Flow-induced glycocalyx formation and cell alignment of HUVECs compared to iPSC-derived ECs for tissue engineering applications

The relevance of cellular in vitro models highly depends on their ability to mimic the physiological environment of the respective tissue or cell niche. Static culture conditions are often unsuitable, especially for endothelial models, since they completely neglect the physiological surface shear stress and corresponding reactions of endothelial cells (ECs) such as alignment in the direction of flow. Furthermore, formation and maturation of the glycocalyx, the essential polysaccharide layer covering all endothelial surfaces and regulating diverse processes, is highly dependent on applied fluid flow. This fragile but utterly important macromolecular layer is hard to analyze, its importance is often underestimated and accordingly neglected in many endothelial models. Therefore, we exposed human umbilical vein ECs (HUVECs) and human induced pluripotent stem cell-derived ECs (iPSC-ECs) as two relevant EC models in a side-by-side comparison to static and physiological dynamic (6.6 dyn cm−2) culture conditions. Both cell types demonstrated an elongation and alignment along the flow direction, some distinct changes in glycocalyx composition on the surface regarding the main glycosaminoglycan components heparan sulfate, chondroitin sulfate or hyaluronic acid as well as an increased and thereby improved glycocalyx thickness and functionality when cultured under homogeneous fluid flow. Thus, we were able to demonstrate the maturity of the employed iPSC-EC model regarding its ability to sense fluid flow along with the general importance of physiological shear stress for glycocalyx formation. Additionally, we investigated EC monolayer integrity with and without application of surface shear stress, revealing a comparable existence of tight junctions for all conditions and a reorganization of the cytoskeleton upon dynamic culture leading to an increased formation of focal adhesions. We then fabricated cell sheets of EC monolayers after static and dynamic culture via non-enzymatic detachment using thermoresponsive polymer coatings as culture substrates. In a first proof-of-concept we were able to transfer an aligned iPSC-EC sheet to a 3D-printed scaffold thereby making a step in the direction of vascular modelling. We envision these results to be a valuable contribution to improvements of in vitro endothelial models and vascular engineering in the future.

To define optimal parameters for alignment and elongation of endothelial cells (ECs) under flow conditions, HUVECs were seeded on ibidi µ-slides (channel heights 0.2 and 0.4 mm) and cultured under different volume flow rates. Various shear stresses in a physiological range from 0 (static culture) to 10 dyn cm -2 were applied in a preliminary screening for up to 96 h. After 48 h, cellular alignment and elongation was already observed for ≥6.6 dyn cm -2 , whereas cells cultured with 3.3 to 5 dyn cm -2 required 72 h for alignment. Alignment and elongation were observed at ≥1.7 dyn cm -2 after 96 h. High shear stress conditions (10 dyn cm -2 ) resulted in a reduction of cellular confluency after 96 h, as indicated by black areas. Based on these results, we finally decided to apply 6.6 dyn cm -2 throughout the remaining experiments, which at the same time yields a fast alignment without being apparently harmful to cells for ≥96 h. Adaption of the experiment to iPSC-ECs demonstrated their high capacity for alignment and elongation, as the cells were (already) aligned after 24 h under all flow conditions.

Directionality analysis
To qualitatively analyze and visually highlight the influence of varying culture conditions on the orientation of ECs as already shown in section 1 of this supplementary material, phase contrast images of ECs after static and dynamic culture were pre-processed using ImageJ and subsequently an orientation/angle-specific color coding was added using the OrientationJ plugin [1]. For preprocessing, the background was subtracted from the raw phase contrast images (rolling ball, radius 100 px), a Gaussian filter (σ = 1 µm, scaled), and a bandpass filter (down to 20 µm, no lower limit) were applied to compensate for inhomogeneous illumination and smoothen the image [2]. The image is binarized via (auto)-thresholding according to Huang et al. [3], and finally processed/colored using the OrientationJ plugin (local window: σ = 3 pixel). Color map is showing hue: orientation, saturation: coherency and lightness of the original image. ImageJ was further used to quantify the effect of culture conditions on cellular alignment/orientation. Initially, the background of phase contrast images was subtracted (rolling ball, radius = 100 pixels) and the contrast of images was increased via gamma correction (γ = 0.8) and contrast stretching (ImageJ's enhance contrast function, cutoff parameter = 0.7% saturated pixels). Using the find maxima function (prominence = 25, excluded on edges, light background), the cell centers were determined and the image was segmented into single cell areas accordingly. The Extended Particle Analyzer of the BioVoxxel toolbox was used to determine/measure the main orientation of single cells 1 [4].

Evaluation of native glycocalyx thickness
To estimate glycocalyx thickness(es) under native, non-fixed conditions, ECs after static culture were stained with wheat germ agglutinin (WGA, labelled with Alexa™ 555, ThermoFisher) as well as CMFDA and subsequently imaged via confocal microscopy. Due to the behavior of thermoresponsive polymers and despite using an incubation chamber as well as rapid sample preparation, unfixed cells on PGE-coated substrate started to detach during image acquisition (data not shown). The experiment was therefore performed on tissue culture-treated polystyrene slides (TCPS). To exclude any influence of the cell culture substrate on glycocalyx expression, HUVECs and iPSC-ECs were seeded on TCPS and PGE-coated PS, stained with WGA and imaged via confocal microscopy (after fixation with PFA). As demonstrated in Supplementary Figure S4, no difference in glycocalyx expression as indicated by mean fluorescence intensity of WGA-staining was observed between both substrates for each cell type, respectively.

Evaluation of the PBMC adhesion assay
A peripheral blood mononuclear cell (PBMC) adhesion assay was established to investigate glycocalyx functionality. Supplementary Figure S6A illustrates the processing of the raw data and analysis. HUVECs were incubated with fluorescently labeled PBMCs, washed, fixed and subsequently imaged using fluorescence confocal microscopy. The images were processed with ImageJ (conversion from 16 to 8 bit, application of a gaussian filter, σ = 1 µm and manual thresholding ranging from 30 to 255) and analyzed using ImageJ's Particle Analyzer, counting the number of PBMCs per image (size: 30-125 pixel 2 ). Various PBMC concentrations were tested and ECs were differentially treated to validate the assay. As a positive control, HUVECs were incubated for 6 hours with 10 or 50 ng mL -1 of the proinflammatory cytokine TNF-α, which is known to increase leukocyte adhesion to ECs via upregulation of adhesion molecules (CAMs), and concomitant shedding of their glycocalyx [5,6]. In addition, a concentration-dependent increase in adherent PBMCs was demonstrated. Finally, an apparent increase in PBMC adhesion after mechanical disruption of the glycocalyx via excessive pipetting demonstrated the assay's suitability to examine glycocalyx functionality.

Supplementary
Subsequently, an analysis of PBMC adhesion to HUVECs and iPSC-ECs after static culture on PGEcoated PS as culture substrate was conducted. Results demonstrated no significant difference in leucocyte count on the monolayer surface for both cell types (Supplementary Figure S7).
Supplementary Figure S7. Comparison of relative PBMC adhesion to HUVECs and iPSC-ECs after 96 h of static culture. Relative data are presented as mean ± SEM with respect to "HUVECs static" (n=4, Mann-Whitney Test).

Cell sheet detachment after 48 h
As previously shown in Supplementary Figure S1, the application of 6.6 dyn cm -2 for 48 h is sufficient for a pronounced alignment of both, HUVECs and iPSC-ECs on TCPS. Since one objective of this study was to generate confluent cell sheets of different endothelial cells, both under static (non-aligned) and dynamic (aligned) culture conditions, we attempted to detach cell sheets from PGE-coated PS first after 48 h. Supplementary Figure S8A demonstrates an elongation and alignment of HUVECs and iPSC-ECs on PGE-coated PS comparable to TCPS. By transferring the slides into room temperature PBS -, we triggered singularization and partial detachment of ECs under all culture conditions (Supplementary Figure S8B).
Supplementary Figure S8. Detachment of HUVECs and iPSC-ECs after 48h of static or dynamic culture. Phase contrast images were taken (A) before and (B) after (incomplete) detachment. (Arrow indicates flow direction. Scale bars: 100 µm, n=2).

Formation of stress fibers due to fluid flow
The application of shear stress to ECs results in the formation of stress fibers, which are known to increased cellular adhesion to the underlying substrate [7]. Supplementary Figure S9 displays confocal images of endothelial focal adhesions after static and dynamic culture conditions, stained with phalloidin (red) and vinculin (green), demonstrating the excessive formation of stress fibers after exposure to shear stress.
Supplementary Figure S9. Confocal images showing focal adhesions after static and dynamic culture of HUVECs and iPSC-ECs for 96 h. Cells were stained for phalloidin (red) and vinculin (green). Nuclei were counterstained with Hoechst 33342 (blue). (Arrow indicates flow direction. Scale bars: 10 µm).