Structural and functional changes underlying activation of monocytes in heparin-induced thrombocytopenia

Abstract

Background:

Heparin-induced thrombocytopenia (HIT) is an antibody-mediated disorder associated with thrombosis developing in response to anticoagulation with heparin. Monocytes targeted by HIT antibodies contribute to the prothrombotic state, but structural and functional alterations of the activated monocytes have not been described.

Aim:

To study morphological and functional changes in monocytes caused by HIT antibodies interacting with membrane-associated PF4 in vitro.

Methods:

THP-1, isolated human or FcγRIIA-positive and FcγRIIA-negative mouse monocytes were incubated with recombinant human PF4 and/or anti-PF4/heparin antibodies followed by scanning electron microscopy and confocal microscopy.

Results:

Binding of PF4 to monocytes induced formation of “knobs” ~150 nm in size that protruded from the cell surface. Addition of pathogenic HIT-like monoclonal antibodies (KKO) caused profound remodeling of the cell membrane and time-dependent formation and clustering of KKO/PF4/glycosaminoglycan complexes into large “blebs” ranging in size from 500–1200 nm. Dynamic confocal microscopy revealed formation of monocyte-derived microvesicles in response to PF4 and KKO. In contrast, RTO, a monoclonal antibody that blocks PF4 oligomerization and prevents thrombocytopenia/thrombosis in an animal HIT model, inhibited PF4-induced modification of monocyte surfaces. Comparing monocytes from transgenic mice expressing hFcγRIIA to wildtype mice lacking FcγRIIA indicated that bleb formation results from clustering of knobs caused by bivalent HIT antibodies through crosslinking of FcγRIIA.

Conclusions:

Binding of pathogenic HIT antibodies to PF4-containing antigenic complexes assembled on the monocyte surface promotes large-scale plasma membrane remodeling as part of cell activation through the FcγRIIA receptors, resulting in the release of procoagulant microvesicles, which together may contribute to thrombosis in HIT.

Introduction

Heparin-induced thrombocytopenia (HIT) is a pathological condition in which thrombosis is associated with antibody-mediated thrombocytopenia that develops in a subset of patients treated with heparin. These antibodies recognize multimolecular complexes formed by specific interactions between endogenous platelet factor 4 (PF4), a positively charged protein secreted by activated platelets, and exogenously infused negatively charged heparin, or other polyanions, such as cell surface glycosaminoglycans (GAGs) [1, 2]. HIT antibodies cause platelet activation via FcγRIIA receptors with multiple functional consequences, including intracellular signaling via Syk, exposure of phosphatidylserine, expression of P-selectin, and release of procoagulant microvesicles [3–7] followed by platelet death via complex apoptotic and non-apoptotic pathways [8].

It has become increasingly evident that HIT antibodies activate not only platelets, but also monocytes [9–12], neutrophils [13–15] and endothelial cells [16, 17]. The activation mechanism implies binding of PF4 to glycosaminoglycans (GAGs) on the cell membrane, forming cell-surface PF4/GAG complexes targeted by HIT antibodies [18–20]. Monocytes are extremely sensitive to HIT antibodies because their glycocalyx contains a complex mixture of GAGs, including heparan sulfate and dermatan sulfate, with relatively high affinities for PF4, which is necessary to form stable antigenic complexes [21, 22]. Monocytes are present in HIT thrombi and play a major role in the pathogenesis of HIT-associated thrombosis [19]. Even in the absence of heparin, pathogenic human HIT-associated and monoclonal HIT-like antibodies (KKO) bind to PF4 complexed with GAGs on the surface of monocytes, initiating cell activation in vitro and thrombus formation in vivo [18, 19]. A procoagulant role of monocytes activated by HIT antibodies dependent on the presence of PF4, rather than heparin, is well established [9–11, 19]. HIT activated monocytes express tissue factor [10–12, 19, 23, 24], release procoagulant microvesicules [12] and promote formation of procoagulant coated platelets [9]. Microvesicles are submicron particles released from activated or apoptotic cells that express phosphatidylserine in their membranes, thus serving as a procoagulant surface. Microvesicles can also serve as a source of membrane or intracellular proteins that affect thrombosis through diverse mechanisms, transfer biological information to other blood cells, and alter clot structure and properties [25]. Their prothrombotic properties are enhanced in the presence of tissue factor [26].

Although the cell-associated PF4/GAG complexes are critically important targets for HIT antibodies, the mechanism of their formation and subsequent alterations of cell morphology and function have received little study. Here, we show that monocyte membrane-bound PF4/GAG structures are heterogeneous, dynamic, and form surface-attached ultralarge immune complexes with pathogenic HIT-like antibodies. These structures induce dramatic remodeling of the plasma membrane, contributing to cell activation and release of microvesicles, which promotes a prothrombotic cell surface.

Materials and Methods

Chemicals and reagents

The following items were purchased: Ficoll^®-Paque Premium (GE Healthcare, USA); anti-phycoerythrin (PE) MicroBeads (Miltenyi Biotec, Germany); LS Columns (Miltenyi Biotec, Germany); RosetteSep^™ Human Monocyte Enrichment Cocktail (StemCell, USA); EasySep Mouse Monocyte Isolation Kit (Stemcell, USA); bovine serum albumin fatty acid-free (Sigma-Aldrich, USA); heparin (Fresenius Kabi, USA); hexamethyldisilazane (Electron Microscopy Sciences, USA); PE-conjugated mouse anti-human CD41a antibodies (BD Pharmingen, USA); Vybrant^™ DiI or DiD Cell-Labeling Solution (Life Technologies, USA); dimethyl sulfoxide (DMSO), anhydrous (Life Technologies, USA); Hoechst 33342 (Life Technologies, USA); all other chemicals and reagents were analytical grade from Sigma-Aldrich, USA. Recombinant human PF4 (hPF4) cloned into the plasmid pMT/BiP/V5-His (Invitrogen, USA) was expressed using the Drosophila Expression System (Invitrogen, USA), purified, and characterized as described [27]. Total protein concentrations were determined using the bicinchoninic acid protein assay (Pierce, USA) with bovine serum albumin as the standard and then tested by dynamic light scattering to exclude self-aggregation and for the ability to form ultralarge complexes with heparin as described previously [28]. hPF4 was conjugated with fluorescein isothiocyanate (FITC) after its heparin-binding sites were protected by binding to heparin-agarose beads (Sigma, USA) and tested for immune complex formation as described[20]. Monoclonal anti-PF4 antibodies KKO and RTO [29, 30] and FcγRIIA-blocking antibody IV.3 [31] were produced in hybridoma cells and purified on a Protein G column (Invitrogen, USA). HIT specific human IgG was isolated from a 61-year-old male patient who was strongly positive by an immunoassay for anti-PF4/heparin IgG {optical density of OD= 2.196, by the ZYMUTEST HIA MonoStrip IgG; Hyphen BioMed, Paris France; OD normal range: 0.000 – 0.500) and the serotonin release assay (LabCorp assay using liquid chromatography/tandem mass spectroscopy). Total IgG was precipitated from plasma by ammonium sulfate as described[32]. Patient IgG specific to PF4/GAG complex was then isolated using beads with PF4 bound to heparin immobilized on diamino-dipropylamine agarose (Pierce Chemical), as previously described[33]. Control hIgG was isolated from pooled plasma of 20 healthy volunteers on a protein G column (Invitrogen) as described by the manufacturer.

THP-1 cells culture

Monocyte-like cell line derived from leukemia THP-1 cells were cultured in phenol red free RPMI-1640 media supplemented with fetal bovine serum (10%), penicillin (100 U/mL), streptomycin (100 μg/mL), and β-mercaptoethanol (0.05 mM). Cells were kept at density from 0.2 × 10⁶/mL to 1.0 × 10⁶/mL at 37°C under 5% CO2.

Isolation of and characterization of human monocytes

Human blood was collected after informed consent from healthy, aspirin-free volunteers using a 19-gauge butterfly needle in 129 mM sodium citrate (10:1, v/v) under a protocol approved by CHOP’s Institutional Review Board for studies involving human subjects (IRB 07–5565) and immediately centrifuged at 190g for 10 minutes at room temperature. Platelet-rich plasma (PRP) was removed and replaced by Tyrode-EDTA buffer (134 mM NaCl, 3 mM KCl, 0.3 mM NaH₂PO₄, 2 mM MgCl₂, 5 mM HEPES, 5 mM glucose, 12 mM NaHCO₃, 3% BSA; 1 mM EDTA). Monocytes were then isolated by negative selection using RosetteSep^™ Human Monocyte Enrichment Cocktail according to the manufacturer’s instructions. Remaining platelets were labeled with PE-conjugated anti-human CD41a antibody and depleted using anti-PE MicroBeads according to the manufacturer’s instructions. Finally, monocytes were washed in Tyrode buffer (190g, 10 minutes, 4°C), re-suspended in the RPMI serum-free medium or in Tyrode buffer, and counted using CBC Hemavet 950 FS (Drew Scientific, USA) and in a hemocytometer. Cell viability was assessed by a standard method of supravital staining with trypan blue. On average, monocytes (CD14+) comprised 92±3% of all CD45-positive leukocytes as assessed by flow cytometry (Supplemental Figure 1A).

Isolation and characterization of mouse monocytes

Transgenic mice expressing human FcγRIIA on monocytes and platelets were described previously [19]. Mouse blood was drawn from the inferior vena cava using a protocol approved by the Institutional Animal Care and Use Committee at the Children’s Hospital of Philadelphia into a syringe containing sodium citrate (final concentration 12.9 mM) and immediately diluted 1:2 (v/v) with modified Tyrode buffer containing 2 mM EDTA. Diluted blood was centrifuged at 150g for 5 minutes at room temperature followed by removal of PRP. The pellet was mixed with the cold ACK lysing buffer 1:25 (v/v) and monocytes were isolated using negative magnetic separation with EasySep Mouse Monocyte Isolation Kit (Stemcell, USA) following manufacturer’s instructions. Isolated monocytes were resuspended in EDTA-free Tyrode buffer containing 0.5% bovine serum albumin and 2 mM CaCl₂. On average, monocytes (CD115+) comprised 90±3% of all CD45-positive leukocytes as assessed by flow cytometry (Supplemental Figure 1B).

Confocal microscopy of human monocytes or THP-1 cells

THP-1 cells or freshly isolated human monocytes were labeled with a DiI-membrane stain and Hoechst 33342-nuclear stain. Purified recombinant PF4 (50 μg/ml) was added to monocytes 10 minutes before KKO (a mixture containing 10 μg/ml of Alexa 488-labeled KKO and 40 μg/ml unlabeled KKO). To see the effect of PF4 itself on monocytes, cells were incubated with purified recombinant PF4 alone (a mixture containing 10 μg/ml FITC-labeled PF4 and 40 μg/ml unlabeled PF4). Untreated monocytes without addition of PF4 or KKO were used as a negative control. The cell suspension was deposited on the Glass Bottom Microwell Dishes (MatTec Corporation, USA) after addition of reagents and imaged immediately.

Fluorescent images of the untreated and variously treated monocytes were taken every 5 minutes for 1 hour with an inverted confocal microscope using a water immersion 40x objective (Zeiss LSM710, Germany) equipped with a Zeiss ZEN 2011 software. In dynamic experiments, monocytes were imaged at 10, 25, 70 and 120 minutes after a corresponding treatment using an inverted spinning disc confocal microscope with an oil immersion 40x objective (Leica Stellaris 5 DMi8, Germany) equipped with a LAS X software. The degree of co-localization of antibodies and cell plasma membrane was quantified using the ImageJ software (NIH) with a JACoP plugin.

Scanning electron microscopy (SEM) of monocytes and image analysis

Isolated human or mouse monocytes were incubated for 60 minutes at 37°C without or with PF4 (50 μg/ml). Part of the PF4-treated samples were then incubated with KKO (50 μg/ml) added 10 minutes after PF4. After incubation, the monocytes were fixed for 30 minutes at room temperature in a 2% glutaraldehyde solution in 50 mM sodium cacodylate buffer (pH 7.5) containing 100 mM NaCl. Fixed monocytes were deposited on a 0.4-μm-pore size polycarbonate filter by centrifugation at 150g for 15 minutes. The cells were rinsed three times for 5 minutes with the 50 mM sodium cacodylate buffer, dehydrated in ascending concentrations of ethanol, then immersed into hexamethyldisilazane and dried overnight. A thin film of gold-palladium was layered on the samples using a sputter coater (Polaron e5100, Quorum Technologies, UK). Micrographs were taken with a scanning electron microscope Quanta 250 FEG (FEI, Hillsdale, Oregon, USA). Using SEM images of human monocytes, the following morphometric parameters were measured with the ImageJ software: areas of ruffled and smooth surfaces per monocyte as well as dimensions of “knobs” (defined as the membrane-associated spherical particles <500 nm in diameter) and “blebs” (the membrane-associated spherical particles >500 nm in diameter) on the cell surface. The morphometry was performed on more than 40 individual cells for each experimental condition. At least 8 randomly selected images, each containing 3–7 individual monocytes, were analyzed for each experimental condition.

Statistical analyses

Statistical analyses were performed using the GraphPad Prism 8 software package (GraphPad Software, USA). Normality of data distribution was assessed with the Shapiro–Wilk and D’Agostino-Pearson criteria. Data arrays were analyzed using the unpaired Student’s t-test, Mann–Whitney U test or ordinary one-way ANOVA with Tukey’s multiple comparisons test for multiple comparisons. The χ²-test was used to analyze morphological data in categorical values. The level of statistical significance was 95% (p<0.05).

Results

PF4 binds to monocytes and changes the plasma membrane morphology

When THP-1 cells were incubated with fluorescently-labeled PF4, confocal microscopy revealed that PF4 bound robustly to the cell surface within 5 minutes (Figure 1, A and B). The cell-bound PF4 appeared as clustered fluorescent “knobs”, reflecting PF4 assembly on the cell plasma membrane, presumably due to interaction with surface GAGs. These protruding knobs were also visualized with SEM after incubation of isolated human monocytes with hPF4 (Figure 1, C and D). The knobs were observed on the surface of 65% of the PF4-treated monocytes compared to only 26% (p<0.001) of the control untreated monocytes (Table 1, Supplemental Figure 3). The presence of knobs on about ¼ of the control monocytes might be due to PF4 released by activated platelets during isolation of monocytes and/or to other membrane-associated macromolecular complexes. The average knob size on the PF4-treated human monocytes was 148±27 nm (n=808).

Figure 1. PF4 binds to monocytes and THP-1 monocytic cell line and induces morphological changes of the plasma membrane.

(A, B) Representative fluorescent confocal micrographs of THP-1 cells labeled with Hoechst 33342 (nuclear stain, blue) incubated with a mixture of FITC-labeled recombinant PF4 (10 μg/ml) (green) and unlabeled PF4 (40 μg/ml) added to prevent excessive fluorescence for 1 min (A) or 10 min (B). Membrane-bound PF4 appears as clustered fluorescent “knobs”. (C, D) Representative SEM images of human monocytes, either untreated (C) or treated with hPF4 (D), showing formation of “knobs” (purple arrows) along with a loss of ruffles (partial smoothening) of the cell membrane.

Table 1. Membrane remodeling of monocytes treated with purified PF4 alone or in combination with the HIT-like pathogenic (KKO) and non-pathogenic (RTO) antibodies against PF4/heparin.

Incidence of the structural elements formed on the surface of monocytes under various experimental conditions as visualized by SEM.

Identified membrane structures	Experimental conditions
	Control untreated monocytes (n=41)	Monocytes treated with PF4 (n=58)	Monocytes treated with PF4+KKO (n=51)	Monocytes treated with PF4+RTO (n=54)
Percentage (number) of cells bearing the particular membrane structures
Cells with knobs (<500nm)	26% (11)	65% (38)***	20% (10)	23% (12)
Cells with blebs (>500nm)	2% (1)	2% (1)	65% (37)***	0% (0)
Cells with ruffles	97% (40)	82% (48)*	67% (34)**	89% (48)
Cells without ruffles	2% (1)	17% (10)	34% (17)***	12% (6)

n – total number of individual cells analyzed under each experimental condition.

^*p<0.05;

^**p<0.01;

^***p<0.001 compared to control untreated monocytes; χ²-test.

In addition to knob formation, exposure to PF4 changed the overall cell morphology, causing a partial loss of the ruffles on the monocyte plasma membrane, as revealed by quantitative analysis of the SEM images (Tables 1 and 2). Untreated monocytes typically had an almost continuous ruffled surface (Figure 1C), while monocytes treated with PF4 started to lose this prevailing ruffled surface (Figure 1D), with 17% of the cells having smooth plasma membrane without ruffles compared to only a single unruffled cell in the control (Table 1, Supplemental Figure 3). This visual observation was confirmed by morphometry (Table 2), showing that the average ruffled surface area per cell in monocytes treated with PF4 was about 15% less than unexposed control monocytes (p<0.01).

Table 2. Membrane remodeling of monocytes activated by the HIT-like pathogenic antibody (KKO).

Morphometry of the surface of individual monocytes under various experimental conditions as visualized by SEM.

Structural elements	Control untreated monocytes (n=42)	Monocytes treated with PF4 alone (n=58)	Monocytes treated with PF4+KKO (n=51)
Smooth surface area/cell, μm2	8.7±0.6	9.8±0.7	12.1±1.0**
Ruffled surface area/cell, μm2	23.5±0.8	20.3±0.8**	19.8±1.4*
Smooth/ruffled surface ratio calculated for individual cells	0.41±0.06	0.51±0.04	0.83±0.14*
Surface area covered by blebs and knobs/cell, μm2	0.32±0.05	0.36±0.03	1.25±0.26**
Number of blebs/cell	0.15±0.06	0.02±0.01	1.9±0.4**
Number of knobs/cell	13.1±1.8	17.3±1.8	6.5±0.9**

n – total number of individual cells analyzed under each experimental condition.

The results are presented as a mean ± SEM;

^*p<0.05;

^**p<0.01 compared to control untreated monocytes; unpaired Student’s t-test.

Monocyte membrane remodeling induced by the interaction of KKO with PF4

Activation of human monocytes by the pathogenic HIT-like antibody KKO in the presence of PF4 leads to the expression of tissue factor and increased procoagulant activity [18] However, tissue factor expression requires 3–6 hours of stimulation [10, 12, 19], while the membrane remodeling starts within 10 minutes after binding of PF4 (Figure 1A). To study the morphologic changes attendant with the binding of KKO, PF4-treated monocytes were incubated with KKO and analyzed using SEM. KKO caused formation of large blebs (>500 nm) found on 65% of monocytes (Figure 2A and B, Table 1) and disappearance of the smaller knobs (<500 nm), which were now seen on only 20% of the cells (p<0.001). On average, there were ~2 blebs formed per cell treated with PF4 and KKO (Table 2). At the same time, following the combined application of PF4 and KKO, the average number of knobs per one cell (6.5) was reduced ~3-fold compared to the monocytes exposed to PF4 alone (17.3) (Table 2).

Figure 2. Monocyte membrane remodeling induced by KKO added to the cells pre-coated with PF4.

(A, B) Representative SEM images showing multiple human monocyte surface-associated blebs (purple arrows) as well as smooth and porous cell membrane (green arrows). (C, D) Representative fluorescent confocal micrographs of THP-1 cells labeled with Hoechst 33342(blue-nuclei); DiD (red-membrane) and incubated with unlabeled PF4 for 20 min; at 6 (C) and 12 (D) minutes after addition of KKO (a mixture of Alexa 488 labeled KKO, 10 μg/ml and unlabeled KKO, 40 μg/ml to prevent excessive fluorescence). The images show the time-dependent formation of membrane-bound PF4-KKO ultra large immune complexes appearing as green large fluorescent “blebs”. (E, F) Human monocytes, untreated (E) or treated with RTO after PF4 (F), which are morphologically indistinguishable, as they both have ruffled plasma membrane without knobs and blebs.

To follow the formation of knobs and blebs over time, we used confocal microscopy of THP-1 cells and live unfixed human monocytes imaged at 3 min intervals for 30 min (Figure 2C and D, Supplemental Figure 2). Binding of fluorescently labeled KKO to the PF4-coated monocyte membrane occurred within 6 minutes of incubation with a very high extent of co-localization between KKO and the plasma membrane (Pearson colocalization coefficient 0.89±0.06). Within these 6 minutes, relatively small knobs formed that enlarged over time and reached the size of large blebs, similar to those seen by SEM (Figure 2D, Supplemental Figure 2).

Addition of KKO to PF4-treated monocytes also enhanced disappearance of the membrane ruffles, which is the morphology typical of quiescent, untreated monocytes. After combined PF4/KKO exposure, the fraction of ruffled monocytes (67%) was significantly less than the unexposed cells (97%, p<0.01) or cells exposed to PF4 alone (82%, p<0.05) (Table 1, Supplemental Figure 3). Many monocytes incubated with PF4 and KKO had the characteristics of activated cells with smooth (unruffled) surface (34% cells), compared to the control (2%) or monocytes incubated with PF4 alone (17%) (Table 1, Supplemental Figure 3). These visual qualitative differences in the membrane smoothness (categorical values) were confirmed by the measurement of the relative areas occupied by smooth and ruffled surfaces per cell (Table 2). The PF4/KKO-treated monocytes had ~40% larger average smooth area than the control untreated cells (p<0.01) and ~25% larger smooth area than the cells incubated with PF4 alone (p<0.01). Consequently, the average ruffled area in the PF4/KKO-exposed monocytes was ~16% less (p<0.05) than in control untreated cells (Table 2). Remarkably, some monocytes incubated with PF4 and KKO had open pores in the plasma membrane (Figure 2B), suggesting degranulation [34]. There were no significant changes in the size of monocytes in SEM images (Supplemental Figure 4A), but the loss of ruffles altered the scattering properties evaluated by flow cytometry (Supplemental Figure 4B,C). These changes of the monocyte surface were also accompanied by expression of phosphatidyl serine, as assessed by Annexin V binding using flow cytometry (Supplemental Figure 4D) at the same time as these morphological changes observed by SEM.

Inhibition of PF4-induced remodeling of the monocyte membrane by RTO

To determine if the changes induced in monocyte membrane morphology were specific for PF4 complexes and the HIT-like monoclonal antibody KKO, we incubated PF4-coated monocytes with an isotype-matched, non-activating, anti-human PF4 monoclonal antibody RTO [29], which is known to disaggregate PF4 into monomers that do not form complexes with polyanions [28, 30]. After addition of RTO (50 μg/ml) to PF4-treated monocytes, knobs were found on 23% of cells, which is similar to unexposed monocytes (26%) and significantly less than on monocytes exposed to PF4 alone (65%, p<0.001) (Figure 2E, F; Table 1). In other words, RTO essentially prevents the formation of knobs on the cell membrane. The inhibitory effect of RTO on the formation of membrane knobs was confirmed by the morphometry of individual monocytes. The average number of knobs per monocyte after incubation with PF4 in the presence of RTO (8.6±2.1) was ~2-fold less than in the absence of RTO (17.3±1.8, p<0.01). In addition, PF4-coated monocytes incubated with RTO had no blebs on their surface (0%), similar to control monocytes (2%), and monocytes coated with PF4 (2%), and in contrast to monocytes incubated with PF4 and KKO (67%). Finally, 89% of monocytes incubated with PF4 and RTO had ruffles, similar to control monocytes (97%, p>0.05) (Table 1, Figure 2E, F).

Formation of monocyte-derived microvesicles induced by KKO and human HIT IgG in the presence of PF4

In addition to the knobs and blebs associated with the plasma membrane, the monocyte preparations exposed to PF4 and KKO contained many extracellular vesicular particles found on the background surface in SEM images (Figure 3A). The rate of formation and release of these microvesicles was followed in real time with spinning-disk fluorescent confocal microscopy, and the structures were identified as DiD-stained fluorescent extracellular submicron-sized objects. Monocytes were followed continuously in the same microscopic fields and the number of microvesicles was counted at 25 and 120 minutes after stimulation with either PF4 and KKO or KKO alone. Exposure of monocytes to KKO in the presence of PF4 caused significantly more extensive microvesicle formation compared to the untreated or KKO-treated cells (Figure 3B). The rate of microvesicle formation, determined as the difference in the number of extracellular particles formed at 120 minutes versus 25 minutes after normalization by the number of monocytes visualized, was also significantly higher when monocytes were incubated with PF4 and KKO compared to KKO alone. The size distribution of these microvesicles (Figure 3C), measured from SEM images (Figure 3A), revealed a broad range from 0.1 μm to >1 μm with a peak at 0.23 μm, consistent with the known size of monocyte-derived microvesicles [35, 36]. Similar results were seen when human PF4 coated monocytes were incubated with human HIT IgG isolated from a patient with HIT (Figure 3E and D).

Figure 3. Extracellular microvesicles are formed by monocytes treated with HIT IgG added to the cells pre-coated with PF4.

(A) A representative SEM image showing extracellular microvesicles on and around monocytes treated with PF4 and KKO. (B) Rate of microvesicle formation represented as an average number of microvesicles per monocyte formed between 125 and 25 minutes of incubation without (untreated) or with addition of KKO over PF4 or KKO alone. (C) Size distribution of extracellular microvesicles (n=452) measured from SEM images like shown in A. (D,E) Microvesicles formed by monocytes treated with human HIT IgG identified by fluorescent confocal microscopy (D) and scanning electron microscopy (E). N= 3–4 biological replicates for each condition in B, C D and E. Data were analyzed by ordinary one-way ANOVA with Tukey’s multiple comparisons test (B,E) or paired t-test (D)

Monocyte membrane remodeling by the PF4/KKO immune complexes requires FcγRIIA

To evaluate the role of FcγRIIA receptors in monocyte activation by KKO in the presence of PF4, monocytes were isolated from the blood of wild type mice that naturally lack FcγRIIA (FcγRIIA⁻) and transgenic mice expressing the human FcγRIIA receptor (FcγRIIA⁺). The FcγRIIA⁺ and FcγRIIA⁻ monocytes pre-coated with human PF4 were incubated in parallel for 60 minutes with KKO and compared using SEM (Figure 4). Blebs were found on the plasma membrane of 72% of the FcγRIIA⁺ monocytes, but only 14% of the FcγRIIA⁻ monocytes had a blebbed surface (p<0.001). Moreover, ~50% of the FcγRIIA⁺ monocytes incubated with PF4/KKO had no surface ruffles, while only 11% of the FcγRIIA⁻ monocytes had none. The two types of monocytes were also distinct in their propensity to aggregate: no FcγRIIA⁻ cells were found within aggregates, while 42% of the FcγRIIA⁺ monocytes formed aggregates, consistent with cell activation. Similar results were obtained with human monocytes incubated with FcγRIIA-blocking antibody IV.3 (Figure 4 H, I). Blocking receptors prevented loss of ruffles (21% of monocytes were smooth when activated in the sample pre-incubated with IV.3 vs 69% of monocytes in the sample incubated with PF4 + KKO alone, p < 0.0001; Supplemental Table 1) and bleb formation (22% monocytes with blebs in the sample pre-incubated with IV.3 vs 94 % of monocytes in sample incubated with PF4 + KKO without blocking FcγRIIA, p < 0.0001; Supplemental Table 2)

Figure 4. Monocyte membrane remodeling induced by the KKO and PF4 requires FcγRIIA.

(A-C) Representative SEM images of monocytes isolated from a WT FcγRIIA⁻ mouse, either untreated (A) or incubated with PF4 alone (B) or with PF4 and KKO (C) without morphological signs of activation. (D-G) Representative SEM images of monocytes isolated from a transgenic FcγRIIA⁺ mouse, either untreated (D) or incubated with PF4 alone (E) or with PF4 and KKO (F). Figure 4F and especially Figure 4D clearly show that the FcγRIIA⁺ monocytes treated with PF4 and KKO, unlike the FcγRIIA⁻ cells, show strong signs of activation, namely blebs on the plasma membrane, smooth membrane, formation of microvesicles, and the propensity to aggregate. (H) FcγRIIA blocking antibody IV.3 prevented loss of ruffles and bleb formation on human monocytes, compared to (I) samples without FcγRIIa blocking antibody.

Platelet-monocyte interactions in the presence of pathogenic HIT-like antibodies and PF4 and the effect of heparin.

To study the interaction between activated monocytes and platelets, human PRP was added to isolated human monocytes and incubated in the absence or presence of PF4 and KKO. Analysis of SEM images showed that in the untreated samples most monocytes were ruffled without knobs and blebs (Figure 5A). Morphological alterations of monocytes caused by PF4 and KKO in the presence of added platelets were typical for activated monocytes, as they were smooth, had knobs, blebs, and formed many microvesicles on the background surface (Figure 5B). The platelet morphology and physical interaction between monocytes and platelets differed. Without PF4 and KKO, mostly round or discoid platelets attached to the background surface and no particulate matter was seen (Figure 5A). In the presence of PF4 and KKO, platelets underwent substantial morphological changes induced by HIT antibodies, namely they lost their discoid shape and formed long filipodia (Figure 5B), confirming the pronounced KKO-induced platelet activation described previously [8, 37]. Remarkably, the platelets physically interacted with monocytes via filipodia, including bridging adjacent monocytes. This functional change of interactions between monocytes and platelets contributes to both thrombosis and inflammation [38]. As shown in Figure 4 H and I, inhibition of FcγRIIa by blocking antibody IV.3 also prevented the formation of platelet monocyte aggregates.

Figure 5. Platelets and monocytes interaction upon exposure to PF4 and KKO and the effect of heparin.

(A) A representative SEM image of an untreated mixture of human monocytes and platelets. (B) A mixture of human monocytes and platelets incubated with KKO in the presence of PF4, showing interactions between smooth monocytes and platelets via filipodia as well as many microvesicles on the background surface. (C, D) Effect of heparin. Human monocytes treated with PF4 (100μg/ml) + KKO (50μg/ml) + heparin (1U/ml) showing signs of activation (loss of ruffles, membrane with pores) surrounded with clusters of ULC of PF4/heparin/KKO.

In the presence of heparin (1U/ml) monocytes showed the features of activation (loss of ruffles, smooth membrane with pores) but with few knobs and blebs on the membrane. In contrast, there were numerous polymeric network-like clusters surrounding the monocytes and interacting platelets (Figure 5 C, D). These clusters likely represent ultralarge immune complexes described by us previously[28]. These complexes do not appear to be integral to the cell membrane, which is consistent with our previous finding that heparin dissociates HIT complexes from the cell surface[18–20]

Discussion

HIT is an autoimmune disorder characterized by mild thrombocytopenia but a high propensity for recurrent thromboembolic events caused by immune complexes composed of PF4/polyanions/antibodies that activate cells, platelets, monocytes, and neutrophils through FcγRIIA receptors. We have previously demonstrated that PF4/polyanion complexes, which serve as targets for activating HIT antibodies, form on cell surfaces in association with membrane glycosaminoglycans [18, 19]. Additionally, we have shown that monocyte activation plays a crucial role in the thrombogenicity associated with HIT antibodies [9, 10, 19]. We now directly demonstrate binding of PF4 to the surface of monocytes in a model system that simulates activation of monocytes by human PF4-containing pathogenic immune complexes. The proposed sequence of events and mechanisms are illustrated in Figure 6. First, PF4 binds to GAGs on the cell surface, forming ~200-nm knobs, the cellular equivalent of the “ultralarge complexes” of human PF4 with unfractionated heparin in solution described earlier [39, 40]. Pathogenic HIT antibodies, represented by the HIT-like monoclonal antibody KKO or human HIT patient IgG in our studies, bind to the antigenic PF4/GAG complexes formed on the cell surface. The resultant immune complexes composed of human PF4, GAGs and the pathogenic HIT antibodies likely cross-link and activate FcγIIA receptors, which induce extensive morphological changes on the monocyte plasma membrane. These changes include loss of membrane ruffles and concurrent membrane “smoothing” with concurrent formation of membrane protrusions in the form of blebs. It is noteworthy that the HIT-like antibodies bind to the surface of monocytes only in the presence of surface-bound PF4, indicating formation of specific immune complexes attached directly to the plasma membrane. Moreover, the alterations of the plasma membrane described here occur within minutes after addition of antibodies, suggesting that they are driven by relatively fast rearrangements of the membrane-associated cytoskeleton in response to the intracellular signaling. In the presence of heparin at concentrations shown to promote cell activation by HIT immune complexes[10, 12] but also to displace PF4 from cell surfaces[18–20], SEM images reveal a polymeric network of what are likely PF4/heparin/KKO ultralarge complexes intermingled with platelets.

Figure 6. Schematic diagram showing the observed effects of pathogenic HIT antibodies on monocytes and the observed changes in HIT.

Quiescent monocytes have ruffles (top left) and have GAGs and FcγRIIA on their surface (top middle). Upon addition of PF4 (top right), the PF4 forms high molecular weight complexes with surface GAGs residing on the monocyte plasma membrane (glycocalyx) and forms <500-nm knobs. These knobs are an antigenic target for pathogenic HIT antibodies. Addition of HIT antibodies leads to the formation of large immune complexes of PF4/GAG/antibodies appearing as >500-nm blebs attached to the cell surface. These blebs optimally require the presence of FcγRIIA receptors (bottom right). Bleb formation then causes monocyte activation associated with release of microvesicles and smoothening of the surface associated with the loss of membrane ruffles (bottom middle). The activated monocytes and monocyte-derived microvesicles are highly procoagulant and promote thrombosis in HIT (bottom left). Created in BioRender. Andrianova, I. (2021).

From the observed time course of events, it seems that formation of plasma membrane-associated knobs induced by PF4, and the blebs induced by PF4 plus the HIT-like antibodies comprises an intermediate step of membrane remodeling that leads ultimately to shedding submicron-size extracellular vesicles. Others have shown that activated monocytes in HIT express tissue factor, release microvesicles and, therefore, are prothrombotic [10–12, 19]. Here we demonstrate that these changes are accompanied by phosphatidylserine expression, which produces a highly prothrombotic surface. Shedding of microvesicles may comprise the mechanism by which activated monocytes lose much of their typical ruffled membrane surface and become smoother. Thus, these monocytes with a smooth surface may be considered “residual bodies” and may represent a structural biomarker for monocyte activation in HIT as well as in other proinflammatory and prothrombotic states.

The ability of the non-pathogenic mAb RTO to interfere with the formation of PF4 surface complexes, as shown by the decreased presence of “knobs” on the monocyte surface and lack of markers of further monocyte activation, is consistent with its potential to destabilize PF4 tetramers [30] and to break HIT antigenic complexes [28]. Moreover, these results provide further evidence that the modulation of HIT antigen and immune complex formation occurs not only in solution but also on the cell surface and may be a potential disease-specific intervention.

The contrasting behavior of the pathogenic antibody KKO and the non-pathogenic antibody RTO raises the question of whether pathogenic (platelet-activating) antibodies from patients with HIT and thrombosis behave differently compared to anti-PF4 antibodies that test negative in functional platelet assays. Here, we demonstrate that HIT-specific pathogenic antibodies isolated from patient plasma behave similarly to KKO. In contrast, non-pathogenic anti-PF4/polyanion-specific antibodies may be more heterogeneous and vary in their effects. Some may display RTO-like activity and dissociate PF4/polyanion complexes, while others may induce membrane remodeling, stabilizing PF4/GAG complexes on the cell surface without engaging FcγRIIA and therefore preventing cell activation as we have shown with deglycosylated KKO [41]. Additional experiments are needed to further elucidate how the biological activity of HIT antibodies affects monocyte membrane dynamics.

There has been a controversy in the literature as to whether activation of monocytes and induction of tissue factor expression in monocytes is mediated by the FcγRI [12] or FcγRIIA [9] receptors. To address this issue, we compared the morphology of human monocytes, monocytes from the wild type mice lacking FcγRIIA receptors (FcγRIIA-negative) and from transgenic mice expressing the human FcγRIIA receptors (FcγRIIA-positive). Unlike the human and mouse FcγRIIA-positive monocytes, FcγRIIA-negative mouse monocytes respond to the pathogenic HIT-like monoclonal antibodies by forming significantly smaller blebs and having less propensity to aggregate (Figure 4). These results confirm that the FcγRIIA receptors are essential for the full structural alterations in human monocytes that develop in response to the HIT-related immune complex activation. Nevertheless, the presence of some blebs on the monocytes from the FcγRIIA-negative mouse monocytes suggests that the FcγRI present on mouse monocytes may contribute to their activation.

The monocyte glycocalyx contains GAGs, including heparan sulfate and dermatan sulfate [21], with a higher affinity for PF4 than the platelet surface, which is composed mostly of chondroitin sulfate[19]. These GAGs are necessary for the formation of stable antigenic complexes [21, 22]. Moreover, monocytes also have many more copies of FcγRIIA on their surface than do platelets (reported values 8×10⁴-1×10⁵ for a human monocyte [42] vs. 900–5,000 for a platelet [43, 44] although the surface density is similar (80–100 molecules per 1 μm² for monocytes vs. ~20–120 molecules per 1 μm² for platelets (see calculations in the Supplement). However, additional membrane receptors in addition to higher affinity GAGs may play a role. We, therefore, expect that the observed surface changes would occur earlier or be more prominent on monocytes than on platelets.

Formation of heterotypic platelet-leukocyte aggregates is a pathophysiological mechanism that bridges inflammation and blood coagulation and can enhance expression of tissue factor by activated monocytes [45]. Elevated levels of circulating platelet-monocyte aggregates in the blood have been found in a number of (pro)thrombotic inflammatory conditions, such as antiphospholipid syndrome [46], COVID-19 [47, 48] and other viral infections [49], tuberculosis [50], sepsis [51], and HIT [52]. The formation of platelet-monocyte aggregates is not just a marker of inflammatory platelet or monocyte activation, but it appears to be a strong risk factor for cardiovascular disease [53, 54]. We have shown that the interaction of platelets with monocytes in the presence of HIT antibodies is a mechanism for transactivation that amplifies the prothrombotic cellular reactions [9]. In this study, the structural basis for the physical interaction between activated platelets and monocytes has been demonstrated, suggesting how platelet-monocyte aggregates might form in the blood of HIT patients. In particular, co-aggregation of platelets and monocytes was shown to be mediated by long thin membrane protrusions or filipodia formed by the KKO/PF4-treated platelets in the presence of monocytes. At the molecular level, platelet-monocyte interactions are mediated by binding of P-selectin to PSGL-1 and CD15, followed by integrin activation on monocytes, resulting in additional binding interactions through CD40L–CD40, TREM-1 ligand–TREM-1, and CD36–CD36 via thrombospondin [38, 55]. Binding of activated platelets to monocytes has been shown to increase the production of various proinflammatory mediators and expression of tissue factor [56].

In summary, our data show that PF4 forms high molecular weight complexes on the surface of monocytes that are observable as <500 nm knobs. In the presence of HIT-like antibodies and involving the surface FcγRIIA receptors, the plasma membrane undergoes remodeling, leading to the formation of >500 nm blebs followed by microvesiculation, as well as formation of platelet-monocyte aggregates. Monocytes (like most blood cells) release prothrombotic microvesicles when they are stimulated. Taken together, these morphological alterations underlie direct immune activation of monocytes and transactivation between monocytes and platelets in HIT. These morphological changes in monocytes found here may not be specific to HIT immune complexes, but may represent a broader mechanism by which monocyte activation contributes to thrombosis and additional pathological outcomes in other immune complex mediated disorders. More generally, we propose that similar processes may be involved in microvesiculation by monocytes and other cells in diverse states associated with thromboinflammation.