Trained immunity causes myeloid cell hypercoagulability

Abstract

The pathogenic basis for increased thrombotic risk in individuals with inflammatory diseases is poorly understood. Myeloid cell "trained immunity" describes persistent innate immune cell memory arising from prior exposure to an inflammatory stimulus, leading to an enhanced immune response to subsequent unrelated stimuli. We identify enhanced myeloid cell prothrombotic activity as a maladaptive consequence of trained immunity. Lipopolysaccharide (LPS) stimulation of macrophages trained previously with β-glucan or heme exhibited significantly enhanced procoagulant activity compared to macrophages stimulated with LPS alone, which was mediated by enhanced acid sphingomyelinase-mediated tissue factor decryption. Furthermore, splenic monocytes isolated from β-glucan-trained mice revealed enhanced procoagulant activity up to 4 weeks after β-glucan administration compared to monocytes from control mice over the same time period. Moreover, hematopoietic progenitor cells and bone marrow interstitial fluid from β-glucan-trained mice had enhanced procoagulant activity compared to control mice. Trained immunity and associated metabolic perturbations may therefore represent an opportunity for targeted intervention in immunothrombotic disease development.

Fig. 1.. β-Glucan–induced trained immunity up-regulates coagulation-associated gene expression.

(A) Schematic diagram outlining the protocol for inducing trained immunity in BMDMs. Briefly, BMDMs were pretreated with media or whole β-glucan particles (100 μg/ml) and left for 24 hours before cells were washed three times with PBS and left to rest for 1 week. On day 7, cells were restimulated with LPS (100 ng/ml) for 6 hours before RNA isolation. (B) Bulk RNA sequencing was performed on LPS-treated (n = 3) and β-glucan–trained BMDMs restimulated with LPS (n = 3), and differential gene expression analysis was performed between the groups. (C) Gene Ontology enrichment analysis of the gene sets significantly up-regulated in LPS-restimulated β-glucan–trained BMDMs compared to LPS-treated BMDMs. BP, biological processes; CC, cellular component; MF, molecular functions. (D) Heatmap showing coagulation-associated gene expression profiles for LPS-treated (1°Media 2°LPS) and LPS-restimulated β-glucan–trained (1°β-glucan 2°LPS) BMDMs following RT-qPCR analysis. Data were normalized for RPS18 mRNA levels, and results were expressed as fold change compared to untreated BMDMs (1°Media 2°Media). RT-qPCR results for (E) F3, (F) Serpine1, and (G) EGR1 mRNA levels. A one-way ANOVA was used to determine statistical significance with *P ≤ 0.05 and **P ≤ 0.01 for six independent experiments measured in duplicate.

Fig. 2.. β-Glucan–induced trained immunity enhances myeloid cell procoagulant activity.

Cells were pretreated with media or whole β-glucan particles (100 μg/ml) and left for 24 hours before cells were washed three times with PBS and left to rest for 1 week. On day 7, cells were restimulated with LPS (100 ng/ml) for 24 hours. A cell-based TGA was performed with cells as the sole source of TF activity in 80 μl of FXII-deficient plasma, 20 μl of MP reagent, and 20 μl of thrombin substrate added to initiate the reaction. (A) Thrombogram for TGA performed with BMDMs from each group and (B) associated lag times calculated. ns, not significant. (C) Thrombogram for TGA performed in the presence of BMDM supernatants (80 μl) and (D) associated lag times. (E) Thrombogram for TGA performed with β-glucan–trained primary human monocytes and (F) associated lag times. (G) A FXa generation assay was performed in the presence of primary human monocytes incubated with FVIIa (0.5 μg/ml) and FX (10 μg/ml) for 30 min. An FXa substrate was added to detect FXa generated as a consequence of TF activity. (H) PAI-1 protein levels measured by ELISA in β-glucan–trained BMDMs. (I) tPA-mediated plasmin generation in the presence of β-glucan–trained BMDMs was measured using a plasmin-specific fluorogenic substrate, and the fluorescence reading after 60 min (PG⁶⁰) was determined. (J) BMDMs were transfected with Serpine1 siRNA or nontargeting siRNA (both 20 nM) before β-glucan training and LPS treatment. tPA-mediated plasmin generation in the presence of these cells was then assessed. A paired t test or one-way ANOVA was used where appropriate to determine statistical significance with *P ≤ 0.05 and **P ≤ 0.01 for three to six independent experiments measured in duplicate.

Fig. 3.. Free heme induces trained hypercoagulability in myeloid cells via the protoporphyrin IX (PPIX) ring.

BMDMs and human monocytes were trained with 50 μM heme, 50 μM PPIX, 50 μM Fe-NTA, or DMSO (dimethyl sulfoxide) vehicle control. After 24 hours, cells were washed and rested for 7 days. Cells were then restimulated with LPS (100 ng/ml) for 6 hours for mRNA analysis and 24 hours for protein analysis and functional assays. (A) F3 mRNA levels in heme-trained BMDMs were determined by RT-qPCR. Data were normalized to RPS18 mRNA levels, and results were expressed as fold change relative to untreated BMDMs (1°Media 2°Media). (B) Thrombogram for TGA performed in the presence of heme-trained BMDMs and (C) associated lag times. (D) Thrombogram for TGA performed in the presence of PPIX- or Fe-NTA–trained BMDMs and (E) associated lag times. (F) Thrombogram for TGA performed in the presence of heme-trained BMDM supernatants and (G) associated lag times. (H) Thrombogram for TGA performed in the presence of heme-trained human monocytes and (I) associated lag times. (J) PAI-1 levels were measured by ELISA in heme-, PPIX-, and Fe-NTA–trained BMDMs. (K) tPA-mediated plasmin generation in the presence of heme-, PPIX-, and Fe-NTA–trained BMDMs was measured using a plasmin-specific fluorogenic substrate, and the fluorescence reading after 60 min (PG⁶⁰) was determined. (L) BMDMs were transfected with Serpine1 siRNA or nontargeting siRNA (both 20 nM) before heme training and LPS treatment. tPA-mediated plasmin generation in the presence of treated BMDMs was then performed. A paired t test or one-way ANOVA was used where appropriate to determine statistical significance with *P ≤ 0.05 and **P ≤ 0.01 for three to six independent experiments measured in duplicate.

Fig. 4.. β-Glucan–induced trained hypercoagulability requires epigenetic and metabolic reprogramming.

(A) Schematic diagram illustrating chemical epigenetic inhibitor targets. (B) BMDMs were incubated with 1 mM MTA, 100 μM EGCG, or 12 mM PG for 1 hour before training with β-glucan (100 μg/ml) for 24 hours. Cells were washed three times, and growth media were supplemented with epigenetic inhibitors added for the rest period (100 μM MTA, 100 μM EGCG, and 12 mM PG). On day 7, cells were restimulated with LPS (100 ng/ml). TNFA and F3 mRNA levels were determined by RT-qPCR, and PAI-1 levels were determined by ELISA. (C) BMDMs were pretreated for 3 hours with 10 mM 2-DG before the β-glucan training protocol. After LPS restimulation on day 7, cells were assessed in a TGA to generate a thrombogram and (D) lag times. (E and F) BMDMs or (G and H) BMDM supernatants were incubated with anti-TF antibody (1H1) for 1 hour before TGA analysis and lag-time measurement. (I) ASMase activity in trained BMDMs was measured by fluorogenic ASMase activity assay. Trained cells were restimulated with LPS for 30 min, 1 hour, or 3 hours before cell lysis. (J and K) ASMase cell surface expression on BMDMs was determined by flow cytometry in F4/80⁺ BMDM populations. (L) β-Glucan–trained and (M) untreated BMDMs were incubated with 10 μM desipramine or imipramine for 1 hour before LPS restimulation on day 7. TGA analysis was then performed, and the (N) lag time was determined. A paired t test or one-way ANOVA was used where appropriate to determine statistical significance with *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001 for three to seven independent experiments measured in duplicate.

Fig. 5.. Splenic monocytes from β-glucan–administered mice display enhanced procoagulant activity weeks after administration.

(A) Schematic depiction of the experimental outline to invoke trained immunity in vivo. Briefly, mice were injected with either PBS or whole β-glucan particles 1 to 4 weeks before sacrifice. CD115⁺ splenic monocyte population was then isolated. Splenic monocytes were left untreated or restimulated with LPS (100 ng/ml) ex vivo for 24 hours. ip, intraperitoneally. (B) TNFα and (C) IL-6 levels released from splenic monocytes were determined by ELISA. (D) A TGA was performed in the presence of splenic monocytes isolated from mice administered β-glucan 1 to 4 weeks previously. (E) Lag time, (F) ETP, and (G) peak thrombin were also determined. (H) In parallel, mice were fed a standard diet or a diet supplemented with dietary β-glucan. Splenic monocytes were then isolated and assessed by TGA to give (I) a thrombogram, (J) lag time, (K) peak thrombin, and (L) ETP. A paired t test or one-way ANOVA was used where appropriate to determine statistical significance with *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ***P ≤ 0.0001 for four to eight mice with samples measured in duplicate.

Fig. 6.. β-Glucan–induced central trained immunity results in a BM environment with enhanced procoagulant activity.

(A) Briefly, mice were injected with either PBS or whole β-glucan particles 1 to 4 weeks before sacrifice. Plasma was collected via the retro-orbital sinus. BM interstitial fluid and HSCs were harvested. BM HSPC populations were analyzed by flow cytometry for (B) % LSK⁺ cells in BM and (C) % MPP3 myeloid-biased progenitor cells in the LSK⁺ compartment. (D) IL-1β, (E) TNFα, and (F) IL-6 cytokine levels were determined in BM interstitial fluid by ELISA. (G) IL-6 and (H) TNFα levels released from cultured BMDMs without or with ex vivo LPS restimulation were determined by ELISA. Cultured BMDMs without or with ex vivo LPS restimulation were included in a TGA to generate (I) a thrombogram, (J) lag time, and (K) ETP values. (L) TGA was performed in the presence of BM interstitial fluid to give a thrombogram and (M) associated lag times. (N) TGA analysis was performed with mouse platelet-poor plasma isolated from trained and control mice. Each well contained 20 μl of plasma, 20 μl of TBS, and 20 μl of FluCa thrombin substrate, and TGA parameters (O) ETP and (P) peak thrombin values are shown. A paired t test or one-way ANOVA was used where appropriate to determine statistical significance with *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ***P ≤ 0.0001 for four to eight mice with samples measured in duplicate.