Heme triggers TLR4 signaling leading to endothelial cell activation and vaso-occlusion in murine sickle cell disease

Abstract

Treatment of sickle cell disease (SCD) is hampered by incomplete understanding of pathways linking hemolysis to vaso-occlusion. We investigated these pathways in transgenic sickle mice. Infusion of hemoglobin or heme triggered vaso-occlusion in sickle, but not normal, mice. Methemoglobin, but not heme-stabilized cyanomethemoglobin, induced vaso-occlusion, indicating heme liberation is necessary. In corroboration, hemoglobin-induced vaso-occlusion was blocked by the methemoglobin reducing agent methylene blue, haptoglobin, or the heme-binding protein hemopexin. Untreated HbSS mice, but not HbAA mice, exhibited ∼10% vaso-occlusion in steady state that was inhibited by haptoglobin or hemopexin infusion. Antibody blockade of adhesion molecules P-selectin, von Willebrand factor (VWF), E-selectin, vascular cell adhesion molecule 1, intercellular adhesion molecule 1, platelet endothelial cell (EC) adhesion molecule 1, α4β1, or αVβ3 integrin prevented vaso-occlusion. Heme rapidly (5 minutes) mobilized Weibel-Palade body (WPB) P-selectin and VWF onto EC and vessel wall surfaces and activated EC nuclear factor κB (NF-κB). This was mediated by TLR4 as TAK-242 blocked WPB degranulation, NF-κB activation, vaso-occlusion, leukocyte rolling/adhesion, and heme lethality. TLR4(-/-) mice transplanted with TLR4(+/+) sickle bone marrow exhibited no heme-induced vaso-occlusion. The TLR4 agonist lipopolysaccharide (LPS) activated ECs and triggered vaso-occlusion that was inhibited by TAK-242, linking hemolysis- and infection-induced vaso-occlusive crises to TLR4 signaling. Heme and LPS failed to activate VWF and NF-κB in TLR4(-/-) ECs. Anti-LPS immunoglobulin G blocked LPS-induced, but not heme-induced, vaso-occlusion, illustrating LPS-independent TLR4 signaling by heme. Inhibition of protein kinase C, NADPH oxidase, or antioxidant treatment blocked heme-mediated stasis, WPB degranulation, and oxidant production. We conclude that intravascular hemolysis in SCD releases heme that activates endothelial TLR4 signaling leading to WPB degranulation, NF-κB activation, and vaso-occlusion.

Figure 1

Hemolysis and plasma heme liberated from Hb induce stasis in transgenic sickle mice. (A) Percent stasis was measured in the subcutaneous venules of NY1DD, HbSS, HbAS, HbAA, and C57BL/6 mice with DSFCs. Flowing venules were selected and mapped at baseline (20-35 venules per mouse). Mice were given a bolus infusion (0.012 mL/g) of the following treatments at the indicated Hb doses: saline (control), water (to induce hemolysis in vivo), HbA, metHbA, cyanometHbA, methylene blue (2 mg/kg, IV) + HbA, haptoglobin (3.2 µmol/kg, IV) + HbA, hemopexin (1.6 µmol/kg, IV) + HbA, or HbS. Percent stasis was measured using intravital microscopy 1 hour after infusion. The numbers of mice (n) in each treatment group are indicated. Bars are mean % stasis + standard deviation (SD) with mean stasis values written above the bars. (B) Percent stasis was measured in the subcutaneous venules of NY1DD and C57BL/6 mice with DSFCs as described in panel A. Mice were given a bolus infusion (0.012 ml/g) of the following treatments at the indicated heme dosages: heme, hemopexin (0.4 µmol/kg, IV) + heme, and PPIX (40 µmol/kg, IP 60 minutes preheme) + heme. (C) Correlation between percent stasis and total plasma heme concentrations. NY1DD sickle mice (n = 3/treatment) were infused with saline, heme, Hb, metHb or cyanometHb, or exposed to 1 hour of hypoxia (7%O₂) followed by 1 hour of hypoxia-reoxygenation (H/R). Percent stasis and total plasma heme were measured 1 hour after treatments. Values are mean % stasis and mean total plasma heme. (D) Percent stasis was measured in HbSS and HbAA mice in steady state. DSFCs were implanted on day −3. Flowing venules were selected at baseline on day 0. The same venules were reexamined for stasis on days 1 and 2. HbAA mice were untreated. HbSS mice were untreated or infused with human haptoglobin (900 µg/g) or hemopexin (34 µg/g) on days 0, 1, and 2. n = 3 mice per group; *P < .05 for HbSS vs HbSS + haptoglobin, HbSS + hemopexin or HbAA mice.

Figure 1

Hemolysis and plasma heme liberated from Hb induce stasis in transgenic sickle mice. (A) Percent stasis was measured in the subcutaneous venules of NY1DD, HbSS, HbAS, HbAA, and C57BL/6 mice with DSFCs. Flowing venules were selected and mapped at baseline (20-35 venules per mouse). Mice were given a bolus infusion (0.012 mL/g) of the following treatments at the indicated Hb doses: saline (control), water (to induce hemolysis in vivo), HbA, metHbA, cyanometHbA, methylene blue (2 mg/kg, IV) + HbA, haptoglobin (3.2 µmol/kg, IV) + HbA, hemopexin (1.6 µmol/kg, IV) + HbA, or HbS. Percent stasis was measured using intravital microscopy 1 hour after infusion. The numbers of mice (n) in each treatment group are indicated. Bars are mean % stasis + standard deviation (SD) with mean stasis values written above the bars. (B) Percent stasis was measured in the subcutaneous venules of NY1DD and C57BL/6 mice with DSFCs as described in panel A. Mice were given a bolus infusion (0.012 ml/g) of the following treatments at the indicated heme dosages: heme, hemopexin (0.4 µmol/kg, IV) + heme, and PPIX (40 µmol/kg, IP 60 minutes preheme) + heme. (C) Correlation between percent stasis and total plasma heme concentrations. NY1DD sickle mice (n = 3/treatment) were infused with saline, heme, Hb, metHb or cyanometHb, or exposed to 1 hour of hypoxia (7%O₂) followed by 1 hour of hypoxia-reoxygenation (H/R). Percent stasis and total plasma heme were measured 1 hour after treatments. Values are mean % stasis and mean total plasma heme. (D) Percent stasis was measured in HbSS and HbAA mice in steady state. DSFCs were implanted on day −3. Flowing venules were selected at baseline on day 0. The same venules were reexamined for stasis on days 1 and 2. HbAA mice were untreated. HbSS mice were untreated or infused with human haptoglobin (900 µg/g) or hemopexin (34 µg/g) on days 0, 1, and 2. n = 3 mice per group; *P < .05 for HbSS vs HbSS + haptoglobin, HbSS + hemopexin or HbAA mice.

Figure 2

Multiple EC adhesion molecules are required for stasis including mobilization of WPB constituents P-selectin and VWF to the EC surface. (A) Blocking antibodies to adhesion molecules P-selectin, E-selectin, VWF, VCAM-1, α4β1 integrin, ICAM-1, PECAM-1, αVβ3 integrin, thrombomodulin, or control IgG (30 µg per mouse, IV) were infused into NY1DD sickle mice with DSFCs 5 minutes prior to infusion of heme (3.2 µmol/kg). Percent stasis was measured at 1 hour after heme infusion as described in Figure 1. *P < .001 vs control IgG and #P < .001 antithrombomodulin vs anti-P-selectin, E-selectin, VWF, VCAM-1, α4β1 integrin, ICAM-1, PECAM-1, or αVβ3 integrin. (B) HUVECs were fixed in 4% paraformaldehyde and stained for surface P-selectin (green) or VWF (red) after the indicated incubations. Nuclei were counterstained with DAPI (blue). White bars represent 40 µm. Representative cells are shown. Top row: HUVECs in 0.1% FBS were incubated with heme (20 µM) for 15 minutes, HbA or HbS (20 µM heme) for 24 hours, or HbA for 24 hours followed by heme for 15 minutes. Middle row: HUVECs were incubated with heme + hemopexin (20 μM), HbA + haptoglobin (20 µM), or HbA + hemopexin for 24 hours. Bottom row: HUVECs were incubated for 24 hours in 0.1% FBS with or without HbA or HbS followed by H/R (5% CO₂/95% N₂ for 3 hours and 5% CO₂/95% air for 1 hour). Control cells were incubated in 0.1% FBS for 24 hours. Cells were not permeabilized; all pictures represent cell-surface expression. (C) Normal C57BL/6 and NY1DD sickle mice were infused with saline (12 mL/kg, negative control), histamine (1200 µmol/kg, positive control), or heme (3.2 µmol/kg). Fifteen minutes after infusion, mice were sacrificed in CO₂ and dorsal skin (n = 1/treatment) was removed and fixed in Zamboni’s for immunofluorescence staining of VWF (red) and counterstaining of nuclei with DAPI (blue). White bars represent 30 µm. Representative venules are shown.

Figure 2

Multiple EC adhesion molecules are required for stasis including mobilization of WPB constituents P-selectin and VWF to the EC surface. (A) Blocking antibodies to adhesion molecules P-selectin, E-selectin, VWF, VCAM-1, α4β1 integrin, ICAM-1, PECAM-1, αVβ3 integrin, thrombomodulin, or control IgG (30 µg per mouse, IV) were infused into NY1DD sickle mice with DSFCs 5 minutes prior to infusion of heme (3.2 µmol/kg). Percent stasis was measured at 1 hour after heme infusion as described in Figure 1. *P < .001 vs control IgG and #P < .001 antithrombomodulin vs anti-P-selectin, E-selectin, VWF, VCAM-1, α4β1 integrin, ICAM-1, PECAM-1, or αVβ3 integrin. (B) HUVECs were fixed in 4% paraformaldehyde and stained for surface P-selectin (green) or VWF (red) after the indicated incubations. Nuclei were counterstained with DAPI (blue). White bars represent 40 µm. Representative cells are shown. Top row: HUVECs in 0.1% FBS were incubated with heme (20 µM) for 15 minutes, HbA or HbS (20 µM heme) for 24 hours, or HbA for 24 hours followed by heme for 15 minutes. Middle row: HUVECs were incubated with heme + hemopexin (20 μM), HbA + haptoglobin (20 µM), or HbA + hemopexin for 24 hours. Bottom row: HUVECs were incubated for 24 hours in 0.1% FBS with or without HbA or HbS followed by H/R (5% CO₂/95% N₂ for 3 hours and 5% CO₂/95% air for 1 hour). Control cells were incubated in 0.1% FBS for 24 hours. Cells were not permeabilized; all pictures represent cell-surface expression. (C) Normal C57BL/6 and NY1DD sickle mice were infused with saline (12 mL/kg, negative control), histamine (1200 µmol/kg, positive control), or heme (3.2 µmol/kg). Fifteen minutes after infusion, mice were sacrificed in CO₂ and dorsal skin (n = 1/treatment) was removed and fixed in Zamboni’s for immunofluorescence staining of VWF (red) and counterstaining of nuclei with DAPI (blue). White bars represent 30 µm. Representative venules are shown.

Figure 3

Heme-mediated stasis and P-selectin and VWF expression require NOX, PKC, and oxidants. (A) Thirty minutes prior to infusion of heme (3.2 µmol/kg), NY1DD mice (n = 3 per treatment) were injected with NOX inhibitors apocynin (10 mg/kg, IP) or DPI (30 mg/kg, IP), the PKC inhibitor calphostin C (100µg/kg, IP), the antioxidant NAC (2.4 mg/kg, IP), or the iron chelator S-DFO (6 mL/kg, IV, every third day × 4). Controls include vehicle (0.012 mL/g dimethylsulfoxide [DMSO], IP) or hetastarch (6 mL/kg, IV, every third day × 4). Percent stasis was measured using intravital microscopy 1 hour after heme infusion. Bars are mean percent stasis + SD with mean stasis values written above the bars. *P < .05 vs control. (B) HUVECs were pretreated 30 minutes with apocynin (1 mM) or calphostin C (300 nm) followed by the addition of heme (10 µM) for 15 minutes. Control cells were untreated. Cells were fixed and immunostained for VWF (red) on the cell surface. Nuclei were counterstained with DAPI (blue). (C) Heme-induced oxidative stress was measured in HUVECs. Confluent HUVECs in 1% FBS were pretreated for 1 hour with vehicle (DMSO), apocynin (10 mM), DPI (20 µM), calphostin C (200 nM), NAC (10 mM), quercetin (50 µM), or DFO (100 µM). After 1 hour, cells were washed and loaded with the cell-permeable ROS probe DCFH-DA (100 µM) for 30 minutes. After 30 minutes, cells were washed and incubated with heme (20 µM) or LPS (20 ng/mL) in 0.1% FBS media for 4 hours. After 4 hours, cells were washed and ROS production was measured as accumulated cell fluorescence from oxidized DCFH-DA. Values are means + SD (n = 4 or 5 wells per treatment).

Figure 4

Endothelial TLR4 signaling is required for WPB activation and vascular stasis induced by heme. (A) HUVECs were incubated with vehicle (DMSO), heme (10 µM), or heme plus the TLR4-signaling inhibitor TAK-242 (400 nM). After a 15-minute incubation, cells were fixed and stained for surface P-selectin (green) and VWF (red). Nuclei were counterstained with DAPI (blue). Data are representative of at least 3 experiments. (B) mPVECs were isolated and cultured from TLR4^+/+ and TLR4^−/− mice. mPVECs were incubated with heme (10 µM), Hb (10 µM heme), or LPS (10 ng/ml) for 15 minutes (n = 6 wells per treatment). Cells were fixed and surface VWF was measured by enzyme immunoassay (*P < .05 for TLR4^+/+ vs TLR4^−/−). (C) Percent stasis was measured in the subcutaneous venules of NY1DD and C57BL/6 mice with DSFCs as described in Figure 1. Mice were given a bolus infusion (0.012 mL/g) of the indicated treatments at the specified heme doses: vehicle (DMSO/saline, 1:9 vol/vol) + heme (3.2 µmol/kg), TAK-242 (2 mg/kg, IP 30 minutes before heme) + heme, vehicle + LPS (1 mg/kg), TAK-242 + LPS, isotype control IgG (30 µg/mouse) + LPS, anti-LPS IgG (30 µg per mouse) + LPS, isotype control IgG + heme, anti-LPS IgG + heme, vehicle + H/R, TAK-242 + H/R, or LPS alone. The numbers of mice (n) in each treatment group are indicated. Bars are mean % stasis + SD with mean stasis values written above the bars. *P < .01 vs control; #P < .001 C57 + LPS vs NY1DD vehicle + LPS. (D) Chimeric NY1DD β^S/TLR4^−/− and NY1DD β^S/TLR4^+/+ mice were generated by transplanting BM from NY1DD β^S sickle mice into TLR4^−/− or TLR4^+/+ mice (n = 3 recipients per group). C57BL/6 mice were used as TLR4^+/+ mice. Negative control chimeric C57 β^Murine/TLR4^−/− mice were generated by transplanting BM from C57BL/6 mice into TLR4^−/− mice. The presence of human β^S was confirmed by isoelectric focusing (IEF) at 3 months posttransplant. Representative IEF bands for each group are shown (bottom). Stasis was measured in transplanted mice 1 to 2 weeks after IEF determination after IV infusion of Hb (0.32 µmol heme/kg) (*P < .01 NY1DD β^S/TLR4^−/− vs NY1DD β^S/TLR4^+/+). Nontransplanted NY1DD β^S mice (n = 2) served as positive controls. All stasis values are mean percent stasis + SD. (E) Leukocyte rolling flux was measured in venules of NY1DD mice with DSFCs before (baseline) and 1 hour after infusion of heme (3.2 µmol/kg). Half of the mice (n = 4) were treated with TAK-242 after baseline (+TAK-242, 2 mg/kg, IP, 30 minutes before heme). Control mice (n = 4) were untreated with TAK-242 (−TAK-242). Values are mean number of rolling cells per minute + SD. (F) Leukocyte adhesion was measured in the same venules as described in panel C. Values are mean number of adherent cells per 100 µm + SD.

Figure 4

Endothelial TLR4 signaling is required for WPB activation and vascular stasis induced by heme. (A) HUVECs were incubated with vehicle (DMSO), heme (10 µM), or heme plus the TLR4-signaling inhibitor TAK-242 (400 nM). After a 15-minute incubation, cells were fixed and stained for surface P-selectin (green) and VWF (red). Nuclei were counterstained with DAPI (blue). Data are representative of at least 3 experiments. (B) mPVECs were isolated and cultured from TLR4^+/+ and TLR4^−/− mice. mPVECs were incubated with heme (10 µM), Hb (10 µM heme), or LPS (10 ng/ml) for 15 minutes (n = 6 wells per treatment). Cells were fixed and surface VWF was measured by enzyme immunoassay (*P < .05 for TLR4^+/+ vs TLR4^−/−). (C) Percent stasis was measured in the subcutaneous venules of NY1DD and C57BL/6 mice with DSFCs as described in Figure 1. Mice were given a bolus infusion (0.012 mL/g) of the indicated treatments at the specified heme doses: vehicle (DMSO/saline, 1:9 vol/vol) + heme (3.2 µmol/kg), TAK-242 (2 mg/kg, IP 30 minutes before heme) + heme, vehicle + LPS (1 mg/kg), TAK-242 + LPS, isotype control IgG (30 µg/mouse) + LPS, anti-LPS IgG (30 µg per mouse) + LPS, isotype control IgG + heme, anti-LPS IgG + heme, vehicle + H/R, TAK-242 + H/R, or LPS alone. The numbers of mice (n) in each treatment group are indicated. Bars are mean % stasis + SD with mean stasis values written above the bars. *P < .01 vs control; #P < .001 C57 + LPS vs NY1DD vehicle + LPS. (D) Chimeric NY1DD β^S/TLR4^−/− and NY1DD β^S/TLR4^+/+ mice were generated by transplanting BM from NY1DD β^S sickle mice into TLR4^−/− or TLR4^+/+ mice (n = 3 recipients per group). C57BL/6 mice were used as TLR4^+/+ mice. Negative control chimeric C57 β^Murine/TLR4^−/− mice were generated by transplanting BM from C57BL/6 mice into TLR4^−/− mice. The presence of human β^S was confirmed by isoelectric focusing (IEF) at 3 months posttransplant. Representative IEF bands for each group are shown (bottom). Stasis was measured in transplanted mice 1 to 2 weeks after IEF determination after IV infusion of Hb (0.32 µmol heme/kg) (*P < .01 NY1DD β^S/TLR4^−/− vs NY1DD β^S/TLR4^+/+). Nontransplanted NY1DD β^S mice (n = 2) served as positive controls. All stasis values are mean percent stasis + SD. (E) Leukocyte rolling flux was measured in venules of NY1DD mice with DSFCs before (baseline) and 1 hour after infusion of heme (3.2 µmol/kg). Half of the mice (n = 4) were treated with TAK-242 after baseline (+TAK-242, 2 mg/kg, IP, 30 minutes before heme). Control mice (n = 4) were untreated with TAK-242 (−TAK-242). Values are mean number of rolling cells per minute + SD. (F) Leukocyte adhesion was measured in the same venules as described in panel C. Values are mean number of adherent cells per 100 µm + SD.

Figure 5

Heme and LPS induce EC NF-κB activation via TLR4 signaling. (A) HUVECs were incubated with LPS (10 ng/ml), TNF-α (10 ng/ml), heme (10 µM), or TAK-242 (400 nM) + heme for the indicated times (1-8 hours). At the end of the incubation period, nuclear extracts were prepared, run on a western blot, and stained for NF-κB phospho-p65. (B) mPVECs from TLR4^+/+ or TLR4^−/− mice were incubated with heme (10 µM), LPS (10 ng/ml), TNF-α (10 ng/ml), or TAK-242 (400 nM) + heme (10 µM) for 1 hour. At the end of the incubation period, nuclear extracts were prepared, run on western blot, and stained for NF-κB phospho-p65.

Figure 6

Intravenous heme is lethal in HbSS, but not HbAA, mice and inhibition of TLR4 signaling with TAK-242 rescues HbSS mice. All HbSS and HbAA mice (n = 6 mice per treatment) were given a bolus infusion of heme (32 µmol/kg) at time zero. One treatment group received TAK-242 (2 mg/kg, IP) 30 minutes before heme infusion. Time of death after heme infusion was recorded.

Figure 7

Proposed model of heme-induced stasis. Intravascular hemolysis in SCD leads to release of heme from methemoglobin S. Heme release can be blocked by haptoglobin or the liberated heme can be bound and removed by hemopexin. When plasma levels of haptoglobin and hemopexin are depleted, heme can activate TLR4 signaling at the EC membrane independently of LPS, but in manner similar to LPS + LPS binding protein that is dependent on cofactors CD14 and MyD88. TLR4 signaling, which can be inhibited by TAK-242, leads to the activation of NF-κB phospho-p65 and the transcription of NF-κB–responsive genes in ECs including VCAM-1, P-selectin, E-selectin, interleukin-1, interleukin-6, interleukin-8, and tissue factor. TLR4 signaling also activates an acute stress sentinel pathway that involves PKC activation leading to the production of ROS by endothelial NOX and EC degranulation of WPB P-selectin and VWF to the cell surface. Rapid mobilization of P-selectin and VWF to the cell surface, triggers vaso-occlusion. Vaso-occlusion also requires the expression of NF-κB–responsive adhesion molecules VCAM-1, ICAM-1, and E-selectin that are chronically upregulated on the endothelium of transgenic sickle mice in steady state.