Figure 1.
Analysis of MDA-MB-231 mfp cells sorted based on EPCR expression.
A Cells were double stained for CD44 and CD24, or EPCR and TF using directly labeled antibodies and analyzed by FACS. B Non-permeabilized cells were stained for EPCR and TF, confirming the presence of EPCR+/TFlow and EPCR−/TFhigh subpopulations in adherent MDA-MB-231 mfp cells; two examples of stained unsorted populations are shown (scale bar 15 µm). C Western blotting of sorted MDA-MB-231 mfp cells confirmed the EPCR+/TFlow and EPCR−/TFhigh phenotypes, respectively, and excluded intracellular pools of EPCR in EPCR− cells. A representative blot and quantification of EPCR and TF levels in cells from 4 independent sorts are shown (*** p<0.001, ** p<0.005, T-test; mean±SD). D EPCR+ and EPCR− populations were isolated in 6 consecutive FACS sorts and profiled for mRNA expression. The sorts of passage 22 and 23 cells were treated as replicates in the data analysis and the clustering for the mean value for these samples and 4 additional earlier and later sorts is shown. Hierarchical Clustering Subbranches: The dendrogram was cut at a defined height to yield 10 gene clusters. Of these, 6 had transcripts upregulated in EPCR+ cells (Branch 1–6) and 4 had transcripts upregulated in EPCR− cells (Branch 7–10). Gene counts were: Branch 1, n = 405; Branch 2, n = 198; Branch 3, n = 59; Branch 4, n = 54; Branch 5, n = 624; Branch 6, n = 2; Branch 7, n = 107; Branch 8, n = 844; Branch 9, n = 85; Branch 10, n = 264 (see Table S1 for gene lists). The genes for each cluster were used as input into Ingenuity IPA to identify pathway annotations or representation of gene signatures (see Table S2).
Figure 2.
EPCR supports tumor growth in a murine model of spontaneous and carcinoma breast cancer model.
A Cohorts of PyMT mice expressing very low levels of EPCR (EPCRLow/Low, n = 26) and control mice (EPCRlow/WT, n = 29) were followed weekly for appearance of palpable tumors (upper panel) and tumor growth. Tumor appearance was not different between cohorts, but cumulative tumor volumes were reduced in EPCRLow/Low versus EPCRlow/WT mice (middle panel, *p<0.05, Mann-Whitney, mean±SEM). EPCRlow/WT heterozygous control mice are eliminated from the cohort due to large tumor sizes earlier, resulting in significantly increased survival for EPCRLow/Low mice (lower panel, ***p<0.001 Log Rank Test). Note that EPCRlow/WT mice had similar tumor sizes as C57BL/6 mice followed at the same time in the same facility, indicating that there was no gene dose effect in heterozygous mice. B PyMT-WT cells (2×106 cells/mouse) were injected into the mammary gland of EPCRlow/WT or EPCRLow/Low mice and a typical experimental outcome for tumor growth is shown (no difference between groups, n = 8 mice/group; mean±SD; confirmed in an independent experiment). C Western blotting for EPCR and TF of cell lysates from PyMT-EPCRflox/flox cells treated twice with 1000 particles/cell (ppc) of Ad5 control or Ad5 cre recombinase D. Cells depicted in C (4×106 cells/mouse) were implanted into the mammary fat pad of heterozygous EPCRlow/WT female mice (** p<0.005, T-test, n = 7 mice/group; mean±SD). Final tumor weights (*** p<0.001, T-test; mean±SD). E An independently established PyMT-EPCRflox/flox line was selected by passage through the mammary fat pad (mfp), treated with Ad5 control or Ad5 Cre virus, and injected at a dose of 1×105 cells/mouse into the mfp of C57BL/6 female mice for tumor growth monitoring (delayed tumor appearance * p<0.05, Log-rank Test, and reduced tumor weights, **p<0.005, T-test, n = 8; mean±SD).
Figure 3.
EPCR+ MDA-MB-231 mfp cells have stem cell-like properties.
A EPCR+ and EPCR− subpopulations isolated based on the gates in Fig. 1A were seeded at 104 cells/well and grown in mammosphere medium for 10 days. Growth was quantified by cell counting B MDA-MB-231 mfp cells were sorted based on EPCR. Sorted EPCR+ and EPCR− cells, as well as parental cells were cultured for 3 days, harvested and stained with the ALDH kit with DEAB addition as a specificity control, and analyzed by FACS. EPCR+ cells had higher ALDH activity than EPCR− or parental cells. At the time of analysis, the parental populations had 8% EPCR+ cells. The experiment was repeated with the same outcome. C Tumor cells were isolated from freshly harvested tumors and immediately analyzed by FACS for TF and EPCR. A representative FACS profile and quantification for 12 EPCR+ cell-derived tumors and 4 EPCR− cell-derived tumors are shown (mean±SD).
Table 1.
Tumor-initiating capacity of EPCR+ and EPCR− cells in the mfp.
Figure 4.
Blocking EPCR alters proliferation in vitro and tumor take of MDA-MB-231 mfp cells in vivo.
A Sorted cells were stained for integrin subunits α4, α6, β1 and β4 and analyzed by FACS. EPCR+ cells are α4β1 positive whereas EPCR− cells are α4 negative, but express β1. Conversely, EPCR− cells expressed α6β4, and EPCR+ expressed α6 but not β4 integrin. B FACS-isolated EPCR+ or EPCR− cells were seeded in 48-well plates coated with CS-1 (5 µg/ml), or serum-free 805 G cell supernatant diluted 1∶2 with DMEM, and cultured in the presence of 100 µg/ml control IgG, αEPCR-1500, or αEPCR-1535 for 48 hours. Cell numbers were quantified by MTT assay (different between control IgG and αEPCR-1535, *p<0.05 ANOVA, followed by Bonferroni test, n = 3, mean±SD). C MDA-MB-231 mfp FACS-isolated EPCR+ cells (104/mouse) were injected into the mammary fat pad in matrigel containing 0.5 mg of either control IgG, non-blocking αEPCR-1500 or blocking αEPCR-1535. The blocking antibody significantly reduced tumor take (** p<0.005, Log-rank Test, n = 7 for IgG and αEPCR-1500, n = 8 for αEPCR-1535) and final tumor sizes (** p<0.005 versus control IgG, ANOVA with Bonferroni post test, mean±SD).
Figure 5.
Blocking EPCR decreases tumor growth of human breast cancer cells.
A MDA-MB-231 mfp cells (1×106 cells/mouse) were mixed with 1 mg control IgG or αEPCR-1535 and injected into the mammary fat pad of SCID mice. Blocking EPCR significantly reduced tumor growth and final tumor weights (**p<0.05, T-test, mean±SD, n = 8 mice/group). B MDA-MB-231 mfp cells (1×106 cells/mouse) were injected in the mammary fat pad of SCID mice and mice were treated at days 4, 7 10, 12, and 15 by intraperitoneal injections of 1 mg of control IgG, αEPCR-1500, or αEPCR-1535. Treatment with inhibitory αEPCR-1535 significantly reduced tumor growth and final tumor weights (ANOVA followed by Bonferroni posttest, * p<0.05, ** p<0.005, *** p<0.001 IgG versus αEPCR-1535; # p<0.05 αEPCR-1500 versus αEPCR-1535, § p<0.05 IgG versus αEPCR-1500).
Figure 6.
Expression analysis of PyMT tumor macrophages.
PyMT tumors from heterozygous PyMT-EPCRLow/WT mice were dispersed and macrophages were separated with αCD11b (A) or αCD11c (B) paramagnetic beads from tumor cells and stromal cells. Tumor macrophages were CD11b+/CD11c+/F4/80+ by FACS and CD11c was used to assess the efficiency of selection. Expression of the indicated mRNAs was determined by RT-PCR. Coagulation factor mRNA was normalized to a standard curve of normal mouse liver mRNA (* p<0.05, ** p<0.005 T-test, mean±SD, n = 5–8).