Exp. Eye Res. (1998) 66, 783–790Article Number : ey980484

Temperature-Sensitive Interactions Between RPE and Rod Outer

Segment Surface Proteins

FANG YAN, NIGEL G. F. COOPER BARBARA J. MLAUGHLIN*

Department of Ophthalmology & Visual Sciences, Kentucky Lions Eye Center, and the Department of

Anatomical Sciences & Neurobiology, University of Louisville School of Medicine, Louisville, KY, U.S.A.

(Received Cleveland 19 August 1997 and accepted in revised form 31 January 1998)

Phagocytosis of rod outer segments by the retinal pigment epithelium is distinguished by the two distincttemperature-dependent steps of binding and ingestion. This study was designed to see if retinal pigmentepithelial (RPE) plasma membrane proteins interact with ROS plasma membrane proteins at temperaturesfavoring either binding or ingestion. A modified blot overlay assay was used whereby Western blots ofRPE plasma membrane proteins were overlaid with biotinylated ROS plasma membrane proteins.RPE}ROS interactions were detected by streptavidin-HRP and the ECL method at 25°C (ingestion), 15°C(binding), and 4°C (little or no binding or ingestion). Unlabeled ROS proteins served as the negativecontrol. Competition with excess unlabeled ROS proteins were used to test the specificity of the proteininteractions. Some protein interactions were somewhat temperature dependent. For example, two RPEplasma membrane proteins (200 kDa and 173 kDa) interacted with ROS plasma membrane proteins atboth 25°C and 15°C, but not at 4°C. A strongly labeled protein at 50 kDA protein was present at 25°Cbut weakly labeled at 15°C and at 4°C. Other protein interaction were more clearly temperaturedependent. For example, a 110 kDa RPE protein interacted with ROS proteins only at 25°C. Another RPEprotein (55 kDa) interacted only at 15°C. These latter data provide correlations between binding eventsin the assay and previously described stages of phagocytosis. # 1998 Academic Press

Key words : retinal pigment epithelium; rod outer segment; phagocytosis ; plasma membrane; receptor.

1. Introduction

Retinal pigment epithelial (RPE) cells participate in the

normal visual cycle by phagocytizing photoreceptor

outer segments (ROS) during their ongoing renewal.

Several studies have demonstrated that ROS phago-

cytosis is mediated by a specific ligand-receptor

interaction. Kinetic studies have revealed that binding

and ingestion of ROS by cultured rat RPE cells reaches

saturation with increasing ROS concentration (Hall

and Abram, 1987). In addition to saturability,

Mayerson and Hall (1986) have demonstrated that

cultured RPE cells selectively bind and ingest ROS over

other particles, such as red blood cells, bacteria, and

yeast. Furthermore, competition experiments have

shown that phagocytosis of iodinated ROS can be

inhibited 40% by a two-fold excess of unlabeled ROS

(Laird and Molday, 1988).

Several RPE receptors have been proposed for ROS

phagocytosis, with the first candidate being the

mannose receptor (Tarnowski, Shepherd and

McLaughlin, 1988; Boyle et al., 1991; Shepherd,

Tarnowski and McLaughlin, 1991; McLaughlin et al.,

1994). This receptor has been isolated from human

RPE cells as a 175 kDa cell surface protein which can

be labeled by an antiserum raised against a human

* Corresponding author: Barbara J. McLaughlin, Department ofOphthalmology and Visual Sciences, Kentucky Lions Eye Center,University of Louisville School of Medicine, Louisville, KY 40202-1594, U.S.A.

alveolar macrophage mannose receptor (Shepherd et

al., 1991). The same antiserum inhibits ROS phagocy-

tosis by cultured rat RPE cells (Boyle et al., 1991). A

second receptor candidate is CD36, an 88 kDa glyco-

protein expressed on rat RPE (Ryeom et al., 1994,

1995, 1996a). When non-phagocytic cells are trans-

fected with RPE CD36 cDNA, these cells bind and

ingest ROS (Ryeom et al., 1994, 1995). CD36 antibody

also inhibits ROS binding and ingestion by cultured rat

RPE cells (Ryeom et al., 1996a). A third receptor

candidate, integrin, has also been suggested to

function in ROS phagocytosis (Anderson, Johnson and

Hageman, 1995; Lin and Clegg, 1996; Miceli,

Newsome and Tate, 1997). The vitronectin receptor

(αvβ&

integrin) has been localized on both the apical

surface of RPE cells and the ROS plasma membrane

(Anderson et al., 1995). Functionally, a RGD peptide

has been shown to inhibit ROS binding and ingestion

by cultured human RPE cells (Lin and Clegg, 1996).

Recently, Hall’s group has shown that an antiserum

against RPE plasma membrane glycoproteins between

51 kDa and 62 kDa strongly inhibits ROS phagocytosis

and suggests that glycoproteins in this molecular

weight range may serve as receptors for phagocytosis

(Hall et al., 1996).

Phagocytosis can be divided into three phases, (1)

recognition and binding, (2) ingestion, and (3)

digestion of shed ROS in phagosomes or phago-

lysosomes (Philp and Bernstein, 1981; Chaitin and

Hall, 1983; Bok, 1985). Kinetic studies of ROS

phagocytosis have shown that binding and ingestion

0014–4835}98}060783­08 $25.00}0 # 1998 Academic Press

784 F. YAN ET AL.

events can be separated by varying the temperature

(Hall and Abram, 1987). When the temperature is

between 20°C to 37°C, the number of ingested ROS by

normal RPE cells is four times higher than the number

of bound ROS. However, at 10°C–17°C, little ingestion

takes place, and the number of bound ROS is higher

than the number of ingested ROS. Almost no binding

or ingestion occurs when the temperature is below

4°C. These data support the suggestion that binding

and ingestion are two distinct and separate com-

ponents of phagocytosis.

The present study is designed to determine whether

there are different RPE plasma membrane proteins

which interact with ROS proteins at temperatures that

favor binding versus ingestion. Of particular interest is

whether dystrophic rat RPE cells, in which ROS

binding is intact but ingestion is absent (Chaitin and

Hall, 1983), display different protein interactions

during ingestion as compared to normal RPE.

2. Materials and Methods

Biotinylation of ROS Plasma Membrane Proteins

Bovine neural retinas were peeled off the eye-cups

and incubated with 0±4 mg ml−" of Sulfo-NHS-Biotin

(Pierce, Rockford, IL, U.S.A.) in a 50 m sodium

bicarbonate buffer, pH 8±5, containing 20% sucrose

for 2 hr at 4°C (20 neural retinas in 10 ml of this

solution). Biotinylated ROS were subsequently isolated

according to the method previously described (Chaitin

and Hall, 1983). A phagocytic assay was carried out

to confirm that cultured RPE bound and ingested

biotinylated ROS at the same level as unlabeled ROS.

Total (bound and ingested) ROS and bound ROS were

counted by the double immunoflourescent method

(Chaitin and Hall, 1983). A two-way T Test was

performed between treatments to determine significant

(P!0±05) differences. Biotinylated ROS were

solubilized with an overlay buffer (20 m sodium

phosphate, pH 7±5, containing 1% NP-40, 0±1% SDS,

0±5%s sodium deoxycholate, 150 m NaCl, and

0±02% sodium azide). The total ROS protein con-

centration was tested by the bicinchoninic acid (BCA)

method and kept at 0±2 mg ml−".

Detection of Biotinylated ROS Proteins

Biotinylated ROS proteins were resolved using SDS-

polyacrylamide gel electrophoresis (Laemmli, 1970).

Samples were diluted 1:1 in the sample buffer

(125 m Tris buffer, pH 6±8, containing 4% SDS, 20%

glycerol, 10% 2-mercaptoethanol) at room tempera-

ture for 30 min. ROS proteins resolved on SDS-

PAGE were transferred to PVDF membrane (Millipore,

Bedford, MA, U.S.A.) (Towbin, Staehelin and Gordan,

1979). Non-specific binding sites were blocked by in-

cubating the membrane with 2% gelatin in Tris-

buffered Saline (TBS) containing 0±1% Tween-20 for

1 hr at room temperature. Biotinylated ROS plasma

membrane proteins were visualized by incubating

the blot with 1:10000 dilution of streptavidin-HRP

(Amersham, Arlington Heights, IL, U.S.A.), for 1 hr at

room temperature and developed with the ECL method

(Amersham, Arlington Heights, IL, U.S.A.). The

specificity of biotinylation was checked by using

unlabeled ROS as a negative control. Biotinylated ROS

proteins were also compared to ROS plasma membrane

proteins isolated by the ricin-agarose method (Yan et

al., 1995).

Testing the Purity of Isolated RPE Plasma Membranes

RPE cells were scraped from culture dishes by a cell

lifter. Cell homogenates were prepared by solubilizing

cells in 0±1% Triton X-100 in 10 m Tris, pH 7±4. RPE

plasma membranes were then isolated by differential

centrifugation (Ottonello and Maraini, 1984) and

solubilized in the same solution. A series of enzyme

activities in each fraction was tested to determine the

relative purity of the plasma membrane fraction as

compared to the cell homogenates. The marker

enzyme, 5«-nucleotidase, was measured for detecting

plasma membranes, acid phosphatase for lysosomes,

and NADPH cytochrome C-reductase for endoplasmic

reticulum. Aliquots of each fraction were removed for

protein determination by the BCA method (Pierce,

Rockford, IL, U.S.A.), and enzyme activity was

calculated by multiplying the specific enzyme activity

by the protein concentration.

Blot Overlay Assay

Primary cultures of normal and dystrophic RPE cells

were used to isolate whole cell proteins or RPE plasma

membrane proteins. Solubilized RPE proteins from

either cell homogenates or plasma membrane fractions

were resolved on 8% slab SDS-PAGE gels (Laemmli,

1970) and transferred to PVDF membranes (Towbin et

al., 1979). The blots were incubated with 2% gelatin

in TBS-T (TBS containing 0±1% of Tween-20) for 3 hr

at room temperature in order to block nonspecific

background binding. The membrane strips were then

overlaid with solubilized biotinylated ROS plasma

membrane proteins in the overlay buffer (0±2 mg ml−"

of total ROS proteins) overnight at three experimental

temperatures : 4°C, 15°C and 25°C. RPE proteins

interacting with biotinylated ROS proteins were

visualized by incubating the strips with 1:10000

dilution of streptavidin-HRP in TBS-T for 1 hr, and

developed by the ECL method. Unlabeled ROS proteins

(0±2 mg ml−") served as a negative control. Between

the two incubating steps, the strips were extensively

rinsed three times by using TBS-T.

To verify the interaction of ROS with RPE proteins,

anti-rhodopsin antibodies were used instead of

streptavidin-HRP to label the blot overlay assays. Such

blots were incubated with a 1:1000 dilution of mouse

anti-rhodopsin antibody (kindly provided by Dr P.

RPE-ROS PROTEIN INTERACTIONS 785

Hargrave) in TBS-T for 2 hr. Rhodopsin-containing

micelles were visualized by incubating the trips with a

1:10000 dilution of antimouse IgG-HRP in TBS-T

for 1 hr.

A competition experiment was performed to test the

specificity of the interaction between RPE proteins on

the PVDF membrane with solubilized ROS proteins.

Excess unlabeled ROS proteins (one, five and ten times

excess unlabeled ROS proteins over labeled ROS

proteins) were added to the overlay buffer to com-

petitively inhibit the protein interaction between

biotinylated ROS and RPE proteins.

3. Results

Analysis of Isolated RPE Plasma Membrane Fraction

Isolated RPE plasma membrane fractions were

evaluated using enzyme markers to check the purity of

the plasma membrane fractions (Table I). Approxi-

mately 68% of the total 5«-nucleotidase activity, 13%

of the total acid phosphatase activity and 5% of the

NADPH cytochrome C reductase activity was

recovered in the solubilized plasma membrane frac-

tion. The enzyme assays confirmed that the pre-

dominant membrane proteins present in this fraction

were from plasma membrane.

Comparison of Proteins in Biotinylated and Non-

biotinylated Plasma Membrane Fractions

Biotinylated ROS proteins were compared to ROS

plasma membrane proteins isolated by the ricin-

agarose method (Molday and Molday, 1987; Yan et

al., 1995). Silver stained total ROS proteins [Fig. 1(A)]

were compared to silver-stained ROS plasma mem-

brane proteins isolated by the ricin-agarose method

[Fig. 1(B)]. Biotinylated ROS proteins visualized with

streptavidin-HRP were shown in Fig. 1(C). Protein

molecular weights were calculated from each gel

using the molecular weight standard used with that

gel. Lines were drawn to connect proteins form

different gels which showed similar molecular weights.

There were no bands in the control lane when non-

T I

Marker enzymes in RPE plasma membrane fractions

and cell homogenates

HomogenatePlasma

membrane

Proteinu 2±3 1±35«-Nucleotidase* 69±5 83±4 (68%)†Acid phosphatase 24±1 5±9 (13%)Cytochrome c-reductase 0±032 0±003 (5%)

u Protein is expressed as mg ml−".* Enzyme activities are expressed as µmol mg−" min−".† Values in parenthesis are recovery of activity in percent of

activity in homogenate corrected by protein concentration in eachfraction.

F. 1. Comparison of biotinylated ROS plasma membraneproteins to those isolated by ricin-agarose beads. Lane A:Total ROS proteins (silver stained) ; Lane B: ROS plasmamembrane proteins isolated by ricin-agarose beads (silverstained) ; Lane C: Biotinylated ROS proteins (streptavidin-HRP and ECL) ; Lane D: Control, non-biotinylated ROSproteins (streptavidin-HRP, ECL). Lines connect proteinswith similar molecular weights. The pattern of biotinylatedproteins was similar to those plasma membrane proteinsisolated by the ricin-agarose bead method.

biotinylated ROS proteins and streptavidin-HRP were

used [Fig. 1(D)]. The pattern of biotinylated ROS

proteins was similar to the affinity-purified, plasma

membrane gel and indicated that the majority of

surface proteins were biotinylated.

Because biotinylated ROS were to be used to detect

interactions between ROS and potential RPE receptors,

it was necessary to test if biotinylation had any effect

on ROS phagocytosis. Phagocytic assays were per-

formed on rat RPE cells in cultures challenged with

biotinylated ROS (ROS*) or unlabeled ROS. The

number of bound and ingested ROS (labeled and

unlabeled) was counted and compared. RPE cells

ingested 545³30 cm−# unlabeled ROS as compared to

537³18 of labeled ROS (Fig. 2). Binding of labeled

and unlabeled ROS was similar (163 vs. 166). There

were no differences in ingested or bound ROS between

these two groups, which indicated that biotinylation of

ROS did not interfere with ROS phagocytosis. These

data also supported the conclusion that biotinylation

would not interfere with protein–protein interaction.

Temperature-sensitive Interactions between RPE and

ROS

Blot overlay assays were performed initially with

homogenates of RPE cells (Fig. 3). Two RPE proteins

(200 and 173 kDa) interacted with ROS plasma

786 F. YAN ET AL.

F. 2. The effect of biotinylation on ROS phagocytosis. (A)Normal RPE cells in cultures were challenged with eitherbiotinylated (ROS*) or non-biotinylated ROS, and total(bound and ingested) and bound ROS were detected byimmunofluorescence ; (B) Quantitation of the number ofbound and ingested ROS and ROS* showed no differences inthe number of ingested and bound ROS between these twogroups indicating that biotinylation of ROS did not interferewith ROS phagocytosis.

membrane proteins at both 25°C and 15°C, but not at

4°C (Bands 1 and 2 in Fig. 3). One RPE protein with

molecular weight of 110 kDa (Band 3 in Fig. 3)

interacted with ROS proteins at 25°C, but not at 15°Cor 4°C. The 55 kDa RPE protein (Band 5 in Fig. 3)

interacted with ROS plasma membrane proteins at

15°C, but not at 25°C or 4°C. Two RPE proteins at

50 kDa and 60 kDa (Bands 4 and 6 in Fig. 3) were

detected at all temperatures. Two RPE proteins

(130 kDa and 80 kDa) (bands a and b in Fig. 3) were

detected in all conditions of the blot over assays. These

F. 3. Blot of RPE proteins from cell homogenatesoverlaid with biotinylated ROS proteins at 25°C, at 15°C andat 4°C. Non-biotinylated ROS proteins were used in thecontrol lane. Bands 1–6 represented RPE proteins specificallyinteracting with ROS proteins. Bands a and b representednon-specific staining by streptavidin.

F. 4. RPE blot from plasma membrane fraction overlaidwith biotinylated ROS proteins at 25°C, at 15°C and at 4°C.Non-biotinylated ROS proteins were used in the control lane.All RPE proteins (Bands 1–6) interacting with biotinylatedROS in Fig. 3 were also present in the plasma membraneprotein fraction.

two proteins were also detected in the control lane

which was interacted with streptavidin but which did

not contain biotinylated ROS proteins.

In the next experiments, blot overlay assays were

used on purified RPE plasma membrane protein

fractions (Fig. 4). Six RPE proteins were also detected

in these assays. The most notable difference found in

the plasma membrane fraction versus cell homogenate

RPE-ROS PROTEIN INTERACTIONS 787

F. 5. Lane A: a blot overlay assay at 25°C which wasincubated with rhodopsin antibody followed by HRP-conjugated secondary antibody; Lane B: rhodopsin antibodywas preabsorbed with ROS before incubation with the blotoverlay assay; Lane C: only secondary antibody was usedwith the RPE blot overlay assay; Lane D: no ROS proteinswere overlaid on the RPE blots. Only specific bindinginteractions between ROS proteins and RPE (Bands 1, 2, 3,4 and 6) were labeled.

was that band 6 was intensely stained at 25°C,

compared to that at 15°C and 4°C (Fig. 4). It was

noteworthy that band 3 which was present only at

25°C, and band 5 which was present only at 15°C,

F. 6. Competition experiments performed at 25°C with increasing concentrations of non-biotinylated ROS proteins inhibitedinteractions between RPE proteins and biotinylated ROS* proteins.

were detected again in these assays. The same two

nonspecific interactions were also detected in these

assays (Bands a and b in Fig. 4).

ROS protein interactions with RPE protein were

checked by labeling the blot overlay assay with

rhodopsin antibodies. These experiments were based

on the assumption that some ROS membrane micelles

used in the blot overlay assays would contain some

rhodopsin as a protein constituent. As shown in Fig. 5,

bands 1, 2, 3, 4 and 6 were labeled at 25°C, and non-

specific bands (Bands a and b in Fig. 3) were not

labeled. Several control experiments were performed.

The five bands were absent when the rhodopsin

antibody was preabsorbed with ROS before immuno-

blotting the blot overlay assay [Fig. 5(B)]. The five

bands were not present when primary antibody was

omitted [Fig. 5(C)] or when ROS protein was omitted

from the overlay [Fig. 5(D)]. In addition, a competition

experiment was performed to test the specificity of the

interaction between RPE and ROS proteins. In this

experiment, excess non-biotinylated ROS proteins

were used to inhibit the interaction between RPE

proteins and biotinylated ROS proteins. As the con-

centration of non-biotinylated ROS proteins increased,

the labeling density of bands 1, 2, 3, 4 and 6 decreased

(Fig. 6).

Dystrophic RPE Blot Overlay Assays

RPE cells from RCS rats have been shown to bind

ROS, but fail in ingestion (Chaitin and Hall, 1983).

Because this animal model provides a natural way to

separate out the binding and ingestion phases of

phagocytosis, blot overlay assays with dystrophic RPE

proteins were performed to see if this assay could

788 F. YAN ET AL.

F. 7. Comparison of normal and dystrophic RPE blots from cell homogenates overlaid with biotinylated ROS proteins at25°C. Non-biotinylated ROS proteins were used in the control lanes. Dystrophic RPE blots showed the same protein interactionswith biotinylated ROS as compared to normal RPE blots.

detect any differences between normal and RCS rats.

At ingestion temperature, the same bands, 1, 2, 3, 4

and 6, were observed when dystrophic RPE blots were

overlaid with biotinylated ROS proteins (Fig. 7). The

same nonspecific bands (a and b) detected previously

were also present. These results indicated that there

were no qualitative differences in the protein inter-

actions between dystrophic RPE and ROS proteins.

4. Discussion

The blot overlay assay used in this study detects

protein–protein interactions and provides a useful

method to identify potential receptors and their ligands

(Crawford and Beckerle, 1994). Such methods can

result in underestimates of the number of protein–

protein interactions because some of the proteins

which are denatured for gel electrophoresis may not

become fully renatured in the assay (Carr and Scott,

1992). On the other hand, the method could result in

the detection of too high a number of protein–protein

interactions due to non-specific binding. In the current

study, non-specific interactions are minimized by the

use of blocking reagents, through the incorporation of

controls and through the use of a competition assay.

Some non-specific binding between streptavidin and

protein bands in blots was detected through the use of

these controls (Fig. 3).

The current study is a modification of the overlay

method and uses labeled ROS surface proteins as

ligands to overlay RPE proteins in an attempt to

identify prospective phagocytic receptors in the RPE. It

is necessary to point out that while we have used this

assay to differentiate between the binding and in-

gestion phases of phagocytosis, that both events are

evident in these assays as binding events. While it is

possible that the results of these binding assays do not

reflect biological processes occurring in vivo, they do

correlate with the previously described temperature-

dependent phases of binding and ingestion (Hall and

Abram, 1987). Further analysis will be required to

link particular proteins seen in this assay with

particular aspects of phagocytosis.

In the blot overlay assay, two RPE plasma

membrane-associated proteins are uniquely identified

by their temperature-specific interactions with ROS

proteins. A 55 kDa protein interacts with ROS proteins

only at 15°C, whereas a 110 kDa RPE protein inter-

acts with ROS proteins only at 25°C. Therefore, these

proteins are correlated with the binding and ingestion

phases of phagocytosis. In addition, a 50 kDa RPE

protein is also identified by its greater binding of

ROS proteins at 25°C than at 15°C. Therefore, this

protein also is correlated with the ingestion, rather

than the binding phase of phagocytosis. The other two

RPE proteins (173 kDa and 200 kDa) interact with

ROS proteins at 15°C and at 25°C and are potentially

correlated with both binding and ingestion of ROS (see

Table II).

Hall et al. (1996) isolated RPE plasma membrane

glycoproteins by a combination of Concanavalin A

affinity purification and an affinity column made with

antiserum against the RPE plasma membrane-

enriched fraction. Adsorption of the antiserum to

blotted RPE proteins and elution of adsorbed IgGs

produced a fraction from the 51–62 kDa range which

inhibited ROS phagocytosis. In the present study, a

50 kDa RPE protein interacted with ROS at both

RPE-ROS PROTEIN INTERACTIONS 789

T II

RPE plasma membrane proteins interaction with ROS

proteins

25°C(Ingestion)

15°C(Binding) 4°C

200 (1)* 200 (1)173 (2) 173 (2)110 (3)60 (4) 60 (4) 60 (4)

55 (5)50 (6)† 50 (6) 50 (6)

* Numbers in parenthesis represent protein bands in Figs 3–7.† Intensely labeled as compared to that at 15°C and at 4°C.

binding and ingestion temperatures, and a 55 kDa

RPE protein interacted with ROS only at binding

temperatures. Therefore, two different experimental

approaches have shown that RPE proteins in the

50–62 kDa molecular weight range are implicated in

phagocytic interactions and could be candidates for

receptors.

Another candidate receptor for phagocytosis is the

173 kDa RPE protein, which interacts with ROS at

both binding and ingestion temperatures in the blot

overlay assay and may be related to the mannose

receptor. The mannose receptor has been found to be

present at the RPE apical membrane and can be

labeled as a 175 kDa protein in immunoblots of RPE

apical membrane by an antiserum against human

alveolar macrophage mannose receptor (Boyle et al.,

1991; Shepherd et al., 1991). In addition, this receptor

has been shown to play a role in ROS phagocytosis by

the RPE (Tarnowski et al., 1988; Boyle et al., 1991).

Furthermore, rhodopsin and other mannose-contain-

ing glycoproteins isolated from ROS plasma mem-

branes have been demonstrated to bind to a mannose

receptor column and could be potential ligands for

mannose receptor-mediated ROS phagocytosis (Lutz,

Guo and McLaughlin, 1995; Yan et al., 1995). In this

regard, it is of interest that labeling the blot overlay

assay with a rhodopsin antibody demonstrates inter-

actions between rhodopsin-containing micelles and

RPE proteins. We speculate that one of these inter-

actions at the 173 kDa region is related to the rat RPE

mannose receptor.

Earlier studies by Hall and colleagues have also

implicated high molecular weight plasma membrane

glycoproteins on the RPE as candidates for phagocytic

receptors (Clark and Hall, 1986; Colley, Clark and

Hall, 1987). In the latter study, two glycoproteins at

160 kDa and 214 kDa were consistently removed by

protease digestion of RPE cell surfaces that resulted in

a loss in phagocytic ability. In the former study, a

183 kDa glycoprotein was singled out in phago-

cytically-deficient RPE cells as being abnormally

glycosylated and therefore a possible phagocytic

candidate in normal RPE. Similarly, the blot overlay

assay used in the present study showed that two RPE

plasma membrane proteins, 173 kDa and 200 kDa,

interacted with ROS proteins and strongly suggested

that these two high molecular weight RPE proteins

may be candidate receptors for ROS phagocytosis.

Cooper and colleagues have looked also for defects

in ingestion proteins in the rat dystrophic RPE and

found that 175 kDa and 86 kDa glycoproteins have

reduced affinities for certain lectins which are

expressed at the onset of the phagocytic defect (Tien,

McLaughlin and Cooper, 1991). This has led them to

suggest that changes in RPE glycoproteins are

associated with the defect in ingestion. The present

study has found that dystrophic RPE blot overlays

exhibit similar protein interactions with ROS proteins

at ingestion temperatures as compared to normal RPE.

Although dystrophic RPE proteins can interact with

ROS in this assay, it is possible that they might still

have some defect in their structure which allows

binding to ROS ligands but does not activate the

intracellular signaling pathways required for phago-

cytosis (Hall, Abram and Mittag, 1991; Heth and

Schmidt, 1991).

There are no RPE proteins interacting with ROS

proteins in the blot overlay assays that show similar

molecular weights to CD36, another candidate re-

ceptor for ROS phagocytosis. CD36 is an 88 kDa cell

surface glycoprotein. Antibodies to CD36 inhibit ROS

binding and ingestion in cultured rat RPE cells

(Ryeom, Sparrow and Silverstein, 1996a), and when

non-phagocytic cells are transfected with CD36 cDNA,

these cells bind and ingest ROS (Ryeom et al., 1994,

1995). Its absence in the present study can be

explained by the fact that CD36 ligands are thought

to be lipids, such as phosphatidylserine and phos-

phatidylinositol (Ryeom et al., 1996b). In the blot over-

lay assay used here, only protein–protein interactions

are detected.

The data presented show that there are several RPE

proteins (200 kDa, 173 kDa, and 50 kDa) which

interact with ROS at both binding and ingestion

temperatures. If either of these proteins is involved in

phagocytosis it would indicate that the binding and

ingestion phases are components of a sequential

process whereby the binding phase triggers the

ingestion phase using the same receptor protein. Such

a model is supported by the finding that when ROS-

binding is inhibited by an antiserum raised against

RPE plasma membrane proteins, ROS-ingestion is also

inhibited (Gregory and Hall, 1992). However, the data

presented here clearly demonstrate that a single RPE

protein band at 110 kDa interacts with ROS only at

ingestion temperatures and another protein band at

55 kDa interacts with ROS only at binding tempera-

tures. If these two proteins are involved in phagocytosis

then these data would support a model in which the

ingestion and binding phases are mediated by two

distinct proteins. In this model, one phase could be

blocked while the other phase remains intact and this

790 F. YAN ET AL.

model could provide an explanation for why ROS

binding is intact in the dystrophic animal model while

ingestion is inhibited (Chaitin and Hall, 1983).

Acknowledgements

This research was supported by NIH grant EY02853, andby grants from the Kentucky Lions Eye Foundation andResearch to Prevent Blindness Inc.

References

Anderson, D. H., Johnson, L. V. and Hageman, G. S. (1995).Vitronectin receptor expression and distribution atphotoreceptor-retinal pigment epithelial surface. J.Comp. Neurol. 360, 1–16.

Bok, D. (1985). Retinal photoreceptor-pigment epitheliuminteractions. Invest. Ophthalmol. Vis. Sci. 26, 1659–94.

Boyle, D., Tien, L., Cooper, N. G. F., Shepherd, V. L. andMcLaughlin, B. J. (1991). A mannose receptor isinvolved in retinal phagocytosis. Invest. Ophthalmol. Vis.Sci. 32, 1464–70.

Carr, D. W. and Scott, J. D. (1992). Blotting and band-shifting: techniques for studying protein–protein inter-actions. Trends in Biochem. Sci. 17, 246–9.

Chaitin, M. H. and Hall, M. O. (1983). Defective ingestion ofrod outer segments by cultured dystrophic rat pigmentepithelial cells. Invest. Ophthalmol. Vis. Sci. 24, 812–20.

Clark, V. M. and Hall, M. O. (1986). RPE Cell SurfaceProteins in Normal and Dystrophic Rats. Invest.Ophthalmol. Vis. Sci. 27, 136–44.

Colley, N. J., Clark, V. M. and Hall, M. P. (1987). Surfacemodification of retinal pigment epithelial cells : effectson phagocytosis and glycoprotein composition. Exp. EyeRes. 44, 377–92.

Crawford, A. W. and Beckerle, M. C. (1994). Blot overlayassay: a method to detect protein–protein interactions.In Cell biology, a laboratory handbook. Vol. 3, Pp. 301–8.(Celis, J. E. Ed.) Academic Press Inc: London U.K.

Gregory, C. Y. and Hall, M. O. (1992). The phagocytosis ofROS by RPE cells is inhibited by an antiserum to rat RPEcell plasma membranes. Exp. Eye Res. 54, 843–51.

Hall, M. O. and Abram, T. A. (1987). Kinetic studies of therod outer segment binding and ingestion by cultured ratRPE cells. Exp. Eye Res. 45, 907–22.

Hall, M. O., Abram, T. A. and Mittag, T. W. (1991). ROSingestion by RPE cells is turned off by increased proteinkinase C activity and by increased Calcium. Exp. EyeRes. 52, 591–8.

Hall, M. O., Burgess, B. L., Abrams, T. A., Ershov, A. V. andGregory, C. Y. (1996). Further studies on the identi-fication of the phagocytosis receptor of rat retinalpigment epithelial cells. Exp. Eye Res. 63, 255–64.

Heth, C. A. and Schmidt, S. Y. (1991). Phagocytic challengeinduces changes in phosphorylation of retinal pigmentepithelium proteins. Curr. Eye Res. 10, 1049–57.

Laemmli, U. K. (1970). Cleavage of structural proteinsduring assembly of the head of bacteriophage T4.Nature 227, 680–5.

Laird, D. W. and Molday, R. S. (1988). Evidence against therole of rhodopsin in rod outer segment binding to RPEcells. Invest. Ophthalmol. Vis. Sci. 29, 419–28.

Lin, H. and Clegg, D. O. (1996). Arg-gly-asp containingpeptide inhibits ROS phagocytosis by cultured humanRPE. Invest. Ophthalmol. Vis. Sci. (Suppl), 37, S378.

Lutz, D. A., Guo, Y. and McLaughlin, B. J. (1995). Natural,high-mannose glycoproteins inhibit ROS binding andingestion by RPE cells. Exp. Eye Res. 61, 487–93.

Mayerson, P. L. and Hall, M. O. (1986). Rat retinal pigmentepithelial cells show specificity of phagocytosis in vivo.J. Cell Biol. 103, 299–308.

McLaughlin, B. J., Cooper, N. G. F. and Shepherd, V. L.(1994). How good is the evidence to suggest thatphagocytosis of ROS by RPE is receptor mediated? Prog.Ret. Eye Res. 1, 147–64.

Miceli, M. V., Newsome, D. A. and Tate, D. J. (1997).Vitronectin is responsible for the serum stimulateduptake of rod outer segments by cultured retinalpigment epithelial cells. Invest. Ophthalmol. Vis. Sci. 38,1588–97.

Molday, R. S. and Molday, L. L. (1987). Differences in theprotein composition of bovine retinal rod outer segmentdisk and plasma membranes isolated by a ricin-gold-dextran density perturbation method. J. Cell Biol. 105,2589–601.

Ottonello, S. and Maraini, G. (1984). The isolation by selfforming gradients of Percoll of plasma membraneenriched fractions from bovine retinal pigment epi-thelium. Curr. Eye Res. 3, 1085–96.

Philp, N. J. and Bernstein, M. H. (1981). Phagocytosis byretinal pigment epithelium explants in culture. Exp. EyeRes. 33, 47–53.

Ryeom, S. W., Sparrow, J. R. and Silverstein, R. L. (1994).CD36 is expressed on retinal pigment epithelium (RPE)and mediates phagocytosis of photoreceptor outersegments (ROS). Invest. Ophthal. Vis. Sci. (Suppl) 35,S2140.

Ryeom, S. W., Silverstein, R. L. and Sparrow, J. R. (1995).The absence of CD36, a molecular mediating phago-cytosis of photoreceptor outer segments (ROS), corre-lates with a lack of ROS uptake by retinal pigmentepithelial cells (RPE). Invest. Ophthal. Vis. Sci. (Suppl)36, S815.

Ryeom, S. W., Sparrow, J. R. and Silverstein, R. L. (1996a).CD36 participates in the phagocytosis of rod outersegments by retinal pigment epithelium. J. Cell Sci. 109,387–95.

Ryeom, S. W., Sparrow, J. R. and Silverstein, R. L. (1996b).Photoreceptor outer segments (ROS) interactions withCD36: structure}function relationships. Invest. Ophthal.Vis. Sci. (Suppl) 37, S920.

Shepherd, V. L., Tarnowski, B. I. and McLaughlin, B. J.(1991). Isolation and characterization of a mannosereceptor from human pigment epithelium. Invest.Ophthal. Vis. Sci. 32, 1779–84.

Tarnowski, B. I., Shepherd, V. L. and McLaughlin, B. J.(1988). Expression of mannose receptors for pinocytosisand phagocytosis on rat retinal pigment epithelium.Invest. Ophthal. Vis. Sci. 29, 742–8.

Tien, L. F., McLaughlin, B. J. and Cooper, N. G. F. (1991).Glycoprotein in the retinal pigment epithelium ofnormal and dystrophic rats. Invest. Ophthal. Vis. Sci. 32,319–26.

Towbin, H., Staehelin, T. and Gordon, J. (1979). Electro-phoretic transfer of proteins from polyacrylamide gels tonitrocellulose sheets : Procedure and some applications.Proc. Natl. Acad. Sci. U.S.A. 76, 4350–4.

Yan, F., Lutz, D. A., Shepherd, V. L., Boyle, D. andMcLaughlin, B. J. (1995). Characterization of rod outersegment plasma membrane proteins which bind to themannose receptor. Curr. Eye Res. 14, 465–71.