Self-sustaining IL-8 loops drive a prothrombotic neutrophil phenotype in severe COVID-19

Abstract

Neutrophils provide a critical line of defense in immune responses to various pathogens, inflicting self-damage upon transition to a hyperactivated, procoagulant state. Recent work has highlighted proinflammatory neutrophil phenotypes contributing to lung injury and acute respiratory distress syndrome (ARDS) in patients with coronavirus disease 2019 (COVID-19). Here, we use state-of-the art mass spectrometry-based proteomics and transcriptomic and correlative analyses as well as functional in vitro and in vivo studies to dissect how neutrophils contribute to the progression to severe COVID-19. We identify a reinforcing loop of both systemic and neutrophil intrinsic IL-8 (CXCL8/IL-8) dysregulation, which initiates and perpetuates neutrophil-driven immunopathology. This positive feedback loop of systemic and neutrophil autocrine IL-8 production leads to an activated, prothrombotic neutrophil phenotype characterized by degranulation and neutrophil extracellular trap (NET) formation. In severe COVID-19, neutrophils directly initiate the coagulation and complement cascade, highlighting a link to the immunothrombotic state observed in these patients. Targeting the IL-8-CXCR-1/-2 axis interferes with this vicious cycle and attenuates neutrophil activation, degranulation, NETosis, and IL-8 release. Finally, we show that blocking IL-8-like signaling reduces severe acute respiratory distress syndrome of coronavirus 2 (SARS-CoV-2) spike protein-induced, human ACE2-dependent pulmonary microthrombosis in mice. In summary, our data provide comprehensive insights into the activation mechanisms of neutrophils in COVID-19 and uncover a self-sustaining neutrophil-IL-8 axis as a promising therapeutic target in severe SARS-CoV-2 infection.

Keywords: COVID-19; Cytokines; Neutrophils; Proteomics; Vascular Biology.

Figure 1. Neutrophil proteomics reveal immature neutrophils with a strong IFN response in COVID-19.

(A) Principal component analysis of neutrophils of patients, by group. Proteins with adjusted P value of less than 0.1 were included for this PCA. (B) Positions of patient neutrophil samples on principal component axis 1 (PC 1). One-way ANOVA. (C) Box plot of CD10 abundance (log-scaled median-MAD normalized abundance) of neutrophils by group. Abundance is normalized with median-MAD method. Significance is adjusted P values. (D) Heat map of mean IFN-stimulated genes (ISG) protein abundance on neutrophils by group. (E) Box plot of ISG score calculated from log-scaled abundance values (see methods) on neutrophils by group. One-way ANOVA with post hoc Holm-Sidak’s multiple comparisons test to control. (A–E), n = 9 healthy control, n = 5 pneumonic control, n = 5 severe COVID-19 and n = 9 intermediate patients with COVID-19. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 2. Severe COVID-19 neutrophils upregulate proteins implicated in IL-8 signaling.

(A) ClueGo upregulated neutrophil proteome pathway grouping of severe COVID-19 compared with intermediate COVID-19. (B) Heatmap of mean IL-8 pathway protein abundance on neutrophils by group. (C) Box plot of IL-8 score calculated from log-scaled abundance values (see methods) by group. One-way ANOVA with post hoc Tukey’s multiple comparisons test between all groups. (A–C) n = 9 healthy control, n = 5 pneumonic control, n = 5 severe COVID-19 and n = 9 intermediate patients with COVID-19. (D and E) Linear regression of Horowitz index (PaO2/FiO2) or clinically measured D-dimer (μg/mL) and IL-8 score of patients with COVID-19. P value signifies slope significantly non-zero. 95% confidence interval shown in gray. n = 5 severe and 9 intermediate patients with COVID-19. (F) Box plot of normalized serum IL-8 plasma levels of COVID-19 and control patients. n = 26 control patients without COVID-19, n = 78 patients with mild-moderate COVID-19 (WHO Grade 1–4), n = 29 severe COVID-19 (WHO Grade 5–8) patients. One-way ANOVA with post hoc Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 3. A systemic and autocrine neutrophil IL-8 loop in COVID-19.

(A and B) UMAP of cells from scRNA-Seq bronchioalveolar lavage data in Wauters et al. (26). Annotated cell type in UMAP and feature plot of IL-8 expression by patient population. (C) Violin plot of neutrophil IL-8 expression by patient population in Wauters et al. (26) (n [neutrophils] = 431 for mild and n = 6117 for severe COVID-19, n = 788 for mild non–COVID-19 pneumonia and n = 1 for severe non–COVID-19 pneumonia). Since only 1 severe non–COVID-19 pneumonia neutrophil was detected in BALF derived from healthy patients, no violin plot depicting IL-8 production in this patient population is shown. Unpaired Student’s t tests. (D) Violin plot of IL-8 expression by cell type and patient population of Wauters et al. (26), including neutrophils. (E) IL-8 production assay schematic. (F) IL-8 production assay results. One-way ANOVA with post hoc Dunnett’s multiple comparisons test comparing IL-8 to the other conditions. n = 4 neutrophil donors. Mean ± SEM is shown. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 4. COVID-19 neutrophils are characterized by IL-8–induced degranulation and drive a systemic prothrombotic phenotype.

(A) NETing and degranulated neutrophils as percentage of all neutrophils with control treatment of IL-8. Paired 2-sided Student’s t test. (B) Representative images of control and IL-8 treated neutrophils. Stars show degranulation, arrows NETs. Scale bar: 10 μm. ALPL, alkaline phosphatase; MPO, myeloperoxidase. (C) Heatmap of mean azurophilic granule protein abundance by group. (D) Box plot of overall granule score calculated from log-scaled abundance values (see methods) by group. One-way ANOVA with post hoc Tukey’s multiple comparisons test between all groups. (E) Box plots of all 4 granule scores calculated from log-scaled abundance values (see Methods) by group. Two-sided unpaired Student’s t test between controls and intermediate or severe COVID-19. (F) Linear regression of clinically measured D-dimer [μg/mL] and azurophilic granule score of neutrophil proteins of patients with COVID-19. (G) Heat map mean of coagulation cascade protein abundance by group. (H) Correlation matrix of clinically measured D-dimer and Horowitz index and neutrophil protein abundance of patients with COVID-19. Pearson r is shown in each box and as a heatmap, P values is in brackets. (I) Linear regression of D-dimer [μg/mL] and neutrophil fibrinogen protein abundance of patients with COVID-19. (F and I) P value signifies slope significantly non-zero. 95% confidence interval shown in gray. n = 5 severe and 9 intermediate patients with COVID-19. (C–E and G) n = 9 healthy control, n = 5 pneumonic controls, n = 5 severe COVID-19, and n = 9 intermediate COVID-19. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 5. Therapeutic blockade of IL-8 reduces COVID-19–associated neutrophil activation in vitro.

(A) IL-8 production and blocking assay similar to Figure 2E. One-way ANOVA with post hoc Dunnett’s multiple comparisons test comparing IL-8 to the other conditions. n = 5 healthy neutrophil donors. (B) Illustration of in vitro experiment: Neutrophils from healthy donors were exposed to plasma from patients with severe COVID-19 and pretreated with vehicle, anti–IL-8 antibody or the CXCR-1/-2 antagonist reparixin. Degranulation, NETosis and surface expression of activation markers were assessed by microscopy and flow cytometry, respectively. (C) NETing and degranulated neutrophils as percent of healthy neutrophils stimulated with COVID-19 plasma and nothing, anti–IL-8 antibodies or reparixin. Paired 2-sided Student’s t test, n = 7 plasma samples from patients with COVID-19. (D) Representative images of COVID-19 plasma and COVID-19 plasma and anti–IL-8 incubated neutrophils. Stars show degranulation, arrows NETing. Scale bar: 10 μm. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 6. Therapeutic blockade of IL-8 signaling attenuates ARDS-related microthrombosis in vivo.

(A) Illustration of in vivo SARS-CoV-2 S1 spike protein–induced lung injury mouse model. (B) Clinical sepsis score at 24 hours after S1 spike protein–induced lung injury for control and reparixin-treated hACE2 mice. (C) Quantification and representative micrographs of fibrinogen-positive/fibrinogen-binding neutrophils in one vessel in the lungs of control or reparixin-treated mice. Scale bar: 10 μm. (D) Quantification and representative micrographs of microthrombi in lungs of vehicle or reparixin-treated mice. Vessel borders are shown with dashed white lines. Scale bar: 50 μm. (B–D) Two-tailed unpaired Student’s t test, n = 4 per group. Data are shown as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.