The polyphosphate-factor XII pathway drives coagulation in prostate cancer-associated thrombosis

Abstract

Cancer is a leading cause of thrombosis. We identify a new procoagulant mechanism that contributes to thromboembolism in prostate cancer and allows for safe anticoagulation therapy development. Prostate cancer-mediated procoagulant activity was reduced in plasma in the absence of factor XII or its substrate of the intrinsic coagulation pathway factor XI. Prostate cancer cells and secreted prostasomes expose long chain polyphosphate on their surface that colocalized with active factor XII and initiated coagulation in a factor XII-dependent manner. Polyphosphate content correlated with the procoagulant activity of prostasomes. Inherited deficiency in factor XI or XII or high-molecular-weight kininogen, but not plasma kallikrein, protected mice from prostasome-induced lethal pulmonary embolism. Targeting polyphosphate or factor XII conferred resistance to prostate cancer-driven thrombosis in mice, without increasing bleeding. Inhibition of factor XII with recombinant 3F7 antibody reduced the increased prostasome-mediated procoagulant activity in patient plasma. The data illustrate a critical role for polyphosphate/factor XII-triggered coagulation in prostate cancer-associated thrombosis with implications for anticoagulation without therapy-associated bleeding in malignancies.

Figure 1

Prostasomes initiate lethal PE in mice via the intrinsic coagulation pathway. (A) PE was induced by intravenous infusion of PC3 cell-derived prostasomes (0.8 µg/g body weight [bw]) in WT, F12^−/−, F11^−/−, Kng1^−/−, Klkb1^−/−, and Bdkrb2^−/− mice or WT animals pretreated with ASIS (WT + ASIS; 2.5 µg/g bw). Mortality was assessed in each group of mice (n = 5-15). Animals alive 30 minutes after challenge were considered survivors. Horizontal bars represent mean values. ***P < .001 and **P < .01 vs untreated WT. (B) Prostasome challenged mice were intravenously infused with Evans blue shortly after the onset of respiratory arrest while the heart was still beating or after 30 minutes for those animals that survived. Lungs were excised, and perfusion defects were analyzed. (C) Immunohistochemical localization of fibrin deposition on sections from lungs of WT, F12^−/−, F11^−/−, and WT + ASIS-treated mice using the fibrin-specific antibody 59D8 (high magnification, lower right). Sections were counterstained with Mayer's hematoxylin (bar, 100 µm). (D) Thrombi per visual field were counted at ×10 magnification from sections such as those in C. Columns are mean ± standard error of the mean (SEM) for 35 fields. (E) Accumulation of fibrin in lungs of prostasome-challenged WT, F12^−/−, F11^−/−, and WT + ASIS mice was analyzed by immunoblotting. (F-G) Pulmonary thromboembolism model as in A induced by injection of 10 µg/g bw seminal prostasomes (F; n = 6) or 150 µg/g bw healthy male or patient prostasomes/exosomes (G; n = 5), *P < .05 vs untreated WT (F) and **P < .01 vs healthy males (G). P values were determined using 1-way analysis of variance (ANOVA); n.s., nonsignificant.

Figure 2

Prostasomes trigger coagulation in a FXII-dependent manner. Comparison of (A-C, upper) seminal and (D-F, lower) PC3 cell prostasomes: TEM images show morphologies of isolated (A) seminal and (D) PC3 prostasomes (bar, 500 nm). Size distributions of (B) seminal and (E) PC3 prostasomes assessed from TEM images as shown in A and D, 6 fields each. Real-time thrombin generation stimulated with increasing concentrations (0-28 µg/mL) of (C) seminal and (F) PC3 prostasomes in platelet-free (solid lanes) and ultracentrifuged (dashed line) plasma. Representative curves of a series of n = 6 are shown. (G-H) Mechanism of prostasome-driven coagulation: real-time thrombin generation initiated by 14 µg/mL PC3 prostasomes in NP, NP supplemented with rHA-Infestin-4 (Inf-4; 500 µg/mL), ASIS (30 nM), or a combination of ASIS and Inf-4 (30 nM and 500 µg/mL) and plasma deficient in FXII (FXII def.) or FXI (FXI def.); n = 3 to 11. (I) PC3 prostasome (1.75 µg/mL)-stimulated recalcification clotting time in ultracentrifuged NP in the presence of inhibitors as in G and H or buffer or in FXII- or FXI-deficient plasma. **P < .01 and ***P < .001 vs NP, n = 5. (J-L) Contact activation was analyzed by conversion of the FXIIa/PK chromogenic substrate S-2302. Buffer stimulated plasma is shown as control. (J) Plasma was incubated with increasing concentrations of PC3 prostasomes (0-200 µg/mL). (K) PC3 prostasome (100 µg/mL)-induced S-2302 conversion in NP or plasma deficient in FXII (FXII def.), prekallikrein (PPK def.), or high-molecular-weight kininogen (HK def.). *P < .05 and ***P < .001 vs NP + buffer. Bars represent the absorbance at λ = 405 nm at 60 minutes; n = 6. (L) Normal (dark columns) or FXII-deficient (light columns) plasma was incubated with PC3 prostasomes and exosomes from various cancer cells (Panc1, BxPC3, Capan2, and HL-60; 100 µg/mL each). Absorbance of cleaved S-2302 in prostasome-treated plasma is given relative to buffer signal at 60 minutes. ***P < .001 NP vs FXII def., n = 3 to 6. P values were determined using 1-way ANOVA (I,K) or Student t test (L). Data are presented as mean ± SEM; n.s., nonsignificant.

Figure 3

Prostasomes, PC cells, and PC tissue expose procoagulant polyP. Isolation of polyP from prostasomes and PC cells: (A-B) PolyP was extracted from PC3 (A) prostasomes and (B) cells by an anion exchanger chromatography, separated by electrophoresis on polyacrylamide/urea gel and visualized by DAPI-negative staining. Synthetic polyP with mean chain lengths of 134, 383, and 637 serves as molecular size standard. Purified polyP and synthetic LC polyP were loaded before and after incubation with phosphatase (Psp, 10 U/mL for 1 hour). (C,E) Epifluorescence images of polyP on PC3 prostasomes using Alexa 594-labeled PPBD. The polymer colocalizes with (C) prostasome surface membrane marker CD63 (anti-CD63) and (E) FXIIa (3F7). (D,F) Confocal laser scanning microscopy of polyP in (D) nonpermeabilized and (F) permeabilized PC3 cells. PolyP is stained with Alexa 488-conjugated PPBD, and DNA was visualized with DAPI. Bar is 500 nm in C and E and 10 µm in D and F. (G-H) PC polyP activates FXII. Platelet-free plasma was incubated with 1.5 µg/mL PC3 (G) prostasome or (H) cell polyP in the absence or presence of PPBD (1 mg/mL), generation 1.0 dendrimer (25 µg/mL), or Psp (1 U/mL). PolyP-treated FXII-deficient plasma (FXII def.) and buffer-stimulated normal plasma (buffer) is shown as control. Hydrolysis of the chromogenic substrate S-2302 measures formed FXIIa and PK. Mean ± SEM, n = 6. (I-J) Real-time thrombin formation generated in platelet-free plasma stimulated with 1.5 µg/mL (I) prostasome or (J) PC3 cell polyP in the absence or presence of PPBD (0.5 and 2 mg/mL) or generation 1.0 dendrimer (25 µg/mL). Thrombin formation in polyP-activated FXII-deficient plasma (FXII def.) and buffer-stimulated normal plasma is blotted for comparison. A representative thrombin generation curve of a series of n = 6 is shown. (K) Mortality associated with intravenous injection of PC3 prostasomes (0.8 µg/g bw) in WT mice pretreated with saline or PPBD (150 µg/g bw) was assessed in each group (n = 5); animals alive at 30 minutes after challenge were considered survivors. *P < .05 vs saline, unpaired Student t test. (L) PolyP detection on seminal, PC3 cell, and patient prostasomes using PPBD binding in an ELISA. Bars represent polyP content relative to seminal prostasomes, mean ± SEM, n = 3. ***P < .001 vs seminal prostasomes, 1-way ANOVA. (M-P′) Immunohistochemistry of FXII, polyP, and fibrin in human PC tissue sections. (M-M′) Hematoxylin/eosin-staining of the adenocarcinoma (arrows denote occluded vessels). Immunohistochemical localization of FXII using (N-N′) anti-FXII antibodies (arrows indicate FXII-positive occluded vessels and cancer tissue, respectively), (O-O′) polyP using recombinant PPBD as a probe (arrows), and (P-P′) fibrin deposits using 59D8 antibody (arrows) in (M-N′) paraffin and (O-P′) cryo sections. Black and yellow asterisks denote tissue areas containing cancer cells and nonmalignant tissue, respectively. Sections were counterstained with Mayer’s hematoxylin. Scale bars, 100 µm.

Figure 4

FXIIa inhibition protects mice from prostasome-induced PE. (A) Real-time thrombin formation in murine platelet-free plasma stimulated with PC size synthetic polyP (LC polyP, 100 µg/mL) in the absence or presence of increasing concentrations of anti-FXIIa antibody 3F7 (n = 3). Molar antibody concentrations are relative to plasma FXII (375 nM). (B) Mortality associated with intravenous injection of PC3 cell prostasomes (0.8 µg/g bw) in WT mice pretreated with saline or 3F7 (4.5 µg/g bw) was assessed in each group (n = 6); animals still alive at 30 minutes after challenge were considered survivors. ***P < .001 vs saline, unpaired Student t test. (C) Mice challenged with prostasomes were intravenously infused with Evans blue shortly after respiratory arrest while lungs were still perfused. Excised lungs show perfusion defects in red. (D) Sections from lungs of saline- and 3F7-treated WT mice were analyzed for fibrin by immunohistochemistry with 59D8 antibody and counterstained with Mayer’s hematoxylin; bar, 100 µm. (E) Thrombi per visual field were counted at ×10 magnification from sections such as those in D. ***P < .001 vs saline, unpaired Student t test. Columns are mean ± SEM for 35 fields. (F-G) 3F7 treatment does not impair hemostatic capacity. (F) Tail bleeding times and (G) total hemoglobin loss assessed by hemoglobin absorbance at λ = 575 nm was determined in saline- and 3F7-infused mice. Mean ± standard deviation, n = 10.

Figure 5

Prostasome/FXIIa-driven increased procoagulant activity in PC patient plasma. (A-D) Real-time thrombin generation in plasma of healthy individuals (Controls) and patients (Patients). Thrombin formation in prostasome-depleted [Patients(−)] and prostasome-reconstituted [Patients(+)] patient plasma is blotted for control; n = 9 each. (E-H) Real-time thrombin generation in normal plasma stimulated by addition of prostasomes/exosomes (250 µg/mL) from healthy female and male controls (n = 10 each) or PC patients (n = 20). Patient prostasome-triggered coagulation was measured in the absence or presence of 3F7 (100 µg/mL). (A,E) Lag time until thrombin formation starts, (B,F) time to peak thrombin, (C,G) maximum thrombin (peak thrombin), and (D,H) total thrombin (ETP). Each symbol represents an individual. Horizontal bars indicate mean values. *P < .05, **P < .01, and ***P < .001, unpaired Student t test; n.s., nonsignificant.