Acute experimental venous thrombosis impairs venous relaxation but not contraction

Abstract

Background

Venous thrombosis (VT) damages the vein wall, both physically with prolonged distension and from inflammation. These factors contribute to post thrombotic syndrome (PTS). Interleukin-6 (IL-6) may play a role in experimental PTS and vein wall responses. Prior assessment of post thrombotic vein wall injury has used static measures such as histology and immunologic assays. The purpose of this study was to use myography to quantify the changes in contraction and relaxation of murine vessels exposed to an acute VT.

Methods

Wild type (WT) C57BL/6 mice were used to determine baseline vein wall passive tension on a DMT 610m myograph, including dosing concentrations of phenylephrine (Phe) and acetylcholine (Ach). WT and IL-6^−/− mice underwent VT using inferior vena cava (IVC) ligation (complete stasis) and stenosis (partial stasis) and no surgery mice as controls. Mice were harvested at 2 days (2D) and analyzed using a myograph. Vessels were stimulated with Phe and Ach to stimulate a contraction and relaxation response. Endothelial responses to VT were quantified by CD31 immunohistochemistry, Greiss assay, PCR, and Evan’s blue assays.

Results

Optimal passive tension was determined to be 2 millinewtons (mN), with the optimal concentrations of Phe and Ach being 7E-3M and 1E-5M respectively. There was no significant difference in contraction when exposed to Phe between the WT control, WT 2D ligation, and WT 2D stenosis IVC segments and the IL-6^−/− mice with or without thrombus (all P > .05). When treated with Ach, there was significantly more relaxation in the non-thrombosed control IVC segments than those mice IVC segments that had a 2D thrombus of either ligation or stenosis derived thrombotic mechanisms in both WT and IL-6^−/−. CD31 staining showed ~20% less luminal endothelium post stasis thrombosis (p = <.01) but no loss in controls (p>.05). Evan’s blue showed a trend of increased leakiness in post-thrombotic vein walls. No significant difference in endothelial gene markers or NO production was found.

Conclusions

When compared to controls, acute thrombosis in total or partial stasis models did not impair IVC contractile responses, suggesting no effect on medial vascular smooth muscle response. The relaxation response was significantly reduced in the post thrombotic groups, likely from direct endothelial injury. These findings suggest, at acute time points, VT impairs the endothelial function of a vein wall while retaining the vascular smooth muscle cell function, and may be a mechanism that promotes PTS.

Table of Contents Summary

Post thrombotic syndrome is complex. We show that experimental acute DVT in the mouse cava causes mild endothelial damage with resultant impaired relaxation by myography but not venous contraction.

Introduction

Deep venous thrombosis (DVT) is a significant health care problem in the United States, with over 250,000 patients affected yearly, and at least 200,000 diagnosed with pulmonary embolism (PE).¹ The incidence of venous thromboembolism (VTE) is similar to stroke and myocardial infarction, with a death rate of 15% and a recurrence rate of 33% at 10 years,² resulting in costs of billions of dollars per year.³ Current DVT therapy primarily involves anticoagulation⁴ which, while being an effective treatment strategy for many patients, carries with it significant bleeding risks and the presentation of contraindications to anticoagulation in many patients.⁵ The late DVT consequence, post-thrombotic syndrome (PTS), affects 6 to 7 million patients,^{4, 5} with between 400,000 and 500,000 of these patients developing skin ulcerations,^{6, 7} and negatively impacts their quality of life.¹⁰

While impaired thrombus resolution in humans correlates with PTS,^{6, 8} no therapies directly address this entity,^{5, 9} with direct-acting oral anticoagulants (DOACs) having little effect on PTS.¹⁰ Invasive approaches to DVT include active pharmacomechanical thrombolysis, but this is a resource-intense and costly therapy.¹¹ Moreover, the CaVenT and ATTRACT thrombolysis trials have failed to produce paradigm changing therapies for PTS,^{12, 13} including no significant benefit compared with anticoagulant therapy in the latter trial.

Experimental VT models have recapitulated that which likely occurs in humans, with areas of occluding stasis DVT and areas of peri-thrombus blood flow.¹⁴ Numerous mechanisms may injure the vein wall, including early mechanical stretch,¹⁵ direct endothelial injury or loss,¹⁶ and pro-inflammatory cytokines and chemokines.^{17, 18} For example, a prototypical cytokine that has human biomarker correlation with PTS as well as potential mechanism is interleukin-6 (IL-6).

Histological and biomechanical assessments of post-thrombotic vein wall injury have been done;¹⁹ however, the post-thrombotic physiological vein wall response has not been characterized, despite the importance of vein contraction and relaxation in its physiological role concerning overall blood volume regulation and blood return. Lack of venous tone may exacerbate valvular insufficiency, contributing to venous hypertension and promotes PTS. The physiological assessments of rat IVC not exposed to thrombus have been characterized, showing increased ex vivo stretch is associated with inflammatory changes in the vein wall.²⁰

This study examined the effects of VT on the contraction and relaxation responses in vessels utilizing myography, in addition to evaluating endothelial damage done to vessel walls.

Methods

Mouse models

Well-described VT models, including complete IVC ligation just below the renal veins (termed ‘stasis’) and IVC stenosis, with a suture tightened just below the renal veins around a 30 gauge needle (termed ‘stenosis’), were utilized to induce a thrombus in experimental animals. C57/BL6J and IL-6^−/− male mice (Jackson Laboratory, Bar Harbor, ME) were utilized for thrombotic models.²¹ Vessels were harvested at two days (2D) post-ligation to simulate acute thrombosis. Groups consisted of no surgery controls, 2D stasis and 2D stenosis animals. Note that the thrombus was gently removed for all assays that follow in those mice who underwent one of the models above

Treatment of all experimental animals was conducted consistent with IACUC standards, with approved protocols. Animals were housed in the animal care facility at the University of Michigan, North Campus Research Complex.

Myography

To ensure vessel reactivity prior to any testing, the vessels were maintained in a physiological saline solution bath supplied with appropriate heat and an O₂/CO₂ gas mixture post-harvest and thrombus removal. Vessels were subjected to a DMT-recommended wake-up protocol to test for vessel viability, the procedure of which is further described in Supplement.

Determination of the passive tension and dose response curves, in addition to the experimental evaluation of both contraction and relaxation responses of IVCs, were conducted utilizing a DMT 610m myograph with data captured using LabChart software (DMT-USA, Inc., Ann Arbor, MI).

Passive tension and appropriate drug dosages were determined as explained in Supplement, allowing the appropriate tension and concentrations to test the vessels to be determined. The vessels were then tested for their contractile and relaxation capabilities by being mounted to the myograph and exposed to Phe and Ach respectively. The vessels were then contracted with Phe, allowing the contraction to occur for 15 minutes and plateau, after which time force readings were determined from LabChart (Contraction Plateau – Baseline). Vessels were then exposed to acetylcholine, with relaxation readings in LabChart being determined after plateau was established after 15 minutes (Relaxation plateau – baseline).

Quantitative Real Time PCR

Harvested IVCs were weighed and then flash frozen with liquid nitrogen until isolation of the RNA could occur. Total RNA from each sample were reverse transcribed to cDNA using the RT2 first Strand Kit (Qiagen). cDNA was analyzed by real time PCR using RT2 SYBR Green Rox qPCR Mastermix (Qiagen). Commercially available primer sets were purchased from Qiagen; β-actin, cat# PPM02945B; vWF, cat# PPM05310E; and eNOS, cat# PPM03801A. The reactions were run on an Applied Biosystems StepOnePlus system (Thermo Fisher, USA) and quantified using the ΔΔCt method with β-actin as the endogenous control gene. Three individual experiments were performed for each sample for each gene.²²

Evan’s Blue Assay

Fresh 0.5% Evan’s Blue (Sigma, St. Louis, Missouri) was made in sterile PSS the day of the harvest of the IVCs. Each mouse was injected intravenously with 0.2mL of the Evan’s Blue solution and given 30 minutes to allow the dye to perfuse the animals. After 30 minutes, the mouse was exsanguinated and the IVC was harvested and measured with a micrometer (Digimatic, Mitutoyo Corp., Japan). The IVC was then placed on weigh paper for 15 minutes, to allow the IVC to dry, providing a more accurate weight for further analysis. The dry weight was taken and the IVC was placed directly into 150μl of formamide (Sigma). All tubes were kept on ice during harvesting of additional IVCs to preserve tissue quality. The IVCs in formamide were placed in a hot bath at 55°C for 48 hours. The solution was vortexed, the IVC removed, and 100pL of the formamide was placed in a 96-well plate and read for absorbance at 610nm with a blank formamide included in each plate. For analysis, the samples were first corrected for the formamide blank, and then corrected for length.

Griess Reaction for Nitrite Determination

The IVC of the mice were excised and placed in a 24 well plate with 700uL of PSS buffer solution. The IVCs were stimulated with recombinant ILl-β (Life Technologies) at 1.4ng/mL for 18 hours at 37°C in humidified 5% CO2.¹⁶ Manufacturer’s instructions were followed (Molecular Probes, G-7921) with small modifications described below. As a control, a blank with 20pL of Griess reagent and 280pL of deionized water was made. Absorbance of the nitrite containing samples was measured at 548nm and quantified relative to the reference sample. The content of tissue nitrite was calculated by a linear standard curve obtained from known amounts of the nitrite standard and normalizing to tissue weight μg/g).²³

Histology

Fresh tissue (IVC with thrombus intact) was fixed in 10% buffered formalin for two hours, transferred to 70% ethanol, and subsequently embedded in paraffin for immunohistochemistry (5 μm tissue sections, 3 sections per mouse). After blocking non-specific sites with species-specific serum, sections were stained for endothelial cells by CD31 (BD Pharmingen, San Diego, CA). A species-specific kit was used. DAB (diaminobenzidine) was used to develop the color and the slides were counterstained with hematoxylin (Vector Laboratories, Inc., Burlingame, CA) and cover slipped. Positive staining was assessed for completeness and presence of endothelial cells around the lumen using ImageJ. Three sections per mouse were analyzed.

Statistical Analysis

GraphPad Prism software version 6.0 was used to analyze the data, which is presented as the mean +/− the standard deviation (SD). Data was first analyzed for normal distribution and then statistical significance between multiple groups was determined using one-way analysis of variance followed by Newman-Keuls post hoc test. All data are representative of at least two independent experiments. A P-value of less than or equal to 0.05 was considered significant.

Results

Myographic evaluation of control and post-thrombotic vessel contraction and relaxation

Passive tension on the mouse IVC was evaluated from 1.5 mN-4 mN, increasing in 0.5 mN intervals. The vessels tested showed the most robust contraction at 2 mN of force, giving the passive tension the vessels were tested at. Data from the passive tension evaluation is in Figure 1a. The Ach dose response curve showed the 1×10⁻⁵M concentration gave the most robust relaxation response. Results of the ACh dose response curve are shown in Figure 1b.

Figure 1.

A) Force readings gathered from IVCs at each passive tension were tested, allowing for a determination of the optimal passive tension for evaluation of the contractile and relaxation ability of the vessels. 2 mN was found to be optimal, which is shown above as the most robust contraction. Contractions were elicited utilizing 96 mM KPSS solution. B) The IVC response to increasing concentrations of acetylcholine, with the relaxation response being most prominent at 1E-5 M. Vessels were contracted prior to Ach exposure by Phe at 7E - 3M.C) The IVC response to increasing concentrations of phenylephrine to elicit contraction is shown above. 7E - 3M was chosen to avoid potential damage at higher dosages of phenylephrine.

Analysis of the Phe dose response curve in Figure 1c showed the 7×10⁻³M concentration resulted in the best contraction and stable contraction. Higher doses of phe result in an oscillatory behavior of the vessels between contraction and relaxation when exposed to Ach following the phenylephrine.

The vessels were exposed to a thrombus for a two-day time period to simulate the effects of an acute VT.²¹ Upon examination of the contractile responses using Phe dosing as stated above, in both the WT and IL-6^−/− groups, there was no statistically significant difference between the groups (Figure 2a).

Figure 2.

A) There were no significant differences between the no surgery, ligation, and stenosis animals in IL-6^−/− mice as compared to WT. B) Significant differences were found in the relaxation response of both the WT and IL-6^−/− mice when the controls were compared to the stasis and stenosis groups. N = 7 - 8. P<.05 by ANOVA with Newman Kuels post-test.

The relaxation responses using Ach at 7×10-3M showed marked impairment, with the WT controls relaxing significantly more than the WT 2D ligation (p = 0.0051) and WT 2D stenosis groups (p = 0.0006) (Figure 2b). The IL-6^−/− control group also relaxed significantly more than both the IL-6^−/− 2D ligation (p = 0.0036) and IL-6^−/− 2D stenosis groups (p = 0.0059). There was no statistically significant difference in relaxation between the ligation and stenosis groups for both the WT (P=0.066) and IL-6^−/− (P=0.723) (Figure 1C).

Thrombus sizes were determined in separate groups of mice, and no significant differences were found at 2 days comparing WT stasis, WT stenosis, IL-6−/− stasis and IL-6−/− stenosis (stasis = 20 ± 1, stenosis = 12 ± 4, IL-6−/− stasis = 14 ± 1, IL-6−/− stenosis = 12 ± 2; mg/cm, N = 5-8, P = NS by ANOVA with Bonferonni multiple comparisons test).

Given that the myographic responses were similar between the stasis and stenosis VT models, we only performed further physiological and gene expression analysis on stasis model mice that follows.

Evans blue assay

Evans blue assay was performed to determine the ‘leakiness’ of the endothelium as the dye is circulated prior to sacrifice to allow for perfusion of the animal. Evans blue stains sub-endothelial tissue but not endothelium. Post-thrombotic vein walls demonstrated a trend toward increased vessel leakage as shown with greater vein wall eluted dye in both the ligation and stenosis VT groups (p = .08) overall and significantly more dye eluted in between the WT no surgery and ligation groups (P=0.0475) (Figure 3a).

Figure 3.

A) Evans blue assay showed increased release of bound dye, consistent with less endothelial presence in the stasis and stenosis groups, but was not significant by ANOVA (N = 4-5). B) Greiss reaction for Nitrite showed no difference in control or post thrombotic groups (N = 4-5). C) Endothelial cells stained positive in the lumen of both control and post thrombotic 2D stasis animals, showing the cells were present in both groups of animals. Evaluation of staining revealed 22% less endothelium in the 2D stasis mice when compared with controls (n = 3; P=.002 by t-Test).

Griess assay

Another measure of endothelial function is nitric oxide (NO) production. Evaluation of the ex vivo vein wall for production of NO by measurement of nitrate was assessed by the Griess reaction assay. We found in both models there were no significant differences in the capacity of the vein to produce NO when stimulated with interluken-1 beta (IL-1β) post-thrombosis (Figure 3b) as compared with controls.

CD31 staining

The luminal brown staining indicates CD31⁺ cells, confirming the presence of endothelial cells, and suggested neither the harvesting process nor the myographic testing caused intimal damage. However, about 22% less luminal staining was found in 2d post-thrombotic ligation samples (N = 3 - 4) (p =.002) (Figure 3c).

qPCR

Several endothelial markers were examined by gene expression with qPCR. We found there was no significant difference between the controls and the stasisand stenosis models, evaluating eNOS, VEGFR2, and von Willebrand factor (VWF n = 3 - 4) in WT mice (data not shown).

Discussion

We report for the first time the effect of acute VT on the normal physiological vein wall function, as well as the specific techniques to assess mouse IVC myographic responses. We demonstrated that relaxation, but not contraction, was significantly impaired, possibly due to partial loss and impaired endothelial function. These findings also suggest that the early post-thrombotic vein wall, if thrombus is exogenously removed, can respond to contractile stimuli but may have prolonged relaxation impairment. These findings may have clinical ramifications for humans. When there is rapid DVT resolution, PTS may be less likely, whereas prolonged obstruction with valve dysfunction is associated with PTS.^{5, 24} However, whether this translates in humans is much less clear, particularly given the recent publication of the ATTRACT trial.¹³ How a post thrombotic vein with impaired relaxation may ultimately affect overall venous function is not clear, but could potentiate venous hypertension through decreased vessel compliance.

To avoid bias, we used two models of VT – the total stasis, with little peri-thrombus blood flow and the stenosis model, that has peri-thrombus blood flow. These models recapitulate the clinical presentations of DVT in humans, simulating areas of total occlusive DVT and DVT with blood flow around it.²⁵ The stenosis model resolves the VT more quickly at later time points such as day 8 and beyond, with more mice having a smaller VT or no VT as compared with total stasis (unpublished data). Both models produced the same effect measured by myography; namely impaired relaxation. Consistently, we found that the VT sizes were not significantly different in stasis or stenosis at 2 days, suggesting that a threshold effect, likely related to stretch and thrombus-vein wall contact. Thus, we speculate that given the impaired venous relaxation, but intact nitric oxide production and Evan’s blue assay, that the limited endothelial injury from the 2 day thrombus was the cause. Note however that the harvest and myography did not itself cause endothelial damage.

We assessed the vein wall responses in IL6^−/− mice as well, due to the potential role of this cytokine pathway in humans.²⁶ We found that the global lack of IL-6 does not seem to impact the early thrombus response on the vein wall, at least as determined by myographic assessment. We only assessed IL6^−/− mice by myography, as there were no phenotypic differences between the responses in WT and IL-6^−/− mice; thus the genetic, histological, and Griess reactions were not done in this group.

In conclusion, we believe the myograph assessment provides a new tool to assess post-thrombotic injury and will allow testing of novel agents. Further analysis will focus on the sub-chronic and chronic time points after thrombus development for contraction and relaxation responses, with and without novel therapies, to provide insights into the development of PTS and its prevention.

ARTICLE HIGHLIGHTS.

Type of Research

Basic laboratory investigation into non thrombosed and acute post thrombosed mouse vena cava using myography and other assays.

Key Findings

Mild endothelial damage was observed in those veins exposed to two-day thrombus, created by two models. This was associated with significant impairment in vein relaxation but not contraction.

Take home Message

Post thrombotic experimental vein wall injury affects mainly the relaxation function of the vein, but not contraction.

Venous thrombosis is a common disease that may lead to pulmonary embolism and/or post thrombotic syndrome. The pathophysiology of vein wall damage involves leukocyte driven processes that may damage the vein wall, as well as the intima from the mechanical effects. To date, the resulting physiologic damage of contraction and relaxation has not been examined. Herein, we show acute thrombosis damages the intima to a small extent, but significantly impairs normal relaxation but not contraction. These findings are important as an assay for post thrombotic damage, and lend further insight into post thrombotic syndrome.