Platelet-TLR7 mediates host survival and platelet count during viral infection in the absence of platelet-dependent thrombosis

Abstract

Viral infections have been associated with reduced platelet counts, the biological significance of which has remained elusive. Here, we show that infection with encephalomyocarditis virus (EMCV) rapidly reduces platelet count, and this response is attributed to platelet Toll-like receptor 7 (TLR7). Platelet-TLR7 stimulation mediates formation of large platelet-neutrophil aggregates, both in mouse and human blood. Intriguingly, this process results in internalization of platelet CD41-fragments by neutrophils, as assessed biochemically and visualized by microscopy, with no influence on platelet prothrombotic properties. The mechanism includes TLR7-mediated platelet granule release, translocation of P-selectin to the cell surface, and a consequent increase in platelet-neutrophil adhesion. Viral infection of platelet-depleted mice also led to increased mortality. Transfusion of wild-type, TLR7-expressing platelets into TLR7-deficient mice caused a drop in platelet count and increased survival post EMCV infection. Thus, this study identifies a new link between platelets and their response to single-stranded RNA viruses that involves activation of TLR7. Finally, platelet-TLR7 stimulation is independent of thrombosis and has implications to the host immune response and survival.

Figure 1

TLR7 is present in human and mouse platelets and TLR7 activation induces a reduction in platelet count. (A) TLR7 mRNA levels in human platelets from participants in the FHS (n = 1889, 839 men and 1050 women) screened by qPCR. (B) TLR7 protein levels in human platelets of 6 healthy donors (not involved in the FHS) screened by western-blot analysis (F, females; M, men). (C) TLR7 mRNA levels in murine platelets screened by qPCR (n = 5 males, n = 6 females). (D-G) WT (baseline platelet count 729 ± 167 × 10³ platelets/μL) and TLR7KO male mice (baseline platelet count 780 ± 147 × 10³ platelets/μL) at 14 weeks of age were injected (once) with TLR7 agonist, Loxo or at 17 weeks of age with EMCV. Percent of baseline platelet count in WT (D) and TLR7KO (E) mice injected with Loxo and WT (F) and TLR7KO (G) mice infected with EMCV. SEMs (H) and TEMs (I) of human platelets treated with EMCV. White circles in H encompass viral particles. Bar represents 2 μm in H and 0.2 μm in I. Data are average ± SD and were analyzed by Student t test using n = 5 animals/group (C-H).

Figure 2

Murine platelet-TLR7 mediates aggregation of platelets with leukocytes. Male mice were injected with Loxo. One hour postinjection, blood was drawn by cheek puncture, and aggregates between platelets (labeled with CD41) and leukocytes (labeled with Ly6G) were measured by flow cytometry. (A) Dot plots of each condition used with no gating. (B) Quantitation of CD41, Ly6G aggregates, gated around the granulocyte population. (C-D) Male mice were injected with EMCV as detailed in “Methods.” (C) Dot plots of each condition used with no gating. (D) Quantitation of CD41, Ly6G-positive aggregates, gated around the granulocyte population. Data are average ± SD and were analyzed by Student t test using n = 6 (3 males, 3 females)/group/condition (B) and n = 4 (D). (E) Isolated platelets or neutrophils (abbreviated as PMN) were labeled, pretreated, and then mixed for 15 minutes with the other untreated population. The Ly6G, CD41-positive population was quantified by flow cytometry n = 3 (2 mice per isolation were pooled for n = 1 in either genotype). KO denotes TLR7KO in E. Data are average ± SD and were analyzed by 1-way ANOVA (P < .0001) followed by Bonferroni test.

Figure 3

Human platelet-TLR7 mediates aggregation of platelets with granulocytes through endosomal signaling. Blood from human donors was treated with Loxo immediately after draw and stained with CD41 (platelets) and CD14 (leukocytes) antibodies. Aggregates between platelets and granulocytes were measured by flow cytometry. (A) Dot plots gated around the granulocyte population. FSC-H, forward scatter; SSC-H, side scatter. (B) Quantitation of CD41, CD14-positive aggregates from A. The following conditions were applied: R837 (2 μg/mL; n = 8); Loxo (1 mM; n = 5); Pam₃CSK₄ (10 μg/mL; males n = 10). (C) CD41, CD14 aggregates in human blood (n = 3) posttreatment with specific antagonist (IRS661, 2.8 μM) or the endosomal inhibitor chloroquine (Cl, 10 μM). (D-E) Platelets (denoted Plt) and neutrophils (denoted PMN) were isolated from human blood and labeled with CD41 and CD14, respectively. (D) Quantitation of CD41, CD14-positive aggregates in each fraction stimulated with either TLR7 agonist (Loxo) or pretreated with IRS661 or Cl (for 30 minutes) and then stimulated with Loxo for 15 minutes. The fractions were mixed together for 15 minutes, and aggregates were measured (n = 4) in each experiment. (E) SEM images of platelets and WBC stimulated together with different agonists. (F) SEM image of a WBC and a platelet with viral particles at the end of its pseudopodia (see circles). Bar in E and F represents 4 μm, except in the magnified image, where it is 2 μm. Data are average ± SD and were analyzed by Student t test (3B) or 1-way ANOVA (C, P < .05; E, P < .0001) followed by Bonferroni test (P values are noted on the graph).

Figure 4

TLR7 activation in platelets leads to Akt and p38-MAPK phosphorylation and translocation of P-selectin to the cell surface. (A-D) Platelets were isolated from humans and mice and treated with TLR7 agonist (Loxo) or pretreated (for 30 min) with the TRL7 antagonist, IRS661, and then stimulated with Loxo for the indicated time intervals. Protein was isolated and resolved by western-blot analysis. Phosphorylation (A) and quantitation (B) of kinases involved in α-granule release in human platelets. (C-D) Phosphorylation and quantitation, respectively, of kinases involved in α-granule release in murine platelets. (E) P-selectin surface expression and (F) quantitation in human platelets, post TLR7 agonist (Loxo) or thrombin (IIA) stimulation, resolved by flow cytometry. IC, isotype control. (G) Platelet P-selectin interacts with PSGL-1 on the granulocyte population measured by FlowSight image cytometer. (H-I) Five-week-old female WT and P-selectin (SELP) KO mice were injected with Loxo. (H) Heterotypic aggregates between CD41-platelets and Ly6G-PMNs and platelet count (I) 2 hours after Loxo injection. Data are average ± SD and were analyzed by Student t test, except the quantitation of the experiment with the inhibitors, analyzed by ANOVA (P < .0001). Analysis is based on n = 4 humans (A-B), n = 3 groups of mice (C-D), n = 4 for Loxo and n = 3 for thrombin (E), n = 4 (F), and n = 3 (G), n = 3-4 mice/group (H-I).

Figure 5

Rapid thrombocytopenia induced by TLR7 stimulation is caused by granulocyte (neutrophil) internalization of platelets and continued leukocyte aggregate formation. (A) Confocal microscopy of murine blood postinjection with Loxo. Mice were transfused with CFSE-labeled platelets; blood was drawn, fixed, and stained with 4,6 diamidino-2-phenylindole (DAPI). (B) Confocal microscopy of human blood prestained with CD41-FITC and CD45-CY5 (CD41-green stains are platelets and CD45-red stains are leukocytes) and stimulated with Loxo. DAPI (blue) stains the nucleus and was added postfixation. Cells are identified as neutrophils according to their lobularity and intensity of CD45. (C) Internalization coefficient measured by IDEAS-software (Amnis FlowSight Cytometry) in human blood. In these studies, CD45 was used to define the outside of the neutrophils, and CD41 is the internalization probe (n = 4). (D-E) Confocal images of human or murine blood showing large aggregates of platelets and WBCs. (D) Murine blood of mice transfused with CFSE-labeled platelets and injected with Loxo, fixed, and stained with DAPI. (E) Human blood treated ex vivo with Loxo (15 minutes) and stained with CD41 (green)-platelets, CD45 (red)-leukocytes, or DAPI (blue)-nucleus. In all cases, pictures are representative of: n = 3 (A), n = 5 (B); n = 4 (C), n = 3 (D), and n = 5 (E). Data in C are average ± SD and were analyzed by Student t test. Images in A-B and D-E were taken with spinning disk confocal microscope and Metamorph 7.4.2 software and merged by ImageJ (NIH) software. Pictures were taken with ×100 Plan Apo oil lens and the scale bar is 10 μm.

Figure 6

Platelets and platelet-TLR7 mediate infection-affected mortality. WT male mice (17 weeks) were depleted of platelets by injecting anti-GPIb-α antibody (2 μg/g of mouse), and at 12 hours postinjection they were infected with EMCV. (A) Survival of EMCV-infected, platelet-depleted mice (n = 13) compared with IgG-injected and infected controls (n = 12) evaluated by Mantel-Cox test. Noninfected anti-IgG (n = 4) or anti-Plt (n = 4) injected mice had a 100% survival. (B-F) Platelets from WT or TLR7KO mice (abbreviated KO here) were isolated, labeled with CFSE, and transfused in recipient mice (all mice had similar CFSE-labeled platelets posttransfusion assessed by flow cytometry) as labeled on the graphs. The control mice were transfused with buffer in which platelets were concentrated. Mice were injected with EMCV 2 hours posttransfusion. (B-C) Survival curves of the injected mice determined by Mantel-Cox test and based on: n = 5 for WT->KO and n = 6 for buffer->KO, and n = 6 for KO->WT and n = 5 for buffer->WT. (D) Reduction in platelet count in the transfused KO (see Figure 1F) mice as a result of platelet-TLR7 presence. (E) Platelet count in the WT mice as a result of KO platelet transfusion. (F) Heterotypic aggregates induced in the WT and the KO mice post buffer transfusion. Data for D-F are average ± SD and were analyzed by 2-tail Student t test, based on the following n: (D-E) n = 4 (WTplt->KO), n = 5 (buffer->KO), n = 5 (KOplt->WT), and n = 3 (buffer->WT) (F) WT groups: n = 3 (control), n = 3 (EMCV), n = 3 (control), and n = 5 (EMCV).

Figure 7

TLR7-activated platelets adhere to collagen but do not aggregate or form significant thrombi. Fluorescently labeled and stimulated, washed platelets were run over collagen-coated slides. (A) Human platelets treated with TLR7 ligand (TLR7-L) (R837, 2 μg/mL or 8.3 μM) or with thrombin (0.5 U/mL). (B) Quantitation of (A) for n = 6 individuals (3 females and 3 males). (C) Murine platelets treated with thrombin (0.5 U/mL) or TLR7-L: Loxo (100 μM) or R837 (2 μg/mL). (D) Quantitation of (C) using n = 3 (Loxo), n = 5 (R837), and n = 5 (thrombin). (E) SEM images of platelets treated with different agonists. (F) Aggregation of isolated human or (G) murine platelets treated with various concentrations of agonists depicted in the graph. Data are average ± SD and were analyzed by Student t test using n = 5 for human platelets (except Loxo, n = 3) and n = 3 for each condition in mice, as each n is pooled platelets from 4 to 5 mice. (H) PAC-1 binding to isolated human platelets post isotype control, control, Loxo (1 mM), and thrombin (0.005 U/mL) tested by flow cytometry. Histogram is representative of n = 4 individuals. PAC-1 tests for epitope on αII_bβ₃ integrin of human platelets, necessary for fibrinogen binding and aggregation. Bars in A and D represent 100 μm and in E represent 4 μm.