Neutrophil CD14 is a driver and a therapeutic target for deep vein thrombosis

Key Points

• •Multiomic analyses (RNA sequencing, proteomics, flow cytometry) of neutrophils at 3 hours after DVT revealed robust overexpression of CD14.
• •Neutrophil-specific CD14 knockdown using a novel lentiviral short hairpin RNA approach significantly reduced DVT severity.

Visual Abstract

Abstract

Neutrophil-mediated persistent inflammation and neutrophil extracellular trap formation (NETosis) are critical in the pathogenesis of deep vein thrombosis (DVT). Identifying the mechanisms controlling these proinflammatory and prothrombotic functions is critical for designing and developing new therapeutics for DVT. However, how neutrophils acquire these phenotypes during the early stages of DVT remains poorly understood. Here, multiomic analyses (RNA sequencing, proteomics, and flow cytometry) of bone-marrow neutrophils 3 hours after DVT, revealed consistent overexpression of CD14. Concurrently, mice with DVT exhibited significantly elevated plasma granulocyte colony-stimulating factor (G-CSF) and neutrophil hyperactivation. Mechanistic studies using a geometric deep learning model (DeepPBS) and chromatin immunoprecipitation revealed that elevated G-CSF drives CD14 overexpression by upregulating the transcription factor C/EBPα (leucine zipper CCAAT-enhancer binding protein α). Importantly, neutrophil-specific CD14 knockdown using a novel lentiviral short hairpin RNA approach significantly improved DVT outcomes by lowering thrombus burden, thrombosis incidence, and intrathrombus neutrophil and citrullinated histone H3 accumulation. Studies in G-CSF stimulated primary human neutrophils revealed that CD14 inhibition reduces markers of inflammation and NETosis (activated gasdermin D, citrullinated histone H3, and S100A8/A9), while increasing apoptosis, thus demonstrating translational relevance. Collectively, our study uncovers an essential role of neutrophil CD14 in DVT pathogenesis and establishes it as a promising therapeutic target for DVT.

Introduction

Deep vein thrombosis (DVT) is a complex process involving a cascade of events where neutrophils, monocytes, platelets, and endothelial cells cooperatively promote thrombus formation.^1,2 Patients with DVT exhibit an acute proinflammatory response, characterized by increased systemic leukocytosis and inflammatory cytokines.3, 4, 5 Elevated leukocytes, particularly neutrophils, are strongly associated with an increased risk of DVT and mortality.3, 4, 5 Neutrophils drive venous thrombus propagation by releasing neutrophil extracellular traps (NETs), secreting inflammatory mediators, activating endothelium, and altering coagulation and fibrinolysis.6, 7, 8, 9, 10 Elucidating the upstream regulators of these diverse prothrombotic neutrophil functions is critical for identifying and developing novel therapeutic interventions.

Potential upstream regulators of neutrophil activity include factors such as granulocyte colony-stimulating factor (G-CSF), which is a critical regulator of granulopoiesis and mature neutrophil activation.^11,12 G-CSF signaling activates the transcription factor C/EBPα, primarily by enhancing its ability to bind target gene promoters, thereby driving neutrophil gene expression programs.¹³ Nevertheless, it is not known whether this G-CSF/C/EBPα pathway activates neutrophils in the early stages of DVT.

This study aimed to investigate the mechanisms responsible for the increased prothrombotic activity of neutrophils during the early phase of DVT. Employing multiomic analyses of neutrophils isolated from mice with inferior vena cava (IVC) stenosis (3 hours after induction), we observed significant upregulation of CD14, a coreceptor for toll-like receptor 4 (TLR4). Although CD14 is known to be highly expressed by monocytes and macrophages, it is also actively synthesized by neutrophils.¹⁴ Furthermore, the broader CD14/TLR4 signaling pathway has been implicated in various aspects of thromboinflammation.¹⁵ Interestingly, we also observed significantly increased plasma G-CSF levels during the early phase of DVT. However, whether G-CSF drives neutrophil CD14 upregulation via the transcription factor C/EBPα and the role of neutrophil-specific CD14 in DVT pathogenesis remain unknown. Therefore, we investigated the mechanistic link between G-CSF, C/EBPα, and CD14 expression; determined the functional impact of neutrophil-specific CD14 knockdown on DVT pathogenesis; and evaluated the effects of CD14 inhibition using in vitro human neutrophil assays.

Methods

Detailed methods are provided in the supplemental File.

Mice

All animal procedures were approved by the institutional animal care and use committees of Louisiana State University Health Sciences Center–Shreveport (approval no. P-23-002). Ten- to 12-week-old male C57BL/6J (WT) mice were purchased from The Jackson Laboratory.

IVC stenosis model

The mouse IVC stenosis model of DVT was performed in mice, as previously described.16, 17, 18, 19, 20 Only male mice were used for this model, as ligation in female mice may result in necrosis of the reproductive organs.^18,21 Briefly, a midline laparotomy was made, IVC side branches were first ligated. For stenosis, a space holder (30-gauge) was positioned on the outside of the exposed IVC, and a permanent narrowing ligature was placed below the left renal vein. Next, the needle was removed to restrict blood flow to 80% to 90%. IVC thrombi were harvested 48-hour after stenosis, detached from the vessel wall, dried, weighed in a microbalance, and imaged. The operator was blinded with respect to the treatment of mice.

RNA sequencing (RNA-seq)

Neutrophils were isolated from bone marrow (3 hours after IVC stenosis) as previously described by our group.22, 23, 24, 25 Neutrophil cell pellets were lysed, and RNA was isolated using the RNeasy Mini Kit (Qiagen, catalog no. 74106) as per manufacturer’s instructions. Libraries were prepared with the Stranded mRNA Prep, Ligation Kit (Illumina). One microgram of RNA was processed for each sample and mRNA was purified and fragmented. cDNA was synthesized and 3' ends were adenylated. Anchor sequences were ligated to each sample and a limited-cycle polymerase chain reaction (PCR) was performed to amplify and index the libraries. The average library size was determined using an Agilent TapeStation D1000 assay (Agilent Technologies) and libraries were quantitated with quantitative PCR (qPCR) (Bio-Rad CFX96 Touch Real-Time PCR, NEB Library Quant Kit for Illumina). Libraries were normalized to 0.5nM and pooled. The library pool was denatured and diluted to ∼100pM. A 1% library of 2.5pM PhiX was spiked in as an internal control. Paired end 76 × 76 base pair sequencing was performed on an Illumina NovaSeq 6000.

Peptide and protein identification

Protein identification from bone-marrow derived neutrophils (3 hours after IVC stenosis) was evaluated by commercial service from Cell Signaling Technology, TMT10plex Total Proteome profiling service, that provides accurate global profiling of protein abundance in cells and uses multiplexed sample labeling with Tandem Mass Tags (TMT) and liquid chromatography tandem mass spectrometry. Mass spectra were evaluated by Cell Signaling Technology using SEQUEST and the GFY-Core platform (Harvard University). Searches were performed against the 20180718 update of the UniProt Homo sapiens database with a mass accuracy of ±50 ppm for precursor ions and 0.02 Da for product ions. Total proteome data were filtered to a 1% peptide-level false discovery rate with mass accuracy ±5 ppm on precursor ions and presence ions; Immobilized Metal Affinity Chromatography (IMAC) data were filtered to samples with a phosphorylated residue prior to filtering to a 1% protein-level false discovery rate. All IMAC quantitative results were generated using Skyline 16 to extract the integrated peak area of the corresponding peptide assignments. Accuracy of quantitative data was ensured by manual review in Skyline or in the ion chromatogram files. TMT quantitative results were generated using the GFY-Core platform (Harvard University).

Lentivirus preparation for neutrophil-specific CD14 knockdown

Lentiviral vectors expressing either a Cd14-targeting short hairpin RNA (shRNA) (pLV[miR30]-mCherry->EGFP:{mCd14[miR30-shRNA#2]}) or a scramble control shRNA (pLV[miR30]-mCherry-{Human MRP8}>EGFP:{scramble shRNA}), both driven by the human MRP8 promoter, were generated by VectorBuilder. The constructs were assembled using the gateway cloning system. Specifically, 3 entry clones were created: 1 containing the human MRP8 promoter, 1 containing the EGFP reporter, and 1 containing the 5' miR-30E-{mCd14[miR30-shRNA#2]}-3' miR-30E cassette. These entry clones were recombined into a lentiviral backbone containing the mCherry marker via LR recombination. The vector map is shown in supplemental Figure 1.

Human samples collection and peripheral neutrophil and monocyte isolation

Peripheral blood was collected at the Louisiana State University Health Sciences Center–Shreveport from healthy volunteers and patients with stroke (approved Institutional Review Board [IRB] protocol, 0002176). Samples were processed within 15 minutes from their collection, and neutrophils were isolated as we previously reported.⁸ Briefly, peripheral venous blood was collected into sodium citrate blood collection tubes, incubated for 20 minutes at room temperature with 5:1 volume of HetaSep (Stemcell catalog no. 07806) to allow red blood cell sedimentation. The upper layer was then gently loaded on top of Histopaque-1077 and centrifuged at 400g for 30 minutes at room temperature, no brake. A pellet, containing the neutrophils and RBCs, was further subjected to red blood cell lysis. Neutrophil purity was determined by flow cytometry and found consistently to be >90%. For monocyte isolation monocytes were enriched using the EasySep Human Monocyte Enrichment Kit (Stemcell Technologies, catalog no. 19059).

Statistical analysis

For analysis, GraphPad Prism software (10.4.0) was used. Normality and equal variance were tested using the Shapiro-Wilk and Bartlett tests, respectively. Normally distributed data were analyzed by Student t test, 1-way or 2-way ANOVA (analysis of variance) followed by Sidak multiple comparisons test, and nonnormally distributed data were analyzed using the Mann-Whitney test (for 2-group) or nonparametric 2-way ANOVA followed by Fisher least significant difference test. Thrombosis incidence data were analyzed using the Fisher exact test. The results were considered significant at P < .05.

Results

DVT induces elevated G-CSF levels and early systemic neutrophilia

Increased G-CSF levels and higher blood neutrophil counts have been reported in several inflammatory conditions.^11,12 To evaluate the effect of DVT on G-CSF levels and neutrophil counts, we subjected C57BL/6J mice to DVT (IVC stenosis) or sham surgery. Compared to sham controls, mice with DVT showed significantly increased plasma G-CSF (at 3 and 6 hours; Figure 1A-B), elevated blood neutrophils, a higher neutrophil-to-lymphocyte ratio, and reduced lymphocytes (at 3 hours; supplemental Figure 2A-C). Platelet and monocyte count, however, remained unchanged (supplemental Figure 2D-E). These alterations were accompanied by increased plasma markers of neutrophil activation in mice with DVT, including cell-free DNA, elastase, and myeloperoxidase (supplemental Figure 3A-C). Thus, DVT triggers increased plasma G-CSF and systemic markers indicative of neutrophil activation.

Figure 1.

Unbiased RNA-seq reveals increased CD14 expression on neutrophils after DVT. (A) Schematic of experimental design. (B) Plasma G-CSF levels in mice with IVC stenosis and with sham surgery at indicated time points. (C-F) Unbiased transcriptomics of bone marrow neutrophils isolated from mice 3 hours after sham or DVT surgery. (C) Principal component analysis. (D) Volcano plots of differentially expressed genes (DEGs) based on RNA-seq. (E) GSEA based on RNA-seq showing the top 10 upregulated pathways in response to DVT surgery. (F) Chord diagram showing relationships between gene ontology terms and DEGs. All data are expressed as mean ± SD and analyzed by 2-way repeated measures ANOVA (Kruskal-Wallis test) followed by Fisher least significant difference test (panel B). n = 3 (panel B); n = 5 (panels C-F). IL-8, interleukin-8; LPS, lipopolysaccharide; MPO, myeloperoxidase; NE, Neutrophil elastase; TNF, tumor necrosis factor.

Multiomic profiling reveals increased CD14 expression on neutrophils after DVT

To investigate the mechanisms behind heightened neutrophil thrombogenicity at the onset of thrombosis, we analyzed the transcriptome and proteome of murine bone-marrow neutrophils collected 3 hours after DVT induction. RNA-seq of neutrophils from mice with DVT revealed distinct transcriptional profiles compared to controls, as demonstrated by principal component analysis (Figure 1C) and differential gene expression analysis (Figure 1D). Upregulated genes in neutrophils from DVT mice were associated with inflammation (Cd14, Nos2, Saa3, Acod1, and Plscr1), antiapoptosis (Muc1, Bcl3, Snai1, Bcl2, Ddit4, and Cd47), and chemotaxis (Cxcl3, Cxcl2, Ccl2, and Pded4d). Gene ontology analysis revealed enrichment of several Cd14-associated biological processes (Figure 1E-F). To determine whether these transcriptional changes were reflected at the protein level, we performed proteomic analysis and identified 764 differentially regulated proteins in the DVT group compared to controls (Figure 2A). Consistent with the RNA-seq data, neutrophils from DVT mice exhibited an approximately threefold increase in CD14, placing it among the top 5 most highly upregulated proteins (Figure 2B). CD14 is a glycosylphosphatidylinositol-anchored membrane protein. Although CD14 is highly expressed in monocytes, it also plays a functionally significant role in neutrophils.¹⁴ Next, using flow cytometry, we confirmed our proteomics data, observing significantly increased CD14 expression on circulating neutrophils from mice with DVT, but not on monocytes (Figure 2C; supplemental Figure 4). Thus, our data from RNA-seq, proteomics, and flow cytometry uncover increased CD14 expression by neutrophils, but not monocytes, from mice with DVT.

Figure 2.

G-CSF induces CD14 upregulation via C/EBPα. (A-B) Total proteome profiling of bone-marrow neutrophils isolated from mice with sham surgery or DVT using multiplexed sample labeling with TMT and liquid chromatography tandem mass spectrometry. (A) Volcano plots of differentially expressed proteins based on the proteomics data. (B) Top 10 upregulated proteins in mice with DVT. (C) CD14 expression on circulating monocytes and neutrophils by flow cytometry analysis 3 hours after sham or DVT-surgery (left: flow cytometry histogram; right: quantification). (D) The mRNA expression level of CEBPA (C/EBPα) in primary human neutrophils in response to G-CSF (10 ng/mL) for 6 hours is determined by quantitative Reverse Transcription Polymerase Chain Reaction (RT-PCR). (E) AlphaFold3 predicted structure of C/EBPα bound to the promoter sequence of interest, which is input to DeepPBS. DeepPBS predicted relative heavy atom-level importance scores (RI scores) are shown as spheres, revealing key protein residues involved in DNA recognition. (F) DeepPBS predicted binding specificity logo, that is Position Weight Matrix.²⁶ The x-axis sequence represents the putative C/EBPα binding motif within CD14 promoter region. (G) Chromatin immunoprecipitation (ChIP)–qPCR is performed on primary human neutrophils exposed to G-CSF (10 ng/mL) for 6 hours with antibody to C/EBPα and the target promoter region of CD14 is amplified by qPCR. ChIP-qPCR data are presented as relative enrichment over input ± SD of 3 biological repeats. All data are expressed as mean ± SD and analyzed by 2-way ANOVA followed by Holm-Šídák multiple comparisons (panel C) or Mann-Whitney test (panels D-G). n = 4 (panels A-B,G), n = 6 (panels C-D). IL-1RB, interleukin-1RB.

G-CSF induces CD14 upregulation via C/EBPα

Next, we evaluated the mechanisms responsible for increased CD14 expression on human neutrophils upon G-CSF exposure. Given that G-CSF levels increased after DVT (Figure 1B) and G-CSF is known to drive neutrophil gene expression by activating C/EBPα binding to target promoters,¹³ we first assessed the mRNA expression of C/EBPα in neutrophils in the absence and presence of G-CSF. We found that G-CSF significantly increased CD14 (supplemental Figure 5A) and CEBPA (C/EBPα) expression in human neutrophils (Figure 2D), but not in monocytes (supplemental Figure 5B-C). In line with these results, circulating neutrophils isolated from mice with DVT surgery also exhibited increased Cd14 and Cebpa expression levels (supplemental Figure 6A-B).

To mechanistically explore how C/EBPα regulates CD14 expression, we performed a structure-based analysis of the CD14 promoter using AlphaFold3²⁷ and a geometric deep learning model (DeepPBS)²⁶ (Figure 2E). The analysis revealed a predicted C/EBPα binding motif within the CD14 promoter, consistent with the known C/EBPα binding sequence (JASPAR MA0102.5)¹³ (Figure 2F). To validate this, we next performed a chromatin immunoprecipitation assay using neutrophils exposed to G-CSF or control. qPCR (chromatin immunoprecipitation–qPCR) targeting the putative binding site in the CD14 promoter showed a significant increase in C/EBPα binding upon G-CSF exposure (Figure 2G), suggesting that G-CSF promotes CD14 expression via C/EBPα-mediated transcriptional regulation. Collectively, these findings establish that G-CSF signaling induces neutrophil CD14 upregulation via C/EBPα and neutrophils CD14 is a viable therapeutic target to reduce DVT severity.

Neutrophil-specific CD14 knockdown attenuates DVT severity in mice

Next, to assess the pharmacological efficacy of targeting neutrophils CD14, we utilized an established model of IVC stenosis.^16,24,25 To enable therapeutically relevant, neutrophil-specific CD14 knockdown, we developed and utilized a novel MRP8 promoter-driven lentiviral shRNA system (Figure 3A). Lentiviral-mediated shRNA delivery is a well-validated approach commonly employed for achieving in vivo durable knockdown.²⁸ A single IV dose (2.5 × 10⁷ transducing units) was sufficient to achieve ∼65% CD14 knockdown up to 2-week in neutrophils whereas monocytes CD14 levels were not affected (Figure 3B-C). Moreover, baseline circulating neutrophil counts, tail bleeding time (Figure 3D-E), complete blood counts (supplemental Figure 7), and neutrophil surface marker expression (CD11b and CD62L; supplemental Figure 8A-B) remain unaffected in mice injected with CD14 shRNA. To evaluate DVT outcomes, male C57BL/6J mice were randomly assigned to receive either scramble or CD14 shRNA, and IVC stenosis was induced on day 7 (Figure 3F). Neutrophil-specific CD14 knockdown significantly reduced thrombus burden, and more importantly significantly lowered thrombosis incidence (5/15 vs 12/15 in controls, Figure 3G-I), highlighting the causative role of neutrophil CD14 in DVT pathogenesis. We next evaluated the effects of CD14 knockdown in IVC stasis (complete ligation) induced DVT model. In contrast to the significant protective effect observed in the stenosis model, CD14 knockdown failed to improve DVT outcomes in mice with IVC stasis-induced DVT (supplemental Figure 9). This key differential finding strongly suggests that the role of neutrophil CD14 in venous thrombosis is model-specific and highlights its crucial, nonredundant involvement in the pathobiology of flow-restricted (stenosis) thrombosis, but not in flow-arrest (stasis) thrombosis.

Figure 3.

Neutrophil-specific CD14 knockdown attenuates DVT severity in mice. (A) Schematic representation of the lentiviral-based system for neutrophil-specific Cd14 knockdown. A single IV dose of lentivirus (2.5 × 10⁷ transducing units per mouse) was administered, and analyses were performed on day 7 after injection. (B) Cd14 expression analysis in neutrophils: left, Cd14 mRNA fold change relative to Gapdh; right, surface CD14 expression determined by flow cytometry. (C) Cd14 mRNA fold change in circulating monocytes relative to Gapdh. (D) Blood neutrophil count on day 7 of scramble of CD14-shRNA administration. (E) Tail bleeding time on day 7 of scramble of CD14-shRNA administration. (F) Schematic of experimental design. (G) Representative ultrasound image of thrombus from CD14 knockdown and control mice 48 hours after DVT. (H) Quantification of thrombus area and (I) thrombosis incidence from n = 15 mice from each group. Only the data from mice with thrombus are shown in panel H. All data are expressed as mean ± SD and analyzed by Mann-Whitney test (panels B-E,H), Fisher exact test (I). n = 4-6 (B-E), n = 5 (panels D-E), n = 15 (panels G-I), DVT surgery was performed in n = 15 mice from each group.

Neutrophil CD14 promotes DVT severity by driving neutrophil accumulation and procoagulant platelet formation

CD14 activation on neutrophils, initiates proinflammatory cascades, leading to the release of prothrombotic damage-associated molecular patterns like S100A8/A9.^29,30 Consistent with a critical role in DVT pathogenesis, we observed that mice treated with CD14 shRNA exhibited significantly reduced mean plasma S100A8/A9 levels (supplemental Figure 10). Because S100A8/A9 is known to drive neutrophil activation³¹ and the formation of procoagulant platelets,³² both key factors in DVT severity, we next determined the mechanistic basis for the improved DVT outcomes in CD14-knockdown mice.

To dissect the contribution of neutrophil-specific CD14 in DVT pathogenesis, we first evaluated thrombus composition 48 hours after DVT. Immunohistochemistry revealed that CD14 shRNA–treated mice displayed significantly reduced neutrophil content and decreased citrullinated histone H3 expression, indicating attenuated NET formation (NETosis; Figure 4A-B). This reduction in intrathrombus accumulation was linked to altered neutrophil trafficking, as we observed significantly reduced expression of the chemokine receptor Cxcr2 in day 2 thrombi samples of mice treated with CD14 shRNA (Figure 4C). Furthermore, the expression of endothelial cell adhesion molecules (Vcam1, Icam1, Selp, and Sele) was unchanged (supplemental Figure 11), supporting a neutrophil-intrinsic trafficking defect rather than altered endothelial adhesion.

Figure 4.

Neutrophil CD14 promotes DVT severity by driving neutrophil accumulation and procoagulant platelet formation. (A-B) Representative immunofluorescence images of sectioned IVC with thrombus (48 hours after stenosis) from CD14 knockdown and control mice. (A) Left: neutrophil accumulation (Ly6G 1A8, green; DAPI [4′,6-diamidino-2-phenylindole], blue), right: quantification. (B) Left: anti-histone H3 (citrulline R2 + R8 + R17, NETs, red and DAPI blue), right: quantification. Original magnification ×20; scale bar, 100 μm. (C) Quantitative RT-PCR analysis of Cxcr2 mRNA expression in IVC thrombi harvested 48 hours after DVT surgery from CD14-knockdown and scramble control mice. Gene expression levels were normalized to Gapdh. (D) Representative immunofluorescence images of IVC sections containing thrombi 48 hours after stenosis in CD14-knockdown and scramble control mice. Left: anti-CD41 (platelets, red; DAPI, blue), right: quantification. Original magnification ×20; scale bar, 100 μm. (E) Detection of procoagulant platelets (CD62P⁺ annexin⁺) in IVC thrombi samples from control and CD14-knockdown mice 24 hours after DVT surgery. Left: representative flow cytometry plots showing CD62P and annexin V double-positive platelets; right: quantitative analysis. Quantitative RT-PCR analysis in IVC thrombi harvested 48 hours after DVT surgery from CD14-knockdown and scramble control mice. (F) Fibrinolytic markers (Plat, Plau, Serpine1, and Mmp9), and (G) proinflammatory cytokines (Il1b, Tnfa, and Il6). Gene expression levels were normalized to Gapdh. Data are presented as mean ± SD and analyzed using an unpaired Student t test or Mann-Whitney test, n = 5 to 6 (panels A-G). MFI, Mean Fluorescence Intensity.

Next, we assessed the impact of CD14 knockdown on overall thrombus structure and function. Immunofluorescence analysis showed a significant reduction in Ter119⁺ red blood cells, fibrin deposition (supplemental Figure 12A-B), and platelet (CD41⁺) accumulation (Figure 4D) in the CD14-knockdown group compared with controls. Moreover, expression of extracellular matrix genes (Col1a3 and Mmp2) was unchanged (supplemental Figure 12C), indicating that the structural composition of the thrombus was not significantly altered. Functionally, flow cytometry revealed a significant decrease in procoagulant platelets (CD62P⁺ annexin V⁺) in CD14-knockdown thrombi (Figure 4E). These data collectively suggest that neutrophil CD14 contributes to DVT severity by promoting neutrophil accumulation, which in turn enhances procoagulant platelet formation and alters overall thrombus composition.

Finally, we explored how CD14 influences thrombus remodeling and inflammation. We assessed the expression of fibrinolytic markers (Plat, Plau, Serpine1, and Mmp9) and proinflammatory cytokines (Il1b, Tnfa, and Il6). As shown in Figure 4F, Plau expression was significantly increased, whereas Serpine1 was reduced in CD14-knockdown thrombi, suggesting a favorable shift toward a profibrinolytic state. Consistent with attenuated neutrophil accumulation and inflammation, transcripts for the proinflammatory cytokines Il1b and Tnfa were also markedly decreased (Figure 4G). Collectively, these findings demonstrate that neutrophil CD14 knockdown lowers DVT severity by reducing neutrophil accumulation, inhibiting procoagulant platelet formation, and establishing a proresolving intrathrombus environment.

G-CSF signaling drives neutrophil activation and exacerbates DVT via the CD14 upregulation

To establish a mechanistic link between G-CSF signaling and neutrophil activation and to confirm its translational relevance in humans, we stimulated primary human neutrophils with recombinant human G-CSF (10 ng/mL, 6 hours) in the presence or absence of a CD14-neutralizing antibody. Immunoblotting and immunofluorescence analyses revealed that G-CSF treatment strongly induced markers of neutrophil activation and NETosis. Specifically, G-CSF significantly increased total and cleaved gasdermin D (GSDMD; Figure 5A-B), increased citrullinated histone H3 (Figure 5C-D), reduced apoptosis (TUNEL⁺ cells; Figure 5C-D), while simultaneously increasing the secretion of S100A8/A9 (Figure 5E). Importantly, CD14 blockade partially but significantly reversed these effects (Figure 5A-E). Furthermore, given that oxidized phospholipids (oxPAPC) serves as an endogenous CD14 ligand and is known to promote neutrophil activation (including Reactive Oxygen Species [ROS] generation, cytokine release, and NETosis), we evaluated if its effects were also CD14-dependent. Stimulating control neutrophils with oxPAPC significantly increased GSDMD, NETosis, and S100A8/A9 release while reducing caspase 3 activity (supplemental Figure 13A-E). Importantly, an anti-CD14 antibody significantly attenuated these effects (supplemental Figure 13A-E). Altogether, these data highlight translational relevance of CD14 signaling in mediating prothrombotic signals.

Figure 5.

G-CSF signaling drives neutrophil activation and exacerbates DVT via the CD14 upregulation. Primary human neutrophils (2 × 10⁶) were pretreated for 30 minutes with anti-CD14 antibody (15 μg/mL), followed by G-CSF stimulation (10 ng/mL) for 6 hours. (A) Representative western blot showing full-length GSDMD (FL-GSDMD), cleaved Gasdermin D N-terminal domain (GSDMD-NT), and ACTIN in neutrophil lysates. (B) Densitometric quantification relative to ACTIN. (C) Representative immunofluorescence images showing (CitH3, green; DAPI, blue) and TUNEL⁺ cells (red; DAPI, blue). (D) Quantification. Original magnification ×10; scale bar, 50 μm. (E) S100A8/A9 levels in neutrophil supernatants measured by enzyme-linked immunosorbent assay. (F) Representative western blot showing FL-GSDMD, cleaved GSDMD-NT, cleaved caspase-3, and ACTIN in peripheral neutrophils isolated from sham and DVT mice at 6 hours after surgery. (G) Densitometric quantification of FL-GSDMD and GSDMD-NT relative to ACTIN. (H) Schematic representation of the experimental design. (I) Flow cytometric analysis of CD14 expression on circulating neutrophils in DVT mice 24 hours after treatment with anti–G-CSFR antibody. Left: representative flow cytometry histograms; right: quantitative analysis. (J) Thrombus weight. Anti–G-CSFR antibody (300 μg/kg) or an isotype control were administered intraperitoneally to mice 30 minutes prior and 24 hours after surgery. Data are presented as mean ± SD and analyzed using 1-way ANOVA (panels B,D-E), Mann-Whitney test (panels G,J), or unpaired Student t test (I). n = 6 (panels A-E), n = 4 (panels F-G), n = 5 (I), and n = 15 (panel J). CitH3, citrullinated histone H3; MFI, Mean Fluorescence Intensity.

Because we observed significant increase in plasma G-CSF in DVT mice (Figure 1B), we next evaluated whether G-CSF signaling drives neutrophil activation and thereby promotes DVT severity in vivo. We first confirmed the GSDMD pathway activation in a murine DVT model. Circulating neutrophils isolated from DVT mice 6 hours after surgery showed a significant increase in cleaved GSDMD and cleaved caspase-3 compared with sham controls (Figure 5F-G), indicating rapid activation of the CD14-GSDMD pathway. Total GSDMD and levels were not significantly different (not shown). Notably, an increased caspase-3 level was observed in DVT mice; however, the time point analyzed (6 hours after DVT induction) likely reflects an early phase in which caspase-3 is partially cleaved or modestly active without triggering apoptosis. This early activation may occur downstream of ROS generation, as previously reported.^33,34

Finally, to define the causative role of G-CSF signaling in DVT pathology, we administered a neutralizing anti–G-CSF receptor (G-CSFR) antibody (300 μg/kg) or an isotype control to mice 30 minutes prior to DVT induction (Figure 5H). Mice treated with the anti–G-CSFR antibody exhibited significantly reduced CD14 expression on circulating neutrophils (Figure 5I), and, more importantly, a markedly reduced DVT severity (Figure 5J). Collectively, these comprehensive data demonstrate that G-CSF signaling drives CD14 upregulation in neutrophils, activating the GSDMD pathway, and consequently exacerbating DVT severity.

Discussion

The pathogenesis of DVT is a highly inflammatory process involving complex cross talk among endothelial cells, platelets, monocytes, and neutrophils.^1,2 Although neutrophils are recognized as critical mediators of venous thrombus formation,^7,10,35, 36, 37, 38, 39 the specific, druggable pathways regulating their prothrombotic functions are not fully characterized. Here, we identify neutrophil CD14 as a critical driver of DVT and reveal the upstream signaling axis responsible for its upregulation following DVT induction by IVC stenosis. We demonstrate that DVT triggers a rapid increase in plasma G-CSF, which induces the expression of CD14 via C/EBPα specifically on circulating neutrophils. Mechanistically, CD14 acts as a key signaling hub in neutrophils, promoting activation, NETosis, and the release of prothrombotic factors (like S100A8/A9) through the GSDMD pathway. Most importantly, our findings establish neutrophil CD14 as a viable therapeutic target. We found that neutrophil-specific CD14 knockdown significantly attenuates DVT severity and thrombosis incidence in mice by reducing neutrophil accumulation and inhibiting the formation of procoagulant platelets. These data collectively position the G-CSF/CD14 axis as a novel inflammatory pathway that mediates neutrophil-driven venous thrombus formation, offering a highly specific strategy for DVT prevention and treatment.

The clinical significance of the G-CSF signaling pathway in thrombosis and inflammation is supported by extensive clinical and genetic evidence. Although G-CSF is essential for granulopoiesis and host defense against infection, it often plays a detrimental role in sterile inflammatory diseases by promoting excessive neutrophil activation and survival.⁴⁰ This is clinically evident because patients with cancer receiving recombinant G-CSF exhibit a significantly higher incidence of VTE^41,42 and evidence from Mendelian randomization analyses indicates that genetically determined elevated G-CSF levels increase the risk of pulmonary embolism.^43,44 Furthermore, G-CSF administration in healthy volunteers increases circulating histone-DNA complexes,⁴⁵ a well-established surrogate marker for NETs critically linked to human thrombosis. Our finding that blocking G-CSF receptor signaling drastically reduces DVT severity is consistent with reports showing G-CSF antagonism mitigates disease in models of rheumatoid arthritis⁴⁶ and chronic obstructive pulmonary disease.⁴⁷ However, the novelty of our work lies in defining the specific molecular pathway that links systemic G-CSF elevation directly to the prothrombotic state. Unlike previous studies that broadly implicated G-CSF in promoting survival or migration via STAT3 and MAPK signaling,^46,47 we pinpoint the C/EBPα-CD14-GSDMD axis as the critical axis by which G-CSF specifically drives the NETotic and procoagulant transformation of the neutrophil.

Our most significant mechanistic finding is the involvement of GSDMD (a key inducer of pyroptosis and NETosis),^48,49 downstream of CD14 activation in neutrophils. Although GSDMD deficiency is known to protect against DVT,⁵⁰ that effect has been attributed primarily to inhibiting pyroptotic death of monocytes and macrophages and the subsequent release of tissue factor.⁵⁰ In contrast, our data reveal a neutrophil-intrinsic CD14-GSDMD axis that mediates NETosis without necessarily leading to complete lytic cell death. Specifically, we observed a rapid increase in cleaved GSDMD and markers of NETosis in circulating neutrophils from DVT mice, effects that were significantly attenuated by CD14 blockade in human neutrophils. This is consistent with evidence showing that CD14 and its endogenous ligand oxPAPC promote inflammasome-related activation in immune cells, leading to interleukin-1β secretion without full pyroptotic cell death.⁵¹ Importantly, recent work in thrombotic microangiopathy models also highlights that GSDMD primarily affects neutrophils by driving their maturation and β2-integrin activation, leading to enhanced recruitment and subsequent necroinflammation.⁵² We propose that the G-CSF/CD14 signal primes GSDMD cleavage in the neutrophil lineage, driving activation and release of inflammatory components (like S100A8/A9) that promote the procoagulant state, representing a GSDMD-dependent, nonlytic form of neutrophil activation distinct from the canonical monocyte/macrophage pyroptosis observed in DVT.

Targeting inflammatory neutrophils has emerged as a promising avenue for DVT management, moving beyond conventional anticoagulation.^7,39,53 Current strategies focus on regulating neutrophil activity, for example, by blocking NETosis via inhibitors of PAD4 (peptidylarginine deiminase 4) or GSDMD, or promoting NET degradation using DNases.^7,39,53 Furthermore, the recent US Food and Drug Administration–approved DPP1 inhibitor brensocatib successfully targets neutrophil maturation by disarming serine proteases,⁵⁴ demonstrating the feasibility of neutrophil-centric therapeutics. Consistent with the consensus that neutrophils and NETosis are essential drivers of venous thrombosis, our data demonstrated that CD14 knockdown significantly protected against DVT in the flow-restricted IVC stenosis model. However, this protective effect was entirely lost in the complete ligation (stasis) model. This differential finding is highly informative, indicating that the prothrombotic function of neutrophil CD14 is dependent on the model of injury. The involvement of CD14 appears crucial in the context of vessel wall injury and turbulent flow (stenosis) but becomes dispensable during flow-arrest coagulation (stasis). Our study also introduces a unique advantage by targeting the upstream regulator, neutrophil CD14. The success of neutrophil-specific CD14 knockdown in drastically reducing thrombus size and incidence validates CD14 as a therapeutic target. Moreover, in agreement with the known role of PLAUR+ neutrophils in fibrinolysis during flow reduction–induced thrombosis,⁵⁵ here we demonstrate that CD14 knockdown in neutrophils promotes a proresolving phenotype. This is characterized by increased intrathrombus Plau (urokinase-type plasminogen activator) expression and reduced Serpine1 (plasminogen activator inhibitor-1) expression, suggesting that neutrophils can be reprogrammed for enhanced fibrinolytic activity. This strategy offers the benefit of acting high in the signaling cascade (G-CSF→CD14), which enables effective control of multiple pathological downstream events, including GSDMD cleavage, NETosis, fibrinolysis, and S100A8/A9 release, simultaneously. By specifically targeting an upstream, cell-surface molecule upregulated in response to DVT, our approach maximizes therapeutic effect in the context of DVT while minimizing potential off-target effects on essential, common neutrophil effector functions, thereby laying the foundation for a novel class of highly targeted antithrombotic agents.

Our data provide unique insights into the transcriptional and proteomic changes in the bone-marrow neutrophils during early DVT pathogenesis, which may contribute to venous thrombus propagation and disease severity. Bone marrow neutrophils are the ideal target for this analysis because they express high levels of the G-CSFR (CSF3R) and reside adjacent to the fenestrated sinusoidal endothelium, allowing plasma G-CSF to readily access the marrow and directly modulate their transcriptional landscape.56, 57, 58 Our unbiased analyses shed new light on our understanding of neutrophil biology following DVT, and that may facilitate the development of effective therapeutics for DVT management. A major strength of this work is the novel MRP8-driven shRNA approach, which allowed for neutrophil-specific CD14 targeting in vivo, circumventing the limitation of lacking Cd14^fl/fl mice and providing a robust preclinical platform with direct translational implications for therapy development. Furthermore, the conservation of this CD14-mediated pathway was confirmed in human primary neutrophils in our in vitro studies. Despite these strengths, our study has limitations, particularly those inherent to translating findings from the murine model. For example, mouse blood differs from human blood, where neutrophils are far more abundant (representing most white blood cells in humans).⁵⁹ Another limitation is that the scope of our model was limited to young, healthy male mice, and DVT severity was assessed primarily by terminal thrombus weight and incidence. Future investigations are warranted to evaluate the therapeutic effect of CD14 targeting in females, aged, prothrombotic mouse models using more physiologically relevant outcomes such as IVC patency and embolism. Finally, the murine IVC stenosis model primarily reflects thrombus initiation driven by flow restriction, which is only one component of the multifactorial nature of human DVT pathogenesis.

In conclusion, this study demonstrates that the G-CSF–C/EBPα axis drives neutrophil CD14 expression during early DVT, and crucially, that targeting neutrophil CD14 attenuates DVT severity by mitigating neutrophil hyperactivation and NETosis. These findings collectively establish neutrophil CD14 as a driver of DVT and a promising therapeutic target.

Conflict-of-interest disclosure: The authors declare no competing financial interests.