Metabolism of thrombospondin 2. Binding and degradation by 3t3 cells and glycosaminoglycan-variant Chinese hamster ovary cells.

Thrombospondin 1 (TSP1) and thrombospondin 2 (TSP2) are members of the thrombospondin family that have a similar structural organization but somewhat different functional activities. Iodinated recombinant mouse TSP2 bound to NIH 3T3 cells and was internalized and degraded through a chloroquine-inhibitable pathway. TSP2 degradation was saturable, specific, and similar to the kinetics of degradation of TSP1. Human platelet TSP1, recombinant mouse TSP1, and recombinant mouse TSP2 cross-competed with one another for degradation by 3T3 cells. Degradation of TSP2 was less sensitive to inhibition by heparin than degradation of TSP1. This is in agreement with differences in heparin-binding affinity of the two TSPs. Degradation of TSP2 was slower in cultures of Chinese hamster ovary (CHO) cells lacking heparan sulfate proteoglycans than in wild type CHO cells or in cultures of 3T3 cells treated with heparitinase than in untreated 3T3 cells. Degradation of TSP2 was inhibited by antibodies against the low density lipoprotein receptor-related protein (LRP) or by the 39-kDa receptor-associated protein, a known antagonist of LRP. This study indicates that TSP2 and TSP1 are metabolized by a common internalization and degradation pathway involving heparan sulfate proteoglycan and LRP. Competition for this pathway is a possible mechanism whereby cells can control the levels and ratio of TSP1 and TSP2 in the extracellular milieu.

Thrombospondin 1 (TSP1) and thrombospondin 2 (TSP2) are members of the thrombospondin family that have a similar structural organization but somewhat different functional activities. Iodinated recombinant mouse TSP2 bound to NIH 3T3 cells and was internalized and degraded through a chloroquine-inhibitable pathway. TSP2 degradation was saturable, specific, and similar to the kinetics of degradation of TSP1. Human platelet TSP1, recombinant mouse TSP1, and recombinant mouse TSP2 cross-competed with one another for degradation by 3T3 cells. Degradation of TSP2 was less sensitive to inhibition by heparin than degradation of TSP1. This is in agreement with differences in heparinbinding affinity of the two TSPs. Degradation of TSP2 was slower in cultures of Chinese hamster ovary (CHO) cells lacking heparan sulfate proteoglycans than in wild type CHO cells or in cultures of 3T3 cells treated with heparitinase than in untreated 3T3 cells. Degradation of TSP2 was inhibited by antibodies against the low density lipoprotein receptor-related protein (LRP) or by the 39-kDa receptor-associated protein, a known antagonist of LRP. This study indicates that TSP2 and TSP1 are metabolized by a common internalization and degradation pathway involving heparan sulfate proteoglycan and LRP. Competition for this pathway is a possible mechanism whereby cells can control the levels and ratio of TSP1 and TSP2 in the extracellular milieu.
The structural modules of TSP2 are similar to those of TSP1, with an increasing gradient of sequence identity from the NH 2terminal module (38% identity) to the COOH-terminal domain (82% identity) (4,6). The patterns of expression of TSP1 and TSP2 mRNAs are distinct in tissues of embryonic and developed mice (56,57). To compare the structure and function of TSP1 and TSP2, we have expressed mouse TSP2 (mTSP2) in a baculovirus system as a disulfide-bonded homotrimer (58). mTSP2 supports adhesion for endothelial cells, osteosarcoma cells, and colon carcinoma cells by mechanisms similar but not identical to those of TSP1. Adherence to both TSPs appears to utilize heparan sulfate proteoglycans and ␣ v ␤ 3 integrin, and is regulated by Ca 2ϩ and reduction. One major difference between adhesive activities of TSP1 and TSP2 is the differential sensitivity to inhibition of adhesion by heparin (58). In another adhesion system where adrenocortical cells are used as a source, bovine TSP2 (also known as corticotropin-induced secreted protein, CISP) shows an antiadhesive activity (59). TSP2, like TSP1, binds TGF-␤1 (55). TSP2 does not, however, activate latent TGF-␤1 (55). This lack of activity apparently is due to substitution of the activating RFK sequence in TSP1 (55) with the trypsin-susceptible, nonactivating RIR sequence in TSP2 (55,58). TSP2 inhibits the activation of latent TGF-␤1 by TSP1, presumably through the common TGF-␤1 binding sequence GGWSHW present in both TSP1 and TSP2 (55). Therefore TSP2 may act as a buffer to the activation of latent TGF-␤1 by TSP1.
Because TSP1 has potent biological activities, and TSP2, besides its own functions, may regulate TSP1 function, one may postulate that the ratio of TSP1 and TSP2, as well as their amount, is regulated by their expression and half-life. TSP1 is secreted from platelet ␣-granules upon activation (19,60). It is also produced by a variety of normal and transformed cell lines (19,61,62). Whereas TSP1 production is up-regulated dramatically by serum or growth factors, TSP2 expression is constitutive (5,6). It has been shown that platelet TSP1 can bind to cells and incorporate into the extracellular matrix (63,64), or be cleared by cells via endocytosis and lysosomal degradation (30,(63)(64)(65)(66)(67)(68). Cell surface heparan sulfate proteoglycans are required for binding and degradation of platelet TSP1 (30,(63)(64)(65)(66)(67)(68). Low density lipoprotein receptor-related protein (LRP) has been shown recently to synergize with heparan sulfate proteoglycans in mediating internalization and degradation of platelet TSP1 (67,68). One may hypothesize, based on the homology between TSP1 and TSP2, that their metabolism is similar. However, many homologous proteins have different receptors, and the NH 2 -terminal region that mediates binding to heparin is the part with the lowest sequence identity between TSP1 and TSP2. Thus, one may also hypothesize that the metabolism of TSP1 and TSP2 is different. To evaluate these hypotheses, we carried out experiments to investigate whether recombinant mouse TSP2 is metabolized by a mechanism similar to that of platelet or recombinant TSP1 and whether TSP1 and TSP2 compete for the same degradation pathway.

MATERIALS AND METHODS
Proteins and Reagents-Human platelet TSP1 (hTSP1) and recombinant mTSP2 produced with baculovirus were purified as described previously (58). Production and purification of recombinant mTSP1 were similar to those used for mTSP2. Briefly, mouse TSP1 cDNA including bases 51-3751 flanked by MluI linkers in the pJDM eukaryote expression vector (a generous gift from Dr. Vishva Dixit) (69) was used. NcoI and EcoRI digestion was used to generate a 5Ј mouse TSP1 fragment containing bases 210-1404 which lacked excess 5Јuntranslated region. A 3Ј fragment was generated by EcoRI digestion and incomplete BamHI digestion which cut at the BamHI site in the pJDM multiple cloning region but not inside the mouse TSP1 3Ј cDNA. These two fragments were subcloned into the baculovirus transfer vector pAcSG2 (Pharmingen, San Diego, CA) linearized by NcoI and BglII, utilizing the compatible cohesive ends of BamHI and BglII. Recombinant mTSP1 virus was generated using the BaculoGold transfection system (Pharmingen) with Lipofectin per the manufacturer's instructions. After transfection, recombinant viruses were plaque-purified once, and third passage virus in serum-free medium SF900 II (Life Technologies, Inc.) was used to infect Spodoptera frugiperda cells.
Generation of Polyclonal Antibodies against mTSP2-Purified mTSP2, 30 g, was transferred to nitrocellulose paper after electrophoresis in SDS. The nitrocellulose paper containing the mTSP2 band was cut out, washed with H 2 O, air-dried, frozen in a dry ice/isopropanol bath, and crushed into a fine powder with a glass rod. The powder was emulsified with complete Freund's adjuvant and injected subcutane-ously into two New Zealand White male rabbits (30 g each). The rabbits were boosted three times at 1-month intervals with electrophoretically repurified mTSP2, 35 g, on nitrocellulose in incomplete Freund's adjuvant, followed by boosts with purified soluble mTSP2 every 4 -6 weeks. Antibody titers and antibody specificities were checked by enzyme-linked immunosorbent assay (ELISA) on 96-well plates coated with hTSP1, mTSP1, or mTSP2. Antibodies to hTSP1 were produced by a similar protocol. The specificities and species crossreactivities of rabbit antibodies to mTSP2 or hTSP1 were determined by direct ELISA. The anti-mTSP2 antisera had titers against mTSP2 of 1:20,000 -1:40,000, against hTSP1 of 1:600 and against mTSP1 of 1:1,500. Anti-TSP1 antisera had a titer against mTSP1 of 1:15,000 and against mTSP2 of 1:1,200.
Iodination of Proteins-Purified TSP, 100 g, was iodinated with 0.5 mCi of Na 125 I in the presence of 0.5 mM chloramine-T as described previously (63,65). After 1 min, phenylmethylsulfonyl fluoride-treated bovine albumin was added to a concentration of 10 mg/ml. 125 I-TSP was repurified by affinity chromatography on heparin-agarose and eluted by 1 M NaCl in 0.3 mM Ca 2ϩ and 10 mM Tris, pH 7.4. Albumin was added to a final concentration of 2 mg/ml, and 125 I-TSPs were stored as small aliquots at Ϫ70°C until use. Iodinated TSPs had the expected mobilities in autoradiograms of polyacrylamide gels after electrophoretic separation in SDS without and with reduction. Specific activities were 1.2-5.2 mCi/mg TSP. Iodinated TSPs had 2-7% trichloroacetic acidsoluble radioactivity.
Binding and Degradation of 125 I-TSPs-NIH 3T3 cells were obtained from the American Type Culture Collection (Rockville, MD) and maintained in high glucose Dulbecco's modified Eagle medium (DME) containing 10% fetal bovine serum at 37°C in an incubator containing 8% CO 2 . Chinese hamster ovary (CHO) cells were maintained as described previously (66) in a 5% CO 2 incubator. Cells were grown to confluence on 24-well tissue culture plates (Costar, Cambridge, MA) and washed three times with DME before assays. Binding and degradation assays were carried out in DME with 0.2% bovine albumin containing 100 units/ml penicillin G and 150 g/ml streptomycin sulfate according to procedures described before (63)(64)(65)(66). Binding medium containing 125 I-TSP was incubated with cells in the CO 2 incubator at 37°C for various times. After the incubation, binding medium was removed and mixed with trichloroacetic acid at a final concentration of 10%. After incubation on ice for 15 min, the precipitate was removed by centrifugation. The increase in trichloroacetic acid-soluble radioactivity above the baseline value during the incubation was taken as TSP that had been degraded by the cells. The negligible increase in trichloroacetic acidsoluble radioactivity during incubation in plates without cells was considered as the baseline. At the end of the incubation, cell layers were washed with cold Tris-buffered saline three times and dissolved in 0.2 N NaOH to determine the amount of total cell protein and 125 I-TSP that was associated with the cell layers.
In some experiments, cells were treated with heparitinase (0.4 unit/ ml), or heparitinase and heparinase I (0.4 unit/ml), or/and chondroitinase ABC (0.1 or 0.05 unit/ml used in two separate experiments) in the binding medium containing 25 mM HEPES, pH 7.0, for 45 min at 37°C before 125 I-TSPs were added, and the incubation was carried out for additional 4 h.
Immunofluorescence Studies-Cells for immunofluorescence studies were grown on glass coverslips to subconfluence. Cells were washed three times with DME. Some coverslips with cells were incubated with 20 g/ml TSP in DME containing 0.2% albumin for 1 h, with or without blocking reagents, before being washed three times and fixed with 3.7% paraformaldehyde. Cells on one set of coverslips were permeabilized with chilled acetone for 5 min. Coverslips were stained with primary antibodies against TSP1 or TSP2 at 1:1000 dilution for 1 h at 37°C followed by a 45 min incubation at 37°C with rhodamine conjugated goat anti-rabbit IgG at 1:100 dilution. Coverslips were rinsed three times with Tris-buffered saline before mounting on glass slides in glycerol gelatin (Sigma). Slides were observed and photographed on a Nikon microscope equipped with epifluorescence and phase contrast. Controls for specific staining included nonimmune serum at the same dilution and immune serum whose signals were blocked by preincubation with either hTSP1 or mTSP2.

RESULTS AND DISCUSSION
Internalization and Degradation of TSP2 by 3T3 Cells-When iodinated mTSP2 was incubated with NIH 3T3 cells, it became rapidly associated with the cell layers (Fig. 1A). Trichloroacetic acid-soluble radioactivity began to appear in the supernatant after about 30 min and increased linearly over the 4-h incubation period (Fig. 1B). No intermediate degradation products were detected either in the culture medium or in the cell associated pool by polyacrylamide gel electrophoresis in SDS followed by autoradiography (data not shown). The increase in trichloroacetic acid-soluble radioactivity in the medium was blocked to near background by chloroquine, an inhibitor of endogenous acidification and the lysosomal degradation pathway (75, 76) (data not shown). We compared the metabolism of mTSP2 to that of human platelet TSP1 and, in order to account for any differences that might arise from species differences or cell of origin, recombinant full length mTSP1 expressed by the same methods used to express mTSP2. Iodinated hTSP1 or mTSP1 was metabolized with kinetics similar to that of mTSP2 (Fig. 1). From these data, we conclude that mTSP1 and mTSP2, like human platelet TSP1 (30,(63)(64)(65)(66)(67)(68), bind to cells, become internalized, and are degraded through a lysosomal pathway.
Degradation of TSP2 by 3T3 Cells: Saturability and Specificity-When increasing concentrations of mTSP2 were incubated with 3T3 cells, degradation was saturable (Fig. 2). Replotting of the degradation data to a double reciprocal plot yielded a straight line, with an apparent "K m " of 47 Ϯ 21 g/ml (104 Ϯ 47 nM) and a maximum turnover rate of 9 Ϯ 3 g of mTSP2 degraded/mg cell protein per 4 h (data are expressed as mean Ϯ S.E. of three separate experiments, each with duplicate numbers). The values for hTSP1 and mTSP1 were 130 Ϯ 48 g/ml (289 Ϯ 106 nM) and 17 Ϯ 6 g of TSP degraded/mg of cell protein/4 h, and 15 Ϯ 8 g/ml (33 Ϯ 18 nM) and 4 Ϯ 2 g of TSP degraded/mg of cell protein/4 h, respectively. Degradation of each 125 I-TSP at 1 g/ml was inhibited by approximately 50% by 50 g/ml of the same unlabeled TSP in the binding medium (Table I). PF4, 15 g/ml, inhibited degradation of 125 I-labeled mTSP1 and TSP2 to 69 Ϯ 12% and 68 Ϯ 7% of control, respectively (data not shown). Degradation was not inhibited by 100 g/ml fibronectin, fibrinogen, native or ureatreated vitronectin, or type I collagen (data not shown).
Competition of TSP2 and TSP1 for Degradation by 3T3 Cells-Incubation with mTSP2 at 50 g/ml reduced degradation of labeled hTSP1 and mTSP1 to 49 Ϯ 8%, and 71 Ϯ 6%, respectively (Table I). When 50 g/ml unlabeled hTSP1 was incubated with 125 I-TSPs, degradation of labeled mTSP1 and mTSP2 by 3T3 cells was reduced to 64 Ϯ 4% and 50 Ϯ 9% of control, respectively (Table I). Incubation with 50 g/ml of mTSP1 reduced degradation of labeled hTSP1 and mTSP2 to 51 Ϯ 11% and 45 Ϯ 11% of control, respectively (Table I). Due to limited amounts and solubility of proteins, we were not able to test higher concentrations of mTSP2 or mTSP1 for cross inhibition studies. Incubations with hTSP1 at 100 g/ml resulted in greater inhibition of degradation of labeled mTSP1 or mTSP2. 3T3 cells appeared necrotic after incubation with hTSP1 at concentrations higher than 150 g/ml. The fact that TSP2 and TSP1 cross-inhibited degradation of each other in rough approximation to the calculated K m values of degrada- tion indicates that the two proteins are internalized and degraded through similar pathways.
Inhibition of TSP2 Degradation by Heparin-It was shown previously that high capacity binding and degradation of hTSP1 by cells in culture is inhibited by heparin (30,(63)(64)(65)(66)(67)(68). TSP2 differs most from TSP1 in the NH 2 -terminal PARP module that mediates binding to heparin (77, 78) and binds heparin less avidly than TSP1 as assessed by salt concentration re-quired to disrupt binding (58,69). We therefore compared the effect of heparin on the binding and degradation of mTSP1 and mTSP2. Binding (Fig. 3A) and degradation (Fig. 3B) of mTSP2 were inhibited by heparin. The EC 50 for heparin inhibition of degradation of 1 g/ml mTSP2 was 4 g/ml. Binding and degradation of mTSP1 and hTSP1 by 3T3 cells, in contrast, were inhibited by heparin at a 6 -10-fold lower concentration. Greater than 50% inhibition was achieved at a heparin concentration of 0.4 -0.6 g/ml or roughly 1:1 molar concentrations of heparin and TSP1 subunit (Fig. 3).
Effect of Heparitinase on Degradation of TSP2 by 3T3 Cells-Inhibition of degradation by PF4, a heparin-binding protein like TSP1 and TSP2, and by heparin itself suggests that TSP2 interacts with a heparan sulfate proteoglycan at some point in the degradative pathway. 3T3 cells treated with heparitinase degraded less mTSP2 (Fig. 4). The decrease in degradation of TSP2 was similar to results of parallel experiments done with hTSP1 (Fig. 4). Chondroitinase ABC had no effect on degradation of either TSP. Addition of chondroitinase ABC or heparinase I to heparitinase did not cause a further decrease below that seen with heparitinase alone (data not shown). The decrease noted in the heparitinase-treated culture was not complete, but the value is similar to the decrease in binding of TSP1 to endothelial cells achieved after treatment with heparitinase (65).

Metabolism of TSP2 by CHO Cells Defective in Heparan Sulfate Proteoglycans-CHO cells defective in heparan sulfate
have been shown to be defective in adhesion to (28) or degradation of (66) platelet TSP1. The following cells were used to probe further the role of heparan sulfate proteoglycans in the metabolism of TSP2: K1 wild type cells, mutant 745 cells defective in xylosyltransferase and deficient in several glycosaminoglycans, mutant 803 cells specifically lacking heparan sulfate chains, and mutant 677 cells lacking heparan sulfate and having increased levels of chondroitin sulfate. Wild type K1 cells degraded TSP2 (Table II). Mutant 745 cells and 803 cells degraded much less TSP2 compared to the wild type cells. The decrease in degradation of TSP2 by the mutant 677 cells was less profound (Table II), suggesting that the excess chondroitin sulfate chains could compensate for the deficiency of heparan were then added to cells without removal of the enzyme and the incubation was carried out for additional 4 h. Degradation was measured as described as for Fig. 1. To relate effects on TSP1 and TSP2, data with the two proteins are plotted against one another normalized to control (no enzyme). Error bars indicate mean Ϯ S.D. of these determinations.

TABLE I Cross-inhibition of TSP degradation by each other
Each 125 I-TSP, 1 g/ml, was incubated with NIH 3T3 cells containing 50 g/ml of the indicated unlabeled TSP for 4 h at 37°C. Degradation was measured as in Fig. 1 sulfate chains as previously shown for hTSP1 (66). Parallel studies of platelet hTSP1 and recombinant mTSP1 yielded similar results to those obtained with mTSP2 (Table II). These results, along with the heparitinase experiments, suggest that heparan sulfate proteoglycans are required for degradation of TSP2.

Role of Low Density Lipoprotein Receptor-related Protein in
Degradation of TSP2-LRP is an endocytic receptor for the internalization and subsequent degradation of apolipoprotein E, lipoprotein lipase-enriched ␤-very low density lipoprotein, very low density lipoprotein, plasminogen activators, and ␣ 2macroglobulin-proteinase complexes (79). LRP was also recently shown to be involved in the internalization and degradation of hTSP1 as assessed by the ability of RAP, a known antagonist of LRP ligand binding, to inhibit hTSP1 degradation by human lung fibroblasts and human smooth muscle cells (67,68). RAP blocked degradation of TSP2 by 3T3 cells with an EC 50 of 40 -60 nM, similar to the concentrations required to block degradation of TSP1s (Fig. 5). Polyclonal anti-LRP IgG, when used at 75 g/ml, inhibited TSP2 degradation by 3T3 cells, whereas a control IgG did not show any inhibition (data not shown). The cells did not show any obvious morphological change in the presence of RAP or the anti-LRP antibodies. These data indicate that LRP synergizes with heparan sulfate proteoglycans to mediate the degradation of TSP2 by 3T3 cells.
Immunofluorescence Studies-Immunofluorescence studies were carried out to look for microscopic evidence of the binding and degradative events found with iodinated TSPs.
Mouse 3T3 cells synthesized endogenous mTSP1 and mTSP2 as judged by indirect immunofluorescence of permeabilized cells with anti-TSP1 or anti-TSP2 antibodies (Fig. 6, row A). This staining was relatively weak compared to staining of cells exposed to exogenous TSPs. When 20 g/ml of mTSP2, mTSP1, or hTSP1 was incubated with intact 3T3 cells and cells were examined without permeabilization of membrane, TSP was detected bound to cell surfaces (not shown). When cells were permeabilized, a bright punctate staining pattern for the exogenously added TSP was evident that was not seen on nonpermeabilized cells incubated with exogenous TSP or permeabilized cells without incubation with exogenous TSP (Fig. 6, row  B). This result indicates that TSP2 and TSP1 are endocytosed by the cells. Heparin inhibited binding and internalization of TSP2 and TSP1 to near background (Fig. 6, row C). When binding was done in the presence of 1 M RAP, the process of FIG. 6. Indirect immunofluorescence of the turnover of TSPs bound to NIH 3T3 cells. Subconfluent NIH 3T3 cells were incubated with DME, 0.2% albumin alone (row A), or with 20 g/ml hTSP1 (h1), mTSP1 (m1), or mTSP2 (m2) in DME, 0.2% albumin for 1 h (row B). Some cells were incubated with TSP in the presence of either heparin, 250 g/ml (row C); RAP, 1 M (row D); or chloroquine, 0.1 mM (row E). At the end of incubation, cells were washed with DME, fixed, permeabilized with chilled acetone, and stained with antibodies against hTSP1 (for hTSP1 and mTSP1 sets) or mTSP2 (for mTSP2 set) as described under "Materials and Methods." No staining was seen if nonimmune serum was used. internalization of TSP from the cell surface bound pool to the intracellular pool was inhibited, as shown by a decrease in the punctate intracellular staining with a concomitant increase in cell surface staining (Fig. 6, row D). When chloroquine was included in the binding medium, the intracellular punctate staining was more intense and the labeled structures were larger, presumably because chloroquine inhibited the lysosomal degradation but not the internalization of TSPs (Fig. 6, row E). When cells were incubated with TSPs for 45 min and then chased for 3 h, the intracellular punctate staining disappeared, whereas if chloroquine was included in the chase medium, the punctate staining persisted (not shown).
Conclusion-Binding and degradation of TSP2 by 3T3 cells were inhibited by heparin and RAP. Heparitinase-treated 3T3 cells and CHO cells defective in synthesis of heparan sulfate proteoglycan were slower in degradation of TSP2. Binding and degradation of TSP1 and TSP2 by 3T3 cells showed different sensitivities to inhibition by heparin, consistent with sequence differences in the heparin-binding sequences of the PARP module of the two proteins (4,6), the difference in salt concentration needed to displace TSP1 and TSP2 from immobilized heparin (58,69), and the differential effects of heparin on cell adhesion to TSP1 and TSP2 (58). The effect of soluble heparin is probably only of mechanistic importance. Heparin potentially can exert its inhibitory effect at several steps in the TSP binding and degradation pathway. It could block the initial binding of TSPs to cell surface heparan sulfate proteoglycans, and thus inhibit the subsequent steps of uptake and degradation. Consistent with this hypothesis, heparin caused less 125 I-TSP to be associated with cell layers and less cell surface TSP binding in the immunofluorescence studies. The fact that the sensitivities of degradation of the two TSPs correlate with the other measures of the TSP-heparin interaction indicates that this is the major mechanism of inhibition. Heparin could also inhibit the association of cell surface bound TSPs with LRP. It was shown previously that degradation of TSP1 was inhibited by heparin oligosaccharides of 8 to 10 units that had no inhibitory effect on binding (30). One possibility is that the short heparin oligosaccharides act like RAP (67,68) to inhibit association of TSP or heparan sulfate proteoglycan with LRP.
In addition to demonstrating that TSP1 and TSP2 have a similar degradation pathway, we found that TSP1 and TSP2 could compete with each other for degradation. This degradation is inhibited by RAP, which probably is an important physiological modulator. It was shown previously that TSP2, when existing alone, has potentially important adhesive functions (58). When coexisting with TSP1, TSP2 regulates at least one TSP1 function, viz. activation of latent TGF-␤ (55). Competing for the same degradation pathway provides a mechanism for the fine control of the levels of the two TSPs in wounds and in tissues in which expression of TSP1 and TSP2 may overlap. Because the expression of TSP2 by fibroblasts is constitutive compared to the dramatic regulation of TSP1 expression by serum or growth factors (5,6), binding and degradation of TSP2 by cells may be the primary mechanism for control of the extracellular concentration of TSP2. One can envision a scenario whereby newly expressed or secreted TSP1, in the process of being degraded, blocks the degradation of constitutively expressed TSP2, thus buffering cells from the effect of TSP1.