P-selectin deficiency attenuates tumor growth and metastasis

Abstract

Selectins are adhesion receptors that normally recognize certain vascular mucin-type glycoproteins bearing the carbohydrate structure sialyl-Lewisx. The clinical prognosis and metastatic progression of many epithelial carcinomas has been correlated independently with production of tumor mucins and with enhanced expression of sialyl-Lewisx. Metastasis is thought to involve the formation of tumor-platelet-leukocyte emboli and their interactions with the endothelium of distant organs. We provide a link between these observations by showing that P-selectin, which normally binds leukocyte ligands, can promote tumor growth and facilitate the metastatic seeding of a mucin-producing carcinoma. P-selectin-deficient mice showed significantly slower growth of subcutaneously implanted human colon carcinoma cells and generated fewer lung metastases from intravenously injected cells. Three potential pathophysiological mechanisms are demonstrated: first, intravenously injected tumor cells home to the lungs of P-selectin deficient mice at a lower rate; second, P-selectin-deficient mouse platelets fail to adhere to tumor cell-surface mucins; and third, tumor cells lodged in lung vasculature after intravenous injection often are decorated with platelet clumps, and these are markedly diminished in P-selectin-deficient animals.

Figure 1

Genotyping and phenotyping of P-selectin and Rag2 double null mutants. (A) Genomic DNA from mouse tail clips were used as templates for PCR genotype screening for Rag2 allele and P-selectin allele. The blackened region represents the neomycin A resistance gene that was targeted into both wild-type alleles; the arrows represent the two forward primers and the single reverse primer used in each reaction. Each PCR contained the three primers. F1 represents the double heterozygote from the first cross between Rag2 −/− and P-selectin −/− mice. (B) Peripheral blood from F1 mice and double null mutants was labeled with anti-CD4-phycoerythrin or anti-CD8-fluorescein isothiocyanate and analyzed by using a Becton Dickinson FACScan flow cytometer. (C) Frozen lung sections were probed with biotinylated anti-mouse P-selectin antibody and followed with streptavidin-peroxidase. The enzyme conjugate was detected with metal-enhanced diaminobenzidine substrate. The arrows indicate lung endothelium.

Figure 2

Effects of P-selectin deficiency on primary tumor growth rate and survival. (A) The length and width of tumors arising from s.c. injected LS180 cells were measured and their volumes were calculated. Average tumor size for each mouse group (n = 9 per group) is plotted against time. (B) The growth rate at day 30 for each mouse was measured when all were still alive and was analyzed for statistical significance by using the Wilcoxon rank–sum test. The null hypothesis of equal ranks was rejected (P < 0.05). (C) Survival curve of mice with s.c. tumors (n = 9 per group). The trend toward improved survival did not achieve full statistical significance at 60 days, when institutional animal welfare guidelines required termination of the experiment.

Figure 3

P-selectin facilitates the development of micrometastasis in the lung. (A) LS180 cells were injected intravenously as described, and the mice were sacrificed after 4 weeks. Examples of hematoxylin/eosin-stained lung sections from each mouse type are shown. Note the micrometastasis evident in the P-selectin +/+ lung. (B) Several cross-sections of the lung from each mouse were analyzed for histological evidence of metastatic cells. (C) Micrometastases were quantitated by PCR analysis with primers for human Alu sequences as described in Materials and Methods and analyzed for statistical significance by using the Wilcoxon rank–sum test.

Figure 4

P-selectin affects the seeding of intravenously injected tumor cells. Mice were injected with [³H]thymidine-labeled LS180 cells. After 3 hr, tissues were homogenized and proteolyzed, and radioactivity was monitored, as described under Materials and Methods. To correct for differences in recovery of radioactivity per mouse, the results are shown as the ratios of radioactivity found in various organs relative to that found in the brain (which had 4–7% of the radioactivity in all mice). Residual radioactivity in the circulating blood was similar in both groups. Statistical significance values were generated by using Wilcoxon rank–sum test.

Figure 5

P-selectin-deficient platelets fail to rosette on colon cancer cells in a calcium- and mucin-dependent manner. Calcein-labeled mouse platelets were studied for their interactions with cultured LS-180 cells as described in Materials and Methods. Left, fluorescence; Right, phase-contrast images. (A) P-selectin +/+ platelets. (B) P-selectin −/− platelets. (C) P-selectin +/+ platelets in 1 mM EDTA. (D) P-selectin +/+ platelets, tumor cells pretreated with OSGPase. (Bar = 50 μm.)

Figure 6

In vivo interactions of intravenously injected tumor cells with platelets is P-selectin-dependent. Mice were intravenously injected with fluorescently labeled LS180 cells, and, after 30 min, frozen sections of the lungs were studied for the presence of tumors cells and associated platelets (anti-CD41) as described under Materials and Methods.