(Received for publication, June 6, 1996, and in revised form, November 6, 1996)
From the Holland Laboratory, Departments of
Biochemistry and
Molecular Biology, American Red
Cross, Rockville, Maryland 20855, and the Laboratories of
§ Biochemistry and ¶ Pathology, NCI, National
Institutes of Health, Bethesda, Maryland 20892
Thrombospondin-1 (TSP-1) is a large modular trimeric protein that has been proposed to play a diverse role in biological processes. Newly synthesized TSP-1 either is incorporated into the matrix or binds to the cell surface where it is rapidly internalized and degraded. TSP-1 catabolism is mediated by the low density lipoprotein receptor-related protein (LRP), a large endocytic receptor that is a member of the low density lipoprotein receptor family. Using adenovirus-mediated gene transfer experiments, we demonstrate that the very low density lipoprotein receptor can also bind and internalize TSP-1. An objective of the current investigation was to identify the portion of TSP-1 that binds to these endocytic receptors. The current studies found that the amino-terminal heparin binding domain (HBD, residues 1-214) of mouse TSP-1, when prepared as a fusion protein with glutathione S-transferase (GST), bound to purified LRP with an apparent KD ranging from 10 to 25 nM. Recombinant HBD (rHBD) purified following proteolytic cleavage of GST-HBD, also bound to purified LRP, but with an apparent KD of 830 nM. The difference in affinity was attributed to the fact that GST-HBD exists in solution as a dimer, whereas rHBD is a monomer. Like TSP-1, 125I-labeled GST-HBD or 125I-labeled rHBD were internalized and degraded by wild type fibroblasts that express LRP, but not by fibroblasts that are genetically deficient in LRP. The catabolism of both 125I-labeled GST-HBD and rHBD in wild type fibroblast was blocked by the 39-kDa receptor-associated protein, an inhibitor of LRP function. GST-HBD and rHBD both completely blocked catabolism of 125I-labeled TSP-1 in a dose-dependent manner, as did antibodies prepared against the HBD. Taken together, these data provide compelling evidence that the amino-terminal domain of TSP-1 binds to LRP and thus the recognition determinants on TSP-1 for both LRP and for cell surface proteoglycans reside within the same TSP-1 domain. Further, high affinity binding of TSP-1 to LRP likely results from the trimeric structure of TSP-1.
Thrombospondin-1 (TSP-1)1 is a large glycoprotein composed of three identical subunits, which are covalently linked by interchain disulfide bonds. This molecule is a part of a family of proteins that currently consists of five members (for reviews, see Refs. 1-3). Each subunit of TSP-1 is composed of an amino-terminal heparin binding domain, a type I procollagen homology region, three properdin repeats, three epidermal growth factor-like repeats, seven calcium binding repeats, and a carboxyl-terminal domain (4). TSP-1 has been proposed to play a diverse role in numerous biological processes; these include embryonic development (5), angiogenesis (6), and hemostasis (7, 8). The ability of TSP-1 to influence these biological processes has been attributed to its capacity to bind to matrix proteins, proteinases, growth factors, and cell surface receptors.
Given that TSP-1 appears to influence a large number of biological processes, it is not surprising that levels of TSP-1 are carefully regulated. Protein levels of TSP-1 are controlled at the level of RNA synthesis and stability (9) and by the rapid catabolism of the protein by an endocytic receptor that leads to its degradation (10). Recent work (11, 12) has found that the low density lipoprotein receptor-related protein (LRP) is an endocytic receptor responsible for TSP-1 uptake and degradation.
LRP is a member of the LDL receptor family, which includes the LDL
receptor (13), the VLDL receptor (14), gp330/megalin (15), and a newly
described apolipoprotein E receptor 2 (16). A 39-kDa protein, termed
the receptor-associated protein (RAP) (17), binds to several members of
the LDL receptor family (18-20) and blocks their ligand binding
capacity. LRP mediates the cellular uptake and subsequent degradation
of proteinases, such as tissue-type plasminogen activator (21) and
urinary-type plasminogen activator (22), proteinase-inhibitor
complexes, such as 2-macroglobulin-proteinase complexes
(23, 24), soluble amyloid precursor protein (25), and apolipoprotein
E-enriched lipoproteins (26, 27). LRP is expressed in many tissues, and
is a major endocytic receptor in the liver and in the central nervous
system (for reviews, see Refs. 28-30).
The goal of the present study was to identify additional LDL receptor family members capable of binding TSP-1, and to identify the region on TSP-1 responsible for its interaction with these receptors. These studies confirm that in addition to LRP and gp330/megalin, the VLDL receptor can also bind TSP-1 and mediate its cellular catabolism. Further, the studies identify the amino-terminal heparin binding domain of TSP-1 as a region responsible for mediating the catabolism of TSP-1 via interaction with LRP, and indicate that high affinity binding of TSP to LRP results from the trimeric structure of TSP.
TSP-1 was purified from thrombin-activated platelets as described (31). The NH2-terminal heparin binding domain of TSP-1 was purified by chromatography on heparin-Sepharose and size-exclusion chromatography following digestion of thrombospondin with chymotrypsin as described (32). LRP was isolated from human placenta as described by Ashcom et al. (23). Human RAP, expressed in bacteria as a fusion protein with glutathione S-transferase, and glutathione S-transferase were prepared and purified as described previously (18). Bovine serum albumin, fraction V (BSA) and heparin, fraction II (Mr 16,000) were purchased from Sigma. Thrombin was purchased from Enzyme Research Laboratories, Inc. (South Bend, IN). Proteins were labeled with [125I]iodine to a specific activity ranging from 2 to 10 µCi/µg of protein using IODOGEN (Pierce).
Expression and Purification of the HBDThe heparin binding
domain of TSP-1 was expressed as a fusion protein with glutathione
S-transferase in Escherichia coli using pGEX-2T
expression vector (Promega) as described previously (18). cDNA
fragments encoding residues 1-214 of mouse TSP-1 were prepared by
polymerase chain reaction using 21-base synthetic oligonucleotide primers and cDNA encoding mouse TSP-1 kindly provided by Vishva Dixit (University of Michigan Medical School, Ann Arbor, MI). The
product was sequenced to confirm that no errors were introduced during
polymerase chain reaction. The recombinant heparin binding domain
includes Cys-150 and Cys-214, which participate in intrachain disulfide
bond formation, but does not include cysteines involved in
trimerization of the thrombospondin subunits. GST-HBD was purified to
homogeneity by chromatography on GSH-Sepharose and/or chromatography on
heparin-Sepharose, followed by gel filtration on Superdex-75. Recombinant HBD (rHBD) was purified by chromatography on
heparin-Sepharose and size-exclusion chromatography after digestion of
the fusion protein with thrombin. Amino acid sequencing analysis
yielded a single, expected sequence. Protein concentrations were
determined by using an extinction coefficient of 12,700 M1 cm
1 at 280 nm for
recombinant HBD and 53,700 M
1
cm
1 for GST-HBD, which were calculated as described (33).
The molecular mass values of 23,500 and 49,000 daltons for rHBD and
GST-HBD, respectively, were calculated from the amino acid sequence.
Molecular mass values of 500,000, 23,500, and 97,000 daltons were used
to calculate the molar concentrations of TSP, rHBD, and GST-HBD, respectively.
Polyclonal antibodies against the heparin binding domain of mouse thrombospondin were prepared by immunizing a rabbit with rHBD (residues 1-214 of mouse TSP-1, preceded by amino-terminal glycine and serine). For immunization, rHBD was prepared by digestion of GST-HBD with thrombin. GST and rHBD were then separated by SDS-polyacrylamide gel electrophoresis. Gel slices containing rHBD were used for immunization. Antibodies were purified by chromatography on protein G-Sepharose. The specificity of the antibodies was confirmed by immunoblot analysis. An enzyme-linked immunosorbent assay (ELISA) confirmed that the antibody bound to rHBD and GST-HBD, as well as to purified human TSP-1.
Solid-phase Binding AssaysELISAs were performed as detailed elsewhere (22). Briefly, microtiter wells were coated with purified LRP at 5 µg/ml in 50 mM Tris, 150 mM NaCl, pH 8.0 (TBS), for 4 h at 37 °C and then blocked with 3% BSA in TBS. The LRP-coated microtiter wells were then incubated with increasing concentrations of TSP-1, GST-HBD, and rHBD in 3% BSA, TBS, 5 mM CaCl2, for 18 h at 4 °C. Bound proteins were detected with the rabbit polyclonal anti-HBD IgG (20 µg/ml) and goat anti-rabbit IgG (Bio-Rad) conjugated to horseradish peroxidase. Nonspecific binding was determined by measuring the binding to BSA-coated wells, and was subtracted from the data prior to nonlinear regression analysis. Data were analyzed by nonlinear regression analysis using the following equation.
A is the absorbance at 650 nm, Amax is the absorbance value at saturation, Amin is the background absorbance in the absence of ligand, [ligand] is the molar concentration of free ligand, and KD is the dissociation constant. Since the free ligand concentration is unknown in these experiments, the use of this equation assumes that the total amount of added ligand is far greater than the amount of ligand bound to the microtiter wells. Under these conditions, the amount of free ligand is approximately equal to the total ligand concentration, and thus total ligand concentrations can be substituted into Equation 1.
![]() |
(Eq. 1) |
Cell Binding and Cellularly Mediated Ligand Internalization and Degradation Assays
A normal mouse embryonic fibroblast line (MEF) and a mouse embryonic fibroblast cell line that is genetically deficient in LRP biosynthesis (PEA13) were obtained from Dr. Joachim Herz (University of Texas Southwestern Medical Center, Dallas, TX) and maintained as described (34). Cells were seeded into 12-well dishes (Corning, Corning, NY) at 1 × 105 cells/well 1 day prior to the assay and allowed to grow for 24 h at 37 °C, 5% CO2. Cellular internalization and degradation assays were conducted as described previously (11). Surface binding, internalization, and degradation of 125I-labeled proteins by cells was measured after incubation for indicated time intervals at 37 °C in 0.5 ml of Dulbecco's modified medium (Mediatech, Washington, DC) containing 0.3 mg/ml BSA. Surface binding and internalization is defined as radioactivity that is sensitive or resistant, respectively, to release from cells by trypsin (50 µg/ml) and proteinase K (50 µg/ml) (Sigma) in buffer containing 5 mM EDTA. Degradation is defined as radioactivity in the medium that is soluble in 10% trichloroacetic acid. In all experiments a control was included, in which the amount of degradation products generated in the absence of cells was also measured and subtracted from each data point. The binding of 125I-labeled TSP-1 to MEF was conducted at 4 °C as described (35).
Adenovirus-mediated Gene TransferRecombinant
replication-deficient adenoviral vectors were prepared as described
(36). The recombinant adenoviruses used were Ad-LacZ, an adenovirus
containing lacZ cDNA under the human cytomegalovirus
enhancer and 3-promoter, and Ad-VLDLR (also called Ad.CMVVLDLR in Ref.
36), an adenovirus containing the human VLDL receptor under the human
cytomegalovirus enhancer and 3
-promoter. PEA13 cells were infected as
described previously (37). To assess the effectiveness of the infection
with Ad-VLDLR, immunoblotting and RAP ligand blotting of cell extracts
were routinely employed.
Tryptophan fluorescence was monitored using an SLM8000 photon-counting spectrofluorimeter (Aminco). Protein (1 µM initial concentration) in TBS, pH 7.4, was denatured by stepwise titration with 10 M urea solution in TBS in a 4-mm pathlength cuvette (Hellma) with continuous stirring. Temperature was maintained at 25 °C by using a thermostated cuvette holder connected to a Neslab water bath. The excitation wavelength was 285 nm, and emission spectra were recorded for each urea concentration. The spectrum of each sample was corrected by subtraction of the buffer alone and normalized to 1 µM protein concentration.
Equilibrium SedimentationEquilibrium sedimentation measurements were performed using a Beckman XL-A Optima Analytical Ultracentrifuge equipped with absorbance optics and a Beckman An-60Ti rotor. Proteins were loaded at three concentrations (0.25, 0.5, and 1 µM) in 20 mM Tris, 150 mM NaCl, pH 7.5, into a six-hole centerpiece and spun at 11,000 rpm (GST-HBD) or 24,000 rpm (rHBD and GST) for 24 h. Twenty data sets for three concentrations were averaged and jointly fitted for a singular molecular weight. Compositional partial specific volumes for the proteins were calculated according to Zamyatnin (38).
Previous work has suggested that LRP plays an important role
in the catabolism of TSP-1 (11, 12). These studies documented that
TSP-1 binds to LRP in vitro and that cells expressing LRP are able to catabolize TSP-1. The catabolism of
125I-labeled TSP-1 was inhibited by antibodies against LRP
and by RAP (11), a known antagonist of LRP activity, providing evidence that LRP-mediated endocytosis is a major pathway for TSP-1 catabolism. To determine if other mechanisms besides LRP-mediated endocytosis exist
that lead to the degradation of TSP-1, the catabolism of 125I-labeled TSP-1 was measured in cells that are
genetically deficient in LRP (PEA-13 mouse embryonic fibroblasts) (39)
and was compared to the catabolism of 125I-labeled TSP-1 in
wild type mouse embryonic fibroblasts (MEF) expressing LRP. For these
experiments, 125I-labeled TSP-1 was incubated with cells at
37 °C, and at selected time intervals, the amount of radioactivity
that was released with proteinase treatment (Fig.
1A), internalized (Fig. 1B), and released back into the medium as trichloroacetic acid-soluble fragments
(Fig. 1C) was measured. When added to cells at 37 °C, 125I-labeled TSP-1 demonstrated a slow,
time-dependent association with cells or matrix components,
regardless of whether or not the cells expressed LRP. Heparin prevented
the association of TSP-1 with the cell surface. In contrast, RAP did
not inhibit this process; in fact, increased association of TSP-1 with
cells expressing LRP was noted in the presence of RAP (Fig.
1A). We attribute this accumulation of
125I-labeled TSP-1 to blocking of the catabolic pathway for
TSP-1 in the presence of RAP. In contrast, the internalization (Fig. 1B) and degradation (Fig. 1C) of
125I-labeled TSP-1 strongly correlated with LRP expression.
LRP-containing fibroblasts rapidly internalize 125I-labeled
TSP-1 leading to its degradation, and these processes were inhibited by
RAP. Cells deficient in LRP are inefficient in internalizing and
degrading TSP-1 (Fig. 1, panels B and C). These
data indicate that LRP-mediated endocytosis is a major pathway for the
cellular catabolism of TSP-1.
The VLDL Receptor Is Also Capable of Mediating the Cellular Uptake of TSP-1
Like LRP, the VLDL receptor also binds multiple ligands
(37). However, in contrast to LRP, the VLDL receptor is expressed in
the endothelium (37, 40). Since TSP-1 is catabolized by endothelial
cells (41), it was of interest to determine if the VLDL receptor could
mediate the cellular catabolism of TSP-1. For these experiments, an
adenoviral vector was utilized to express the VLDL receptor in PEA-13
fibroblasts, a cell line that does not express this receptor or LRP. As
a control, PEA13 fibroblasts were infected with an adenovirus vector,
Ad-LacZ, containing the lacZ cDNA instead of the VLDL
receptor cDNA. Fig. 2 shows that cells expressing
the VLDL receptor are able to mediate the cellular degradation of
125I-labeled TSP-1 (left panel). In contrast,
cells that lack the VLDL receptor (PEA-13 fibroblasts infected with
Ad-LacZ, middle panel) are unable to mediate the degradation
of 125I-labeled TSP-1. For comparison, the amount of
cellular degradation of 125I-TSP observed in wild type
fibroblasts, which express LRP, is also shown (right panel).
These experiments demonstrate that like LRP, the VLDL receptor can bind
and mediate the cellular internalization of TSP-1 leading to its
degradation, and suggest that the RAP-sensitive receptor in endothelial
cells that catabolizes TSP-1 (12) very likely is the VLDL receptor.
Recombinant Heparin Binding Domain Assumes a Stable Folded Conformation
The next goal of our studies was to identify the portion of TSP-1 responsible for binding to LRP. Based on the observation (12)2 that heparin interferes with the binding of TSP-1 to LRP in vitro, and on the recent observation (42) that a truncated form of TSP-1 containing the amino-terminal portion of the molecule is catabolized by cells, we speculated that heparin binding portions of TSP-1 may participate in the binding of TSP-1 to LRP. It is well known that the amino-terminal domain of TSP-1, consisting of residues 1-214, contains a high affinity heparin binding site (43, 44). To investigate the possibility that this region (called the heparin binding domain or HBD) may itself bind to LRP, we expressed this portion (residues 1-214) of mouse TSP-1 as a fusion protein with glutathione S-transferase.
Fig. 3 demonstrates an analysis by SDS-polyacrylamide
gel electrophoresis under nonreducing conditions of GST-HBD (lane
1), cleaved rHBD (lane 2), HBD isolated following
chymotryptic cleavage of human TSP-1 (lane 3), and GST
(lane 4). Under nonreducing conditions, the rHBD migrated
close to its calculated molecular mass of 23.5 kDa. The reduced rHBD
displayed a slightly slower electrophoretic mobility than nonreduced
rHBD (data not shown), an observation consistent with the notion that
the cysteine residues in positions 150 and 214 form the expected
intrachain disulfide bond. rHBD bound to heparin-Sepharose and was
eluted at an ionic strength similar to that required to elute the HBD
generated from TSP-1 by proteolytic digestion, which suggests that the
rHBD is appropriately folded. To assess the folding properties of rHBD,
the effect of increasing urea concentrations upon the intrinsic
fluorescence of rHBD was monitored. The results (Fig. 4)
demonstrate that the environment of the tryptophan residues in rHBD is
dramatically altered as the urea concentration is increased. In the
presence of urea, the tryptophan fluorescence is quenched and the
wavelength of maximal fluorescence emission is shifted to that expected
for a more polar environment surrounding the tryptophan residue. These data are consistent with unfolding of the rHBD. A replot of the fluorescent intensities versus urea concentration (Fig. 4,
inset) demonstrates that a cooperative unfolding transition
was observed. These results provide compelling evidence that the rHBD
is folded into a compact structure that can be disrupted with urea. A
similar titration curve was obtained for HBD isolated by chymotryptic digestion of TSP-1 (Fig. 4, inset), although it appears to
denature at a slightly lower concentration of urea, possibly due to the presence of carbohydrate.
To assess whether or not rHBD exists in solution as a monomer, we utilized analytical ultracentrifugation. The apparent weight-average molecular weight (Mrapp) of rHBD, GST-HBD, and GST was determined from sedimentation equilibrium measurements. The results of these studies are summarized in Table I, and confirm that the rHBD is a monomer in solution. On the other hand, GST-HBD forms a stable dimer in solution, due to the self-association properties of GST.
|
To determine if the HBD
was capable of binding to LRP, an ELISA was utilized. In this assay,
microtiter wells were coated with LRP, and increasing concentrations of
TSP, GST-HBD, and rHBD were added. The binding of each of these
molecules to the LRP-coated microtiter wells was detected with a
polyclonal antisera raised against purified rHBD. The results of this
assay (Fig. 5) demonstrate that TSP, GST-HBD, and rHBD
bind to microtiter wells coated with LRP in a
dose-dependent manner. The apparent affinity for the interaction of these molecules was determined by nonlinear regression analysis. A KD of 7 nM was obtained for
TSP, which is in excellent agreement with the KD
determined previously (11). In the case of GST-HBD, a
KD of 10 nM was obtained. In several
experiments, this value ranged from 10 to 25 nM. The KD for rHBD was considerably larger, with a value of 830 nM determined from analysis of the data. Since our
studies indicate that rHBD is folded, the weaker affinity of rHBD for LRP-coated microtiter wells most likely arises from the fact that rHBD
exists as a monomer in solution, whereas GST-HBD exists as a dimer. The
association of dimeric GST-HBD with multiple binding sites on LRP, or
with multiple LRP molecules is likely to result in a complex less prone
to dissociate, thus increasing the affinity of the interaction.
Binding of 125I-Labeled TSP-1 to MEF at 4 °C Is Inhibited by GST-HBD
The ability of GST-HBD to compete for cell
surface binding of 125I-labeled TSP-1 was next measured.
These experiments were performed at 4 °C to prevent cellularly
mediated catabolism of the labeled TSP-1. The results demonstrate that
increasing concentrations of GST-HBD inhibit the binding of
125I-labeled TSP-1 to cells (Fig.
6A), while GST had no effect in the assay.
Fig. 6B shows the effect of other molecules on the binding
of 125I-labeled TSP-1 to MEF. RAP (800 nM) was
unable to inhibit the binding of 125I-labeled TSP-1 to the
cells, while heparin was effective in inhibiting binding. These data
indicate that, although RAP binds to heparin (45, 46) and appears
capable of interacting with cell surface proteoglycans (47), it is
unable to compete for TSP-1 binding sites on MEF. Furthermore, the
inability of RAP to compete for 125I-labeled TSP-1 binding
to these cells indicates that LRP is not the prominent cell surface
binding site for 125I-labeled TSP-1.
rHBD Is Internalized and Degraded by Fibroblasts Expressing LRP, but Not by LRP-deficient Fibroblasts
Having demonstrated an
in vitro interaction between LRP and the HBD of TSP, we next
sought to determine if LRP could also mediate the degradation of this
molecule. Fibroblasts expressing LRP rapidly internalized
125I-labeled GST-HBD (Fig. 7A)
and subsequently degraded this molecule (Fig. 7B). Both the
internalization and degradation were blocked when RAP was included
during the incubation. In contrast, cells that lack LRP were unable to
internalize or degrade 125I-labeled GST-HBD.
125I-Labeled GST was not internalized or degraded by either
of these two cell lines (data not shown).
Studies were also conducted to examine the ability of mouse embryonic
fibroblasts to internalize and degrade rHBD. Fig. 8 demonstrates that 125I-labeled rHBD is readily internalized
and degraded by fibroblasts expressing LRP, but not by cells that lack
LRP. Both the internalization and degradation was effectively blocked
by heparin and RAP. The internalization and degradation were also
reduced by excess unlabeled rHBD. These studies indicate that rHBD,
like GST-HBD, is also internalized by LRP-expressing cells. Together,
these data provide convincing evidence that the HBD of TSP-1 is
internalized and degraded by cells via LRP-mediated uptake.
Cellular Catabolism of 125I-Labeled TSP-1 Is Inhibited by GST-HBD and rHBD
To determine if the catabolism of TSP-1 is
blocked by GST-HBD, 125I-labeled TSP-1 was incubated with
LRP-expressing fibroblasts in the presence of increasing concentrations
of GST-HBD or RAP, and the extent of cellularly mediated degradation
measured (Fig. 9A). The results demonstrate
that both GST-HBD and RAP completely inhibit the degradation (Fig.
9A) of 125I-labeled TSP-1, while GST alone had
no effect on the catabolism of TSP-1. Fig. 9B demonstrates
that the catabolism of 125I-labeled TSP-1 is inhibited in
the presence of rHBD in a dose-dependent manner. In this
case, high concentrations of rHBD were required, consistent with the
weaker affinity of rHBD for LRP. These data confirm that the
amino-terminal HBD can block the cellular catabolism of TSP.
Cellular Catabolism of 125I-Labeled TSP-1 Is Inhibited by Antibodies Prepared against the rHBD
To further investigate
the possibility that the HBD of TSP-1 is responsible for mediating its
catabolism, the effect of antibodies prepared against rHBD on the
catabolism of 125I-labeled TSP-1 was investigated. The
result of these studies indicate that the antibodies reduce both the
uptake (Fig. 10A) and degradation (Fig.
10B) of 125I-labeled TSP-1 in a
dose-dependent manner. In contrast, a control IgG had no
effect on TSP-1 catabolism.
TSP-1 is a modular protein capable of interacting with cells
through distinct domains. Previous studies have shown that TSP-1 is
rapidly taken up by cells and degraded via LRP-mediated endocytosis (11, 12). Given that TSP-1 is known to bind to various integrin receptors including 3
1 (48),
4
1,
5
1
(49),
v
3 (50), as well as CD36 (51), it
was of interest to explore whether receptors besides LRP could mediate
the uptake and degradation of TSP-1. To examine this possibility, we
have investigated the catabolism of 125I-labeled TSP-1 in
mouse embryonic fibroblasts genetically deficient in LRP (39). Our
results indicate that, although these cells are capable of binding to
TSP-1 to a similar extent as control mouse fibroblasts, genetic
deletion of LRP results in the complete loss of the ability of these
cells to internalize or degrade 125I-labeled TSP-1. This
convincingly demonstrates that LRP is the prominent receptor that
mediates TSP-1 endocytosis in mouse embryonic fibroblasts. These
observations, in conjunction with our previous reports demonstrating
that LRP antibodies are potent inhibitors of TSP-1 catabolism (11, 12),
indicate the LRP-mediated endocytosis is likely a major pathway for the
catabolism of TSP-1.
Our data also document that, like LRP and gp330, the VLDL receptor can also mediate the cellular catabolism of TSP-1. While these receptors have overlapping ligand specificity, they exhibit a different pattern of expression. LRP is widely expressed but prominent in liver and brain (52); VLDL receptor in heart, muscle, adipose tissue, and brain (53); glycoprotein 330 is prominently expressed in specialized epithelia of the brain, lung, and kidney (54, 55). Recent studies (11, 40) indicate that the VLDL receptor is expressed in the endothelium and probably represents the RAP-sensitive receptor responsible for the catabolism of TSP-1 in these cells.
TSP-1 is a modular protein composed of several domains with different activities. A major objective of the current investigation was to identify the portion of TSP-1 that is able to interact with LRP. Our initial work focused on the amino-terminal heparin binding domain, since heparin appears to influence the binding of TSP-1 to LRP. The high affinity heparin binding site of TSP-1 resides in the amino-terminal domain encompassing residues 1-218 (43, 44). This portion of the molecule has been previously expressed in E. coli, and appears to be appropriately folded, as evidenced by its high affinity for heparin and by its ability to partially block binding of 125I-labeled TSP-1 to carcinoma cells (43).
The present studies indicate that the TSP-1 heparin binding domain contains an LRP-recognition determinant. This conclusion is supported by several independent lines of evidence. First, both GST-HBD and rHBD bind to purified LRP in an in vitro binding assay. Second, 125I-labeled GST-HBD or rHBD is endocytosed and degraded by fibroblasts that express LRP, but not by fibroblasts that are deficient in LRP. Third, the cellular degradation of 125I-labeled GST-HBD and rHBD is blocked by RAP, an inhibitor of LRP function. Fourth, we demonstrated that the GST-HBD or rHBD completely blocked catabolism of 125I-labeled TSP-1 in a dose-dependent manner. Finally, antibodies prepared against the rHBD inhibited the cellularly mediated uptake and degradation of 125I-labeled TSP in a dose-dependent manner. Taken together, these data provide compelling evidence that the amino-terminal domain of TSP-1 binds to LRP, and therefore represents a domain of TSP-1 that is responsible for its cellular catabolism.
Binding measurements noted that GST-HBD bound to LRP with an affinity comparable to that measured for the binding of TSP-1 to LRP. However, when cleaved from GST, the rHBD still bound to LRP, but with a 40-fold reduced affinity. The weaker affinity is likely due to the fact that rHBD exists as a monomer in solution, whereas GST-HBD is a dimer. Dimerization appears to result from the GST moiety, since this molecule also exists in solution as a dimer. At this time it is not known if the dimeric GST-HBD binds to multiple sites on a single LRP molecule, or if it binds to multiple LRP molecules. The data suggest that the high affinity of TSP-1 for LRP is likely a result of its trimeric structure, which may result in interactions of TSP with multiple binding sites on a single LRP molecule, or in the interactions of a single TSP molecule with multiple LRP molecules.
These data also highlight that the recognition determinants on TSP-1 for both LRP and for proteoglycans reside within the same TSP-1 domain. This is of significance since studies on the catabolism of 125I-labeled TSP-1 by cells have observed that, while the process is mediated by LRP, cell surface proteoglycans seem to play an important role by facilitating this process. Thus, treatment of WI-38 fibroblasts and human smooth muscle cells with heparitinase reduces the amount of 125I-labeled TSP-1 degraded by 80% and 91%, respectively (11, 12). Furthermore, Chinese hamster ovary cells defective in glycosaminoglycan synthesis show a reduction in the amount of TSP-1 that is internalized and degraded (56). These data suggest that localization of TSP-1 on the cell surface via binding to cell surface proteoglycans is an important component of the LRP-mediated catabolism of TSP-1.
In this regard, the catabolism of TSP-1 resembles that of another LRP
ligand, lipoprotein lipase. Lipoprotein lipase not only binds LRP, but
also promotes the binding of VLDL to LRP (57). Incubation of human
fibroblasts with heparinase reduces the surface binding,
internalization, and degradation of lipoprotein lipase-enriched VLDL,
but does not alter the catabolism of
125I-2M-proteinase complexes, which is known
to be mediated by LRP. Thus, proteoglycans also facilitate the
lipoprotein lipase-promoted catabolism of VLDL by cultured cells.
Interestingly, like the TSP-1 interaction with LRP, proteoglycans are
not required for binding of lipoprotein lipase-enriched VLDL to
purified LRP in a cell-free system. Finally, like TSP-1, an LRP and
proteoglycan recognition determinant of lipoprotein lipase also resides
within the same domain (58).
While the exact role of proteoglycans is not known, proteoglycans may sequester ligands, such as TSP-1 and lipoprotein lipase on the cell surface, increasing the local concentration of the ligand. At this time it is not known if the proteoglycans are internalized along with LRP and its ligand, or if the ligand first dissociates from the proteoglycan prior to its LRP-mediated internalization. Since the residues within the HBD of TSP-1 that are important for binding to heparin have been identified (59), it will be possible to determine if these residues are distinct from those that interact with LRP. This should give insight into the cooperative role between cell surface proteoglycans and LRP in mediating the binding and cellular catabolism of TSP-1.
We appreciate the help of Evan Behre in preparing RAP, and we thank Dr. J. Herz (University of Texas Southwestern Medical Center) for providing the mouse embryonic fibroblast cell lines that are genetically deficient in LRP. We thank Dr. C. Vinson (National Institutes of Health) for help in cloning of recombinant proteins.