RESEARCH ARTICLE
199
Glycosaminoglycan synthesis and secretion by the
retinal pigment epithelium: polarized delivery of
hyaluronan from the apical surface
Preenie deS Senanayake1, Anthony Calabro2, Kazutoshi Nishiyama1, Jane G. Hu3, Dean Bok3
and Joe G. Hollyfield1,*
1Cole Eye Institute and 2Biomedical Engineering, The Cleveland Clinic Foundation, Cleveland, OH 44195, USA
3Jules Stein Eye Institute, Brain Research Institute and Department of Neurobiology, University of California Los
Angeles, Los Angeles, CA 90095,
USA
*Author for correspondence (e-mail: hollyfj@ccf.org)
Accepted 30 October; published on WWW 11 December 2000
Journal of Cell Science 114, 199-205 © The Company of Biologists Ltd
SUMMARY
Hyaluronan and chondroitin sulfate glycosaminoglycan
secretion from retinal pigment epithelial cells was
established in confluent cultures with high transepithelial
resistance. Cell cultures were maintained on Millicell-PCF
culture plates, which allow separation of culture medium
exposed to apical and basal epithelial surfaces. Following
various times in culture, apical and basal culture
media were sampled at three day intervals and the
glycosaminoglycan content was quantified. Samples were
digested with proteinase K to free the glycosaminoglycans
from their core proteins, the glycosaminoglycans were
ethanol precipitated, and subjected to hyaluronidase SD
and chondroitinase ABC digestion to release hyaluronan
and chondroitin sulfate disaccharides. Disaccharides
were fluorotagged with 2-aminoacridone, separated on
polyacrylamide gels and the molar fluorescence in each
disaccharide band quantitated. Hyaluronan in the apical
medium was significantly higher than in the basal medium
(5-12 times) at all recovery intervals (P<0.0001). In
contrast, the distribution of unsulfated chondroitin,
4-sulfated chondroitin and 6-sulfated chondroitin
disaccharides in apical and basal media was non-polar.
Confocal microscopy of cultures probed with a hyaluronanspecific fluorotag established that the HA evident in these
cultures is restricted to the apical border of the RPE
cultures. Collectively, these data indicate that hyaluronan
synthesized by the retinal pigment epithelium is secreted
preferentially from the apical surface, suggesting that this
tissue is an important source of hyaluronan present in the
interphotoreceptor matrix.
INTRODUCTION
of the RPE prevents the passive movement of extracellular
molecules between basal (choroidal connective tissue stroma)
and apical (IPM) compartments (Cunha-Vaz et al., 1966;
Shiose, 1970), which in turn allows the transport machinery of
the RPE to control the volume and composition of fluid in the
IPM (Hughes et al., 1998).
A number of studies document the synthesis by cultured
RPE cells of extracellular matrix molecules; including laminin,
fibronectin, integrin, heparin sulfate proteoglycan, type IV
collagen, tissue inhibitor of matrix metalloproteinase-3 (TIMP3), pigment epithelial derived factor (PEDF) and hyaluronan
(HA) (Edwards, 1982; Philip and Nachmias, 1987; Stramm,
1987; Newsome et al., 1988; Campochiaro et al., 1989; Ruiz
et al., 1996). Recent analysis of matrix molecules present on
the apical side of the RPE (within the IPM) identify several
molecules, including HA, matrix metalloproteinases (MMPs),
interphotoreceptor matrix retinoid binding protein (IRBP),
PEDF, and the novel IPM molecules SPACR and SPACRCAN
(also called IPM150 and IPM200) (Acharya et al., 1998;
Acharya et al., 2000). The finding that some of these molecules
bind HA, led us to propose that HA may function as the
primary scaffold of the insoluble IPM (Hollyfield, 1999).
The retinal pigment epithelium (RPE) is a neuroepithelium
derived from the outer wall of the optic cup during early
organogenesis. In the adult eye, the RPE is positioned between
the photoreceptor layer of the neurosensory retina and the
choriocapillaris of the choroid. The apical and basolateral
surfaces of the RPE are in contact with distinctly different
extracellular matrices. The apical surface interacts with the
interphotoreceptor matrix (IPM) and the distal tips of the
photoreceptor outer segments, whereas the basal surface abuts
Bruch’s membrane, a pentalaminate matrix separating the RPE
from the capillary endothelial cells of the choroid.
Proximal to the apical surface, the lateral borders of the
RPE cells are linked to each other via complex junctional
attachment zones (Hudspeth and Yee, 1973). Tight junctions
(zonula occludens) that are a part of this complex function to
prevent the passive diffusion of molecules between the apical
and basal compartments separated by the RPE. Since the
capillary endothelial cells below Bruch’s membrane are
fenestrated, macromolecules of the serum bathe the matrix
below the basolateral borders of the RPE. The barrier function
Key words: Retinal pigment epithelium, Hyaluronan, Chondroitin
sulfate, Polarity
200
JOURNAL OF CELL SCIENCE 114 (1)
Although a number of previous studies indicate that RPE cells
in culture synthesize HA (Edwards, 1982; Hewitt and
Newsome, 1987; Stramm, 1987), the polarity of HA secretion
has not been addressed. Because of the potential role of HA in
the organization of the IPM, it is critical to define the origin of
HA in this compartment and to understand the dynamics of its
synthesis and turnover. In the present study we establish the
distribution of HA and chondroitin GAGs in apical and basal
medium recovered from confluent RPE cultures that have
established high resistance junctions.
MATERIALS AND METHODS
Reagents
2-aminoacridone HCl (AMAC) was from Molecular Probes (Eugene,
Oregon USA). D-Glucose, D-mannose, mercuric acetate, glacial
acetic acid (99.99+%), dimethylsulfoxide (DMSO, 99.9%), sodium
cyanoborohydride (95%), and glycerol (99.5%) were from AldrichSigma (PO Box 355, Milwaukee, WI 53201, USA). Proteinase K and
phenol red (0.5% w/v) were from Gibco (Grand Island, NY 14072,
USA). MONOTM composition gels (#60100) and MONOTM gel
running buffer (#70100) were from Glyko Inc (81 Digital Drive
Novato, California 949949, USA). Dowex AG50W-X8 (200-400
mesh) Bio-Rad Laboratories (2000 Alfred Nobel Drive, Hercules,
CA 94547, USA). Hyaluronidase SD, chondroitinase ABC, and
unsaturated hyaluronan, the chondroitin sulfate disaccharide standards
and the biotin-conjugated HA binding protein (bHABP) were from
Seikagaku, America (Ijamsville, MD 21754, USA). High purity
hyaluronan (Healon®, the trade name for highly purified HA,
Pharmacia, Maersk Medical Ltd, UK).
Preparation of ∆disaccharide standards for fluorescent
derivatization
The AMAC derivatized HA and chondroitin ∆disaccharide standards
(with and without mercuric acetate treatment) were prepared as
previously described (Calabro et al., 2000b). For quantitation, AMAC
derivatized ∆Di2S standards were calibrated with AMAC derivatized
glucose standards as previously described (Calabro et al., 2000a).
RPE cell cultures
The use of all human tissues followed the tenets of the Declaration of
Helsinki, and the donors or their guardians gave consent for donation
of the tissues. Institutional human experimentation committee
approval was obtained for the use of human tissues. The human RPE
culture conditions have been described previously (Frambach et al.,
1990; Hu et al., 1994) and are summarized as follows: human RPE
was collected and grown in Eagle’s minimum essential medium
without calcium (MEM (Joklik); Sigma Chemical, St Louis, MO)
with additives described previously until it proliferated, reached
confluence and released cells into the medium. These non-attached
cells were collected and grown on Millicell chambers with
polycarbonate membranes (Millipore, Bedford, MA) coated with
mouse laminin (Collaborative Research, Bedford, MA). The Millicell
chambers were maintained in multiwell plates, which allowed the
separate exposure of apical and basal medium. The normal calcium
medium used to maintain established cultures consisted of Eagle’s
minimum essential medium (MEM, Irvin Scientific, Santa Ana, CA)
with supplements for optimal human RPE cell culture as described
previously (Hu and Bok, 2000). The transepithelial electrical
resistance (TER) of each confluent monolayer was measured with an
epithelial voltohmmeter (World Precision Instruments, New Haven,
CT) as previously reported (Hu et al., 1994). Culture medium from
apical and basal compartments was changed every three days and the
respective apical and basal media were retained. The recovered
medium was stored at −80°C until used for carbohydrate analysis as
described below. Apical and basal culture medium was collected from
two series of cultures maintained for over 140 days. In addition,
individual samples including the RPE cells with adherent matrix and
the culture medium from apical and basal compartments were
recovered on days 14, 17, 56, 59, 84, and 87 days of culture for
carbohydrate analysis.
Microscopy
Cell density measurements were performed on representative cultures
as previously described (Bok et al., 1992). Briefly, cultures on
polycarbonate supports were fixed with 4% formaldehyde in
phosphate buffer, dehydrated, mounted with glass coverslips and
viewed with a ×20 objective and ×10 ocular lenses. One of the oculars
was fitted with a calibrated micrometer square, and the cells within
the area counted. Counts from 10 fields in four separate cultures were
determined and the mean values were expressed as the cell density
extrapolated to the total area occupied by the RPE on the Millicell
plate.
The cultured RPE cells, along with the polycarbonate membranes,
were cut out from their chambers and fixed for 24 hours in 1%
formaldehyde and 1% glutaraldehyde in 0.1 M phosphate buffer, pH
7.4, at 4°C and then in 1% OsO4 in the same buffer for 30 minutes.
The monolayers and their supports were sequentially dehydrated in
70-100% ethanol, treated with propylene oxide and embedded in
Araldite 502 (Ted Pella Inc., Redding, CA). Ultrathin sections were
cut with a diamond knife, stained with uranium and lead salts and
imaged in a Zeiss 910 electron microscope.
For confocal microscopy, RPE cultures with high resistance
junctions were fixed in chilled methanol for ten minutes following a
brief rinse of the RPE/filter complex in 1× PBS. RPE cells were
blocked in 5% (w/v) bovine serum albumin. Filters were then
incubated with biotin-conjugated HA binding protein overnight at
4°C. This incubation was followed by sequential incubation with
FITC-conjugated streptavidin (Jackson ImmunoResearch, West
Grove, PA) for 1 hour at room temperature. The immunostained filters
were mounted on glass slides using an antifading medium (Vector,
Burlington, CA) containing DAPI, a dye specific for nuclei. The
cultures were observed using a TCS-SP/Leica confocal microscope
(Heidelberg, Germany). The optimal dilutions for the bHABP and
secondary antibody were 1:100 and 1:200, respectively. To control for
probe specificity and eliminate possible autofluorescence in RPE
cells, a neighboring portion of the filter was stained identically except
that the bHABP was omitted.
Medium and cell matrix preparation
The methods used for sample preparation and carbohydrate analysis
have been described in detail (Calabro et al., 2000a; Calabro et al.,
2000b). The basis of this analysis is a novel fluorophore-assisted
carbohydrate electrophoresis (FACE) procedure for the qualitative and
quantitative analysis of glycosaminoglycans. The method uses
enzymes that cleave glycosaminoglycans to create products, primarily
mono-and disaccharides, characteristic of the enzyme specificity.
Each cleavage exposes a new reducing terminus on the carbohydrate
that is fluorotagged by reductive animation with 2-aminoacridone.
The tagged products are then displayed by electrophoresis, identified
by their characteristic migration, and quantitated by their molar
fluorescence.
Culture medium and cell matrices were digested with proteinase K
(50 µl, 2.5 mg/ml) at 60°C to free GAGs from their core protein. A
second 50 µl aliquot of a fresh solution of proteinase K was added,
and digested for an additional 2 hours at 60°C. The proteinase K
digests were then boiled for 10 minutes to inactivate the proteinase
K. The digests were concentrated in a vacuum concentrator to 300 µl
and 1 ml of absolute ethanol was added and the sample stored at
−20°C. The ethanol extracts were centrifuged at 10,000 g for 15
minutes at 4°C to pellet any macromolecular material including the
HA and chondroitin sulfate chains, and the supernatant solution
Hyaluronan synthesis by the RPE
containing the low molecular mass molecules such as glucose and
salts were aspirated to waste. The pellet was washed with 1 ml of
–20°C absolute ethanol and centrifuged at 10,000 g for 15 minutes
at 4°C. The supernatant solution was aspirated to waste, and the
precipitate dried for 5 minutes on a vacuum concentrator. The
lyophilized precipitate was re-suspended in 100 µl, 0.0005% phenol
red, 100 nM ammonium acetate, pH 7 (Calabro et al., 2000a; Calabro
et al., 2000b).
GAG digestions
The ammonium acetate extracts were digested for 1 hour at 37°C with
100 mU/ml of hyaluronidase SD followed by 3 hours at 37°C with
the addition of 100 mU/ml of chondroitinase ABC. The enzyme
digests were dried on a vacuum concentrator prior to derivatization
with AMAC as described below. Representative samples were
mercury treated to distinguish internal disaccharides with their
unsaturated hexuronic acid from non-reducing terminal disaccharides
with saturated hexuronic acid residues or hexosamines (Calabro et al.,
2000a).
Fluorotagging procedures
Hyaluronidase SD and chondroitinase ABC digestion products were
derivatized by addition of 20 µl of 12.5 mM AMAC (500 nmoles) in
85% DMSO/15% acetic acid followed by incubation for 15 minutes
at room temperature. Then 20 µl of 1.25 M sodium cyanoborohydride
(50,000 nmoles) in ultrapure water was added followed by incubation
for 16 hours at 37°C. After derivatization, 10 µl of glycerol (20% final
concentration) was added to each sample prior to electrophoresis. All
derivatized samples were stored in the dark at −70°C (Calabro et al.,
2000a).
Electrophoresis, imaging and quantitation
The derivatized unknown samples to be run in separate lanes were
spiked with one AMAC-derivatized ∆Di2S standard containing 62.5,
Fig. 1. (A) A diagrammatic
representation of the RPE cell
Millicell culture unit used in this
study is presented. Apical and
basal medium are separated by
the RPE cultures on the
permeable membrane.
(B) Representative Millicell
chambers with (left) and without
(right) RPE cells are shown. The
black color of the left chamber is
due to the presence of melanin in
the RPE cells. (C) Cultured RPE
cells at confluence. (D) An
electron micrograph of an RPE
cell on a porous Millicell
transmembrane filter showing the
free apical surface exposed to the
apical medium and the basal
surface on the support membrane.
The arrows indicate pores in the
membrane which communicate to
the basal chamber.
201
31.3, 15.6, 7.8, 3.9, or 1.9 pmoles per 5 µl as an internal standard.
Samples were run on MONOTM composition gels with MONOTM gel
buffer. The samples were electrophoresed at 4°C for 80 minutes at a
constant 500 V with a starting current of 25 mA /gel, and a final
current of 10 mA /gel. The gels in their glass supports were
illuminated with UV light (365 nm) from an Ultra Lum
Transilluminator, and imaged with a Quantix cooled CCD camera
(Roper Scientific/Photometrics). The images were analyzed using the
Gel-Pro AnalyzerTM program version 3.0 (Media Cybernetics). The
digital images shown in the results section depict oversaturated pixel
intensity for the major derivatized structures in order to allow
visualization of less abundant derivatized carbohydrates. Quantitation
was performed on gel images having all pixels within a linear 12-bit
intensity depth range. The pixel density of the ∆Di2S standard
concentrations listed above were 2.86, 1.31, 0.8, 0.45, 0.36, and 0.31,
respectively. The efficiency of recovery of HA, 0S, 4S and 6S
glycosaminoglycans using this protocol was 68%, 74%, 72% and
70%, respectively.
RESULTS
The RPE cultures
A diagram of the chamber used for culturing RPE is presented
in Fig. 1A. As is evident in this illustration, RPE cultures
established on the porous support of the inner chamber allows
isolation of the two medium compartments, with one exposed
only to the apical surface of the RPE (the inner compartment)
and the other exposed to the basaolateral surface via the open
channels in the supporting membrane. The key factor in
separation of the apical and basal culture media requires cellcell interaction with the establishment of high resistance
junctions. Chambers containing no cells or RPE cells prior to
202
JOURNAL OF CELL SCIENCE 114 (1)
reaching confluence have no resistance to current passage
between electrodes placed in the inner and outer chambers. The
resistance of RPE cultures used in these studies ranged
between 800-1500 ohms/cm2, indicating the physical
separation by the RPE of medium on apical and basal borders.
The RPE cells at confluence become heavily melanized (Fig.
1B) and when viewed with low magnification microscopy,
showed a cobblestone appearance with most cells forming
a hexagonal border (Fig. 1C). Cell counts made from
representative cultures reflect an average density of 395,000 ±
28,800 RPE cells/mm2 (mean ± s.d.). When examined with
transmission electron microscopy, the RPE exists as a single,
low cuboidal layer, with apical microvillae and prominent
melanin granules in the apical cytoplasm (Fig. 1D). The
cultures illustrated were prepared on identical Millicell
supports used for the culture medium assays. Below the basal
plasma membrane, a basal lamina usually separates the RPE
from the supporting membrane. Pores in the supporting
membrane, which communicate with the basal compartment,
are evident in longitudinal profile (Fig. 1D).
Carbohydrate analysis
Electrophoretic patterns of AMAC-labeled carbohydrates
from apical and basal media preparations are presented in
Fig. 2A. The patterns of carbohydrates that accumulated in
the medium over the 3 day period between culture medium
changes can be compared directly to the authentic
standards present in the lane labeled MW. Note the
prominent hyaluronan disaccharide band (∆DiHA) in the
apical medium lanes (A1, A2 and A3) relative to the
density of the ∆DiHA band in the basal medium (B1,
B2 and B3). In contrast, the unsulfated chondroitin
disaccharide bands (∆Di0S) show similar density in apical
and basal media samples. The single prominent broad band
midway between the ∆Di4S and ∆Di6S standards,
represents these two chondroitin disaccharides, which do
not separate well in these preparations. The composition of
the unresolved broad band was determined to consist of
∆Di4S and ∆Di6S by mercuric acetate treatment of
representative apical and basal RPE culture media as
previously described (Calabro et al., 2000a). The most
prominent bands in these preparations represent glucose
Fig. 2. Fluorograms showing the separation of saccharides
recovered from the culture media (A) and RPE cultures/matrices
(B). The images shown depict oversaturated pixel intensity for
the major derivatized components in order to allow visualization
of less abundant derivatized components. The lane labeled MW
shows the separation of a standard mixture of 12 AMAC
derivatized saccharides that gives a wide separation of wellresolved bands. (A) Saccharides present in apical (A1-A3) and
basal (B1-B3) medium from three separate cultures (1-3).
Control lane shows a derivatized medium sample without cell
cultures for background corrections. ∆Di2S standards at
progressive dilutions from 62.5-1.9 pmoles (as determined by
hexuronic acid analysis) were added to the second (labeled A1)
through the seventh lane (labeled B3) to generate an internal
concentration curve in each gel (R=0.991, P=0.0001) for
quantification of resolved disaccharides. (B) Saccharides present
in RPE cultures-matrices (1-3) and a control polycarbonate filter
with medium only maintained for the same peroid of time as the
RPE cultures.
remaining in the culture medium. Usually, the basal medium
contained more remaining glucose than the apical medium.
The carbohydrate composition of cells from representative
RPE cultures was also analyzed. As is evident in Fig. 2B, no
∆DiHA or ∆Di0S is evident in the cellular material. In
contrast, some ∆Di4S and ∆Di6S is evident above control
levels. Three unknown bands of variable intensity (x1, x2 and
x3) are also present. Band x1 was novel to the cellular
carbohydrates and was not seen in the medium samples
(compare with Fig. 2A).
The first quantitative assessment of the medium samples
was performed on a single culture maintained for 99 days.
The medium from apical and basal compartments recovered
at three day intervals from day 54 through day 99 of this
culture period was processed and the ∆DiHA, ∆Di0S and
combined ∆Di4S and ∆Di6S disaccharides quantified (Fig.
3). The amount of ∆DiHA in the apical medium is
consistently 5 to 10 times higher than the amount present in
the basal medium. In contrast, the amount of ∆Di0S is similar
in apical and basal medium throughout the period of analysis,
while ∆Di4S-∆Di6S levels are slightly more abundant in the
apical medium.
Hyaluronan synthesis by the RPE
Fig. 3. Disaccharide levels in apical and basal medium from a set of
pooled human RPE cell cultures at confluence with high resistance
junctions maintained for 100 days in culture. Beginning on day 51
the medium was collected every three days. The analytical period
presented extended over a 45 day period.
To provide quantitative data on additional cultures that
would allow more definitive statistical comparison, the
carbohydrate profiles from seventeen separate RPE cultures
were analyzed (Fig. 4A). Data from the two cultures
designated A and B were from medium samples collected at
three day intervals over 100 days of culture. Those in cultures
designated C-K were individual medium samples from single
collection times. Although the amounts of carbohydrates vary
between the profiles shown, ∆Di4S-∆Di6S are most abundant,
followed by ∆Di0S, and then ∆DiHA. The amount of ∆DiHA
in apical medium is 4 to 6 times higher than that present in the
basal medium. Although the magnitude of the difference varied
from culture to culture, in each of these comparisons, the
differences in HA concentration was highly significant
(P<0.0001). The levels of ∆Di4S-∆Di6S and ∆Di0S in the
apical medium as compared to the basal concentrations were
not statistically significant.
The differences in relative amounts of the ∆-disaccharides
in apical and basal medium are more evident when the data is
expressed as an apical/basal concentration ratio (Fig. 4B). The
data used in this figure are the same as was plotted by
concentration in Fig. 4A. Equal concentration in apical and
basal medium would reflect a ratio of 1, which is near the level
of ∆Di4S-∆Di6S and ∆Di0S ratios. In contrast, the ∆DiHA
ratios range from 5 to 12, indicating 5 to 12 times the amount
of HA disaccharides in the medium exposed to the apical
surface of the RPE cultures as compared to that present in the
basal compartment.
203
Fig. 4. Quantitation of disaccharide levels in apical and basal media
from human RPE cell cultures at confluence with high resistance
junctions maintained for 100 days. The medium was changed every
three days. In A, cultures A and B represent pooled apical and basal
medium from series A (sampling period over 96 days; n=32) and
series B (sampling period over 45 days; n=14), respectively. Cultures
C-K represent individual cultures (sampling period over 89 days;
n=30). The data are presented as mean ± sem (P<0.05 was
considered to be significant). In B the saccharide levels from A are
plotted as a ratio of apical to basal concentrations. The height of the
shaded background above the abscissa represents a ratio of 1, when
the concentration ratio in the two compartments are equal.
Hyaluronan distribution
The use of the biotinylated HA-binding fragment derived from
aggrecan, in conjunction with confocal microscopy, allows the
direct visualization of HA associated with the RPE cells (Fig.
5). When viewed en face, green fluorescence, from the
streptavidin-FITC that binds to bHABP, was readily evident on
the cultures, but the staining was not uniformly distributed (Fig.
5A). In some regions the whole of the epithelial surface appeared
labeled in either a uniform or a punctate pattern (Fig. 5B). In
other areas, FITC fluorescence was associated with bright
strands or cables of HA. Some areas of the cultures were free of
FITC fluorescence, with only the DAPI-stained nuclei evident.
In viewing the en face images, it was evident that much of
the FITC-fluorescence was above the DAPI-labeled RPE
nuclei. When Z-axis images were generated from the cultures,
all FITC-fluorescence was found at the level of the apical
surface of the RPE (Fig. 5C). No areas were observed with
FITC-fluorescence below the DAPI-labeled nuclei.
204
JOURNAL OF CELL SCIENCE 114 (1)
Fig. 5. Confocal images of HA localization in RPE cultures. In A,
the en face image shows green fluorescence (HA) in variable
densities over the continuous epithelium (nuclei are blue). Strandlike cables of HA are evident in the upper left, while the center and
lower left of the image shows near continuous coverage of HA. The
boxed area in the lower right is enlarged in B. The punctate
distribution of HA in this enlargement may represent HA associated
with hyaluronan synthase in the cell membrane. A reconstruction of
the Z-axis distribution of HA from a 10 µm wide line through one of
the areas of uniform HA density is presented in C. Note that all HA
fluorescence (green) is located on the apical side of the nuclei (blue).
DISCUSSION
The data reported in this study indicate that HA secreted by
the RPE in confluent cultures with high resistance junctions
is highly polarized, with 82-95% of the HA delivered into
the apical medium. To our knowledge this is the first
demonstration in any epithelium of polarized HA secretion.
HA is unique among GAGs in that it is synthesized at the
cell surface as a free GAG (Prehm, 1984; Weigel et al., 1997).
In contrast, all other GAGs are covalently linked to a core
protein, and their biosynthetic pathway involves the Golgi
apparatus, followed by vesicular transport to the cell surface
for delivery into the extracellular compartment (Hascall and
Hascall, 1981). Hyaluronan synthase (HAS), the enzyme
responsible for polymerization of HA, is located in the plasma
membrane. As UDP-precursor sugars (glucuronic acid and Nacetylglucosamine) are brought to the cytoplasmic side of the
enzyme, the assembled HA polymer is extruded to the exterior
of the cell. HAS has been cloned in human and mouse. In both
species three distinct genes are present (HAS1, HAS2 and
HAS3), which are located at three different positions in the
genome (chromosomes 19q13.3, 8q24.12, and 16q22.1,
respectively, for human; and chromosomes 17, 15 and 8,
respectively, for mouse) (Spicer et al., 1997). The finding that
the bulk of the HA synthesized by the RPE is present in the
apical medium suggests that most of the hyaluronan synthase
in these cells resides in the apical plasma membrane.
A single molecule of HA can have a molecular mass
approaching 10 million Daltons, which would contain 25,000
repeat disaccharides, supporting an extended length of around
25 µm (Hascall and Hascall, 1981). One might argue that
because of its large size, it may be difficult for HA secreted
from the basolateral RPE surface to move through the pores in
the Millicell plate and have access to the basal medium.
Evidence against this possibility comes from the finding that
chondroitin sulfate GAGs show near equal distribution in
apical and basal culture medium. It is also unlikely that HA
could be trapped below the RPE or be restricted from entering
the basal medium if secreted from the basolateral borders of
the RPE cells. Since the GAG content of the RPE cells and
adherent matrix was extracted directly from the filter, any HA
present in or on the filter should have been detected, but this
analysis indicates that no detectable HA is associated with the
filter (see Fig. 2B). The final evidence excluding the possibility
that HA is restricted from moving through the filter comes from
the observation that HA was always present in the basal
medium, albeit in small amounts (Fig. 3). We conclude that the
distribution of HA observed in apical and basal medium
reflects the amounts of HA secreted from apical and basolateral
borders of the RPE during the period between culture medium
changes.
bHABP was useful in independently defining the
distribution of HA in the RPE cultures. After application of
bHABP, all streptavidin-FITC fluorescence was associated
with the apical surface of the RPE (Fig. 5). Unexpected was
the variation in distribution of HA in different regions of the
cultures. The absence of HA fluorescence in some areas could
be due to retraction of the HA in response to the methanol
fixation or in regional loss of HA by the brief PBS rinse prior
to fixation. The punctate areas of HA fluorescence present in
many areas on the apical surface of the RPE were particularly
intriguing (Fig. 5A-B). It is possible that these may represent
short segments of HA that remain associated with single or
small clusters of HAS at the apical surface of the epithelium.
Additional studies using HAS antibodies and bHABP are
currently underway to evaluate this possibility.
In the eye, HA delivered from the RPE apical surface would
become a part of the IPM. Early biochemical studies
demonstrate that small amounts of HA can be rinsed from the
IPM (Berman, 1964; Berman and Bach, 1968; Berman, 1969;
Adler and Klucznik, 1982), suggesting that HA may be only a
minor component of this matrix. More recently we have
experimentally demonstrated that the glycoprotein SPACR and
the proteoglycan SPACRCAN bind HA (Acharya et al., 1999;
Acharya et al., 2000), which lead us to propose that HA may
function in the IPM as the basic scaffold to which these and
other IPM molecules are anchored. It is likely that this binding
to HA is mediated through RHAMM-type HA-binding motifs
present in the primary sequence of the SPACR and
SPACRCAN polypeptide (Hollyfield, 1999). Thus the presence
of HA in the IPM and the associations made by HA with other
Hyaluronan synthesis by the RPE
molecules present in and around this matrix may be of
fundamental importance to the stability of macromolecules
comprising the insoluble IPM as well as the attachment of the
retina to the RPE.
Although our analysis clearly identifies the RPE as a
potential source for HA in the IPM, it is possible that the retina
may also contribute HA to this compartment. Recent studies
on isolated retinal cells indicate that HA is synthesized by the
radial glial cells of the retina (Müller cells), but not by retinal
neurons (Normand et al., 1998). Müller cells traverse the full
thickness of the retina with their apical border abutting the IPM
and their basal end feet terminating at the vitreal surface
(Hogan et al., 1971). Additional studies of the polarity of
retinal HA secretion will be required to define the destination
of HA synthesized by Müller cells.
In summary, FACE analysis of HA and chondroitin sulfate
GAGs delivered into apical and basal culture medium by the
RPE was established. HA was 5-12 times more abundant in the
medium exposed to the apical surface of the epithelium than
in the medium exposed to the basolateral border. In contrast,
the amount of chondroitin sulfate was similar in both
compartments. These data indicate that hyaluronan synthesized
by the RPE is secreted preferentially from the apical surface.
Supported by grants from The Foundation Fighting Blindness (D.B.
and J.G.H.), Hunt Valley, Maryland; Retina Research Foundation
(J.G.H.), Houston, Texas; and the National Institutes of Health (D.B.
and J.G.H.), Bethesda, Maryland. D.B. is also supported by the Dolly
Green Chair at UCLA.
REFERENCES
Acharya, S., Rodriguez, I. R., Moreira, E. F., Midura, R. J., Misono, K.,
Todres, E. and Hollyfield, J. G. (1998). SPACR, a novel interphotoreceptor
matrix glycoprotein in human retina that interacts with hyaluronan. J. Biol.
Chem. 273, 31599-31606.
Acharya, S., Foletta, V., Lee, J., Rayborn, M., Rodriguez, I., Young, S. III
and Hollyfield, J. (2000). SPACRCAN, a novel human interphotoreceptor
matrix hyaluronan-binding proteoglycan synthesized by photoreceptors and
pinealocytes. J. Biol. Chem. 275, 6945-6955.
Adler, A. J. and Klucznik, K. M. (1982). Proteins and glycoproteins of the
bovine interphotoreceptor matrix: composition and fractionation. Exp. Eye
Res. 34, 423-434.
Berman, E. (1964). The biosynthesis of mucopolysaccharides and
glycoproteins in pigment epithelial cells of bovine retina. Biochem. Biophys.
Acta 83, 371-373.
Berman, E. R. and Bach, G. (1968). The acid mucopolysaccharides of cattle
retina. Biochem. J. 108, 75-88.
Berman, E. R. (1969). Mucopolysaccharides (glycosaminoglycans) of the
retina: identification, distribution and possible biological role. Mod. Probl.
Ophthal. 8, 5-31.
Bok, D., O’Day, W. and Rodriguez-Boulan, E. (1992). Polarized budding of
vesicular stomatis and influenza virus from cultured human and bovine
retinal pigment epithelium. Exp. Eye Res. 55, 853-860.
Calabro, A., Benavides, M., Tammi, M., Hascall, V. and Midura, R.
(2000a). Microanalysis of enzyme digests of hyaluronan and
chondroitin/dermatan sulfate by fluorophore-assisted carbohydrate
electrophoresis (FACE). Glycobiology 10, 273-281.
205
Calabro, A., Hascall, V. and Midura, R. (2000b). Adaptation of FACE
methodology for microanalysis of total hyaluronan and chondroitin sulfate
composition from cartilage. Glycobiology 10, 283-293.
Campochiaro, P. A., Sugg, R., Grotendorst, G. and Hjelmeland, L. M.
(1989). Retinal pigment epithelial cells produce PDGF-like proteins and
secrete them into their media. Exp. Eye Res. 49, 217-227.
Cunha-Vaz, J., Shakib, M. and Ashton, N. (1966). Studies on the
permeability of the blood-retinal barrier. I. On the existence, development,
and site of a blood-retinal barrier. Br. J. Ophthalmol. 50, 441-453.
Edwards, R. B. (1982). Glycosaminoglycan synthesis by cultured human
retinal pigmented epithelium from normal postmortem donors and a
postmortem donor with retinitis pigmentosa. Invest. Ophthalmol. Vis. Sci.
23, 435-446.
Frambach, D. A., Fain, G. L., Farber, D. B. and Bok, D. (1990). Beta
adrenergic receptors on cultured human retinal pigment epithelium. Invest.
Ophthalmol Vis Sci. 31, 1767-1772.
Hascall, V. and Hascall, G. (1981). Proteoglycans. In Cell Biology
of the Extracellular Matrix (ed. E. Hay), pp. 39-63. New York: Plenum
Press.
Hewitt, A. T. and Newsome, D. A. (1987). Altered glycoconjugates in
cultures of retinitis pigmentosa retinal pigment epithelium. In Degenerative
Retinal Disorders: Clinical and Laboratory Investigations (ed. J. G.
Hollyfield, R. E. Anderson and M. M. LaVail), pp. 79-92. New York: Alan
R. Liss, Inc.
Hogan, M., Alvarado, J. and Weddell, J. (1971). Histology of the human
eye. Philadelphia: W. B. Saunders Co.
Hollyfield, J. (1999). Hyaluronan and the functional organization of the
interphotoreceptor matrix. Invest. Ophthalmol. Vis. Sci. 40, 2767-2769.
Hu, J., Gallemore, R., Bok, D., Lee, A. and Frambach, D. (1994).
Localization of NaKATPase on cultured human retinal pigment epithelium.
Invest. Ophthalmol. Vis. Sci. 35, 3582-3588.
Hu, J. G. and Bok, D. (2000). A cell culture medium that supports
differentiation of human retinal pigment epithelium into highly polarized
monolayers. Mol. Vision (submitted).
Hudspeth, A. and Yee, A. (1973). The intercellular junctional complexes of
the retinal pigment epithelia. Invest. Ophthalmol. Vis. Sci. 12, 354-365.
Hughes, B., Gallemore, R. and Miller, S. (1998). Transport mechanisms in
retinal pigment epithelium. In The Retinal Pigment Epithelium (ed. M.
Marmor and T. Wolfensberger), pp. 103-134. New York: Oxford University
Press.
Newsome, D., Pfeffer, B., Hewitt, A., Robey, P. and Hassell, J. (1988).
Detection of extracellular matrix molecules synthesized in vitro by monkey
and human retinal pigment epithelium: influence of donor age and multiple
passages. Exp. Eye Res. 46, 305-321.
Normand, G., Hicks, D. and Dreyfus, H. (1998). Neurotrophic growth
factors stimulate glycosaminoglycan synthesis in identified retinal cell
populations in vitro. Glycobiology 8, 1227-1235.
Philip, N. J. and Nachmias, V. T. (1987). Polarized distribution of integrin
and fibronectin in retinal pigment epithelium. Invest. Ophthalmol. Vis. Sci.
28, 1275-1280.
Prehm, P. (1984). Hyaluronate is synthesized at the plasma membrane.
Biochem. J. 220, 597-600.
Ruiz, A., Brett, P. and Bok, D. (1996). TIMP-3 is expressed in human retinal
pigment epithelium. Biochem. Biophys. Res. Commun. 226, 467-474.
Shiose, Y. (1970). Electron microscopic studies on blood-retinal and bloodaqueous barriers. Japan. J. Ophthalmol. 14, 78-87.
Spicer, A., Seldin, M., Olsen, A., Brown, N., Wells, D., Doggett, N., Itano,
N., Kimata, K., Inazawa, J. and McDonald, J. (1997). Chromosomal
localization of the human and mouse hyaluronan synthesis genes. Genomics
41, 493-497.
Stramm, L. E. (1987). Synthesis and secretion of glycosaminoglycans in
cultured retinal pigment epithelium. Invest. Ophthalmol. Vis. Sci. 28, 618627.
Weigel, P. H., Hascall, V. C. and Tammi, M. (1997). Hyaluronan synthases.
J. Biol Chem. 272, 13997-14000.