Tissue factor deficiency causes cardiac fibrosis and left ventricular dysfunction

Abstract

Exposure of blood to tissue factor (TF) activates the extrinsic (TF:FVIIa) and intrinsic (FVIIIa:FIXa) pathways of coagulation. In this study, we found that mice expressing low levels of human TF (≈1% of wild-type levels) in an mTF^−/− background had significantly shorter lifespans than wild-type mice, in part, because of spontaneous fatal hemorrhages. All low-TF mice exhibited a selective heart defect that consisted of hemosiderin deposition and fibrosis. Direct intracardiac measurement demonstrated a 30% reduction (P < 0.001) in left ventricular function in 8-month-old low-TF mice compared with age-matched wild-type mice. Mice expressing low levels of murine FVII (≈1% of wild-type levels) exhibited a similar pattern of hemosiderin deposition and fibrosis in their hearts. In contrast, FIX^−/− mice, a model of hemophilia B, had normal hearts. Cardiac fibrosis in low-TF and low-FVII mice appears to be caused by hemorrhage from cardiac vessels due to impaired hemostasis. We propose that TF expression by cardiac myocytes provides a secondary hemostatic barrier to protect the heart from hemorrhage.

Expression of tissue factor (TF) by adventitial fibroblasts and vascular smooth muscle cells surrounding blood vessels provides a hemostatic barrier that activates coagulation when vascular integrity is disrupted (1). TF is also expressed by cardiac muscle but not by skeletal muscle (1). TF functions as the high-affinity cellular receptor for FVII/VIIa (2). The coagulation protease cascades are comprised of the extrinsic (TF:FVIIa) and intrinsic (FVIIIa:FIXa) pathways, which together maintain hemostasis (3).

Many murine models of coagulation have been generated that provide new insights into the role of the various procoagulant and anticoagulant proteins in hemostasis (4). For instance, FV^{Leiden/Leiden} mice, which express an FV variant that is resistant to inactivation by activated protein C, and TM^Pro/Pro mice, which express a mutated version of thrombomodulin (TM) with reduced thrombin binding, both exhibit prothrombotic phenotypes with increased fibrin deposition in select tissues (5–7). Mice with prohemorrhage phenotypes include models of hemophilia A (FVIII^−/−) and B (FIX^−/−), as well as fibrinogen-deficient mice (Fbg^−/−) and thrombocytopenic mice (NF-E2^−/−) (8–12). Mice with complete deficiencies in TF, FVII, FX, FV, and prothrombin die in utero or shortly after birth (4). We and others have generated mice expressing low levels (<0.1–1% of wild-type levels) of human TF, murine FVII, and murine FV (13–15). We have shown that low-TF mice have impaired uterine hemostasis (16). A similar phenotype is observed with low-FVII mice.

In this study, we performed a detailed characterization of low-TF mice. These mice exhibited shorter lifespans than wild-type mice. Histological analysis of various tissues of low-TF mice revealed hemosiderin deposition and fibrosis selectively in their hearts. Our data suggest that cardiac fibrosis in low-TF mice is caused by hemorrhage from cardiac vessels due to impaired hemostasis.

Materials and Methods

Mice.

Low-TF mice (line 47) were analyzed on either a mixed genetic background (62.5% C57BL/6J, 25% 129Sv, and 12.5% BALB/c) or a C57BL/6J background (≥97%). mTF^Δcyt/Δcyt mice express normal levels of murine TF lacking the cytoplasmic domain (17). Low murine FVII mice (FVII^{tTA-FVII/tTA-FVII}) were generated by replacing the entire FVII gene with a transgene expressing FVII under the control of a tTA responsive promoter (E.R., Z. Liang, A. Martin, and F.C., unpublished data). Fbg^−/− and FIX^−/− mice have been described (11, 18). All studies were approved by The Scripps Research Institute Animal Care and Use Committee and comply with National Institutes of Health guidelines.

Measurement of Cell Counts and Clotting Activity.

White cell, red cell and platelet counts, and hemoglobin and hematocrit levels were determined by LabCorp (San Diego), using blood collected from the inferior vena cava. Activated partial thromboplastin times (APTTs) and prothrombin times (PTs) were performed using Automated APTT Reagent and Thromboplastin Reagent (Organon Teknika), respectively, and clotting times determined using a START4 Coagulation Analyzer (Diagnostica Stago, Parsippany, NJ). Levels of TAT in the plasma were determined using a commercial ELISA (Enzygnost, Dade Behring, Marburg, Germany). The procoagulant activity of heart tissue extracts added to mouse plasma was determined using a one-stage clotting assay as described (13) and converted to activity units by comparison to a standard curve generated using mouse brain thromboplastin. Factor VII assays were performed using a modification of the Coaset FVII assay (Chromogenix, Milan). A standard curve was generated by combining wild-type mouse plasma with human FX-deficient plasma at varying ratios.

Histology.

Tissue sections were stained with hematoxylin/eosin (H&E), Prussian Blue, or Masson's Trichrome. Macrophages were identified with a monoclonal antibody MOMA-2 (1:1,000; Serotec; ref. 20).

Measurement of Left Ventricular (LV) Function of the Hearts of Low-TF Mice.

LV function of hearts of low-TF mice (8 months of age) and age-matched C57BL/6J mice was measured as described (20). LV function (dp/dt) and LV systolic pressures were obtained at a constant heart rate range of 480–510. The heart rate was decreased by increasing the level of isoflurane anesthesia.

Data Analysis.

Statistical analysis was performed using a two-tailed unpaired Student's t test, and differences were determined to be statistically significant at a P value of <0.05.

Results

Characterization of Low-TF Mice.

Low-TF mice (mTF^−/−/hTF⁺ line 47) contain a minigene (hTF) that directs cell type-specific expression of human TF that is similar to the expression of murine TF (13). Immunohistochemical analysis indicated that human TF was expressed by adventitial cells surrounding blood vessels (data not shown). However, quantitation of the procoagulant activity of tissue extracts from various tissues indicated that these mice expressed low levels of TF (≈1% of wild-type levels; ref. 13). For example, the procoagulant activity of tissue extracts from hearts of low-TF mice was very low compared with control mice (Table 1), indicating a deficiency of TF in the heart.

Table 1.

Clotting activity in low-TF mice

Genotype	mTF+/+	mTF+/−/hTF+	mTF−/−/hTF+
Heart PCA, arbitrary units	232  ± 80 (3)	98  ± 80 (3)	0.2  ± 0.2 (3)
PT, s	10.47  ± 0.15 (3)	10.48  ± 0.05 (4)	10.53  ± 0.32 (4)
APTT, s	26.37  ± 0.90 (3)	29.85  ± 0.76 (4)	31.53  ± 0.90 (4)
TAT, ng/ml	1.78  ± 0.44 (3)	0.74  ± 0.28 (5)	0.04  ± 0.02 (5)

PT, prothrombin time; APTT, activated partial thromboplastin time; PCA, procoagulant activity. Numbers in parentheses indicate the number of mice used in each analysis.

We evaluated the clotting activity and response to hemostatic challenge of low-TF mice (4–8 weeks of age) on a C57BL/6J background. No significant differences were found in whole blood samples collected from low-TF mice (n = 5) and mTF^+/−/hTF⁺ (n = 5) littermate mice with regard to platelet, red cell, and white cells counts, and hematocrit and hemoglobin (not shown). In addition, low-TF mice had APTTs and PTs that were similar to control mice (Table 1). These results indicate that the low levels of TF in these mice maintain hemostasis under normal conditions. In contrast, low-TF mice exhibited an abnormal response to hemostatic challenge. Low-TF mice showed a significantly prolonged occlusion time (99 ± 40 min, mean ± SD, n = 7; P < 0.001) compared with C57BL/6J mice (44 ± 18 min, n = 9) in a Rose Bengal model of carotid artery injury (S. Day, J. Reeve, B.P., N.M., and W. Fay, unpublished data). We also observed that low-TF mice have low levels of circulating TAT complexes compared with mTF^+/−/hTF⁺ littermate mice and wild-type C57BL/6J mice (Table 1), indicating that low-TF mice generate lower levels of thrombin. Taken together, these results suggest that low-TF mice would be prone to excessive hemorrhage in the event of vessel injury.

Survival of Low-TF Mice.

In initial studies, we observed that low-TF mice on a mixed genetic background had shorter lifespans (Fig. 1) compared with C57BL/6J mice (21). Low-TF mice had even shorter lifespans on a C57BL/6J background compared with a mixed background (Fig. 1). Autopsies revealed that 17% of the low-TF mice died of spontaneous hemorrhages in the brain, lung, and gastrointestinal tract (not shown). Although we cannot exclude the possibility that we have underestimated the death rate due to acute hemorrhage, it appears that the majority of low-TF mice are dying prematurely of another cause.

Fig 1.

Survival of low-TF mice on mixed and C57BL/6J backgrounds. Kaplan–Meier plots showing survival profiles of 77 low-TF mice on a mixed background (Mixed) and 61 low-TF mice on a C57BL/6J background (C57).

Fibrosis in the Hearts of Low-TF Mice.

Histological examination of the major organs (brain, lung, heart, liver, kidney, and spleen) of low-TF mice revealed fibrosis in the heart but no defects in the other organs. Cardiac fibrosis increased with age, with the rate of fibrosis being faster in the hearts of low-TF mice on the C57BL/6J background compared with the mixed background (Fig. 2A). Fibrosis was initially observed in the subepicardium and then became widespread throughout the myocardium (Fig. 2B).

Fig 2.

Fibrosis in hearts of low-TF mice. (A) The degree of fibrosis was scored (0–5) in heart sections stained with Masson's Trichrome. Fibrosis was scored in hearts from 35 mice on a mixed background and 31 hearts on a C57BL/6J background. (B) Cross-sections of hearts stained with Masson's Trichrome demonstrate subepicardial and myocardial fibrosis in the hearts of low-TF mice (8 and 23 months) and no fibrosis in control mice (8 months). Normal myocardium stains red-brown and fibrotic tissue stains blue. Original magnification of the panels from the cross-sections was ×100. Hearts shown are from low-TF mice on a mixed genetic background. LV, left ventricle.

Decreased LV Function in the Hearts of Low-TF Mice.

Hemodynamic studies revealed that low-TF mice had a marked impairment of heart contractility manifested by a significant decrease in dp/dt at every heart rate examined (Fig. 3). At normal mouse heart rates (480–510 beats per min), the dp/dt and the LV pressure were decreased by 30% (P < 0.001). These results indicate that the cardiac fibrosis in the hearts of low-TF mice significantly impairs LV function.

Fig 3.

LV function in the hearts of low-TF mice. LV function was performed on 8-month-old low-TF mice (n = 6) on a C57BL/6J background and age-matched C57BL/6J mice (n = 5). (Left) LV function was measured at different heart rates. (Center and Right) LV function and LV pressure at normal heart rates (480–510 beats per min).

Detailed Histological Analysis of the Hearts of Low-TF Mice.

We examined hearts from >100 low-TF mice at different ages. We observed golden-brown granular deposits in the myocardium in tissue sections stained with H&E (Fig. 4A). We suspected that these deposits were hemosiderin, which is an insoluble protein produced by phagocyte digestion of hematin. We used Prussian Blue staining to confirm the presence of iron in the deposits in serial sections (Fig. 4B). Immunohistochemical studies showed that the hemosiderin-laden cells stained with MOMA-2 (not shown), which specifically recognizes macrophages (19). Hemosiderin deposits were associated with areas of fibrosis (Fig. 4C). No hemosiderin was seen in 1-month-old mice (not shown). In 3-month-old mice, hemosiderin was observed in the subepicardium and within the myocardium (Fig. 5A; see also Fig. 6A). In older mice (8 months), hemosiderin was observed throughout the myocardium selectively deposited around capillaries (Fig. 5B) and larger vessels (not shown). No hemosiderin was observed in the hearts of wild-type C57BL/6J mice (not shown).

Fig 4.

Hemosiderin deposits in the hearts of low-TF mice. Serial sections of a low-TF mouse on a mixed background (8 months of age) were stained with H&E (A), Prussian Blue (B), and Masson's Trichrome (C). Hemosiderin appears as a golden-brown deposit with H&E and stains blue with Prussian Blue. (Original magnification ×400.)

Fig 5.

Histological analysis of the hearts of low-TF mice. Heart sections were stained with H&E. (Original magnifications: A, ×100; B, ×1,000; C, ×250; D–F, ×400.) Hearts of low-TF mice (line 47) at 3 months (A), 5 months (C), 8 months (B and D), 11 months (F), and 14 months (E) of age are shown. Hemosiderin (brown) is observed subepicardially (A) and perivascularly (B). (C and D) Interstitial hemorrhages and necrosis of a cardiac myocyte (nuclear pyknosis and loss of cross striations; arrow; D). (E) Leukocyte infiltration and cardiac myocyte necrosis. (F) Fibrosis (light pink) and associated hemosiderin deposition with some remaining cardiac myocytes (dark pink). Hearts shown are from low-TF mice on a mixed background. Blood vessels (bv), epicardium (epi), and hemosiderin (arrowheads) are indicated.

Fig 6.

Analysis of hearts of mice with deficiencies in the extrinsic pathway of coagulation. Heart sections were stained with Prussian Blue (A–C) or Masson's Trichrome (D). Shown are heart sections from an mTF^−/−/hTF⁺(line 47) mouse (12 weeks of age; A), an mTF^−/−/hTF⁺(line 31) mouse (8 weeks of age; B), and an FVII^{tTA-FVII/tTA-FVII} mouse (9 weeks of age; C and D).

Hemosiderin in the heart is most likely derived from erythrocytes that have hemorrhaged into the myocardium. Indeed, we identified interstitial hemorrhages in the hearts of low-TF mice (Fig. 5 C and D), and these hemorrhages were often associated with cardiac myocyte necrosis (Fig. 5D). In other hearts, we observed infiltration of leukocytes into the myocardium (Fig. 5E). Fibrosis and hemosiderin were seen throughout the hearts of older low-TF mice (Fig. 5F). These results suggest that hemosiderin deposition is derived from erythrocytes hemorrhaging into the myocardium.

Hemosiderin Deposition in Hearts of an Independent Line of Low-TF Mice.

Previously, we demonstrated that we could rescue mTF^−/− embryos with two independent transgenic lines that express low levels (line 47) and very low levels (line 31) of human TF (13, 14). Indeed, line 47 provided long-term survival of mTF^−/−/hTF⁺(line 47) mice (see Fig. 1), whereas all mTF^−/−/hTF⁺(line 31) mice die within 8 weeks of birth (14). The majority (77%) of the mTF^−/−/hTF⁺(line 31) mice succumb to fatal hemorrhages in the brain and lung. Fig. 6 A and B shows hemosiderin deposition in hearts of mTF^−/−/hTF⁺(line 47) and mTF^−/−/hTF⁺(line 31) mice, respectively. These results indicate that an independent line of low-TF mice have hemosiderin deposition in their hearts.

Analysis of Mice Expressing TF Lacking the Cytoplasmic Domain.

We have shown that TF is located within the specialized adhesion junctions (intercalated discs) between cardiac myocytes (22), suggesting that TF may contribute to the structural integrity of the myocardium. In addition, the TF cytoplasmic domain has been shown to bind to the cytoskeleton via actin binding protein 280 (23). We tested the hypothesis that TF plays a structural role in the heart via interaction of the TF cytoplasmic domain with the cytoskeleton by examining the hearts of mTF^Δcyt/Δcyt mice that express normal levels (100%) of murine TF lacking the cytoplasmic domain (17). Importantly, these mice have normal hemostasis. No hemosiderin or fibrosis was observed in 6-month-old mTF^Δcyt/Δcyt mice (not shown), indicating that deletion of the TF cytoplasmic domain does not result in hemosiderin deposition in the heart.

Hemosiderin Deposition in Hearts of Low-FVII Mice.

Low-FVII mice were generated by replacing the FVII gene with a transgene that expresses FVII under the control of a tTA-responsive promoter. These mice express low levels of FVII (<1% of wild-type levels) and 50% of these mice die within 45 days of birth mostly because of brain hemorrhages (E.R., unpublished data). We observed hemosiderin deposition and fibrosis in the hearts of low-FVII mice (Fig. 6 C and D). Interstitial hemorrhages were also observed in the hearts of low-FVII mice (not shown). Thus, the cardiac phenotype of low-FVII mice was remarkably similar to that of low-TF mice.

Analysis of Hearts of FIX^−/− and Fbg^−/− Mice.

We analyzed the hearts of FIX^−/− and Fbg^−/− mice to determine whether hemosiderin deposition in the heart was a common phenotype in prohemorrhagic mice. We did not observe hemosiderin in the hearts of either FIX^−/− (13 weeks of age) or Fbg^−/− (6 months of age) mice (not shown). These results indicate that hemosiderin in the heart is selective for deficiencies in the extrinsic pathway of coagulation.

Discussion

We used low-TF mice to test the hypothesis that TF expression by cardiac myocytes contributes to hemostasis in the heart. Low-TF mice had shortened lifespans compared with wild-type mice, in part, because of spontaneous hemorrhages. However, fatal hemorrhages did not appear to account for the death of the majority of the mice. Analysis of the major organs of these low-TF mice revealed striking fibrosis in their hearts. LV function of 8-month-old low-TF mice was reduced by 30% (P < 0.001) compared with age-matched wild-type C57BL/6J mice. This decrease was not sufficiently large to induce congestive heart failure, but the extensive fibrosis may cause fatal arrhythmias in the low-TF mice and contribute to their shortened lifespan.

Low-TF mice had normal platelet counts and hemoglobin levels, but had reduced TAT levels and impaired responses to hemostatic challenge, suggesting that these mice may be prone to excessive hemorrhage in the event of blood vessel injury. Of note, low-TF mice had low levels of functional procoagulant activity in their hearts, indicating reduced TF expression by cardiac myocytes. Studies with low-TF mice on two different genetic backgrounds demonstrated a reduced survival on a C57BL/6J background compared with a mixed background. Interestingly, FV^{Leiden/Leiden} mice developed intravascular thrombosis in the perinatal period on a mixed (129Sv-C57BL/6J) background, but not on a C57BL/6J background (7). Similarly, TM^Pro/Pro mice had more fibrin deposition on a mixed (129Sv-C57BL/6J) background than on a C57BL/6J background (24). These results suggest that the C57BL/6J background is prohemorrhagic compared with the mixed background and may explain why low-TF mice on a C57BL/6J background have an increased rate of cardiac fibrosis and a reduced lifespan.

The cause of the cardiac fibrosis was investigated by detailed histological analysis of the hearts of low-TF mice at different ages. Hemosiderin deposition in younger mice was predominantly subepicardial. In older mice, hemosiderin was observed around capillaries and larger vessels throughout the myocardium. Hemosiderin may result from hemorrhage of erythrocytes into the myocardium. Indeed, we observed interstitial hemorrhages in the myocardium. Importantly, low-FVII mice exhibited a remarkably similar phenotype with hemosiderin and fibrosis in their hearts. Mice expressing normal levels of murine TF lacking the cytoplasmic domain had no hemosiderin in their hearts. Taken together, we propose that the most likely cause of cardiac fibrosis in the hearts of low-TF and low-FVII mice is hemorrhage from cardiac vessels resulting from impaired hemostasis.

Why do low-TF and low-FVII mice exhibit a hemostatic deficit selectively in the heart? The most likely explanation is that the heart represents a special tissue in which there is repetitive minor mechanical injury to blood vessels, especially at the surface of the heart, that produces hemorrhage in mice with severe TF or FVII deficiency. Additionally, the presence of the TF:FVII complex at the interface between cardiac myocytes and endothelial cells may contribute to the stability of capillaries. Future studies of mice with a cardiac-specific knockout of TF will directly address the role of cardiac myocyte TF in the heart.

We propose a model of tissue-specific hemostasis (Fig. 7). In this model, blood vessels in all tissues, such as skeletal muscle, have a primary hemostatic barrier due to TF expression by pericytes, vascular smooth muscle cells, and adventitial fibroblasts (1, 25). In addition to this primary hemostatic barrier, the heart has a secondary hemostatic barrier due to TF expression by cardiac myocytes (Fig. 7). Other tissues that have a secondary TF-dependent hemostatic barrier include the brain and uterus. Our model may explain why mice with deficiencies in either the extrinsic (low TF and low FVII) or the intrinsic (FVIII^−/− and FIX^−/−) pathways exhibit different phenotypes. For instance, low-TF and low-FVII mice have severely impaired uterine hemostasis, whereas FVIII^−/− and FIX^−/− mice have normal uterine hemostasis (10, 12, 16). The current study strongly suggests that hemostasis in the heart is primarily regulated by the TF:FVIIa-driven extrinsic pathway and is independent of FVIIIa:FIXa, because hemosiderin is observed in hearts of low-TF and low-FVII mice, but not in hearts of FIX^−/− mice. At present, it has not been determined whether severely FVII-deficient individuals surviving into adulthood develop cardiac fibrosis.

Fig 7.

Model showing how TF contributes to primary and secondary hemostatic barriers in cardiac and skeletal muscle. Skeletal muscle contains only a primary TF hemostatic barrier surrounding blood vessels (dark pink), whereas cardiac muscle contains both primary and secondary (light pink) TF hemostatic barriers. Endothelial cell layer is shown in green and blood is shown in red.

In skeletal muscle, we propose that coagulation is initiated by the TF:FVIIa complex, but requires the FVIIIa:FIXa complex to generate sufficient levels of FXa and thrombin to maintain hemostasis. Our model may explain why hemophiliacs bleed at sites of low-TF expression, such as joints and skeletal muscle (26), where the secondary TF-dependent hemostatic barrier is low or absent. Low-TF and low-FVII mice do not bleed in skeletal muscle, suggesting that there are sufficient levels of the TF:VIIa complex formed at this site to maintain hemostasis. It is notable that Fbg^−/− mice and patients classified as afibrinogenemic exhibit a relatively low frequency of spontaneous bleeding events, which may be because of sufficient platelet activation and platelet plug formation to control the bleeds (8). Similarly, the absence of hemosiderin deposition or fibrosis in the hearts of Fbg^−/− mice may be due to normal platelet activation, which would compensate for the loss of fibrinogen and prevent a hemostatic defect in the heart. Deficiencies in TF or FVII would result in reduced levels of thrombin, which would affect both fibrin deposition and platelet activation. In sum, our data demonstrates that the intrinsic and extrinsic pathways of blood coagulation do not contribute equally to hemostasis in different tissues.

Clinically, administration of recombinant FVIIa (NovoSeven) at high doses restores hemostasis in individuals with deficiencies in FVII, FVIII, or FIX, and is an effective therapy for treating hemorrhages in hemophiliacs with inhibitory antibodies to FVIII (27, 28). Conversely, targeting the TF:FVIIa complex with new antithrombotic drugs holds promise for reducing life-threatening thrombosis (29–31). Our data from low-TF and low-FVII mice suggest that prolonged use of these drugs could cause adverse effects on cardiac hemostasis.

Abbreviations

• Fbg, fibrinogen

• H&E, hematoxylin/eosin

• LV, left ventricular

• TF, tissue factor