- Email: [email protected]
Induction of proliferative vitreoretinopathy by a unique line of human retinal pigment epithelial cells
BASIC SCIENCE STUDIES
Induction of proliferative vitreoretinopathy by a unique line of human retinal pigment epithelial cells Christian A. Wong,*t MSc; Michael J. Potter,* MD; Jing Z. Cui,* MD; Tom S. Chang,* MD; Patrick Ma, * MD; Alan L. Maberley,* MD; William H. Ross,* MD; Valerie A. White,*:j: MD; Arif Samad,ll MD; William Jia,*§ PhD; Dan Homan,* MB BS; Joanne A. Matsubara,*t PhD ABSTRACT • RESUME Background: The most widely used models of proliferative vitreoretinopathy (PVR) rely on injection of cells into the vitreous of animals. Using retinal pigment epithelial (RPE) cells from human PVR membranes may produce a more accurate model of human PVR. We performed a study to determine whether human RPE cells derived from a single epiretinal membrane (ERM) are capable of inducing the same disease in the rabbit eye, and whether the induced ERMs had cellular components similar to those of human PVR membranes. Methods: Cells were harvested from a human ERM obtained at surgery for PVR. RPE cells were cultured from the membrane and injected into the right eye of 24 New Zealand albino rabbits. The left eyes served as controls. The eyes were examined by indirect ophthalmoscopy over 4 weeks. The enucleated eyes were then examined by means of microscopy and histochemical analysis. Results: By day 7, PVR had developed in all but I of the 24 experimental eyes, with 8 progressing to localized tractional retinal detachment. By day 21, localized tractional retinal detachment had developed in 17 eyes; I eye progressed to extensive tractional retinal detachment by day 28. lmmunostaining showed that mostly RPE cells, but also myofibroblasts, glial cells and collagen, were present in the newly formed rabbit PVR membranes. Interpretation: Human RPE cells cultured from a PVR membrane appear to be capable of inducing PVR in rabbits. The resultant ERMs are similar to those formed in human PVR and consist mainly of RPE cells. Contexte : Les modeles de vitreoretinopathie proliferante (VRP) les plus utilises dependent de !'injection de cellules dans le corps vitre animal. L'utilisation de
From the Departments of *Ophthalmology, t Anatomy, :j:Pathology and §Surgery, University of British Columbia, Vancouver, BC, and lithe Department of Ophthalmology, Dalhousie University, Halifax, NS
Department of Ophthalmology, University of British Columbia, 2550 Willow St., Vancouver BC V5Z 3N9; fax (604) 875-4663; jms@ interchange.ubc.ca
Originally received May 31, 2001 Accepted for publication Nov. 30, 2001
This article has been peer-reviewed.
Correspondence to: Dr. Joanne A. Matsubara, Eye Care Centre,
Can J Ophthalmol 2002;37:21 1-20
Induction of PVR-Wong et al
211
Induction of PVR-Wong et al
cellules epitheliales pigmentaires de Ia retine (EPR) provenant des membranes VRP humaine peut produire un modele plus exact de Ia VRP. Nous avons done mene une etude pour determiner si les cellules EPR derivees d'une simple membrane epiretinienne (MER) pouvaient induire Ia maladie dans l'ceil du lapin et si les composants cellulaires des MER ainsi induits etaient semblables a ceux des membranes de Ia VRP chez les humains. Methodes : On a preleve des cellules d'une MER au cours d'une chirurgie de VRP. Les cellules EPR ont ete cultivees a partir de Ia membrane puis injectees dans l'ceil droit de 24 lapins albinos de Ia Nouvelle-Zelande, l'ceil gauche servant de temoin. L'on a examine les yeux par ophtalmoscopie indirecte sur une periode de 4 semaines. Les yeux enuclees furent ensuite soumis a un examen microscopique et a une analyse histochimique. Resultats: Le ye jour, Ia VRP s'etait developpee dans 23 des 24 yeux assujettis a I' experience et, chez huit sujets, elle progressait vers un decollement tractionnel et localise de Ia retine. Le 21 e jour, le decollement tractionnel et localise de Ia retine s'etait developpe dans 17 yeux; dans un des cas, Ia maladie avait pris de l'ampleur au 28e jour. L'immunocoloration a montre que les nouvelles membranes VRP des lapins consistaient principalement en cellules EPR mais aussi en myofibroblastes, cellules gliales et collagene.
Interpretation : Les cellules EPR cultivees a partir d'une membrane VRP humaine semblent pouvoir induire Ia VRP chez les lapins. Les MER qui en resultent ressemblent a celles qui se forment dans Ia VRP chez les humains et consistent principalement en cellules EPR.
P
roliferative vitreoretinopathy (PVR) is the most common cause of failure of retinal detachment surgery. PVR is characterized by the formation of epiretinal membranes (ERMs), which contract, causing retinal tears and detachments. Over the last 30 years much effort has been devoted to developing an experimental model that closely resembles PVR. An accurate model of PVR could ultimately lead to prevention of this complication, for which there is no adequate treatment or prophylaxis. The incidence of PVR remains at approximately 5% to 10% of retinal detachments that come to surgery. 1 The most widely used models of PVR rely on injection of cells into the vitreous of animals. Cells injected include those found in PVR membranes, such as retinal pigment epithelial (RPE) cells, fibroblasts, macrophages and glial cells.2-5 The cells are usually cultured from healthy, homologous eyes or tissue. Investigators have found that RPE cells isolated from human PVR membranes have different gene expression and morphologic characteristics from healthy human RPE cells in vitroN It is believed that growth factors and cytokines in the vitreous are responsible for this transdifferentiation. s-10 For example, interleukin-6, a cytokine thought to play an important role in the pathogenesis of PVR, is produced in higher levels by RPE cells from human PVR membranes.
212
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
Therefore, using RPE cells from human PVR membranes may produce a more accurate model of human PVR. We performed a study to determine whether human RPE cells derived from a single ERM, from a patient with a retinal detachment and PVR, are capable of inducing the same disease in the rabbit eye. We also examined whether the induced ERMs had cellular components similar to those of human PVR membranes. METHODS
Explant culture
of proliferative vitreoretinopathy
membranes
A single ERM from a patient with a retinal detachment and PVR was surgically removed and placed in Ham's Fl2 medium (Gibco BRL, Grand Island, NY). The membrane was washed with Hanks' balanced salt solution (BSS), dissected into 20 pieces measuring 0.3 em by 0.3 em and placed into separate 35-mm culture dishes. Ham's Fl2 medium, 30% fetal bovine serum (Gibco BRL), penicillin G sodium (100 units/ mL) and streptomycin (100 mg/mL) were added to the dishes, which were then placed in a humidified 37°C incubator containing 5% carbon dioxide. After cell migration from the membrane explants had occurred,
Induction of PVR-Wong et al the cells were dissociated with 0.25% trypsin (Gibco BRL) and placed into culture dishes containing growth medium, consisting of Ham's F12 medium supplemented with 10% fetal bovine serum and antibiotics as described above.
Cell culture immunohistochemistry Subconfluent cultures were fixed with cold methanol/acetone (1: 1 v/v) at -20°C for 15 minutes. After fixation, the cultures were rinsed three times with phosphate-buffered saline (PBS). Their cell membranes were made permeable by treating with 1% Triton X-100 for 5 minutes, and, to prevent nonspecific antibody binding, the cultures were treated with 10% normal horse serum diluted with 1% bovine serum albumin (BSA) in PBS. The fixed cultures were then incubated with a primary antibody in BSA/PBS for 1 hour at room temperature. The primary monoclonal antibody was targetted against human pancytokeratin identifying human keratins 1, 4, 5, 6, 8, 10, 13, 18 and 19 (1:200; Sigma-Aldrich, Oakville, Ont.), a-smooth muscle actin (u-SMA) (1:50; Dako, Carpinteria, Calif.) or glial fibrillary acidic protein (GFAP) (1:2000; Dako). Bound antibody was detected by incubation with a 1:500 dilution in BSA/PBS of biotin-labelled horse antimouse IgG (Vector Laboratories, Burlingame, Calif.), followed by horseradish peroxidase conjugated to avidin. Labelled cells were visualized by means of a standard diaminobenzidine reaction. To rule out nonspecific staining, control cultures were analysed in the same way, except that the primary antibody was replaced with 1% normal horse serum in BSA/PBS. In addition, human ovarian adenocarcinoma (OVCAR-3, US National Institutes of Health) cells, which are known to express cytokeratin, were used as a positive control.''
Collagen gel contraction assay A collagen gel contraction assay was adapted from the method described by Raymond and Thompson.l2 Two groups of RPE cells were investigated: cells at passage 5 and cells at passage 24. The collagen gel was prepared using Vitrogen-100 (3 mg/mL type I bovine dermal collagen, Collagen Corp., Palo Alto, Calif.) mixed 16:1 v/v with lOX PBS and brought to neutral pH with 0.2M sodium hydroxide at 4 °C. The RPE cells were harvested from the culture dishes, centrifuged at 200g for 3 minutes and resuspended in 1 mL of Ham's Fl2/10 medium with lOmM
N-2-hydroxyethylpiperazine-N' -2-hydroxypropanesulfonic acid (HEPES) buffer. Next, 9 mL of the prepared collagen mixture was combined with 1 mL of RPE cells/medium to produce a collagen-RPE cell mixture with a cell concentration of 1.5 x 1Q5 cells/mL and a collagen concentration of 2.5 to 2. 7 mg/mL. Then 500 J1L of each collagenRPE cell mixture was placed in 6 wells of a 24-well plate. Six additional wells for each group served as controls, receiving 500 mL of collagen gel alone (no cells) at a concentration of 2.5 to 2.7 mg/mL. The plates were then placed in a 5% carbon dioxide-95% air incubator at 37°C for 15 minutes to promote solidification of the gel. On removal from the incubator, the collagen gel was separated from the side of the well with a 25gauge needle. Each well then received 1.5 mL of Ham's Fl2/10 medium, added by the drop down the side of the well. The medium was changed every 2 to 3 days by aspirating from the edge of each well, care being taken not to disrupt the gel. The wells were observed on days 1, 4 and 8, and the area of the gel was determined by measuring the longest and shortest diameters of the gel over a millimetre grid.
Retinal pigment epithelial cell preparation and injection The RPE cells used for all cell injections were of intermediate passage (approximately passage 12). Subconfluent (and thus still proliferating) RPE cells were dissociated with 0.25% trypsin and washed twice with BSS. Cell death was confirmed with trypan blue staining. Virtually all cells stained negatively for trypan blue. Approximately 1 X IQ6live cells were resuspended in 0.2 mL of BSS. The cells were prepared no more than 20 minutes before being injected into the eyes. Twenty-four New Zealand albino rabbits weighing 2.5 to 3.5 kg were used for this study. The right eye of each rabbit was injected with cells, and the left eye served as a control. Following pupillary dilation with two eyedrops containing 5% phenylephrine and 1% tropicamide, the rabbits were anesthetized with ketamine (50 mg/kg) and xylazine (10 mg/kg). A 25gauge needle was passed through the sclera 3 mm posterior to the limbus, and 0.2 mL of BSS containing approximately 1 X 1Q6 RPE cells was injected directly over the optic disc of the right eye. To prevent an increase in intraocular pressure, 0.2 mL of aqueous was removed simultaneously from the anterior
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
213
Induction of PVR-Wong et al chamber with a 30-gauge needle. For the left eye, 0.2 mL of BSS not containing RPE cells was injected. After surgery 1% chloramphenicol ointment (Vetcom Inc., Upton, Que.) was applied topically. The rabbits were examined by indirect ophthalmoscopy for inflammation during the first 6 days after cell injection, and for inflammation and PVR on days 7, 14, 21 and 28. PVR was graded according to the scheme described by Fastenberg and colleagues. 3 One animal with an ERM without retinal traction or tractional retinal detachment was observed for 3 months. On day 28, 23 of the animals were deeply anesthetized and killed with an injection of 1.5 mL of sodium pentobarbital (240 mg/mL) into an auricular vein; the remaining rabbit was killed at 3 months. All the animals were treated according to the Declaration of Helsinki, the guidelines of the Canadian Council on Animal Care and the Statement for the Use of Animals in Ophthalmic and Vision Research by the Association for Research in Vision and Ophthalmology.
visualized by means of exposure to the biotin complexed with aminoethylcarbazole (Vectastain Elite ABC Kit, Vector Laboratories). The cryosections were also counterstained with Mayer's modified hematoxylin solution. Three primary monoclonal antibodies were used on each right and left eye. The antibodies targetted against human pancytokeratin, GFAP and a-SMA. Control cryosections from both right and left eyes were subjected to an immunohistochemical assay in which the primary antibody was omitted (i.e., sections were incubated with diluent [1% normal horse serum in BSA/PBS] only) to rule out nonspecific staining. The presence of collagen was assessed with Masson's trichrome stain. 13 Human cardiac tissue, which is known to contain collagen,I4 served as the positive control. RESULTS
In vitro studies
Cell culture immunohistochemistry Development
of proliferative vitreoretinopathy
Both eyes of 13 rabbits were enucleated at 28 days and from a further rabbit at 3 months. The globes were fixed in 10% neutral-buffered formalin for 48 hours and wholly embedded in paraffin wax. Sections were made at a thickness of 12 11m and then stained with hematoxylin and eosin. Membrane histochemistry and trichrome staining To determine the cellular components of the induced ERMs, both eyes of the remaining 10 rabbits were enucleated at 28 days. The anterior segments were removed, and the eyecups were placed in TissueTek OCT (VWR Canlab, Toronto) before being immersed in liquid nitrogen. Then 8-11m sections were cut, mounted on glass slides and placed in a -85°C freezer for storage. After thawing, the sections were air-dried and fixed in acetone for 5 minutes. The cryosections were then treated with 1% BSA/PBS for 15 minutes to block nonspecific binding sites and immersed in 0.3% hydrogen peroxide. The cryosections were exposed to primary antibody for 30 minutes and then washed twice in PBS. Bound antibody was detected by incubation with 1:500 biotin-labelled horse antimouse IgG (Vector Laboratories) in PBS/BSA, followed by horseradish peroxidase conjugated to avidin. Labelled cells were
214
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
All sample cells from passage 4 cultures, passage 17 cultures and passage 24 cultures stained positively for cytokeratin (Fig. 1) but weakly for a-SMA and GFAP, comparable to staining produced when the primary antibodies were replaced with PBS (negative controls). The positive control for cytokeratin (OVCAR-3 cells) stained for cytokeratin, as expected. These findings suggested that the strong immunoreactivity for cytokeratin was due to specific binding and that the cells were of RPE origin.
Fig. !-Positive cytokeratin immunoreactivity in cells from passage 4 cultures obtained from human epiretinal membrane (ERM) (scale bar = 5 ~m).
Induction of PVR-Wong et al
100 -;:!2.
80
ctf
60
ctl Q)
40
0
a> ....
20 0 0
2
4
3
6
5
8
7
Time, d Fig. 2-Reduction in collagen gel area due to contraction mediated by retinal pigment epithelial (RPE) cells for passage 5 (•) and passage 24 (o) cells compared to control wells (no cells).
Table 1-Stage of proliferative vitreoretinopathy* induced in 24 rabbit eyest Day Stage
7
14
21
28
0
0
9 6 8 0
4
2 5 17 0
0 I 5 17
No proliferative vitreoretinopathy I
2 3
4
10 10 0
*Graded according to the stages described by Fastenberg and colleagues3 as follows: stage I, intravitreal membrane; stage 2, focal traction, localized vascular changes; stage 3, localized detachment of medullary ray; stage 4, extensive retinal detachment, total medullary ray detachment; stage 5, total retinal detachment, retinal folds and holes. tThe control eyes were normal at all time points.
Collagen gel contraction assay The reduction in collagen gel area due to RPEmediated contraction for both passage 5 and passage 24 RPE cells is summarized in Fig. 2. At day 1 there was 49% reduction with the passage 5 cells and 25% reduction with the passage 24 cells compared to the control wells (no cells added). By day 8 the collagen gel area in both experimental groups was 23% that of the control group.
Fig. 3-Fundus photographs of rabbit eyes with induced proliferative vitreoretinopathy (PVR). A: Experimental eye 7 days after cell injection, showing ERM and underlying localized tractional retinal detachment (arrows). B: Experimental eye 28 days after cell injection, showing ERM exerting retinal traction (arrows indicate elevation of the medullary ray). C: Control eyes were normal at 28 days.
In vivo studies At day 3 there was minor conjunctivitis in both the experimental and control eyes. The anterior chamber was clear. By day 7 the conjunctivitis had cleared completely in both groups, and the eyes appeared normal.
By day 7 PVR had developed in all but one of the experimental eyes; nine eyes had ERMs alone, and eight had localized tractional retinal detachments with ERMs present (Fig. 3, A, and Table 1). By day 28, 17
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
215
Induction of PVR-Wong et al
Fig. 5-Retinal folds (arrows) in experimental eye, providing evidence that ERM-induced retinal traction has occurred (hematoxylin-eosin; scale bar = IS f.Jm).
C). The experimental eye observed for 3 months remained at stage 1 (ERM); the control eye was normal at 3 months. Histopathological findings All eyes in which localized or extensive tractional retinal detachments were diagnosed on ophthalmoscopy showed the presence of an ERM (Fig. 4, top), the presence of subretinal fluid (Fig. 4, top), evidence of retinal traction (Fig. 5) and degenerated photoreceptor outer segments (Fig. 6, A) on histologic evaluation. All control eyes displayed normal retinal morphologic characteristics (Fig. 6, B), with the absence of subretinal fluid (Fig. 4, bottom). Artefactual retinal detachment occurred in the control eyes (Fig. 4, bottom) after enucleation as a result of the fixation procedure.IS These histochemical findings confirmed the ophthalmoscopic observations.
Fig. 4-Top: Experimental eye 28 days after cell injection. The presence of subretinal fluid (SF) and ERM (arrowheads) provides evidence that retinal detachment has occurred (hematoxylin-eosin; scale bar = 2 mm). Bottom: Control eye. Subretinal fluid and ERMs were not present in any control eye 28 days after injection. Arrowheads indicate artefactual retinal detachment (hematoxylin-eosin; scale bar = 2 mm).
eyes (71%) had localized tractional retinal detachments, 5 (21%) had retinal traction (Fig. 3, B), and 1 (4%) had an extensive tractional retinal detachment. None progressed to stage 5 (total retinal detachment). All control eyes were normal through 28 days (Fig. 3,
21 6
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
Membrane immunohistochemistry and trichrome staining All induced ERMs examined displayed immunoreactivity for cytokeratin (Fig. 7, A), c:x-SMA (Fig. 7, B) and GFAP (Fig. 7, C), which indicated that RPE cells, myofibroblasts and glial cells respectively were present. RPE cells and myofibroblasts were the dominant cell types. The experimental and control eyes stained for cytokeratin in the RPE layer and for c:x-SMA in the choroidal blood vessels. All negative controls, in which the primary antibody was omitted, displayed background levels of immunoreactivity.
Induction of PVR-Wong et al
Fig. 6-Photomicrographs 28 days after cell injection (hematoxylin-eosin; scale bar = 5 1Jm). A: Experimental eye shows degenerated photoreceptor outer segments (OS) (arrows), which suggests that photoreceptors lost their association with RPE monolayer before day 28. This provides additional support to the clinical findings that a retinal detachment had occurred in this eye before enucleation. B: Control retina shows healthy photoreceptor OS (arrows). IN = inner segments, ENL = external nuclear layer, EPL = external plexiform layer, INL = internal nuclear layer, IPL = internal plexiform layer, GCL = ganglion cell layer.
© Fig. ?-Immunohistochemical and trichrome staining of ERM in experimental eye with stage 3 PVR (localized detachment of medullary ray). A: At day 28, positive immunoreactivity for cytokeratin reflects the presence of RPE cells throughout the ERM {black arrow). The rabbit RPE monolayer (white arrow) also showed immunoreactivity for cytokeratin (scale bar = 25 IJm). B, C: Positive a -smooth muscle actin and glial fibrillary acidic protein (GFAP) immunoreactivity throughout ERM indicates the presence of myofibroblasts (B) and glial cells (C) respectively. The aminoethylcarbazole staining for GFAP is seen as a light aminoethylcarbazole reaction product (counterstained with hematoxylin; scale bar = 25 1Jm). 0: Positive trichrome staining indicates the presence of collagen throughout the ERM, as seen by the blue reaction product (scale bar = I00 1Jm).
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
21 7
Induction of PVR-Wong et al
Trichrome staining for collagen was seen in our ERMs (Fig. 7, D) and in the cardiac tissue control sections. INTERPRETATION
Our results suggest that human RPE cells taken from an ERM can induce PVR in a rabbit model. We observed a progression in the clinical manifestations of PVR, from the early appearance of a membrane to later signs of retinal traction and detachment. This is consistent with the picture in human PVR, although the time course is shorter than for the development of PVR after retinal detachment surgery. The short time course of PVR in our model is useful for experimental purposes, as it will allow for rapid assessment of potential therapeutic or prophylactic agents against PVR. It may also shed light on the mechanism of PVR. In our model, injection resulted in a sudden influx of about a million RPE cells into the vitreous, whereas in human PVR, RPE cells migrate more gradually through retinal tears or as a result of surgical procedures during retinal detachment surgery.16·17 More extensive retinal tears are associated with more rapid onset of PVR.18,19 The shorter time course of PVR in our model may also be explained by differences between RPE cells in PVR membranes and those in the RPE layer in the retina. RPE cells that migrate from their retinal monolayer are known to undergo a number of changes, collectively termed dedifferentiation, metaplasia or transdifferentiation.2.8-10,20 First, it is thought that RPE cells cultured from the retinal monolayer assume transdifferentiated (myofibroblast-like) characteristics and that these changes are necessary for increased contractile activities.2,9,21,22 This ability likely plays a key role in the development of tractional retinal detachment due to PVR. In this study we showed that RPE cells, cultured from the same human PVR membrane used in the intravitreal injections, were able to rapidly contract a collagen matrix. This may help explain the short time course of development of PVR in our model. Second, RPE cells isolated from PVR membranes are known to express different genes - and, thus, proteins -from those in the retinal monolayer (e.g., interleukin-6). 6 The genes are likely to code for proteins that are important in the pathogenesis of PVR, such as growth factors and cytokines and their receptors. Cytokines, such as interleukin-6, may function as chemoattractants for other cell types involved, includ-
218
CAN
J OPHTHALMOL-VOL. 37, NO. 4, 2002
ing the glial cells and macrophages seen in our model. Growth factors are also believed to play a vital role in the transdifferentiation of RPE cells and thus in PVR. 8 Indeed, our group has found that a number of growth factors and their receptors are expressed in the unique culture of RPE cells used in this study (unpublished data, 2000). This also helps explain why our model had a shorter time course than human PVR after retinal detachment surgery and provides important clues for the development of potential therapies. The RPE cells used in this study have unique qualities, which also may have contributed to the shorter time course of PVR that was observed. The RPE cells used in our model are unique in that they were derived from a single PVR membrane and were previously shown to complete several population doublings (at least 48) without showing signs of replicative senescence. It is unusual for cells cultured from human PVR membranes to survive more than a few passages. Even human RPE cells cultured de novo from donor eyes tend to become senescent after 15 to 20 passages. It is likely that the RPE cells used in this study had spontaneously transformed. RPE transformation could serve as a benefit in providing an "immortalized" culture of cells for future studies. Conversely, transformed RPE cells may not accurately reflect RPE cells found in other PVR membranes. We previously demonstrated that our RPE cells retained epithelial-like characteristics and contractile properties up to passage 24. 11 However, additional experiments should be performed to further characterize this unique culture of cells. These could include studying the expression of structural proteins and growth factors and their receptors, and evaluating the stability of both morphology and gene expression in successive population doublings. Based on our immunohistochemical characterization of the cultured cells and to the best of our knowledge, the only human cells injected into the vitreous of the rabbit eyes were RPE cells. Therefore, glial cells found in the ERMs at day 28 must have originated from the rabbit. It is likely that these glial cells invaded the vitreous through the tractional tears in the retina. It is also likely that most of the cells within the developing ERM in our model originated from the human RPE cells injected. Since the injection did not break the inner limiting membrane, rabbit cells did not have access to the vitreous at this stage; they gained access to the vitreous cavity only after the tractional retinal tears had occurred. This is an advantage of our model because therapeutic agents that tar-
Induction of PVR-Wong et al
get RPE proliferation and contraction can be tested specifically. Previous animal models of PVR focused on the injection of cells into the vitreous or on a more physical intervention, including injection of particles, gas, albumin or enzymes, and mechanical tearing of the retina. Cell injection models have involved administering whole blood, erythrocytes, platelets, leukocytes, fibroblasts or glial cells. Although some of these models have demonstrated clinical characteristics of PVR, few of the groups published histologic data. 2-4.23-32 In 1975 Mandelcorn and associates2 injected adult human RPE cells into the vitreous of the primate eye. They observed metaplasia and proliferation but were not able to induce tractional retinal detachment. In 1986 Grierson and coworkers33 injected human RPE cells, derived from either an adult or fetal cell line, into the monkey eye. They also found that the adult cells, cultured from the RPE monolayer of a donor eye, produced an insubstantial ERM and no tractional retinal detachment. However, they observed that fetal RPE cells produced extensive membranes and traction retinal detachments. These studies support the unique nature of the RPE cells used in our model. It is likely that our RPE cells already possess the characteristics of transdifferentiation, as they were able to induce traction retinal detachments. These characteristics logically include active proliferation and expression of the necessary structural proteins and growth factors and their receptors. Like the models proposed by Mandelcorn and associates2 and Grierson and coworkers,33 our model involved the injection of xenogenic cells (human RPE cells) into the vitreous cavity of the rabbit eye. After surgery, we observed minor inflammation in both the experimental and control eyes, which resolved completely by day 4; this suggested that the inflammation was due to the surgery itself and not to the human cells. Furthermore, in one eye we allowed an ERM, which did not cause a tractional retinal detachment, to survive for 3 months. There was no evidence of ocular inflammation throughout this period. The lack of inflammation is consistent with anterior chamber-associated immune deviation (immune privilege in several parts of the eye, including the vitreous cavity).34 Owing to this immune privilege, a systemic immune response can be induced without causing ocular inflammation, as the mediators of immunogenic inflammation delayed hypersensitivity and complement fixation are quenched. Our data illustrate that in our model,
tractional retinal detachment occurred owing to ERM contraction, not inflammation. SUMMARY
We found that human RPE cells cultured from a patient with PVR are capable of inducing the same disease in the rabbit eye. Our model is unique and has distinct advantages that make it an excellent tool for studying the pathogenesis of, and potential treatments for, PVR. We thank Dr. Cal Roskelly for his generous gift of the human ovarian carcinoma cells; Howard Meadows, Dr. Zheng Chen, Eleanor To, Dawn Lam and Tara Stewart for their expert technical advice and assistance; Virginia Booth, Dr. Farzin Forooghian and Dr. Peter Leong-Sit for their advice and assistance with the ocular injections; and Drs. Wayne Yogi and John Church for their helpful comments and suggestions. REFERENCES
1. Retina Society Terminology Committee. The classification of retinal detachment with proliferative vitreoretinopathy. Ophthalmology 1983;90:121-5. 2. Mandelcom MS, Machemer R, Fineberg E, Hersch SB. Proliferation and metaplasia of intravitreal retinal pigment epithelium cell autotransplants. Am J Ophthalmol1975 ;80: 227-37. 3. Fastenberg DM, Diddie KR, Dorey K, Ryan SJ. The role of cellular proliferation in an experimental model of massive periretinal proliferation. Am J Ophthalmol 1982;93: 565-72. 4. Peters MA, Burke JM, Clowry M, Abrams GW, Williams GA. Development of traction retinal detachments following intravitreal injections of retinal Miiller and pigment epithelial cells. Graefes Arch Clin Exp Ophthalmol 1986; 224:554-63. 5. Hui YN, Sorgente N, Ryan SJ. Posterior vitreous separation and retinal detachment induced by macrophages. Graefes Arch Clin Exp Ophthalmo/1984;225:279-84. 6. AbeT, Durlu YK, Tarnai M. The properties of retinal pigment epithelial cells in proliferative vitreoretinopathy compared with cultured retinal pigment epithelial cells. Exp Eye Res 1996;63:201-10. 7. Hiscott PS, Grierson I, Hitchins CA, Rahi AH, McLeod D. Epiretinal membranes in vitro. Trans Ophthalmol Soc UK 1983;103:89-102. 8. Vidaurri-Leal J, Hohman R, Glaser BM. Effect of vitreous on morphologic characteristics of retinal pigment epithelial cells. A new approach to the study of proliferative vitreoretinopathy. Arch Ophthalmol1984; 102:1220-3. 9. Casaroli-Marano RP, Pagan R, Vilaro S. Epithelial-mesenchymal transition in proliferative vitreoretinopathy: intermediate filament protein expression in retinal pigment
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
219
Induction of PVR-Wong et al
10.
11.
12.
13.
14.
15.
16.
17. 18.
19. 20.
21.
22.
23.
epithelial cells. Invest Ophthalmol Vis Sci 1999;40: 2062-72. Kirchhof B, Sorgente N. Pathogenesis of proliferative vitreoretinopathy. Modulation of retinal pigment epithelial cell functions by vitreous and macrophages. Dev Ophthalmoll989;16:1-53. Wong CA, Cui JZ, Matsubara JA, Retina Group of UBC. Human RPE cells derived from a PVR membrane that retains epithelial characteristics and contractile properties at high passage numbers [abstract]. Invest Ophthalmol Vis Sci 1999;40:S463. Raymond MC, Thompson JT. RPE-mediated collagen gel contraction. Inhibition by colchicine and stimulation by TGF-beta. Invest Ophthalmol Vis Sci 1990;31: 1079-86. Burkitt HG, Young B, Heath JW. Notes on staining techniques. In: Killgore J, editor. Wheater's basic histology. 3rd ed. Edinburgh: Churchill Livingstone; 1993. p. 400. Burkitt HG, Young B, Heath JW. Circulatory system. In: Killgore J, editor. Wheater's basic histology. 3rd ed. Edinburgh: Churchill Livingstone; 1993. p. 140-52. Margo CE, Lee A. Fixation of whole eyes: the role of fixative osmolarity in the production of tissue artifact. Graefes Arch Clin Exp Ophthalmol1995;233:366-70. Charteris DG. Proliferative vitreoretinopathy: pathobiology, surgical management, and adjunctive treatment. Br J Ophthalmoll995;79:953-60. Pastor JC. Proliferative vitreoretinopathy: an overview. Surv Ophthalmoll998;43:3-l8. Bonnet M. Clinical factors predisposing to massive proliferative vitreoretinopathy in rhegmatogenous retinal detachment. Ophthalmologica 1984;188:148-52. Scott JD. Equatorial giant tears affected by massive vitreous retraction. Trans Ophthalmol Soc UK 1976;96:309-12. MUller-Jensen K, Machemer R, Azarnia R. Autotransplantation of retinal pigment epithelium in intravitreal diffusion chamber. Am J Ophthalmoll975;80:530-7. Vidaurri-Leal JS, Glaser BM. Effect of fibrin on morphologic characteristics of retinal pigment epithelial cells. Arch Ophthalmoll984;l02:1376-9. Grisanti S, Guidry C. Transdifferentiation of retinal pigment epithelial cells from epithelial to mesenchymal phenotype. Invest Ophthalmol Vis Sci 1995;36:391-405. Algvere P, Wallow IH, Martini B. The development of vit-
220
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
reous membranes and retinal detachment induced by intravitreal carbon microparticles. Graefes Arch Clin Exp Ophthalmoll988;226:471-8. Frenzel EM, Neely KA, Walsh A W, Cameron JD, Gregerson DS. A new model of proliferative vitreoretinopathy. Invest Ophthalmol Vis Sci 1988;39:2157-64. Hitchins CA, Grierson I, Hiscott PS. The effects of injections of cultured fibroblasts into the rabbit vitreous. Graefes Arch Clin Exp Ophthalmol1985;223:237-49. Hsu HT, Dorey K, Sorgente N, Ryan SJ. Surgical removal of the vitreous. Its effect on intraocular fibroblast proliferation in the rabbit. Arch Ophthalmol1984;102:605-7. Lean LS, Van der Zee WAM, Ryan SJ. Experimental model of proliferative vitreoretinopathy (PVR) in the vitrectomised eye: effect of silicone oil. Br J Ophthalmol 1984;68:332-5. Miller B, Miller H, Ryan SJ. Experimental epiretinal proliferation induced by intravitreal red blood cells. Am J Ophthalmol 1986; 102:188-95. Shabo AL, Maxwell DS. Experimental immunogenic proliferative retinopathy in monkeys. Am J Ophthalmol 1977;83:471-80. Topping TM, Abrams GW, Machemer R. Experimental double-perforating injury of the posterior segment in rabbit eyes; the natural history of intraocular proliferation. Arch Ophthalmoll979;97:735-42. Machemer R, Laqua H. Pigment epithelium proliferation in retinal detachment (massive periretinal proliferation). Am J Ophthalmoll975;80: 1-23. Machemer R, Van Horn D, Aaberg TM. Pigment epithelial proliferation in human retinal detachment with massive periretinal proliferation. Am J Ophthalmol 1978;85: 181-91. Grierson I, Boulton M, Hiscott P, Hitchins C, Gilbert D, McLeod D. Human retinal pigment epithelial cells in the vitreous of the owl monkey. Exp Eye Res 1986;43: 491-502. Jiang LQ, Streilein JW. Immune privilege extended to allogenic tumor cells in the vitreous. Invest Ophthalmol Vis Sci 1991;32:224-8.
Key words: proliferative vitreoretinopathy, human retinal pigment epithelial cells, rabbit model, retinal detachment
Induction of proliferative vitreoretinopathy by a unique line of human retinal pigment epithelial cells Christian A. Wong,*t MSc; Michael J. Potter,* MD; Jing Z. Cui,* MD; Tom S. Chang,* MD; Patrick Ma, * MD; Alan L. Maberley,* MD; William H. Ross,* MD; Valerie A. White,*:j: MD; Arif Samad,ll MD; William Jia,*§ PhD; Dan Homan,* MB BS; Joanne A. Matsubara,*t PhD ABSTRACT • RESUME Background: The most widely used models of proliferative vitreoretinopathy (PVR) rely on injection of cells into the vitreous of animals. Using retinal pigment epithelial (RPE) cells from human PVR membranes may produce a more accurate model of human PVR. We performed a study to determine whether human RPE cells derived from a single epiretinal membrane (ERM) are capable of inducing the same disease in the rabbit eye, and whether the induced ERMs had cellular components similar to those of human PVR membranes. Methods: Cells were harvested from a human ERM obtained at surgery for PVR. RPE cells were cultured from the membrane and injected into the right eye of 24 New Zealand albino rabbits. The left eyes served as controls. The eyes were examined by indirect ophthalmoscopy over 4 weeks. The enucleated eyes were then examined by means of microscopy and histochemical analysis. Results: By day 7, PVR had developed in all but I of the 24 experimental eyes, with 8 progressing to localized tractional retinal detachment. By day 21, localized tractional retinal detachment had developed in 17 eyes; I eye progressed to extensive tractional retinal detachment by day 28. lmmunostaining showed that mostly RPE cells, but also myofibroblasts, glial cells and collagen, were present in the newly formed rabbit PVR membranes. Interpretation: Human RPE cells cultured from a PVR membrane appear to be capable of inducing PVR in rabbits. The resultant ERMs are similar to those formed in human PVR and consist mainly of RPE cells. Contexte : Les modeles de vitreoretinopathie proliferante (VRP) les plus utilises dependent de !'injection de cellules dans le corps vitre animal. L'utilisation de
From the Departments of *Ophthalmology, t Anatomy, :j:Pathology and §Surgery, University of British Columbia, Vancouver, BC, and lithe Department of Ophthalmology, Dalhousie University, Halifax, NS
Department of Ophthalmology, University of British Columbia, 2550 Willow St., Vancouver BC V5Z 3N9; fax (604) 875-4663; jms@ interchange.ubc.ca
Originally received May 31, 2001 Accepted for publication Nov. 30, 2001
This article has been peer-reviewed.
Correspondence to: Dr. Joanne A. Matsubara, Eye Care Centre,
Can J Ophthalmol 2002;37:21 1-20
Induction of PVR-Wong et al
211
Induction of PVR-Wong et al
cellules epitheliales pigmentaires de Ia retine (EPR) provenant des membranes VRP humaine peut produire un modele plus exact de Ia VRP. Nous avons done mene une etude pour determiner si les cellules EPR derivees d'une simple membrane epiretinienne (MER) pouvaient induire Ia maladie dans l'ceil du lapin et si les composants cellulaires des MER ainsi induits etaient semblables a ceux des membranes de Ia VRP chez les humains. Methodes : On a preleve des cellules d'une MER au cours d'une chirurgie de VRP. Les cellules EPR ont ete cultivees a partir de Ia membrane puis injectees dans l'ceil droit de 24 lapins albinos de Ia Nouvelle-Zelande, l'ceil gauche servant de temoin. L'on a examine les yeux par ophtalmoscopie indirecte sur une periode de 4 semaines. Les yeux enuclees furent ensuite soumis a un examen microscopique et a une analyse histochimique. Resultats: Le ye jour, Ia VRP s'etait developpee dans 23 des 24 yeux assujettis a I' experience et, chez huit sujets, elle progressait vers un decollement tractionnel et localise de Ia retine. Le 21 e jour, le decollement tractionnel et localise de Ia retine s'etait developpe dans 17 yeux; dans un des cas, Ia maladie avait pris de l'ampleur au 28e jour. L'immunocoloration a montre que les nouvelles membranes VRP des lapins consistaient principalement en cellules EPR mais aussi en myofibroblastes, cellules gliales et collagene.
Interpretation : Les cellules EPR cultivees a partir d'une membrane VRP humaine semblent pouvoir induire Ia VRP chez les lapins. Les MER qui en resultent ressemblent a celles qui se forment dans Ia VRP chez les humains et consistent principalement en cellules EPR.
P
roliferative vitreoretinopathy (PVR) is the most common cause of failure of retinal detachment surgery. PVR is characterized by the formation of epiretinal membranes (ERMs), which contract, causing retinal tears and detachments. Over the last 30 years much effort has been devoted to developing an experimental model that closely resembles PVR. An accurate model of PVR could ultimately lead to prevention of this complication, for which there is no adequate treatment or prophylaxis. The incidence of PVR remains at approximately 5% to 10% of retinal detachments that come to surgery. 1 The most widely used models of PVR rely on injection of cells into the vitreous of animals. Cells injected include those found in PVR membranes, such as retinal pigment epithelial (RPE) cells, fibroblasts, macrophages and glial cells.2-5 The cells are usually cultured from healthy, homologous eyes or tissue. Investigators have found that RPE cells isolated from human PVR membranes have different gene expression and morphologic characteristics from healthy human RPE cells in vitroN It is believed that growth factors and cytokines in the vitreous are responsible for this transdifferentiation. s-10 For example, interleukin-6, a cytokine thought to play an important role in the pathogenesis of PVR, is produced in higher levels by RPE cells from human PVR membranes.
212
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
Therefore, using RPE cells from human PVR membranes may produce a more accurate model of human PVR. We performed a study to determine whether human RPE cells derived from a single ERM, from a patient with a retinal detachment and PVR, are capable of inducing the same disease in the rabbit eye. We also examined whether the induced ERMs had cellular components similar to those of human PVR membranes. METHODS
Explant culture
of proliferative vitreoretinopathy
membranes
A single ERM from a patient with a retinal detachment and PVR was surgically removed and placed in Ham's Fl2 medium (Gibco BRL, Grand Island, NY). The membrane was washed with Hanks' balanced salt solution (BSS), dissected into 20 pieces measuring 0.3 em by 0.3 em and placed into separate 35-mm culture dishes. Ham's Fl2 medium, 30% fetal bovine serum (Gibco BRL), penicillin G sodium (100 units/ mL) and streptomycin (100 mg/mL) were added to the dishes, which were then placed in a humidified 37°C incubator containing 5% carbon dioxide. After cell migration from the membrane explants had occurred,
Induction of PVR-Wong et al the cells were dissociated with 0.25% trypsin (Gibco BRL) and placed into culture dishes containing growth medium, consisting of Ham's F12 medium supplemented with 10% fetal bovine serum and antibiotics as described above.
Cell culture immunohistochemistry Subconfluent cultures were fixed with cold methanol/acetone (1: 1 v/v) at -20°C for 15 minutes. After fixation, the cultures were rinsed three times with phosphate-buffered saline (PBS). Their cell membranes were made permeable by treating with 1% Triton X-100 for 5 minutes, and, to prevent nonspecific antibody binding, the cultures were treated with 10% normal horse serum diluted with 1% bovine serum albumin (BSA) in PBS. The fixed cultures were then incubated with a primary antibody in BSA/PBS for 1 hour at room temperature. The primary monoclonal antibody was targetted against human pancytokeratin identifying human keratins 1, 4, 5, 6, 8, 10, 13, 18 and 19 (1:200; Sigma-Aldrich, Oakville, Ont.), a-smooth muscle actin (u-SMA) (1:50; Dako, Carpinteria, Calif.) or glial fibrillary acidic protein (GFAP) (1:2000; Dako). Bound antibody was detected by incubation with a 1:500 dilution in BSA/PBS of biotin-labelled horse antimouse IgG (Vector Laboratories, Burlingame, Calif.), followed by horseradish peroxidase conjugated to avidin. Labelled cells were visualized by means of a standard diaminobenzidine reaction. To rule out nonspecific staining, control cultures were analysed in the same way, except that the primary antibody was replaced with 1% normal horse serum in BSA/PBS. In addition, human ovarian adenocarcinoma (OVCAR-3, US National Institutes of Health) cells, which are known to express cytokeratin, were used as a positive control.''
Collagen gel contraction assay A collagen gel contraction assay was adapted from the method described by Raymond and Thompson.l2 Two groups of RPE cells were investigated: cells at passage 5 and cells at passage 24. The collagen gel was prepared using Vitrogen-100 (3 mg/mL type I bovine dermal collagen, Collagen Corp., Palo Alto, Calif.) mixed 16:1 v/v with lOX PBS and brought to neutral pH with 0.2M sodium hydroxide at 4 °C. The RPE cells were harvested from the culture dishes, centrifuged at 200g for 3 minutes and resuspended in 1 mL of Ham's Fl2/10 medium with lOmM
N-2-hydroxyethylpiperazine-N' -2-hydroxypropanesulfonic acid (HEPES) buffer. Next, 9 mL of the prepared collagen mixture was combined with 1 mL of RPE cells/medium to produce a collagen-RPE cell mixture with a cell concentration of 1.5 x 1Q5 cells/mL and a collagen concentration of 2.5 to 2. 7 mg/mL. Then 500 J1L of each collagenRPE cell mixture was placed in 6 wells of a 24-well plate. Six additional wells for each group served as controls, receiving 500 mL of collagen gel alone (no cells) at a concentration of 2.5 to 2.7 mg/mL. The plates were then placed in a 5% carbon dioxide-95% air incubator at 37°C for 15 minutes to promote solidification of the gel. On removal from the incubator, the collagen gel was separated from the side of the well with a 25gauge needle. Each well then received 1.5 mL of Ham's Fl2/10 medium, added by the drop down the side of the well. The medium was changed every 2 to 3 days by aspirating from the edge of each well, care being taken not to disrupt the gel. The wells were observed on days 1, 4 and 8, and the area of the gel was determined by measuring the longest and shortest diameters of the gel over a millimetre grid.
Retinal pigment epithelial cell preparation and injection The RPE cells used for all cell injections were of intermediate passage (approximately passage 12). Subconfluent (and thus still proliferating) RPE cells were dissociated with 0.25% trypsin and washed twice with BSS. Cell death was confirmed with trypan blue staining. Virtually all cells stained negatively for trypan blue. Approximately 1 X IQ6live cells were resuspended in 0.2 mL of BSS. The cells were prepared no more than 20 minutes before being injected into the eyes. Twenty-four New Zealand albino rabbits weighing 2.5 to 3.5 kg were used for this study. The right eye of each rabbit was injected with cells, and the left eye served as a control. Following pupillary dilation with two eyedrops containing 5% phenylephrine and 1% tropicamide, the rabbits were anesthetized with ketamine (50 mg/kg) and xylazine (10 mg/kg). A 25gauge needle was passed through the sclera 3 mm posterior to the limbus, and 0.2 mL of BSS containing approximately 1 X 1Q6 RPE cells was injected directly over the optic disc of the right eye. To prevent an increase in intraocular pressure, 0.2 mL of aqueous was removed simultaneously from the anterior
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
213
Induction of PVR-Wong et al chamber with a 30-gauge needle. For the left eye, 0.2 mL of BSS not containing RPE cells was injected. After surgery 1% chloramphenicol ointment (Vetcom Inc., Upton, Que.) was applied topically. The rabbits were examined by indirect ophthalmoscopy for inflammation during the first 6 days after cell injection, and for inflammation and PVR on days 7, 14, 21 and 28. PVR was graded according to the scheme described by Fastenberg and colleagues. 3 One animal with an ERM without retinal traction or tractional retinal detachment was observed for 3 months. On day 28, 23 of the animals were deeply anesthetized and killed with an injection of 1.5 mL of sodium pentobarbital (240 mg/mL) into an auricular vein; the remaining rabbit was killed at 3 months. All the animals were treated according to the Declaration of Helsinki, the guidelines of the Canadian Council on Animal Care and the Statement for the Use of Animals in Ophthalmic and Vision Research by the Association for Research in Vision and Ophthalmology.
visualized by means of exposure to the biotin complexed with aminoethylcarbazole (Vectastain Elite ABC Kit, Vector Laboratories). The cryosections were also counterstained with Mayer's modified hematoxylin solution. Three primary monoclonal antibodies were used on each right and left eye. The antibodies targetted against human pancytokeratin, GFAP and a-SMA. Control cryosections from both right and left eyes were subjected to an immunohistochemical assay in which the primary antibody was omitted (i.e., sections were incubated with diluent [1% normal horse serum in BSA/PBS] only) to rule out nonspecific staining. The presence of collagen was assessed with Masson's trichrome stain. 13 Human cardiac tissue, which is known to contain collagen,I4 served as the positive control. RESULTS
In vitro studies
Cell culture immunohistochemistry Development
of proliferative vitreoretinopathy
Both eyes of 13 rabbits were enucleated at 28 days and from a further rabbit at 3 months. The globes were fixed in 10% neutral-buffered formalin for 48 hours and wholly embedded in paraffin wax. Sections were made at a thickness of 12 11m and then stained with hematoxylin and eosin. Membrane histochemistry and trichrome staining To determine the cellular components of the induced ERMs, both eyes of the remaining 10 rabbits were enucleated at 28 days. The anterior segments were removed, and the eyecups were placed in TissueTek OCT (VWR Canlab, Toronto) before being immersed in liquid nitrogen. Then 8-11m sections were cut, mounted on glass slides and placed in a -85°C freezer for storage. After thawing, the sections were air-dried and fixed in acetone for 5 minutes. The cryosections were then treated with 1% BSA/PBS for 15 minutes to block nonspecific binding sites and immersed in 0.3% hydrogen peroxide. The cryosections were exposed to primary antibody for 30 minutes and then washed twice in PBS. Bound antibody was detected by incubation with 1:500 biotin-labelled horse antimouse IgG (Vector Laboratories) in PBS/BSA, followed by horseradish peroxidase conjugated to avidin. Labelled cells were
214
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
All sample cells from passage 4 cultures, passage 17 cultures and passage 24 cultures stained positively for cytokeratin (Fig. 1) but weakly for a-SMA and GFAP, comparable to staining produced when the primary antibodies were replaced with PBS (negative controls). The positive control for cytokeratin (OVCAR-3 cells) stained for cytokeratin, as expected. These findings suggested that the strong immunoreactivity for cytokeratin was due to specific binding and that the cells were of RPE origin.
Fig. !-Positive cytokeratin immunoreactivity in cells from passage 4 cultures obtained from human epiretinal membrane (ERM) (scale bar = 5 ~m).
Induction of PVR-Wong et al
100 -;:!2.
80
ctf
60
ctl Q)
40
0
a> ....
20 0 0
2
4
3
6
5
8
7
Time, d Fig. 2-Reduction in collagen gel area due to contraction mediated by retinal pigment epithelial (RPE) cells for passage 5 (•) and passage 24 (o) cells compared to control wells (no cells).
Table 1-Stage of proliferative vitreoretinopathy* induced in 24 rabbit eyest Day Stage
7
14
21
28
0
0
9 6 8 0
4
2 5 17 0
0 I 5 17
No proliferative vitreoretinopathy I
2 3
4
10 10 0
*Graded according to the stages described by Fastenberg and colleagues3 as follows: stage I, intravitreal membrane; stage 2, focal traction, localized vascular changes; stage 3, localized detachment of medullary ray; stage 4, extensive retinal detachment, total medullary ray detachment; stage 5, total retinal detachment, retinal folds and holes. tThe control eyes were normal at all time points.
Collagen gel contraction assay The reduction in collagen gel area due to RPEmediated contraction for both passage 5 and passage 24 RPE cells is summarized in Fig. 2. At day 1 there was 49% reduction with the passage 5 cells and 25% reduction with the passage 24 cells compared to the control wells (no cells added). By day 8 the collagen gel area in both experimental groups was 23% that of the control group.
Fig. 3-Fundus photographs of rabbit eyes with induced proliferative vitreoretinopathy (PVR). A: Experimental eye 7 days after cell injection, showing ERM and underlying localized tractional retinal detachment (arrows). B: Experimental eye 28 days after cell injection, showing ERM exerting retinal traction (arrows indicate elevation of the medullary ray). C: Control eyes were normal at 28 days.
In vivo studies At day 3 there was minor conjunctivitis in both the experimental and control eyes. The anterior chamber was clear. By day 7 the conjunctivitis had cleared completely in both groups, and the eyes appeared normal.
By day 7 PVR had developed in all but one of the experimental eyes; nine eyes had ERMs alone, and eight had localized tractional retinal detachments with ERMs present (Fig. 3, A, and Table 1). By day 28, 17
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
215
Induction of PVR-Wong et al
Fig. 5-Retinal folds (arrows) in experimental eye, providing evidence that ERM-induced retinal traction has occurred (hematoxylin-eosin; scale bar = IS f.Jm).
C). The experimental eye observed for 3 months remained at stage 1 (ERM); the control eye was normal at 3 months. Histopathological findings All eyes in which localized or extensive tractional retinal detachments were diagnosed on ophthalmoscopy showed the presence of an ERM (Fig. 4, top), the presence of subretinal fluid (Fig. 4, top), evidence of retinal traction (Fig. 5) and degenerated photoreceptor outer segments (Fig. 6, A) on histologic evaluation. All control eyes displayed normal retinal morphologic characteristics (Fig. 6, B), with the absence of subretinal fluid (Fig. 4, bottom). Artefactual retinal detachment occurred in the control eyes (Fig. 4, bottom) after enucleation as a result of the fixation procedure.IS These histochemical findings confirmed the ophthalmoscopic observations.
Fig. 4-Top: Experimental eye 28 days after cell injection. The presence of subretinal fluid (SF) and ERM (arrowheads) provides evidence that retinal detachment has occurred (hematoxylin-eosin; scale bar = 2 mm). Bottom: Control eye. Subretinal fluid and ERMs were not present in any control eye 28 days after injection. Arrowheads indicate artefactual retinal detachment (hematoxylin-eosin; scale bar = 2 mm).
eyes (71%) had localized tractional retinal detachments, 5 (21%) had retinal traction (Fig. 3, B), and 1 (4%) had an extensive tractional retinal detachment. None progressed to stage 5 (total retinal detachment). All control eyes were normal through 28 days (Fig. 3,
21 6
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
Membrane immunohistochemistry and trichrome staining All induced ERMs examined displayed immunoreactivity for cytokeratin (Fig. 7, A), c:x-SMA (Fig. 7, B) and GFAP (Fig. 7, C), which indicated that RPE cells, myofibroblasts and glial cells respectively were present. RPE cells and myofibroblasts were the dominant cell types. The experimental and control eyes stained for cytokeratin in the RPE layer and for c:x-SMA in the choroidal blood vessels. All negative controls, in which the primary antibody was omitted, displayed background levels of immunoreactivity.
Induction of PVR-Wong et al
Fig. 6-Photomicrographs 28 days after cell injection (hematoxylin-eosin; scale bar = 5 1Jm). A: Experimental eye shows degenerated photoreceptor outer segments (OS) (arrows), which suggests that photoreceptors lost their association with RPE monolayer before day 28. This provides additional support to the clinical findings that a retinal detachment had occurred in this eye before enucleation. B: Control retina shows healthy photoreceptor OS (arrows). IN = inner segments, ENL = external nuclear layer, EPL = external plexiform layer, INL = internal nuclear layer, IPL = internal plexiform layer, GCL = ganglion cell layer.
© Fig. ?-Immunohistochemical and trichrome staining of ERM in experimental eye with stage 3 PVR (localized detachment of medullary ray). A: At day 28, positive immunoreactivity for cytokeratin reflects the presence of RPE cells throughout the ERM {black arrow). The rabbit RPE monolayer (white arrow) also showed immunoreactivity for cytokeratin (scale bar = 25 IJm). B, C: Positive a -smooth muscle actin and glial fibrillary acidic protein (GFAP) immunoreactivity throughout ERM indicates the presence of myofibroblasts (B) and glial cells (C) respectively. The aminoethylcarbazole staining for GFAP is seen as a light aminoethylcarbazole reaction product (counterstained with hematoxylin; scale bar = 25 1Jm). 0: Positive trichrome staining indicates the presence of collagen throughout the ERM, as seen by the blue reaction product (scale bar = I00 1Jm).
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
21 7
Induction of PVR-Wong et al
Trichrome staining for collagen was seen in our ERMs (Fig. 7, D) and in the cardiac tissue control sections. INTERPRETATION
Our results suggest that human RPE cells taken from an ERM can induce PVR in a rabbit model. We observed a progression in the clinical manifestations of PVR, from the early appearance of a membrane to later signs of retinal traction and detachment. This is consistent with the picture in human PVR, although the time course is shorter than for the development of PVR after retinal detachment surgery. The short time course of PVR in our model is useful for experimental purposes, as it will allow for rapid assessment of potential therapeutic or prophylactic agents against PVR. It may also shed light on the mechanism of PVR. In our model, injection resulted in a sudden influx of about a million RPE cells into the vitreous, whereas in human PVR, RPE cells migrate more gradually through retinal tears or as a result of surgical procedures during retinal detachment surgery.16·17 More extensive retinal tears are associated with more rapid onset of PVR.18,19 The shorter time course of PVR in our model may also be explained by differences between RPE cells in PVR membranes and those in the RPE layer in the retina. RPE cells that migrate from their retinal monolayer are known to undergo a number of changes, collectively termed dedifferentiation, metaplasia or transdifferentiation.2.8-10,20 First, it is thought that RPE cells cultured from the retinal monolayer assume transdifferentiated (myofibroblast-like) characteristics and that these changes are necessary for increased contractile activities.2,9,21,22 This ability likely plays a key role in the development of tractional retinal detachment due to PVR. In this study we showed that RPE cells, cultured from the same human PVR membrane used in the intravitreal injections, were able to rapidly contract a collagen matrix. This may help explain the short time course of development of PVR in our model. Second, RPE cells isolated from PVR membranes are known to express different genes - and, thus, proteins -from those in the retinal monolayer (e.g., interleukin-6). 6 The genes are likely to code for proteins that are important in the pathogenesis of PVR, such as growth factors and cytokines and their receptors. Cytokines, such as interleukin-6, may function as chemoattractants for other cell types involved, includ-
218
CAN
J OPHTHALMOL-VOL. 37, NO. 4, 2002
ing the glial cells and macrophages seen in our model. Growth factors are also believed to play a vital role in the transdifferentiation of RPE cells and thus in PVR. 8 Indeed, our group has found that a number of growth factors and their receptors are expressed in the unique culture of RPE cells used in this study (unpublished data, 2000). This also helps explain why our model had a shorter time course than human PVR after retinal detachment surgery and provides important clues for the development of potential therapies. The RPE cells used in this study have unique qualities, which also may have contributed to the shorter time course of PVR that was observed. The RPE cells used in our model are unique in that they were derived from a single PVR membrane and were previously shown to complete several population doublings (at least 48) without showing signs of replicative senescence. It is unusual for cells cultured from human PVR membranes to survive more than a few passages. Even human RPE cells cultured de novo from donor eyes tend to become senescent after 15 to 20 passages. It is likely that the RPE cells used in this study had spontaneously transformed. RPE transformation could serve as a benefit in providing an "immortalized" culture of cells for future studies. Conversely, transformed RPE cells may not accurately reflect RPE cells found in other PVR membranes. We previously demonstrated that our RPE cells retained epithelial-like characteristics and contractile properties up to passage 24. 11 However, additional experiments should be performed to further characterize this unique culture of cells. These could include studying the expression of structural proteins and growth factors and their receptors, and evaluating the stability of both morphology and gene expression in successive population doublings. Based on our immunohistochemical characterization of the cultured cells and to the best of our knowledge, the only human cells injected into the vitreous of the rabbit eyes were RPE cells. Therefore, glial cells found in the ERMs at day 28 must have originated from the rabbit. It is likely that these glial cells invaded the vitreous through the tractional tears in the retina. It is also likely that most of the cells within the developing ERM in our model originated from the human RPE cells injected. Since the injection did not break the inner limiting membrane, rabbit cells did not have access to the vitreous at this stage; they gained access to the vitreous cavity only after the tractional retinal tears had occurred. This is an advantage of our model because therapeutic agents that tar-
Induction of PVR-Wong et al
get RPE proliferation and contraction can be tested specifically. Previous animal models of PVR focused on the injection of cells into the vitreous or on a more physical intervention, including injection of particles, gas, albumin or enzymes, and mechanical tearing of the retina. Cell injection models have involved administering whole blood, erythrocytes, platelets, leukocytes, fibroblasts or glial cells. Although some of these models have demonstrated clinical characteristics of PVR, few of the groups published histologic data. 2-4.23-32 In 1975 Mandelcorn and associates2 injected adult human RPE cells into the vitreous of the primate eye. They observed metaplasia and proliferation but were not able to induce tractional retinal detachment. In 1986 Grierson and coworkers33 injected human RPE cells, derived from either an adult or fetal cell line, into the monkey eye. They also found that the adult cells, cultured from the RPE monolayer of a donor eye, produced an insubstantial ERM and no tractional retinal detachment. However, they observed that fetal RPE cells produced extensive membranes and traction retinal detachments. These studies support the unique nature of the RPE cells used in our model. It is likely that our RPE cells already possess the characteristics of transdifferentiation, as they were able to induce traction retinal detachments. These characteristics logically include active proliferation and expression of the necessary structural proteins and growth factors and their receptors. Like the models proposed by Mandelcorn and associates2 and Grierson and coworkers,33 our model involved the injection of xenogenic cells (human RPE cells) into the vitreous cavity of the rabbit eye. After surgery, we observed minor inflammation in both the experimental and control eyes, which resolved completely by day 4; this suggested that the inflammation was due to the surgery itself and not to the human cells. Furthermore, in one eye we allowed an ERM, which did not cause a tractional retinal detachment, to survive for 3 months. There was no evidence of ocular inflammation throughout this period. The lack of inflammation is consistent with anterior chamber-associated immune deviation (immune privilege in several parts of the eye, including the vitreous cavity).34 Owing to this immune privilege, a systemic immune response can be induced without causing ocular inflammation, as the mediators of immunogenic inflammation delayed hypersensitivity and complement fixation are quenched. Our data illustrate that in our model,
tractional retinal detachment occurred owing to ERM contraction, not inflammation. SUMMARY
We found that human RPE cells cultured from a patient with PVR are capable of inducing the same disease in the rabbit eye. Our model is unique and has distinct advantages that make it an excellent tool for studying the pathogenesis of, and potential treatments for, PVR. We thank Dr. Cal Roskelly for his generous gift of the human ovarian carcinoma cells; Howard Meadows, Dr. Zheng Chen, Eleanor To, Dawn Lam and Tara Stewart for their expert technical advice and assistance; Virginia Booth, Dr. Farzin Forooghian and Dr. Peter Leong-Sit for their advice and assistance with the ocular injections; and Drs. Wayne Yogi and John Church for their helpful comments and suggestions. REFERENCES
1. Retina Society Terminology Committee. The classification of retinal detachment with proliferative vitreoretinopathy. Ophthalmology 1983;90:121-5. 2. Mandelcom MS, Machemer R, Fineberg E, Hersch SB. Proliferation and metaplasia of intravitreal retinal pigment epithelium cell autotransplants. Am J Ophthalmol1975 ;80: 227-37. 3. Fastenberg DM, Diddie KR, Dorey K, Ryan SJ. The role of cellular proliferation in an experimental model of massive periretinal proliferation. Am J Ophthalmol 1982;93: 565-72. 4. Peters MA, Burke JM, Clowry M, Abrams GW, Williams GA. Development of traction retinal detachments following intravitreal injections of retinal Miiller and pigment epithelial cells. Graefes Arch Clin Exp Ophthalmol 1986; 224:554-63. 5. Hui YN, Sorgente N, Ryan SJ. Posterior vitreous separation and retinal detachment induced by macrophages. Graefes Arch Clin Exp Ophthalmo/1984;225:279-84. 6. AbeT, Durlu YK, Tarnai M. The properties of retinal pigment epithelial cells in proliferative vitreoretinopathy compared with cultured retinal pigment epithelial cells. Exp Eye Res 1996;63:201-10. 7. Hiscott PS, Grierson I, Hitchins CA, Rahi AH, McLeod D. Epiretinal membranes in vitro. Trans Ophthalmol Soc UK 1983;103:89-102. 8. Vidaurri-Leal J, Hohman R, Glaser BM. Effect of vitreous on morphologic characteristics of retinal pigment epithelial cells. A new approach to the study of proliferative vitreoretinopathy. Arch Ophthalmol1984; 102:1220-3. 9. Casaroli-Marano RP, Pagan R, Vilaro S. Epithelial-mesenchymal transition in proliferative vitreoretinopathy: intermediate filament protein expression in retinal pigment
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
219
Induction of PVR-Wong et al
10.
11.
12.
13.
14.
15.
16.
17. 18.
19. 20.
21.
22.
23.
epithelial cells. Invest Ophthalmol Vis Sci 1999;40: 2062-72. Kirchhof B, Sorgente N. Pathogenesis of proliferative vitreoretinopathy. Modulation of retinal pigment epithelial cell functions by vitreous and macrophages. Dev Ophthalmoll989;16:1-53. Wong CA, Cui JZ, Matsubara JA, Retina Group of UBC. Human RPE cells derived from a PVR membrane that retains epithelial characteristics and contractile properties at high passage numbers [abstract]. Invest Ophthalmol Vis Sci 1999;40:S463. Raymond MC, Thompson JT. RPE-mediated collagen gel contraction. Inhibition by colchicine and stimulation by TGF-beta. Invest Ophthalmol Vis Sci 1990;31: 1079-86. Burkitt HG, Young B, Heath JW. Notes on staining techniques. In: Killgore J, editor. Wheater's basic histology. 3rd ed. Edinburgh: Churchill Livingstone; 1993. p. 400. Burkitt HG, Young B, Heath JW. Circulatory system. In: Killgore J, editor. Wheater's basic histology. 3rd ed. Edinburgh: Churchill Livingstone; 1993. p. 140-52. Margo CE, Lee A. Fixation of whole eyes: the role of fixative osmolarity in the production of tissue artifact. Graefes Arch Clin Exp Ophthalmol1995;233:366-70. Charteris DG. Proliferative vitreoretinopathy: pathobiology, surgical management, and adjunctive treatment. Br J Ophthalmoll995;79:953-60. Pastor JC. Proliferative vitreoretinopathy: an overview. Surv Ophthalmoll998;43:3-l8. Bonnet M. Clinical factors predisposing to massive proliferative vitreoretinopathy in rhegmatogenous retinal detachment. Ophthalmologica 1984;188:148-52. Scott JD. Equatorial giant tears affected by massive vitreous retraction. Trans Ophthalmol Soc UK 1976;96:309-12. MUller-Jensen K, Machemer R, Azarnia R. Autotransplantation of retinal pigment epithelium in intravitreal diffusion chamber. Am J Ophthalmoll975;80:530-7. Vidaurri-Leal JS, Glaser BM. Effect of fibrin on morphologic characteristics of retinal pigment epithelial cells. Arch Ophthalmoll984;l02:1376-9. Grisanti S, Guidry C. Transdifferentiation of retinal pigment epithelial cells from epithelial to mesenchymal phenotype. Invest Ophthalmol Vis Sci 1995;36:391-405. Algvere P, Wallow IH, Martini B. The development of vit-
220
CAN J OPHTHALMOL-VOL. 37, NO. 4, 2002
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
reous membranes and retinal detachment induced by intravitreal carbon microparticles. Graefes Arch Clin Exp Ophthalmoll988;226:471-8. Frenzel EM, Neely KA, Walsh A W, Cameron JD, Gregerson DS. A new model of proliferative vitreoretinopathy. Invest Ophthalmol Vis Sci 1988;39:2157-64. Hitchins CA, Grierson I, Hiscott PS. The effects of injections of cultured fibroblasts into the rabbit vitreous. Graefes Arch Clin Exp Ophthalmol1985;223:237-49. Hsu HT, Dorey K, Sorgente N, Ryan SJ. Surgical removal of the vitreous. Its effect on intraocular fibroblast proliferation in the rabbit. Arch Ophthalmol1984;102:605-7. Lean LS, Van der Zee WAM, Ryan SJ. Experimental model of proliferative vitreoretinopathy (PVR) in the vitrectomised eye: effect of silicone oil. Br J Ophthalmol 1984;68:332-5. Miller B, Miller H, Ryan SJ. Experimental epiretinal proliferation induced by intravitreal red blood cells. Am J Ophthalmol 1986; 102:188-95. Shabo AL, Maxwell DS. Experimental immunogenic proliferative retinopathy in monkeys. Am J Ophthalmol 1977;83:471-80. Topping TM, Abrams GW, Machemer R. Experimental double-perforating injury of the posterior segment in rabbit eyes; the natural history of intraocular proliferation. Arch Ophthalmoll979;97:735-42. Machemer R, Laqua H. Pigment epithelium proliferation in retinal detachment (massive periretinal proliferation). Am J Ophthalmoll975;80: 1-23. Machemer R, Van Horn D, Aaberg TM. Pigment epithelial proliferation in human retinal detachment with massive periretinal proliferation. Am J Ophthalmol 1978;85: 181-91. Grierson I, Boulton M, Hiscott P, Hitchins C, Gilbert D, McLeod D. Human retinal pigment epithelial cells in the vitreous of the owl monkey. Exp Eye Res 1986;43: 491-502. Jiang LQ, Streilein JW. Immune privilege extended to allogenic tumor cells in the vitreous. Invest Ophthalmol Vis Sci 1991;32:224-8.
Key words: proliferative vitreoretinopathy, human retinal pigment epithelial cells, rabbit model, retinal detachment