Defective angiogenesis delays thrombus resolution: a potential pathogenetic mechanism underlying chronic thromboembolic pulmonary hypertension

Abstract

Objective

Restoration of patency is a natural target of vascular remodeling following venous thrombosis that involves vascular endothelial cells and smooth muscle cells as well as leukocytes. Acute pulmonary emboli usually resolve within six months. However, in some instances, thrombi transform into fibrous vascular obstructions, resulting in occlusion of the deep veins, or in chronic thromboembolic pulmonary hypertension (CTEPH). We proposed that dysregulated thrombus angiogenesis may contribute to thrombus persistence.

Approach and Results

Mice with an endothelial-cell-specific conditional deletion of vascular endothelial growth factor receptor 2/kinase insert domain protein receptor (VEGF-R2/Kdr) were utilized in a model of stagnant flow venous thrombosis closely resembling human deep vein thrombosis. Biochemical and functional analyses were performed on pulmonary endarterectomy specimens from patients with CTEPH, a human model of non-resolving venous thromboembolism. Endothelial cell-specific deletion of Kdr and subsequent ablation of thrombus vascularization delayed thrombus resolution. In accordance with these findings, organized human CTEPH thrombi were largely devoid of vascular structures. Several vessel-specific genes such as KDR, vascular endothelial cadherin and podoplanin were expressed at lower levels in white CTEPH thrombi than in organizing deep vein thrombi and organizing thrombi from aortic aneurysms. In addition, red CTEPH thrombi attenuated the angiogenic response induced by VEGF.

Conclusions

In the present work, we propose a mechanism of thrombus non-resolution demonstrating that endothelial cell-specific deletion of Kdr abates thrombus vessel formation, misguiding thrombus resolution. Medical conditions associated with the development of CTEPH may be compromising early thrombus angiogenesis.

Introduction

Restoration of vascular patency is a key target of vascular remodeling following thrombosis ^1-2. Apart from plasmin-driven fibrinolysis as a major proteolytic system in venous thrombus resolution, angiogenesis ³ and leukocyte recruitment ⁴ hallmark this process, with the expression of vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF) ⁵, and recruitment of bone marrow-derived Tie-2 expressing cells ^6-7.

Acute pulmonary thromboemboli undergo almost complete resolution within six months ⁸, however, in 0.1-9.1% of cases they persist in the major pulmonary arteries and trigger a vascular remodeling process with medial fibrosis and intimal hyperplasia leading to vascular stenosis and occlusions, clinically translating into chronic thromboembolic pulmonary hypertension (CTEPH) ^9-10. All currently available evidence indicates that CTEPH is primarily caused by pulmonary thromboembolism, as opposed to primary pulmonary in situ vascular thrombosis ¹¹. We speculate that pulmonary embolism may be followed by a pulmonary vascular remodeling process modified by infection ¹², immune phenomena ¹³, inflammation ¹⁴, circulating and vascular-resident progenitor cells ^15-16, thyroid hormone replacement and malignancy ¹⁷. Hypercoagulation, “sticky” red blood cells, high platelet count and uncleavable fibrinogen ¹⁸ also contribute to obliteration of large and small vessels in CTEPH. Thus, CTEPH may serve as a human model disease for venous thrombus non-resolution. CTEPH thrombus classically represents a cast of the pulmonary vascular bed, consisting of endothelium, smooth muscle cells and fibroblasts ^{16, 19}.

Previous animal studies demonstrated that venous thrombus recanalization may occur within 24 hours of thrombus formation ³, and several of these studies focused on the effect of administration of pro-angiogenic agents on thrombus resolution ^{4, 20-21}. In rats, a single bolus injection of recombinant VEGF protein into newly formed venous thrombi under reduced flow conditions resulted in enhanced recanalization and organization ²¹. VEGF naked DNA gene transfer and adenovirus-mediated VEGF gene therapy facilitated thrombus recanalization and resolution in both rats and mice ^22-23. In another study, induction of angiogenesis with bFGF and epithelial neutrophil activating protein (ENA-78) increased neovascularization, but did not affect experimental thrombus resolution ²⁰. Thus, controversy remains regarding the role of angiogenesis in venous thrombus resolution.

In this work we hypothesized that (i) angiogenesis plays a key role for thrombus resolution, and that (ii) CTEPH may result from a condition of decreased thrombus vascularization leading to thrombus non-resolution. We utilized an experimental model resembling human deep vein thrombosis in transgenic mice conditionally deficient in kinase insert domain protein receptor (Kdr), the predominant cellular receptor for VEGF in endothelial cells, to prove that thrombus resolution is critically dependent on Kdr. Furthermore, we aimed to study thrombus angiogenesis and vascularization at different stages of thrombosis in CTEPH. We harvested red “fresh” and white “fibrotic” CTEPH thrombus material, and compared those tissues with parent unthrombosed pulmonary artery and with a spectrum of venous and arterial thrombus samples. Parent, unthrombosed CTEPH pulmonary artery is a small piece of artery obtained from a proximal, apparently disease-free site [see supplement for full description].

Materials and Methods

Materials and Methods are available in the online-only Supplement.

Results

Characterization of mice with endothelial cell-specific deletion of Kdr (Kdr_Δend)

Kdr loci and the position of the probe used for Southern blot analysis are shown in Figure 1A. Total DNA was isolated from lungs, kidneys (Figure 1B, lanes 1 and 2), livers and hearts (data not shown) of controls (n=8) and Kdr_Δend (n=8, Figure 1B, lanes 3 and 4). Digestion with Spe I resulted in a 4.9 kb fragment for the floxed allele and a 15.3 kb fragment for the CRE deleted allele (Δ allele) (Figure 1B) ²⁴. The floxed allele contains a diagnostic Spe I site which is absent in the Δ allele (Figure 1A). While only the floxed allele could be detected in the organs of controls (Figure 1B, lanes 1 and 2), both the floxed and the Δ alleles were discernible in Kdr_Δend (Figure 1B, lanes 3 and 4). Kdr expression in endothelial cells isolated from lungs, kidneys and liver of Kdr^flox/flox/Tie-2CreER was significantly decreased after tamoxifen (TX) induction compared to cells from animals that did not receive TX (p<0.05, Figure 1C).

Figure 1. Characterization of Kdr_Δend.

(A) The wild-type Kdr locus, the floxed and Δ alleles as well as the position of the probe used for Southern blot analysis are shown. (B) Southern blot analysis indicates the presence of only the floxed allele in lung and kidney homogenates from control (lanes 1 and 2), whereas the Δ and floxed alleles were found in the corresponding organs of Kdr_Δend (lanes 3 and 4). (C) Kdr expression in endothelial cells (n=4 each, bars 1 and 2) and in monocytes (n=6 each, bars 3 and 4) of Kdr^flox/flox/Tie-2CreER after TX treatment (filled bars), compared with untreated mice (hatched bars). Kdr gene expression in endothelial cells of untreated mice was set to 0. * indicates p<0.05. (D) Monocytes in percent of total leukocytes, and monocytes expressing Tie-2 and KDR in percent of total monocytes of controls and Kdr_Δend (n=8 each) are shown. (E) Tail bleeding times from controls and Kdr_Δend (n=10 each). (F) The percentage of total and activated leukocyte/platelet aggregates (LPA) of total leukocytes, and the percentage of total and activated monocyte/platelet aggregates (MPA) of total monocytes in controls and Kdr_Δend (n=8 each). Controls are represented by open boxes/bars, and Kdr_Δend by filled boxes/bars.

KDR expression was also examined in Tie-2 expressing monocytes of Kdr_Δend and controls (n=8 each). Flow cytometry analyses demonstrated that the proportion of KDR⁺ monocytes of total monocytes and of Tie-2⁺ monocytes did not differ between Kdr_Δend and controls (Figure 1D). Only less than 0.25% monocytes expressed KDR, less than 0.25% monocytes expressed Tie-2, and less than 0.1% monocytes co-expressed Tie-2 and KDR (Figure 1D). Mean fluorescence intensity (MFI) of KDR in KDR⁺ monocytes and in Tie-2⁺ KDR⁺ monocytes was not different (in Kdr_Δend: 1851±746 and 1686±1335 MFIs, and in controls: 1780±870 and 1644±1343 MFIs respectively). Kdr mRNA gene expression level of isolated monocytes was equal in Kdr^flox/flox/Tie-2CreER before and after TX treatment (Figure 1C). Kdr mRNA expression was 1000 times less in monocytes than in endothelial cells (Figure 1C), leading support to the expectation that the Kdr_Δend phenotype was primarily endothelial cell-dependent.

Kdr_Δend did not display a prothrombotic phenotype, because tail bleeding times did not differ from controls (n=10 each, Figure 1E). Regarding platelet function, the percentage of total and activated leukocyte/platelet aggregates and the percentage of total and activated monocyte/platelet aggregates were similar in Kdr_Δend and controls (n=8 each, Figure 1F).

Delayed thrombus resolution in Kdr_Δend

To study thrombus resolution, a murine stagnant flow inferior vena cava (IVC) thrombosis model was employed. Thrombosis was induced by partial ligation of IVC in Kdr_Δend and controls, and thrombi were harvested 3, 7, 14 and 28 days after surgery (Figure 2A-H). We were able to harvest thrombi in 73% of Kdr_Δend and in 70% of controls (n=8 or 9 per group and individual time point). Weights, cross-sectional areas and volumes of thrombi from Kdr_Δend were significantly increased at all time points compared with controls (Figure 2I-K). Mean relative thrombus volume changes are illustrated in Figure 2L. Thrombus volumes from Kdr_Δend decreased at a slower rate between days 3 to 7 and days 7 to 14 than control thrombi.

Figure 2. Thrombus resolution in a mouse model of stagnant flow venous thrombosis in Kdr_Δend and controls.

(A-D) Trichrome stains of representative thrombus cross-sectional areas from Kdr_Δend and (E-H) the corresponding control thrombi, scale bar represents 100 μm. (I) Weights, (J) cross-sectional areas and (K) volumes (L) volume changes of thrombi formed in caval veins of Kdr_Δend (filled square symbols/bars) compared with controls (open square symbols/bars). * indicates p<0.05.

Endothelial cell-specific deletion of Kdr abates thrombus angiogenesis

Thrombus microvessels (expressed as percentage of isolectin B4 positive cells per total cells within a cross-sectional sample of the thrombus) were quantified by immunohistochemistry. On day 3 after ligation, microvessel density was similar in both groups. In thrombi of controls, microvessel density steadily increased until day 28. By contrast, microvessel density in Kdr_Δend thrombi was significantly decreased by days 7 and 14 compared with controls (Figure 3A-E, Supplemental Table I). By day 28 thrombus microvessel density was similar to controls.

Figure 3. Thrombus endothelial cells and macrophages in Kdr_Δend and controls.

(A) Isolectin B4 staining of a representative day 7 thrombus from Kdr_Δend and (B) from control. (C) and (D) show respective high power field images, scale bars represent 100 μm, respectively. (E) Microvessel counts (expressed as isolectin B4-positive cells of total cells per thrombus cross-sectional area) in thrombi from Kdr_Δend and controls. Relative gene expression of (F) Vegfa (G) Cdh5 and (H) Kdr of thrombi in Kdr_Δend and controls. (I) F4/80 staining of a representative day 7 thrombus from Kdr_Δend and (J) from control. (K) and (L) show the respective high power field images, scale bars represent 100 μm, respectively. (M) Macrophage counts (expressed as F4/80 positive cells in percent of total cells per thrombus cross-sectional area) in thrombi from Kdr_Δend and controls. Relative gene expressions of (N) CD68 (O) Ptprc and (P) Ctgf in thrombi of Kdr_Δend and controls. Data of Kdr_Δend are represented by filled square symbols and those of controls by open square symbols. * indicates p<0.05.

To verify immunohistochemical findings, expression of the angiogenic markers vascular endothelial-(VE−) cadherin (Cdh5), VEGF (Vegfa) and Kdr was analyzed in thrombi by quantitative Real-time PCR. In controls, these three markers were highly expressed on days 3 and 7, and decreased significantly by days 14 and 28. By contrast, expression of Cdh5 and Vegfa in Kdr_Δend remained low at all time points (Figures 3F and 3G). As expected, Kdr expression was very low on days 3, 7 and 14 in Kdr_Δend (Figure 3H).

Diminished thrombus macrophage numbers in Kdr_Δend

By day 7 control thrombi had higher counts of macrophages (expressed as percentage of F4/80 positive cells of total cells per thrombus area) than Kdr_Δend thrombi. No significant differences in macrophage counts were observed on days 3, 14 and 28 (Figure 3M, Supplemental Table I). CD68 mRNA gene expression was equal in thrombi of Kdr_Δend and controls at all time points, except for day 7, where control thrombi displayed more CD68 mRNA than Kdr_Δend thrombi (Figure 3N). This was in good agreement with immunohistochemistry findings. In controls, common leukocyte antigen CD45 (Ptprc) was highly expressed on days 3 and 7, and decreased significantly by days 14 and 28. In Kdr_Δend thrombi, Ptprc expression remained low on days 3, 7 and 14, however by day 28 expression had increased (Figure 3O). mRNA levels of connective tissue growth factor (Ctgf), a marker for organization and fibrotic transformation was increased on days 3 and 7 in controls, and decreased by days 14 and 28. In Kdr_Δend thrombi, Ctgf expression was low on days 3, 7 and 14, however by day 28 expression had increased (Figure 3P).

Patients

Consecutive CTEPH patients were consented for tissue analyses at the General Hospital of Vienna and the Kerckhoff Clinic in Bad Nauheim. Mean age of CTEPH patients was 58 ± 12 years, 46 % were female, and general characteristics matched those of a typical European CTEPH population (Table 1).

Table 1. Baseline Clinical and Hemodynamic Characteristics of CTEPH patients consented for this study (n=24, means ± SD).

Age (years)	58 ± 12
Sex (female, n)	11
6 minute walking distance (m)	440 ± 103
BDS *	4 ± 2
mPAWP (mmHg) †	12 ± 5
mPAP (mmHg) ‡	49 ± 13
CO (l/min) §	4.4 ± 0.8
PVR (dynes/sec/cm5) ∥	725 ± 268
NYHA functional class (% patients) **
I	0
II	12.5
III	70.8
IV	16.7
C-reactive protein (mg/dl)	0.97 ± 1.14
Fibrinogen (mg/dl)	369 ± 78
Leukocytes (g/l)	7.8 ± 4.5

^*Borg Dyspnea Score

^†mean pulmonary arterial wedge pressure

^‡mean pulmonary artery pressure

^§cardiac output

^∥pulmonary vascular resistance

^**New York Heart Association functional class

Patients from whom venous and arterial thrombi, acute coronary thrombi and carotid endarterectomy specimens had been harvested were gender and age-matched with CTEPH patients (n=46).

Low expression of angiogenic factors in CTEPH thrombi

By histology, mouse IVC thrombi were very similar to human CTEPH thrombi (Figure 4A-D). Von Willebrand factor (vWF) and platelet endothelial cell adhesion molecule-1 (PECAM-1)-positive cell counts were lower in white and red CTEPH thrombi than in parent unthrombosed pulmonary arteries (Figure 4F-L). Gene expression levels of angiopoietin-2 (ANGPT2), PECAM1, CDH5, KDR, podoplanin (PDPN), Tie-2 (TEK) and VEGFA were decreased in white CTEPH thrombi compared with respective parent unthrombosed pulmonary arteries as reference standards (Figure 5A). Furthermore, expression of factors involved in proliferative pathways of vascular cells such as bone morphogenetic protein receptor type 2 (BMPR2) or transforming growth factor-β1 (TGF-β1, TGFB1) was decreased. By contrast, the thrombogenic molecule plasminogen activator inhibitor-1 (PAI-1, SERPINE1) was increased in white CTEPH thrombi (Figure 5A). There was no difference in gene expression of CTGF and leukocyte specific genes (PTPRC, CD68 and myeloperoxidase MPO) between white CTEPH thrombi and parent unthrombosed pulmonary arteries. When other vascular thrombi (acute femoral thrombi, coronary aspirates, organizing aortic thrombi, carotid thrombendarterectomies, acute pulmonary emboli, subacute and organizing deep vein thrombi) were analyzed, white CTEPH thrombi displayed more vessel-specific gene expression than any acute thrombi, but significantly less than any organizing thrombi (Figures 5B and 5C).

Figure 4. Analyses of vascular structures in mouse thrombi, and in red and white portions of PEA specimens.

(A) Trichrome stains of representative IVC mouse thrombi on days 3 (B) and 28, showing close histological similarities to representative (C) red and (D) white CTEPH thrombi. (E) Typical PEA specimen with red (arrowhead) and white (arrow) thrombus components. Trichrome (middle panel) and von Willebrand factor (vWf) immunohistochemical stains (lower panel) of (F, J) unthrombosed pulmonary artery, (G, K) red CTEPH thrombus and (H, L) white CTEPH thrombus from the same patients. (I) Vessel counts (indicated as numbers of cells positive for vWf or PECAM-1 per mm² tissue) in red CTEPH thrombi, white CTEPH thrombi and unthrombosed CTEPH pulmonary arteries (n=10 each), respectively. * indicates p<0.05 compared with red and white CTEPH thrombi. Note that in (K) vWf immunoreactivity is mainly extracellular. Scale bars represent 100 μm in (A), (B), (F), (G), (H), (J), (K), (L), 500 μm in (C), (D) and 1 cm in (E).

Figure 5. Real-time PCR analyses of white CTEPH thrombi, using different vascular tissue comparators (arterial thrombi in panel B, venous thrombi in panel C).

(A) Relative expression of several vessel and leukocyte-specific genes in white CTEPH thrombi compared with unthrombosed pulmonary arteries from the same patients (n=15), with pulmonary artery signals set to 0. (B) Relative expressions of CTGF and several vessel-specific genes (CDH5, KDR, PDPN) in in vitro whole blood clots from CTEPH patients (n=6), in acute femoral thrombi (n=4), in coronary aspirates from acute myocardial infarctions (n=7), in organizing thrombi from aortic aneurysms (n=6), in carotid endarterectomies (n=8) and in red CTEPH thrombi (n=8) compared with white CTEPH thrombi (n=15). (C) Relative expression of vessel-specific genes as in panel B in in vitro whole blood clots from CTEPH patients (n=6), in acute pulmonary emboli (n=5), in subacute deep vein thrombi (n=4), in organizing venous thrombi (n=6) and in red CTEPH thrombi (n=8) with white CTEPH thrombi signal set to 0, and normalized to 18S RNA. * indicates p<0.05 compared with white CTEPH thrombi. CTEPH in vitro clots and red CTEPH thrombi are shown in both data sets.

Low ratio of angiogenesis gene expression/CTGF expression in white CTEPH thrombi

Figures 5B and 5C display angiogenesis gene expression levels in arterial and venous thrombi at different stages of organization, in comparison with white CTEPH thrombi. CTGF expression levels were rising in venous thrombi in the following order: in vitro clots from CTEPH patients, acute pulmonary emboli, subacute deep vein thrombi and organizing venous thrombi occluding deep vein segments that were harvested during variceal surgeries. In arterial thrombi, CTGF gene expression levels were increasing in the following order: acute femoral thrombi, coronary aspirates from acute myocardial infarction and organizing thrombi from aortic aneurysms. In comparator thrombus tissues, but not in white CTEPH thrombi, gene expressions of CDH5, KDR and PDPN were increasing with increasing CTGF levels. Ratios of CDH5/CTGF, KDR/CTGF and PDPN/CTGF expression were lowest in white CTEPH thrombi. Respective gene expressions in red CTEPH thrombi were similar to acute arterial and venous thrombi.

Angiogenic inhibitory activity of red CTEPH thrombus

Following 2-D gel electrophoresis of extracts of representative fresh red and organized white CTEPH thrombi (Figure 4E) as well as of pulmonary artery tissue from the same patient (Supplemental Figure I), protein fingerprinting revealed significant differences in protein profiles from these three tissues. Organized CTEPH thrombi demonstrated the least variety of protein bands (Supplemental Figure IB). To find out whether these lysates contain cell-stimulatory factors, a BrdU assay was employed to analyze DNA synthesis rates of human umbilical vein endothelial cells (HUVEC). Here, stimulation for 24 h with lysates of fresh red and white CTEPH thrombus material as well as from parent pulmonary artery tissue in the absence or presence of VEGF was assessed, while DNA synthesis rates with basal medium served as control. Material from white CTEPH thrombi and CTEPH pulmonary artery tissue were found to stimulate DNA synthesis in HUVEC in a dose-dependent manner over 24 h (Figure 6A), similar to lysates from healthy pulmonary artery tissue (data not shown). Heat-induced denaturation of proteins in these lysates reversed their stimulatory effect (data not shown), indicating that thrombus-bound proteins were responsible for the mitogenic activity. The stimulatory activity of lysates at 40 μg/ml was equivalent to VEGF alone at 50 ng/ml, and both, lysates together with VEGF, exhibited an additive effect on DNA synthesis rates in HUVEC. By contrast, red CTEPH thrombi did not influence DNA synthesis in HUVEC.

Figure 6. Proliferative and sprouting activity of red CTEPH thrombus, white CTEPH thrombus and CTEPH pulmonary artery lysates.

(A) Activity of protein lysates from CTEPH thrombi and pulmonary arteries was verified by their effect on DNA synthesis rates of HUVEC in a BrdU assay. The DNA synthesis rates after 24 h incubation with different concentrations of homogenized red CTEPH thrombi, white CTEPH thrombi or CTEPH pulmonary artery tissue (n=3 each) in absence or presence of 50 ng/ml VEGF (positive control) are demonstrated. Values determined in the absence of any additives to basal medium (BM) were set at 100% (control). (B) Inhibitory effects of red CTEPH thrombus on HUVEC sprouting. HUVEC spheroids were stimulated with red CTEPH thrombi, white CTEPH thrombi, or CTEPH pulmonary arteries (n=3 each) in the absence or presence of 25 ng/ml VEGF (positive control) for 48 h. HUVEC spheroids stimulated in the absence of any additives to BM served as controls. Bars illustrate mean total lengths of sprouts. * indicates p<0.05 compared with red CTEPH thrombi, ▼ compared with BM, ▲ compared with VEGF in addition to BM.

In an in vitro 3-D angiogenesis assay isolated tissue lysates did not induce sprout formation, whereas red thrombus material at 40 μg/ml inhibited VEGF-induced sprouting by about 15% (Figure 6B). Material from white CTEPH thrombi as well as unthrombosed pulmonary artery tissue did not affect VEGF-induced endothelial cell sprouting.

Discussion

VEGFR-2 (KDR) is a type III transmembrane kinase receptor ²⁵ and the predominant endothelial cell receptor to promote VEGF functions through specific intracellular signaling cascades leading to proliferation, migration, survival and increased permeability, each of which contributes to the angiogenic response ²⁶. Endothelial-cell-specific deletion of this pathway allowed us to demonstrate the effects of angiogenesis in a model of vascular thrombosis. We were able to confirm the results derived from the animal studies in CTEPH which is a human model disease for thrombus non-resolution.

The Tie-2 receptor is expressed by endothelial cells and hematopoietic stem cells ²⁷. Transgenic mice expressing inducible CRE recombinase under the control of a Tie2 promoter/enhancer quantitatively excise loxP-flanked (floxed) genes in endothelial cells ^28-30. Tie-2 expressing monocytes and macrophages have been shown to regulate revascularization of the ischemic limb ³¹. However, we do not believe that in our model monocytes and macrophages are affected by Kdr deletion because there was no difference in gene and protein expression levels between Kdr^flox/flox/Tie-2CreER before and after TX treatment and between Kdr_Δend and control littermates (Figures 1C and 1D). The targeted gene deletion in this model primarily addressed angiogenesis. We found that baseline KDR expression in monocytes was extremely low in our model; values were similarly low as in a previously reported study ³². Only less than 0.25% monocytes expressed KDR. Kdr mRNA expression was 1000 times less in monocytes than in endothelial cells (Figure 1C). In addition, it has been shown that human monocytes express only VEGFR-1 (Flt-1) and not KDR, whereas endothelial cells express both receptors ^33-34.

In our study, Kdr_Δend were employed in a stagnant flow venous thrombosis model ¹², where thrombus formation and thrombus resolution are inextricably overlapping. Kdr_Δend did not have a prothrombotic phenotype, as tail bleeding times and the percentage of total and activated leukocyte/platelet aggregates and monocyte/platelet aggregates did not differ from controls (Figures 1E and 1F), but the lack of Kdr impacted thrombus early on, leading to larger thrombus by days 1-3 (Figure 2L). Decreased numbers of microvessels in Kdr_Δend were primarily apparent during days 7-14, while vessel density was increasing by day 28 (Figure 3E, Supplemental Table I) which may be enhanced by thrombus shrinkage in later stages of this model. Because VE-cadherin as well as VEGF and Kdr-mRNA levels were significantly decreased between ligation and day 7 (Figure 3F-H), we conclude that early compromise of angiogenesis in this model accounts for the difference in thrombus size. This concept is further supported by the analysis of changes in thrombus volumes with the largest difference between days 3 and 7 (Figure 2L), and by published data showing that recanalization may occur within 24 hours of thrombus formation ⁷.

Leukocyte recruitment and especially macrophages play a central role in thrombus resolution, and impaired monocyte recruitment delayed thrombus resolution ^35-36. Monocytes secrete chemoattractants, growth factors, and proteases ⁷, thus directing the chemokine/cytokine milieu to resolution. Areas of venous thrombi containing large numbers of monocytes, are also rich in neovascular channels ⁷. In our model, expression of CD68 and Ptprc was attenuated during the early phase of thrombus resolution in Kdr_Δend (Figures 3N and 3O). There was no evidence that monocytic Kdr is targeted by the gene deletion, suggesting that angiogenesis may have to occur first, to enable monocyte recruitment into resolving thrombi.

A possible intrinsic effect of TX and CRE recombinase on angiogenesis and thrombus resolution was excluded by experiments showing similar thrombus resolution rates in Kdr^flox/flox (with and without TX administration) and in Kdr^flox/flox/Tie-2CreER mice compared with Kdr^flox/flox (both without TX administration), respectively (data now shown). Repeated injections of TX after ligation might have led to a more complete suppression of Kdr expression and might have prevented the re-appearance of endothelial cells and Kdr expression by day 28 (Figure 3H). TX treatment was terminated one week prior to ligation to minimize toxicity and avoid a direct effect of the compound on venous thrombus resolution. The re-emergence of Kdr expression by day 28 in Kdr_Δend may derive from other cells, such as macrophages and bone-marrow derived endothelial progenitor cells (Figure 3H) that invade the late thrombus, and may be enhanced by thrombus shrinkage in later stages of this model. Many other signaling pathways beyond VEGF participate in the formation of new vessels, e.g. notch/delta, ephrin/Eph receptor, roundabout/slit, and netrin/UNC receptor families, as well as intracellular proteins such as hedgehog and sprouty ³⁷. It is possible that Kdr_Δend compensated the loss of Kdr by one of these angiogenic signaling pathways, contributing to an increase in microvessels in Kdr_Δend by day 28.

After a successful proof-of-concept demonstration that angiogenesis in the animal model is a key component for resolution of experimental thrombosis, angiogenesis was investigated in the vascular obstructive material of CTEPH. CTEPH is a rare and late sequela of venous thromboembolism (VTE), which is characterized by non-resolving thrombi in the pulmonary arteries (Figure 4E), and is therefore an ideal model disease to study the vascular remodeling of thrombosis.

Parent pulmonary artery is a traditional comparator tissue for CTEPH thrombi because it eliminates patient-to-patient variability. In addition, pulmonary artery is characterized by the same main components, i.e., endothelial cells, smooth muscle cells, fibroblasts, and occasional inflammatory cells as CTEPH vascular occlusions. Previously published work of our group has utilized parent pulmonary artery as control tissue for gene expression studies in CTEPH thrombus ¹⁹. Red appositional CTEPH thrombus corresponds to erythrocyte-rich non-organized thrombus that is thought to originate from slow-flow. In addition, red thrombus composition is characterized by very few vascular resident cells such as smooth muscle cells and myofibroblasts (Figure 4), and significantly differs from white thrombus by proteomic analyses (Supplemental Figure I).

Lower gene expression levels of KDR were found in white CTEPH thrombi compared to pulmonary arteries form the same patients (Figure 5A). Not only KDR, but also other genes important for endothelial cells (e.g. ANGPT2, PECAM1, CDH5, PDPN, TEK, VEGFA, BMPR2 and TGFB1) were expressed at lower levels in CTEPH thrombi compared with pulmonary arteries. Changes in gene expression may reflect different proportions of cell types (especially of endothelial cells) present in CTEPH vascular scars versus relatively normal pulmonary artery walls. Lymphangiogenesis appears to be more important in arterial thrombosis (Figure 5), and remains to be characterized in the vascular remodeling of venous thrombosis. In accord with reported data, PAI-1 showed an increased expression level in CTEPH thrombi ¹⁹ (Figure 5A). In addition to its role as the main fibrinolysis inhibitor in humans, PAI-1 was shown to inhibit cerebral angiogenesis ³⁸ and sprout formation of HUVEC and human lung microvascular endothelial cells ³⁹. Moreover, pharmacologic inhibition of PAI-1 promoted angiogenesis in a mouse model of hind-limb ischemia ⁴⁰.

Because CTEPH is a rare disease we may assume that acute thrombi that were harvested as controls were unlikely not to resolve. Under these assumptions, we performed relative gene expression analyses using CTGF as a marker for fibrotic transformation (i.e., thrombus age) and KDR, CDH5 and PDPN as markers for angiogenesis. The relative quantity of CTGF mRNA as a molecular marker of fibrosis has demonstrated a strong correlation with thrombus collagen content ¹². The first observation was that fresh red appositional CTEPH thrombi contained more CTGF than acute pulmonary emboli. In addition, white CTEPH thrombi at the same level of CTGF expression as organizing deep vein thrombi were characterized by significantly reduced angiogenic gene expression. Comparisons of white CTEPH thrombi with vascular thrombi of different origin and age suggested discordant gene expressions, with a lower angiogenic gene/CTGF expression ratio of white CTEPH thrombi.

Despite islands of vessels, particularly at the borders of thrombi (data shown in ¹⁹), CTEPH thrombi were overall less vascularized than pulmonary arteries, which was confirmed by decreased vWf and PECAM-1-positive cell counts per mm² (Figure 4I). Our observations are corroborated by recent data showing angiostatic factors such as platelet factor 4, collagen type I and interferon-gamma-inducible 10 kD protein (IP-10) within CTEPH thrombi ⁴¹.

Therefore, experiments were designed to understand the biochemical composition of pulmonary vascular thrombus in CTEPH. Thrombus fingerprinting by 2-D gel analysis illustrated that red CTEPH thrombus was significantly different from white organized CTEPH thrombus, and thrombus-free CTEPH pulmonary artery (Supplemental Figure I). DNA synthesis rates of HUVEC in response to incubation with PEA material served to illustrate biological activity of homogenates. When the activity of lysates was tested in an in vitro 3-D angiogenesis assay, red CTEPH thrombi inhibited VEGF-induced HUVEC spheroids sprouting, while white clots and pulmonary arteries were neutral (Figure 6B). There are two explanations: either white CTEPH thrombi and parent unthrombosed pulmonary artery tissue have a stimulatory factor not present in red CTEPH thrombi, or all three have a stimulatory factor and red CTEPH thrombi also carry an inhibitory factor. As previously reported ¹⁹ healthy pulmonary artery did neither differ from unthrombosed CTEPH pulmonary artery in a BrdU assay nor in an in vitro 3-D angiogenesis assay (data not shown). These results suggest that the paucity of vessels in white CTEPH thrombus may be enhanced by the angiostatic activity of red thrombus.

One of the limitations of the present work is that we have drawn conclusions on thrombus biology in CTEPH, which is a chronic condition that is evolving over months to decades after the initiating VTE, based on experimentation in an acute thrombosis model in the mouse. In addition, despite striking histological similarities between mouse and human thrombi (Figures 4A-D), stagnant flow venous thrombosis in the caval vein may not completely replicate pulmonary embolism. A further limitation of the study is the use of tissue lysates in the in vitro experiments, and the focus on soluble proteins rather than cellular components of endarterectomy specimens. Comparisons with arterial thrombosis may be biased by a component of atherosclerotic inflammation which may be distinct from venous thrombosis. Plasmin and matrix metalloproteinase-related mechanisms play important roles in venous thrombus resolution ⁴², yet those remain to be studied in Kdr_Δend.

The results of our study suggest that angiogenesis and an intact angiogenic response during the early phase of thrombus resolution are key mediators of normal thrombus resolution. Analysis of PEA specimens illustrates a paucity of vessels and low ratio between thrombus angiogenesis and fibrosis, supporting the hypothesis that deficient angiogenesis is a key biological mechanism of occlusive vascular remodeling after VTE. Clinical conditions that have been found associated with CTEPH, e.g., splenectomy, infection or phospholipid-antibodies must be examined for their possible interaction with the angiogenic response during thrombus organization. Insights into these processes might be helpful in tailoring more specific and effective therapies for VTE, preventing CTEPH.

Significance.

Mechanisms underlying the resolution of vascular thrombi remain unclear. In the present work, we show that endothelial cell-specific deletion of vascular endothelial growth factor receptor-2 entails a deficiency of thrombus-associated vessels, resulting in thrombus persistence. Chronic thromboembolic pulmonary hypertension (CTEPH) is a human model of venous thrombus non-resolution that closely resembles the vascular remodeling of thrombosis created in the present transgenic mouse model. Findings in human endarterectomy specimens from CTEPH patients lend support to the concept of misguided thrombus resolution in CTEPH as a sequela of insufficient thrombus angiogenesis.

Abbreviations

Cdh5/CDH5

vascular endothelial cadherin

CRE

Cre recombinase

CTEPH

chronic thromboembolic pulmonary hypertension

Ctgf/CTGF

connective tissue growth factor

HUVEC

human umbilical vein endothelial cells

IVC

inferior vena cava

Kdr/KDR

kinase insert domain protein receptor

PEA

pulmonary endarterectomy

PDPN

podoplanin

Ptprc/PTPRC

protein tyrosine phosphatase, receptor type, C (common leukocyte antigen CD45)

TX

tamoxifen

VTE

venous thromboembolism

Vegfa/VEGFA

vascular endothelial growth factor A