Polyphosphate expression by cancer cell extracellular vesicles mediates binding of factor XII and contact activation

Abstract

Extracellular vesicles (EV) have been implicated in diverse biological processes, including intracellular communication, transport of nucleic acids, and regulation of vascular function. Levels of EVs are elevated in cancer, and studies suggest that EV may stimulate thrombosis in patients with cancer through expression of tissue factor. However, limited data also implicate EV in the activation of the contact pathway of coagulation through activation of factor XII (FXII) to FXIIa. To better define the ability of EV to initiate contact activation, we compared the ability of EV derived from different cancer cell lines to activate FXII. EV from all cell lines activated FXII, with those derived from pancreatic and lung cancer cell lines demonstrating the most potent activity. Concordant with the activation of FXII, EV induced the cleavage of high molecular weight kininogen (HK) to cleaved kininogen. We also observed that EVs from patients with cancer stimulated FXII activation and HK cleavage. To define the mechanisms of FXII activation by EV, EV were treated with calf intestinal alkaline phosphatase or Escherichia coli exopolyphosphatase to degrade polyphosphate; this treatment blocked binding of FXII to EVs and the ability of EV to mediate FXII activation. In vivo, EV induced pulmonary thrombosis in wild-type mice, with protection conferred by a deficiency in FXII, HK, or prekallikrein. Moreover, pretreatment of EVs with calf intestinal alkaline phosphatase inhibited their prothrombotic effect. These results indicate that polyphosphate mediates the binding of contact factors to EV and that EV-associated polyphosphate may contribute to the prothrombotic effects of EV in cancer.

Graphical abstract

Figure 1.

Characterization of EV. Particle numbers (A) and relative protein concentrations (B) in the fractions eluted following qEV size-exclusion chromatography. Data are mean ± SEM. (C) Tetraspanin (CD9, CD63, and CD81) content of purified EVs was analyzed in the indicated elution fractions by immunoblotting. (D) Size distribution of pooled fractions 7 to 10 from HDFs (left panel) and L3.6 cells (right panel). (E) Electron microscopy image of EV from HDFs (left panel) and L3.6 cells (right panel).

Figure 2.

polyP on EV derived from cancer cells mediates ligand binding. (A) The amount of polyP associated with cancer cell–derived EV was estimated by comparison with a standard curve (supplemental Figure 2A). (B) Detection of polyP on L3.6-derived EVs following incubation with buffer (HBS), DNase I and RNase A (D/R), or CIP. polyP of different chain lengths was used to estimate the size of EV-associated polyP. L3.6-derived EVs were preincubated with buffer or E coli PPX, and the binding of PPX-Δ12–AF488 (C), anti-CD81–AF488 (D), and or FXII-AF488 (E) was measured. The number of total EV was determined in scatter mode, and the number of ligand-binding EV was determined in fluorescence mode using a 488-nm excitation laser. The differences in fluorescent particle numbers were compared by estimating the area under each curve (AUC) using GraphPad Prism 8. Bars represent means ± SEM. ***P < .001, 1-way ANOVA with multiple comparisons. L, long; M, medium; S, short.

Figure 3.

FXII activation by EVs derived from cancer cell lines. (A) Activation of FXII induced by EV from cancer cells (L3.6, H1975, HT29, and U937) and HDFs. (B) FXIIa generation by cancer cell–derived EV. EV-induced S-2302 hydrolysis was estimated at 90 minutes after addition of the indicated amount of EV (determined by protein concentration) and comparison with a FXIIa standard curve (supplemental Figure 3). (C) FXIIa generated by cancer cell–derived EV treated with DNase I (10 U/mL), RNase A (10 U/mL), and CIP (1, 10, and 20 U/mL). (D) Effect of CTI on FXII activation induced by L3.6 EV in NHP and FXII-deficient (FXII-def.) plasma. Hydrolysis of S-2302 in plasma incubated with EV in the presence or absence of CTI was compared. (E) Activation of purified FXII (375 nM) by EV. EVs (20 or 40 µg protein) were incubated in the presence of FXII (375 nM) and S-2302 and A₄₀₅ were monitored for 90 minutes. (F) FXII cleavage in the presence of L3.6 EV was assessed by immunoblotting using the reaction mixture in (D). The FXII heavy chain (HC) and light chain (LC) were detected under reducing conditions using goat anti-human FXII polyclonal antibody. Bars represent means ± SEM. *P < .05, **P < .01, ***P < .001, 2-way ANOVA with multiple comparisons. ns, not significant.

Figure 4.

Cleavage of HK caused by cancer cell–derived EV. (A) HK cleavage in NHP incubated with cancer cell–derived EV (0.5 and 1 μg/mL) and analyzed by immunoblotting. (B) HK cleavage by EVs from (A) expressed as the percentage of cHK relative to intact HK. (C) HK cleavage 60 minutes after the addition of L3.6 EV following treatment of EV with CIP, DNase I, or RNase A. DS was used as a positive control. (D) Effect of CIP, DNase I, and RNase A on the cleavage of HK by EV from cancer cells and HDFs; the ratio of cHK/HK was determined by immunoblotting and infrared quantification. (E) Effect of CTI on L3.6 EV-induced cHK generation. L3.6 EV were incubated with NHP in the presence or absence of CTI (10 μg/mL) for 60 minutes. For HK cleavage analysis, a polyclonal rabbit anti-human HK antibody (D5; supplemental Figure 1) was used to detect HK and cHK. Bars represent means ± SEM. ***P < .001, 2-way ANOVA with multiple comparisons. LC1, light chain 1 (54 kDa); LC2, light chain 2 (47 kDa). ns, not significant.

Figure 5.

Pulmonary emboli in mice treated with EV derived from L3.6 cells. (A) Paraffin section of lung from a WT mouse after L3.6 EV infusion into the IVC (original magnification ×5; H&E stain). Scale bar, 500 μm. Frozen sections of lung following EV infusion into the IVC and staining for platelet GP1b (CD42c, red) or fibrin (monoclonal antibody 59D8, green) at original magnification 320 (scale bar, 100 mm). (B) At original magnification 310 (scale bar, 200 mm. (C) DAPI (blue) stains cell nuclei. (D) Number of thrombi observed in a random 10× field following infusion of L3.6 EVs into WT C57BL/6, f12^−/−, kng1^−/−, and klkb1^−/− mice. (E) Effect of preincubation of L3.6 EV with buffer alone (untreated) or CIP (treated) prior to IVC infusion on density of lung thrombi, measured per ×10× field. (F) Effect of preincubation of L3.6 EV with buffer alone (untreated) or CIP (treated) on time to death after IVC infusion. Each point represents an individual animal. Bars in graphs depict 95% confidence intervals , and comparisons between groups were performed using ANOVA and Student t tests. hpf, high-powered field; ns, not significant.

Figure 6.

Increased levels of cHK in plasma from patients with cancer. (A) HK cleavage in plasma from patients with pancreatic cancer and healthy donors (normal) was determined by immunoblotting using antibody to HK domain 5. (B) The same plasmas as in (A) were analyzed using an antibody specific for the C terminus of the free HK heavy chain (HC) after reduction (HKa1; supplemental Figure 1). (C) Ratio of cHK/HK in plasma from healthy individuals and patients with pancreatic cancer (PANC) (n = 26), determined by Wes capillary immunoblotting. (D) Ratio of cHK/HK in healthy individuals and patients presenting to the cancer thrombosis clinic (n = 21) with symptoms of VTE, as determined by Wes capillary immunoblotting. Bars represent means ± SEM. ***P < .001, ****P < .0001, unpaired Student t test with Welch’s correction. (E) Log relative hazard of VTE or death vs cHK/total HK ratio from the Cox proportional hazard modeling of patients with different types of cancer. Each red dot indicates a VTE or death event, and each black dot indicates a censored event; the shaded area is the 95% confidence interval for the log hazard. (F) Log relative hazard of VTE only (censored at death) vs cHK/total HK ratio from the Cox proportional hazard modeling of patients with different types of cancer. Each red dot indicates a VTE event, and each black dot indicates a censored event; the shaded area is the 95% confidence interval for the log hazard. (G) FXII activation by EVs immunopurified from a healthy donor or from patients with pancreatic or colon cancer. Data are means ± SEM. *P < .05, ***P < .001, 1-way ANOVA with a unpaired Student t test with multiple comparisons.