A novel role for factor VIII and thrombin/PAR1 in regulating hematopoiesis and its interplay with the bone structure

Abstract

Analysis of hematopoietic stem cells (HSCs) in factor VIII knockout (FVIIIKO) mice revealed a novel regulatory role for the coagulation cascade in hematopoiesis. Thus, HSCs in FVIIIKO mice had reduced proportions of CD34(low) cells within Lin(-)Sca(+)Kit(+) progenitors, and exhibited reduced long-term repopulating capacity as well as hyper granulocyte-colony-stimulating factor (G-CSF)-induced mobilization. This disregulation of HSCs is likely caused by reduced levels of thrombin, and is associated with altered protease-activated receptor 1 (PAR1) signaling, as PAR1 KO mice also exhibited enhanced G-CSF-induced mobilization. Analysis of reciprocal bone marrow (BM) chimera (FVIIIKO BM into wild-type recipients and vice versa) and the detection of PAR1 expression on stromal elements indicates that this phenotype is likely controlled by stromal elements. Micro-computed tomography analysis of distal tibia metaphyses also revealed for the first time a major impact of the FVIII/thrombin/PAR1 axis on the dynamic bone structure, showing reduced bone:tissue volume ratio and trabecular number in FVIIIKO and PAR1KO mice. Taken together, these results show a critical and novel role for the coagulation cascade, mediated in part by thrombin-PAR1 interaction, and regulates HSC maintenance and a reciprocal interplay between HSCs and the dynamic bone structure.

Figure 1

LSK cell levels in the BM of FVIIIKO vs wild-type (WT) mice. (A) Percentage of LSK cells in FVIIIKO (white) compared with WT (C57BL/6) mice (black) (0.976% ± 0.007% vs 0.946% ± 0.019% (n = 10, P = .671). (B) Representative FACS histograms comparing isotype control staining to CD34 staining on C57BL and FVIIIKO LSK are shown. (C) In 4 independent experiments, a total of 8 C57BL/6 and 7 FVIIIKO samples were stained for CD34 and the data analyzed as follows: first, the average CD34 median of the C57BL samples in each experiment was calculated. This value was then used as a reference for comparison against individual median CD34 values in the same experiment (ie, individual median of each mouse was divided by the average median staining of the C57BL group in the same experiment). FVIIIKO mice (gray) exhibited a significantly higher value (1.43 ± 0.226, N = 7) compared with the median value of C57BL (black) (1.0 ± 0.067, N = 8), P < .001. Error bars are means ± SD. (D-G) FVIII KO HSCs exhibit similar short-term but reduced LTR capacity. (d) Schematic representation of the competitive repopulation assay. Donor-type chimerism was defined by FACS in each recipient, and the average level was calculated based on the values found in the 6 recipients of every individual donor. In control experiments, the BM mixture was composed of similar numbers of B6SJL (CD45.1) and C57BL (CD45.2) WT BM cells. (E) Chimerism levels at 6 and 22 weeks after transplantation of C57BL/6 BM (black) and FVIIIKO BM (white). (F) Chimerism levels at different time points after transplantation of FVIIIKO BM (dashed line) and C57BL BM (black). Each line is based on the average values found in 6 recipients receiving BM from the same donor. Total number of mice was 30. (G) Average linear trend line slopes of donor-type chimerism over time in mice receiving FVIIIKO BM (square) vs mice receiving C57BL/6 BM (diamond). Error bars are means ± SD (n = 30 mice).

Figure 2

G-CSF–induced splenomegaly and LSK cell mobilization is enhanced in FVIIIKO mice. (A) Macroscopic view of selected spleens of FVIIIKO and C57BL/6 mice either with or without 7 days of G-CSF treatment. (B) Average spleen weight (mg) of C57BL/6 (left) and FVIIIKO (right) mice in the presence or absence of G-CSF treatment (2-way ANOVA, P < .01). (C) Number of CFUs before and after G-CSF treatment in the spleen of FVIIIKO (white) or C57BL/6 mice (black). (D) Number of LSK cells in PB of control and G-CSF–treated splenectomized FVIIIKO and C57BL/6 mice. Error bars are means ± SD.

Figure 3

Thrombin\PAR1 axis is associated with enhanced G-CSF stimulation in FVIII KO mice. (A) Thrombin blood plasma levels (nM) in FVIIIKO (gray) and C57BL (black) mice at different time points as measured by the TGA. Summary of the results is depicted on the right (n = 4). (B) The number of LSK cells in PB of G-CSF–treated C57BL/6 (black), PAR1KO (gray), and PAR2KO (white) mice. (C) CXCL12 protein BM plasma levels detected by ELISA with and without G-CSF treatment in C57BL (black), FVIIIKO (white), and PAR1KO (gray) mice. Data are represented in (ng/mL); n = 12 per group.

Figure 4

FVIIIKO and PAR1 KO mice exhibit impaired interaction between stroma cells and hematopoietic niche. (A) A scheme depicting the procedure in the assay used to distinguish between the roles of PAR1 on stromal cells vs PAR1 on HSCs. Host mice were lethally irradiated by split-dose total body irradiation (total of 10 Gy on day 3 and 4 Gy on day 1) and transplanted (day 0) with 5 × 10⁶ donor cells. One month later, mice were mobilized by G-CSF for 7 days. (B) Spleen weight was examined for the 3 different chimeras tested as indicated. Bars represent average values ± SD.

Figure 5

PAR1 expression by stromal niche cells. (A) A representative immunohistological staining of frozen femur from C57BL/6 mouse is shown. Sample was stained for Osteopontin (red) and PAR1 (green). Arrowheads point to some of the positive PAR1 staining. PAR1-positive cells were detected within the bone trabeculae, along the endosteum and as scattered cells areas inside the hematopoietic compartment (magnification, ×10). (B) Two larger magnification (×40) samples of staining for extracellular OPN together with membrane bound PAR1 staining are shown. Isotype control staining is depicted at the inset, using an irrelevant isotype-matched primary Ab for both PAR1 and OPN, followed by the secondary Ab used for the specific staining. OPN, osteopontin. (C) PAR1 specific expression on osteoblasts population is shown. Osteocalcin (green) is used as osteoblast specific intracellular marker. PAR1 is represented in red. Merge (yellow) demonstrates PAR1 expression on osteocalcin positive osteoblasts. Nuclei are denoted by Hoechst staining (blue). (D) FACS analysis of PAR1 expression on stromal cells. PAR1-positive signal was detected on CD45⁻CD31⁻Ter119⁻ + Sca1⁻PDGFRα⁺ subpopulation in C57BL/6 mice (red line). Similar cells of PAR1 KO mice were used as negative control (green line). (E) PAR1 expression was determined in GFP-CXCL12 and GFP-Nestin mice. Flushed fraction of BM was stained for CD45, CD31, and Ter119 and the negative fraction was examined for GFP signal. As can be seen, PAR1 expression was determined in both GFP CXCL12 (green) and GFP-Nestin (blue) mice. Isotype antibody staining was used as a negative control.

Figure 6

Aberrant bone structure in FVIIIKO and PAR1KO mice. (A) Bone connective tissue staining with sirius red for collagen synthesis is depicted. C57BL, FVIIIKO and PAR1KO are shown, respectively, at NB and 14 weeks of age. Arrows point to diminished amount of trabeculae in FVIIIKO and PAR1KO mice compared with C57BL control. NB, newborn. (B) Quantification of the structural parameters of the tibial metaphysis such as BMD, BV:TV, Tb.N, Tb.Th, and Tb.Sp. in C57BL (black), FVIIIKO (white), and PAR1KO (blue) are shown graphically. Graphs show mean value ± SD (#P < .001; *P < .05, n = 4/group). (C) Quantification of serum osteocalcin measured by ELISA. C57BL (black), FVIIIKO (white), and PAR1KO (green) are shown. Bars depict mean value ± SD (**P < .05, n = 10 per group).