Heme induces endothelial tissue factor expression: potential role in hemostatic activation in patients with hemolytic anemia

Authors

  • B. N. Y. SETTY,

    1. Marian Anderson Comprehensive Sickle Cell Anemia Care and Research Center, Department of Pediatrics, Division of Research Hematology, Jefferson Medical College, Thomas Jefferson University, Philadelphia, PA, USA
    Search for more papers by this author
  • S. G. BETAL,

    1. Marian Anderson Comprehensive Sickle Cell Anemia Care and Research Center, Department of Pediatrics, Division of Research Hematology, Jefferson Medical College, Thomas Jefferson University, Philadelphia, PA, USA
    Search for more papers by this author
  • J. ZHANG,

    1. Marian Anderson Comprehensive Sickle Cell Anemia Care and Research Center, Department of Pediatrics, Division of Research Hematology, Jefferson Medical College, Thomas Jefferson University, Philadelphia, PA, USA
    Search for more papers by this author
  • M. J. STUART

    1. Marian Anderson Comprehensive Sickle Cell Anemia Care and Research Center, Department of Pediatrics, Division of Research Hematology, Jefferson Medical College, Thomas Jefferson University, Philadelphia, PA, USA
    Search for more papers by this author

B. N. Yamaja Setty, Department of Pediatrics, Thomas Jefferson University, Medical College Building, Suite #727, 1025 Walnut Street, Philadelphia, PA 19107, USA.
Tel.: +1 215 955 9821; fax: +1 215 955 8011.
E-mail: yamaja.setty@jefferson.edu

Abstract

Summary. Objectives: We explored the possibility that heme, an inflammatory mediator and a product of intravascular hemolysis in patients with hemolytic anemia including sickle cell disease, could modulate hemostasis by an effect on endothelial tissue factor (TF) expression. Methods: Levels of TF mRNA, protein and procoagulant activity were measured in heme-treated endothelial cells. Results: Heme induces TF expression on the surface of both macrovascular and microvascular endothelial cells in a concentration-dependent manner, with 12-fold to 50-fold induction being noted (enzyme-linked immunosorbent assay) between 1 and 100 μm heme (P < 0.05). Complementary flow cytometry studies showed that the heme-mediated endothelial TF expression was quantitatively similar to that of tumor necrosis factor-alpha (TNF-α). Heme also upregulated the expression of TF mRNA (8-fold to 26-fold), protein (20-fold to 39-fold) and procoagulant activity (5-fold to 13-fold) in endothelial cells in a time-dependent manner. The time-course of heme-mediated TF antigen expression paralleled the induction of procoagulant activity, with antibody blocking studies demonstrating specificity for TF protein. Interleukin (IL)-1α, and TNF-α are not involved in mediating the heme effect, as antibodies against these cytokines and IL-1-receptor antagonist failed to block heme-induced TF expression. Inhibition of heme-induced TF mRNA expression by sulfasalazine and curcumin suggested that the transcription factor nuclear factor kappaB is involved in mediating heme-induced TF expression in endothelial cells. Conclusions: Our results demonstrate that heme induces TF expression by directly activating endothelial cells, and that heme-induced endothelial TF expression may provide a pathophysiologic link between the intravascular hemolytic milieu and the hemostatic perturbations previously noted in patients with hemolytic anemia including sickle cell disease.

Introduction

Numerous studies have identified that hemostasis is perturbed in patients with hemolytic anemias including sickle cell disease (SCD). Hemostatic abnormalities observed in SCD include evidence for enhanced thrombin generation [1], increased circulatory levels of tissue factor (TF) antigen and procoagulant activity [2,3], and circulating endothelial cells and endothelial- and monocyte-derived microparticles that are positive for TF [4,5]. TF, a cell surface receptor for factor VII/VIIa, is the physiologic initiator of blood coagulation, forming a complex with circulating FVIIa, and activating FX with subsequent thrombin generation [6]. Although TF is not constitutively expressed on endothelium or monocytes in vivo, its expression can be induced by a variety of agonists with pathophysiologic relevance to hemolytic anemias, including the inflammatory cytokines interleukin (IL)-1 and tumor necrosis factor-alpha (TNF-α), hypoxia, shear stress, growth factors, endotoxin and reperfusion injury [7–9].

Erythrocytes, which undergo hemolysis intravascularly, release heme and hemoglobin into the circulation, with documented studies suggesting that both heme and cell-free hemoglobin may modulate disease severity in patients with SCD by decreasing the bioavailability of the cytoprotective mediator nitric oxide (NO) [10–12]. Other studies have shown that heme can activate endothelial cells in vitro, upregulating the expression of surface adhesion molecules such as vascular cell adhesion molecule-1 (VCAM-1), intercellular adhesion molecule-1 (ICAM-1) and E-selectin [13]. As many cytokine-induced downstream signaling events in endothelial cells are common to both TF induction and adhesion molecule expression, we hypothesized that heme could affect hemostasis by modulating endothelial TF expression. We report that heme at pathologically relevant concentrations induces TF mRNA and protein expression in endothelial cells. We show that TF produced in response to heme is expressed on the cell surface, and that it is functional.

Materials and methods

Materials

Three TF-specific antibodies, including a rabbit polyclonal antibody and two mouse monoclonal (clones hTF-1 and TF9-10H10) antibodies, were used in this study. The rabbit polyclonal antibody and the mouse monoclonal antibody hTF-1 were kindly provided by R. Bach (VA Medical Center, Minneapolis, MN, USA), and clone TF9-10H10 was obtained from Calbiochem (La Jolla, CA, USA). Details of these antibodies and other reagents used are described in Supporting information.

Culture of endothelial cells and THP-1 cells

Human umbilical vein endothelial cells (HUVECs) were obtained from Clonetics (Lonza, Walkersville, MD, USA) and cultured according to the manufacturer’s protocol. Cells at passages 2–4 were used. All experiments were performed using endothelial growth medium containing 50 μg mL−1 polymyxin-B (EGM-polyB), unless otherwise indicated. Representative experiments also were performed using human lung microvascular endothelial cells (HLMECs; passages 2–4) from Clonetics. Endothelial cells were incubated in the absence or presence of the desired agonist (1–100 μm heme, 10 ng mL−1 IL-1α, or 10 ng mL−1 TNF-α) for various times (0.5–8 h), and then analyzed for TF mRNA, TF protein or TF procoagulant activity. As a positive control for inducible TF expression, we used THP-1 cells (a human monocytic leukemia cell line; ATCC, Manassas, VA, USA), activated with lipopolysaccharide (LPS) in the absence of polymyxin-B.

Preparation of heme

Fresh working stocks of 10 mm heme were prepared in 0.05 m NaOH, diluted to the desired concentration in EGM-polyB, and pH adjusted to 7.4. Endotoxin-free water was used in the preparation of all reagents. Endotoxin levels in reagents and media were assessed using a limulus assay kit (Lonza).

Analysis of endothelial TF protein

Cell surface TF expression was assessed using two complementary methods, including an enzyme-linked immunosorbent assay (ELISA)-based assay (intact cell monolayer) and flow cytometry (cell suspension), as previously described for endothelial adhesion receptors [14]. Total TF protein level was assessed by using western blotting of cellular proteins. The monoclonal anti-TF antibodies, TF9-10H10 and hTF-1, were used in ELISA and/or flow cytometry procedures, and the rabbit polyclonal anti-TF antibody was employed in western blot analysis (for details refer to Supporting information).

Analysis of endothelial TF mRNA

TF mRNA levels in control and treated endothelial cultures were measured using a two-step semiquantitative reverse transcription polymerase chain reaction (RT-PCR) assay employing the primers 5′-TGT-GAC-CGT-AGA-AGA-TGA-ACG-GAC-3′ (forward) and 5′-CCA-CTC-CTG-CCT-TTC-TAC-ACT-TGT-3′ (reverse), which yielded a 380-bp PCR product for TF. The identity of the PCR fragment from the endothelial TF transcript was confirmed by sequencing the purified PCR fragment, and matching the sequence with published TF mRNA (for details refer to Supporting information).

TF functional activity

TF procoagulant activity on intact endothelial monolayers or in cell lysates prepared from control and treated cultures was measured using a TF activity assay kit (American Diagnostica, Stanford, CT, USA). To confirm that the conversion of FX to FXa was TF-mediated, in parallel experiments either the cell monolayers (∼ 50 000 cells) or endothelial lysates (2.5 μg of protein) were preincubated with the monoclonal anti-TF antibody hTF-1 (15 μg) for 30 min, and assayed for TF activity.

Statistical analysis

Statistical evaluation was performed using Sigmastat (Jandel, San Rafael, CA, USA). All values presented are mean ± standard deviation. Multiple group comparison was done using either one-way anova or the Kruskal–Wallis test, as appropriate, and if significant, group-wise comparisons were performed with the Bonferroni or the Dunn test. Differences at P < 0.05 were considered to be statistically significant. Paired-group comparison was performed using either the paired t-test or the Mann–Whitney rank sum test, as appropriate.

Results

Effects of heme on endothelial surface TF antigen expression

Using flow cytometry and ELISA-based assays, we investigated whether heme induced endothelial surface TF expression. As shown in Fig. 1A, heme upregulated TF expression on intact HUVECs in a concentration-dependent manner, with 12-fold, 20-fold and 50-fold induction being noted at 1, 10 and 100 μm heme, respectively, as compared to media controls as assessed by ELISA. The effects noted at all heme concentrations were statistically significant at P < 0.05 or P < 0.01. A similar concentration-dependent effect of heme on TF expression also was noted on microvascular endothelial cells, using HLMECs as a representative cell system (Fig. 1B). In parallel experiments, the positive control TNF-α increased TF expression by ∼ 80-fold on both macrovascular and microvascular endothelial cells. Cell surface TF expression was also confirmed using complementary flow cytometry employing two different monoclonal anti-TF antibodies: TF9-10H10 (Fig. 1C) and hTF-1 (Fig. 1D). The results demonstrate that significant numbers of cells acquired TF-positivity following activation with heme (22% and 21% of cells with TF9-10H10 and hTF-1, respectively). Cells stimulated with the positive control TNF-α exhibited 27% and 26% TF-positivity, respectively.

Figure 1.

 Effects of heme and tumor necrosis factor-alpha (TNF-α) on surface expression of tissue factor (TF) protein on endothelial cells. (A, B) Analysis by enzyme-linked immunosorbent assay (ELISA). Following incubation of human umbilical vein endothelial cells (HUVECs) (A) or human lung microvascular endothelial cells (B) for 4 h in the presence or absence of the indicated concentrations of heme or TNF-α, surface TF expression was assessed by ELISA. Results presented are the means ± standard deviation from six (A) or three (B) experiments. All experiments presented were performed with unfixed cells. We also performed ELISA assays using fixed cells, and found that the TF expression profile in response to activation with heme and cytokines was similar to that of unfixed cell preparations (i.e. increased TF expression in activated endothelial cells as compared to unstimulated medium control). Increased expression of TF on unstimulated control cells was, however, noted on unfixed cells (an increase of approximately 20%) as compared to fixed cells. *P < 0.05 or P < 0.01 as compared to media control. (C, D) Analysis by flow cytometry. Following incubation of HUVECs for 4 h in the absence (Ci/Di) or presence of 10 ng mL−1 TNF-α (Cii/Dii) or 100 μm heme (Ciii/Diii), cells were harvested, labeled with anti-TF antibody TF9-10H10 (C) or hTF-1 (D), or an equivalent amount of an isotype-matched negative immunoglobulin control, and analyzed. Histogram profiles of fluorescein isothiocyanate fluorescence from cells labeled with anti-TF antibody (gray histogram) and isotype-matched negative control (solid black line) are shown. The positive histogram region, M1, is defined using cells labeled with the isotype control. Positivity, shown in each panel, is the difference in M1 between the cells labeled with anti-TF and the isotype control. Results presented are from a representative experiment repeated three times with similar results.

Effects of heme on endothelial TF mRNA expression

In preliminary experiments, we found that heme induced TF mRNA expression in a concentration-dependent manner, with 8-fold, 15-fold and 26-fold upregulation being noted at 1, 10 and 100 μm heme, respectively. As maximal effects of heme on both TF mRNA and TF protein expression (Fig. 1) were found at 100 μm, our subsequent experiments were performed using 100 μm heme. This agonist induced TF mRNA expression in a time-dependent manner, with maximal responses between 2 and 3 h (Fig. 2A). TF induction in response to the positive controls IL-1α and TNF-α occurred earlier when compared to heme, with TF mRNA being already present in endothelial cells within 30 min of cytokine exposure. Profiles of TF mRNA expression in endothelial cells and THP-1 cells are shown in Fig. 2B. The results show that total RNA from activated endothelial cells and THP-1 cells contained TF mRNA yielding an RT-PCR fragment of 380 bp. The identity of the endothelial PCR fragment was confirmed by DNA sequence analysis, which showed 100% identity with the published TF mRNA sequence. Densitometric analyses demonstrated that incubation of endothelial cells with 100 μm heme for 2 h upregulated TF mRNA levels by approximately 17-fold (n = 4) as compared to unstimulated control cells (Fig. 2B), whereas increases of 29-fold and 48-fold were noted in cells incubated with TNF-α and IL-1α, respectively (Fig. 2B). The effects of polymyxin-B on LPS-, heme- and IL-1α-induced endothelial TF expression were evaluated. Whereas polymyxin-B had no inhibitory effect on either heme- or IL-1α-induced endothelial TF expression, LPS-induced (1 μg mL−1) TF mRNA expression was inhibited by 86% (Fig. S1). In addition, the heme used in our experiments was free of any detectable endotoxin as assessed by the limulus assay (< 0.01 EU mL−1 in test medium containing 100 μm heme).

Figure 2.

 Effects of heme and cytokines on tissue factor (TF) mRNA and protein levels in human umbilical vein endothelial cells (HUVECs). (A) Time-dependent changes in TF mRNA levels. Following incubation of HUVECs with the desired agonist [10 ng mL−1 tumor necrosis factor-alpha (TNF-α), 10 ng mL−1 interleukin-1alpha (IL-1α), or 100 μm heme] for the times indicated, total RNA was isolated, and analyzed for TF mRNA by reverse transcription polymerase chain reaction (RT-PCR). The experiment was repeated twice with similar results. (B) Heme and cytokines upregulate endothelial TF mRNA expression. HUVECs were incubated in the absence (lane 1) or presence of 10 ng mL−1 TNF-α (lane 2), 10 ng mL−1 IL-1α (lane 3), or 100 μm heme (lane 4) for 2 h, and total RNA was then isolated and analyzed for TF mRNA. RT-PCR products of TF mRNA from THP-1 cells (a positive control for TF induction) incubated in the absence (lane 5) or presence (lane  6) of 100 ng mL−1 lipopolysaccharide (LPS) are shown for comparison. β-Actin mRNA, a constitutively expressing transcript, was coamplified with TF mRNA as an endogenous control for TF mRNA quantitation, using previously described primers [14]. The RT-PCR products of TF mRNA (380 bp) and β-actin mRNA (687 bp) from a representative gel are shown at the top. Band intensities of PCR fragments were determined densitometrically and expressed as a ratio of TF to β-actin mRNAs. The message ratio in the unstimulated medium control was 0.26 ± 0.12. Values (expressed as fold change as compared to media control) are the means ± standard deviation (SD) from four experiments. *P < 0.05 as compared to control. (C) Time-dependent changes in TF protein levels. Following incubation of HUVECs with the desired agonist (10 ng mL−1 TNF-α, 10 ng mL−1 IL-1α, or 100 μm heme) for the times indicated, total cellular proteins were isolated, and analyzed for TF protein by western blotting. The experiment was repeated three times with similar results. (D) Heme and cytokines upregulate endothelial TF protein expression. HUVECs were incubated in the absence (lane 1) or presence of 10 ng mL−1 TNF-α (lane 2), 10 ng mL−1 IL-1α (lane 3) or 100 μm heme (lane 4) for 4 h and then analyzed for TF protein. Immunoblots from THP-1 cells incubated in the absence (lane 5) or presence (lane 6) of 100 ng mL−1 LPS are shown for comparison. Equal protein loading was checked by reprobing the stripped immunoblots for β-tubulin protein. TF and β-tubulin (βT) protein bands from representative immunoblots are shown at the top. Intensities of protein bands were determined densitometrically and expressed as a ratio between TF and β-tubulin protein. This ratio in the unstimulated medium control was 0.06 ± 0.06. Values (expressed as fold change as compared to media control) are the means ± SD from four experiments. *P < 0.05 as compared to control.

Effects of heme on endothelial total TF protein levels

Heme induced TF protein expression in a time-dependent manner, with 20-fold to 39-fold stimulation over controls being noted between 4 and 7 h, and a peak response at 4 h similar to that noted with cytokines. The latter agonists, however, induced significant protein expression by 2 h (Fig. 2C). Western blots from representative experiments showing the profiles of TF protein and β-tubulin (the loading control) from endothelial cells and THP-1 cells are shown in Fig. 2D. The results show that only activated endothelial cells and THP-1 cells expressed TF protein. Densitometric analyses of the blots showed that incubation of endothelial cells with 100 μm heme for 4 h upregulated TF protein expression by approximately 35-fold (n = 4), as compared to 70-fold and 88-fold increases in cells incubated with TNF-α and IL-1α, respectively (Fig. 2D).

Effects of heme on endothelial TF functional activity

TF expressed in endothelial cells in response to heme (100 μm) was functionally active, as assessed by the ability of cell lysates to convert FX to FXa in the presence of exogenously provided FVIIa. Heme-treated endothelial cells expressed TF procoagulant activity in a time-dependent manner, with 5-fold to 13-fold increases being seen over control between 4 and 7 h, with peak activity at 4 h (Fig. 3A). The time-course of TF activity paralleled the time-course of TF antigen expression in heme-treated cells (Figs 3A and 2C). Similar time-course profiles were noted with TNF-α and IL-1α (Fig. 3A). Blocking experiments with anti-TF (Fig. 3B) showed that more than 90% of procoagulant activity in cells treated with all three agonists (heme, TNF-α, and IL-1α) was specifically TF-associated. As TF procoagulant activity in cell lysates reflects total cellular activity, including cell surface, encrypted and intracellular TF, we further evaluated whether heme induces the expression of functional procoagulant TF on intact endothelium. The results show that both heme and cytokines induced cell surface TF procoagulant activity, with 2.6-fold, 7.2-fold and 11.9-fold increases being noted over media controls (Fig. 3C). Measured surface procoagulant activity was, however, < 10% of the total activity in both heme-activated and cytokine-activated endothelial cells (Fig. 3B,C).

Figure 3.

 Effects of heme and cytokines on endothelial tissue factor (TF) functional activity. (A) Time-dependent changes in TF procoagulant activity. Following incubation of human umbilical vein endothelial cells (HUVECs) with the desired agonist [10 ng mL−1 tumor necrosis factor-alpha (TNF-α), 10 ng mL−1 interleukin-1alpha (IL-1α), or 100 μm heme] for the times indicated, total cellular proteins were extracted and analyzed for TF procoagulant activity. The experiment was repeated three times with similar results. (B) Heme and cytokines upregulate endothelial TF procoagulant activity in cell lysates. Following incubation of HUVECs for 4 h with the desired agonist (10 ng mL−1 TNF-α, 10 ng mL−1 IL-1α, or 100 μm heme), cell lysates were prepared. Total cellular proteins (2.5 μg) were incubated in the presence (black bars) or absence (cross-hatched bars) of a mouse monoclonal anti-TF antibody (hTF-1, 15 μg) for 30 min, and then assayed for TF functional activity. Results are the means ± standard deviation from five experiments. **P < 0.05 or P < 0.01 as compared to media control. *P < 0.005 by paired t-test compared to respective controls with no antibody treatment. (C) Heme and cytokines upregulate endothelial TF procoagulant activity on intact cell monolayers. Following activation of HUVECs (∼ 50 000 cells in wells of a 96-well plate) for 4 h with the desired agonist (10 ng mL−1 TNF-α, 10 ng mL−1 IL-1α, or 100 μm heme), monolayers were incubated in the presence (black bars) or absence (cross-hatched bars) of a mouse monoclonal anti-TF antibody (hTF-1, 15 μg) for 30 min, and then assayed for TF functional activity. The results presented are from a representative experiment repeated twice with similar results.

Effects of curcumin and sulfasalzine on heme-induced TF mRNA expression

In an attempt to assess whether the transcription factor nuclear factor kappaB (NF-κB) was involved in mediating heme-induced TF expression, we evaluated the effects of inhibitors of NF-κB activation, including sulfasalazine and curcumin [15,16]. Following pretreatment of endothelial cells for 30 min with sulfasalazine (0.2 mm) or curcumin (10 μm), monolayers were incubated for 2 h in the absence or presence of 100 μm heme. The results presented in Fig. 4 show that both sulfasalazine and curcumin inhibited heme-induced TF mRNA expression in endothelial cells, suggesting that NF-κB is involved in mediating heme-induced TF expression.

Figure 4.

 Effects of sulfasalazine and curcumin on heme-induced endothelial tissue factor (TF) mRNA expression. Following pretreatment of human umbilical vein endothelial cells (HUVECs) for 30 min with sulfasalazine (SULFA) (0.2 mm, lane 3) or curcumin (CURC) (10 μm, lane 4), monolayers were incubated for 2 h in the absence (lane 1) or presence (lanes 2–4) of 100 μm heme. Total RNA was extracted and analyzed for TF mRNA. β-Actin mRNA was coamplified with TF mRNA as an endogenous control for TF mRNA quantitation. The reverse transcription polymerase chain reaction products of TF mRNA (380 bp), and β-actin mRNA (687 bp) are shown in the top corner. Bars represent densitometric readings of the ratio between TF and β-actin messages. Results presented are from a representative experiment repeated twice with similar results.

Discussion

We explored the possibility that heme, a product of intravascular hemolysis in patients with hemolytic anemias, could modulate hemostasis by an effect on endothelial TF expression. Our results show that heme, at pathologically relevant concentrations, induces TF expression (both mRNA and protein) in endothelial cells, and that the TF is biologically active.

TF, a 47-kDa transmembrane glycoprotein, the cellular receptor and cofactor for FVII/VIIa, and the predominant initiator of blood coagulation [6], can be expressed on activated monocytes and endothelial cells [7]. Expression on the endothelium can be induced by a variety of agonists, including the inflammatory cytokines IL-1 and TNF-α [7]. Whereas multiple nuclear transcription factors are involved in TF induction in endothelial cells, the agonists IL-1α and TNF-α mediate their effects through activation of activator protein-1 (AP-1) and NF-κB [7,17]. As heme, like IL-1 and TNF-α, appears to activate NF-κB, inducing the expression of the adhesion molecules ICAM-1, VCAM-1 and E-selectin on endothelial cells in vitro [13,18], we hypothesized that heme could modulate TF expression. We showed that heme induces expression of both TF mRNA and functionally active TF protein with time-courses not too dissimilar to that of IL-1α and TNF-α (Figs 2A,C and 3A), although the cytokine response occurred somewhat earlier. Furthermore, the heme effect on endothelial TF expression was not mediated via release of the endogenous cytokines IL-1α and TNF-α from the test endothelial cells (see Supporting information). In vitro studies have shown that both sulfasalazine and curcumin inhibit NF-κB activation in epithelial cells and endothelial cells [15,16], and curcumin, in addition, may affect activation of AP-1, Egr-1 and SP-1 [16]. In preliminary experiments, we further assessed whether heme-induced TF expression was modulated by sulfasalazine and curcumin. Both modulators blocked heme-induced TF mRNA expression (Fig. 4), suggesting that NF-κB is one of the transcription factors involved in mediating the observed heme effects. In this context, it is interesting to note that a recent study by Hasan and Schafer demonstrated that induction of Egr-1 expression by heme caused TF expression in vascular smooth muscle cells [19]. Using flow cytometry and ELISA, we also show that heme induces cell surface expression of TF antigen on both macrovascular and microvascular endothelial cells (Fig. 1A and B), and that the cell surface expression of TF induced by heme was comparable to that induced by TNF-α (Fig. 1C,D). As plasma heme levels in patients with intravascular hemolytic anemias such as SCD and paroxysmal nocturnal hemoglobinuria (PNH) have been documented to range between 20 and 600 μm [10,20], and as the significant changes in TF expression reported here were noted at heme concentrations as low as 1 μm, the heme effects observed in our study appear to be pathologically relevant. In addition, their relevance may also be extended to situations associated with acute elevations of levels of plasma hemoglobin, such as cardiopulmonary bypass. These conditions carry an increased thrombotic risk [20–22].

As endotoxin can induce TF expression in endothelial cells even at very low concentrations, the use of endotoxin-free reagents and media in our experiments was crucial to delineate the heme effect. Previous studies have shown that polymyxin-B inhibits endotoxin-induced cell activation, including TF expression, in human aortic endothelial cells [23]. We have therefore performed all studies in media containing polymyxin-B. In addition, in preliminary experiments, we evaluated the effects of polymyxin-B on agonist-induced TF expression, and confirmed that this agent blocked LPS-mediated, but not heme-induced or cytokine-induced, TF expression in HUVECs (Fig. S1). Moreover, limulus assay on the media used in our studies showed that they were free of any detectable endotoxin. We further evaluated whether heme (like hemoglobin [24]) could potentiate LPS-induced TF expression. We found no such enhancement (see Supporting information).

Although both laboratory and clinical findings show enhanced thrombin generation in patients with hemolytic anemia including SCD, the factor(s) that mediate these abnormalities have not been identified. Our results show that heme induces TF expression at concentrations that are potentially achievable in the microcirculation. Gladwin and coworkers have shown that intravascular hemolysis produces a state of resistance to endogenously produced, or exogenously delivered NO, owing to the stoichiometric oxidation of endogenous NO by cell-free plasma hemoglobin with resultant inhibition of vasodilatation [11]. Decreased NO bioavailability, with consequential impairment of the salutary effects of NO-induced inhibition of platelet aggregation, may be operative in the recently reported enhancement of platelet activation in SCD-related pulmonary hypertension [25]. Our findings serve to further expand the paradigm of hemolysis-induced NO-dependent endothelial dysfunction to additionally include abnormalities in hemostasis based on heme-induced perturbations in fluid-phase hemostasis. In this context, it is interesting to note that the study by Solovey et al. [9] has demonstrated pulmonary endothelial TF expression in SCD transgenic mice following experimental reperfusion injury, a model for SCD-related vaso-occlusion. Although no studies to date have reported a cause and effect relationship between intravascular hemolysis and thrombogenic risk in patients with hemolytic anemia, a previous preliminary report cited an association between the prothrombotic fragment F1.2 and abnormal transcranial Doppler flow velocities in children with SCD [26]. In addition, a recent study by deLatour et al. addressed the thrombotic history of 460 patients with PNH [21]. These authors reported episodes of thromboses in 116 of 454 accessible patients, including the Budd–Chiari syndrome, thrombosis of the central nervous system, and leg thrombosis suggestive of an association between hemolysis and hemostatic perturbations. Other relevant heme effects reported to date include modulation of inflammatory tone by activation of neutrophils, triggering their oxidative burst, and the production of reactive oxygen species [18,27,28]. Our results, taken together with other heme-related published findings may thus provide a critical link between the hemolytic milieu of SCD, coagulation activation and inflammation, and further reiterate that interventions aimed at minimizing intravascular hemolysis may provide therapeutic benefit.

Addendum

B. N. Y. Setty was involved in hypothesis formulation, designed the study, supervised and performed selected experiments, performed statistical analyses of the data, and wrote the paper. S. G. Betal performed mRNA-related work and provided assistance with western blot analyses. J. Zhang provided assistance with western blot analyses. M. J. Stuart was involved in hypothesis formulation, reviewed the study design and data, and provided a critical review of the manuscript.

Acknowledgements

We thank S. Surrey for his critical review of the manuscript, and R. Bach for the gift of TF antibodies. P. Setty provided secretarial assistance and prepared illustrations. This work was supported by grants U54 HL70585 and R01 HL73944 to B. N. Y. Setty and M. J. Stuart from the National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.

Ancillary