Megakaryocytes Internalize and Are Activated by Immune Complexes

Abstract

BACKGROUND:

Receptors that bind antibodies are essential for protective adaptive immune responses against antibody-opsonized pathogens, yet their engagement by antibody-autoantigen complexes can drive chronic inflammation in autoimmune diseases. Megakaryocytes, the precursor cells of platelets, express such receptors. However, their response to immunoglobulin G antibodies remains unclear.

METHODS:

We used both systemic lupus erythematosus and COVID-19 as relevant examples of autoimmune and infection-driven contexts in which antibodies are involved to characterize human and mouse megakaryocyte responses.

RESULTS:

We found that megakaryocytes internalized immune complexes composed of autoantigens or SARS-CoV-2. In both human and mouse megakaryocytes, immune complexes triggered the release of chemokines and procoagulant extracellular vesicles. This process required FcγRIIA (Fc gamma receptor IIA) engagement, downstream Syk (spleen tyrosine kinase) signaling, and protein translation. A detailed analysis revealed that megakaryocyte-derived extracellular vesicles did not contain organelles and were largely indistinguishable from a subset of small-sized extracellular vesicles released by activated platelets. In FcγRIIA-transgenic mice, we analyzed megakaryocytes in both the bone marrow and lungs in a lupus model, whereas megakaryocytes were examined in the lungs in a COVID-19 model. In all cases, immunoglobulins were detected in close proximity to FcγRIIA-expressing megakaryocytes. Notably, the chemokine CXCL2 (C-X-C motif ligand-2) was increased in FcγRIIA-expressing mice under disease conditions. Tissue spatial analysis revealed that CXCL2 predominantly localized to megakaryocytes, supporting these cells as a major source. Furthermore, SARS-CoV-2 stimulated megakaryocytes to release CXCL2 only in the presence of IgG from SARS-CoV-2 immune individuals, and this response was strictly dependent on FcγRIIA expression.

CONCLUSIONS:

These findings suggest that megakaryocytes contribute to adaptive immune responses through FcγRIIA-mediated signaling.

Novelty and Significance.

What Is Known?

• Immune complexes are molecular clusters formed when antibodies bind to targets, such as viruses, during infections or self-antigens in autoimmune diseases.

• Immune complexes activate immune cells, which contributes to pathogen clearance but can also promote chronic inflammation.

• Megakaryocytes, the precursor cells of platelets, are increasingly recognized as active contributors to immune responses, but their response to immune complexes is unknown.

What New Information Does This Article Contribute?

• Megakaryocytes can directly recognize and internalize immune complexes, including those formed by virus and autoantigens, through a specific receptor called FcgRIIA.

• FcgRIIA activation in megakaryocytes promotes protein translation, which is under the control of Syk (spleen tyrosine kinase) and downregulated by G6b-B.

• FcgRIIA-activated megakaryocytes produce procoagulant extracellular vesicles and inflammatory molecules (chemokines), which may respectively promote thrombosis and activate other immune cells.

This work redefines megakaryocytes as active immune-sensing cells, not just platelet precursors. These findings provide a new understanding of how antibody-driven responses may contribute to thrombo-inflammation during adaptive immunity in conditions such as viral infections and autoimmunity.

Meet the First Author, see p e000762

Editorial, see Article by Tun and Morrell

Megakaryocytes are large polyploid cells (≈20–100 µm) primarily known for their role in platelet production.^1,2 In addition to producing platelets, megakaryocytes continuously release extracellular vesicles (EVs), which are abundant in circulation, although their exact function remains unclear.^1,3

Megakaryocytes are best known for their importance in hemostasis, but transcriptomic analyses suggest that they have a previously unstudied diversity and that they may also contribute to immunity.⁴ As part of their proposed role in host defense, they can generate cytokines and process proteins for antigen presentation through the MHC (major histocompatibility classes)-1 and MHC-2.^1,5,6 Megakaryocytes express TLRs (Toll-like receptors) that mediate immune responses to pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns.⁴ Furthermore, antigen-antibody molecular scaffolds, called immune complexes, participate in opsonization, phagocytosis, and cellular activation through binding to the family of FcγR (Fcγ receptors).⁷ FcγRIIA is expressed by megakaryocytes and platelets in humans, but not in mice, which lack the gene.^8,9 FcγRIIA promotes platelet activation by immune complexes in heparin-induced thrombocytopenia, and in infectious and autoimmune diseases, such as sepsis and systemic lupus erythematosus (SLE).^10–12 However, it is unknown whether FcγRIIA binding to immune complexes regulates megakaryocyte functions.

A unifying feature of both infectious and autoimmune diseases is the capacity of IgG to drive immunothrombotic responses.¹³ In COVID-19, increased levels of platelet-derived EVs and enhanced thrombosis have been widely reported,^14–20 alongside elevated inflammatory cytokines in the lungs of patients with severe disease, with chemokines among the most abundant mediators in bronchoalveolar lavages.²¹ Most individuals now harbor antispike IgG due to vaccination and recurrent viral exposure. In SLE, there is overproduction of antibodies that target extracellular self-antigens, forming immune complexes that play a pathogenic role through activation of FcγRs.²² Platelet activation is also a hallmark of SLE, with increased circulating platelet-derived EVs potentially contributing to the elevated cardiovascular risk, while FcγRIIA expression aggravates disease in lupus-prone mice.¹⁰

Utilizing both COVID-19 and SLE as relevant examples of infectious and autoimmune-driven contexts in which antibodies are involved, we characterized FcγRIIA-driven megakaryocyte activation.

Methods

Data Availability

All primary data and reagents can be made available on reasonable request to the corresponding author. Mouse strains must be shared in accordance with institutional material transfer agreement.

Figure schematics were created by Isabelle Allaeys (isabelle.allaeys@crchudequebec.ulaval.ca) in 2025 (https://BioRender.com) using the Biorender open access academic license of Department of Microbiologie-Infectiologie et Immunologie at Université Laval.

Ethics

Mice

Mice were housed and bred in a pathogen-free animal facility. Guidelines of the Canadian Council on Animal Care were followed in a protocol approved by the Animal Welfare Committee at Laval University (Projects: 2021-819, 2020-594).

Humans

Human Megakaryocytes

The study was approved by the responsible institutional review boards (local ethics committee of Umbria-Comitato Etico Regionale Umbria-approval CE-2775/25) and was performed in conformity with the Declaration of Helsinki.

Megakaryocyte Differentiation

The isolation of mouse megakaryocytes is described in the Supplemental Material and adapted from prior literature.²³ Human megakaryocytes were differentiated from CD34⁺ cells isolated from peripheral blood of healthy subjects.²⁴

SARS-CoV-2 Isolation

Acute respiratory syndrome coronavirus 2 (SARS-CoV-2; strain Laboratoire de Santé Publique du Québec, B1 lineage) was propagated in lung epithelial A549-hACE2 (human angiotensin-converting enzyme 2) cells.¹⁸

Analysis of EVs From Mouse Megakaryocytes and Platelets

Platelet isolation and stimulation were adapted from a prior publication,¹⁸ and detailed methods for EV quantifications are included in the Supplemental Material.

Quantification and Statistical Analysis

GraphPad Prism software (version 10) was used for statistical analyses. Student paired t test was used to compare data from 2 groups with normal distribution and pairing. Two-way ANOVA with Sidak or Dunnett multiple comparisons test and 1-way ANOVA with Dunnett or Tukey multiple comparisons test were used for multiple group data analysis. Data were tested for normal distribution by the Shapiro-Wilk test. Data that did not fit the assumption of normal distribution were compared using the nonparametric analysis indicated in the respective figure legends: Mann-Whitney U test for comparison of data from 2 groups, or Kruskal-Wallis test with Dunn multiple comparisons test for comparison of data from multiple groups. Data were tested for outliers by Robust regression and Outlier removal. Data are presented as mean±SD. Relevant significant P values (<0.05) are indicated in the figures. P>0.1 are considered nonsignificant. Primary analyses were performed post hoc.

Results

Bona Fide TLR Ligands Promote Release of Cytokines by Mouse Megakaryocytes

Antibodies can opsonize pathogens, thereby delivering PAMPs to endosomal TLRs. Likewise, in autoimmunity, autoantibodies bind damage-associated molecular patterns that are also capable of triggering TLR signaling. In both settings, immune complexes therefore engage TLRs and FcγRs. Before dissecting FcγRIIA-mediated signaling in megakaryocytes, we thus established whether TLR engagement alone is sufficient to induce megakaryocyte activation.

To characterize megakaryocyte response to PAMPs or SARS-CoV-2, we generated megakaryocytes by the differentiation of lineage-negative progenitor cells isolated from bone marrow of C57BL6/J mice (Figure 1A). After density gradient–based purification, cells were analyzed by flow cytometry. Cells displayed the expected megakaryocyte-specific surface CD (cluster of differentiation) 42b and CD41 (Figure 1B) and a high degree of polyploidy (>50% 16 N and 32 N), and had a diameter of ≈50 µm as determined by light-microscopy imaging (Figure 1A and 1B).

Figure 1.

Megakaryocytes (MKs) release cytokines in response to TLR (Toll-like receptor) ligands, but not to SARS-CoV-2 exposure. A, Schematic representation of mouse MK differentiation. Representative brightfield image of mature MKs, scale bar 100 µm. B, MK maturation markers (left) and ploidy (right) at day 4. C, Heat map of cytokine/chemokine expression in MK supernatant after stimulation. CONT: control vehicle-only, TLR ligands, SARS-CoV-2 (CoV-2) or conditioned Media (CMed). D and E, CXCL2 expression in MK supernatants after stimulation with TLR ligands (TLR7, TLR9; D) or heat-killed bacteria (E): Escherichia coli (10⁷·mL⁻¹), Listeria monocytogenes (10⁸·mL⁻¹), Mycobacterium tuberculosis (10 µg·mL⁻¹), Pseudomonas aeruginosa (10⁷·mL⁻¹), Staphylococcus aureus (10⁸·mL⁻¹), and Salmonella typhimurium (10⁸·mL⁻¹). C, N=4 to 6, 2-way ANOVA with Dunnett multiple comparisons test. D, N=5 to 9, Kruskal-Wallis test. E, N=3, 1-way ANOVA with Dunnett multiple comparisons test. CCL indicates chemokine C‑C motif ligand; D4, day 4; G-CSF, granulocyte colony‑stimulating factor; GM-CSF, granulocyte‑macrophage colony‑stimulating factor; HMW, high molecular weight; IFN, interferon; IL, interleukin; KC, keratinocyte‑derived chemokine; LIF, leukemia inhibitory factor; Lin, lineage; LMW, low molecular weight; M-CSF, macrophage colony‑stimulating factor; ns, nonsignificant; TNF, tumor necrosis factor; TPO, thrombopoietin; and VEGF, vascular endothelial growth factor. Created in BioRender. Allaeys, I. (2026) https://BioRender.com/rjxrue2.

We incubated isolated megakaryocytes with ligands of TLR4 (lipopolysaccharide), TLR2 (Pam3CSK4), and TLR3 (Poly(I:C) high or low molecular weight, and quantified 32 cytokines in supernatants, including chemotactic cytokines of the CC and CXC family collectively referred to as chemokines. Although TLR4 and TLR2 ligands were proficient at inducing multiple inflammatory mediators (G-CSF [granulocyte colony‑stimulating factor], IL [interleukin]-6, CXCL1, CCL3 [chemokine C‑C motif ligand 3], CCL4) in addition to CXCL2, which was the most abundant mediator, TLR3 ligands failed to do so at any of the concentrations tested (Figure 1C; Figure S1A). CXCL2 was selected for the following analysis because it was the predominant chemokine expressed by megakaryocytes and thus provided the greatest sensitivity. Both TLR7 (Imiquimod) and TLR9 (CpG ODN2395) ligands also induced CXCL2 release (Figure 1D). This suggests that endosomal TLR7/9 and the surface TLRs generally involved in recognition of bacterial PAMPs may be prone to activate signaling in megakaryocytes.

Depending on the cellular lineage or the type of endotoxin involved, TLR4 activation by endotoxins such as lipopolysaccharide can require CD14 expressed on the cell surface or in serum, and MD2 (Myeloid Differentiation protein‑2) present in serum.²⁵ To determine whether these molecules are required for megakaryocytes to recognize bacterial PAMPs, megakaryocytes were exposed to lipopolysaccharide found in Gram-negative bacteria and the TLR2 ligand lipoteichoic acid found in Gram-positive bacteria in the presence or absence of serum. Both TLR ligands induced CXCL2 in the absence of serum, but the response to lipopolysaccharide was further enhanced by serum, pointing to a role for MD2 or CD14 in presentation of lipopolysaccharide to TLR4 in megakaryocytes (Figure S1B). The strong response mediated by TLR2 and TLR4 suggests that megakaryocytes are highly responsive to bacterial stimuli. To confirm this, we incubated megakaryocytes with heat-inactivated Escherichia coli O111:B4, Listeria monocytogenes, Mycobacterium tuberculosis, Pseudomonas aeruginosa, Staphylococcus aureus, and Salmonella typhimurium, and quantified CXCL2 as a marker of megakaryocyte activation. CXCL2 was abundantly released in response to all tested bacteria except S. typhimurium (Figure 1E), confirming that TLRs in megakaryocytes are functional.

To determine whether SARS-CoV-2 activates megakaryocytes, we exposed them to SARS-CoV-2. Because viruses must propagate by infecting cells, we also used a negative control conditioned medium obtained from noninfected lung epithelial cells, as described elsewhere,^18,26 and added it to megakaryocytes in the same proportion. Neither SARS-CoV-2 nor the negative control induced cytokines across the 32 assessed (Figure 1C), suggesting that, in contrast to bacteria, which strongly activate megakaryocytes, immune response to SARS-CoV-2 alone, if present, is not strongly mediated by TLRs.

Murine Megakaryocytes Are Activated by Immune Complexes Through FcγRIIA–Immunoreceptor Tyrosine-Based Activation Motif Signaling

Whether FcγRIIA can mediate megakaryocyte activation is unknown. Because the FcγRIIA receptor is absent in mice,²⁷ we compared megakaryocytes differentiated from mice expressing the FcγRIIA transgene (FcγRIIA^tgn)²⁷ to megakaryocytes from wild-type mice (FcγRIIA^null). Flow cytometry showed that megakaryocytes differentiated from FcγRIIA^tgn mice, but not FcγRIIA^null mice, expressed FcγRIIA in addition to CD41 (Figure S2), thereby better mimicking human megakaryocytes.

We utilized heat-aggregated IgG (HA-IgG) as an immune complex surrogate, as it allows the precise repercussion of FcγR signaling, without the risk of using an antigen that may activate other receptors such as TLRs, which we have shown to be sensitive to activation (Figure 1C). HA-IgG are 160±7 nm in diameter^12,28,29 and were confirmed free of detectable endotoxin (<0.1 Endotoxin Units·mL⁻¹). FcγRIIA^tgn and FcγRIIA^null megakaryocytes were incubated with HA-IgG, and cytokines and chemokines were measured in cell culture supernatants (Figure 2A). Out of the 32 molecules tested, only CXCL2 was significantly induced by the stimulation of FcγRIIA.

Figure 2.

Megakaryocytes (MKs) recognize immune complexes in an FcγRIIA-dependent manner. A, Heatmap of cytokine/chemokine expression in MK supernatants after heat-aggregated IgG (HA-IgG) stimulation. N=4. Two-way ANOVA with Sidak multiple comparisons test. B, CXCL2 expression measured in FcγRIIA^tgn MK supernatants incubated with or without stimuli. N=3. One-way ANOVA with Dunnett multiple comparisons test. C, CXCL4 expression in FcγRIIA^tgn MK supernatants from cells incubated or not with deposited HA-IgG, lipopolysaccharide (LPS) or thrombin. N=4. One-way ANOVA with Dunnett multiple comparisons test. D, CXCL2 release by MKs on deposited HA-IgG stimulation in the presence of translation inhibitor (cycloheximide [CHX]) or Syk (spleen tyrosine kinase) inhibitor (R406). N=5. Two-way ANOVA, uncorrected Fisher’s LSD (Least Significant Difference). E, CXCL2 measured in MK supernatants from cells with different genetic background incubated with or without deposited HA-IgG. N=4 (FcγRIIA^tgn;β3^−/−), N=5 (FcγRIIA^tgn; ALOX12 [12-lipoxygenase]^−/−), N=7 (FcγRIIA^null and FcγRIIA^tgn; Mpig6b)^−/− ) N=9 (FcγRIIA^tgn;Treml1^−/−), and N=19 (FcγRIIA^tgn). Two-way ANOVA with Sidak multiple comparisons test. F, Representative images of HA-IgG internalization by MKs. Cells were incubated with HA-IgG (yellow), labeled with CellTracker (purple), then PERM (permeabilized) or not, and fixed. Nuclei (Hoechst, blue), scale bars 20 µm. G, The graph shows the mean fluorescence intensity (MFI) of IgG per MK. N=13 (FcγRIIA^null_noPERM), N=18 (FcγRIIA^tgn_PERM), N=19 (FcγRIIA^null_PERM) and N=20 (FcγRIIA^tgn_noPERM). Mixed-effect model (REML [Restricted Maximum Likehood]) with Sidak multiple comparisons test. CCL indicates chemokine C-C motif ligand; CONT, control vehicle-only; CXCL, C-X-C motif ligand; GM-CSF, granulocyte-macrophage colony-stimulating factor; IL, interleukin; LIF, leukemia inhibitory factor; M-CSF, macrophage colony-stimulating factor; Mpig6b, megakaryocyte and platelet inhibitory receptor G6b; ns, nonsignificant; TNF, tumor necrosis factor; and VEGF, vascular endothelial growth factor.

Immune complexes may engage receptors as soluble entities or deposit on blood vessel walls and tissues under conditions of persistent antigen availability, sustained antibody production, or defective immune complex clearance. To test the relative potency of immobilized and soluble immune complexes, FcγRIIA^tgn megakaryocytes were incubated in wells coated with HA-IgG or in the presence of soluble HA-IgG (Figure 2B). Although megakaryocytes produced CXCL2 under both conditions, deposited immune complexes induced a higher response than the soluble ones, which may be consistent with the reported higher valency needed to activate FcγRIIA in platelets³⁰ (Figure 2B). In contrast, exposure of megakaryocytes to deposited fibrinogen or collagen did not induce CXCL2 release, and no synergistic effect was observed when fibrinogen or collagen was combined with soluble HA-IgG (Figure 2B). This was not due to a failure of the megakaryocytes to bind these molecules, as light-microscopy confirmed that megakaryocytes adhered similarly to deposited HA-IgG, fibrinogen, and collagen (Figure S3A). Moreover, both FcγRIIA and TLR4 activations promoted CXCL2 with similar potency, whereas thrombin failed to induce any CXCL2 from megakaryocytes (Figure S3B). Like CXCL2, release of CXCL4 (PF4 [platelet factor 4] thereafter) was also induced by FcγRIIA activation in these conditions (Figure 2C; Figure S3B). In contrast, mouse platelets efficiently released PF4 in response to both immune complexes and thrombin (Figure S3C), but platelets did not release CXCL2 under any of the tested conditions (Figure S3C). Inhibition of translation in megakaryocytes completely blocked CXCL2 release, and reduced that of PF4 to its constitutive levels (Figure 2D; Figure S3D), indicating chemokine neosynthesis in response to immune stimulation in megakaryocytes.

Studies of platelets suggest that integrin αIIbβ3 may synergize with FcγRIIA to enhance its signaling.³¹ Furthermore, the ALOX12 (12-lipoxygenase) is an enzyme acting downstream of FcγRIIA that regulates platelet activation and thrombosis.^13,32 To assess the regulatory role of these proteins in megakaryocytes, we crossed the FcγRIIA^tgn mice with mice lacking β3 (FcγRIIA^tgn::β3^−/−) or ALOX12 (FcγRIIA^tgn::ALOX12^−/−). Megakaryocytes differentiated from these mice showed intact FcγRIIA stimulation (Figure 2E), suggesting that activation in these cells occurs independently of β3 and ALOX12.

FcγRIIA contains an immunoreceptor tyrosine-based activation motif in its cytoplasmic domain.⁷ The pharmacological inhibition of Syk (spleen tyrosine kinase) blocked CXCL2 and PF4 release (Figure 2D; Figure S3D), validating the involvement of Syk in FcγRIIA-mediated immunoreceptor tyrosine-based activation motif signaling in megakaryocytes.³³ G6b-B (Megakaryocyte and platelet inhibitory receptor G6b- isoform B) and TLT-1 (Triggering receptor expressed on myeloid cells-like transcript-1) are immunoreceptor tyrosine-based inhibition motif–containing receptors with demonstrated regulatory activity in platelet activation.^34,35 To assess the role of these immunoreceptor tyrosine-based inhibition motif–containing receptors in megakaryocyte response to antibodies, we crossed FcγRIIA^tgn mice with Mpig6b^−/− or Treml1^−/− mice, lacking G6b-B and TLT-1, respectively, and examined CXCL2 release in response to FcγRIIA activation. We found that the absence of G6b-B, but not TLT-1, enhanced megakaryocyte response (Figure 2E), suggesting that G6b-B negatively regulates FcγRIIA activity in megakaryocytes.

Murine Megakaryocytes Internalize Immune Complexes

We used confocal microscopy to compare the ability of FcγRIIA^tgn and FcγRIIA^null megakaryocytes to internalize HA-IgG. Isolated megakaryocytes were incubated with HA-IgG, and the uptake of HA-IgG was monitored by measuring the mean intensity of fluorescently conjugated anti-IgG used to label the HA-IgG in intact or permeabilized megakaryocytes. A high signal intensity was observed in permeabilized FcγRIIA^tgn cells, but not in the nonpermeabilized FcγRIIA^tgn cells (Figure 2F and 2G), suggesting that the vast majority of the immune complexes were internalized by megakaryocytes and were thus almost absent from the surface. Furthermore, only basal levels of fluorescence were detected in all FcγRIIA^null megakaryocytes, confirming the primary role of FcγRIIA in both the interaction with and internalization of immune complexes (Figure 2F and 2G).

Release of Procoagulant EVs by Mouse Megakaryocytes Is Increased by FcγRIIA, but Not TLR Activation

Studies suggest that the majority of EVs in blood are constitutively released from megakaryocytes.^3,36 The stimulation of FcγRIIA on platelets promotes the release of EVs, some of which contain mitochondria or proteasomes.^37–39 It is, however, unknown whether immune-relevant stimuli would also promote EV release from megakaryocytes.

EVs from megakaryocytes, identified using the CD41 integrin, were quantified by flow cytometry in the supernatants of megakaryocytes. To validate the quantitative reliability of the experimental results, supernatants from cultures with varying concentrations of megakaryocytes were used. Megakaryocytes released impressive amounts (≈1×10³ CD41+ EVs per megakaryocyte (Figure 3A through 3C). Quantification of CD41+ EVs by flow cytometry or nanoparticle tracking analysis yielded similar results (Figure S5). These CD41+ EVs were sensitive to detergent (Figure 3A), confirming that actual EVs with a membrane moiety were detected. EVs were labeled with annexin V, which binds exposed phosphatidylserine in a calcium-dependent manner, and this binding was abrogated by calcium chelation with EDTA, further confirming the specificity of the flow cytometry analyses (Figure 3B).

Figure 3.

Megakaryocytes (MKs) constitutively release procoagulant extracellular vesicles (EVs) that can be increased by FcγRIIA but not TLR (Toll-like receptor) activation. A, Detection of CD41+ megakaryocyte-derived EVs (MKEVs; representative dot plot, left) in cell supernatant and their sensitivity to detergent (Triton, right). Data are expressed as percentage of control (no detergent). Apparent sizes of Megamix beads are illustrated in black. B, Phosphatidylserine (PS)–positive MKEVs were detected with annexin V in the presence or absence of EDTA (right). Data are expressed as percentage of control (no EDTA). N=4, Paired t test. C, MKEVs released per MK were quantified by flow cytometry (CD41+EV) using 3 different concentrations of MKs. N=4. D, Cryo-electron microscopy of MK supernatants. Heterogeneous singular MKEVs and larger aggregates range from 200 to 500 nm (scale bars 250 nm). Note the resolved lipid bilayer in the bottom right image. The bottom left image shows an elongated particle (red arrows). Top images represent singular and clustered MKEVs. Black dots (white arrows) correspond to 10 nm gold particles conjugated to annexin V, markers of PS-exposing MKEVs. White asterisks indicate thin threads of the carbon support film. E, MKEVs (CD41+, flow cytometry) in MK supernatants in the presence or absence of heat-killed bacteria (left) or SARS-CoV-2 (CoV-2) and conditioned Media (CMed; right). N=3. F, MKEVs in cell supernatants stimulated in the presence (orange) or absence (yellow) of deposited heat-aggregated IgG (HA-IgG). N=7. Two-way ANOVA with Sidak multiple comparisons test. G, Thrombin generation curve of MKEVs from nonstimulated MKs (n=4) or platelet-derived EVs (PEVs) obtained from HA-IgG stimulated FcγRIIA^tgn platelets (N=4; left). CD41+PS+EV concentration was determined by flow cytometry, and concentrations used in the assay were equalized. Negative controls (right) were medium-only, HA-IgG, and HBS buffer only. Medium spiked with phospholipids (PL) served as a positive control. H, Thrombin generation peak was measured using different concentrations of EVs (N=4). CD indicates cluster of differentiation; and CONT, control vehicle-only.

To qualitatively characterize EVs produced by megakaryocytes, we used cryo-electron microscopy. Megakaryocytes generated EVs with diameters that varied between 200 and 500 nm, as well as clusters of EVs, all of which were densely labeled with annexin V gold particles, confirming the exposed phosphatidylserine (Figure 3D). EVs contained some internal material and resembled platelet-derived EVs.⁴⁰ Some elongated particles, which looked almost empty except for some fibrous material, and were not labeled with annexin V gold particles, were also observed (Figure 3D).

To test whether immune stimuli would enhance EV release from megakaryocytes, EVs in supernatants were quantified after the incubation of megakaryocytes with bacterial lysates or SARS-CoV-2. In contrast to CXCL2, which was promptly induced by bacterial PAMPs (Figure 1E), EV release was unaffected by TLR stimulation or SARS-CoV-2 (Figure 3E). However, FcγRIIA activation was associated with EV release from megakaryocytes, as the incubation of megakaryocytes with immobilized HA-IgG promoted release only when megakaryocytes were differentiated from FcγRIIA^tgn mice, but not for megakaryocytes differentiated from FcγRIIA^null mice (Figure 3F). This release of EVs implicated both protein translation and Syk signaling (Figure S5B).

The expression of phosphatidylserine, a negatively charged phospholipid, on megakaryocyte-derived EVs suggested a procoagulant potential. Indeed, megakaryocyte-derived EVs supported thrombin generation to a similar degree as the positive control (media with spiked-in phospholipids), although they were not as potent as platelet-derived EVs (Figure 3G through 3H). Together, this indicated that megakaryocytes robustly release procoagulant EVs and that this can be enhanced by engaging FcγRIIA, but not TLRs.

Megakaryocyte-Derived EVs Are Largely Indistinguishable From the Small-Sized EVs Shed by Platelets

Different agonists may change how platelets express molecules in EVs.⁴¹ We thus assessed whether megakaryocyte-relevant stimuli would also affect surface expression of proteins and organelle content in EVs.

We obtained both megakaryocytes and platelets from FcγRIIA^tgn mice for comparison. Megakaryocytes and platelets were incubated in the presence or in the absence of HA-IgG, lipopolysaccharide, or thrombin. Prothrombin-deficient plasma was also included to provide MD2 or CD14 without enabling thrombin generation as a bias. We collected megakaryocyte and platelet supernatants and quantified EVs based on CD41 expression and size. Deposited HA-IgG was the only stimulus that enhanced release of megakaryocyte-derived EVs (MKEVs), and these EVs were relatively uniform in size (Figure 4A). In contrast, very few platelet-derived EVs (PEVs) were found in the absence of activation or using a TLR4 trigger (Figure 4B). HA-IgG or thrombin stimulation generated 2 distinct subsets of PEVs, classified as larger and smaller PEVs based on their size and internal content (Figure 4B).

Figure 4.

Megakaryocyte-derived extracellular vesicles (MKEVs) are largely indistinguishable from platelet-derived EVs (PEVs). Megakaryocyte (MK) supernatants were obtained from MK overnight stimulation with vehicle (CONT) deposited heat-aggregated IgG (HA-IgG), lipopolysaccharide (LPS), or thrombin. Platelets were incubated for 60 minutes under the same conditions. A and B, Quantification of CD41+extracellular vesicles (EVs) in MK (A) or platelet (B) supernatants. EVs per MK or per platelet are normalized to the number of cells or platelets at the beginning of the experiment. Representative dot plots of megakaryocyte-derived extracellular vesicles (MKEVs) obtained in control (left) or HA-IgG–stimulated cells (right). Apparent sizes of Megamix beads are illustrated in black. Small EVs, large PEVs and platelets are indicated. N=4. One-way ANOVA with Dunnett multiple comparisons test. C and D, Comparison of MKEVs (blue), small PEVs (orange), large PEVs (purple) and platelet (green) contents. C, Principal component analysis shows similarities between MKEVs and small PEVs. D, Heatmaps representing percentage of CD41+EVs or platelets that are positive for the analyzed protein, phospholipid, or organelle content in different conditions. N=4. Two-way ANOVA with Dunnett multiple comparisons test. Radar charts illustrate similarities between MKEVs (blue) and small PEVs (orange). CD indicates cluster of differentiation; CD62P, P-selectin; CLEC, C-type lectin domain containing; CONT, control vehicle-only; Dim, dimension; GPVI, glycoprotein VI; LAMP, lysosome-associated membrane protein; Mito, mitochondria; Prot, proteasome; and PS, phosphatidylserine.

We thoroughly characterized size, granularity, surface proteins, organelle (mitochondria and proteasomes)³⁹ content, and phosphatidylserine exposure by MKEVs, PEVs, and platelets using flow cytometry (Figure 4). In particular, we labeled EVs and platelets with antibodies detecting CD41, CD49b, glycoprotein VI (GPVI), CD9, CD62P, CLEC-2 (C‑type Lectin‑like receptor 2), LAMP1 (Lysosome‑Associated Membrane Protein 1), and activated αIIbβ3, as well as annexin V to detect phosphatidylserine, MitoTracker to detect mitochondrial content, and a fluorescent probe that binds active proteasomes (the gating strategy is described in Figures S6 through S8).³⁹

The different groups were compared by principal component analysis, which showed that small PEVs and MKEVs resembled each other and were distinct from larger PEVs and platelets, whatever the stimulus examined (Figure 4C and 4D). Larger PEVs contained organelles (proteasomes and mitochondria), but could be distinguished from platelets by size, a higher baseline level of CD62P, and LAMP1 expression. Moreover, unlike platelets, small PEVs lacked activated integrin αIIbβ3 expression under thrombin-stimulation conditions. MKEVs and small PEVs mostly expressed the same proteins, but MKEVs appeared to be mostly negative for CD62P and CLEC-2, while a minority of small PEVs expressed these surface proteins (Figure 4D). Thus, although larger CD41+ PEVs can be distinguished from MKEVs by the detection of CD62P, CLEC-2 and organelle content, MKEVs cannot be reliably distinguished from smaller PEVs, which are the most numerous population of EVs produced by activated platelets. This suggests that no current tools can accurately distinguish whether small EVs originate from platelets or megakaryocytes in vivo.

Human Megakaryocytes Internalize Immune Complexes and Release Both Chemokines and EVs

We generated human megakaryocytes by differentiating blood-derived stem cells to validate key findings obtained with murine megakaryocytes (Figure 5; Figure S9). Human megakaryocytes were incubated with deposited HA-IgG and included an FcγRIIA blocker in a subset of conditions. A panel of 49 cytokines was examined, revealing that activated human megakaryocytes primarily produce chemokines. Notably, CXCL8, the functional human analog of murine CXCL2,⁴² as both bind the CXCR2 receptor, was by far the most abundantly released chemokine, and its production required FcγRIIA (Figure 5A and 5B). Moreover, HA-IgG were efficiently internalized into human megakaryocytes (Figure 5C and 5D). Similar to mouse megakaryocytes, human megakaryocytes constitutively released MKEVs that lacked CD62P and were largely devoid of organelles. MKEV release increased on FcγRIIA activation and was reduced in the presence of a receptor blocker (Figure 5E through 5G; Figure S9).

Figure 5.

Human megakaryocytes (MKs) internalize immune complexes and release cytokines and extracellular vesicles in response to immune complexes. A, Heatmap of cytokine/growth factor and chemokine expression in human MK (hMK) supernatants incubated in the presence of vehicle (CONT), deposited heat-aggregated IgG (HA-IgG), coated COLL (collagen) or coated FIB (fibrinogen). N=3. Each molecule was analyzed individually with 1-way ANOVA (Friedman test, with Dunn multiple comparisons test). B, Heatmap representing log 2-fold changes in chemokine expression in response to HA-IgG in hMK supernatants in the presence of isotypic control (Iso) or FcyRIIA-blocking antibody (IV.3). N=3, Wilcoxon test was realized for each molecule. C and D, Human MKs internalize immune complexes. MKs were incubated for 1 hour with HA-IgG (yellow), labeled with anti-CD61 (purple), then permeabilized or not, and fixed. Nuclei were stained with Hoechst (blue), scale bars 20 µm. N=2. D, Mean fluorescence intensity (MFI) of IgG per MK. N=59 (noHA-IgG_noPERM), N=62 (HA-IgG_noPERM), N=65 (noHA-IgG_noPERM) and N=72 (HA-IgG_PERM). Mixed-effect model REML (Restricted Maximum likehood) with Sidak multiple comparisons test. E, Representative dot plots of human megakaryocyte-derived extracellular vesicles (MKEVs) obtained in supernatant from control cells (top left) or deposited HA-IgG–stimulated cells (top right). Apparent sizes (Side Scatter [SSC]) of Megamix beads are illustrated in black. CD41+MKEVs were quantified (bottom left) in cell supernatants incubated under the same conditions as A. N=3, 1-way ANOVA (Friedman test) with uncorrected Dunn multiple comparisons test. CD41+MKEV sensitivity to detergent (Triton, bottom right) is expressed as percentage of control (no detergent), n=3, paired t test. F, Heatmap shows the percentage of CD41+MKEVs positive for CD62P or organelles (proteasome [Prot], mitochondria [Mito]) under the same conditions as A. N=3. Two-way ANOVA with Dunnett multiple comparisons test. G, CD41+MKEVs in response to deposited HA-IgG in the presence of Iso or FcγRIIA-blocking antibody (IV.3). N=3. Two-way ANOVA with uncorrected Fisher multiple comparisons test. CCL indicates chemokine C-C motif ligand; CONT, control vehicle-only; CXCL, C-X-C motif ligand; EGF, epidermal growth factor; FGF, fibroblast growth factor; FLT-3L, Fms-like tyrosine kinase 3 ligand; GM-CSF, granulocyte-macrophage colony-stimulating factor; IFN, interferon; IL, interleukin; PDGF, platelet-derived growth factor; PERM, permeabilized; TNF, tumor necrosis factor; and VEGF, vascular endothelial growth factor.

Megakaryocytes Interact With Immune Complexes in SLE In Vivo

We investigated whether megakaryocytes may interact with immune complexes in rheumatic diseases. In SLE, neutrophils and platelets release mitochondria, which can be recognized by antimitochondrial antibodies to form immune complexes.^22,29,43 Using autoantibodies from individuals with SLE and isolated mitochondria (IgG titration is illustrated in Figure S4), we found that opsonized mitochondria were internalized exclusively in megakaryocytes that expressed FcγRIIA (Figure 6A and 6B), thus highlighting the potential pathophysiological relevance of FcγRIIA-mediated internalization of immune complexes in megakaryocytes.

Figure 6.

Megakaryocytes (MKs) interact with immune complexes in systemic lupus erythematosus (SLE) in vivo. A and B, MKs (FcγRIIA^null, N=56 or FcγRIIA^tgn, N=36) were incubated for 1 hour with Mitochondrial Immune Complexes (Mito-IC), subsequently permeabilized, fixed, and labeled for CD41 (purple), nuclear dye (Hoechst, blue), and IgG (yellow; A), scale bars 20 µm. B, Quantification of the mean fluorescent intensity (MFI) of IgG per MK. Mixed-effect model REML (Restricted Maximum likehood) with Sidak multiple comparisons test. C and D, Immunofluorescence staining of immune complexes (C) in lungs and bone marrow (BM) of SLE New Zealand black×New Zealand White first generation (NZB/WF1) mice expressing (NZB/WF1^tgn) or not (NZB/WF1^null) FcγRIIA: CD41 (MK, purple), IgG (yellow) and nuclei (blue). White arrows indicate punctate IgG in MK; scale bars 20 µm. Representative of N=4 to 6. D, Percentage of MKs containing immune complexes was evaluated in lungs (left, N=4) and BM (right, N=6). Mann-Whitney U test. E through H) Immunofluorescence staining of CXCL2 in lungs (E and F) or BM (G and H) of SLE NZB/WF1^null or NZB/WF1^tgn mice: CD41 (MK, purple), CXCL2 (yellow) and nuclei (blue), scale bars 20 µm, representative of N=4 to 6. F, CXCL2 expression in MKs (MFI) was quantified in lungs. N=153 MKs for NZB/WF1^null and N=137 NZB/WF1^tgn mice, Mann-Whitney U test. H, CXCL2 expression in MKs (MFI) was quantified in BM. N=274 MKs for NZB/WF1^null and N=291 MKs for NZB/WF1^tgn mice, Mann-Whitney U test. CD indicates cluster of differentiation; and CXCL, C-X-C motif ligand.

We used a mouse model of lupus in which the FcγRIIA transgene is expressed (New Zealand Black×New Zealand White first generation×FcγRIIA^tgn), at 28 to 32 weeks of age, a stage characterized by the presence of autoantibodies and established nephritis.¹⁰ The addition of FcγRIIA in these mice does not alter the circulating levels of autoantibodies, although it accelerates nephritis, thrombosis, and death.¹⁰ We examined megakaryocytes in the bone marrow and lungs, 2 anatomic sites where megakaryocytes are known to reside.^44,45 By using tissue immunofluorescence, we observed that megakaryocytes closely interact with IgG in these tissues, with IgG internalization further enhanced by the expression of FcγRIIA (Figure 6C and 6D). Immunofluorescence determined that CXCL2 protein was principally expressed by megakaryocytes in lungs and bone marrow in lupus-prone mice, with its expression further enhanced by FcγRIIA (Figure 6E through 6H). Taken together, these findings indicate that interactions between megakaryocytes and immune complexes occur in vivo in systemic autoimmune disease, driving megakaryocyte activation.

FcγRIIA Expression Amplifies Platelet Activation in COVID-19

Studies suggest that anti–SARS-CoV-2 antibodies, particularly afucosylated IgG, are more abundant in patients with severe COVID-19 and may enhance FcγRIIA-mediated cytokine responses.⁴⁶ However, whether FcγRIIA contributes to COVID-19 severity remains unknown and has never been studied in vivo in a COVID-19 model. To investigate the role of FcγRIIA, we crossed K18- hACE2 mice, which are susceptible to SARS-CoV-2 infection, with FcγRIIA^tgn mice to generate FcγRIIA^tgn::hACE2 double transgenic mice. For comparison, we used FcγRIIA^null::hACE2 mice.

Mice housed in a biosafety level 3 confinement were either intranasally injected with diluent or with SARS-CoV-2 Delta variant (250 TCID50 [Tissue Culture Infectious Dose 50%]) to induce COVID-19,^26,47 and were euthanized at 3, 5, and 7 DPI (days postinfection). Weight loss, conventionally used to predict lethality in mouse models, occurred similarly in both genotypes (Figure 7A). We assessed the presence of antibodies and viral particles. Infectious viral load in the lungs peaked at day 3 and steadily declined over time, regardless of genotype (Figure 7B). By day 7, antispike IgG and IgM were detectable in the blood of both genotypes (Figure 7C; Figure S13B), indicating that IgG-opsonized virus may have formed in these mice at this time point, although this is impossible to prove.

Figure 7.

FcγRIIA^tgn, but not FcγRIIA^null, lung megakaryocytes (MKs) persist in COVID-19, correlating with increased viral protein and IgG presence in infected lung tissues. A, Change in body weight compared with day 0. B through H, FcγRIIA^null (blue, circles) and FcγRIIA^tgn (yellow, triangles). B, Infectious viral load obtained from infected lungs. C and D, Detection of antispike IgG (C) and CD41+extracellular vesicles (EVs; D) in plasma. E, Representative images from noninfected and infected mice. Legend: nuclear staining (Hoechst, blue), CD41 (yellow), SARS-CoV-2 (purple), IgG (turquoise), scale bar 100 µm. Arrows indicate MKs (CD41 with nuclear signal), asterisks indicate SARS-CoV-2, and arrowheads indicate IgG. F through H, Quantification of SARS-CoV-2 (F), endogenous IgG (G), and MKs (H) in lungs. A through G, 2-way ANOVA with Sidak multiple comparisons test, N=4 to 6. CD indicates cluster of differentiation; DPI, days postinfection; MFI, mean fluorescence intensity; and NI, not infected.

To evaluate the potential impact of FcγRIIA and opsonized SARS-CoV-2, we monitored leukocytes and platelets in the blood (the gating strategy is shown in Figure S10). Leukocyte populations were similarly affected regardless of FcγRIIA activation. Circulating CD45+B220+B cells and CD45+CD3+T cells (CD4⁺ and CD8⁺) decreased over the course of infection (DPI 5 and DPI 7; Figure S11A and 11B) in both genotypes, likely due to their migration into the lungs (Figure S11C). Neutrophils and neutrophil-platelet aggregates (CD45+Ly6Ghigh+CD41+) increased in the blood by day 7 in both genotypes (Figure S11A and 11B), indicating that this process can occur independently of FcγRIIA.

The absence of phenotypic differences in leukocytes contrasted with observations made on platelets (Figure S11D). We monitored CD62P and activated αIIbβ3* expression by platelets, and found an increase in platelet activation in FcγRIIA-expressing mice (Figure S11E). Furthermore, CD41+ EVs, but not EVs from leukocytes, were significantly increased by day 7 in FcγRIIA-expressing mice compared with FcγRIIA^null mice (Figure 7D; Figure S12). Together, these results indicate that FcγRIIA^tgn expression in platelets or megakaryocytes is associated with a higher degree of platelet activation in COVID-19.

FcγRIIA-Expressing Megakaryocytes Coexist With Antibodies and Viral Particles in the Lungs During COVID-19

We examined the lungs of infected mice to assess the effects of FcγRIIA expression on the primary site of SARS-CoV-2 infection. Regardless of the genotype, viral protein levels were highest at 3 and 5 DPI (Figure 7E and 7F). Viral proteins, potentially representing noninfectious/dead virus particles, were undetectable by day 7 in FcγRIIA^null mice, but persisted in FcγRIIA^tgn mice (Figure 7E and 7F). We also quantified IgG deposition in the lungs. Notably, IgG levels began increasing by day 5 and reached statistical significance by day 7 postinfection in FcγRIIA^tgn mice, but not in FcγRIIA^null mice (Figure 7E and 7G).

Megakaryocytes were recognized phenotypically as nucleated CD41+ cells in the lungs (Figure 7E and 7H). We observed that megakaryocytes in FcγRIIA^tgn mice were present through the course of infection. The abundance of megakaryocytes in FcγRIIA^null mice was lower by DPI 7 compared with FcγRIIA^tgn mice (Figure 7H). Together, these findings suggest that SARS-CoV-2-derived particles may be opsonized by immunoglobulins in the lungs of FcγRIIA^tgn mice 7 days after infection, and that FcγRIIA-expressing megakaryocytes are still present at that time.

FcγRIIA Expression Amplifies Chemokine Expression by Megakaryocytes

We measured levels of multiple cytokines and chemokines in the lungs of noninfected and infected mice at DPI 3, 5, and 7. Strikingly, most of the detected mediators (G-CSF, IFNγ [interferon gamma], IL-6, LIF [leukemia inhibitory factor], TNF [tumor necrosis factor], chemokines: CCL11 [eotaxin], CXCL1 [KC], CCL2 [MCP-1 (monocyte chemoattractant protein 1)], CCL3 [MIP-1α], CXCL2 [MIP-2]) were significantly increased at day 7 in FcγRIIA^tgn lungs but not in FcγRIIA^null lungs (Figure 8A). CXCL2, the most abundant inflammatory mediator detected in the lungs, was observed exclusively at day 7 when anti–SARS-CoV-2 are present, was increased by FcγRIIA expression, and was specifically restricted to lung megakaryocytes, further supporting these cells as the principal source of this chemokine (Figure 8B and 8C). We also measured cytokine levels in plasma and observed that G-CSF, IL-6, KC, CCL2 and CCL3, but not CXCL2, were significantly increased in FcγRIIA^tgn mice compared with FcγRIIA^null mice at DPI 7 (Figure S13A). Together, the findings suggest that CXCL2 is locally produced by megakaryocytes in the lungs, rather than systemically in the circulation, during the adaptive immune response and its production depends on FcγRIIA.

Figure 8.

Megakaryocytes (MKs) can contribute to increased cytokine load in COVID-19 through FcγRIIA-dependent interaction with SARS-CoV-2 immune complexes. A, Heatmap of cytokine/chemokine expression in lungs of FcγRIIA^null or FcγRIIA^tgn mice at 3, 5, or 7 days postinfection (DPI) or without infection (NI). N=4 to 5. Cytokines were separated into low, medium (Mid), and high expression. P values indicated within the heatmap cells represent comparisons in a genotype (DPI 7-FcγRIIA^null compared with NI-FcγRIIA^null or DPI 7-FcγRIIA^tgn compared with NI-FcγRIIA^tgn). Kruskal-Wallis with Dunn multiple comparisons test. P values indicated on the right represent comparisons between genotypes at DPI 7 (DPI 7-FcγRIIA^tgn compared with DPI 7-FcγRIIA^null). Mixed-effect model REML (Restricted Maximum likehood) with Sidak multiple comparisons test. B, Representative images of CXCL2 immunofluorescence in lung MKs at DPI 7 in FcγRIIA^null and FcγRIIA^tgn mice: CXCL2 (yellow), CD41 (turquoise) and nuclear dye (Hoechst, blue); arrows indicate MKs in lung (scale bar 20 µm, representative of N=3 NI (Not Infected) and N=5 DPI 7). C, Violin plots representing the expression of CXCL2 (mean fluorescent intensity [MFI]) in lung MKs. N=50 (NI null); N=43 (NI tgn); N=92 (DPI 7 null); N=82 (DPI 7 tgn). Two-way ANOVA with Tukey multiple comparisons test. D through E, Representative images of MKs incubated for 1 hour with SARS-CoV-2 immune complexes (CoV-2–IC), permeabilized, fixed, and labeled for CD41 (purple), nuclear dye (Hoechst, blue), and IgG (yellow; scale bar 20 µm, representative of n=3). E, MFI quantification of IgG per MK. N=16 (FcγRIIA^null) and N=25 (FcγRIIA^tgn). Paired t test. F, CXCL2 measured in MK supernatants from cells incubated or not with CoV-2–IC; the dotted line indicates the level of CXCL2 released by MKs in the presence of SARS-CoV-2 incubated with non–CoV-2 IgG. N=3 to 5. Mixed-effect model REML with Uncorrected Fisher’s Least Significant Difference (LSD). CCL indicates chemokine C-C motif ligand; CD, cluster of different iation; CXCL, C-X-C motif ligand; G-CSF, granulocyte-colony stimulating factor; GM-CSF, granulocyte-macrophage colony-stimulating factor; LIF, leukemia inhibitory factor; M-CSF, macrophage colony-stimulating factor; and tgn, transgenic.

To determine whether SARS-CoV-2 induces megakaryocyte activation and CXCL2 release by an FcγRIIA-dependent mechanism, we purified anti–SARS-CoV-2 IgG from sera of individuals with high anti–SARS-CoV-2 neutralizing titers collected after vaccination. Individuals having no SARS-CoV-2 antibody neutralizing titers were used as negative controls. Efficient opsonization of SARS-CoV-2 by immunoglobulins was confirmed by titration (Figure S14). We incubated FcγRIIA^null and FcγRIIA^tgn megakaryocytes with opsonized SARS-CoV-2 and measured internalization by detecting IgG uptake using immunofluorescence microscopy (Figure 8D). We observed greater uptake of SARS-CoV-2 by FcγRIIA^tgn megakaryocytes compared with FcγRIIA^null megakaryocytes (Figure 8D), suggesting that FcγRIIA expression indeed promotes interaction with and uptake of opsonized SARS-CoV-2 by megakaryocytes. In these conditions, CXCL2 was induced by SARS-CoV-2, but it strictly required IgG from SARS-CoV-2 immune individuals and FcγRIIA (Figure 8E and 8F), validating that megakaryocytes are a source of CXCL2 induced by SARS-CoV-2 through an adaptive immune response.

Discussion

Our findings highlight the role of megakaryocytes as a source of inflammatory mediators, particularly chemokines, in response to TLR activation. By investigating FcγRIIA in megakaryocytes, we further demonstrated that this receptor promotes both internalization of immune complexes and chemokine translation and release, a process involving Syk and that is negatively regulated by the immunoreceptor tyrosine-based inhibition motif–containing receptor G6b-B. Both platelets and megakaryocytes released procoagulant EVs on FcγRIIA activation. EVs derived from megakaryocytes lack organelles, similar to the smallest EVs released by platelets.

The release of proinflammatory cytokines and chemokines by megakaryocytes, notably by the activation of surface TLR2 and TLR4, suggests that megakaryocytes may be effector cells in bacterial responses. Although TLR activation may occur in an infected bone marrow, such as in osteomyelitis, it may also involve the stimulation of these TLRs by damage-associated molecular patterns, such as S100A8/A9 or HMGB1 (high mobility group box 1) that can activate TLR4, or oxidized lipids that can activate TLR2.^48,49 Furthermore, TLR activation may occur in the lungs during a bacterial infection, or due to dysbiosis of the lung microbiota. As megakaryocytes are found in blood circulation and in greater numbers in the spleen during sepsis, we further suggest that bacterial PAMPs may promote release of cytokines/chemokines in sepsis.^50,51

The presence of IgG within platelet α-granules has been recognized for over 75 years.⁵² Platelets can also secrete IgG in response to thrombin stimulation, and platelet-derived IgG can neutralize viruses, including SARS-CoV-2.⁵³ These observations suggest that granule-associated IgG may be acquired directly from the circulation, or supplied to platelets by megakaryocytes. Consistent with this latter possibility, our findings demonstrate that megakaryocytes can internalize IgG, supporting a model in which megakaryocytes contribute to the IgG content of platelets.

Studies suggest that SARS-CoV-2 activates platelets through direct receptor interactions such as TLRs, αIIbβ3, or C-type lectin member 2.^54–56 In our study, SARS-CoV-2 failed to directly activate megakaryocytes, ruling out the contribution of TLR4 to cellular activation by this virus. Stimulation of endosomal TLRs 7 and 9, which recognize internalized nucleic acid, such as those present during viral infections or in autoantigens in SLE, led to CXCL2 release, demonstrating that TLRs in endosomes in megakaryocytes are functional. Thus, although HA-IgG (without the presence of viral genome or autoantigen) sufficed to promote megakaryocyte activation, our findings suggest that FcγRIIA-mediated internalization of opsonized PAMPS and damage-associated molecular patterns, or whole virus such as SARS-CoV-2 (Figure 8D), may deliver these ligands to endosomal compartments where TLRs reside, thereby further amplifying and diversifying megakaryocyte responses to immune complexes. This may lead to the production of less abundant yet highly potent mediators such as interferons.⁵⁷ Indeed, interferon-stimulated genes are induced in megakaryocytes in viral infections and lupus.^10,57–59

The use of transgenic mice expressing FcγRIIA to model human lupus revealed megakaryocytes surrounded by immune complexes in both the bone marrow and lungs. Interactions between megakaryocytes and immune complexes may alter expression of inflammatory mediators in these cells, which may recruit leukocytes to the inflamed tissue and contribute to explaining changes to platelet functions and systemic inflammation in SLE.⁶⁰ Alternatively, megakaryocytes may internalize immune complexes and thereby contribute to their clearance. While the regulatory mechanism of FcγRIIA activation has been better characterized in platelets, we found that its activation in megakaryocytes is under the control of Syk and is downregulated by G6b-B. Conversely, our data showed that TLT-1 does not regulate FcγRIIA in megakaryocytes, which may be consistent with its localization into α-granules and might require translocation to the surface to modify FcγRIIA signaling.⁶¹

We further studied the effect of FcγRIIA on the adaptive immune response by assessing anti–SARS-CoV-2 antibody onset using in vivo experiments that incorporated FcγRIIA into a double transgenic mouse model of SARS-CoV-2 infection, revealing the role of megakaryocytes in chemokine and EV release. As FcγRIIA promotes immune complex internalization, it may contribute to viral persistence, or instead favor viral clearance by alerting leukocytes through chemokines. These findings may extend to other viruses known to activate FcγRIIA, such as herpes simplex virus-1,¹² which can circulate in the blood, influenza H1N1,⁶² and flaviviruses including Dengue and Zika.⁶³

SARS-CoV-2 genome, or part of it, can be found inside platelets of individuals with COVID-19, but how exactly platelets capture SARS-CoV-2 is unclear.^19,58,64 Platelets may interact directly with SARS-CoV-2,⁶⁵ or viral particles may be taken up by megakaryocytes and then transmitted to daughter platelets.⁶⁶ Bone marrow megakaryocytes are less likely to interact directly with SARS-CoV-2 than lung megakaryocytes, although studies in a murine COVID-19 model show that bone marrow megakaryocytes exhibit an altered transcriptome despite the absence of the virus in this location, suggesting that inflammation can affect megakaryocytes.⁴⁷ Our findings suggest that FcγRIIA-mediated uptake of opsonized SARS-CoV-2 by megakaryocytes may induce antiviral transcriptional programs. These transcriptional changes may then be inherited by daughter platelets, potentially explaining the IFN signature observed in platelets from patients with COVID-19.^58,67 The high abundance of platelets in circulation, combined with their potent ability to capture circulating DNA and RNA,⁶⁸ may also explain how SARS-CoV-2 genetic material was detected in platelets.

We confirmed that megakaryocytes are profuse producers of procoagulant EVs. These results partially support findings from prior studies that used LAMP1, CD62P, and GPVI as markers to differentiate megakaryocyte-derived EVs from platelet-derived EVs and that suggested that the majority of so-called platelet-derived (CD41⁺) EVs in circulation actually originate from megakaryocytes rather than platelets.^3,36 However, while larger PEVs could be readily distinguished from MKEVs based on the above markers, the more abundant and smaller PEVs were nearly indistinguishable from MKEVs. This discrepancy with prior literature^3,36 may be attributed to advancements in current technologies, which allow for improved detection of the smallest EVs.

Organelles such as mitochondria and proteasomes were largely absent in MKEVs but were readily identified in the larger PEVs. It may be logical for megakaryocytes to package their organelles only into their daughter platelets and not disperse them into MKEVs, as platelets may require them for their functions.⁶⁹ Intriguingly, protein translation contributed to EV release, which may suggest that the proteome, beyond surface markers or organelles, may differ in FcγRIIA-induced MKEV. It would be interesting to determine how the organelle packaging principles in megakaryocytes differ between budding platelets and budding EVs.^2,70

Using FcγRIIA-expressing and human megakaryocytes, we identified murine CXCL2 and human CXCL8 as dominant mediators. These chemokines are also the most abundant in lungs in COVID-19 in both mice (Figure 8) and humans,²¹ and their release by SARS-CoV-2-exposed megakaryocytes⁷¹ points to chemotaxis as a conserved megakaryocyte function. Both human CXCL8 and murine CXCL2 bind the CXCR2 receptor, and are potent neutrophil attractants and important in inflammatory responses and immune defense in the lung.⁴² Neutrophils in the lungs of SARS-CoV-2–infected mice were abundant but were not significantly more elevated in the presence of FcγRIIA (Figure S11C), pointing to other mechanisms promoting neutrophil recruitment. The stimulation of CXCR2 also induces neutrophil egress from bone marrow⁹ or may promote thrombopoietin release.⁷² CXCR2 is involved in emperipolesis,^73,74 which inhibition reduces myelofibrosis.⁷⁵ Megakaryocytes may potentially be involved in manifestations affecting the lungs in SLE.⁷⁶

CXCL2 profiling in lungs and bone marrow unequivocally identified megakaryocytes as the main source in both COVID-19 and lupus models. However, determining the exact contribution of megakaryocyte-derived CXCL2 to disease remains challenging, primarily due to the genetic backgrounds of the mice (ACE2^tgn mice or NZB×NZWF1). An additional source of uncertainty was that the expression of human ACE2 in K18-hACE2 transgenic mice may not fully replicate its natural distribution in humans.⁷⁷ Despite these limitations, the model we used was ideal for our study, which investigated the role of innate versus adaptive immune receptors expressed by megakaryocytes.

As the immune response we revealed occurs later in the course of SARS-CoV-2 infection, when antibodies are present, it might represent immunity mechanisms that have become more prevalent at the time of writing, when the majority of the population has been immunized against SARS-CoV-2, and may also help explain symptoms of long-COVID-19. Furthermore, the incorporation of diverse PAMPs and immune complexes relevant to autoimmunity, such as those found in SLE, alongside opsonized viruses, extends our findings beyond COVID-19. Given that megakaryocytes have been described as antigen-presenting cells,^1,5,6 the internalization of antigen-containing immune complexes is likely to further modulate and expand their immunologic functions. Our findings provide compelling evidence that megakaryocytes are highly plastic cells capable of sensing and responding to opsonized targets through defined signaling and translational programs. Together, these data position megakaryocytes as active and functionally relevant participants in adaptive immunity.

Article Information

Acknowledgments

The authors are grateful to the technical assistance of Tania Lévesque and the Laboratoire de Santé Publique du Québec and the Public Health Agency of Canada’s National Microbiology Laboratory for providing the SARS-CoV-2 isolates used in this study.

Author Contributions

F. Puhm, L. Flamand, and E. Boilard conceived the work and designed the study. F. Puhm and I. Allaeys designed experiments and assays. F. Puhm, I. Allaeys, E. Petito, L. Bury, A. dos Santos Pereira Andrade, E. Doré, Y. Senis, F. Tupin, and M. Vaillancourt participated in the study design, performed experiments, and were involved in data extraction. A.R. Brisson performed electron microscopy analyses, interpreted the data, and contributed to writing the article. M. Dieudé, M.J. Hébert, and B.I. Florea contributed critical reagents, biospecimens, and instruments. P. Trépanier, A.S. Wolberg, and P.R. Fortin gave critical advice on experiments. F. Puhm and E. Boilard wrote the article. All authors read and approved the final article.

Disclosures

None.

Supplemental Material

Ethic Approvals

Supplemental Methods

Figures S1–S14

Major Resources Table

Reference 78

Supplementary Material

res-139-e327999-s001.pdf ^{(7.8MB, pdf)}

Funding Statement

F. Puhm and F. Tupin are recipients of postdoctoral fellowships from FRQS (Fonds de Recherche du Québec en Santé). E. Doré is recipient of doctoral fellowship from the Arthritis Society and the FRQS. E. Boilard is the recipient of a Merit Award from the FRQS. The work was supported by grants from the Canadian Institutes of Health Research (E. Boilard and L. Flamand).