JOURNAL OF CELLULAR PHYSIOLOGY 143:196-203 (19901

Propagation of Fetal Human RPE Cells: Preservation of Original Culture Morphology After Serial Passage MI-KYOUNG SONG AND GE MlNC LUI*

Department of Ophthalmology, University of California San Francisco, San Francisco, California 94 143 The permissive effects of extracellular matrix (ECM) on in vitro growth and differentiation of fetal human retinal pigment epithelial (RPE) cells have been studied. Factors which enhanced the effect of ECM to support cell division were also examined, including growth factors, culture media, and serum requirement. Under the specific culture conditions we have defined, it is possible to propagate these RPE cells at low density (less than 20 cellslmm2)with excellent growth properties for greater than 72 doublings (fourteen passages) in serial culture. Later-passaged cells maintained the morphological appearance of early-passaged cultures. ECM produced by bovine corneal endothelial cells was by far the most predominant factor in promoting rapid cell proliferation and viability over repeated passaging. Basic fibroblast growth factor (bFGF) exerted a substantial effect on the rate of cell division at different serum concentrations on plastic dishes. In addition, this factor showed profound synergistic effect when RPE cells were maintained on ECM, both in the preservation of cell morphology and also in long term viability. Other growth factors, such as epidermal growth factor (EGF) and transforming growth factor-beta (TCF-B), were also tested, but EGF effects were less prominent than those observed with bFGF, and TGF-€3 had an inhibitory effect at high concentrations.The ability to obtain a relatively large number of human RPE cells in vitro which preserve the appearance of early passage cells may provide useful opportunitiesto study the physiological properties and pathological alterations involving this important cell type. Retinal pigment epithelium (RPE) plays a vital role in maintaining viability of the neural retina. It also fulfills various physical and metabolic functions which contribute to the well being of the visual system (Zinn and Benjamin-Henkind, 1979). Diseases and dysfunction of the RPE may result in irreversible loss of vision. The ability to serially culture human RPE cells, retaining in vivo biological functions such as pigmentation, phagocytic behavior, and fluid transport capability while preserving their morphology, would be extremely valuable since it may provide a sufficient amount of cells to study the biochemical nature of the RPE. Subsequently, the tissue culture system can be extended t o elucidate the mechanisms of various disease states such as macular degeneration and proliferative vitreous retinopathy. Various investigators have successfully established pure primary cultures of human RPE cells by dissection (Albert et al., 1972; Flood et al., 19801, vacuum suction (Del Monte and Maumenee, 19801, and by enzymatic dissociation (Mannaugh et al., 1973). RPE explants have also been maintained in culture for extended periods of time (Nicolaissen et al., 1982). Despite these initial successes, there are only a few reports of attempts to culture non-transformed human RPE cells in continuous passages at clonal density (Aronson, 1983). Under standard culture conditions, GI 1990 WILEY-LISS, INC

RPE cell division has been observed to be quite limited, with alterations in the cells such as depigmentation and some scant morphological changes occurring with successive passage, especially at high split ratios. We report that use of an extracellular matrix (ECM) substrate and basic fibroblast growth factor (bFGF) markedly enhances the ability to propagate human fetal RPE cells and prolongs their viability in long-term serial culture even when plated at clonal densities. Since certain culture media and serum requirements may be vital in maximizing the RPE cells’ response to ECM substrates and growth factors, we studied in detail the interaction of these individual factors for their potential value in the propagation of human RPE cells.

MATERIALS AND METHODS Dulbecco’s modified Eagle’s media (DME H16, DME H21), Medium 199 with Earle’s Balanced Salt Solution (M199 EBSS), Ham’s F12, RPMI 1640, MCDB 170, Eagle’s minimum essential medium (MEM), phosphate buffered saline (PBS), phosphate buffered saline calcium and magnesium free (PBS-CMF), STV solution,

Received August 14, 1989; accepted December 27, 1989. *To whom reprint requestdcorrespondence should be addressed.

HUMAN RETINAL PIGMENT EPITHELIAL CELL PROPAGATIONS

and Pen-Strep were obtained from Gibco (Grand Island, NY). Fetal calf serum (FCS) and calf serum (CS) were obtained from Sterile Systems, Inc. (Logan, UT). Tissue culture plates were from Falcon Plastics (Lincoln Park, NJ). Fungizone was from Squibb (Princeton, NJ). Glutamine, dextran, gentamicin, gelatin, and fibronectin were from Sigma Chemical Co. (St. Louis, MO). Basic FGF (bFGF) and acidic FGF (aFGF) were purified from bovine pituitary and brain (Gospodarowicz, 1975; Gospodarowicz et al., 1976; Gospodarowicz et al., 1984). Human recombinant epidermal growth factor (hEGF) was purchased from Amgen (Thousand Oaks, CAI. Laminin was from Collaborative Research, Inc. (Bedford, MA). Transforming growth factor-beta was obtained from R & D systems, Inc. (Minneapolis, MN).

RPE CELL CULTURE Four- to five-month fetal eyes were obtained following therapeutic abortion and stored refrigerated in McCarey-Kaufman media. Dissections were performed within 12 hours in a laminar flow hood. The exterior of the eyes was rinsed with Neosporin and a n equatorial cut was made. The vitreous was removed and the retina was excised. Tissue containing RPE cells was dissected free from choroidal cell contamination under a Zeiss operating microscope. The RPE tissues were transferred to 35 mm ECM-coated plates containing M199 EBSS supplemented with 15%FCS. After primary outgrowth of RPE cells was observed, the media were changed, and bFGF (1 ng/ml) was added every other day. The specimen was maintained without disturbance until a t least 300-500 cells had migrated from the tissue. The cells were dissociated from the dish by STV solution (0.05% trypsin, 0.02% EDTA in 0.9% NaC1, pH 7.4), and enzyme action was neutralized by serum-supplemented media. Cell number was determined in a Coulter Counter, and the cells were seeded onto new ECM-coated plates for further propagation. Subsequent cultures of RPE cells were maintained in DME H16 supplemented with 15%FCS, 300 pg/ml glutamine, 50 pg/ml gentamicin, 100 U/ml Pen-Strep, and 2.5 pg/ml fungizone. bFGF a t concentration of 1 ng/ml was added to the culture every other day. Confluent plates of RPE cells were either passaged at a split ratio of 1:64 for experimental studies or frozen in liquid nitrogen for future stock. PREPARATION OF ECM Steer eyes were obtained from a local slaughterhouse and transported on ice. Primary cultures of bovine corneal endothelial (BCE) cells were established a s previously described (Gospodarowicz and Ill, 1980b) by gently scraping the endothelial surface of the cornea with a groove director and maintaining them in DME H16 supplemented with 10% FCS, 5% CS, 300 pg/ml glutamine, 50 pg/ml gentamicin, 100 Ulml Pen-Strep, and 2.5. pg/ml fungizone. bFGF (1 ng/ml) was added every other day. For the preparation of ECM-coated dishes, confluent stock plates of BCE cells were subcultured in serum supplemented media with 3% dextran. Upon confluency BCE cells were treated with 0.02 M ammonium hydroxide for 5 minutes and washed five times with

197

PBS. The ECM-coated plates were stored at 4°C in PBS.

CELL GROWTH CONDITIONS The influence of culture media on differentiation of RPE cells was reported by Israel e t al. (1980). Interaction between substrate and culture media on cell growth was investigated by Gospodarowicz and Lui (1981b). These reports prompted us to screen various media for their ability to promote optimum cell growth. DME H16, DME H21, M199 EBSS, RPMI 1640, Ham's F12, MCDB 170, MEM, and mixture of H16/F12 were supplemented with 10% FCS, and RPE cells were seeded at a n initial density of 2 x lo4 cells per 35 mm plastic or ECM-coated plates (or 20 cells/mm2) in the presence or absence of bFGF (1 ng/ml). Media were changed every other day. Triplicate plates of each condition were counted in a Coulter Counter twelve days after seeding. The effect of varying concentrations of FCS ranging from 1to 30%on RPE cell growth was examined using DME H16 medium. As in the previous experiment, 2 x lo4 cells were seeded on 35 mm plastic (20 cells/mm2) or ECM-coated plates in the presence or absence of bFGF (1ng/ml). Triplicate plates were counted a t day 6 in a Coulter Counter. To compare the permissive effect of different substrates on cell proliferation, RPE cells were plated on 35 mm plastic, gelatin-coated, fibronectin-coated, laminin-coated, and ECM-coated dishes. Gelatinized plates were prepared by incubating the dishes with 0.5% gelatin in PBS-CMF for two hours at 4°C. Fibronectinand laminin-coated plates were made by coating 100 pg of the respective material constituted in water per 35 mm plate for 2 hours a t 4°C. Cells were plated on different substrates at a n initial density of 20 cells/ mm2 and maintained in DME H16 supplemented with 15% FCS with or without bFGF (1 ng/ml). Cell counts were done on Day 16. Since recent reports indicated that RPE cells produce bFGF in vitro (Schweigerer et al., 19871, the response of RPE cells to various growth factors was examined. The dose respone of RPE cells to highly purified bFGF (10-100,000 pg/ml), aFGF (100-1,000,000 pg/ml), EGF (10-10,000 pglml), and TGF, (10-10,000 pg/ml) was carried out on 35 mm ECM-coated dishes seeded with 20,000 cells. Triplicate plates were counted on day 9 with a Coulter Counter. CELL GROWTH MEASUREMENT The effect of various substrates and bFGF on the rate of RPE cell division was demonstrated in the growth study. RPE cells were seeded on 35 mm plastic and ECM-coated plates at a density of 20 cells/mm2 in DME H16 supplemented with 15% FCS. Media was changed every other day and 1ng/ml bFGF was added to half of each set of substrates. Triplicate plates of each condition were counted every other day for 2 weeks. The ability of RPE cells to undergo long-term serial culture was studied by repeatedly passaging the cells on plastic and ECM-coated plates in the presence or absence of bFGF (1 ng/ml). For this study, primary cultures of RPE cells were seeded a t initial density of 20 cells/mm2 on quadruplicate plates for each condition. After seven days, duplicate plates of each condi-

198

SONG AND LUI

“T

0- FGF

(D

I

3.0

0 7

X

L11

2.1

W

m I 3 z

2.0

1 1 W

u

1.5

1.o

n

*

m

c

a

Q

A

l

l

N

~

w

m

By

Fig. 1. Effects of substrate and bFGF on the proliferation of fetal human RPE cells. Cells were seeded at 2 x lo4cells per 35 mm dishes containing 2 ml of medium (DME H16) supplemented with 15% FCS. bFGF (shaded bar) was added every other day to half of the experimental dishes at concentration of 1 ng/ml. Cells were counted a t day 16 and the average of triplicate plates were presented. Standard deviation was all within 10% of the average and is indicated by error bars in the diagram.

tion were counted to assess the number of generations. One dish of cells from each of the four conditions was used for replating under the same conditions in the next passage. The remaining plate was maintained for an additional three weeks to note any changes in cell morphology and the occurrence of repigmentation. This procedure was performed over a period of 100 days and pictures of the cells were taken with a Nikon inverted phase contrast microscope equipped with an Olympus automatic camera. RESULTS Response of fetal retinal pigment epithelial cells to various substrates, media and increasing serum concentrations with or without bFGF The importance of substrate on cell growth has been reported previously (Gospodarowicz, 1981a). When fetal RPE cells were exposed to plastic, gelatin-coated, fibronectin-coated, laminin-coated, and ECM-coated dishes, the ECM-coated dishes were superior to the Fig. 2. Morphology of fetal human RPE cells maintained on different other conditions in promoting cell proliferation (shown substrates. Sixteen days old cultures of low initial seeding density (2 in Fig. 1). Shortly after reaching confluency, the cells x 104/35mm dish) were maintained on plastic (P),gelatin (G),lamion ECM-coated dishes attained a columnar epithelial nin (L),fibronectin (F), and ECM-coated dishes (El.Cells were grown configuration with appearance of pigmented granules in DME-HI6 with 15% FCS supplement, in the presence or absence of after one week. Cells maintained on fibronectin-coated bFGF (1ngiml). Phase contrast microscopy, x 100. and laminin-coated dishes achieved nearly the same growth rate as those on the ECM-coated dishes, but failed to achieve the epithelial-like morphology exhib- that Medium 199, DME H16, and DME H21 supported ited by cells maintained on ECM-coated dishes (Fig. 2). active growth of RPE cells on plastic plates in the presFetal RPE cells on plastic and gelatin-coated plates ence of bFGF (Fig. 3A). Dulbecco’s minimum essential showed slower growth rate. At confluency, they adopt- medium and the mixture of DME H16/Ham’s F12 meed a larger and more flattened shape with irregular dia were sufficient, although not optimal, for maintaincell borders, and failed to form the tall columnar con- ing cell proliferation, while Ham’s F12, RPMI 1640, figuration even long after confluency (Fig. 2). Addition and MCDB 170 were clearly lacking in their ability to of bFGF induced faster cell proliferation and improved support RPE cell growth. With the combination of cell morphology regardless of the substrate tested. The bFGF and ECM, all the media except MCDB 170 (i.e., combination of bFGF and ECM appeared to further M199, DME H16, DME H21, MEM, RPMI 1640, FP2/ enhance growth and subsequent morphological differ- H16 and F12) showed almost equivalent propensity to entiation of the cells at confluency. promote RPE cell growth (Fig. 3B). Among all test,ed Systematic screening of eight culture media showed media, both M199 and DME H16 appeared to preserve

199

HUMAN RETINAL PIGMENT EPITHELIAL CELL PROPAGATIONS

A

T

PLAST1C

IJ

[r

W

3.0

m I

t

3

z

YlW

‘3

T

LEY

Fl2

HlI/F12

llRl

UCOE

HlI

HZ1

yD1

Fl2

HlI/FlZ

WM

YUI)

HlI

HZ1

ECM

CO

3.3

3.0

2.5

LO

1.5

1 .o

YlOe

Fig. 3. Effects of different media and bFGF on the proliferation of fetal human RPE cells maintained on plastic and ECM-coated dishes. RPE cells were seeded at 2 x lo4 cells per 35 mm dishes containing 2 ml of medium supplemented with 10% FCS. A shows the response of RPE cells to various media when they were maintained on plastic culture plates with 1 ng/ml of bFGF added every other day. B com-

pares the response of RPE cells to various media on ECM-coated plates, also with the adhtion of 1 ngiml of bFGF every other day. Cells were counted a t day 12 and the average of triplicate plates were presented. Standard deviation was within 10% of the average in all cases and is indicated with error bars in the figure.

the cell morphology most effectively. Nevertheless, cells maintained in DME H16 regained their pigment a t the earliest time (2 weeks post-confluency). DME H16 was subsequently used for further growth studies. Serum effects on RPE cell division are shown in Figure 4. RPE cells maintained on plastic dishes without bFGF exhibited a linear dose response to increasing concentrations of fetal calf serum. At serum concentrations greater than 25 percent, the rate of cell proliferation plateaued at 850 cells/mm2.However, addition of bFGF to RPE cultures allowed cell proliferation in 10%

serum to the same cell density (850 cellslmm’). ECM in the absence of bFGF allowed cell proliferation in only 10% FCS, to a cell density of 950 cells/mm2, even higher than the condition in which cells were maintained on plastic plates plus bFGF. When bFGF was combined with ECM, only 5% fetal calf serum was needed to yield the same cell number. In comparison to plastic dishes without bFGF, cells exposed to bFGF under the same conditions reached a higher cell density (1,100 cells/mm2) 6 days after seeding. When ECM was used in the absence of bFGF, further increase in satu-

200

SONG AND LUI 0-0

0-0

PLASTIC PLASTIC

- FGF + FGF

ECM ECM

A-A A-A

- FGF

0-0

+ FGF

BFGF

0-0

AFGF

W

I

0 c

X

K W

m 2 3

Z

1 J W

0

0

a

10

15

m

w

a0

% SERUM CONCENTRATION

Fig. 4. Influence of varying concentrations of fetal calf serum on human fetal RPE cell growth. The stimulation of RPE cell growth by increasing concentration of FCS (0.5,1,5,10,15,20,25,and 30%)was tested on ECM-coated and non-coated dishes in the absence and presence of bFGF (1 ngiml). 2 x lo4 cells were seeded per 35 mm dish containing 2 ml of medium and the appropriate amount of FCS. Experiment was terminated after 6 days and triplicate dishes of each condition were counted. The result represents the average of the three counts with standard deviation falling within 10% of the mean, as indicated by error bars in the graph.

B ul

-

I

0

x

w [L

m I 3 z J W i

0

4

lo2

101

rating cell density was observed (1,200 cells/mm2).The combination of bFGF and ECM acted synergistically on RPE cells, allowing maximal response to 15% FCS stimulation to achieve a final cell density of 1,400 cells/ mm2. Response of fetal retinal pigment epithelial cells to growth factors To investigate further the observed growth factor effects on our RPE cultures, increasing concentrations of bFGF as well as other growth factors were examined for potential stimulatory or inhibitory activities. Figure 5A shows the response of RPE cells to basic bFGF and acidic bFGF in a log-dose manner. On ECM-coated dishes, basic bFGF reached maximum stimulatory effect between 1and 5 ng/ml. Acidic bFGF was much less potent than basic bFGF and only achieved maximal effect at 500-1,000 ng/ml. In comparison, EGF (Fig. 5B) was less effective than bFGF, but more potent than aFGF, in promoting cell proliferation with optimal stimulation at 50-100 ng/ml. TGFB(Fig. 5C) displayed significant inhibitory effect a t high doses (>500 pg/ml); at low doses (10 pg/ml) the inhibitory effect was not observed.

103

1d4

I o5

EGF CONCENTRATION (pg/ml)

0

7 t

\

I

101

102

103

1o4

TGF-8 CONCENTRATION (pg/ml)

Fig. 5. Mitogenic activity of different growth factors on human fetal RPE cells. Low density cultures of RPE cells (2 x lo4 cells/35 mm dish) were seeded in 2 ml of DME-HI6 medium supplemented with 5% FCS on ECM-coated dishes. Growth factorb) at various concentrations as indicated in the graph were added every other day. Experiments were terminated on day 9 and triplicate plates were counted. Each point represents the average of three counts with standard deviations less than 10%of the mean, as shown by the error bars in each graph. A shows the stimulatory activity of basic (bFGF) and acidic (aFGF) fibroblast growth factor with range for 10' to lo5 pglml and 10' to lo6 pg/ml, respectively. B represents the activity of EGF tested (ranging from 50 to 1 X lo5 pg/ml). C shows the inhibitory effect of TGF, (10' to lo4 pg/ml) tested alone on ECM-coated plates without the addition of other growth factors. The initial density of cells seeded for the TGF, test was 5 x lo4 celld35 mm dish in order to show the inhibiting effect.

Comparison of the growth rate of low density retinal pigment epithelial cells maintained on plastic or ECM-coated dishes The effect of the ECM substrate on RPE cell proliferation compared to plastic substrate was evaluated with and without bFGF supplement. As shown in Figure 6, cultures maintained on plastic dishes exhibited doubling times of 22 hours during their exponential growth phase until they plateaued at 1,300 cells/mm2 mation of a more tightly packed monolayer by the cel'ls, on day 10. Addition of bFGF to the culture medium resulting in higher final cell density (2,000 cells/mm2). When cells were maintained on ECM-coated dishes improved the rate of proliferation and shortened the doubling time to 18 hours. bFGF also facilitated for- in the absence of bFGF, a maximal growth rate of 16-

201

HUMAN RETINAL PIGMENT EPITHELIAL CELL PROPAGATIONS

6 4x10

W

m I

0-0

PIASTIC-FGF PLASTIC+FGF

4-b

A-a

I

3 2x106

.--.

PLASTIC - FGF PLASTIC FGF AA ECM - FGF A-A ECM FGF

ECM-FGF ECM+FGF

0-

4-/

4

3x10

LT

-

0-0

+

+

i 0

0

Mll

2

4

6

a

10

12

14

DAYS Fig. 6. Proliferation of low density human fetal RPE cells maintained on plastic or ECM-coated dishes as a function of time and exposure to bFGF. Cultures were seeded at 2 x lo4 cells per 35 mm dishes containing 2 ml of DME-H16 medium supplemented with 15% FCS with or without bFGF. bFGF was added every other day at concentration of 1 ngiml. Triplicate sets of plates representing each conditions (P-F, P + F, ECM-F, and ECM + F ) were counted every other day for 14 days in total. Results were the average of three counts with less than 10% standard deviation and is indicated by error bars in the graph.

u

0 0

17 hours per doubling was observed. In this case, cells were able to attain a much higher final cell density (3,000 cells/mm2)and the culture was composed of nonoverlapping, distinctly smaller and tightly packed cells containing pigment granules. The doubling time for RPE cells on ECM-coated dishes in the presence of bFGF was about 15-16 hours, but the final cell density did not significantly differ from those maintained on ECM alone.

Serial culture of fetal retinal pigment epithelial cells on plastic and ECM-coated dishes in the presence and absence of bFGF The influence of bFGF and ECM on the serial passage of RPE cells was investigated. As shown in Figure 7, the effects of bFGF and ECM were again apparent under these test conditions. When RPE cells were maintained on plastic substrate in the absence of bFGF, the passaged RPE cells initially went through rapid division until the sixth passage. However, under this condition RPE cells progressively lost their capacity to divide and their lifespan failed to exceed 36 generations. Cell morphology also tended to deteriorate early on. The presence of bFGF improved viability of the passaged cells on plastic substrate for a few more passages, and the cells were observed to maintain cell divisions for up to 45 generations, although at a substantially slower rate. The epithelial-like morphology of the cell was retained for a longer period of time (up to 40 generations), but underwent rapid degeneration thereafter. In both cases, RPE cells failed to regain their pigment after the fourth passage, even when retained in culture for an extended period of time. Long-term viability of the RPE cells was markedly enhanced when cells were maintained on ECM-coated dishes. The cells displayed an active rate of prolifera-

7 14 21 28 35 42 49 56 63 70 77 84 91 98

DAYS Fig. 7. Long-term passage of fetal human RPE cells on plastic or ECM-coated dishes in the presence and absence of bFGF. 2 x lo4 cells from primary RPE culture were seeded in 35 mm dishes (plastic or ECM-coated) containing 2 ml of DME-H16 medium plus 15% fetal calf serum. Four plates were prepared for each culture condition (P-F, P + F , ECM-F, and ECM + F). Medium was changed every other day and bFGF added to the designated dishes immediately after each medium change. After 7 days triplicate plates of each condition were counted to obtain the cell number for determining the number of generations that the cell went through. The fourth dish was used to seed the stock plates for the next passage. The procedure was repeated every seven days. (Roman numerals represent the cell passage number.)

tion for up to 14 passages with no signs of cellular senescence. When both bFGF and ECM were present, RPE cells showed a faster rate of growth resulting in more cell doublings in fourteen passages than cells maintained on ECM alone. Cell morphology on ECM at late passages retained a normal differentiated epithelial appearance, with hexagonal cell borders and highly developed cytoplasmic pigment granules (Fig. 8). Therefore, the ability to obtain potentially useful RPE cells in serial culture appears to be considerably enhanced when both ECM and bFGF are employed together.

DISCUSSION The studies reported here demonstrate the feasibility of extracellular matrix in providing a favorable environment supporting human fetal RPE cell in uitro growth and differentiation in long-term serial culture. Much work has been done in the establishment of human adult RPE primary culture in vitro (Mannaugh et al., 1973; Flood et al., 1980; Albert et al., 1972; and

202

SONG AND LUI

Fig. 8. Morphological appearance of second (P,) and sixth (P,) passage fetal human RPE cells maintained as described in Figure 7. Fourteen-day-oldculture seeded at low density (2 x lo4 cells/35 mm dish) were grown on either A) Plastic-F, B) Plastic+F, C) ECM-F, and D) ECM + F and exposed to 2 ml of DME-H16 medium plus 15% fetal calf serum. bFGF was added to appropriate dishes every other day. Phase contrast microscopy, x 100.

Del Monte et al., 1980), and some efforts have been devoted to long-term culture of these cells (Aronson, 1983; Nicolaissen et al., 1986; and Nicolaissen et al., 1989). Despite the success in establishing human RPE cultures, inability to sustain a constant rate of cell proliferation and to maintain their fully differentiated phenotype over repeated passages is still an obstacle. The most noted phenomenon has been the decreased capacity of these cells to regenerate pigment granules after continuous subculturing (Whittaker, 1967; Boulton and Marshall, 1985). An attempt to define conditions allowing serial prop-

agation of human RPE cells which retain their differentiated appearance, even after rapid cell division, should help in the investigation of this cell type. The role of extracellular matrix and growth factor in controlling cell growth and morphogenesis in tissue culture models has been previously described by Gospodarowicz (1982) and Gospodarowicz et al. (1981a) in other cell types. Interestingly, RPE cells were themselves reported to secrete growth factor (Schweigerer et al., 1987), and may exert autocrine control of their own growth. The modulation of cell shape and the orientation of cell contact among themselves and with the basal surface has been an area of active research (Gospodarowicz and Tauber, 1980~).Retinal pigment epithelium in vivo consists of a monolayer of cuboidal cells resting on Bruch’s membrane, which serves numerous functions in supporting the well-being of the neural retina. It is therefore conceivable that preservation of the in vivo cuboidal configuration may be essential to RPE cells in vitro. Our results show that maintaining RPE cells on an ECM substrate allows the retention of their tall columnar configuration soon after confluency, and this continue after repeated subculturing at low seeding density (2 x lo4 cell435 mm dish). In contrast, although the columnar shape was preserved in early passages, RPE cells grown on plastic culture dishes rapidly lost their cuboidal configuration and became squamous after four to five passages. This alteration of cell configuration correlated with the decrease in rate of cell proliferation and eventually led to senescence. In both cases the cells retained apical-basal polarity with the presence of apical villi as well as highly developed intracellular junctions (desmosomal junctions and tight junctions) which are specific markers for epithelial cells (Lui et al., submitted for publication, 1990). Another possible mechanism by which ECM may modulate RPE cell growth is by enhancing cell responsiveness to nutrients and factors present in the serum. Previous study has demonstrated the permissive effect of ECM in enabling cultured endothelial cells to respond to the mitogenic effects of plasma as well as serum (Gospodarowicz and Ill, 1980a). In our system, ECM lowers the serum concentration required for optimal RPE cell growth on plastic dishes from 25% to lo%, even in the absence of supplemental growth factors. There is evidence that a t least one growth factor, bFGF, is associated with extracellular material deposited by corneal endothelial cells, and this ECM-bound bFGF was shown to retain biological activity (Vlodavsky et al., 1987). Therefore the intrinsic effect of ECM in promoting RPE cell growth in lieu of a higher serum concentration requirement may be due to the association of one or more of these mitogens to the matrix. The cell-attachment proteins fibronectin and laminin are also present in ECM. We found that RPE cells seeded on fibronectin- and laminin-coated dishes could also attain a high rate of cell proliferation, although to a lesser degree than ECM. Growth factors such as bFGF and EGF also appeared to exert a mitogenic effect on RPE cells even in the presence of ECM. Both growth factors stimulated RF’E cell growth in a log-dose manner. In our culture system bFGF was shown to be five times more potent than

HUMAN RETINAL PIGMENT EPITHELIAL CELL PROPAGATIONS

EGF (Fig. 5A,B). Moreover, when RPE cells were simultaneously exposed to both ECM and bFGF, a synergistic effect was observed in both enhanced growth rate and improvement of cell morphology. This synergistic effect was also manifested in the ability of the cells to undergo long-term serial culture (up to 14 passages) without showing signs of senescence (Fig. 7). Since TGF-B was also detected in the vitreous (Lansing et al., 19891, we preliminarily investigated the effect of this factor on the growth of human RPE cells. Studies on other cell types showed that TGF-B acted as an inhibitor (Schroder et al., 1986; Takehara et al., 1987; and Bensaid et al., 1989a). Our results also demonstrated that TGF-B inhibited RPE cell growth in a dose-related manner. Whether this inhibitory effect resulted from direct antagonism of bFGF stimulation in RPE cells is presently undetermined. Furthermore, TGF-B was also highly effective in inhibiting RPE cell growth on the ECM. There are important implications for the ability to establish in vitro models which duplicate the in vivo environment in its capability to support growth and differentiation of fetal human RPE cells. Besides offering the advantage of an unlimited supply of cultured human RPE cells for functional characterization and transplantation studies, it provides a system for testing the response of retinal pigment epithelium to pharmacological agents prior to use of an animal model. Moreover, since the current system enables fetal human RPE cells to regain their pigment granules even after repeated passages, it will be intriguing to apply this approach to maintenance of adult human RPE cells to determine if similarly encouraging results can be achieved.

ACKNOWLEDGMENTS This work was supported in part by NIH grants EY(03980) and That Man May See, Inc., and Research to Prevent Blindness. The authors wish to thank Paul Goulart for his invaluable help in the preparation of this manuscript. LITERATURE CITED Albert, D.M., Tso, M.O.M., and Rabson, A.S. (1972) In vitro growth of pure cultures of retinal pigmented epithelium. Arch. Ophthalmol., 88:63-69. Aronson, J.F. (1983) Human retinal pigment cell culture. In Vitro, 19:642-650. Bensaid, M., Malecaze, F., Bayard, F., and Tauber, J.P. (1989a) Opposing effects of basic fibroblast growth factor and transforming growth factor-B on the proliferation of cultured bovine retinal capillary endothelial (BREC) cells. Exp. Eye Res., 48:791-799. Boulton, M., and Marshall, J. (1985) Repigmentation of human retinal pigment epithelial cells in vitro. Exp. Eye Res., 41:209-218. Del Monte, M.A., and Maumenee, I.H. (1980) New techniques for in uitro culture of human retinal pigment epithelium. Birth Defects, 16:327-338. Flood, M.T., Gouras, P., and Kjeldbye, H. (1980) Growth characteristics and ultrastructure of human retinal pigment epithelium in vitro. Invest. Ophthalmol. Vis. Sci., 19:1309-1320. Gospodarowicz, D. (1975) Purification of a fibroblast growth factor from bovine pituitary. J. Biol. Chem., 250:2515-2520.

203

Gospodarowicz, D., Moran, J.S., and Bialecki, H. (1976) Mitogenic factors from the brain and pituitary: Physiological significance. In: Growth Hormone and Related Peptides. Excerpta Med. Int. Congr. Ser., 381, pp. 141-155. Gospodarowicz, D., and Ill, C.R. (1980aJ Do plasma and serum have different abilities to promote cell growth? Proc. Natl. Acad. Sci. U.S.A., 77:2726-2730. Gospodarowicz, D., and 111, C.R. (1980b) The extracellular matrix and control of proliferation of vascular endothelial cells. J . Clin. Invest., 65:1351-1354. Gospodarowicz, D., and Tauber, J.P. ( 1 9 8 0 ~Growth ) factors and the extracellular matrix. Endocr. Rev., 1:201-227. Gospodarowicz, D., Fujii, D.K., Giguere, L., Savion, N., Tauber, J.P., and Vlodavsky, I. (1981aJ The role of the basal lamina in cell attachment, proliferation, and differentiation. Tumor cells vs normal cells. In: The Prostatic Cell : Structure and Function. G.P. Murphy, A.A. Sandberg, and J.P. Karr, eds. Alan R. Liss, Inc., New York, pp. 95-132. Gospodarowicz, D., and Lui, G.M. (1981b) Effect of substrata in vitro on bovine aortic endothelial cells. J. Cell. Physiol., 109:69-81. Gospodarowicz, D. (1982) The control of proliferation of mammalian cells: Role of plasma factors and substrata. In: Functional Regulation of the Cellular and Molecular Levels. R.A. Corradino, ed. ElsevierINorth Holland, Inc., New York, pp. 13-40. Gospodarowicz, D., Cheng, J., Lui, G.M., Baird, A., and Bohlen, P. (1984) Isolation by heparin-sepharose affinity chromatography of brain fibroblast growth factor: Identity with pituitary fibroblast growth factor. Proc. Natl. Acad. Sci. U.S.A., 81:6963-6967. Isreal, P., Masterson, E., Goldman, A.I., Wiggert, B., and Chader, G.J. (1980) Retinal pigment epithelial cell differentiation in vitro. Invest. Ophthalmol. Vis. Sci., 19:720-727. Lansing, M., Jerdan, J., Dugan, J.,Maglione, A,, and Glaser, B. (1989) The bifunctional role of transforming growth factor Beta- l(TGFbetal) and Beta-2 in neovascularization in the mature animal. Invest. Ophthalmol. Vis. Sci. 30(Supp2.):392. Lui, G.M., Gordon, W.C., Song, M.K., and Bazan, N.G. (1990) Morphological analysis and phagocytic properties of cultured human fetal RPE cells. Curr. Eye Res. (Submitted for publication). Mannaugh, J., Arya, D.V., and Irvine, A.R. (1973) Tissue culture of human retinal pigment epithelium. Invest. Ophthalmol. Vis. Sci., 12:52-64. Nicolaissen, B., Nicolaissen, B.E., Berahi, K., Kolstad, A., Arnesen, K., and Armstrong, D. (1982) Monolayered explants in the study of retinal pigment epithelial behavior in culture. Acta Ophthalmol. (Copenh.), 60t873-880. Nicolaissen, B., Allen, C., Nicolaissen, A., and Arnesen, K. (1986) Human retinal pigment epithelium in long term explant culture. Acta Ophthal. (Copenh.), 64:l-8. Nicolaissen, B., Vishafer, R., Allen, C., Nicolaissen, A,, and Rubin, M.L. (1989) Behavior ofhuman RPE cultured on Bruch’s membrane and on necrotic debris. Invest. Ophthalmol. Vis. Sci., 30:813-822. Schroder, M.F., Muller, G., Birchmeier, W., and Bohlen, P. (1986) Transforming growth factor-B inhibits endothelial cell proliferation. Biochem. Biophys. Res. Commun., 137.295-302. Schweigerer, L., Malerstein, B., Neufeld, G., and Gospodarowicz, D. (1987) Basic fibroblast growth factor is synthesized in cultured retinal pigment epithelial cells. Biochem. Biophys. Res. Comm. 143: 934-940. Takehara, K., LeRoy, E.C., and Grotendorst, G.R. (1987) TGF-B inhibition of endothelial cell proliferation: Alteration of EGF binding and EGF-induced growth regulatory (competence)gene expression. Cell, 49:415-422. Vlodavsky, I., Folkman, J., Sullivan, R., Friedman, R., IshaiMichaeli, R.. Sasse, J., and Klagbrun, M. (1987) Endothelial cellderived basic fibroblast growth fictor: synthesis and deposition into sub-endothelial extracellular matrix. Proc. Natl. Acad. Sci. U.S.A., 84:2292-2296. Whittaker, J.R. (1967) Loss of melanotic phenotype in vitro of retinal pigment cells: Demonstration of Mechanisms involved. Dev. Biol., 15:553 -574. Zinn, K.M., and Benjamin-Henkind, J.V. (1979) Anatomy of the human retinal pigment epithelium. In: The Retinal Pigment Epithelium. K.M. Zinn and M. Marmor, eds. Harvard University Press, Cambridge, pp. 3-31.