Type I interferon is induced by hemolysis and drives antibody-mediated erythrophagocytosis in sickle cell disease

Abstract

Patients with sickle cell disease (SCD) suffer from intravascular hemolysis-associated vascular injury and tissue damage. Classical monocytes (CMo), which are the most abundant of circulating monocytes, are activated in SCD, but the cause and consequences of activation remain incompletely understood. We found a positive correlation between total plasma heme levels and circulating interferon-α (IFN-α) in patients with SCD along with upregulation of the type I IFN (IFN-I) inducible genes in sort-purified SCD patients' CMo by transcriptome analysis. We demonstrated that hemolysis led to IFN-I expression, predominantly by mouse liver monocyte and macrophages (Mⲫ), primarily through Tank kinase binding 1 (TBK1)/IκB kinase-ε (IKKε) but not TLR4. In response to hemolysis-induced IFN-I, mouse CMo migrated to the liver and differentiated into monocyte-derived Mⲫ, increasing their numbers by sixfold with acute hemin treatment. Hemolysis-driven IFN-I activity also led to the induction of Fc receptor CD64 expression on monocyte and Mⲫ populations, enhancing alloantibody-mediated erythrophagocytosis in SCD both in vivo in mice and in in vitro human cultures. Altogether, these data demonstrate IFN-I response to hemolysis as a novel activation pathway in monocytes and Mⲫ in SCD, opening the possibility for development of IFN-I-based diagnostics and therapeutics against alloantibody-mediated erythrophagocytosis.

Graphical abstract

Figure 1.

IFN-I can be induced by hemolysis. (A) Fold change in 14 IFN-I inducible genes as determined by RNA-Seq (IFI44, IFI44L, Mx1, Mx2, LY6E, EPSTI1, RSAD2, IFI6, OAS3, IFIT1, IFIT3, HERC5, ISG15, and SIGLEC1) in SCD CMo (n = 13) relative to HD CMo (n = 6). (B) Plasma IFN-α levels in HD (n = 19) and SCD patients at steady state on a chronic transfusion protocol (n = 36). (C) Plasma IFN-α levels in control and sickle mice (n = 5-6). (D) Scatter plot analysis showing correlation relationship between plasma IFN-α levels and plasma total heme levels in patients with SCD (n = 36 as in panel B). (E) Plasma IFN-α levels in WT mice at 20 hours after IV injection with PBS as control (200 µL/20 g body weight), water (300 µL/20 g body weight), RBC lysate (17.5 µmol hemoglobin/kg body weight in 200 µL), hemoglobin (17.5 µmol/kg body weight in 200 µL), or hemin (17.5 µmol/kg body weight in 200 µL) (n = 4-8). (F) Plasma IFN-β levels in WT mice at 20 hours after IV injection with PBS as control (200 µL/20 g body weight), or hemin (35 µmol/kg body weight in 200 µL) (n = 3-5). (G) Plasma IFN-α levels in WT mice at 20 hours after IV injection with hemin at doses of 0, 4.4, 8.8, 17.5, or 35 µmol/kg body weight all in 200 µL (n = 3-10). (H) Plasma IFN-α levels in WT mice at time point of 0, 2, 6, 20, and 72 hours after IV injection with hemin (35 µmol/kg body weight in 200 µL) (n = 3-7). Data represent mean ± SEM; means were compared using 2-tailed Student t test. *P < .05.

Figure 2.

Hemolysis-induced IFN-I is primarily produced by liver macrophages and monocytes through ROS and TBK1/IKKε. (A) Representative dot plots showing frequencies of IFN-β/YFP⁺ macrophages (Mⲫ) in liver and spleen from IFN-β/YFP reporter mice at 20 hours after IV injection with PBS as control (200 µL/20 g body weight) or hemin (35 µmol/kg body weight). (B) Frequencies of IFN-β/YFP⁺ cells in leukocyte subsets in liver, spleen, bone marrow (BM), and blood from hemin-treated mice shown as in panel A (n = 3). (C) The absolute number of IFN-β/YFP⁺ cells in leukocyte subsets from mice as shown in panel B (n = 3). (D) Plasma IFN-α levels in WT mice at 20 hours after IV injection with PBS as control (200 µL/20 g body weight) or hemin (35 µmol/kg body weight), after pretreatment with Clo-lipo or PBS liposome (300 µL/20 g body weight, 1 day before hemin treatment) (n = 3-7). (E) Plasma IFN-α levels in WT mice and TLR4^−/− mice at 20 hours after IV injection with PBS as control (200 µL/20 g body weight), or hemin (35 µmol/kg body weight) (n = 3-7). (F) Plasma IFN-α levels in WT mice at 20 hours after IV injection with PBS as control (200 µL/20 g body weight) or hemin (35 µmol/kg body weight), after pretreatment with TAK-242 (2 mg/kg body weight), amlexanox (50 mg/kg body weight) or NAC (250 mg/kg body weight), or SnPPIX (35 µmol/kg body weight) (n = 3-10). (G) Plasma IFN-α levels in WT mice at 20 hours after IV injection with hemin (35 µmol/kg body weight) and pretreated serially diluted amlexanox (from 50 to 3 and 0 mg/kg body weight) (n = 5-7). (H) Plasma IFN-α levels in sickle mice treated with amlexanox (50 mg/kg body weight/d for 3 consecutive days) or vehicle as control (n = 4). Data represent mean ± SEM; means were compared using 2-tailed Student t test. *P < .05.

Figure 3.

Hemolysis leads to IFN-I-dependent recruitment of monocytes to the liver. Frequencies of (A) CMo (gating strategy in supplemental Figure 2A-B) and (B) Ly-6C⁺MHC-II⁺ monocyte in total CD45⁺ leukocytes in blood, spleen, and liver of control mice and sickle mice (n = 3-4). Frequencies of (C) CMo and (D) Ly-6C⁺MHC-II⁺ monocyte in total CD45⁺ leukocytes in blood and liver of WT mice and ifnar1^−/− mice at 20 hours after IV injection with PBS as control (200 µL/20 g body weight) or hemin (35 µmol/kg body weight) (n = 4-5). (E) Plasma CCL2 levels in control mice and sickle mice (n = 4-5). (F) Plasma CCL2 levels in mice shown in panel C (n = 4-5). (G) Representative dot plot and histogram showing the gating strategy for analysis of total Mⲫ (F4/80⁺CD11b^low Mⲫ), monocyte derived Mⲫ (Tim-4⁻ MoMⲫ), and resident Mⲫ (Tim-4⁺ Mⲫ) in mouse liver and spleen. (H) Frequencies of MoMⲫ in total Mⲫ of spleen and liver in control and sickle mice (n = 3-6). (I) Representative histograms showing the gating strategy for liver Tim-4⁻ MoMⲫ, and Tim-4⁺ Mⲫ in WT mice, ifnar1^−/− mice, and ccr2^−/− mice at 72 hours after IV injection with PBS as control (200 µL/20 g body weight) or hemin (35 µmol/kg body weight). (J) Frequencies of MoMⲫ in total Mⲫ in livers of mice as shown in I (n = 3-6). Means were compared using 2-tailed Student t test. *P < .05.

Figure 4.

IFN-I increases erythrophagocytosis of low-level alloantibody sensitized RBCs in SCD. (A) Mean fluorescence intensity (MFI) of CD64 on purified CD14⁺ monocytes from HD (n = 14) and SCD (n = 7) after overnight culture with media or IFN-α (1000 IU/mL). (B) Percentage of CFSE⁺ monocytes in cultures of purified monocytes from HD (n = 10) and SCD (n = 9) following overnight co-incubation with CFSE-stained RhD⁺ RBCs coated using serially diluted anti-D antibody (from 4 to 0.06 and 0 µL in 100 µL staining buffer). (C) Percentage of CFSE⁺ monocytes in cultures of purified monocytes from HD (n = 10) and SCD (n = 9) as in panel B using low-dose anti-D antibody (0.25 µL in 100 µL staining buffer) in the absence (media alone) or presence of IFN-α (1000 IU/mL). (D) MFI of CD64 in liver CMo, Ly-6C⁺MHC-II⁺ monocyte, Tim-4⁻ MoMⲫ, and Tim-4⁺ Mⲫ of WT mice and ifnar1^−/− mice at 20 hours after IV injection with PBS as control (200 µL/20 g body weight) or hemin (35 µmol/kg body weight) (n = 4-5). (E) Survival of transfused CFSE-labeled RBCs from hGPA-Tg mice at 1 day in recipient WT and ifnar1^−/− mice treated with anti-hGPA antibody (0.75 µg/20 g body weight, 1 hour after RBC transfusion) or isotype IgG as control, after pretreatment with hemin (35 µmol/kg body weight, 1 day before RBC transfusion) or PBS as control. (F) MFI of CD64 in liver CMo, Ly-6C⁺MHC-II⁺ monocyte, Tim-4⁻ MoMⲫ, and Tim-4⁺ Mⲫ of control and sickle treated with anti-IFNAR1 antibody (2 mg/20 g body weight for 3 consecutive days) or isotype IgG as control (n = 3). (G) Survival of transfused CFSE-labeled hGPA RBCs in mice as shown in panel F at 1 day after receiving transfused RBC and anti-hGPA antibody (0.75 µg/20 g body weight), or isotype IgG as control (n = 5-6). Data represent mean ± SEM; means were compared using 2-tailed Student t test. *P < .05.