Murine platelet production is suppressed by S1P release in the hematopoietic niche, not facilitated by blood S1P sensing

Abstract

The bioactive lipid mediator sphingosine 1-phosphate (S1P) was recently assigned critical roles in platelet biology: whereas S1P₁ receptor-mediated S1P gradient sensing was reported to be essential for directing proplatelet extensions from megakaryocytes (MKs) toward bone marrow sinusoids, MK sphingosine kinase 2 (Sphk2)-derived S1P was reported to further promote platelet shedding through receptor-independent intracellular actions, and platelet aggregation through S1P₁ Yet clinical use of S1P pathway modulators including fingolimod has not been associated with risk of bleeding or thrombosis. We therefore revisited the role of S1P in platelet biology in mice. Surprisingly, no reduction in platelet counts was observed when the vascular S1P gradient was ablated by impairing S1P provision to plasma or S1P degradation in interstitial fluids, nor when gradient sensing was impaired by S1pr1 deletion selectively in MKs. Moreover, S1P₁ expression and signaling were both undetectable in mature MKs in situ, and MK S1pr1 deletion did not affect platelet aggregation or spreading. When S1pr1 deletion was induced in hematopoietic progenitor cells, platelet counts were instead significantly elevated. Isolated global Sphk2 deficiency was associated with thrombocytopenia, but this was not replicated by MK-restricted Sphk2 deletion and was reversed by compound deletion of either Sphk1 or S1pr2, suggesting that this phenotype arises from increased S1P export and S1P₂ activation secondary to redistribution of sphingosine to Sphk1. Consistent with clinical observations, we thus observe no essential role for S1P₁ in facilitating platelet production or activation. Instead, S1P restricts megakaryopoiesis through S1P₁, and can further suppress thrombopoiesis through S1P₂ when aberrantly secreted in the hematopoietic niche.

Graphical abstract

Figure 1.

MK S1P, S1P₁S1P, and the S1P gradient are dispensable for platelet production in mice. (A) Current literature suggests that S1P supports platelet production by 2 independent mechanisms: S1P₁ senses the S1P gradient to promote PP extensions toward blood sinusoids (purple) and further supports fragmentation, and S1P supports platelet fragmentation by receptor-independent promotion of Src family kinase (SFK) expression and activation in MKs (orange). Removal of the S1P gradient and MK S1P production would thus be predicted to have cumulative effects on platelet production. Target cells of Cre alleles used in this study are indicated. (B) Peripheral blood platelet and lymphocyte counts in mice with combined loss of lymphatic endothelium and hematopoietic S1P production (Sphk1^f/−:2^f/−:Mx1Cre⁺) and with alternative gradient ablation by impaired S1P breakdown (Sgpl1^−/). (C) Relative changes in the same cell populations after supplying the functional S1P₁ antagonist fingolimod (2 mg/L) in the drinking water of wild-type mice for 1 week. (D) Platelet and lymphocyte counts in mice after deletion of S1pr1 in hematopoietic and other cells (postnatal induction, Mx1Cre⁺), in all hematopoietic cells (constitutive deletion, Vav1 Cre⁺), in MKs (constitutive deletion; Pf4Cre⁺), or endothelial cells and MKs (postnatal induction, PdgfbiCreERT2⁺). (E-F) Platelet and lymphocyte counts in adult S1pr1^f/f:Mx1Cre^+/− mice 1 month after 3 consecutive injections of Poly IC (E) or in lethally irradiated wild-type mice 1 month after transfusion of S1pr1^f/f:Mx1Cre^+/− BM cells (F). (G) Relative change in platelet counts 24 hours after injections of the S1P₁ agonist SEW2871 or antagonist W146 or respective vehicle controls, as indicated (W146 was injected either as a single bolus [middle] or at 0, 6, 12, and 18 hours [right]). Lymphocyte counts and acute effects of drug treatment in supplemental Figure 1. (H) Platelet half-life, mean platelet volume (MPV), and plasma S1P levels in S1pr1^f/f:Mx1Cre⁺ mice. (I) RBC counts in mice with hematopoietic deletion of S1pr1 or Sphk1&2 (the same mice as in Figure 1D, 1F, 1E, and 1B, respectively). (J) Representative microcomputed tomography images of femurs from S1pr1^f/f:Mx1Cre⁺ mice and littermate controls. Representative coronal and transverse sections (approximate area indicated) are shown, quantification in supplemental Figure 2. (K) BM progenitors as percentage of total bone marrow cells and MK density in BM and spleen of S1pr1^f/f:Mx1Cre⁺ mice. Representative images in supplemental Figure 3. All animals are compared with their respective littermate controls, n indicates the number of animals from which samples were obtained, mean + standard error of the mean shown. Statistical analyses by Mann-Whitney U test. ns, not significant.

Figure 2.

S1P₁is not expressed in murine MKs. (A) Expression of S1P₁ (red) and the MK marker CD41 (green) in spleen of mice with or without MK-selective (S1pr1^f/f:Pf4Cre⁺) or pan-hematopoietic (S1pr1^f/f:Mx1Cre⁺) S1P₁ deletion. Note S1P₁ expression in blood vessels and white pulp, but not in MKs, irrespective of gene deletion. Scale bars represent 50 μm. (B) Constitutive (left) and fingolimod-induced (1 mg/kg, 24 hours; right) S1P₁ signaling in spleens of S1P₁ signaling reporter mice. Note constitutive and fingolimod-enhanced S1P₁ signaling (reflected by nuclear GFP accumulation in green) in blood vessels (in red) and cells within the white pulp, but not MKs (in blue). Scale bars represent 50 μm. (C) Abundance of S1PR transcripts in BM-derived MKs from S1pr1-deficient mouse lines relative to Gapdh. Note lack of S1pr1 expression (mRNA) or compensatory upregulation of other receptors after 3 days of culture. (D) Abundance of nonexcised S1pr1 in genomic DNA (gDNA) from BM-derived MKs from S1pr1-deficient mouse lines after 5 days of culture relative to pooled S1pr1^f/f littermate controls. (E) Relative abundance of nonexcised S1pr1 in genomic DNA from freshly isolated BM cells from S1pr1-deficient mice passed through 70-μm filters. Statistical analysis by Mann-Whitney U test. n.d., not detectable.

Figure 3.

Platelet S1P₁is dispensable for platelet aggregation and spreading in mice. Platelets from mice in which S1pr1 was deleted in MKs (S1pr1^f/f:Pf4Cre⁺; green), littermate controls (orange), or wild-type mice (gray) were isolated, washed, and tested for their capacity to aggregate (A) and spread (B). (A, upper) Platelet aggregation in response to submaximal concentrations (25 μM) of PAR4-AP (thrombin receptor agonist) in the absence (left) or presence of exogenous S1P (0.1-10 μM; middle) or of S1P₁ agonist SEW-2871 (0.5 μM; right). (A, lower) Platelet aggregation in response to submaximal PAR4-AP in the presence of S1P (0.1 μM) in the presence or absence of S1P₁ antagonist W146 (10 μM; left) or in response to the weak platelet agonist ADP (2 μM; right), with and without exogenous S1P (10 μM) or P2Y12 antagonism (2MeSAMP, 40 μM) to address potential redundancy with P2Y12, which, similar to S1P₁, is Gαi coupled. (B) Representative scanning electron microscopy images (upper) showing the extent of platelet spreading 15 and 30 minutes after plating on fibrinogen in the presence of S1P (0.5 μM; quantification in supplemental Figure 7). Note that although S1P did not trigger aggregation (not shown), it slightly increased PAR4-AP induced aggregation. However, neither aggregation nor spreading was influenced by selective S1P₁ modulation or S1pr1 deficiency. S1P also could not compensate for the absence of functional P2Y12 by alternative engagement of Gαi. Statistical analyses by 2-way analysis of variance or the Mann-Whitney U test, as appropriate. Mean ± standard deviation is shown, symbols represent the number of mice.

Figure 4.

Sphk2 deficiency induces thrombocytopenia by redirection of sphingosine to Sphk1. (A-B) Platelet counts and MPV from Sphk2 heterozygous intercrosses in C57BL/6J:129SVJ mixed background. (B) Effect of MK (Pf4Cre)- and lymphatic endothelium/CD45+ (Lyve1Cre)–selective Sphk2 deletion on platelet counts. (C) Thrombocytopenia in Sphk2^−/− mice could be explained by redistribution of sphingosine to Sphk1 rather than by loss of Sphk2-derived S1P. This, in turn, could impair MK maturation by a receptor-dependent mechanism after S1P export by Spns2 or Mfsd2b, depending on cell type. (D) Impact of Sphk2 deficiency on the expression of Sphks and S1PRs and levels of sphingosine in total bone marrow cells (S1P was below the detection threshold). (E) Effect of deletion of Sphk1 in Mx1Cre-sensitive cells on Sphk2 deficiency-induced thrombocytopenia and MPV. (F) Effect of transplantation of BM cells from mice lacking Sphk1&2 in Mx1Cre-sensitive cells to lethally irradiated wild-type recipients and vice versa on platelet counts in the host. Note that the rescue conferred by Sphk1 deficiency is BM cell-derived, as the Sphk2^−/− phenotype itself. (G) Effect of deletion of Sphk1 in MKs on Sphk2 deficiency-induced thrombocytopenia and MPV. (H) Effect of compound Sphk1 deficiency on MKP and platelet life span in Sphk2^−/− mice. Statistical analyses by Mann-Whitney U test or 2-way analysis of variance.

Figure 5.

Sphk2 deficiency induces thrombocytopenia by aberrant S1P₂activation. (A) Effects of S1P₁ (W146, 10 mg/kg, left) or S1P₂ (JTE-013, 1.2 mg/kg) antagonism on Sphk2 deficiency-induced thrombocytopenia (24-hour platelet counts). (B-D) Platelet counts and MPV in litters from independent intercrosses of S1pr2^+/− in a wild-type background (B), S1pr2^+/− in a Sphk2-deficient background (C), and of Sphk2^+/− in a S1P₂-deficient background (D). Note that although S1P₂ deficiency does not itself affect platelet counts, it rescues Sphk2 deficiency-induced thrombocytopenia. (E) Transmission electron micrographs of bone marrow from Sphk2^+/+, Sphk2^−/−, and Sphk2^−/−:S1pr2^−/− mice. Although the majority of MKs from Sphk2^+/+ and Sphk2^−/−:S1pr2^−/− mice were singular and large, with a mature appearance and well-defined demarcation membrane systems (upper), MKs in Sphk2^−/− were highly heterogeneous, with clusters of immature MKs or mature MKs with limited DMS next to blood sinusoids (middle), low-contrast MK “ghosts” that appeared to be undergoing necrosis (bottom left, next to a normal MK) and platelet release within the bone marrow (bottom right). Representative images from n = 4 mice per genotype are shown. Scale bars, 2 μm. (F) Effect of a bolus injection of the Rho kinase inhibitor Y27632 (10mg/kg) on platelet counts in Sphk2^−/− and Sphk2^+/+ controls. Note a significant increase in platelet counts only in the knockout. (G) Sphingosine content of Sphk deficient platelets. Statistical analyses by 2-way analysis of variance (A,F) or the Mann-Whitney U test.