Thrombospondin-1 (TSP-1) is a potent inhibitor of tumor growth and angiogenesis. The antiangiogenic activity of TSP-1 has been mapped to the procollagen homology region and the type 1 repeats (TSR) using synthetic peptides. To elucidate the molecular mechanisms that are involved in the inhibition of tumor growth by the TSRs, we have expressed recombinant versions of these motifs and have assayed their ability to inhibit the growth of experimental B16F10 melanomas and Lewis lung carcinomas. Recombinant proteins that contain all three TSRs (3TSR) or the second TSR with (TSR2+RFK) or without (TSR2) the transforming growth factor-β (TGFβ) activating sequence (RFK) have been expressed in Drosophila S2 cells. In addition, recombinant proteins with mutations in either the RFK sequence (TSR2+QFK) or the WSHWSPW sequence [TSR2 (W/T)] of the second TSR have been prepared. Similar to platelet TSP-1, these proteins are potent inhibitors of endothelial cell migration, and 3TSR of human TSP-1 (3TSR/hTSP-1) and TSR2+RFK activate TGFβ. An 81% inhibition of B16F10 tumor growth is observed at 2.5 mg (135 nmol)/kg/day of the recombinant 3TSR/hTSP-1. A comparable level of inhibition is observed with 2.5 mg (360 nmol)/kg/day of TSR2+RFK. By contrast, 3TSR of mouse TSP-2 (3TSR/mTSP-2), TSR2+QFK, and TSR2 are significantly less effective. TSR2+RFK and TSR2 reduce tumor vessel density, but TSR2+RFK has a greater effect on B16F10 tumor cell apoptosis and proliferation. Concurrent treatment of B16F10 tumor-bearing mice with TSR2+RFK and either a soluble form of the TGFβ receptor or an antibody to active TGFβ reduces the inhibition of B16F10 tumor growth to levels that are comparable with those of TSR2 and TSR2+QFK. By contrast, the presence of the TGFβ-activating sequence does not increase the level of inhibition of Lewis lung carcinoma experimental tumor growth. These data indicate that the TSRs inhibit tumor growth by inhibition of angiogenesis and regulation of tumor cell growth and apoptosis. The regulation of tumor cell growth and apoptosis is TGFβ dependent, whereas the inhibition of angiogenesis is not.
Extracellular matrix proteins provide environmental cues that modulate cellular phenotype during development, tissue remodeling, and tumor growth. Neoplasia arise from mutations in oncogenes and tumor suppressor genes and in genes that are involved in the cell cycle and apoptosis (1). Tumor progression is also affected by landscaper genes that make the tumor microenvironment more or less permissive for growth (2). In vitro and in vivo data indicate that the extracellular matrix protein TSP4-1 functions as a landscaper gene in that it inhibits tumor growth. The lack of TSP-1 gene expression in p53-deficient mice results in decreased survival and changes to the spectrum of tumors observed.5 In addition, decreased TSP-1 expression correlates with the loss of p53 expression in human bladder, skin and colon cancer, and in fibroblasts from patients with Li-Fraumeni syndrome (3, 4, 5, 6, 7). In general, decreased TSP-1 expression is observed in transformed cells. Transfection of TSP-1 expression vectors into these cells inhibits the growth of the tumors that form when these cells are implanted into mice (8, 9, 10, 11, 12). Moreover, systemic injection of the intact TSP-1 protein into mice inhibits the growth of B16F10 melanoma experimental lung metastases (13). Taken together, these data indicate that TSP-1 gene expression down-regulates tumor growth. The ability of TSP-1 to inhibit angiogenesis, activate TGFβ, induce apoptosis of endothelial cells, and inhibit tumor cell growth may contribute to this inhibitory effect.
TSP-1 is a potent inhibitor of angiogenesis (reviewed in Ref. 14). It inhibits endothelial cell growth, migration, and tube formation in vitro (15). Intact TSP-1, bacterial fusion proteins containing TSP-1 sequences or synthetic peptides that contain sequences from the type 1 repeats (TSRs) of TSP-1 also induce apoptosis of endothelial cells (16). A peptide from the procollagen homology region and several peptides from the TSRs reportedly contain antiangiogenic activity (17, 18, 19). Recently, it has been reported that the activity of one of these peptides is dependent upon an l- to d-amino acid racemization that occurs during synthesis (20). The peptide that is synthesized with all l-amino acids is inactive. These data indicate that synthetic peptides may not accurately mimic the activities of the native protein.
The antiangiogenic activity of TSP-1 is reportedly mediated by CD36 on the endothelial cell membrane (21, 22). Whereas the inhibitory peptides from the TSRs are close to the VTCG sequence that is involved in CD36 binding, peptides that do not include this sequence are active as inhibitors of angiogenesis (18, 19, 20). These results suggest that other sequences within TSP-1 may interact with CD36 or that other membrane proteins are involved.
The boundary between the first and second TSRs of TSP-1 contains a sequence that binds and activates TGFβ (23, 24). The sequence RFK has been shown to be necessary and sufficient for activation of TGFβ. Mice that are deficient in TSP-1 display an abnormal phenotype in the lungs that is consistent with decreased levels of TGFβ activation (25, 26). TGFβ has been shown to act to suppress tumor growth. Local injection of TGFβ around experimental A549 human lung carcinomas inhibits tumor growth (27). In mice that lack TGFβ and Rag2, spontaneous adenomas and carcinomas occur in the cecum and colon (28). Mice that are heterozygous for a TGFβ-null allele exhibit enhanced tumor formation in response to chemical carcinogens (29). In addition, overexpression of a dominant-negative form of the TGFβ type II receptor accelerates skin carcinoma (30). By contrast, TGFβ has been reported to stimulate the growth of colon, prostate, and melanoma cells in vitro (31, 32, 33). In addition, TGFβ reportedly stimulates angiogenesis in renal cell carcinoma (34). Thus, the activation of TGFβ by TSP-1 may have variable effects on tumor growth. Systemic injection of peptides that contain or lack the RFK sequence inhibits tumor growth in a mouse xenograft model of breast cancer, showing that activation of TGFβ is not essential for the antitumor activity of TSP-1 (19).
In this study, we have prepared recombinant forms of the TSRs in a eukaryotic expression system. These proteins resemble the native protein in that they bind antibodies to the human TSP-1 TSRs, activate TGFβ, and inhibit endothelial cell migration. We have found that these proteins are potent inhibitors of tumor growth in vivo. This inhibition is attributable to a combination of effects including inhibition of angiogenesis, induction of tumor cell apoptosis, and inhibition of tumor cell proliferation. A portion of the inhibitory effect is mediated by the activation of TGFβ in tumor cells that are responsive to TGFβ.
MATERIALS AND METHODS
Human TSP-1 was purified from the supernatant of thrombin-treated platelets as described previously (35). To remove TGFβ, the sucrose density gradient centrifugation step was performed at pH 11 (35, 36). Recombinant proteins that included sequences from the type 1 repeats were prepared by PCR using the full-length cDNA for human TSP-1 or murine TSP-2 as a template. A recombinant protein containing all three TSRs of TSP-1 (3TSR/hTSP-1, amino acids 361–530) was prepared using the forward primer 475htsp1f (GAT GAT CCC GGG GAC GAC TCT GCG GAC GAT GGC) and the reverse primer 476htsp1r (GAT ACC GGT AAT TGG ACA GTC CTG CTT G). A recombinant protein containing all three TSRs of mouse TSP-2 (3TSR/mTSP-2, amino acids 381–550) was prepared using the forward primer 544mtsp2f (GAT GAT CCC GGG GAT GAG GGC TGG TCT CCG) and the reverse primer 545mtsp2r (GAT ACC GGT AAT AGG GCA GCT TCT CTT). A recombinant protein that contains the second TSR (TSR2, amino acids 416–473) was prepared using the forward primer 537htsp1f (GAT GAT CCC GGG CAG GAT GGT GGC TGG AGC) and the reverse primer 515htsp1r (GAT ACC GGT GAT GGG GCA GGC GTC TTT CTT). To evaluate the role of TGFβ activation on the effect of this recombinant protein, a longer version of the second TSR (TSR2+RFK, amino acids 411–473) that includes the RFK sequence was synthesized using the forward primer 514htsp1f (GAT GAT CCC GGG GAC AAG AGA TTT AAA CAG) and the reverse primer 515htsp1r. In addition, a mutant protein that contained the sequence QFK was constructed using the forward primer 605htsp1f (GAT GAT CCC GGG GAC AAG CAA TTT AAA CAG GAT GG) and the reverse primer 515htsp1r. Synthetic peptides that contain glutamine (Q) instead of arginine (R) do not activated TGFβ (26). A mutant protein in which the three conserved tryptophan residues are mutated to threonine [TSR2 (W/T)] was engineered using the forward primer 592htsp1f (GAT GAT CCC GGG CAG GAT GGT GGC ACG AGC CAC ACG TCC CCG ACG TCA TCT TGT TCT) and the reverse primer 515htsp1r. All PCR products were cloned between the XmaI and the AgeI sites of the vector pMT/BiP/V5-HisA (Invitrogen, Carlsbad, CA). The recombinant proteins included the vector-derived amino acid sequence RSPWG at the NH2 terminal and TGHHHHHH at the COOH terminal. The fidelity of the PCR products was verified by nucleotide sequencing. Each expression vector was cotransfected into Drosophila S2 cells with the selection vector pCoHYGRO according to the manufacturer’s protocols (Invitrogen). Transfected cells were selected with hygromycin B, and the expression of recombinant peptides was monitored by Western blotting using the polyclonal antibody R3 that was raised against a fusion protein that contained all three TSRs of TSP-1 (37). For large-scale preparation of recombinant protein, S2 cells were grown in serum-free medium for 5 days. The culture supernatant was centrifuged to remove the cells and dialyzed against 20 mm NaPO4 (pH 7.8) and 500 mm NaCl. The dialysate was applied to a column of ProBond resin (Invitrogen). The column was eluted with 20 mm NaPO4 (pH 6.0), 500 mm NaCl, and 500 mm imidazole. The protein eluted with 500 mm imidazole was dialyzed against 20 mm NaPO4 (pH 7.0) and 500 mm NaCl, and 1% sucrose was added prior to storage.
HDMECs (kindly provided by Dr. Michael Detmar, Massachusetts General Hospital, Boston, MA) were isolated by the procedure of Richard et al. (38). The cells were cultured in Vitrogen precoated dishes and maintained in EBM (Clonetics Corp., San Diego, CA) containing 20% fetal bovine serum, 1 μg/ml hydrocortisone acetate, 5 × 10−5 m dibutyryl-cAMP, 200 units/ml penicillin, 100 units/ml streptomycin, 250 μg/ml amphotericin, and 2–5 ng/ml vascular endothelial growth factor. Murine B16F10 melanoma and murine Lewis lung carcinoma cells were obtained from the American Type Culture Collection and were maintained in DMEM supplemented with 10% fetal bovine serum, 50 μg/ml penicillin, 50 units/ml streptomycin, and 2 mm glutamine (supplemented DMEM). The effect of TGFβ and TSR2+RFK on B16F10 melanoma and Lewis lung carcinoma cell proliferation was determined with the CellTiter96 nonradioactive cell proliferation assay according to protocols supplied by the manufacturer (Promega Biotech, Madison, WI).
Assay for TGFβ Activation.
B16F10 cells (2.5 × 105) were plated in a T25 flask and grown overnight in supplemented DMEM. The cells were rinsed once with 1.0 ml of serum-free DMEM, and 2.5 ml of serum-free DMEM containing 1.0 μmol of the recombinant protein were added. After an overnight incubation, conditioned medium was collected and centrifuged at 12,000 rpm for 5 min to remove cellular debris. Undiluted medium was used to determine the level of active TGFβ. Three independent samples of culture medium from each peptide treatment group were assayed for the level of active TGFβ (see below).
To assay the level of active TGFβ in tumor tissue, cubes of tissue that were 4 mm on a side were homogenized in 1 ml of PBS containing 2 mm phenylmethylsulfonyl fluoride and 1 μg/ml each of aprotinin, leupeptin, and pepstatin and sonicated for 2 min at 0°C. Homogenates were centrifuged at 12,000 rpm for 5 min in a microcentrifuge to remove debris. Three independent tumor samples from each treatment group were evaluated for TGFβ level.
The TGFβ levels in culture supernatants and tumor extracts were assayed using an ELISA kit for active TGFβ according to the manufacturer’s protocols (R&D Systems, Minneapolis, MN). Briefly, 100 μl of sample or TGFβ standard were mixed with 100 μl of diluent and then added to the individual wells of the assay plate. After incubation for 3 h at 22°C, the plates were washed four times, 200 μl of TGFβ conjugate solution were added to each well, and the plates were incubated for 1.5 h at 22°C. The wells were washed four times, and 200 μl of substrate solution were added to each well. After 20 min at 22°C, 50 μl of stop solution were added to each well, and the absorbency at 450 nm was determined.
In Vitro Migration Assay.
HDMECs at passages 7–10 were serum starved and maintained in EBM with 0.1% BSA (control medium) for 20 h before trypsinization to harvest the cells. Cells were washed in EBM twice and resuspended in control medium at the concentration of 1 × 106 cells/ml. Two hundred μl of cells were packed down and kept frozen. They were lysed and serially diluted by CyQuant reagent (Molecular Probe, Eugene, OR) to be used for standard curve construction.
Transwell membrane (24-well polycarbonate membrane, 8-μm pore size; Corning Costar Corp., Cambridge, MA) coated with Vitrogen (30 μg/ml; Collagen Corp., Freemont, CA) on both sides was used for chemotactic migration experiments. Coated transwells were inverted, and 100 μl of cell suspension were applied to the top of the membrane and covered by the bottom plate carefully so that the cell suspension stayed on top of the membrane. The cells were allowed to adhere to the coated membrane for 2 h in an incubator at 37°C with 5% CO2. After the adhesion incubation, the plates were re-inverted, the bottom wells were filled with 0.4 ml of control medium, and the top wells were filled with 0.1 ml of control medium containing testing reagents. The plate was returned to the incubator for 3.5 h for the cells to migrate. At the end of the incubation, the transwells were washed in PBS, and the cells on the bottom side of the membrane (unmigrated cells) were wiped away with a cotton swab. The membranes were cut out by scalpel and placed into 96-well plates and frozen at −80°C overnight. Two hundred μl of CyQuant reagent were added to each well. Fluorescence reading was done 16 h later with a SpectraFluor plate reader with excitation at 485 nm and emission at 535 nm. The number of cells migrated was then calculated based on the standard curve.
Primary Tumor Growth Assay.
The proteins for injection were mixed with Polymyxin B-Agarose (Sigma Chemical Co.) for 30 min at room temperature to remove endotoxin. The endotoxin levels were <0.05 EU/μg as determined using the QCL-1000 assay kit (BioWhittaker, Walkersville, MD). Proteins were filter sterilized, and the protein concentration was determined prior to injection.
C57BL/6 mice (Taconic, Germantown, NY), 5–8 weeks of age, were acclimated and caged in groups of four or fewer, and their backs were shaved. Cultured B16F10 melanoma or Lewis lung carcinoma cells (1 × 106) were inoculated s.c. on the back of each mouse. Each treatment group contained six mice. Tumors were measured with a dial caliper, and the volumes were determined using the formula width2 × length × 0.52. The treatment began 4 days after inoculation of the tumor cells. The therapeutic groups received TSP-1 or recombinant TSP-1 proteins i.p. daily, and the negative control group received comparable injections of saline alone. The experiments were terminated, and mice were sacrificed when the control mice began to die, usually at day 16 after tumor cell injection. In one experiment, the mice were treated with a daily i.v. injection of TSP-1 into the tail vein. To evaluate the role of TGFβ activation, mice were treated with a soluble form of the extracellular domain of mouse TGFβ type II receptor fused to the Fc region of IgG2a (kindly provided by Drs. Phil Gotswal and Victor Koteliansky, Biogen Corp., Cambridge, MA; Ref. 39). One hundred μg of this reagent were injected on days 1 and 7 of TSR2+RFK [1 mg (144 nmol)/kg/day] or saline treatment. Alternatively, mice were injected with the anti-TGFβ antibody 1D11 that was kindly provided by Dr. Steve Ledbetter (Genzyme Corp., Framingham, MA). This reagent (100 μg/mouse) was injected every other day, beginning on the first day of saline or TSR2+RFK treatment. The antibody 13C4 (kindly provided by Dr. Steve Ledbetter) was used as a species- and isotype-matched control.
The tumor tissues were cut, fixed with neutral buffered formaldehyde, and embedded in paraffin according to standard histological procedures. H&E staining was used for tissue morphology examination. Blood vessels were immunochemically stained by anti-CD31 antibody with a Vectastain ABC kit (Vector Laboratories, Burlingame, CA). Tumor cell apoptosis was detected by terminal deoxynucleotidyl transferase-mediated nick end labeling assay (35). Tumor cell proliferation was detected by staining with a monoclonal anti-proliferating cell nuclear antigen (Santa Cruz Biotechnology, Santa Cruz, CA). The number of blood vessels was recorded by counting 10 high-power fields. Tumor cell proliferation and the apoptotic index were estimated by the percentage of cells scored under a light microscope. A minimum of 1000 cells was counted in each tumor sample.
Characterization of Recombinant Proteins.
Five recombinant forms of the type 1 repeats of human TSP-1 and one form of the type 1 repeats of mouse TSP-2 have been produced in S2 cells. The protein that contains all three type 1 repeats of human TSP-1, designated 3TSR/hTSP-1, contains amino acids 361–530. This protein has a predicted molecular weight of 20,520 and an apparent molecular weight of 25,000 on SDS-PAGE, suggesting that carbohydrates are added by posttranslational modification (data not shown). A comparable electrophoretic mobility was observed for all three type 1 repeats of mouse TSP-2 (3TSR/mTSP-2). Each TSR contains six cysteine residues. Consistent with the presence of intrachain disulfide bonds in these proteins, 3TSR/hTSP-1 and 3TSR/mTSP-2 migrate more rapidly in the absence of reducing agent during SDS-PAGE (data not shown; Ref. 40). The constructs, designated TSR2+RFK and TSR2+QFK, contain the second type 1 repeat of human TSP-1 (amino acids 411–473) and include the sequence DKRFK or DKQFK, respectively, in the NH2-terminal region. The sequences of human and mouse TSP-1 are identical in this region. The RFK sequence has been shown to mediate the activation of TGFβ by TSP-1, and synthetic peptides that contain the sequence QFK are inactive (24). An equivalent region of TSP-1 (amino acids 416–473) that excludes the DKRFK sequence, designated TSR2, and a recombinant protein in which the three conserved tryptophan residues in TSR2 were mutated to threonine residues [TSR2 (W/T)] have also been prepared. The average level of protein expression is comparable for all six proteins (∼24 μg of purified protein from 1 ml of conditioned medium). All six proteins react with a polyclonal antibody, designated R3, that was raised against a bacterial fusion protein composed of the type 1 repeats of human TSP-1 fused to β-galactosidase (data not shown; Ref. 37). Thus, the data indicate that the recombinant proteins produced in S2 cells have the expected molecular properties based on their content of the type 1 repeats.
Functionally, the recombinant TSR-containing proteins are similar to the native TSRs in that they activate TGFβ and inhibit endothelial cell migration. Medium conditioned by B16F10 melanoma cells contain 30.3 ± 2.3 pg/ml of active TGFβ (Fig. 1). Addition of 1 μm TSR2+RFK to the conditioned medium increases the level of active TGFβ to 180.6 ± 23.3 pg/ml. A comparable level of TGFβ activation is observed when the conditioned medium is treated with 3TSR/hTSP-1 (Fig. 1). By contrast, addition of recombinant proteins that do not contain the RFK sequence does not result in an increase in the level of active TGFβ in the conditioned medium (Fig. 1).
As shown in Fig. 2, TSP-1 is a potent inhibitor of endothelial cell migration in vitro. This inhibition is dose dependent up to ∼0.5 nm TSP-1, but concentrations of TSP-1 above this level are less effective. This biphasic response has been reported by others (17). The 3TSR/hTSP-1, TSR2, and TSR2+RFK recombinant proteins also inhibit endothelial cell migration with responses that are similar to TSP-1 (Fig. 2).
Inhibition of Experimental Tumor Growth by Intact TSP-1.
In this study, the B16F10 melanoma model of experimental tumor growth has been used to assay the effect of the intact TSP-1 and recombinant proteins. This model has been used extensively to assay the activity of the antiangiogenic proteins endostatin, angiostatin, and anti-thrombin III (41, 42, 43). Systemic injection of TSP-1 [0.25 mg (0.6 nmol)/kg/day] into tumor-bearing mice inhibits tumor growth by 56% on the twelfth treatment day (Fig. 3,A). At higher doses, less inhibition is observed, and a dose of 2.5 mg (6 nmol)/kg/day has no effect on tumor growth. Human platelet TSP-1 reportedly contains low levels of bound TGFβ that can be removed by treatment with pH 11 buffers (36). Removal of TGFβ from the TSP-1 preparations did not affect the dose response for inhibition of experimental B16F10 tumor growth (Fig. 3,A). To determine whether the higher concentrations of TSP-1 are less effective in reaching the circulation, we treated tumor-bearing mice with TSP-1 that lacked TGFβ by tail vein injection. Maximum inhibition of tumor growth is observed at the highest concentrations of TSP-1 [2.5 mg (6 nmol)/kg/day] when the protein is delivered by i.v. injection (Fig. 3 B). These data suggest that the biphasic effect of TSP-1 on tumor growth that is observed with i.p. injection does not reflect direct effects on tumor growth but rather reflects the ability of the protein to reach the bloodstream.
Inhibition of B16F10 Experimental Tumor Growth by the Type 1 Repeats.
The recombinant proteins that contain the type 1 repeats were used to systemically treat B16F10 tumor-bearing mice. The 3TSR/hTSP-1 protein is a less effective inhibitor of tumor growth at 0.25 mg (13.5 nmol)/kg/day than the intact protein (Fig. 4,A). By contrast, at dosages of 1.0 mg (54 nmol)/kg/day and greater, the 3TSR/hTSP-1 protein is more effective than TSP-1 when the proteins are injected i.p. Tumor volume is reduced by 81% by the 3TSR/hTSP-1 protein with a dose of 2.5 mg (135 nmol)/kg/day. Increasing the dose to 10 mg (540 nmol)/kg/day did not result in increased inhibition of tumor growth (Fig. 4 A).
To explore the potential involvement of the RFK sequence in the inhibition of B16F10 tumor growth, TSR2+RFK and TSR2 have been assayed. The tumor inhibition effect of the TSR2+RFK is comparable with that of 3TSR/hTSP-1 on a weight basis at the various doses used (Fig. 4, A and B). By contrast, TSR2 is significantly less effective (Fig. 4,C). At 0.25 mg (40 nmol)/kg/day, the tumor growth in the mice treated with TSR2 is indistinguishable from the control group. At 1.0 mg/kg/day, TSR2+RFK inhibits tumor growth by ∼80%, whereas TSR2 only inhibits growth by ∼38% on the twelfth day of treatment. Because TSR2+RFK has a higher molecular weight than TSR2, 1 mg of protein corresponds to 144 nmol of TSR2+RFK and 160 nmol of TSR2. At the higher doses, the difference is less pronounced; however, TSR2 remains less active (Fig. 4, B and C). Increasing the dose of TSR2 to 5 mg (800 nmol)/kg/day does not produce a level of inhibition that is achieved with TSR2+RFK at 2.5 mg (360 nmol)/kg/day (Fig. 4,B and 5). TSR2 (W/T) did not inhibit tumor growth, suggesting that the three conserved tryptophan residues are important for TSR2 activity or stability (Fig. 5).
We prepared a recombinant protein designated TSR2+QFK, in which the arginine residue in the RFK sequence is mutated to glutamine, to more specifically demonstrate the importance of the RFK sequence. Systemic injection of TSR2+QFK produces levels of B16F10 tumor growth inhibition that are comparable with TSR2 (Fig. 5). In addition, a recombinant protein that contains all three type 1 repeats of murine TSP-2 (3TSR/mTSP-2) is less effective as an inhibitor of tumor growth than 3TSR/hTSP-1 (Fig. 5). The amino acids RIR are found in the murine TSP-2 sequence in the location where the RFK sequence is located in TSP-1.
To further characterize the effect of the recombinant protein treatment on tumor growth, we have determined the rate of proliferation, the apoptotic index, and the capillary density in tumors from mice treated with TSR2, TSR2+RFK, or TSR2+QFK at a dose (1.0 mg/kg/day), where the largest effect of inclusion of the RFK sequence is observed. Tumors displayed a 71 and 63% reduction in capillary density when the mice are treated with TSR2+RFK and TSR2+QFK proteins, respectively (Figs. 6 and 7,A). B16F10 tumors treated with TSR2 exhibited a 64% decrease in capillary density. The tumors from mice that are treated with TSR2+RFK protein display a 4-fold increase in tumor cell apoptosis (P ≤ 0.005), and the tumors from the mice that are treated with TSR2+QFK or TSR2 displayed only a 1.9-fold increase in apoptosis (P ≤ 0.025; Figs. 6 and 7,B). TSR2+RFK also reduces the percentage of proliferating cell nuclear antigen-positive tumor cells. The number of proliferating cells is decreased by 35% by treatment with TSR2+RFK (P ≤ 0.005), whereas treatment with TSR2 reduces tumor cell proliferation by 7.8% (P ≤ 0.05; Fig. 7 C).
The Role of TGFβ Activation in the Inhibition of B16F10 Tumor Growth by TSR2+RFK.
To establish that the additional inhibitory effect of TSR2+RFK as compared with TSR2 or TSR2+QFK is attributable to activation of TGFβ, we have: (a) added antagonists of active TGFβ concomitant to TSR2+RFK treatment; and (b) assayed the level of active TGFβ in tumor tissue. Two antagonists of active TGFβ have been used. Systemic injection of a chimeric protein that includes the Fc portion of human IgG2a fused to the extracellular domain of the type II TGFβ receptor along with TSR2+RFK significantly (P ≤ 0.005) reduced the ability of TSR2+RFK to inhibit tumor growth (Fig. 8). The B16F10 tumor size is increased by ∼15% in the saline-treated control group (Fig. 8). Thus, in the presence of the soluble receptor, TSR2+RFK only inhibited tumor growth by 55% as compared with a value of 81% in the absence of soluble receptor. The apoptotic and proliferative indices in B16F10 tumors from mice that are treated with TSR2+RFK and the soluble TGFβ receptor are comparable with those of mice treated with saline or TSR2+QFK (Fig. 7, B and C).
The level of active TGFβ in extracts of saline-treated B16F10 melanomas is 65.4 ± 6.5 pg/ml. A comparable level of active TGFβ is detected in tumor tissue from mice treated with TSR2 or TSR2+QFK (Fig. 9). By contrast, a 3.2-fold increase in active TGFβ is observed in B16F10 melanoma tissue from mice treated with TSR2+RFK. The amount of active TGFβ is reduced to control levels when the mice are treated concurrently with TSR2+RFK and the soluble form of the TGFβ receptor (Fig. 9).
We have used systemic administration of an antibody to active TGFβ (1D11) as an alternate strategy for inhibiting the activation of TGFβ by TSR2+RFK in vivo. In the control group that received daily saline injections, treatment with that anti-TGFβ antibody increased tumor size by 17% as compared with the treatment group that received no antibody or a species- and isotype-matched control antibody (13C4; Fig. 10). Concomitant treatment with TSR2+RFK and 13C4 reduced B16F10 tumor volume by 83% as compared with the tumors that grow in mice treated with saline and 13C4. By contrast, concomitant treatment with TSR2+RFK and 1D11 only reduced tumor volume by 53% as compared with the tumors that grow in mice treated with saline and 1D11 (Fig. 10). Histological analysis of the tumors that grow in mice treated with TSR2+RFK and 1D11 reveals that the apoptotic index is not increased, and the proliferative index is not decreased (Fig. 7, B and C). Taken together, these data indicate that TSR2+RFK inhibits B16F10 melanoma growth through a combination of TGFβ-dependent and -independent mechanisms.
Effect of Type 1 Repeat Recombinant Proteins on Lewis Lung Carcinoma Growth.
The inhibitory effect of 3TSR/hTSP-1 is also observed in mice bearing experimental tumors that form by s.c. injection of Lewis lung carcinoma cells. At 2.5 mg (135 nmol)/kg/day, 3TSR/hTSP-1 inhibits tumor growth by 73% on treatment day 12 (data not shown). TSR2+RFK is equally as effective as 3TSR/hTSP-1 on a weight basis. By contrast to the B16F10 melanoma tumors, the level of inhibition of tumor growth is equivalent for TSR2+RFK, TSR2, and TSR2+QFK (Fig. 11). TSR2 (W/T) is without activity in the Lewis lung carcinoma model, as it is in the B16F10 model (Fig. 11). The capillary density in experimental Lewis lung carcinomas is reduced by 62–64% in mice treated with TSR2, TSR2+RFK, or TSR2+QFK (Fig. 7,D). All three proteins produce a comparable increase (1.8–1.9-fold) in apoptotic index and had no effect on proliferative index (Fig. 7, E and F). These data indicate that inclusion of the RFK sequence does not increase the ability of the recombinant TSR-containing proteins to inhibit the growth of tumors that form from Lewis lung carcinoma cells. To determine whether the lack of effect of the RFK sequence is attributable to an inability of the Lewis lung carcinoma cells to respond to TGFβ, we have examined the effect of TSR2+RFK and TGFβ on cell growth in culture. As shown in Fig. 12, TGFβ or TSR2+RFK treatment results in a dose-dependent suppression of B16F10 melanoma cell growth in vitro. In addition, 3TSR/hTSP-1 inhibits B16F10 melanoma cell growth, whereas TSR2, TSR2+QFK, and 3TSR/mTSR-2 do not (data not shown). The suppression of B16F10 melanoma cell growth by TSR2+RFK or 3TSR/hTSP-1 is blocked by the soluble TGFβ receptor and by the antibody to active TGFβ (data not shown). By comparison to the B16F10 melanoma cells, the growth of Lewis lung carcinoma cells is unaffected by the presence of TGFβ or TSR2+RFK (Fig. 12).
The data presented here show that systemic injection of recombinant proteins that contain the TSRs of TSP-1 inhibit the growth of experimental tumors. Inhibition has been observed with both B16F10 melanoma and Lewis lung carcinoma cell lines. The 3TSR/hTSP-1 and TSR2+RFK proteins are relatively potent, showing 70–80% inhibition of tumor volume at 2.5 mg/kg/day. This concentration is 100–300 nmol/kg/day. Injection with intact TSP-1 protein at 0.25 mg/kg/day (0.6 nmol/kg/day) results in a 56% reduction of tumor growth, indicating that the intact protein is considerably more active than the recombinant proteins. The increased potency of the intact protein may be attributable to the trimeric structure of TSP-1. The nine TSRs of TSP-1 may cluster receptors and enhance signal transduction. Dimerization of CD36 in response to TSP-1 has been observed (44). This receptor reportedly mediates the antiangiogenic activity of TSP-1 (21, 22). In addition, intact TSP-1 may include other domains that inhibit tumor growth. Sequences within the procollagen homology region and the COOH-terminal domain of TSP-1 reportedly inhibit angiogenesis (17, 45).
When TSP-1 is administered by i.p. injections, higher doses of TSP-1 are less effective in reducing tumor volume. One other study has examined the effect of injection of TSP-1 into tumor-bearing mice (13). In this study, mice with B16F10 melanoma experimental lung metastases were treated with 5 or 10 mg/kg/day of TSP-1 via i.p. injection. Both doses produced an ∼90% inhibition in the number of visible surface metastases. To reconcile these contradictory effects of higher doses of TSP-1, we have used i.v. injection of TSP-1 in mice with s.c. B16F10 melanomas and found that higher doses are indeed active when the protein is injected directly into the vasculature. Thus, the loss of activity at higher doses when the protein is delivered i.p. appears to be attributable to a decreased ability of the TSP-1 to exit the peritoneal cavity.
Several activities of the TSRs may contribute to the inhibition of tumor growth. These include the ability to: (a) inhibit angiogenesis; (b) reduce tumor cell proliferation; (c) activate TGFβ; and (d) regulate extracellular proteases. The results described here are the first to show that systemic treatment with the TSRs of TSP-1 reduces vessel density in tumors. They are consistent with the observation that overexpression of intact TSP-1 in tumor cells decreases the capillary density in tumors that form from these cells (9, 10). Furthermore, B16F10 experimental tumors that are grown in TSP-1-deficient mice display an increase in capillary density.5 Taken together, these data indicate that TSP-1 in the tumor or stromal compartments and exogenously added TSRs are potent inhibitors of tumor angiogenesis. The TSR2+RFK and TSR2 proteins are both effective inhibitors of tumor angiogenesis. This result is consistent with the observation that both proteins are potent inhibitors of endothelial cell migration. These data are also consistent with synthetic peptide studies that have been performed using the chick chorioallantoic membrane assay (19). In both studies, the inhibition of angiogenesis was observed in the presence or absence of the RFK sequence. Thus, the inhibition of angiogenesis by TSP-1 is not dependent upon the activation of TGFβ.
The rate of tumor cell apoptosis is a key factor in the determination of the rate of tumor growth (46, 47). We have found that TSP-1 proteins that include the RFK sequence significantly increase the rate of B16F10 tumor cell apoptosis. Whereas TSP-1 has been reported to inhibit tumor cell proliferation and induce endothelial cell apoptosis, this study represents the first demonstration that the TSR-containing proteins induce tumor cell apoptosis (16, 48). This activity is enhanced by the addition of the DKRFK sequence at the NH2-terminal because TSR2 did not increase the level of apoptosis of the tumor cells to the same extent as TSR2+RFK. An increase in tumor cell apoptosis is frequently associated with antiangiogenic therapy because of the decrease in nutrients that is associated with the decrease in blood supply. Our data indicate that increased tumor cell apoptosis occurs as a direct result of TSR2+RFK treatment rather than indirectly through inhibition of angiogenesis. Whereas TSR2+RFK, TSR2+QFK, and TSR2 reduce vessel density to comparable levels, TSR2+RFK has a more profound effect on B16F10 tumor cell apoptosis. In addition, this protein significantly decreases tumor cell proliferation in vivo. By contrast, TSR2, TSR2+RFK, and TSR2+QFK have comparable effects on apoptosis within tumors that form from Lewis lung carcinoma cells. The increase in apoptosis (1.8–1.9-fold) that is observed in Lewis lung carcinoma tumors with all three proteins is equivalent to that induced by TSR2 or TSR2+QFK (1.9-fold) in B16F10 melanoma. This level of apoptosis is considerably lower than that induced by TSR2+RFK (4-fold) in B16F10 melanoma tumors.
Our data indicate that the ability of RFK-containing type 1 repeat proteins to inhibit B16F10 tumor growth appears to be attributable, in part, to activation of TGFβ. TGFβ has pleiotropic effects on tumor growth. At early stages of tumorigenesis, TGFβ may act as a tumor suppressor gene (28, 29). TGFβ can induce apoptosis of several different tumor cell lines (Ref. 49 and references therein). At later stages of tumor growth, TGFβ can stimulate angiogenesis through the recruitment of inflammatory and stromal cells (50). We have found that deletion or point mutations that affect the RFK sequence result in less active TGFβ in vitro and in vivo. These proteins are less effective inhibitors of B16F10 tumor growth. In addition, antagonists of active TGFβ reduce the antitumor activity of RFK-containing proteins in the B16F10 model. These data are consistent with the observation that the injection of TGFβ into the tissue surrounding experimental A549 lung adenocarcinomas inhibits tumor growth (27). By contrast, the presence of the RFK sequence does not affect the ability of recombinant proteins to inhibit experimental Lewis lung carcinomas. Furthermore, inclusion of the RFK sequence in a synthetic peptide did not increase the antitumor activity toward mammary tumors formed by injection of MDA MB435 cells (19). In vitro, TGFβ or TSR2+RFK inhibit the growth of B16F10 cells but not Lewis lung carcinoma cells, suggesting that the latter have lost the ability to respond to TGFβ. Similar to Lewis lung carcinoma cells, MDA MB435 cell proliferation is not inhibited by TGFβ (18). Many types of tumors have mutations in the TGFβ type II receptors or Smads and become resistant to TGFβ. Thus, although TSR2+RFK or 3TSR/hTSP-1 are potent inhibitors of tumor growth in general, an enhanced inhibition is obtained in those tumor cells in which TGFβ signaling results in growth suppression.
The data presented here indicate that recombinant proteins that include the TSRs have therapeutic potential as inhibitors of tumor growth. They act to inhibit angiogenesis and to induce tumor cell apoptosis. The former activity is TGFβ independent, and the latter is TGFβ dependent. Thus, a knowledge of the TGFβ signaling capacity of the tumor cells is important for the design of therapeutic regimes that use these proteins. Because these mechanisms of action of the TSRs are probably different from other inhibitors of neoplasia, a combinational approach that includes TSRs with other inhibitors of tumor growth may provide an effective treatment for cancer.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Grant HL28749 and HL07893 from the National Heart, Lung, and Blood Institute of the NIH.
The abbreviations used are: TSP, thrombospondin; TSR, TSP type 1 repeat; TGF, transforming growth factor; HDMEC, human dermal microvessel endothelial cell.
J. Lawler, W-M. Miao, M. Duquette, N. Bouck, R. T. Bronson, and R. O. Hynes, Thrombospondin-1 gene expression affects survival and tumor spectrum of p53-deficient mice. Am. J. Pathol., in press, 2001.
We thank Drs. Sareh Parangi, Zhensheng Wang, Corinne Reimer, and Karen Yee for insightful discussions and Jenny Teece and Mingyan Hu for technical support. Dr. Steve Ledbetter kindly provided the anti-TGFβ antibody 1D11 and the control antibody 13C4. The soluble form of the TGFβ type II receptor was kindly provided by Drs. Phil Gotwals and Victor Koteliansky. The HDMECs were kindly provided by Dr. Michael Detmar. The Dana-Farber/Harvard Cancer Center, Rodent Histopathology Core provided assistance with the histology. The manuscript was prepared by Regina Prout and Alexis Bywater.