Background
Glycosaminoglycans (GAGs) are negatively charged
polysaccharides derived from an amino hexose. They are
structural and functional modulators of extracellular
matrices that play important roles in CNS development and
repair. They exhibit both stimulatory and inhibitory
influences on neurite outgrowth and survival. Evidence
demonstrates the ability of heparin sulfates (HSs) to bind
to growth/trophic factors and selectively regulate such
factors' receptors. [ 1 2 ] . They can act as co-receptors
of growth/trophic and survival factors to regulate cell
behavior and/or restrict diffusion and create a relatively
high local concentration of ligand.
Pigment epithelium-derived factor (PEDF) is an
extracellular neuronal differentiation and survival factor
for cells derived from the retina and CNS. It induces
neuronal differentiation in retinoblastoma cells, protects
retina neurons (including photoreceptors) from death by
apoptosis and other insults, and has a morphogenetic effect
on photoreceptor cells [ 3 4 5 6 ] . It also has
neurotrophic effects on neurons from the cerebellum,
hippocampus and spinal cord [ 7 8 9 10 11 12 ] . In the
intact retina, this factor is identified as a secreted
protein associated by ionic interactions with the
interphotoreceptor matrix [ 13 14 ] , where GAGs are the
major polyanionic components.
Biochemically, PEDF is a 50-kDa glycoprotein with
structural homology to members of the serine-protease
inhibitor (serpin) superfamily [ 3 15 ] . However, it has
no inhibitory effects on proteases. Its neurotrophic
activities are independent of its protease inhibition
potential but dependent on its interaction with
cell-surface receptors [ 9 12 16 17 18 ] . PEDF has high
binding affinity for cell-surface receptors in human
retinoblastoma Y-79 cells (K
d
= 2.7 nM), which is mediated by interactions between a
region spanning amino acid positions 78-121 of the PEDF
polypeptide and the extracellular domains of the receptor
protein [ 18 ] . Blockage of these interactions inhibits
the PEDF neurotrophic effects. PEDF also has binding
affinity for GAGs, such as, heparin, heparin- and
chondroitin-sulfates, but this affinity is ~1000-fold lower
than for the receptor (e.g., K
d
4 μM for the heparin-PEDF interactions) [ 14 19 ] .
The binding to GAGs is mediated by ionic interactions
between an area clustered with positively charged lysines
of PEDF and the negatively charged GAGs. In the PEDF
spatial structure, the putative GAG binding domain is
distinct from and non-overlapping with the neurotrophic
active region [ 14 20 ] .
Because PEDF coexists with GAGs in extracellular
matrices and has binding affinity for them, it is of
interest to investigate the role of GAGs on PEDF activity.
Given that binding to cell-surface receptors is the first
step in the biological activity of PEDF, we used human
retinoblastoma Y-79 cells and their conditioned media (CM)
as sources of functional PEDF receptors and extracellular
matrix components, respectively, to examine the GAG content
in CM and their effects on PEDF ligand-receptor
interactions. The data suggest that heparan sulfate
participates in the formation of a PEDF binding complex
with its cell-surface receptor, and constitutes a positive
modulator for the PEDF-receptor interactions.
Results
Complex formation between PEDF and component(s) in
media conditioned by retinoblastoma cells
To determine whether PEDF interacts with component(s)
in media conditioned by retinoblastoma cells (CM), we
used an ultrafiltration assay. In this assay, soluble
PEDF of 50-kDa is filtered through a membrane with an
exclusion limit of MW 100,000, however it is retained
upon formation of a complex larger than this limit [ 14 ]
. The binding reactions were with a given 125I-PEDF
concentration and CM. Incubations were at 4°C to minimize
enzymatic degradation of proteins and glycosaminglycans
during the reaction. We found that 22% of 125I-PEDF was
retained by the membrane in the presence of concentrated
CM, compared to only 4% retention in the presence of
defined medium (non-conditioned medium) concentrated in
an identical fashion. Specific retention, termed PEDF
bound, was calculated by subtracting the retention in
defined media from that in CM. Figure 1shows that PEDF
was specifically retained when mixed with soluble
conditioned media and the value for PEDF bound increased
proportionally to the concentration factor of the media
(Fig. 1A). Interestingly, protease treatment of the CM
did not abolish the binding (Fig. 1B). These observations
revealed that the retained forms in the CM were PEDF
complexes ≥ 100-kDa, and that the majority of these
complexes were formed with soluble CM components other
than proteins.
Heparin and HS in the conditioned media with
affinity for PEDF
GAGs and polyanions in the CM were fractionated by
anion-exchange column chromatography followed by
PEDF-affinity column chromatography (Fig. 2). The GAG
content was followed by staining with Toluidine Blue-O
(Fig. 2C,2D,2E). The final fraction (CM PEDF) contained
components with binding affinity for PEDF that stained
with Toluidine Blue-O and migrated as high molecular
weight GAGs.
To determine the type of sulfated GAG in the media, we
designed a spectrophotometric assay using heparinase and
heparitinase, specific degrading enzymes for heparin and
HS, respectively (Fig. 3). The activities of both GAG
lyases reached a plateau by one hour of incubation (Figs.
3A,3C) and the degradation of GAG substrates between 0-30
μg was linear. Both CM and CM PEDFcontained substrates
for heparinase and heparitinase (Figs. 3B,3D). The amount
of GAGs was determined by comparison of the amount of Δ
4-hexuronate produced with CM samples to the standard
curves with commercial GAGs. The estimated content of
GAGs in CM varied between 12.4-22.7 μg/ml for heparin and
between 9-10 μg/ml for HS-like molecules, among media
conditioned by three different batches of Y-79 cells. The
estimated GAG content in CM PEDFranged 0.2-1.8 μg/μl and
0.1-0.4 μg/μl for heparin and HS, respectively. Similar
assays were followed with chondroitinase ABC, but its
substrates, ΔDi4S, dermatan and ΔDi6S, in CM were below
detection limits. These results demonstrated that CM and
CM PEDFcontained heparin- and HS-like molecules,
demonstrating that Y-79 cells produced GAGs with binding
affinity for PEDF.
Media conditioned by retinoblastoma cells enhances
the 125I-PEDF binding to cell-surface receptors
We have demonstrated previously that biologically
active 125I-PEDF binds specifically, competitively and
with high affinity to cell-surface receptors of Y-79
cells [ 18 ] . Because the reaction conditions were
identical to those used for biological assays, the
binding reactions were performed in the presence of media
conditioned by the cells for 16 hours (CM). We
investigated the effect of components of the CM, a source
of extracellular matrix, on the PEDF-receptor
interactions using CM and non-conditioned defined media
in radioligand binding assays. Comparison of reactions in
the absence and presence of CM showed that the specific
PEDF-binding to Y-79 cell-surface receptors was 6.8-fold
higher with conditioned medium than with defined medium
(Fig. 4). Note that with the binding method used, the
amount of 125I-PEDF retained in CM without cells is the
same to the amount of non-specific 125I-PEDF binding
(reactions with cells and in the presence of 50-fold
molar excess of unlabeled PEDF) [ 18 ] , indicating
retention by PEI-treated glass-fiber filters of
PEDF-receptor or GAG-PEDF-receptor complexes rather than
PEDF-GAG. Similar results were obtained when the
cell-bound 125I-PEDF was separated by centrifugation,
rather than filtration through glass-fiber filters, and
comparing reactions with CM versus those with 1% BSA in
PBS (data not shown). These observations showed that a
component(s) secreted by retinoblastoma cells enhanced
the PEDF-receptor interactions.
Effect of GAG lyases and chlorate on 125I-PEDF
binding to cell-surface receptors
To deplete the Y-79 cell cultures of HS and
heparin-like GAGs, we used heparitinase and heparinase,
respectively. The cultures were pretreated with each GAG
lyase before using them in radioligand binding assays.
The morphology and viability of the cells were not
affected with the GAG lyase treatments. Figure 5Ashows
that specific 125I-PEDF binding decreased significantly
in heparitinase treated cultures compared to untreated
controls, and less drastically in heparinase treated
ones. Hyaluronidase treatment to deplete the cultures of
hyaluronan did not have an effect on the binding. These
results demonstrated that removal of heparin/HS from the
cell cultures decreased the PEDF binding to receptors on
Y-79 cells. Chlorate is a competitive inhibitor of
ATP-sulfurylase, and inhibition of GAG sulfation in cell
cultures can be achieved by pretreatment of the cultures
with 30 mM sodium chlorate [ 21 ] . The effect of
undersulfated GAGs on the PEDF binding to its receptor
was examined. Cells pretreated with sodium chlorate did
not show changes in viability or morphology; however, the
treatment resulted in a decrease in the specific
125I-PEDF binding to about 35% relative to untreated
controls (Fig. 5B). Sodium sulfate was used to recover
the loss of GAG sulfation by chlorate increasing the
binding to 55% maximal with 10 mM sulfate additions. The
data revealed that inhibition of sulfation of GAGs
reduced the PEDF binding to cell-surface receptors of
Y-79 cells, with about 20% of specific inhibition. Thus,
these observations implied that HS/heparin might play a
functional role in the binding of PEDF to its
cell-surface receptor.
Discussion
It has been proposed that the GAG-binding property of
PEDF provides the molecular basis for its association with
extracellular matrices and may serve to localize PEDF
activity in the retina and CNS [ 14 19 ] . However, the
present results point to direct effects these
polysaccharides might have on the biochemical interactions
between PEDF and PEDF receptors on the surfaces of cells
that respond to this neurotrophic factor. We have shown
that the binding of PEDF to receptors in retinoblastoma
cells is enhanced by the presence of extracellular
heparin/HS-like GAGs, which can be found in the culture
medium of retinoblastoma cells. The fact that the binding
of PEDF to cell surfaces decreases with heparin/HS
depletion, implies that heparin/HS molecules might act as
cofactors for PEDF-receptor interactions. Interactions
between PEDF and extracellular GAGs can also explain the
complex formed by PEDF with CM even after protease
treatment of the latter. The PEDF-heparin/HS complex may
somehow facilitate encounters between PEDF and its
receptor,
e.g. , by inducing a conformational
change in PEDF, which might accelerate the ligand-receptor
interactions. In addition, the receptor may also form a
complex with heparin/HS to facilitate interactions with the
ligand.
To our knowledge, this is the first report on the
production of GAGs by retinoblastoma cells. We found that
these cells produce HS/heparin secreted into the culturing
media. The retina and malignant solid tumors also produce
the sulfated GAGs [ 1 2 22 23 24 25 26 ] . Although HSs are
mostly found as proteoglycans associated with the basal
lamina or the plasma membrane, the presence of HSs in the
culturing medium might be a result of shedding or release
of their extracellular domains from the cell membranes as
soluble components. Cell-associated HS proteoglycans can
undergo regulated shedding from the membrane into the
soluble extracellular matrix or culturing medium converting
the membrane anchored molecules into soluble effectors [ 1
2 ] . In the conditioned media of all the tested batches of
retinoblastoma cells, we detected heparin and HS with
binding affinity for PEDF. The estimated concentrations of
these GAGs in the media may vary with cell density and
conditioning time. However, under the conditions used, they
were within the linear range of HS-PEDF complex formation (
EC
50 = 40 μg/ml) [ 14 ] .
The fact that depletion of heparin/HS-like GAGs from the
culturing media results in inhibition of PEDF binding to
cell surface receptors points to functional roles for these
GAGs such as those of positive modulators of PEDF-receptor
interactions. In this regard, we observed that depletion
from the Y-79 cell cultures of heparin with heparinase was
lower than those depleted of HS with heparitinase,
suggesting that the retinoblastoma-derived HS was more
effective than the retinoblastoma-derived heparin. This
observation can be explained by structural, compositional
and functional differences between heparin and HS GAGs.
GAG-binding proteins can be differentially sensitive to
variations in GAG structure [ 27 ] . GAGs produced among
different cell types have structural and compositional
differences and structural changes in GAGs are known to
occur in cells undergoing morphological differentiation
and/or malignant transformations [ 28 29 30 31 ] . Thus, in
the native retina, or other tissue, modulation of the
PEDF-receptor interactions may depend on the expression of
GAGs, which occur during development and pathological
conditions.
Our data offer interesting possibilities of regulation
of the activity of PEDF. The ratio and amount of production
of heparin and HS by cells bearing PEDF receptors may be an
important mechanism to control the activity of PEDF as is
the modulation of the rate of expression of PEDF receptors.
The data obtained so far on the action of PEDF on different
cells may have not considered the presence of these
cofactors in the media. The binding of GAGs to PEDF that
modulate the binding of these factors to its receptor opens
also the possibility that different GAGs may modulate
differently the affinity of PEDF for its receptors, by
increasing or even decreasing it in some cases. Finally,
the fact that the GAG and receptor binding regions are on
opposite regions of the PEDF molecule, suggest the
possibility of the existence of PEDF mutants or engineered
variants that, having lost or decreased GAG-binding
capabilities, still show high affinity for the PEDF
receptor in a way not dependent upon GAGs content or
composition. These possibilities are discussed under the
consideration that some of them may have important
implications if PEDF, or molecules derived from it, are to
be used in the future as therapeutic agents.
The primary consequence of reducing the heparin/HS and
its sulfation was to minimize a binding site required for
PEDF activity. Although GAGs store PEDF in the
extracellular matrix [ 14 ] , a more direct mechanism
appears necessary, namely, its participation in the binding
of PEDF to its receptor. In a spatial structure of PEDF,
the heparin/HS binding domain of PEDF maps to the opposite
side of the neurotrophic active region [ 20 ] , allowing
distinct and non-overlapping interactions with heparin/HS
on one side of the protein, and with the neurotrophic
receptor on its opposite side. Our data suggest that the
intrinsic affinity of the cell surface receptor for PEDF
appears low, whereas the heparin/HS-PEDF complex is
recognized with high affinity. In addition, a direct
interaction between the receptor and GAGs may also be
necessary. Although details of the mechanism remain to be
revealed, it is clear that heparin/HS is required for the
first step of the neurotrophic activity of PEDF, namely the
encounters with its receptor at the cell surface. The
differentiation and survival of cells in vivo may be
regulated not only by the expression of PEDF and its
receptor but also by the temporal and spatial expression of
GAGs.
Methods
Materials
Heparin purified from bovine intestinal mucosa,
chondroitin sulfates A, B and C, chondroitinase ABC,
sodium chlorate, and Toluidine Blue-O were purchased from
Sigma. Subtilisin was from Boehringer Mannheim.
Heparitinase (E.C.4.2.2.8) and heparinase (E.C.4.2.2.7)
purified from
Flavobacterium heparinum were from
ICN Biomedicals, Inc. and alternatively from Seikagaku.
Heparan sulfate (HS) purified from bovine kidney was from
Seikagaku, hyaluronidase (E.C.4.2.2.1) purified from
Streptomyces hyalurolyticus from
ICN Pharmaceuticals, Coomassie Brilliant Blue from
BioRad, and Q-Sepharose from Pharmacia. Recombinant PEDF
was purified from BHK cells containing an expression
vector with human PEDF cDNA, as previously described [ 19
] .
Preparation of conditioned media
Human retinoblastoma Y-79 cells (0.45-5 × 10
6cells/ml) were cultured in defined media (MEM containing
10 mM HEPES, 1 mM Na-pyruvate, 0.1 mM non-essential amino
acids, 1 mM L-glutamine, 1% penicillin/streptomycin
(LifeTechnologies)) at 37°C for 16-24 h. Media exposed to
these conditions is referred as CM. CM was separated from
cells by centrifugation (1000 ×
g for 5 min at 4°C) and
concentrated by ultrafiltration using membrane filters
with MWCO = 10,000 (Amicon YM10 filters). GAGs/polyanions
purification was performed as follows: concentrated CM
(15 ml) was dialyzed against buffer Q (50 mM Tris-HCl, pH
8.0, 0.2 M NaCl, 6 M Urea, 0.5% CHAPS), filtered through
a 0.4 micron membrane and its soluble components
subjected to anion-exchange column chromatography using
Q-Sepharose Fast Flow (1 ml bed volume). The column was
washed with 15-column volumes of buffer Q, the bound
material eluted with 1.2 M NaCl and termed CM a.
Alternatively, a DEAE-Sephacel column was used.
PEDF-affinity column chromatography
To identify components with PEDF-binding affinity,
purified recombinant protein was used to prepare
PEDF-affinity resin with 3 M Emphaze™ Biosupport Medium
(Pierce Chemical) [ 18 ] . CM awas dialyzed against
buffer P (20 mM sodium phosphate pH 7, 150 mM NaCl, 0.5%
CHAPS) and filtrated through 0.4 μm filters. The soluble
dialysate was mixed with PEDF-resin (6 mg PEDF/ml resin)
at a 2:1 volume-to-volume ratio and incubated at 4°C with
gentle rocking for 16 h. The mixture was packed into a 10
ml Polyprep chromatography column (Bio-Rad) and washed
with 10-column volumes of buffer P. The bound material
was eluted with 10-column volumes of 3 M NaCl, desalted
with 20 mM Tris-HCl pH 8.0, 10% glycerol, treated for
protein depletion, dialyzed against deionized water,
lyophilized and resuspended in deionized water. The final
sample was termed CM PEDF. About 100 μl of CM PEDFwere
obtained from 100 ml of CM. Alternatively, concentrated
CM was used as starting material and protein depletion
was omitted.
Radioligand binding assays
PEDF binding to cell-surface receptors was assayed
using biologically active radioligand 125I-PEDF and Y-79
cells [ 18 ] by a widely-used method with a mechanism of
retention of receptors on polyethylenimine-treated
glass-fiber filters based mainly on ionic interactions [
32 ] . Polyethylenimine binds strongly to glass, which is
negatively charged and integral membrane proteins tend to
be acidic. The resultant polycationic
polyethylenimine-coated glass can retain cell membranes
due to their negative charges. Because binding of
cell-surface receptors to polyethylenimine filters is
rather insensitive to ionic strength, the ionic
phenomenon is thought to be supplemented by hydrophobic
forces and hydrogen binding [ 32 ] . The method used with
Y-79 cells and radiolabeled PEDF has been described
before in detail [ 18 ] . Briefly, cells cultured
overnight in serum-deprived medium at 37°C were
transferred to ice/water bath for 10 minutes before the
addition of ligand. The reaction mixtures containing cell
suspensions with given radioligand concentrations in
untreated or treated media were incubated at 4°C for 90
min, unless indicated. The free and bound 125I-PEDF were
separated by filtration through glass-fiber filters and
the bound radioactivity was determined in the filters
using a β-scintillation counter (Beckman, model LS 3801).
Nonspecific binding was calculated from reactions with a
molar-excess of unlabeled ligand (≥ 50-fold) over
radioligand.
Complex-formation assays
Complex formation between PEDF and CM components was
assayed by a method using ultrafiltration through
membranes of 100,000 MW exclusion limit [ 14 ] . Binding
reactions were performed with a given concentration of
125I-PEDF in defined or conditioned media, and
incubations with gentle rotation at 4°C for 2 h. Free and
bound ligand were separated by ultrafiltration through
Microcon-100 (Amicon). The reaction mixtures were diluted
40-fold with cold 20 mM sodium phosphate pH 6.5, 20 mM
NaCl, 10% glycerol and immediately ultrafiltrated,
repeating twice to ensure removal of free ligand from the
complexes. Each Microcon retenate cup was transferred to
scintillation vials, mixed with 5 ml BioSafe II liquid
scintillation solution (Research Products International)
by extensive vortexing, and its radioactivity determined
using a β-scintillation counter. Nonspecific binding,
calculated from reactions with an excess of unlabeled
ligand (100-fold) over radioligand, reached about 40% of
the total binding.
Enzymatic digestion treatments
The presence of GAGs was assayed using specific GAG
lyases,
i.e. , the presence of heparin, HS,
and chondroitin sulfates with heparinase, heparitinase,
and chondroitin ABC respectively. The amount of GAGs was
determined by the amount of Δ 4-hexuronate produced after
the eliminative cleavage of each substrate by the
corresponding GAG lyase. Samples were depleted of
proteins by protease treatment to avoid interference in
absorbance readings of the product. For heparinase and
heparitinase reactions, samples were treated with 5
milliunits of each enzyme in 150 μl of 0.1 M sodium
acetate and 1 mM CaCl
2 , pH 7 and incubations at 37°C for
various time periods. The reactions were stopped by the
addition of 1 ml of 0.06 M HCl. The soluble material was
separated by centrifugation (3000 ×
g , 10 min) and assayed for
absorbance at 235 nm to measure the concentration of
product Δ 4-hexuronate (Molar extinction coefficient =
5500; [ 33 ] ). For the chondroitinase ABC reactions,
each chondroitin sulfates A, B and C substrate (1 mg
each) and concentrated CM, were incubated with 0.12 units
of chondroitinase ABC in 1 ml of 50 mM Tris-HCl pH 8.0,
60 mM sodium acetate, and 0.02% BSA at 37°C. At various
time periods, aliquots of 0.1 ml were removed and mixed
with 0.9 ml of 45 mM KCl pH 1.8 to stop the reaction.
Insoluble material was removed by centrifugation (1000 ×
g , 10 min) and the supernatant
assayed for absorbance of Δ 4-hexuronate at 232 nm. For
protein depletion, CM was mixed with subtilisin at 0.4
μg/ml in 20 mM Tris-HCl pH 8.0, 10% glycerol and
incubated at 37°C for 16 h. Subtilisin was
heat-inactivated at 75°C for 25 min. The protein
concentration after the reaction was less than 0.1% of
the starting material.
To deplete cell cultures of GAGs, Y-79 cells in
defined serum-free medium (as above) at a density of 1.25
× 10 5cells/ml were cultured in 96-well culture plates
(150 μl/well) and incubated at 37°C in a 5% CO
2 environment for 16 h. Hyaluronidase
(>1TRU/μl), heparinase (1 mu/μl) or heparitinase (1
mu/μl) were each added to various wells and incubated at
37°C in a 5% CO
2 environment for 1 h.
Chlorate treatment of cell cultures
To prevent sulfation of GAGs in cell cultures, we used
a method previously described [ 33 ] . Y-79 cells (1.25 ×
10 5cells/ml) were cultured in 48-well plates (300
μl/well) in defined serum-free medium with or without 30
mM sodium chlorate and 10 mM sodium sulfate at 37°C in a
5% CO
2 environment for 24 h.
GAG detection assays
GAGs and proteins resolved by SDS-polyacrylamide gel
electrophoresis in Tricine/SDS buffer, as instructed by
manufacturer (Novex), were detected with specific stains.
For GAG detection, the gels were incubated sequentially
in 0.2% Toluidine Blue-O in ethanol-water-acetic acid
(50:49:1) for 30 minutes, in ethanol-water-acetic acid
(50:49:1) for 1 h, and in H
2 O for 16 h. [ 34 ] . For protein
detection, the gels were sequentially incubated in 0.5%
Coomassie Brilliant Blue in 50% methanol/10% acetic acid
and in 10% methanol/10% isopropanol. GAG detection was
also performed by the Toluidine Blue-O precipitation and
dot-blot method as described previously [ 34 ] . Briefly,
200 μl of sample were applied to each well of a 48-well
manifold (Life Technologies) onto a PVDF membrane without
vacuum. Then, 5 μl of 0.2% Toluidine blue-O were added to
each well and vacuum was applied. The membrane was
removed, washed twice with destaining solution
(ethanol-water-acetic acid, 50:49:1) for 5 min and
air-dried.
Other methods
The protein concentration was determined using BioRad
Protein Assay (BioRad).
Abbreviations
PEDF, pigment epithelium-derived factor; GAG,
glycosaminoglycan; HS, heparan sulfate; CM, media
conditioned by retinoblastoma cells.
Authors' contributions
EMA participated in the design, assay development and
carried out the complex-formation, GAG detection and
receptor binding assays. JEW carried out the purification
and characterization of GAGs, the enzymatic and chemical
treatments of cell cultures for receptor binding studies,
and drafted the manuscript. SPB conceived the study, and
participated in its design and coordination. All authors
read and approved the final manuscript.
