Targeting platelet-derived CXCL12 impedes arterial thrombosis

Abstract

The prevention and treatment of arterial thrombosis continue to be clinically challenging, and understanding the relevant molecular mechanisms in detail may facilitate the quest to identify novel targets and therapeutic approaches that improve protection from ischemic and bleeding events. The chemokine CXCL12 augments collagen-induced platelet aggregation by activating its receptor CXCR4. Here we show that inhibition of CXCR4 attenuates platelet aggregation induced by collagen or human plaque homogenate under static and arterial flow conditions by antagonizing the action of platelet-secreted CXCL12. We further show that platelet-specific CXCL12 deficiency in mice limits arterial thrombosis by affecting thrombus growth and stability without increasing tail bleeding time. Accordingly, neointimal lesion formation after carotid artery injury was attenuated in these mice. Mechanistically, CXCL12 activated via CXCR4 a signaling cascade involving Bruton's tyrosine kinase (Btk) that led to integrin αIIbβ3 activation, platelet aggregation, and granule release. The heterodimeric interaction between CXCL12 and CCL5 can inhibit CXCL12-mediated effects as mimicked by CCL5-derived peptides such as [VREY]4. An improved variant of this peptide, i[VREY]4, binds to CXCL12 in a complex with CXCR4 on the surface of activated platelets, thereby inhibiting Btk activation and preventing platelet CXCL12-dependent arterial thrombosis. In contrast to standard antiplatelet therapies such as aspirin or P2Y12 inhibition, i[VREY]4 reduced CXCL12-induced platelet aggregation and yet did not prolong in vitro bleeding time. We provide evidence that platelet-derived CXCL12 is involved in arterial thrombosis and can be specifically targeted by peptides that harbor potential therapeutic value against atherothrombosis.

Graphical abstract

Figure 1

Platelet-derived CXCL12 promotes arterial thrombosis. (A-B) Thrombus formation was induced by FeCl₃ in the carotid artery of ApoE^−/− mice (n = 22). The time to occlusion (A) was measured by using Doppler sonography, and thrombi were classified into “stable” and “unstable” (B) as specified in the Methods. (C) Isolated mouse blood was activated with collagen (10 µg/mL), and the concentration of CXCL12 from the releasate was determined by enzyme-linked immunosorbent assay (n = 3). (D) Tail bleeding time was assessed (n = 11). (E-I) Multiparameter analysis of thrombus formation in a collagen-coated flow chamber perfused with murine whole blood (1000 s⁻¹): platelet deposition (E), thrombus size (F), thrombus multilayer score (G), thrombus contraction score (H), and phosphatidylserine (PS) exposure (I) were assessed by Annexin V staining. (J) Representative micrographs (n = 11-15); note that Cxcl12^wtlwt mice form large and contracted thrombi, in which individual platelets are barely recognizable (closed arrow heads), whereas Cxcl12^Δplt/Δplt mice tend to generate smaller less contracted thrombi featuring clearly distinguishable individual platelets (open arrow head); scale bar overview, 50 µm; scale bar inlet, 10 µm. (K-L) Platelet activation by collagen (1, 5, and 10 µg/mL) was analyzed by upregulation of activated αIIbβ3 (K) and P-selectin (L) by flow cytometry (n = 6). Data represent mean ± standard deviation from the indicated numbers of independent experiments or mice. *P ≤ .05, **P ≤ .01, ***P ≤ .001, ****P ≤ .0001, as analyzed by using the Mann-Whitney U test (panels A and D), Fischer's exact test (panel B), and unpaired t tests (panels C, E-I, K, and L). MFI, mean fluorescence intensity; SAC, surface area coverage.

Figure 2

The CXCL12–CXCR4 axis functions as a positive feedback loop in human platelet activation. Platelet aggregation was assessed by MEA in human blood activated by collagen (0.2 μg/mL) (A) or human plaque homogenate (B). CXCR4 was inhibited by 100 nM AMD3465 (n = 5-8). (C) Thrombus formation was induced by perfusion (600 s⁻¹) of human blood, preincubated with PBS or 1 μM AMD3465, in a plaque-coated flow chamber and thrombus volume determined by confocal microscopy (n = 7). (D) CXCL12 was visualized in resting human platelets that were permeabilized and double-stained with antibodies against CXCL12 (purple) and CXCL4 antibody (green) by stimulated emission depletion (STED) microscopy (Leica SP8; scale bar, 2 μm). (E) CXCL12 release from isolated human platelets after activation with collagen (5 μg/mL) was determined by enzyme-linked immunosorbent assay (n = 3). (F-G) Platelet aggregation was assessed by MEA of human blood incubated with different concentrations of recombinant CXCL12 (n = 5-10) (F) or combinations of collagen (0.1 μg/mL), recombinant CXCL12 (0.1 μg/mL), and AMD3465 (100 nM) as indicated (n = 6-10) (G). Data represent mean ± standard deviation from the indicated numbers of independent experiments. *P ≤ .05, ***P ≤ .001, ****P ≤ .0001, as analyzed by paired t test (panels A and B), unpaired t test (panels C and E), and one-way analysis of variance with Tukey's multiple comparison test (panels F and G). AU, arbitrary units.

Figure 3

i[VREY]₄, a CCL5-mimicking peptide, binds to CXCL12. (A) Binding kinetics of CXCL12 to i[VREY]₄ by surface plasmon resonance. i[VREY]₄-biot was immobilized onto a neutravidin-conjugated C1 sensor chip (914 RU), and CXCL12 was injected at 62.5 ng/mL, 125 ng/mL, 250 ng/mL, 500 ng/mL, and 1000 ng/mL. Red traces represent the single-site fit to the raw data (blue). Kinetic parameters of 3 independent experiments are indicated as mean ± standard deviation. (B) Expansions of ¹⁵N HSQC spectra are shown for ¹⁵N-labeled CXCL12 in the absence (red peaks) and presence of i[VREY]₄ at concentrations of 20 µM (green peaks), 40 µM (purple peaks), and 200 µM (blue peaks). (C) The interaction of endogenous CXCL12 with i[VREY]₄-biot (1 µM) in human blood was quantified on platelets by proximity ligation with DuoLink by flow cytometry (C) and visualized by confocal microscopy on platelets (scale bar, 2 µm) (n = 3) (D). Data represent mean ± standard deviation from the indicated numbers of independent experiments. **P ≤ .01, ***P ≤ .001 as analyzed by one-way analysis of variance with Dunnett's multiple comparison test (panel C). MFI, mean fluorescence intensity.

Figure 4

i[VREY]₄ inhibits the prothrombotic activity of CXCL12. (A-D) The effects of i[VREY]₄ (5 µM) on platelet aggregation in human blood activated with collagen and recombinant CXCL12 (0.1 µg/mL; n = 15) (A), recombinant CXCL12 alone (0.1 µg/mL) (B), collagen alone (0.2 µg/mL; n = 5) (C), or homogenized human plaque (833 µg/mL, n = 4) (D) were measured by using MEA. (E) Thrombus formation was induced by perfusion of human blood through a plaque-coated flow chamber at 600 s⁻¹. Thrombus volume in the absence and presence of i[VREY]₄ (5 µM) was analyzed by confocal microscopy (n = 6-7). (F) Time to occlusion as in Figure 1A. i[VREY]₄ (100 µg, n = 10) or saline control (n = 9) was injected intraperitoneally 1 hour before induction of thrombosis. (G) Mouse blood from the indicated genotypes was mixed with 1 µg/mL collagen in the presence or absence of 5 µM i[VREY]₄, and platelet aggregation was measured by using MEA (n = 6-8). (H) i[VREY]₄-biot plasma levels were detected at various time points after intraperitoneal (i.p.) injection of 75 µg by enzyme-linked immunosorbent assay. (I) Neutrophil mobilization from the bone marrow of C57BL/6 mice was assessed 1 hour and 2 hours after i.p. injection of PBS with 100 μg i[VREY]₄ or 100 μg AMD3465 by using an automated blood counter (n = 3-7). Data represent mean ± standard deviation from the indicated numbers of independent experiments or mice. *P ≤ .05, **P ≤ .01, ***P ≤ .001, ****P ≤ .0001 as analyzed by using one-way analysis of variance with Tukey's multiple comparison test (panels A, G, and I), unpaired t test (panels B and E), paired t test (panels C and D), and Mann-Whitney U test (panel F). AU, arbitrary units; ns, not significant.

Figure 5

CXCL12-dependent platelet aggregation requires signaling through Btk. (A-B) Blood was pretreated for 30 minutes at 37°C with dimethyl sulfoxide (DMSO) (0.1% solvent control) or remibrutinib (0.1 µM) for Btk inhibition. Platelet aggregation was assessed by MEA after activation with collagen (0.1 µg/mL) and recombinant CXCL12 (0.1 µg/mL) or recombinant CXCL12 alone (1 µg/mL). (C) Phosphorylation of Btk (pBTK) in human platelets treated with CXCL12 (1 µg/mL) was analyzed by flow cytometry (n = 3). (D-F) PRP prepared from human blood was preincubated with DMSO (0.1%, solvent control) or remibrutinib (1 µM) for 30 minutes at 37°C before stimulation with CXCL12 (D), 2.5 µg/mL CRP-XL (E), or CXCL12 and collagen (n = 3) (F). Platelet aggregation was stopped after 1, 2, or 5 minutes by CGS buffer, and representative western blots patterns (upper panels D of and E) and quantification of Btk Y551 phosphorylation compared with total Btk (lower panels) are shown. (F) Phosphorylation of Y223 per total Btk after stimulation with CXCL12 (0.1-10 µg/mL) is shown in a representative immunoblot and densitometric quantification (lower panel) (n = 3). (G-H) Platelet activation was assessed by PAC-1 (activated αIIbβ3) and P-selectin antibody staining with and without Btk inhibition (0.1 µM remibrutinib) before stimulation with indicated combinations of recombinant CXCL12 (0.1 μg/mL) and CRP-XL (0.01 μg/mL). The samples were analyzed by using flow cytometry (n = 6). Platelet aggregation was assessed by MEA after activation with collagen (0.1 µg/mL) and CXCL12 (0.1 µg/mL) or CXCL12 alone (1 µg/mL). Data are represented as mean ± SD. Significant differences between different treatment groups are marked with asterisks *P ≤ .05, **P ≤ .01, ***P ≤ 0.001, ****P ≤ .0001, while differences between time points (D-E) within the same group are marked with hashtags ####P ≤ .0001 as analyzed by paired (A-B) or unpaired (C), t test and two-way analysis of variance (ANOVA) with Dunnett's multiple comparison test (D,E,G,H). *DMSO + CRP-XL vs remibrutinib + CRP-XL of each time point and # at each time point vs time point 0. AU, arbitrary units; MFI, mean fluorescence intensity; ns, not significant.

Figure 6

i[VREY]₄ blocks CXCL12-induced phosphorylation of Btk (pBTK) CXCR4 dependently without affecting CXCR4 internalization. (A) pBTK in human platelets was analyzed by using flow cytometry. Platelets were treated with CXCL12 (1 µg/mL) and as indicated with i[VREY]₄ (n = 3). (B) Changes in CXCR4 expression on human platelets was analyzed by flow cytometry after treatment with recombinant CXCL12 (0.1 µg/mL), collagen (1 µg/mL), and i[VREY]₄ (5 µM). (C-F) CXCL12 on human platelets was detected by flow cytometry. Human blood was treated with CXCL12 (0.1 µg/mL) (C,E) or collagen (1 µg/mL) (D,F), and detection was conducted with directly conjugated monoclonal antibodies (panels C and D, clone K15C, n = 10; panels E and F, clone 79018, n = 8). Binding of i[VREY]₄-biot to platelets from tamoxifen-injected CreErt^wt/wt Cxcr4^flox/flox (wild type [WT]) or CreErt^tg/wt Cxcr4^flox/flox (CXCR4 knockout [KO]) mice was measured by flow cytometry under resting conditions or stimulated with 10 µg/mL collagen (G) (n = 3-5). Data represent mean ± standard deviation from the indicated numbers of independent experiments or mice. *P ≤ .05, **P ≤ .01, as analyzed by one-way analysis of variance with Tukey's multiple comparison test (panels A-F) or unpaired t test (panel G). MFI, mean fluorescence intensity; ns, not significant.

Figure 7

i[VREY]₄ improves the inhibitory effect of standard antiplatelet therapy without increasing the risk of bleeding. (A) The effect of aspirin (300 µg/mL), cangrelor (0.34 µg/mL), and i[VREY]₄ (5 µM) alone or in combination on collagen/epinephrine closure time was measured with the platelet function analyzer-200 device (n = 5). (B-G) Platelet aggregation was assessed by MEA in human blood activated with collagen (0.1 µg/mL) and recombinant CXCL12 (0.1 µg/mL), CXCL12 alone (1 µg/mL), or human plaque homogenate (833 µg/mL). The blood was pretreated for 1 hour either with dimethyl sulfoxide as a control, aspirin (300 µg/mL) alone, or in combination with i[VREY]₄ (5 µM) or cangrelor (0.34 µg/mL) alone or in combination with i[VREY]₄ (n = 8). Data represent mean ± standard deviation from the indicated numbers of independent experiments. *P ≤ .05, **P ≤ .01, ***P ≤ .001, ****P ≤ .0001 as analyzed by repeated measure one-way analysis of variance with Tukey's multiple comparison test. AU, arbitrary units; ns, not significant.