Endothelial Cells Increase Mesenchymal Stem Cell Differentiation in Scaffold-Free 3D Vascular Tissue

Abstract

Abstract

In this study, we present a versatile, scaffold-free approach to create ring-shaped engineered vascular tissue segments using human mesenchymal stem cell-derived smooth muscle cells (hMSC-SMCs) and endothelial cells (ECs). We hypothesized that incorporation of ECs would increase hMSC-SMC differentiation without compromising tissue ring strength or fusion to form tissue tubes. Undifferentiated hMSCs and ECs were co-seeded into custom ring-shaped agarose wells using four different concentrations of ECs: 0%, 10%, 20%, and 30%. Co-seeded EC and hMSC rings were cultured in SMC differentiation medium for a total of 22 days. Tissue rings were then harvested for histology, Western blotting, wire myography, and uniaxial tensile testing to examine their structural and functional properties. Differentiated hMSC tissue rings comprising 20% and 30% ECs exhibited significantly greater SMC contractile protein expression, endothelin-1 (ET-1)-meditated contraction, and force at failure compared with the 0% EC rings. On average, the 0%, 10%, 20%, and 30% EC rings exhibited a contractile force of 0.745 ± 0.117, 0.830 ± 0.358, 1.31 ± 0.353, and 1.67 ± 0.351 mN (mean ± standard deviation [SD]) in response to ET-1, respectively. Additionally, the mean maximum force at failure for the 0%, 10%, 20%, and 30% EC rings was 88.5 ± 36. , 121 ± 59.1, 147 ± 43.1, and 206 ±  0.8 mN (mean ± SD), respectively. Based on these results, 30% EC rings were fused together to form tissue-engineered blood vessels (TEBVs) and compared with 0% EC TEBV controls. The addition of 30% ECs in TEBVs did not affect ring fusion but did result in significantly greater SMC protein expression (calponin and smoothelin). In summary, co-seeding hMSCs with ECs to form tissue rings resulted in greater contraction, strength, and hMSC-SMC differentiation compared with hMSCs alone and indicates a method to create a functional 3D human vascular cell coculture model.

Abstract

The goal of this work is to create an in vitro vascular model that exhibits structural and functional properties similar to those of native vascular tissue. For the first time, we demonstrated that human mesenchymal stem cells cocultured with endothelial cells as 3D cell aggregates, differentiated into smooth muscle cells, exhibited contractile protein expression, and contracted in response to endothelin-1. These tissue rings could be fused together to form cohesive tubular constructs to mimic the geometry of native vasculature. Overall, this study demonstrated a novel method to create and assess 3D human vascular tissue constructs using quantitative metrics.

Introduction

Cardiovascular research relies on the use of 2D cell culture and animal models to develop and test new therapies.¹ While 2D cultures are a necessary first step, they do not accurately reflect the complex 3D environment found in vivo.² Furthermore, in vivo animal models often fail to predict clinical success due to species differences.^3,4 Thus, a need exists for 3D human vascular models to not only serve as screening tools but also quantitatively assess the effects of different cell types and agonists on cell and tissue responses.^5–7 Using functional human tissue models in vitro may allow researchers to accelerate the development of new treatments, eliminate ineffective therapies earlier in the testing process, and reduce resource expenditure and animal use.

Vascular tissue engineering represents one method to create 3D in vitro vascular models.^7–14 Common approaches for the creation of tissue-engineered blood vessels (TEBVs) are (1) a scaffold-based approach that consists of cell-seeded or acellular conduits fabricated from polymer or extracellular matrix (ECM)-derived materials^15–23 and (2) a scaffold-free approach that consists only of cells to form tubular structures.^24–26 While scaffold-based TEBVs have shown promising results as vascular grafts in vivo,^{18,22,23,27–31} they do not mimic native human vascular structures, since their geometry and mechanical integrity relies on the scaffold material. In contrast, scaffold-free TEBVs exhibit greater cell density, enhanced ECM production, and improved biological function,^17,26 all of which are important to create a vascular model.^24,32 Ideally, this model would consist of human cells that form functional smooth muscle cell (SMC) and endothelial cell (EC) layers in a scaffold-free, small diameter (≤2 mm) tubular structure, which responds to agonists.

Scaffold-free tissue-engineered constructs in a variety of shapes (e.g., spheres, rods, rings, and tubes) have been reported, using different cell sources (e.g., fibroblasts, primary smooth muscle cells, hepatocytes, and human bone marrow-derived mesenchymal stem cells [hMSCs]).^{24,26,33–44} Rolled cell sheets have been used to create scaffold-free TEBVs that exhibit burst pressures of ∼200 mmHg and respond to vascular agonists.^24,25,43 However, they require automated equipment, specialized skills, and long culture times (i.e., up to 25 weeks) to fabricate.^24,25,43 To overcome these challenges, a cellular self-assembly approach was developed to create scaffold-free tissue constructs using cell aggregates in fewer than 4 weeks.^26,33–42 For example, Kelm et al. fused spherical microtissues together to form a cell-only tubular structure using human myofibroblasts and ECs.²⁶ However, expression of contractile SMC proteins was limited, and tissue contractility was not tested.²⁶

To create a functional cell-only TEBV requires a proliferative cell type that can differentiate into contractile SMCs. hMSCs meet these criteria since they exhibit increased proliferative capacity, are more readily available than primary somatic SMCs,²⁰ and differentiate into a robust SMC lineage upon exposure to transforming growth factor beta-1 (TGF-β1).^45–49 Our lab previously demonstrated that hMSCs can self-assemble to form vascular tissue constructs and differentiate into hMSC-derived SMCs (hMSC-SMCs). However, contractile protein expression in our previous studies was limited to early-stage SMC markers.³⁷ Thus, the goal of the present work was to create scaffold-free tissues from hMSC-SMCs that exhibit robust SMC contractile protein expression and contract in response to a known vasoconstrictor, endothelin-1 (ET-1).

To achieve this goal, hMSCs were co-seeded with varying percentages of ECs (0%, 10%, 20%, or 30%) in ring-shaped nonadhesive agarose wells and cultured in SMC differentiation medium. Once aggregated, the cells formed tissue rings that were easily harvested for either endpoint analysis or tissue fusion to create TEBVs. ECs were incorporated for two reasons: (1) to mimic the physiological effects of ECs in native human blood vessels and (2) coculturing ECs with hMSCs in 2D has been shown to increase hMSC-SMC differentiation.^50–52 Previous studies using cocultured ECs with hMSCs were limited by their use of a 2D environment, use of only one ratio of hMSCs:ECs, and a lack of hMSC-SMC contractility testing.

To address these limitations, we created 3D hMSC-EC cocultures and tested the effect of increasing EC percentages on hMSC-SMC differentiation and contractile function. We hypothesized that ECs would significantly increase hMSC-SMC differentiation and ring contractility. Finally, tissue rings were fused to form TEBVs, and the effects of ECs on TEBV morphology and hMSC-SMC differentiation were evaluated. We hypothesized that the incorporation of ECs into TEBVs would not significantly affect TEBV morphology but would enhance hMSC-SMC differentiation in tubular constructs. The successful completion of this work would allow quantitative assessment of the effects of different cell types and agonists on scaffold-free 3D vascular tissue structure and function.

Methods

Cell culture

Bone marrow-derived hMSCs were purchased from RoosterBio^®, Inc. and cultured in growth medium (RoosterNourish^TM-MSC, KT-001, RoosterBio, Inc., Frederick, MD) in CellBIND^® T-175 flasks (Corning^®, Corning, NY). Primary human umbilical vein endothelial cells (HUVECs) were purchased from American Type Culture Collection (ATCC^®, Manassas, VA) and expanded in a vascular endothelial growth factor–supplemented growth medium (ATCC^®) in CellBIND^® T-175 flasks (Corning^®). HUVECs were chosen since they are a readily available and commonly used cell source to study the interactions of hMSCs and ECs for vascular tissue engineering applications.^26,50,51,53 Cells were expanded until passage 3–5 prior to creating tissue rings.

Engineered tissue fabrication

Tissue rings were fabricated using a similar method as previously described.^33,34,37,42 A 2% (w/v) mixture of agarose powder (SeaKem^® LE, Lonza, Switzerland) in Dulbecco’s modified Eagle’s medium (DMEM, Corning) was autoclaved, cooled, and pipetted into custom single-mold polydimethylsiloxane negatives.⁴² Once solidified, the agarose wells were transferred into 48-well plates and equilibrated with the RoosterNourish-MSC growth media overnight at 37°C.

At 90% confluency, hMSCs and HUVECs were trypsinized and cells were suspended at 12 × 10⁶ cells/mL in the following hMSC-to-EC ratios: 10:0 (0% EC rings), 9:1 (10% EC rings), 8:2 (20% EC rings), and 7:3 (30% EC rings). Fifty microliters of cell suspension was seeded into each agarose well, resulting in 0.6 × 10⁶ cells/ring (seeding-day 0). After 4 h, 800 µL of RoosterNourish-MSC was added to each well. On day 4 of culture, rings were switched to differentiation medium consisting of DMEM (Corning) supplemented with 5% fetal bovine serum (Gibco, Grand Island, NY), 1% penicillin–streptomycin (Corning), 1% l-glutamine (Gibco), 50 μg/mL ascorbic acid (Sigma-Aldrich, St. Louis, MO), 1% insulin–transferrin–selenium (Gibco), and 5 ng/mL TGF-β1 (PeproTech^®, Cranbury, NJ). Media was changed every 2–3 days for a total of 22 days. Rings were then harvested for histology, Western blotting, wire myography testing, or mechanical testing.

To fabricate TEBVs, tissue rings (0% or 30% ECs) were harvested on day 4, stacked onto a silicone tube mandrel (Liveo^TM, 508–006, 1.47 mm inner diameter, 1.96 mm outer diameter), and cultured for an additional 18 days in differentiation medium. Electrospun polycaprolactone cuffs were fabricated as previously reported³⁵ and placed on either end to allow cell ingrowth. The media was changed every 3 days. Figure 1 shows a schematic of the timeline, fabrication process, and analysis of the engineered tissues.

FIG. 1.

Schematic representation of the methods for testing the effects of ECs on hMSC-derived engineered tissues. DM, differentiation media; EC, endothelial cell; GM, growth media; hMSC, human mesenchymal stem cell; TEBV, tissue-engineered blood vessel. Created with BioRender.com.

Tissue dimension measurements

Brightfield images of the tissue rings were acquired every 3–4 days during culture using a 2× objective on a DMLB2 Leica microscope equipped with a DFC-480 camera (Leica, Weltzer, Germany). Ring thickness was calculated using ImageJ (https://imagej.net/Fiji), and the percent reduction in inner post diameter was calculated by subtracting the final inner area on day 22 from the initial inner area on day 1, divided by the initial inner area. To measure TEBV morphology, brightfield images of the fusing rings were acquired every other day for the first 6 days of fusion culture. ImageJ was used to measure the fusion angles between rings over time. As the rings fuse, the angle between the rings (i.e., fusion angle) should increase, with a final value of 180° indicating complete fusion.^37,41

Histological analysis

Engineered tissues were fixed in 10% neutral buffered formalin for 1 h at room temperature, processed for paraffin embedding, and sectioned at 5 μm. Tissue sections were stained using hematoxylin and eosin and/or Picrosirius Red/Fast Green (PR/FG) to visualize tissue structure and collagen deposition, respectively. Polarized light microscopy was used to visualize collagen fiber organization in PR/FG-stained tissue sections. A 20× objective on a Nikon^® Eclipse-LV100N POL microscope equipped with an AmScope-MU1803 camera was used to acquire images. The percent of the tissue stained positive for collagen fibers (percent collagen deposition) was calculated using ImageJ.

Immunostaining was used to visualize SMC contractile proteins and ECs. Antigen retrieval was performed on deparaffinized sections in 10 mM Tris, 1 mM ethylenediamine tetraacetic acid (EDTA), and 0.05% Tween-20 (pH 9.0) in a pressure cooker for 5 min. Tissue sections were blocked in 5% bovine serum albumin for 1 h. Sections were incubated overnight at 4°C in primary antibodies against SMC contractile proteins and CD31 (Table 1). Tissue sections were incubated in secondary antibodies (AlexaFluor^TM 568 rabbit anti-mouse and AlexaFluor^TM 488 donkey anti-rabbit, Invitrogen, Waltham, MA) for 1 h at room temperature. The CD31 primary antibody was directly conjugated to AlexaFluor 488. Stained sections were mounted using ProLong^TM Gold Antifade with 4′,6-diamidino-2-phenylindole (DAPI, Invitrogen). Either a 20× objective or a 60× immersion objective on an inverted fluorescent microscope (Keyence BZ-X800, Itasca, IL) was used to acquire images. To confirm that the percent of ECs seeded in each ring group persisted, the percent of CD31-positive cells was calculated using ImageJ. The cell density was also calculated using DAPI. To quantify hMSC-SMC differentiation in the 0% and 30% EC TEBVs, the percent area that was positive for smooth muscle α-actin (SMA), transgelin, calponin, and smoothelin was calculated using ImageJ and denoted “percent positive area.”

Table 1.

Summary of Antibodies Used to Perform Immunostaining

Antigen	Vendor	Catalog number	Concentration (mg/mL)	Host species	Dilution
SMA	Agilent Dako	M085101-2	71	Mouse	1/100
TGLN	Abclonal	A6760	3.2	Rabbit	1/100
CALP	Agilent Dako	M355601-2	46.5	Mouse	1/100
SMTN	Santa Cruz	sc-28562	0.20	Rabbit	1/100
CD31a	Abcam	ab275989	0.50	Rabbit	1/100
EDNRA	Abcam	ab178454	1.05	Rabbit	1/100

^a Indicates Alexa Fluor 488 conjugated primary antibody.

CALP, calponin; CD31, platelet endothelial cell adhesion molecule (PECAM); EDNRA, endothelin-1 receptor A; SMA, smooth muscle α-actin; SMTN, smoothelin; TGLN, transgelin (SM22-alpha).

Western blotting

Two tissue rings per group were pooled, flash frozen in liquid nitrogen, and homogenized in radioimmunoprecipitation assay buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.5% w/v sodium deoxycholate, 0.5% Nonidet P-40, 2 mM EDTA, and protease inhibitors [Invitrogen]). Cell lysates were centrifuged for 5 min at 10,000 rpm, and the supernatant was collected. Protein concentrations were determined using the Pierce^TM BCA Protein Assay Kit (Thermo Scientific, Waltham, MA). Four micrograms of each protein sample were subjected to 4–20% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes. Membranes were incubated for 1 h at room temperature in a 5% nonfat dry milk blocking solution and then with primary antibodies against SMC contractile proteins (Table 2) in the blocking solution overnight at 4°C. Secondary antibodies conjugated to horseradish peroxidase (Invitrogen) were incubated for 1 h at room temperature. Membranes were developed using SuperSignal^TM West Dura Extended Duration Substrate (Thermo Scientific) for 5 min and imaged using an Azure 600 chemiluminescent scanner (Azure Biosystems, Dublin, CA). Relative protein expression was calculated by normalization to α-tubulin 1b, and densitometry was performed using ImageJ.

Table 2.

Summary of Antibodies Used to Perform Western Blotting

Antigen	Vendor	Catalog number	Concentration (mg/mL)	Host species	Dilution
SMA	Abclonal	A17910	0.2	Mouse	1/500
TGLN	Abclonal	A6760	3.2	Rabbit	1/500
CALP	Abclonal	A3734	0.4	Rabbit	1/2000
SMTN	Abclonal	A6745	0.94	Rabbit	1/500
TUBA1b	Proteintech	11224–1-AP	0.6	Rabbit	1/2500

TUBA1b, α-tubulin 1b.

Wire myography

To evaluate tissue ring function, ET-1-mediated contraction was measured using wire myography. ET-1 was chosen because it is known as a potent, endothelium-derived vasoconstrictor in vivo^54–56 and a common agonist used to test contractility of vascular tissue-engineered constructs in vitro.^43,57 Tissue rings were mounted onto the 620M Multi Myograph System (Danish Myography Technologies, Denmark) in physiological saline solution (PSS) (0.13 M sodium chloride [NaCl], 4.7 mM potassium chloride [KCl], 1.18 mM potassium phosphate [KH₂PO₄], 1.17 mM magnesium sulfate heptahydrate [MgSO₄ 7H₂O], 5.5 mM glucose, 0.026 mM EDTA, and 1.6 mM calcium chloride [CaCl₂]) at 37°C. The myograph chambers were infused with balanced oxygen and 5% CO₂ gas. Tissue rings were pretensioned to a 1.5 mN tare load and treated with 100 mM KCl to induce ring contraction via voltage-gated ion channel depolarization. After 8 min, rings were washed four times with PSS and dosed with ET-1 (Tocris, Bristol, UK) at increasing doses of 0.01, 0.1, and 1 µM. Contraction was measured for 8 min, after which rings were flash frozen for Western blot analysis. The change in force (“ΔForce”) for each treatment was calculated by subtracting the baseline force value immediately before treatment from the peak force post treatment for each ring sample. An ET-1 dose-response curve was also generated to compare the peak force after each ET-1 dose.

Uniaxial tensile testing

Tissue ring uniaxial tensile testing was performed as previously described.^33,58 Images of rings within the agarose wells were acquired to measure ring thickness. Assuming a toroidal shape, the cross-sectional area was calculated. Tissue rings were mounted onto wire grips submerged in PBS, pretensioned to 10 mN, subjected to eight loading precycles, and pulled to failure at 10 mm/min using an ElectroPuls E1000 (Instron^®, Norwood, MA) with a 1 N load cell. The maximum force before failure, maximum tangent modulus (MTM), strain at maximum stress, and ultimate tensile strength (UTS) were reported. The mean UTS values were used to calculate theoretical burst pressure values using the Laplace equation: $2\times \left(\frac{\mathrm{UTS}\hspace{0.17em}\times \hspace{0.17em}\mathrm{wall}\hspace{0.17em}\mathrm{thickness}}{\mathrm{inner}\hspace{0.17em}\mathrm{diameter}}\right)$, where the wall thickness and inner diameter were calculated from the brightfield images of the rings cultured within the agarose wells after 22 days, assuming a toroidal shape.

Statistical analysis

Statistical analysis was performed using Prism 10 software (version 10.2, GraphPad). A two-way analysis of variance (ANOVA) was used to determine statistical significance (p-value <0.05) for the ring thicknesses and TEBV fusion angles over time. One-way ANOVA was used to determine statistical significance (p-value <0.05) for all other experimental outcomes between hMSC-EC tissue ring groups. Tukey’s post hoc tests were used to compare each experimental group to the control (0% ECs). t-Tests were used to determine statistical significance (p-value <0.05) for the TEBV wall thickness (0% ECs vs. 30% ECs) and the relative area positive for contractile SMC markers. Data are represented as either mean ± standard deviation or mean ± standard error of the mean as indicated in each figure caption.

Results

Effects of ECs on ring morphology

To evaluate the effects of ECs on tissue ring dimensions, brightfield images of the tissue rings were captured every 3–4 days (Fig. 2A). Importantly, there was no significant difference in ring thickness 1 day after seeding (Fig. 2B), suggesting that each experimental group was seeded with a similar cell density. After 22 days in culture, all cocultured rings exhibited significantly greater ring thicknesses compared to the 0% EC control rings with values of 175 ± 11.6, 193 ± 14.3, 209 ± 15.7, and 203 ± 20.7 µm for 0%, 10%, 20%, and 30% EC rings, respectively (Fig. 2B). In contrast, only the 20% and 30% EC rings (14.6 ± 2.20% and 15.3 ± 4.57%, respectively) exhibited significantly greater reduction in inner agarose post area compared with that of the controls (9.67 ± 3.18%) (Fig. 2C). This suggested that tissue rings with higher percentages of ECs contracted around the agarose posts more strongly over time.

FIG. 2.

Effects of EC coculture on hMSC tissue ring thickness as a function of time. (A) Representative brightfield images of tissue rings over the 22 days in culture, scale = 1 mm. (B) Comparison of tissue ring thickness over time between the four groups, n = 18–20 rings per group, ^#p-value <0.05 for 0% vs. 10% EC rings, $†$p-value <0.05 for 0% vs. 20% EC rings, *p-value <0.05 for 0% vs. 30% EC rings; values are mean ± SD. (C) Percent reduction in inner post area, n = 18–20 rings per group, *p-value <0.05, **p-value <0.01; values are mean ± SD. SD, standard deviation.

Effects of ECs on hMSC-SMC differentiation in tissue rings

CD31 immunostaining demonstrated that ECs were distributed throughout the tissue wall (Fig. 3A). The percentage of CD31 positive cells in the 0%, 10%, 20%, and 30% EC ring groups was 0%, 10.9 ± 1.34%, 20.9 ± 1.62%, and 30.2 ± 2.08%, respectively (Fig. 3B), indicating that the expected number of ECs were seeded and retained in each group. No significant difference was observed with respect to cell density (Fig. 3C).

FIG. 3.

EC distribution and persistence within hMSC-SMC tissue rings after 22 days in culture. (A) Representative images of tissue ring sections stained with CD31 (green) and DAPI (blue) captured with a 60× lens, scale = 50 µm, center of ring is on the left in each image. (B) Comparison of percentage of CD31-positive cells between the groups, n = 8 rings per group, ****p-value <0.0001; values are mean ± SEM calculated from two regions of interest per ring. (C) Comparison of overall tissue ring cell density as a function of the % ECs cocultured with hMSCs, n = 8 rings per group; values are mean ± SEM calculated from two regions of interest per ring. DAPI, 4′,6-diamidino-2-phenylindole; EC, endothelial cell; SEM, standard error of mean; SMC, smooth muscle cell.

To evaluate the effects of ECs on hMSC-SMC differentiation, expression of SMC-specific contractile proteins was examined. Immunostaining demonstrated that all tissue rings expressed early- to mid-stage SMC differentiation markers, including the contractile proteins SMA, transgelin, calponin, and smoothelin (Fig. 4). Western blot analysis confirmed these results (Fig. 5A). Quantification of protein levels using densitometry and comparisons with the 0% EC rings revealed 1.40- and 1.48-fold increased expression of transgelin and 2.07- and 2.90-fold increased expression of calponin, in 20% and 30% EC rings, respectively (Fig. 5B). Smoothelin protein levels were increased in all EC-containing rings compared with controls (4.76-, 6.93-, and 16.3-fold increase for 10%, 20% and 30% EC rings, respectively) (Fig. 5C).

FIG. 4.

Effects of EC coculture on hMSC-SMC differentiation assessed using immunostaining. Representative images of tissue sections stained for SMC-specific proteins captured with a 20× lens, scale = 100 µm; images are oriented with central agarose posts to the left for all sections. CALP, calponin; SMA, smooth muscle α-actin; SMTN, smoothelin; TGLN, transgelin.

FIG. 5.

Effect of EC coculture on hMSC-SMC differentiation assessed using western blotting. (A) Representative Western blots to evaluate expression of contractile SMC proteins. (B) Quantification of western blots for SMA, TGLN, CALP, n = 6 pooled ring lysates (two rings pooled/lysate), *p-value <0.05, **p-value <0.01, ****p-value <0.0001; values are mean ± SD. (C) Quantification of western blot for SMTN, n = 6 pooled ring lysates (two rings pooled/lysate), *p-value <0.05, ***p-value <0.0001; values are mean ± SD. CALP, calponin; SMA, smooth muscle α-actin; SMTN, smoothelin; TGLN, transgelin.TUBA1b, α-tubulin 1b.

Effects of ECs on tissue ring function

Tissue ring function was assessed using wire myography with KCl and ET-1 treatment (Fig. 6A). Upon KCl treatment, the 20% and 30% EC rings (0.329 ± 0.0826 and 0.415 ± 0.231 mN, respectively) exhibited significantly greater ΔForce compared with controls (0.0984 ± 0.123 mN) (Fig. 6B). Similarly, both the 20% and 30% EC rings (1.31 ± 0.353 and 1.67 ± 0.351 mN, respectively) exhibited significantly greater ΔForce with ET-1 treatment compared with the rings without ECs (0.745 ± 0.117 mN) (Fig. 6C). Importantly, when the contractile force was normalized to ring thickness, these results persisted (data not shown). The ET-1 dose–response curves also demonstrated that 20% and 30% EC rings exhibited significantly greater contraction on both the second and third doses of ET-1 (100 and 1000 nM) compared with the controls (Fig. 6D). Endothelin-1 receptor A (EDNRA) immunostaining correlated with ET-1 contractile response (Fig. 6E).

FIG. 6.

Effects of EC coculture on tissue ring contractility measured by wire myography. (A) Representative image of a tissue ring loaded onto the wire myograph, white box indicates the tissue ring loaded onto the wires, scale = 2 mm. Comparison of the change in force (Δ Force) of tissue rings before and after stimulation with (B) potassium chloride (KCl) and (C) endothelin-1 (ET-1), n = 8 rings per group, *p-value <0.05, **p-value <0.01, ***p-value <0.001, ****p-value <0.0001; values are mean ± SD. (D) ET-1 dose-response curves for each experimental group, $†$p-value <0.05 for 0% vs. 20% EC rings, *p-value <0.05 for 0% vs. 30% EC rings; values are mean ± SD. (E) Immunostaining of tissue sections for endothelin-1 receptor A (ENDRA); images are oriented with central agarose posts to the left for all sections. EDNRA = green; DAPI = blue. Scale = 50 µm.

Effects of ECs on collagen deposition

All tissue rings exhibited circumferentially aligned nuclei (Fig. 7A). PR/FG staining revealed that collagen was deposited throughout the vascular wall, with mature collagen fibers circumferentially aligned near the outer portion of the vascular wall (Fig. 7A). The 20% and 30% EC rings (19.7 ± 2.30%, and 17.5 ± 1.53%, respectively) exhibited significant increases in percent collagen deposition compared to the rings without ECs (9.53 ± 0.916%) (Fig. 7B).

FIG. 7.

Histological analysis of tissue rings stained for H&E and PR/FG. (A) Representative tissue sections imaged under brightfield and polarized light microscopy captured with a 20× lens. Collagen appears as red stain with PR/FG and birefringence with polarized light, scale = 100 µm. Images are oriented with central agarose posts to the left for all sections. (B) Collagen content quantified as % tissue area exhibiting birefringence, n = 8 rings per group, **p-value <0.01, ***p-value <0.001; values are mean ± SEM calculated from two regions of interest per ring. H&E, hematoxylin and eosin; PR/FG, Picrosirius Red/Fast Green.

Effects of ECs on ring mechanical properties

To determine if the increased collagen expression in the 20% and 30% EC rings contributed to their mechanical strength, uniaxial tensile testing was performed (Fig. 8A). The 20% and 30% EC rings (147 ± 43.1 and 206 ±  0.8 mN, respectively) exhibited significantly increased maximum force before failure compared with the 0% EC rings (88.5 ± 36.  mN) (Fig. 8B). No significant differences were observed with respect to MTM (Fig. 8C). Nevertheless, the 30% EC rings (6.59 ± 2.01) exhibited significantly increased strain at the maximum stress compared to the control rings (3.22 ± 2.02) (Fig. 8D). The 30% EC rings (3.59 ± 0.455 MPa) also exhibited significantly increased UTS compared with the rings without ECs (1.98 ± 0.891 MPa) (Fig. 8E). Representative stress–strain curves underscore these results (Fig. 8F). Using the Laplace equation, theoretical burst pressures were calculated to be 257 kPa (1925 mmHg), 265 kPa (1987 mmHg), 374 kPa (2808 mmHg), and 530 kPa (3975 mmHg) for 0%, 10%, 20%, and 30% EC rings, respectively.

FIG. 8.

Effect of ECs on tissue ring mechanical properties. (A) Representative image of tissue ring mounted on tensile testing wires. (B) Comparison of maximum force experienced before failure, n = 6–8 rings per group, *p-value <0.05, ****p-value <0.0001. (C) Comparison of maximum tangent modulus (MTM), n = 6–8 rings per group. (D) Comparison of strain at maximum stress, n = 6–8 rings per group, ***p-value <0.001. (E) Comparison of ultimate tensile strength (UTS), n = 6–8 rings per group, *p-value <0.05. (F) Representative stress–strain curves from tissue ring uniaxial tensile tests.

Effects of ECs on TEBV morphology

Brightfield images of fusing 0% and 30% EC rings demonstrated that fusion angles in both groups increased over the first 4 days but then plateaued at ∼140° on day 6 (Fig. 9A). On day 2, 30% EC TEBVs exhibited significantly greater fusion angles than 0% EC TEBVs but then equilibrated after day 4 (Fig. 9B).

FIG. 9.

Effect of ECs on TEBV morphology. (A) Representative brightfield images of hMSC rings (with 0% or 30% ECs) on silicone tube mandrels during tissue fusion culture for 6 days captured with a 2× lens, scale = 1 mm. (B) Comparison of fusion angles between hMSC rings cocultured with 0% or 30% ECs, n = 3 TEBVs per group, *p-value <0.05 and ****p-value <0.0001 when compared across time points for both groups, ^#p-value <0.05 when comparing 0% EC vs. 30% EC TEBVs on day 2. Values are mean ± SEM calculated from three fusion angle measurements per TEBV.

Effects of ECs on collagen deposition and hMSC-SMC differentiation in TEBVs

PR/FG-stained TEBV sections suggest that both groups deposited collagen throughout the TEBV wall (Fig. 10A). No difference was observed with respect to TEBV wall thickness (Fig. 10B). Fluorescent images of the TEBVs stained for SMA (red), CD31 (green), and nuclei (blue) indicate that SMA expression was isolated to the outer regions of the TEBV wall, while CD31-positive ECs were predominantly located closer to the TEBV lumen (Fig. 10C). The percentage of CD31-positive cells was calculated as 32.5 ± 2.29%, confirming that the expected number of ECs were seeded and retained. Lastly, the 30% EC TEBVs exhibited a significantly greater relative area that was positive for calponin and smoothelin compared with the 0% EC TEBVs (Fig. 10D).

FIG. 10.

Histological comparison of hMSC-only and hMSC-EC TEBVs cultured for 22 days. (A) Brightfield and polarized light images of PR/FG-stained tissue sections from 0% and 30% EC TEBVs captured with a 20× lens, scale = 100 µm; images are oriented with the central silicone tubes on the bottom for all sections. (B) Comparison of TEBV wall thickness, n = 6 TEBVs per group, values are mean ± SEM calculated from two regions of interest per TEBV. (C) Comparison of relative positive area for SMA, TGLN, CALP, and SMTN, n = 6 TEBVs per group, *p-value <0.05, ***p-value <0.001; values are mean ± SEM calculated from two regions of interest per TEBV. (D) Immunostaining of 0% and 30% EC TEBVs for SMA (red) with CD31 (green), TGLN (green), calponin (red), and SMTN (green) captured with a 20× lens, scale = 100 µm; images are oriented with the central silicone tubes on the bottom for all sections.

Discussion

Scaffold-free tissue engineering presents a unique opportunity to build 3D vascular constructs using multiple cell types with high cell densities. While 2D models have demonstrated increased contractile protein expression when SMCs are cocultured with ECs, there is limited understanding of how ECs affect hMSC-SMC contractile protein expression in 3D. Using a cocultured hMSC-EC cellular self-assembly model, the effects of ECs on tissue ring morphology, hMSC-SMC differentiation, ring collagen content, ring mechanical properties, and ring contraction were evaluated.

Two tissue ring morphological metrics were significantly altered based on the percentage of seeded ECs. First, the cocultured ring thickness was significantly greater than that of the 0% EC rings. To investigate if this observation was due to increased cell density, cell nuclei were visualized and quantified using DAPI staining, but no differences were observed. Second, the percent reduction in the inner post area was significantly increased for the 20% and 30% EC rings compared to that of the control rings. This may be a result of SMC contractile protein expression and tissue ring contraction, thus constricting the inner agarose post and decreasing its area.

While all tissue rings expressed contractile proteins, 20% and 30% EC rings exhibited significantly greater transgelin, calponin, and smoothelin expression compared to the controls. These results are consistent with previous reports in which 2D hMSC-HUVEC cocultures (using a 1:1 ratio) demonstrated increased expression of SMA, transgelin, and calponin compared to hMSC monocultures after 4–7 days in culture.^50,51 These results may be due to the direct cell–cell interactions of hMSCs with ECs via the Notch signaling pathway; it is well established that Notch signaling mediates vascular SMC differentiation and phenotype.^{49,50,59–62} Importantly, Notch and TGF-β1 signaling have been shown to cooperatively induce a molecular and functional contractile phenotype in human aortic SMCs.⁴⁹ This suggests that with TGF-β1 and EC-mediated signaling, the 20% and 30% EC tissue rings may have been able to express significantly more transgelin, calponin, and smoothelin.

Functionally, both the 20% and 30% EC rings exhibited greater contraction in response to KCl and ET-1 compared to the 0% EC rings. The enhanced response to KCl may be due in part to increased expression of contractile proteins^63–65 and enhanced expression of gap junctions, which help propagate the depolarization signal from KCl. Previous work has shown that when hMSCs are cocultured with ECs, the expression of the gap junction protein connexin 43 (Cx43) is increased compared with hMSC monocultures.⁵¹ Cx43 mediates metabolic, electrical, and ionic signals between cells and has been implicated in the uptake of potassium.^66–68

Although tissue rings responded to both KCl and ET-1, their response to ET-1 was greater, unlike native vascular tissue that exhibits maximum contraction with KCl.^69,70 This may be because these engineered tissues do not express as many voltage-gated ion channels as native vasculature tissue. The 20% and 30% EC rings may have exhibited increased ET-1-mediated contraction due to increased expression of calponin, one of the primary functional proteins involved in the pathway,^54,71 and increased expression of the ET-1 receptor, EDNRA. While previous research has demonstrated that hMSC-EC cocultures on a scaffold exhibit greater nonspecific contraction,⁵⁰ to our knowledge, the EC dose-dependent response to ET-1-mediated contraction of scaffold-free tissue constructs represents a novel finding.

Another novel finding of this work is the increased collagen expression in 3D scaffold-free cocultured hMSC-SMCs and ECs. While some studies have demonstrated decreased collagen expression in SMC-EC cocultures in 3D scaffold-based constructs,^72–74 our findings are consistent with previous reports that suggest collagen expression in 2D cocultured SMCs and ECs was increased compared with SMCs alone.^50,52 Interestingly, the percent collagen content of the 20% and 30% EC rings, ranging from 17% to 20%, was similar to that of the human internal mammary artery (IMA), ∼17% of the native tissue.⁷⁵

The IMA is of particular interest here since it is one of the most commonly used blood vessels in coronary bypass surgery. Consistent with the increased collagen expression, the 20% and 30% EC rings exhibited significantly greater maximum failure force compared with the 0% EC rings. When compared with native human vascular tissue, the calculated burst pressures of all tissue rings were greater than that of the saphenous vein (∼1600 mmHg), while only the calculated burst pressure of the 30% EC rings was greater than that of IMA (∼3200 mmHg).⁷⁶ Although these calculated burst pressures are theoretical, they suggest that these scaffold-free tissue engineered constructs may withstand physiological forces.

The versatile tissue ring modules were also used as building blocks to create TEBVs. Rings with 30% ECs yielded the most contractile tissue with the highest UTS; therefore, 30% EC rings were used to create TEBVs and compared with 0% EC TEBVs. Both the 0% and 30% EC rings formed cohesive tubular structures, with no differences observed with respect to ring fusion or wall thickness. Importantly, the TEBVs exhibited wall thicknesses similar to those of the medial layer of native coronary arteries, which range from ∼200 to 350 µm.⁷⁷ The 30% EC TEBVs exhibited increased calponin and smoothelin expression compared with the 0% TEBVs, consistent with the tissue ring experiments.

In summary, this work characterized the effects of varying EC concentrations on engineered hMSC-SMC tissue ring structure and function. A platform for the creation of 3D functional human SMC-EC coculture constructs was established. These constructs can be used to assess human vascular tissue response to agonists and offer an alternative to 2D cell culture models. One limitation of this study was the lack of smooth muscle-myosin heavy chain (SM-MHC) expression in the engineered tissue since SM-MHC is one of the primary functional proteins in native vascular SMCs. Future work will focus on achieving SM-MHC expression via longer culture times and/or mechanical stretch.^78,79 The effects of mechanical stretch will be evaluated on hMSC-EC TEBV morphology, mechanics, and contractile protein expression. The goal of this future work is to create functional human derived TEBVs with luminal flow to mimic circulatory physiology. Additionally, the flexibility in this culture method may also allow for the coculture of other cell types, such as genetically modified SMCs or ECs that exhibit characteristics of vascular disease. In this way, a vascular disease model may be created and used for drug discovery.

Authors’ Contributions

W.G.D. and M.W.R. designed and conceptualized the experiments. W.G.D. executed all experimentation including engineered tissue fabrication, histology, Western blotting, wire myography, and mechanical testing. W.G.D. analyzed the data and wrote the article. A.E.F.-M. assisted with histology, image acquisition, wire myography, mechanical testing, and data analysis. A.K., A.E.H., B.T., L.R., and L.L. provided the materials and equipment necessary to run the wire myography experiments, established the wire myography protocol, supervised and performed the wire myography testing, and assisted with its data analysis. All authors aided in interpreting the results as well as editing and reviewing the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This research was partially supported by National Institutes of Health, The National Heart, Lung, and Blood Institute, R15 HL137197 (M.W.R.).