Jump to
Abstract
Heme oxygenase (HO)-1 (encoded by Hmox1) catalyzes the oxidative degradation of heme to biliverdin and carbon monoxide. HO-1 is induced during inflammation and oxidative stress to protect tissues from oxidative damage. Because intravascular thrombosis forms at sites of tissue inflammation, we hypothesized that HO-1 protects against arterial thrombosis during oxidant stress. To investigate the direct function of HO-1 on thrombosis, we used photochemical-induced vascular injury in Hmox1−/− and Hmox1+/+ mice. Hmox1−/− mice developed accelerated, occlusive arterial thrombus compared with Hmox1+/+ mice, and we detected several mechanisms accounting for this antithrombotic effect. First, endothelial cells in Hmox1−/− arteries were more susceptible to apoptosis and denudation, leading to platelet-rich microthrombi in the subendothelium. Second, tissue factor, von Willebrand Factor, and reactive oxygen species were significantly elevated in Hmox1−/− mice, consistent with endothelial cell damage and loss. Third, following transplantation of Hmox1−/− donor bone marrow into Hmox1+/+ recipients and subsequent vascular injury, we observed rapid arterial thrombosis compared with Hmox1+/+ mice receiving Hmox1+/+ bone marrow. Fourth, inhaled carbon monoxide and biliverdin administration rescued the prothrombotic phenotype in Hmox1−/− mice. Fifth, using a transcriptional analysis of arterial tissue, we found that HO-1 determined a transcriptional response to injury, with specific effects on cell cycle regulation, coagulation, thrombosis, and redox homeostasis. These data provide direct genetic evidence for a protective role of HO-1 against thrombosis and reactive oxygen species during vascular damage. Induction of HO-1 may be beneficial in the prevention of thrombosis associated with vascular oxidant stress and inflammation.
HO-1 is an inducible enzyme with broad tissue expression that is upregulated in response to oxidant stress and inflammatory stimuli and preserves vascular homeostasis.1–3 HO-1 protects tissues during inflammatory stress4,5 through degradation of prooxidative heme,6 production of bilirubin7 and carbon monoxide (CO),8 and regulation of cellular iron.9,10 In contrast, HO-1 inhibition is associated with tissue pathology in atherosclerosis11–13 and other inflammatory conditions associated with intravascular thrombosis such as septic shock,14 hypoxia,15 and graft rejection.16–18 HO-1 and its byproducts regulate inflammatory responses through repression of proinflammatory genes,19,20 suggesting a role for HO-1 in the regulation of thrombosis associated with inflammation.
At a cellular level, HO-1 prevents apoptosis by inhibiting reactive oxygen species (ROS) formation21 and attenuates platelet activation in vitro.22 Vascular smooth muscle cell–derived CO, generated by HO-1, attenuates platelet activation in vitro by increasing platelet cGMP,23 and platelets themselves possess functionally active HO-1.24 Deletion of Hmox1 in humans is characterized by severe and persistent endothelial cell (EC) damage with alterations in coagulation and fibrinolysis, including elevations in circulating von Willebrand Factor (vWF).25 Inhibition of HO-1 during pulmonary ischemia/reperfusion results in additive increases in plasminogen activator inhibitor-1 and fibrin deposition.26
These studies led us to hypothesize that HO-1 would prevent intravascular thrombosis associated with EC damage. We used a genetic model in vivo to directly investigate the role of HO-1 on arterial thrombosis in Hmox1−/− mice. We demonstrate that Hmox1 deletion leads to accelerated thrombosis by several mechanisms, including endothelial disruption and apoptosis, platelet activation, elevations in tissue factor (TF) and vWF, and a failure of HO-1–deficient bone marrow (BM)-derived progenitor cells to prevent thrombosis. This prothrombotic phenotype in Hmox1−/− mice is rescued by CO inhalation and biliverdin administration. Our data indicate that HO-1 directly preserves endothelial integrity and prevents arterial thrombosis during vascular injury.
Materials and Methods
An expanded Materials and Methods section is in the online data supplement at http://circres.ahajournals.org.
Carotid Artery Injury
Photochemical injury was induced in carotid arteries as described27 with approval from the NIH Animal Care and Use Committee.
Tissue Analysis
Transmission electron microscopy was performed on carotid segments using ultrathin sections and examined with a Nikon Eclipse E800 light microscope. An terminal deoxyribonucleotidyl transferase–mediated TUNEL assay was used to detect DNA fragmentation and apoptosis in situ. Protein carbonylation concentration in murine carotid arteries was determined using the Biocell PC test kit (BioCell Corp, Auckland, New Zealand).
Clinical Laboratory Measurements
Prothrombin time, complete blood count, and bilirubin measurements were made using citrated murine plasma or heparin. Bleeding times were evaluated as described previously.28 TF concentration in carotid arteries was determined using the Actichrome TF kit (American Diagnostica, Standford, Conn). Murine plasma was subjected to ELISA using a polyclonal Asserachrom vWF kit (Diagnostica Stago Inc, Parsippany, NJ), a high-affinity means for evaluating circulating murine vWF28 (Molecular Innovations Inc, Southfield, Mich). In vitro platelet aggregation assay was performed on citrated whole blood samples obtained from the inferior vena cava.
p38 Mitogen-Activated Protein Kinase Inhibitor and Biliverdin Treatment
The p38 mitogen-activated protein kinase (MAPK) inhibitor SKF-86002 was injected subcutaneously 30 minutes before photochemical injury.29 Biliverdin was injected intraperitoneally 3 hours before photochemical injury.30
BM Transplantation
BM, derived from Hmox1−/− and Hmox1+/+ mice, was injected intravenously into the tail veins of irradiated recipient mice. Twelve weeks later, successful engraftment was confirmed by quantitative PCR, and engrafted animals underwent photochemical injury.
CO Inhalation
CO inhalation studies were performed in chambers with a 0.4% CO/balanced air mixture blended with balanced air on site to obtain a stable chamber concentration of 500 ppm CO and O2 level of 20% to 21%. CO and O2 levels were continuously monitored inside the chamber for 24 hours or 15 minutes before or during photochemical injury.
Gene Expression Analysis
RNA was isolated and amplified as described.31 Microarray data processing and analysis were conducted on hybridized Affymetrix chips using Affymetrix GCOS version 1.4.32 Transformed data were subjected to a principal component analysis to detect outliers. Paired and unpaired t tests were performed to detect differentially expressed genes between Hmox1−/− and Hmox1+/+ injured and noninjured arteries. To address the multiple comparisons, fold-cut off filters and false-discovery rate analysis filters were applied.31 Two-way hierarchical clustering was used to bring together sets of samples and genes with similar expression patterns. Pathway analysis was performed using Ingenuity Pathway Analysis (Ingenuity Systems Inc, Redwood City, Calif). Gene expression was validated by quantitative real-time PCR.
Statistics
Statistical analyses were performed using ANOVA or t test as appropriate. Results are expressed as means±SEM and were considered significant at P<0.05.
Results
Deletion of Hmox1 Accelerates Thrombus Formation
We hypothesized that HO-1 would prevent intravascular thrombosis during vascular damage. To test this hypothesis, we used an established photochemical injury model and measured time to occlusive thrombosis in the carotid arteries of Hmox1−/−, Hmox1+/−, and Hmox1+/+ mice. Hmox1+/+ carotid arteries occluded at 54.7±3.3 minutes, whereas thrombus formed in Hmox1−/− arteries in a significantly shorter time, 20.9±3.4 minutes (P<0.001). Hmox1+/− mice demonstrated an intermediate time to thrombosis, 39.9±3.4 minutes (P<0.05) (Figure 1A). Occlusive thrombosis was confirmed by hematoxylin/eosin staining of injured vessels (Figure 1B).
Figure 1. Accelerated thrombosis in Hmox1−/− mice. A, Time to arterial occlusion. Carotid arteries from Hmox1+/+, Hmox1+/−, and Hmox1−/− mice were injured, and time to stable cessation of blood flow was recorded. *P<0.05, Hmox1+/− vs Hmox1+/+ and Hmox1+/− vs Hmox1−/−; ***P<0.001, Hmox1−/− vs Hmox1+/+ (n=20 Hmox1+/+; n=8 Hmox1+/−; and n=7 Hmox1−/− mice). B, Demonstration of occlusive thrombosis. Representative arterial thrombus in a carotid artery of a Hmox1−/− mouse, which occluded at 20 minutes. Three vessels from Hmox1−/− and Hmox1+/+ mice were analyzed by hematoxylin/eosin. Magnification, ×100. C, Biliverdin rescues Hmox1−/− prothrombotic phenotype. Mice were injected with biliverdin 3 hours before injury, and time to stable cessation of blood flow was recorded (n=3 Hmox1+/+; and n=3 Hmox1−/− mice).
To further test the hypothesis that accelerated thrombosis is the direct result of Hmox1 deficiency, biliverdin, the byproduct of oxidative degradation of heme by HO-1, was administered to Hmox1+/+ and Hmox1−/− mice 3 hours before photochemical injury. Biliverdin rescued the prothrombotic phenotype in Hmox1−/− mice: time to thrombosis for Hmox1−/−, 50.0±10.0 minutes; versus Hmox1+/+, 59.0±1.0 minutes (P=NS) (Figure 1C). These data suggest that HO-1 directly regulates thrombosis following vascular injury.
Endothelial Damage and Denudation Are Present in Hmox1−/− Mice
We hypothesized that oxidative injury in Hmox1−/− mice leads to EC damage, denudation, exposure of the subintima, and intravascular thrombosis. To test this hypothesis, carotid arteries were examined by electron microscopy 15 to 20 minutes after photochemical injury. Arteries from Hmox1−/− mice had extensive endothelial damage characterized by nuclear condensation and cytoplasmic vacuolization (Figure 2A and 2B). Endothelial denudation was evident with degranulating platelet-rich microthrombi (Figure 2C). In contrast, the endothelium in Hmox1+/+ arteries was intact (Figure 2D). These findings suggest that EC damage and denudation may contribute to rapid thrombosis in Hmox1−/− arteries.
Figure 2. EC damage and thrombus formation in Hmox1−/− mice. A through C, Transmission electron microscopy of carotid arteries from Hmox1−/− mice 15 to 20 minutes after injury. EC damage with nuclear condensation (NC) and cytoplasmic edema (CE) (A, white arrows; magnification, ×1000; B, magnification ×2500) and platelet degranulation were observed (C, black arrows; magnification, ×1000). D, Intact endothelium in Hmox1+/+ arteries 15 to 20 minutes after injury. Magnification, ×1000.
To investigate the mechanism of EC damage and loss, we examined EC apoptosis in Hmox1−/− and Hmox1+/+ arteries, and we delivered a p38 MAPK inhibitor, SKF-86002, to Hmox1+/+ mice before photochemical injury. A significant increase in apoptotic ECs was evident in Hmox1−/− arteries following vascular injury (Hmox1−/−, 8.3±0.4%; versus Hmox1+/+, 1.7±0.7%; P<0.0005) (Figure 3A through 3C). In addition, inhibition of p38 MAPK increased EC apoptosis in Hmox1+/+ injured arteries (7.7±2.6%; versus non–inhibitor-treated Hmox1+/+ arteries, 1.7±0.7%; P<0.05) (Figure 3A) and accelerated thrombosis formation (38.3±6.4 versus 54.7±3.3 minutes; P<0.05). Thus, EC apoptosis contributes to accelerated thrombosis in injured Hmox1−/− arteries and is likely mediated through p38 MAPK.
Figure 3. EC apoptosis in injured arteries from Hmox1−/− mice. A, Percentage of apoptotic ECs in arteries from Hmox1+/+ mice, Hmox1+/+ mice injected with the p38 MAPK inhibitor SKF-86002 30 minutes before injury, and Hmox1−/− mice, all following injury. *P<0.05, Hmox1+/+ with inhibitor vs no inhibitor; ****P<0.0005, Hmox1+/+ vs Hmox1−/−. B, TUNEL staining of ECs following injury in Hmox1−/− arteries (right, black arrows) compared with Hmox1+/+ arteries (left). Magnification, ×400. C, Cross-sections from Hmox1+/+ (left) and Hmox1−/− (right) arteries stained with TUNEL following injury (top) and CD31+ antibodies (bottom). Magnification, ×100. D, Protein carbonylation levels in Hmox1+/+ arteries (left) and Hmox1−/− arteries (right) at baseline and following injury. *P<0.01, Hmox1+/+ baseline vs injury (n=3 mice in each group).
To further pursue the hypothesis that oxidative injury in Hmox1−/− mice leads to EC damage and subsequent thrombosis, we measured levels of ROS in Hmox1−/− and Hmox1+/+ arteries using a protein carbonylation assay. We found a significant elevation in ROS in Hmox1−/− arteries before photochemical injury compared with Hmox1+/+ arteries (1.34±0.24 versus 0.90±0.06 nmol/mg, respectively; P<0.05), consistent with Hmox1 deficiency. Following photochemical injury, ROS rose in Hmox1+/+ mice to 1.66±0.13 nmol/mg (P<0.01) compared with Hmox1+/+ mice at baseline. ROS remained elevated in Hmox1−/− mice following injury 1.54±0.18 nmol/mg (Figure 3D). Elevations in ROS in Hmox1−/− mice were consistent with a deficiency of Hmox1 and superimposed oxidative injury leading to EC damage and thrombosis.
Hemostatic Function in Hmox1−/− Mice
To investigate the mechanisms of thrombosis, we first asked whether abnormalities in hemostasis were present in Hmox1−/− mice. No significant differences were present in bleeding time, platelet counts, or prothrombin times between Hmox1−/− and Hmox1+/+ mice (P=NS), suggesting that abnormalities in primary and secondary hemostasis did not account for accelerated thrombosis (Table). Significant differences were observed, however, in hematocrit, hemoglobin concentration, and mean corpuscular volume of Hmox1−/− compared with Hmox1+/− and Hmox1+/+ mice, indicative of an anemia associated with Hmox1 deficiency.25
Table. Hemostasis Measurements in Hmox1+/+ and Hmox1−/− Mice
Because heme catabolism by HO-1 leads to the formation of the antioxidants bilirubin and CO, which serve protective functions in the vasculature, we assayed for baseline arterial carboxyhemoglobin (COHb) and total plasma bilirubin levels in Hmox1−/−, Hmox1+/−, and Hmox1+/+ mice. No significant differences were observed between the genotypes (baseline and following injury) in total bilirubin, arterial hemoglobin, or COHb. Thus, accelerated arterial thrombosis in Hmox1 deficiency is not attributable to defects in primary or secondary hemostasis, bilirubin, or COHb.
Elevations in TF and vWF in Hmox1−/− Mice
To further investigate the mediators of thrombosis, we assayed arterial TF and circulating vWF, factors known to be associated with oxidant damage and pathological thrombosis.33 Differences in carotid TF levels were not present at baseline between Hmox1−/− and Hmox1+/+ mice, but TF levels did significantly rise in Hmox1−/− arteries 15 minutes after the onset of photochemical injury and before the detection of occlusive thrombus (Hmox1−/−, 1.9-fold increase versus Hmox1+/+; P=0.02) (Figure 4A). Significant differences in circulating vWF levels were not present at baseline, but significant elevations in circulating vWF were observed in Hmox1−/− mice following photochemical injury (Hmox1−/−, 156.4±28.7; versus Hmox1+/+, 71.03±17.1; P<0.05) (Figure 4B). Elevations in TF and vWF following vascular injury may contribute in combination with EC damage to promote thrombus formation in Hmox1−/− mice.
Figure 4. Arterial TF and plasma vWF levels are elevated in Hmox1−/− mice. A, TF levels in Hmox1+/+ arteries (left) and Hmox1−/− arteries (right) following injury. Values are expressed as fold activation from Hmox1+/+ arteries. *P=0.02, Hmox1−/− vs Hmox1+/+ (n=6 mice in each group). B, Plasma vWF levels following injury in Hmox1+/+ mice (left) and Hmox1−/− mice (right). *P<0.05, Hmox1−/− vs Hmox1+/+ (n=6 mice in each group).
BM-Derived Cells Contribute to Accelerated Thrombosis in Hmox1−/− Mice
BM-derived progenitor cells are increasingly recognized as key components in vascular regeneration.34,35 We hypothesized that BM-derived cells lacking HO-1 may be defective in protection of damaged endothelium following injury. Accordingly, we transplanted Hmox1−/− BM into Hmox1+/+ mice, Hmox1+/+ BM into Hmox1+/− mice, and Hmox1+/+ BM into Hmox1+/+ mice. Engraftment was confirmed, photochemical injury was performed, and the time to thrombosis was measured. Transplantation of Hmox1−/− BM into Hmox1+/+ mice led to significantly accelerated thrombosis compared with transfer of Hmox1+/+ BM into Hmox1+/+ mice (42.8±3.4 versus 67.6±9.6 minutes; P<0.05) (Figure 5A), suggesting that Hmox1−/− BM-derived cells contribute to accelerated thrombosis. Transfer of Hmox1+/+ BM into Hmox1+/− mice led to a thrombosis time similar to transfer of Hmox1−/− BM into Hmox1+/+ mice (40.3±5.2 minutes) (Figure 5A), suggesting that the partial absence of HO-1 in the arterial wall created a local milieu in which endothelial damage still led to accelerated thrombosis, even in the setting of BM-derived cells with full HO-1 complement.
Figure 5. Transplantation of Hmox1−/− marrow into Hmox1+/+ mice leads to accelerated thrombosis. A, BM from Hmox1+/+ or Hmox1−/− mice (donor) were transplanted into Hmox1+/+ (left), Hmox1+/− mice (middle), or Hmox1+/+ mice (right) (recipient), and injury was performed 14 weeks later. *P<0.05 compared with Hmox1+/+ BM transplanted into Hmox1+/+ recipients (n=7 to 9 mice in each group). B, TF levels in BM-transplanted arteries, as described in A. Values are expressed as fold activation from the Hmox1+/+ arteries. *P<0.05, compared with Hmox1+/+ BM transplanted into Hmox1+/+ recipients. C, vWF levels in BM-transplanted mice, as described in A. *P<0.05, **P<0.01, compared with Hmox1+/+ BM transplanted into Hmox1+/+ recipients.
To determine whether Hmox1+/+ progenitor cells could rescue the accelerated thrombosis phenotype in Hmox1−/− mice, we transplanted Hmox1+/+ BM into Hmox1−/− mice. Unexpectedly, engraftment did not occur, and injury experiments could not be performed.
We postulated that TF and vWF may also contribute to accelerated thrombosis when Hmox1−/− BM is transplanted into Hmox1+/+ mice. TF was increased 2-fold in this setting compared with TF levels following transplantation of Hmox1+/+ marrow into Hmox1+/+ recipients (P<0.05), and vWF was similarly elevated (136.8±12.4 versus 52.3±8.6; P<0.005; Figure 5B and 5C). TF levels were also elevated 2-fold in Hmox1+/− mice receiving Hmox1+/+ BM, likely reflecting local EC damage in Hmox1+/− mice (Figure 5B and 5C).
To determine whether Hmox1−/− platelets aggregate abnormally independent of Hmox1−/− endothelium, in vitro platelet aggregations assays were performed in response to collagen or ADP. Equivalent numbers of Hmox1−/− and Hmox1+/+ platelets were exposed to agonists, and the extent and rate of aggregation was determined using an optical impedance method. No significant differences were observed in response to collagen (2.5 to 10 μg/mL) or ADP (5 to 20 μmol/L), indicating that HO-1 deficiency does not alter platelet aggregation (data not shown).
CO Inhalation Rescues the Thrombosis Phenotype in HO-1 Deficiency
Because CO attenuates platelet aggregation in vitro,36 we investigated whether inhalation of sublethal doses of CO would rescue the thrombotic phenotype in Hmox1−/− and Hmox1+/+ mice. Hmox1−/− and Hmox1+/+ mice were exposed to balanced room air containing CO at 500 ppm in a controlled chamber for ≈24 hours and underwent photochemical injury, and the time to arterial occlusion was measured. Inhaled CO did not modify thrombosis in Hmox1+/+ mice (room air, 54.7±2.3 minutes; versus inhaled CO, 52.0±10.5 minutes; P=NS) (Figure 6A). In contrast, inhaled CO rescued the prothrombotic phenotype in Hmox1−/− mice by increasing the time to arterial occlusion (room air, 20.9±3.4 minutes; versus inhaled CO, 36.6±1.6 minutes; P<0.05) (Figure 6B). Delivery of CO just before or during injury had no effect on time to thrombosis.
Figure 6. CO inhalation rescues the prothrombotic phenotype in Hmox1−/− mice. Hmox1+/+ mice (A) or Hmox1−/− mice (B) were treated with room air (left) or inhaled CO at 500 ppm (right) for 24 hours and underwent injury. Time to arterial occlusion was measured. P=NS, n=4 mice in each group (A); *P<0.05, inhaled CO vs room air (n=4 mice in each group) (B).
Hmox1 Deficiency Leads to Thrombosis Through a Failure of Arterial Repair Mechanisms
The elevations in ROS in Hmox1−/− arteries led us to further hypothesize that HO-1 induction following arterial injury is critical to mediate the generation of ROS. Accordingly, we investigated HO levels at baseline and following photochemical injury using RT-PCR and microarray methods. We observed an 8.9-fold increase in HO-1 RNA expression in Hmox1+/+ injured arteries compared with noninjured arteries (Figure 7A), suggesting that, indeed, injury is a primary stimulus for the induction of HO-1 in the arterial wall. Consistent with our hypothesis that elevated HO-1 expression is a primary mechanism to respond to ROS generation and promote vascular repair, we observed a 3.7-fold elevation in RNA expression of the cyclin-dependent kinase inhibitor p21Cip1 in injured Hmox1+/+ arteries (Figure 7A). Previous work from our laboratory demonstrated that HO-1 promotes arterial repair by upregulation of p21Cip1.12
Figure 7. Comparison of Hmox1+/+ and Hmox1−/− expression profiles pre- and postinjury. A, RNA fold activation of Hmox1 and p21Cip1 in injured Hmox1+/+ arteries. B, Principal component analysis, PC1 vs PC2, is shown with each group in a different color and 50% confidence ellipses. The percentage of total variance explained by each principal component is shown in parentheses. C, Hierarchical cluster analysis of genes overexpressed in injured Hmox1+/+ and Hmox1−/− arteries. Cluster analysis was applied to gene expression data from 199 probe sets on MOE430A 2.0 chips selected by a t test between injured and noninjured group with a false discovery rate of ≤10% and change of ≥10-fold. The level of gene expression relative to its mean across all samples is represented denoting above average (red), average (black), or below average (green) expression. D, Hierarchical clustering of genes with ≥10-fold upregulation by injury and Hmox1 deficiency. The gene list is selected by comparing (t test) injured arteries with uninjured arteries using a maximum 10% false discovery rate, a minimum 10-fold change cutoff, and a ≥2.5-fold change in Hmox1−/− vs Hmox1+/+. Cluster 1 includes genes induced more by injury in Hmox1−/− arteries (Acta2, Srgn, Alox12, Gata1, Pdzk1IP1, Klhl6, Nrgn, Mpl, Edem1, transmembrane protein 40, Armc8, Adcy1_predicted, Mmrn1, Triobp, Rgs18, F5, F13a1, Slamf1, Gsta2, Plek, Gp1bb, Gas2l1, Clec4g, Gypa, Hbb-y, Gp1bb, Lyz, Slfn4, Itga5); cluster 2, genes expressed more in Hmox1+/+ arteries (Aebp1, Fmo3, Npnt, Dsp, Itga5, Itgbl1, Hspb6, Rrad, Itga8, Csrp1, Myh11, Ntf3, Tagln, Fblim1, Myocd, Lrrfip1, Sgcg, Lmcd1, Mfap4, Rbpms, Igk-C, Nr4a2, S100a8, Cxcl1, Cxcl2, Nfkbiz, IL6, Fosb, Fmod). The complete gene names are provided in supplemental Table I.
We next performed an amplified microarray analysis of noninjured and injured arteries. A principal component analysis was conducted first to determine major changes in gene expression among 4 groups: Hmox1+/+ noninjured, Hmox1+/+ injured, Hmox1−/− noninjured, and Hmox1−/− injured arteries. We observed no major differences in gene expression patterns in noninjured arteries between Hmox1+/+ and Hmox1−/− mice (Figure 7B). Interestingly, photochemical injury itself led to an upregulation of multiple genes in both Hmox1+/+ and Hmox1−/− injured arteries; following injury, 199 genes were upregulated using a 10% false discovery rate and a >10-fold change cutoff. This gene expression pattern was accentuated in Hmox1−/− injured arteries (Figure 7B). Gene expression clusters demonstrate the distinct differences between noninjured and injured arteries in Hmox1+/+ and Hmox1−/− groups (Figure 7C).
To understand the consequences of Hmox1 deficiency in vascular repair, we selected genes that changed more in the absence of Hmox1, by a factor of at least 2.5, than in the presence of Hmox1. This analysis revealed 58 genes that were induced by injury differentially in the Hmox1−/− arteries compared with Hmox1+/+ arteries, including 9 genes in coagulation/thrombosis pathways; 2 genes in oxidative stress response; and 2 genes in apoptosis/cell cycle regulation (Figure 7D). One of the genes in the coagulation pathway is of particular interest; glycoprotein 1b is a platelet surface membrane glycoprotein that functions as a vWF receptor, consistent with our observation of elevated vWF levels in Hmox1−/− mice after injury.
We conclude (1) that in the absence of arterial injury, Hmox1 deficiency is reasonably well tolerated and that gene expression profiles are not significantly different from Hmox1+/+; (2) that following photochemical injury, ROS are generated, which are mitigated by antioxidants in the arterial wall, including the HO-1 pathway; and (3) that in the absence of HO-1, photochemical injury produces ROS, which persist with toxic effects on the artery wall, including damage to the endothelium, release of vWF, endothelial denudation, synthesis and release of TF, transcriptional activation of genes in the coagulation/thrombosis pathways, and subsequent thrombosis.
Discussion
Intravascular thrombosis is a well-recognized complication of vascular inflammation. HO-1 has vascular protective properties, including inhibition of vascular smooth muscle cell growth,8,12,23 induction of vasodilation,12 and protection against inflammation and oxidant damage.1–4,20,21 Although HO-1 guards against platelet-dependent thrombosis,37 the direct role of HO-1 in the prevention of arterial thrombosis, a known complication of vascular oxidant damage, has not been explored. Here, we used a genetic model of Hmox1 deletion to directly investigate the regulation of intravascular thrombosis by HO-1. We found that HO-1 protects against intravascular thrombosis associated with EC damage. Hmox1 deletion leads to accelerated thrombosis by several mechanisms, including EC disruption and apoptosis, platelet activation, elevations in TF and vWF, and a failure of BM-derived progenitor cells to protect the artery. This prothrombotic phenotype in Hmox1−/− mice is rescued by CO inhalation and biliverdin administration, known byproducts of HO-1 metabolism of heme. Our findings demonstrate that HO-1 directly preserves endothelial integrity, attenuates platelet activation, and protects against arterial thrombosis during acute vascular injury.
Previously, the role of HO-1 in thrombosis has been studied indirectly through animal models in which stimuli such as hypoxia, lipopolysaccharide, or xenotransplantation provoke inflammation.3,15–18 In these studies, HO-1 mediates against the induction of proinflammatory cytokines such as tumor necrosis factor-α, IL-1, and monocyte chemoattractant protein-1; it attenuates lipopolysaccharide-induced and monocyte-derived endothelial activation and limits expression of TF and plasminogen activator inhibitor-1, which can result in intravascular thrombosis.19,27 In this study, we directly demonstrate, through genetic approaches, a critical role for HO-1 to mediate the induction of ROS following oxidant injury and prevent arterial thrombosis.
Extensive endothelial damage with elevations in circulating vWF and thrombomodulin has been reported previously in a human case of Hmox1 deficiency.25 TF, which combines with factor VIIa to activate factor X and induce thrombosis, was also significantly elevated in Hmox1−/−, but not Hmox1+/+, arteries following photochemical injury. It is likely that elevations in TF and vWF facilitated thrombosis formation in Hmox1−/− mice through their effects on fibrin and the coagulation cascade. Our observations linking TF to accelerated thrombosis in Hmox1 deficiency is reinforced by the finding that CO inhibits MAPK-driven expression of Egr-1 and subsequent TF expression.20 Indeed, the concept that CO is a cytoprotective product of HO-1 is further supported by the reversal of rapid thrombosis in Hmox1−/− mice following CO inhalation.
In addition to vascular ECs, platelets and other circulating blood cells possess functional HO-1,24 and CO generated by HO-1 in cultured vascular smooth muscle cells attenuates platelet activation via elevations in platelet cGMP.23 Although platelet counts and bleeding times were not significantly different in Hmox1−/− and Hmox1+/+ mice, we hypothesized that the loss of HO-1 in platelets might also contribute to accelerated thrombosis in concert with EC damage. Our findings that BM from Hmox1−/− mice transplanted into Hmox1+/+ recipients leads to accelerated thrombosis compared with Hmox1+/+ mice transplanted with Hmox1+/+ BM suggest that BM-derived cells, which lack HO-1, directly contribute to thrombosis formation as well.
Hmox1−/− platelets did not demonstrate increased aggregation in vitro, and acute inhalation of CO just before and during photochemical injury did not alter the thrombotic phenotype, suggesting that Hmox1−/− platelets are not solely responsible for thrombosis in the setting of Hmox1−/− deficiency. Other studies using indirect in vitro approaches have suggested a possible role for HO-1 in platelet-dependent thrombosis37; these differences may be attributable to technical factors, such as use of Hmox1−/− platelets.
Endothelial progenitor cells, originating from the BM, have been shown to be recruited to sites of vascular injury, where they differentiate into ECs and contribute to adaptive vascular remodeling.34,35,38 Interestingly, thrombosis was significantly reduced following balloon angioplasty of rabbit carotid arteries transplanted with endothelial progenitor cells overexpressing HO-1.38 These findings are consistent with the observations from our BM transplantation findings; that is, BM-derived progenitor cells from Hmox1−/− mice are more susceptible to apoptosis and are associated with rapid thrombus formation. These Hmox1−/− progenitor cells lack the antithrombotic and antiinflammatory properties of HO-1, likely because of a lack of protection against ROS.
CO prevents oxidant-induced endothelial damage in the lung,26 directly protects ECs from apoptosis,21,22 and attenuates platelet aggregation following elevations in cGMP.36 Chronic CO inhalation reversed the prothrombotic phenotype in Hmox1−/− mice but had no effect on Hmox1+/+ mice, suggesting that the endogenous generation of CO by HO-1 is a potent mechanism by which HO-1 protects the vasculature from thrombosis under conditions of oxidative stress. Similarly, biliverdin, another byproduct of heme metabolism by HO-1, rescued the prothrombotic phenotype of Hmox1 deficiency in a p38 kinase-dependent manner, providing an additional mechanism by which HO-1 protects vascular integrity.
We conclude that HO-1 directly preserves vascular integrity and prevents thrombosis during oxidant stress by multiple mechanisms, including preservation of the endothelium, inactivation of platelets, and adaptive remodeling by BM-derived progenitor cells. Administration of byproducts of HO-1, CO, and biliverdin rescues the prothrombotic phenotype of Hmox1 deficiency during oxidative stress. Thus, therapies targeted toward the induction of HO-1 or it cytoprotective byproducts may be beneficial in preventing thrombosis observed in vascular disorders.
Acknowledgments
Sources of Funding
This work was supported by the Division of Intramural Research at the National Heart, Lung, and Blood Institute, NIH.
Disclosures
None.
Footnotes
-
Original received September 20, 2005; resubmission received July 12, 2007; accepted September 12, 2007.
References
This Issue
Jump to
Article Tools
Share this Article
- Heme Oxygenase-1 Deficiency Accelerates Formation of Arterial Thrombosis Through Oxidative Damage to the Endothelium, Which Is Rescued by Inhaled Carbon MonoxidePermalink:
