L .

(Newsome et al., 1987; Mullins et al., 1997; Johnson et al., 2000). Some investigators have hypothesized that drusen may result from the failure to dispose RPE derived ECM molecules, or may be due in part to an inammatory immune reaction (Johnson et al., 2000; Penfold et al., 2001). However, the precise relationship between drusen and AMD is unclear, since drusen are also present in many older individuals with no clinically signicant visual loss. Nevertheless, drusen formation appears to precede the occurrence of visual symptoms in AMD eyes and is therefore an important aspect of the disease. The molecular events that underlie drusen formation are not well established. In general, steady-state levels of ECM components are thought to be regulated by matrix metalloproteinases (MMPs) and their naturally occurring proteinaceous inhibitors (Vu and Werb, 2000). MMPs constitute a family of about 20 Zn2 dependent proteases that cleave ECM proteins at neutral pH (Parsons et al., 1997; Sethi et al., 2000). The proteins are broadly classied into four subgroups based on their substrate specicity: the collagenases (MMP-1, 8 and -13) that cleave brillar collagens (types I, II, III, VII, VIII, X); the gelatinases (MMP-2 and -9) that cleave collagen (types IV, V, VII, X, XI), gelatins, and elastin; the stromelysins [MMP-3, -10 and -11 and MMP-7 (matrilysin)] that cleave a variety of substrates, including gelatinases; and the fourth miscellaneous group consisting of MMP-12 and membrane-type MMPs (MMP-14, -15, -16, -17) (Sethi et al., 2000). All MMPs share amino acid sequences and are secreted as proenzymes that require proteolytic activation. The importance of MMPs is evident from their roles in tissue remodeling during development and in pathogenesis (Parsons et al., 1997). The activity of MMPs is tightly controlled by tissue inhibitors of metalloproteinases (TIMPs). TIMPs comprise a family of four distinct proteins: TIMP-1, -2, -3 and -4 that differ in their sequence, structure, biochemical properties and in vivo and in vitro expression (Gomez et al., 1997; Brew, Dinakarapandian and Nagase, 2000;). In addition, TIMPs have been shown to exhibit differing afnities for pro-MMPs (Bigg et al., 1997). Imbalances in the ratio and extracellular activities of MMP and TIMPs have been linked to pathological tissue destruction in cancer, arthritis, cardiovascular and vitreoretinal diseases (Liotta, Steeg and Stetler-Stevenson, 1991; Brinckerhoff, 1992; Armstrong et al., 1994; Sethi et al., 2000). Recently, the importance of TIMPs in controlling retinal tissue organization was highlighted by the discovery of TIMP-3 mutations that lead to Sorsby's fundus dystrophy (SFD; Weber et al., 1994), an autosomal dominant macular disorder characterized by abnormal deposits in the inner portion of Bruch's membrane and neovascularization, similar to AMD. These mutations give rise to active inhibitors that can form dimers via unpaired cysteine residues (Langton

et al., 2000). This may inhibit TIMP-3 turnover, since SFD patients show increased deposition of TIMP-3 (Fariss et al., 1998; Chong et al., 2000). Although mutations in TIMP-3 have not been detected in AMD (De La Paz et al., 1997), it is possible that defects in balance of TIMP and MMP activity could lead to the accumulation of drusen and progression to AMD. Several studies have investigated the expression of MMPs and TIMPs in the human eye. TIMP-3 mRNA has been localized to the RPE using in situ hybridization (Ruiz, Brett and Bok, 1996), and increases in the TIMP-3 expression have been associated with Simplex Retinitis Pigmentosa, a degenerative ocular disease (Jomary et al., 1997). TIMP-3 immunoreactivity was detected in RPE, Bruch's membrane and drusen of aged eyes (Vranka et al., 1996; Fariss et al., 1997; Jomary et al., 1997), and the expression was shown to increase with age (Fariss et al., 1997, Kamei and Hollyeld, 1999). In contrast, TIMP-2 was expressed in lower levels and TIMP-1 was not detected in normal retina/choroid. (Vranka et al., 1996). Human RPE cells have been shown to secrete MMP-1, -2, -9, 3, and TIMP-1 (Alexander et al., 1990). In a separate study, RPE cells were shown to secrete MMP-2, TIMP1, -2 and -3 (Padgett et al., 1997). Moreover, the secretion of MMPs and TIMPs was dependent on the passage number of RPE cells in culture and composition of the substratum. MMP -1,-2,-3 and -9 have been detected in isolated human Bruch's membrane and choroid, and levels of MMP-2 and -9 were shown to increase with age (Guo et al., 1999). Relatively less is known about how the expression of MMPs and TIMPs is related to AMD. Using in situ hybridization, MMP-2, -9 and TIMP-1, -2 and -3 mRNAs have been detected in subfoveal brovascular membranes of choroidal neovascularization, and TIMP-3 message was also detected in RPE (Steen et al., 1998). MMP-2 mRNA increased after experimentally induced choroidal neovascularization in the rat (Kvanta et al., 2000), and increased MMP-2 has been observed in the interphotoreceptor matrix of AMD eyes (Plantner, Jiang and Smine, 1998). TIMP-3 protein has also been detected in drusen and Bruch's membrane of eyes with AMD at higher levels than controls (Jomary et al., 1997, Kamei and Hollyeld, 1999). Whether other TIMP or MMPs are present in drusen has not been investigated. In this study, cDNA arrays, immunouorescence, and in situ zymography are utilized to investigate the expression of MMPs and TIMPs in cultured RPE, and their distribution in RPE choroid and drusen in donor eyes with and without a history of AMD. Although MMP distribution was widespread, they could not be detected in the drusen interior, which displayed abundant TIMP-3 immunoreactivity. The results support the hypothesis that drusen are `cold spots' for proteolysis due to high concentrations of TIMP-3.

MMPS AND TIMPS IN DRUSEN

143

2. Materials and Methods Human Retinal Pigment Epithelial Cells and Donor Eyes Human fetal RPE cells were cultured in tissue culture dishes as described by Lin and Clegg (1998) and harvested before reaching conuence (CSMD cells), or were subconuent cultures grown on gelatin microcarriers (cells provided by Titan Pharmaceuticals, South San Francisco, CA, U.S.A.). Human eyes were obtained from the Doheney Eye Bank, Santa Barbara, CA, U.S.A.; The Lions Eye Bank, Portland, OR, U.S.A.; and from Drs Jay and Maureen Neitz at the Department of Cell Biology and Anatomy and The Eye Institute, Medical College of Wisconsin, Milwaukee, WI, U.S.A.. For immunohistochemistry, globes were xed using 4 % paraformaldehyde prepared in PBS, pH 7.4, within 6 hr post mortem and dissected to remove the vitreous. Subsequently, they were cut into quadrants and stored in 0.5 % paraformaldehyde at 48C. Dissected eye quadrants consist of sclera, choroid, RPE, with or without the retina. For in situ zymography, eyes were embedded in OCT and frozen within 5 hr post mortem. Table I lists the human donors of RPE choroid used in this study. Donor tissues were selected based on their age and history of AMD. cDNA Array Analysis The human cytokine expression array (375 cDNAs, GA001, R & D Systems, Minneapolis, MN, U.S.A.) was employed in these studies. This cDNA array consists of DNA coding regions from cytokines, chemokines, interleukins, enzymes, and other immunoregulatory factors and receptors spotted in duplicate on charged nylon membrane (www.rndsystems.com). To make probes, total RNA was extracted from cultured RPE cells by exposure to `RLT' lysis buffer (Qiagen, Santa Clara, CA, U.S.A.) and homogenized using a QIA shredder (Qiagen, Santa Clara, CA, U.S.A.). RNA was puried using a silica gel-based spin column that eliminates RNAs 5200 bases (RNeasy Mini Kit;

Qiagen, Santa Clara, CA, U.S.A.). Potentially contaminating DNA was eliminated by exposure to RNase free DNase, followed by repurication of RNA. RNA was quantied by absorbance readings at 260 nm. Five mg of total RNA was primed with array specic primers (i.e. a mixture of primers including one set corresponding to each gene sequence present in the array) and cDNA was generated using MMLV RVTase using the Strip-Ez RT kit (Ambion, Austin, TX, U.S.A.). 33P-labeled cDNA was then separated from the unincorporated nucleotides using a spin gel ltration column (BioGel P6DG, BioRad) yielding 2040 106 cpm of probe (4109 cpm mg 1). Prior to incubation with the labeled probe, the arrays were pre-incubated (428C, overnight) with 5 ml of ULTRAhyb (Ambion, Austin, TX, U.S.A.) containing 5 mg ml 1 denatured Cot-1 DNA (Life Technologies) to block binding to repeated elements. After the prehybridization, the probe was denatured by heating to 958C for 5 min, immediately diluted with 2 ml of ULTRAhyb buffer at 688C, and added directly to the prehybridization solution. The array was then hybridized with the probe for 1620 hr at 428C, and washed once with 150 ml of 2 SSPE (300 mM NaCl, 17 mM NaH2PO4, 2 mM EDTA; pH 7.4) containing 1 % SDS at room temperature, once with 150 ml of 2 SSPE, 1 % SDS at 688C and twice with 150 ml of 0.2 SSPE, 1 % SDS at 688C. There was an insignicant change in the relative hybridization intensity of individual spots at rinse temperatures from 68 to 728C indicating that signals remaining after the nal wash are specic. After washing, the arrays were placed on lter paper presoaked with 0.2 SSPE, 1 % SDS, sealed in plastic wrap and data were collected for 14 days using a phosphor storage screen and imaging on a Molecular Dynamics Storm 640 phosphorimager. The resulting 16-bit grayscale image was imported into an image analysis program (LabWorks, UVP, Inc. Upland, CA, USA) and the intensity of each array spot was determined after background subtraction. The data

TABLE I Human donor eyes used in this study


Sample 1 2 3 4 5 6 7 8 9 10 11 12 Donor WI-2-97 WI-4-97 WI-5-97 WI-6-97 O-6-00 O-10-00 SB-1-97 SB-2-97 SB-4-97 SB-5-97 SB-6-97 SLO-95-96 Age 86 84 82 82 87 83 21 75 66 89 81 42 Sex M F F F M F M M M M F F Drusen Present Present Present Present Present Present Not detected Present Not detected Present Present Not detected Clinical history AMD AMD AMD AMD AMD AMD No history of AMD No history of AMD Unknown ocular history No history of AMD No history of AMD No history of AMD

144

S . T. L E U E T A L .

were then imported into an Excel spreadsheet containing the gene identiers and the expression levels for each gene were determined relative to the total hybridization signal. Hybridization to three Escherichia coli genes (b0658, b1444 and b3535) were used as negative controls. These yielded values of 8.72 + 4.14 and 9.27 + 2.2 for CSMD and microcarrier cultures, respectively. Control experiments established that arrays screened on multiple, separate occasions with probes made from the same RNA preparation yield highly reproducible rank orders of abundance for expressed genes. In order to minimize differences resulting from probe synthesis or array to array variation, duplicate arrays were screened with separate probes generated from the same RNA and the results were averaged. Antibodies Anti-human MMPs (a-MMP) and anti-human TIMPs (a-TIMP) raised in either mouse [MMP Sampler Kit consisting of monoclonal antibodies, Oncogene Research Products (ORP), Boston, MA, U.S.A.] or rabbit (MMP Antibody Panel Starter Kit and TIMP Starter Kit consisting of polyclonal antibodies, Chemicon, Temecula, CA, U.S.A.) were employed in this study. a-MMP-1 (clone 41-1E5), -2 (clone 42-5D11) and -3 (clone 55-2A4) from ORP recognize epitopes within amino acid residues 332350 of human MMP1, 524539 of human MMP-2 and puried human pro-MMP-3, respectively. a-MMP-1, -2, and -3 from Chemicon (Temecula, CA, U.S.A.) were raised against synthetic peptides from human MMP-1, -2 and -3. aTIMP-1, -2, -3 and -4 from Chemicon (Temecula, CA, U.S.A.) recognize sequences in the carboxy terminus of the corresponding proteins. MMP antibodies recognize both the active and inactive forms of the protein. All antibodies have been previously characterized and do not cross react with other members of the corresponding families. Cy3-conjugated goat antimouse IgG [(HL); Jackson Immunoresearch Laboratories, West Grove, PA, U.S.A.], Cy3-conjugated donkey anti-rabbit IgG [(HL); Jackson Immunoresearch Laboratories, West Grove, PA, U.S.A.] and Cy3conjugated goat anti-rabbit IgG [(HL); Chemicon, Temecula, CA, U.S.A.] were used as secondary antibodies. Immunohistochemistry Immunohistochemistry was performed on human RPE choroid using indirect immunouorescence. Human RPE choroid tissue pieces, approximately 2 3 mm 23 mm in size, were dissected from eye quadrants and washed with PBS to remove the xative. The region of the eye from which the RPE choroid was derived was kept constant for a given donor between independent experiments. The tissue piece was embedded in 5 % low melting agarose

(Sigma, St. Louis, MO, U.S.A.) prepared in PBS containing 0.1 % NaN3 (Sigma, St. Louis, MO, U.S.A.). Agarose blocks were sectioned using a vibrotome to generate transverse sections of 100 mm thickness. The sections were blocked with PBS containing 0.5 % BSA and 0.1 % Triton X-100 for a minimum of 2 hr at 48C with gentle shaking. Blocking buffer was also used for subsequent preparations of antibody dilutions and all washes. Sections were treated with primary antibodies at concentrations of 5 mg ml 1 (monoclonal) and 1020 mg ml 1 (polyclonal), respectively, in a total dilution volume of 400 ml. Following incubation with the primary antibody (1216 hr at 48C with gentle shaking), the sections were washed with PBS prior to incubation with the secondary antibody. Following a 1216 hr incubation with the secondary antibody (at 5 7.5 mg ml 1 in a total dilution volume of 400 ml) at 48C with gentle shaking, the sections were washed with PBS and mounted in the anti-fade reagent, Prolong (Prolong Anti-Fade Kit, Molecular Probes, Eugene, OR, U.S.A.). In Situ Zymography Unxed, frozen globes were dissected into quadrants using a razor blade, and 10 mm sections were cut on a cryostat. Sections were incubated at 258C for 24 hr in the presence of 50 mM Tris, pH 7.4, 150 mM NaCl, 5 mM CaCl2, 0.2 mM sodium azide, 40 mg ml 1 DQ gelatin from pig skin (Molecular Probes, Eugene, OR, U.S.A.). This gelatin has been uorescinated to such a degree that the uorescein groups quench one another. Upon proteolysis, the uorescein residues become separated, quenching is relieved, and uorescence results. As a control, sections were incubated without the DQ gelatin. At the end of the incubation, sections were mounted in Prolong, and observed using the confocal microscope. In some experiments, protease inhibitors were added as indicated in the legend of Fig. 7. Because one of these inhibitors, 1,10-phenanthroline, is itself uorescent, an additional wash step using ethanol was carried out for this experiment. This step was also carried out in the absence of inhibitors, and it did not alter the amount of uorescence observed. Confocal Microscopy Analysis of uorescence was done on a Biorad 1024 laser scanning confocal microscope. Gain and black level were adjusted so that background uorescence observed in control sections (secondary antibody alone for immunohistochemistry; no DQ gelatin for in situ zymography) was negligible. Background uorescence was due to lipofuscin autouorescence present in the RPE layer, and, in some experiments, a weak autouorescence in the elastin layer of Bruch's membrane (Fig. 1). Multiple sections

MMPS AND TIMPS IN DRUSEN

145

F IG . 1. Expression of TIMP-3 in human RPEchoroid. Confocal micrographs showing TIMP-3 immunouorescence in donor tissues with [(A)(C), sample 4] and without a clinical history of AMD [(D), sample 8] are shown. (A)(C) Are representative of the variety of staining patterns observed in hard drusen. (D) shows a small drusen from a donor without a clinical history of AMD. (E) and (F) Are negative controls using only secondary antibody for samples 4 and 8, respectively. Immunoreactivity in Bruch's membrane [arrowhead in (A)] and drusen [black and white arrows in (A) (D)] is indicated. Antibody used was the polyclonal anti-TIMP-3 from Chemicon (Temecula, CA, U.S.A.). Abbreviations used are: r, RPE; bm, Bruch's membrane; d, drusen; c, choroid and cc, choriocapillaries. Magnication 600.

from each donor tissue were analyzed. To allow comparisons, similar laser settings were used for all panels within a particular gure. 3. Results cDNA Array Analysis of MMP and TIMP Expression by Cultured RPE Cells To approximate the relative expression of MMPs and TIMPs by RPE cells, multiple preparations of mRNA were isolated from two sources of cultured fetal human RPE and used to generate probes that were

hybridized to a cDNA array containing TIMP and MMP sequences. Expression of all 10 MMPs represented in the arrays could be detected (Table II). When normalized to total hybridization signal (see Materials and Methods), MMP-14, a transmembrane MMP (MT1-MMP, Sato et al., 1994), was consistently found to be the most abundantly expressed MMP. MMP-14 mRNA was about three- to four-fold less abundant than glyceraldehyde-3-phosphate-dehydrogenase, which was the most abundantly expressed mRNA detected. MMP-1 and -3 were also relatively abundant, ranking in the top six, about two-fold lower

146

S . T. L E U E T A L .

TABLE II cDNA array analysis of MMP and TIMP mRNA expression by cultured human RPE
mRNA MMP-14 MMP-3 MMP-15 MMP-10 MMP-12 MMP-1 MMP-7 MMP-9 MMP-13 MMP-8 TIMP-2 TIMP-3 TIMP-1 TIMP-4 TACE Furin SPC4 Caspase-1 Plasminogen GAPDH CSMD RPE 42.7 + 1.8 41.1 + 1.9 33.5 + 0.4 32.4 + 4.4 27.5 + 0.6 27.0 + 0.6 25.3 + 3.3 25.0 + 2.3 15.8 + 1.8 14.0 + 0.8 55.2 + 3.0 44.1 + 1.1 39.9 + 0.2 11.3 + 1.7 32.3 + 1.0 26.4 + 0.1 25.1 + 0.3 24.9 + 0.1 13.0 + 1.4 137 + 13 Microcarrier RPE 54.0 + 12.4 26.7 + 4.9 33.8 + 4.1 26.8 + 7.6 20.6 + 6.5 27.0 + 6.9 21.7 + 6.0 19.8 + 4.9 11.4 + 3.5 14.8 + 2.6 48.8 + 11.5 28.0 + 7.8 58.0 + 16.3 ND 26.9 + 6.8 27.6 + 6.9 27.2 + 9.3 18.0 + 3.4 15.6 + 3.6 253.8 + 63.7

drusen from non-AMD donors are included to allow comparison to AMD samples.

TIMPs Only TIMP-3 was detected in drusen of samples with and without a history of AMD [Fig. 1(A)(D)]. However, the pattern of TIMP-3 immunoreactivity in drusen varied considerably. Some drusen were only partially stained [Fig. 1(A) and (B)], and others were more uniformly positive [Fig. 1(C) and (D)]. Spherical hard drusen were sometimes only stained on their surface [Fig. 1(A)]. TIMP-3 immunoreactivity was also observed in Bruch's membrane and in the basal lamina surrounding the choriocapillaries in AMD tissue. Patchy staining was also observed throughout the choroid. The RPE, when present, showed immunoreactivity slightly above the background uorescence from lipofuscin [compare Fig. 1(A) and (E)]. RPE expression of TIMP-3 was more apparent in non-AMD samples, where the monolayer was more robust [Fig. 1(D)]. Non-AMD samples also manifested abundant TIMP-3 staining in RPE, Bruch's membrane, in drusen where present, and throughout the choroid. These data conrm earlier reports of TIMP-3 deposition in Bruch's membrane and in drusen (Fariss et al., 1997; Jomary et al., 1997; Kamei and Hollyeld, 1999). TIMP-4 immunouorescence was not observed in drusen in AMD samples [Fig. 2(A) and (B)], but staining was present in the RPE monolayer and in the choroid. Staining in the RPE was distributed throughout the cytoplasm. However, not all RPE cells were equally stained. In the choroid, a linear, stratied pattern of staining was observed, probably corresponding to longitudinal sections of collagen bers. Bruch's membrane generally lacked TIMP-4 immunoreactivity. A similar distribution of TIMP-4 was noted in nonAMD samples [Fig. 2(C) and (D)], except that the signal in the RPE monolayer was stronger and more consistent. As above, TIMP-4 was detected throughout the RPE cytoplasm. Drusen found in non-AMD samples were also negative for TIMP-4 [Fig. 2(D)]. The pattern of TIMP-2 immunouorescence was similar to that of TIMP-4. In AMD samples, drusen were generally negative for TIMP-2, although occasional staining was observed at the base of drusen particles [Fig. 3(A)]. Bruch's membrane lacked signicant staining, and the choroid was stained throughout [Fig. 3(A) and (B)]. In non-AMD samples, TIMP-2 appeared to be more concentrated on the basal side of the RPE cells [Fig. 3(C) and (D)]. In contrast to the other TIMPs, TIMP-1 was not strongly expressed in RPE in any samples examined. Immunoreactivity was observed in the choroid, however, in a diffuse pattern similar to that observed for TIMP-2 and -4 (data not shown).

Hybridization values were normalized to the total hybridization signal and are expressed as ( % of total hybridization signal) 100. Data shown are derived from two (CSMD) or three (microcarrier RPE) different probe preparations. The average + the range (CSMD) or the S.D. (microcarrier RPE) is reported. ND Not detected.

than MMP-14. MMP-8 and -13 were the least abundant, about ve-fold lower than MMP-14. Expression of MMPs was in the same range as that of other known proteases (TACE, etc.). Three of the four known TIMPs were detected, with TIMP-2 showing the strongest signal in the CSMD RPE and TIMP-1 in the microcarrier RPE. TIMP-4, the least abundant, was about six-fold lower than TIMP-1.

TIMP and MMP Immunoreactivity in RPE Choroid The distribution of MMPs and TIMPs was investigated in RPE choroid from donors with a documented history of AMD (samples 16) and from donors without a known AMD history (samples 712; summarized in Table I). Tissue from donors with a clinical history of AMD exhibited a range of symptoms characteristic of AMD. For example, sample 4 tissue was characterized by widespread loss of RPE and extensive, large, nearly spherical drusen. These were classied as hard drusen as dened by Sarks and Sarks (2001). In contrast, the RPE monolayer of sample 3 was relatively intact and drusen were less frequent, smaller, dome shaped hard drusen (Sarks and Sarks, 2001). Sections of these samples yielded the best morphology. Eyes without a clinical history of AMD manifested intact RPE monolayers and, on rare occasions, small amounts of drusen. Examples of

MMPS AND TIMPS IN DRUSEN

147

F IG . 2. Expression of TIMP-4 in human RPEchoroid. TIMP-4 immunouorescence in donor tissues with [(A), sample 4; (B), sample 3] and without AMD [(C), sample 10; (D), sample 8] is shown. Immunoreactivity in the RPE (curved arrow) is indicated. Antibody used was the polyclonal anti-TIMP-4 from Chemicon (Temecula, CA, U.S.A.). Abbreviations are as in Fig. 1. Magnication 600 in (A); 900 in (B) and (C); 1200 in (D).

F IG . 3. Expression of TIMP-2 in human RPEchoroid. TIMP-2 immunouorescence in donor tissues with [(A), sample 4; (B), sample 3] and without AMD [(C), sample 10; (D), sample 8] is shown. Immunoreactivity in the RPE (curved arrow) is indicated. Antibody used was the polyclonal anti-TIMP-2 from Chemicon (Temecula, CA, U.S.A.). Abbreviations are as in Fig. 1. Magnication 600.

148

S . T. L E U E T A L .

F IG . 4. Expression of MMP-1 in human RPEchoroid. MMP-1 immunouoresence in donor tissues with [(A) and (B), sample 4; (C), sample 3] and without AMD [(D), sample 10; (E) and (F), sample 8] is shown. Immunoreactivity in the RPE (curved arrows), Bruch's membrane [arrowhead in (D)] and on the surface of drusen [arrow in (A)] is indicated. Antibody used was the monoclonal anti-MMP-1 (clone 41-1E5) from ORP. Abbreviations are as in Fig. 1. Magnication 600.

MMPs Representatives from the three major classes of MMPs, MMP-1, -2, and -3, were also analyzed for expression in RPE choroid. Antibodies used in this study recognize both the inactive precursor and the mature form of these enzymes. The authors were particularly interested to see if any MMPs localized to drusen, where they might form a complex with TIMP-3. MMP-1, a collagenase, was detected on the surface of hard drusen particles in AMD tissue in a punctate staining pattern [Fig. 4(A)]. This staining may be due to remnants of RPE cells above the drusen. RPE cells on the surface of drusen [Fig. 4(B) and (C)] were also immunopositive for MMP-1. Bruch's membrane was weakly stained, and the choroid was stained throughout in AMD tissue. Tissues from donors not diagnosed with AMD showed robust MMP-1 expression in the RPE, with a notable line of staining on the apical surface [Fig. 4(D) and (E)]. Drusen, when detected in these samples, were negative for MMP-1 (data not shown). MMP-1 immunoreactivity was observed throughout the choroid. The gelatinase MMP-2 had a distribution quite similar to MMP-1 (Fig. 5). Punctate staining was observed on the surface of hard drusen [Fig. 5(A) and (B)], in RPE cells [especially on the apical surface, Fig. 5(C) and (D)], and throughout the choroid. No

signicant differences were noted between samples with a history of AMD and those without, although sometimes RPE staining was more robust in those without AMD [Fig. 5(C)]. MMP-3, a member of the stromelysin family, was also localized to the surface of hard drusen in AMD samples, possibly in RPE remnants [Fig. 6(A)]. RPE cells overlying and adjacent to drusen were intensely uorescent throughout their cytoplasm. Bruch's membrane was also stained with these antibodies, with the majority of uorescence localized to the basal lamina proximal to the choriocapillaris [Fig. 6(C)]. The choroid was also brightly stained. Non-AMD tissues showed a similar pattern of MMP3 immunoreactivity, except that the intensity of staining was lower than in the AMD samples [Fig. 6(D)(F)]. In addition, RPE staining was concentrated at the apical and basal surfaces of the monolayer. Bruch's staining was present as in AMD tissues, but was less striking. The choroid had the brightest staining, with the highest signal associated with capillaries.

In Situ Zymography Since the antibodies used in this study detect both active and inactive MMPs, a second method was employed to look at all active gelatinases. In situ

MMPS AND TIMPS IN DRUSEN

149

F IG . 5. Expression of MMP-2 in human RPEchoroid. MMP-2 immunouorescence in donor tissues with [(A), sample 4; (B), sample 3] and without [(C), sample 10; (D), sample 8] is shown. Immunoreactivity in the RPE (curved arrows), Bruch's membrane (arrowhead) and drusen (arrow) is indicated. Antibody used was the monoclonal anti-MMP-2 (clone 42-5D11) from ORP. Abbreviations are as in Fig. 1. Magnication 600.

F IG . 6. Expression of MMP-3 in human RPEchoroid. MMP-3 immunouorescence in donor tissues with [(A), sample 4; (B) and (C), sample 3] and without AMD [(D), sample 10; (E) and (F), sample 8] is shown. Immunoreactivity in the RPE (curved arrows), Bruch's membrane [arrowhead in (C)] and drusen [arrows in (A) and (B)] is indicated. Antibody used was the monoclonal anti-MMP-3 (clone 55-2A4) from ORP. Abbreviations are as in Fig. 1. Magnication 600.

150

S . T. L E U E T A L .

zymography was carried out by coating unxed, frozen sections with DQ-collagen, which becomes uorescent upon cleavage. Fig. 7 shows that active collagenases were detected throughout RPE choroid, except in drusen, which were consistently negative for protease activity [Fig. 7(A) and (B)]. To determine which class of protease was responsible for the activity, in situ zymography was carried out in the presence of protease inhibitors. Addition of PMSF, an inhibitor of serine proteases, resulted in a partial decrease in activity [Fig. 7(C)], indicating that serine proteases are responsible for a small fraction of the activity. Addition of EDTA or phenanthroline, inhibitors of metalloproteinases [Fig. 7(D) and (E)], completely blocked activity. These data indicate that most of the activity detected by this assay is due to metalloproteinases. 4. Discussion In this report, human RPE and samples were analyzed from donors with a documented history of AMD for the expression of metalloproteinases and inhibitors that play crucial roles in the turnover of ECM. Since Sorsby's fundus dystrophy, an ocular disease with symptoms similar to AMD, has been traced to mutations in TIMP-3, it was of interest to determine if AMD samples had altered distributions of biologically related proteins. Abnormalities in MMPs and TIMPs might be involved in the accumulation of drusen, a hallmark of AMD. Thus, it was of interest to determine if any MMPs and TIMPs were localized to drusen. Analysis of MMP and TIMP expression by cultured human RPE shows that these cells express at least 10 of the 21 known MMPs and three of four known TIMPs. The ratio of MMP to GAPDH mRNA is comparable to those observed in cultured human endometrium stromal cells (Nie et al., 1999). The relative order of mRNA abundance was similar in both sources of cultured RPE, with the exception of TIMP-1 which was the most abundant TIMP in the microcarrier RPE cells and the third most abundant in the CSMD cells. No signicant amounts of TIMP-1 immunoreactivity in RPE were detected, although staining was evident in the choroid. A previous report noted low TIMP-1 levels (Vranka et al., 1996) but others have detected TIMP-1 secretion by RPE cells in culture (Alexander et al., 1990; Padgett et al., 1997). In contrast, TIMP-4 immunoreactivity was detected in situ, but little TIMP-4 mRNA was detected in cultured RPE. Differences in sensitivity, or differences between cultured cells and those in situ may explain these conicting results. MMP-14, the most abundantly expressed metalloproteinase, has been noted before in the sclera, cornea, lens, choroid, RPE and retina (Smine and Plantner, 1997). This MMP degrades matrix proteins, but it is also known to activate proMMPs and shed membrane proteins (Kajita et al.,

2001). Results of the array analysis demonstrate the extensive range of MMPs expressed and serves as a preliminary screening tool to identify MMPs and TIMPs that are expressed. Because of the limititations inherent in the use of array data to compare relative abundance of expression, such as variability in probe synthesis, specicity and hybridization, further studies will be necessary to better quantify expression levels and clarify functional roles. Of all the proteins studied, only TIMP-3 was observed in the main body of drusen particles. TIMP-3 immunoreactivity was detected in drusen in samples from AMD and non-AMD tissue. Some variability in staining was noted, but this could be due to incomplete antibody penetration in the thick sections processed for confocal microscopy. Taken together, the results suggest that TIMP-3 is not restricted to a particular subregion of drusen. This conrms earlier studies that reported TIMP-3 immunoreactivity in drusen (Fariss et al. 1997; Kamei and Hollyeld, 1999). TIMP-3 has previously been localized to ECM, including Bruch's membrane, and is known to bind heparan sulfate groups of proteoglycans (Fariss et al., 1997; Gomez et al., 1997). Since drusen is known to contain ECM components, including heparan sulfate (Newsome et al., 1987, Mullins et al., 1997), TIMP-3 may associate with drusen because of these interactions. Alternatively, drusen associated TIMP-3 may be complexed with a MMP other than those analyzed in this study. Previous studies have seen less choroidal staining for TIMP-2 and -3, (Vranka et al., 1996; Kamei and Hollyeld, 1999) and MMP-1 and -2 (Guo et al., 1999). However, in the case of TIMP-2 and -3, Vranka et al. (1996) noted some TIMP-2 staining in the choroid and comment that TIMP-3 immunoreactivity is present on a `portion of large choroidal blood vessels.' For MMP-2, Guo et al. (1999) noted `patchy' choroidal staining and `intense intercapillary column staining'. `Punctate' MMP-1 staining in the choroid was also observed. The widespread MMP immunoreactivity is consistent with the pattern of activity detected by in situ zymography, and consistent with the fact that MMPs and TIMPs are components of plasma (e.g. Noji et al., 2001). The detection methods may be more sensitive than previous studies because of the antibodies used, or because confocal microscopy was used. Comparison of expression patterns in eyes with and without a history of AMD did not reveal any major differences in the distribution and/or abundance of MMPs or TIMPs (Table III). However, some minor differences were observed. First, RPE from samples without AMD yielded more robust staining for TIMP3, -4, MMP-1 and -2. These differences may be a reection of the general health of the RPE monolayers from AMD donors. As noted in the results, the RPE in AMD sample 4 was absent in many places and morphology was disrupted in others, reecting the

MMPS AND TIMPS IN DRUSEN

151

F IG . 7. Localization of active metalloproteinases in human RPEchoroid from AMD eyes. In situ zymography was carried out by incubating unxed 10 mM cryosections from donor tissues with AMD [(A), sample 5; (B)(F), sample 6] with DQ gelatin, which uoresces upon cleavage. Protease inhibitors were added to determine what class of protease was responsible for the activity [(A) and (B), no inhibitor; (C), 10 mM PMSF; 10 mM EDTA; (E), 10 mM 1,10-phenanthroline]. As a control, the DQ gelatin was omitted (F). Abbreviations are as in Fig. 1. Magnication 600.

152

S . T. L E U E T A L .

TABLE III Expression of TIMPs and MMPs in human RPEchoroid


Antigen TIMP-1 TIMP-2 TIMP-3 TIMP-4 MMP-1 MMP-2 MMP-3 Ocular history AMD No AMD AMD No AMD AMD No AMD AMD No AMD AMD No AMD AMD No AMD AMD No AMD Drusen ND ND ND * ND * ND * ND RPE Bruch's Choroid

RPE: retinal pigment epithelium; Bruch's: Bruch's membrane. The intensity of the uorescent signal is indicated by: , negative; , weakly positive; , positive; , strongly positive; ND, not detected. *Surface of drusen, possibly degenerating RPE, was weakly positive.

severity of the disease in this donor. A second difference was noted in the case of MMP-3, where RPE and the choroid immunoreactivity was greater in samples with a history of AMD. This result must be interpreted with caution, however, due to the qualitative nature of immunohistochemistry and differences in post mortem handling, xation time, antibody preparations, and the small number of samples analyzed. A previous study noted more TIMP-3 expression in AMD vs control eyes (Kamei and Hollyeld, 1999). While general increases in TIMP-3 were not observed, AMD eyes would be expected to have more TIMP-3 because it is prevalent in drusen, and there is more drusen in AMD eyes. Interestingly, MMP immunoreactivity was detected only on the surface of drusen. A punctate staining for all three MMPs was visible in the amorphous, acellular debris overlying hard drusen. This debris may derive from degenerating RPE cells. In fact, the central part of drusen particles was about the only place where MMP immunoreactivity was absent. The presence of TIMP-3 and the absence of MMPs led us to speculate that drusen may be `cold spots' for proteases that degrade matrix components. In situ zymography was carried out using a gelatin substrate and it was found that although metal ion dependent gelatinase activity is widespread, it was absent from drusen. The authors do not think that this is due to a lack of access to the drusen interior since TIMP-3 antibodies can gain access (Fig. 1) and since the drusen core should be exposed after sectioning. Most MMPs, with the exception of MMP-1 and -3, and some MT-MMPs are capable of gelatin cleavage (Sethi et al., 2000). This lack of proteolysis could contribute to drusen

formation, which may in turn contribute to the progression of AMD. This idea is consistent with the recent nding that mutations that lead to Sorsby's fundus dystrophy do not abolish TIMP-3 activity (Langton et al., 2000). These mutations could lead to an augmented TIMP-3 function, resulting in decreased matrix turnover. TIMP-3 is also known to possess additional activities, including the stimulation of cell growth (Yang and Hawkes, 1992), and these may also play a role. In later stages of AMD, neovascularization of the retina occurs as new vessels penetrate Bruch's membrane. Since the process of neovascularization requires active MMPs, these enzymes may play a role in the later stages of AMD. It was recently reported that overexpression of TIMP-3 in rat RPE resulted in an inhibition of experimentally-induced choroidal neovascularization (Takahashi et al., 2000), and MMP inhibitors are being investigated as potential AMD therapeutics (Brown, 2000). However, if inhibition of MMPs contributes to drusen formation, and drusen in turn leads to symptoms of AMD, this strategy may prove ineffective. Further studies will be necessary to test these hypotheses. Acknowledgements
This work was supported by research grants NEI EY06916 (SB), NEI EY0973 (DOC), NEI EY11527 (LVJ) and NEI EY11521 (DHA), a Biostar grant from Titan Pharmaceuticals and the University of California (DHA and LVJ), and a grant from Santa Barbara Cottage Hospital (DOC). The authors wish to thank the Lions Eye Bank of Portland, UK and Matt Nealon and Jason Atienza for their technical assistance. They also wish to thank Dr Teresa Burgess for suggesting the use of DQ gelatin.

MMPS AND TIMPS IN DRUSEN

153

References
Abdelsalam, A., Del Priore, L. and Zarbin, M. A. (1999). Drusen in age-related macular degeneration: pathogenesis, natural course, and laser photocoagulationinduced regression. Surv. Ophthalmol. 44, 128. Alexander, J. P., Bradley, J. M. B., Gabourel, J. D. and Acott, T. S. (1990). Expression of matrix metalloproteinases and inhibitors by human retinal pigment epithelium. Invest. Ophthalmol. Vis. Sci. 31, 25208. Allikmets, R. (2000). Simple and complex ABCR: genetic predisposition to retinal disease. Am. J. Hum. Genet. 67, 7939. Allikmets, R., Shroyer, N. F., Singh, N., Seddon, J. M., Lewis, R. A., Bernstein, P. S., Peiffer, A., Zabriskie, N. A., Li, Y., Hutchinson, A., Dean, M., Lupski, J. R. and Leppert, M. (1997). Mutation of the Stargardt's disease gene (ABCR) in age-related macular degeneration. Science 277, 18057. Armstrong, P. W., Moe, G. W., Howard, R. J., Grima, E. A. and Cruz, T. F. (1994). Structural remodeling in heart failure: gelatinase induction. Can. J. Cardiol. 10, 21420. Bigg, H. F., Shi, Y. E., Liu, Y. E., Steffensen, B. and Overall, C. M. (1997). Specic, high afnity binding of tissue inhibitor of metalloproteinases-4 (TIMP-4) to the COOH-terminal hemopexin-like domain of human gelatinase A. TIMP-4 binds progelatinase A and the COOH-terminal domain in a similar manner to TIMP-2. J. Biol. Chem. 272, 15496500. Brew, K., Dinakarpandian, D. and Nagase, H. (2000). Tissue inhibitors of metalloproteinases: evolution, structure and function. Biochim. Biophys. Acta 1477, 26783. Brinckerhoff, C. E. (1992). Regulation of metalloproteinase gene expression: implications for osteoarthritis. Crit. Rev. Euk. Gene Exp. 2, 14564. Brown, P. D. (2000). Ongoing trials with matrix metalloproteinase inhibitors. Expert Opin. Invest. Drugs 9, 216777. Chong, N. H., Alexander, R. A., Gin, T., Bird, A. C. and Luthert, P. J. (2000). TIMP-3, collagen, and elastin immunohistochemistry and histopathology of Sorsby's fundus dystropy. Invest. Ophthalmol. Vis. Sci. 41, 898902. De La Paz, M. A., Pericak-Vance, M. A., Lennon, F., Haines, J. L. and Seddon, J. M. (1997). Exclusion of TIMP3 as a candidate locus in age-related macular degeneration. Invest. Ophthalmol. Vis. Sci. 38, 10605. Egan, K. M. and Seddon, J. M. (1994). Principles and Practice of Ophthalmology: Clinical Practice. (Albert, D. M. and Jakobiec, F. A., Eds.). Philadelphia, PA, U.S.A.: Saunders. Fariss, R. N., Apte, S. S., Luthert, P. J., Bird, A. C. and Milam, A. H. (1998). Accumulation of tissue inhibitor of metalloproteinases-3 in human eyes with Sorsby's fundus dystrophy or retinitis pigmentosa. Br. J. Ophthalmol. 82, 132934. Fariss, R. N., Apte, S. S., Olsen, B. R., Iwata, K. and Milam, A. H. (1997). Tissue inhibitor of metalloproteinase-3 is a component of Bruch's membrane of the eye. Am. J. Pathol. 150, 3238. Gomez, D. E., Alonso, D. F., Yoshiji, H. and Thorgeirsson, U. P. (1997). Tissue inhibitors of metalloproteinases: structure, regulation and biological functions. Eur. J. Cell Biol. 74, 11122. Guo, L., Hussain, A. A., Limb, G. A. and Marshall, J. (1999). Age-dependent variation in metalloproteinase activity of isolated human Bruch's membrane and choroid. Invest. Ophthalmol. Vis. Sci. 40, 267682.

Heiba, I. M., Eiston, R. C., Klein, B. E. K. and Klein, R. (1994). Sibling correlations and segregation analysis of age-related maculopathy: the Beaver Dam Eye Study. Gen. Epidemiol. 11, 5167. Johnson, L. V., Ozaki, S., Staples, M. K., Erickson, P. A. and Anderson, D. H. (2000). A potential role for immune complex pathogenesis in drusen formation. Exp. Eye Res. 70, 4419. Jomary, C., Neal, M. J., Iwata, K. and Jones, S. E. (1997). Localization of tissue inhibitor of metalloproteinases-3 in neurodegenerative retinal disease. NeuroReport 8, 216972. Kajita, M., Itoh, Y., Chiba, T., Mori, H., Okada, A., Kinoh, H. and Seiki, M. (2001). Membrane-type 1 matrix metalloproteinase cleaves CD44 and promotescell migration. J. Cell Biol. 153, 893904. Kamei, M. and Hollyeld, J. G. (1999). TIMP-3 in Bruch's membrane: changes during aging and in age-related macular degeneration. Invest. Ophthalmol. Vis. Sci. 40, 236775. Klein, R., Klein, B. E. and Linton, K. L. (1992). Prevalence of age-related maculopathy. The Beaver Dam Eye Study. Ophthalmology 99, 93343. Kvanta, A., Shen, W. Y., Sarman, S., Seregard, S., Steen, B. and Rakoczy, E. (2000). Matrix metalloproteinase (MMP) expression in experimental choroidal neovascularization. Curr. Eye Res. 21, 68490. Langton, K. P., McKie, N., Curtis, A., Goodship, J. A., Bond, P. M., Barker, M. D. and Clarke, M. (2000). A novel tissue inhibitor of metalloproteinases-3 mutation reveals a common molecular phenotype in Sorsby's fundus dystrophy. J. Biol. Chem. 275, 2702731. Lin, H. and Clegg, D. O. (1998). Integrin alphavbeta5 participates in the binding of photoreceptor rod outer segments during phagocytosis by cultured human retinal pigment epithelium. Invest. Ophthalmol. Vis. Sci. 39, 170312. Liotta, L. A., Steeg, P. S. and Stetler-Stevenson, W. G. (1991). Cancer metastasis and angiogenesis: an imbalance of positive and negative regulation. Cell 64, 32736. Mullins, R. F., Johnson, L. V., Anderson, D. H. and Hageman, G. S. (1997). Characterization of drusenassociated glycoconjugates. Ophthalmology 104, 28894. Newsome, D. A., Hewitt, A. T., Huh, W., Robey, P. G. and Hassell, J. R. (1987). Detection of specic extracellular matrix molecules in drusen, Bruch's membrane and ciliary body. Am. J. Ophthalmol. 104, 37381. Nie, G. Y., Wang, J., Li, Y., Salamonsen, L. A., Li, Y. Y., Feldman, A. M., Sun, Y. and McTiernan, C. F. (1999). Construction and application of a multispecic competitor to quantify mRNA of matrix metalloproteinases and their tissue inhibitors in small human biopsies. Biochem. Biophys. Methods 40, 8199. Noji, Y., Kajinami, K., Kawashiri, M. A., Todo, Y., Horita, T., Nohara, A., Higashikata, T., Inazu, A., Koizumi, J., Takegoshi, T. and Mabuchi, H. (2001). Circulating matrix metalloproteinases and their inhibitors in premature coronary atherosclerosis. Clin. Chem. Lab. Med. 39, 3804. Padgett, L. C., Lui, G. M., Werb, Z. and LaVail, M. M. (1997). Matrix metalloproteinase-2 and tissue inhibitor of metalloproteinase-1 in the retinal pigment epithelium and interphotoreceptor matrix: vectorial secretion and regulation. Exp. Eye Res. 64, 92738. Parsons, S. L., Watson, S. A., Brown, P. D., Collins, H. M. and Steele, R. J. C. (1997). Matrix metalloproteinases. Br. J. Surg. 84, 1606.

154

S . T. L E U E T A L .

Penfold, P. L., Madigan, M. C., Gillies, M. C. and Provis, J. M. (2001). Immunological and aetiological aspects of macular degeneration. Prog. Ret. Eye Res. 20, 385414. Plantner, J. J., Jiang, C. and Smine, A. (1998). Increase in interphotoreceptor matrix gelatinase A (MMP-2) associated with age-related macular degeneration. Exp. Eye Res. 67, 63745. Rivera, A., White, K., Stohr, H., Steiner, K., Hemmrich, N., Grimm, T., Jurklies, B., Lorenz, B., Scholl, H. P., Apfelstedt-Sylla, E. and Weber, B. H. (2000). A comprehensive survey of sequence variation in the ABCA4 (ABCR) gene in Stargardt disease and agerelated macular degeneration. Am. J. Hum. Genet. 67, 80013. Ruiz, A., Brett, P. and Bok, D. (1996). TIMP-3 is expressed in the human retinal pigment epithelium. Biochem. Biophys. Res. Commun. 226, 46774. Sarks, S. H. and Sarks, J. P. (2001). Age-related maculopathy: nonneovascular age-related macular degeneration and the evolution of geographic atrophy. In `The Retina'. (Ryan, S. J., Ed.), third edn, Pp. 106499. St. Louis, MO, U.S.A.: CV Mosby. Sato, H., Takino, T., Okada, Y., Cao, J., Shinagawa, A., Yamamoto, E. and Seiki, M. (1994). A matrix metalloproteinase expressed on the surface of invasive tumour cells. Nature 370, 615. Seddon, J. M., Ajani, U. A. and Mitchell, B. D. (1997). Familial aggregation of age-related maculopathy. Am. J. Ophthalmol. 123, 199206. Seddon, J. M., Willett, W. C., Speizer, F. E. and Hankinson, S. E. (1996). A prospective study of cigarette smoking and age-related macular degeneration in women. J. Am. Med. Assoc. 276, 11416. Sethi, C. S., Bailey, T. A., Luthert, P. J. and Chong, N. H. (2000). Matrix metalloproteinase biology applied to vitreoretinal disorders. Br. J. Ophthalmol. 84, 65466.

Smine, A. and Plantner, J. J. (1997). Membrane type-1 matrix metalloproteinase in human ocular tissues. Curr. Eye Res. 16, 9259. Steen, B., Sejersen, S., Berglin, L., Seregard, S. and Kvanta, A. (1998). Matrix metalloproteinases and metalloproteinase inhibitors in choroidal neovascular membranes. Invest. Ophthalmol. Vis. Sci. 39, 2194200. Takahashi, T., Nakamura, T., Hayashi, A., Kamei, M., Nakabayashi, M., Okada, A. A., Tomita, N., Kaneda, Y. and Tano, Y. (2000). Inhibition of experimental choroidal neovascularization by overexpression of tissue inhibitor of metalloproteinases-3 in retinal pigment epithelium cells. Am. J. Ophthalmol. 130, 77481. Vranka, J. A., Johnson, E., Zhu, X., Shepardson, A., Alexander, J. P., Bradley, J. M. B., Wirtz, M. K., Weleber, R. G., Klein, M. L. and Acott, T. S. (1996). Discrete expression and distribution pattern of TIMP-3 in the human retina and choroid. Curr. Eye Res. 16, 10210. Vu, T. H. and Werb, Z. (2000). Matrix metalloproteinases: effectors of development and normal physiology. Genes Dev. 14, 212333. Weber, B. H., Vogt, G., Pruett, R. C., Stohr, H. and Felbor, U. (1994). Mutations in the tissue inhibitor of metalloproteinases-3 (TIMP3) in patients with Sorsby's fundus dystrophy. Nat. Genet. 8, 3526. Webster, A. R., Heon, E., Lotery, A. J., Vandenburgh, K., Casavant, T. L., Kean, T., Ohl, K. T., Beck, G., Fishman, G. A., Lam, B. L., Levin, A., Heckenlively, J. R., Jacobson, S. G., Weleber, R. G., Shefeld, V. C. and Stone, E. M. (2001). An analysis of allelic variation in the ABCA4 gene. Invest. Ophthalmol. Vis. Sci. 42, 117989. Yang, T. T. and Hawkes, S. P. (1992). Role of the 21-kDa protein TIMP-3 in oncogenic transformation of cultured chicken embryo broblasts. Proc. Natl. Acad. Sci. U.S.A. 89, 1067680.

You might also like