Skip to main content

Correlation Between Numerical Estimation of In Vivo Local Sphingosine-1-Phosphate Concentration and Endothelial Permeability

Final Report Summary - S1PWSSECPERMEABILITY (Correlation Between Numerical Estimation of In Vivo Local Sphingosine-1-Phosphate Concentration and Endothelial Permeability)


Atherosclerosis is a disease in which plaques build up inside arteries. Most people have atherosclerotic plaques from their teenage years onwards [1] and the plaques themselves are not dangerous. However, rupture of these plaques is the primary cause of heart attacks and strokes: worldwide each year 7.2 million people die of coronary heart disease and 5.5 million die of stroke [2]. Atherosclerotic plaques are thought to develop when low density lipoproteins (LDL) penetrate the endothelial layer and then oxidise. Inflammation ensues, and macrophages ingest the LDL leading to fatty streaks.
While the progression of atherosclerosis is relatively well known, and risk factors for the disease have been established, the factors involved in the early initiation of the disease are still not clearly understood. What is known is that atherosclerosis is heterogeneously distributed around the vasculature [3]. This spatial heterogeneity implicates flow mediated effects since flow creates variation in the stresses exerted on the arterial wall and affects mass transfer to the wall.
Flow has many effects on the arteries but of all these effects, those acting on permeability of the endothelium are likely to be potent contributors to initiation of atherosclerosis. A major function of the endothelium is the regulation of the transport of molecules into the arterial wall and surrounding tissues.
While shear stress contributes to the permeability of the endothelium there are also many chemicals which alter the permeability via signalling pathways that act upon the actinomyosin cytoskeleton and changes in the level of expression and phosphorylation of junction proteins. Examples of these are: nitric oxide, cyclic AMP, oxidised LDL and reactive oxygen species [4]. More recently the effects of the signalling molecule sphingosine-1-phosphate (S1P) on the endothelium have been studied [5].
S1P is a signalling sphingolipid synthesized intracellularly from sphingosine via phosphorylation of its primary hydroxyl group. S1P is stored primarily in platelets at a high concentration [6] and is released upon their activation. However there are other sources of plasma S1P including other haematopoietic cells and vascular ECs. Plasma concentrations of S1P are 200-1000 nM [5]. Plasma S1P is bound mainly to HDL (50-70 %), albumin (30 %) and then LDL and VLDL (<10 %).
S1P1 is the major S1P receptor expressed on vascular ECs. It is coupled exclusively via Gi to the Ras-MAP kinase, phosphoinositide (PI) 3-kinase-Akt pathway and induces endothelial migration, angiogenesis and decrease in endothelial permeability [7]. Stimulation of barrier integrity occurs in part via upregulation of VE-cadherin and stimulation of intercellular adhesion.
We therefore hypothesize:
Endothelial permeability correlates inversely with the combined local WSS and S1P concentration.


The research in this project proposal was divided into 4 objectives:
1: To determine the relationship between shear stress and S1P released from platelets.
2: To investigate the effect of S1P on endothelial permeability in vivo.
3: To predict the local WSS and S1P concentration using a computational model.
4: To compare the endothelial permeability with WSS and S1P concentration predictions.

Description of work performed

Objective 1: To determine the relationship between shear stress and S1P released from platelets.
Data from [8] were used to inform a numerical model for the release of S1P by platelets. The numerical model, which is a chain reaction, can be conceptualised as the S1P being transported across the membrane of the platelet.

Objective 2: To investigate the effect of S1P on endothelial permeability in vivo.
Endothelial macromolecular permeability measurements require the following major steps:
A: preparation of fluorescent tracer, rhodamine labelled albumin
B: tail vein injection of rhodamine-albumin
C: blood sample collection from heart (open approach) and cannulation of the left ventricle
D: dissection of artery, tissue preparation and mounting on slide
E: confocal microscopy
I am now proficient in steps A to E which required learning and practising a number of techniques, as well as refining protocols to enable measurements specifically on the entire length of the mouse aorta. Code was written to programme the confocal microscope to efficiently scan the irregular shape of the tissue samples thus allowing the large area to be imaged.
Permeability measurements have now been made on 16 mice without injection of S1P and 11 mice with injection of S1P.

Objective 3: To predict the local WSS and S1P concentration using a computational model.
A reaction-convection-diffusion model for the release and transport of S1P in arteries has been developed. The model has 8 species: unactivated platelets, activated platelets, open transporters, closed transporters, S1P-transporter complex, S1P outside the platelets, endothelial S1P receptor, S1P receptor-S1P complex. A system of reaction rate equations was developed based on S1P release data and a suggested release mechanism from the literature. The model was implemented in the open source computational fluid dynamics software ‘OpenFoam’. First the flow was calculated based on realistic flow boundary conditions and having found the velocity field the reaction-convection-diffusion equations were solved. The model has been applied to 2D models of arterial branches.
Flow through the mouse aortic arch was simulated with assistance from Kitty Chow (BEng project student). The arterial geometry was taken from a micro-CT scan of a plastic resin cast of the mouse aortic arch which was produced by Zahra Mohri (Post doc, Weinberg Group). The Vascular Modelling Toolkit (VMTK) was used to segment the geometry from the CT scan, smooth and mesh the surface. Flow boundary conditions were taken from [9] and the open source CFD software ‘OpenFOAM’ was used to calculate the velocity field and wall shear stress. A next step is to implement the transport model for S1P, as described above, into this 3D model of flow in the aortic arch.
The originally proposed method for measuring the boundary conditions and validating the flow simulation was ultrasound image velocimetry (UIV) (or echo-Particle Image Velocimetry). However, this method is still under development and hence a series of validation studies were first performed. Flow rates found by integrating the velocity profile measured using UIV were compared with the flow rate obtained with a transit time flow meter in a flow phantom (5 mm tube) and pulsatile flow, and in the rabbit aorta. Results from the flow phantom were in good agreement, however, the in vivo experiments were less successful. Two issues were identified: in vivo the particles do not stay correlated as long as in the phantom, and in vivo the contrast is reduced by reflections from the surrounding tissue. To address the correlation issue we have developed a novel scanning technique which effectively reduces the time between image frames. Validation studies showed that we can now measure faster velocities than with the original method (up to at least 1.5 m/s).

Objective 4: To compare the endothelial permeability with WSS and S1P concentration predictions.
Calculations of WSS in the arch section of the aorta were compared with permeability measurements in the first 5 mice. No correlation was found with this small sample size but analysis of the remaining data is ongoing.

Significant Results

Influence of S1P on macromolecular permeability of the mouse aorta
This is the first time macromolecular permeability measurements have been made on the whole length of the mouse aorta. In mice without injection of S1P permeability was highest in the aortic arch and iliac bifurcation, as well as in tear shaped rings around the intercostal branch ostia which included the ‘cushion’ on the upstream side. Analysis of the results, including the comparison to mice with injected S1P is ongoing.
2D simulations of Sphingosine-1-Phosphate (S1P) Receptor Activation in Arterial Branches
Calculations of S1P receptor activation without inclusion of platelet released S1P showed the S1P-S1P1 complex concentration was largely uniform, except when the diameter of the branch approached that of the main artery and then upstream concentration was larger than downstream. Introducing S1P internalization made little difference to the complex distribution except when the rate coefficient for the internalization was made very large in which case upstream concentration was again larger than downstream. When platelet released S1P was included the higher shear stress upstream of the branch resulted in significantly larger complex concentration there.

Development of Ultrasound Imaging Velocimetry (UIV)
The initial validation study showed spatial resolution of the velocity field was 0.1 mm in the direction of the ultrasound beam and 0.6 mm perpendicular. Velocities up to 1 m/s could be measured, twice that reported in literature. Steady flow velocity profiles agreed with the theoretical parabolic profile except near the edges of the tube. Pulsatile flow was validated by integrating the velocity profile to find the flow rate and comparing this with measurement of the flow rate made using a transit time flow meter. Results from the tube phantom were in good agreement (< 15 % difference over 80 % of the cardiac cycle) however issues with contrast and correlation were found in the in vivo experiments.
With the new ‘interleaved’ imaging sequence, which allows the interframe time to be varied, we can now measure velocities of at least 1.4 m/s using the whole width of the ultrasound transducer. Using a larger interframe time allows more precise measurement of the velocity, however, during the deceleration phase of the cardiac cycle when transitional or turbulent flow develops, the particles become decorrelated when using the larger decorrelation times. Future work could involve synchronising the interframe time to the cardiac cycle to take advantage of long interframe times during stable accelerating flows and shorter interframe times during decelerating flow.

Potential Impact of Results

The development of Ultrasound Imaging Velocimetry (UIV) is an important improvement on ultrasonic methods for measuring blood velocity. Doppler ultrasound is the current clinical standard, however, only the component of the velocity parallel with the ultrasound beam can be measured. In situations where the angle between the flow and the ultrasound beam is unknown, the velocity cannot be determined. UIV has the potential to increase diagnostic accuracy and in addition will be a useful tool in biomedical research.
Since atherosclerosis is a disease affecting many people the improvement to length and quality of life of EU citizens by preventing people from getting it could be huge. Understanding the basic science behind endothelial permeability, which has implications for atherosclerotic disease development, is vital for development of pharmaceutical therapies. These could, for example, act on the S1P1 receptor. Another industry which would benefit from better understanding of factors involved in atherosclerotic initiation is the medical device industry. Better understanding of the flow mediated effects on the endothelium could lead to improved stent design which minimises the risk of creating new atherosclerotic lesions downstream from the stented ones. If therapeutic targets were identified and pharmaceuticals developed these could also be incorporated in drug eluting stents. In terms of health care, the public sector will benefit in turn from new drugs, devices or life style advice. Economic benefits for the EU could include money saved on health care and revenues from any drugs developed.


[1] J. P. Strong, G. T. Malcom, C. A. McMahon, R. E. Tracy, W. P. Newman, E. E. Herderick and J. F. Cornhill, "Prevalence and Extent of Atherosclerosis in Adolescents and Young Adults," JAMA, vol. 281, pp. 727-735, 1999.
[2] J. Mackay and G. A. Mensah, "The Atlas of Heart Disease and Stroke," [Online]. Available: [Accessed August 2010].
[3] P. A. VanderLaan, C. A. Reardon and G. S. Getz, "Site Specificity of Atherosclerosis: Site-Selective Responses to Atherosclerotic Modulation," Arterioscler Thromb Vasc Biol, vol. 24, pp. 12-22, 2004.
[4] S. Pan, "Molecular Mechanisms Responsible for the Atheroprotective Effects of Laminar Shear Stress," Antioxidants & Redox Signaling, vol. 11, pp. 1669-1682, 2009.
[5] S. Lucke and B. Levkau, "Endothelial Functions of Sphingosine-1-phosphate," Cellular Physiology and Biochemistry, vol. 26, pp. 87-96, 2010.
[6] N. Kobayashi, T. Nishi, T. Hirata, A. Kihara, T. Sano, Y. Igarashi and A. Yamaguchi, "Sphingosine 1-phosphate is released from the cytosol of rat," J Lipid Research, vol. 47, pp. 614-621, 2006.
[7] Y. Takuwa, Y. Okamoto, K. Yoshioka and N. Takuwa, "Sphingosine-1-phosphate signaling and biological activities in the cardiovascular system," Biochemica et Biophysica Acta, vol. 1781, pp. 483-488, 2008.
[8] S. Aoki, M. Osada, M. Kaneko, Y. Ozaki and Y. Yatomi, "Fluid shear stress enhances the sphingosine 1-phosphate responses in cell–cell interactions between platelets and endothelial cells," Biochem Biophys Res Comm, vol. 358, p. 1054–1057, 2007.
[9] Y. Huo, X. Guo and G. S. Kassab, "The Flow Field along the Entire Length of Mouse Aorta and Primary Branches," Annals of Biomedical Engineering, vol. 36, pp. 685-699, 2008.