Recent efforts to elucidate the molecular regulation of vascular morphogenesis have focused on the generation of three-dimensional micro-tissues that allow interactions between endothelial cells (ECs) and vascular smooth muscle cells (SMCs) to be investigated. For example, Korff et al. (2001), using human umbilical vein ECs (HUVECs) and human umbilical artery smooth muscle cells (HUASMCs), have generated tissue spheroids in carboxymethylcellulose that were characterized by an outer layer of ECs which surrounded an inner core of SMCs. These spheroids have been used to study paracrine interactions between ECs and mural cells (e.g., SMCs) that mediate vascular endothelial cell growth factor (VEGF) signaling (Korff et al.,2001), and most recently, to investigate Tie-2/angiopoietin-2 signaling (Scharpfenecker et al.,2005). Using the same cell types, Kelm et al. (2005) generated spheroids in hanging drop culture characterized by an outer layer of SMCs surrounding an inner mass of ECs that lacked lumens. Using these spheroids, Kelm and colleagues investigated VEGF production and its effects on blood vessel formation.
Here, we sought to generate a spheroid that more closely recapitulated the features of a blood vessel. Specifically, we were interested in generating a spheroid with an outer layer of SMCs surrounding a single inner endothelium with a central lumen. Using exogenous VEGF as a means of promoting the fusion of blood vessels (Drake and Little,1995; Drake and Fleming,2000; Argraves et al.,2002; Dor et al.,2002; Dominguez et al.,2007), we report the generation of mouse allantoic-derived vascular spheroids that are characterized by an outer layer of SMCs (i.e., smooth muscle α-actin [SMαA], SM22-α, smooth muscle myosin II heavy chain [SM-MHC] expressing cells) surrounding an inner endothelium (i.e., PECAM-1, CD34, VE-cadherin expressing cells) defining a central lumen-like cavity. Furthermore, we show that these vascular spheroids exhibit physiologically relevant vasodilatory and contractile responses.
Generation of Vascular Spheroids From Mouse Allantois Tissue
When allantoides isolated from embryonic day (E) 8.5 mouse embryos (Fig. 1A) were cultured in a hanging drops for 18 hr, they took on a spherical shape (Fig. 1B), with a mean outer diameter of 440.5 μm ± 39.08 (n = 17). Analysis of laser scanning confocal microscopy (LSCM) z-series projections showed that the spheroids were comprised of an outer layer of cells expressing SMαA (Fig. 1C, red), SM22-α (data not shown), and smooth muscle myosin (data not shown) that surrounded an inner three-dimensional capillary-like network of PECAM-1–positive (Fig. 1C, green), CD34-positive (data not shown) cells. Examination of three-micron physical sections of the spheroids (Fig. 1D) revealed that the PECAM-1–positive cells (green) surrounded immunonegative areas that may represent either vascular lumens or extravascular spaces that separate PECAM-1–positive cells.
Exogenous VEGF Promotes the Fusion of Spheroid Blood Vessels
Given the established ability of VEGF 165 (VEGF) to promote the fusion of blood vessels to form vessels of increased diameter (Drake and Little,1995; Argraves et al.,2002; Dor et al.,2002; Dominguez et al.,2007), the effect of exogenously added VEGF on the blood vessels of allantoic spheroids was investigated. Analysis of a LSCM z-series acquired from a VEGF-treated spheroid that was immunolabeled with antibodies to PECAM-1 and SMαA showed that VEGF profoundly affected the EC component of the spheroids. Rather than a network of luminized PECAM-1–positive vessels occupying the center of the spheroid, as was seen in the untreated spheroids (Fig. 1D) analysis of optical section(s) acquired from VEGF-treated spheroids showed that treated spheroids were uniluminal; having a single inner layer of PECAM-1 (Fig. 2A, green) expressing ECs that circumscribed a single large central immunonegative area. Conventional fluorescence microscopic analysis of 3 μm sections of paraffin embedded VEGF-treated spheroids immunolabeled with antibodies to PECAM-1 (Fig. 2B, green) and SMαA (Fig. 2B, red) confirmed the overall morphological effects of VEGF, and showed that the endothelium, while continuous, was not uniform. In some areas, the endothelial cells appeared as a simple squamous layer (Fig. 2B, asterisk), whereas in other areas the endothelium was stratified, with PECAM-1–positive projections evident (Fig. 2B, arrowheads).
The Central Space of VEGF-Treated Allantoic Spheroids Is Analogous to a Vascular Lumen
Because the lumen of a normal blood vessel is circumscribed by an endothelium, we next investigated whether the PECAM-1–expressing ECs in VEGF-treated spheroids expressed the EC specific cell–cell junction protein, VE-cadherin. As seen in Figure 3A, LSCM analysis of VEGF-treated spheroids immunolabeled using antibodies to VE-cadherin (green) and SMαA (red) revealed that PECAM-1–positive ECs that surrounded the central space also expressed VE-cadherin. While the above observations suggested that the central space is lined by a vascular endothelium, it remained unclear whether the endothelium was contiguous, and thus able to act as a barrier. To evaluate this, 15- to 20-nm quantum dot nanocrystals (Qdots) were injected into the central cavity of the spheroid. In vivo studies have shown that Qdots injected into the circulatory system of zebrafish are retained within blood vessels (Rieger et al.,2005). As shown in Figure 3C, Qdots injected into spheroids (Fig. 3B) filled the entire central space. Furthermore, observations of injected spheroids showed that the Qdots did not leak out of the central space into the surrounding SMαA-positive cell layer (data not shown). These findings, in conjunction with the finding that the ECs lining the cavity express VE-cadherin, suggest that the cavity is lined by a contiguous endothelium and is, therefore, analogous to the lumen of a blood vessel.
Cells Forming the Outer Layer of the Spheroid Express SMαA, SM22-α, and SM-MHC
Additional immunohistological analysis was performed to determine whether cells of the outer layer of the VEGF-treated spheroids expressed SMC proteins in addition to SMαA (Fig. 4A). As shown in Figure 4C,D, the outer layer of cells expressed both SM22-α and SM-MHC, two well-established SMC proteins (Duband et al.,1993; Miano et al.,1994; Li et al.,1996; Madsen et al.,1998). Furthermore, high magnification LSCM analysis of the SMαA labeled cells showed that the actin proteins were organized in filamentous networks (Fig. 4B), suggestive of a functional contractile apparatus.
Cells of the Outer SMC Layer Are Closely Associated With Cells of the Inner Endothelial Layer
Initial immunohistological analysis of VEGF-treated spheroids suggested that SMαA-positive and PECAM-1–positive cell layers were separated by other nonimmunoreactive cells (Fig. 2). However, high magnification LSCM analysis of VEGF-treated spheroids immunolabeled with SM22-α and PECAM-1 showed that SM22-α–positive cells (SMCs) extend from the outer surface of the spheroid to the inner endothelial layer (Fig. 5). The cytoplasmic immunostaining achieved with SM22-α antibody highlighted the close interaction between the SM22-α–positive cells and ECs (Fig. 5C, arrows).
VEGF-Treated Spheroids Respond to Agents That Regulate Vascular Tone
To evaluate whether the VEGF-treated spheroids exhibited physiologically relevant vasodilatory and contractile responses, the spheroids were treated with well-described regulators of vascular tone. First, spheroids were treated with KCl, a mediator of SMC contraction (Dora et al.,1997; Hadoke et al.,2001; Ross et al.,2006). This treatment resulted in an overall reduction in spheroid diameter. Within 15 min of KCl exposure, the spheroids reached 50% of the maximal contraction, which was achieved after 1 hr of treatment (Fig. 6A). Next, we evaluated the capacity of the EC component of spheroids to modulate the activity of the SMC component. To do this, spheroids that had been precontracted with KCl were treated with acetylcholine (ACh), a vasorelaxant. As seen in Figure 6B, the diameter of precontracted spheroids increased in response to ACh. To evaluate the extent to which the ACh-mediated relaxation of the spheroids was attributable to nitric oxide (NO), an EC-generated mediator of vascular SMC relaxation (Furchgott and Zawadzki,1980), precontracted spheroids were exposed to an inhibitor of endothelial cell NO production, N-nitro-L-arginine methyl ester (L-NAME). As seen in Figure 6C, L-NAME inhibited the ACh-mediated relaxation of the spheroids.
Another important physiological mediator of vascular tone is S1P, a lysophospholipid carried in the blood by high-density lipoproteins (HDL; Murata et al.,2000). Using aortic ring-based assays to evaluate vasorelaxation, Nofer et al. (2004) have demonstrated that HDL promotes vasorelaxation by means of signaling through the S1P receptor, S1P3. To determine whether spheroids exhibit a similar vasodilatory response, we evaluated the response of precontracted spheroids from wild-type and S1P3-deficient mice to HDL. As shown in Figure 7, KCl-contracted spheroids from both wild-type and S1P3 nulls relaxed in response to HDL treatment. However, the magnitude of the relaxation observed in S1P3-deficient spheroids was less (∼50%) than that of wild-type spheroids. The observed HDL effects were similar to those observed using aortic rings in which S1P3-deficient aortic rings relaxed 57% that of wild-type in response to HDL (Nofer et al.,2004).
Here, we describe the development of vascular micro-tissues (vascular spheroids) that are characterized by an outer layer of SMαA-, SM22-α-, and SM MHC-positive cells and an inner cavity lined by PECAM-1-, CD34-, and VE-cadherin–positive cells. ECs lining the cavity exhibit properties of a vascular endothelium as evidenced by VE-cadherin expression and the fact that injected Q-dots are retained within the cavity. The fact that Qdots, which have a diameter similar to that of low-density lipoprotein particles (Rohrer et al.,2006; Teerlink and Scheffer,2007), do not leak from the lumen-like cavity into the subendothelial space may be an indication that the endothelium of the vascular spheroids displays physiological macromolecular barrier function.
The finding that spheroids derived from E8.5 allantoides generate SMCs is of interest considering that SMCs are not present in the allantois in vivo at E8.5. It is possible that the SMCs arose through differentiation of allantoic mesenchymal cells (Downs et al.,2001). It is also possible that the SMCs arose either directly from Flk1-positive cells (Yamashita et al.,2000) or by means of a transdifferentiation of endothelial cells (Frid et al.,2002; Ishisaki et al.,2003). Whatever the progenitor may be, it appears that our culture conditions potentiate the differentiation of SMCs in explanted allantoides.
Our studies also demonstrated that ECs and SMCs of vascular spheroids are capable of physiologically relevant cell–cell signaling. This conclusion is supported by evidence that ECs were able to mediate the relaxation of the smooth muscle layer by means of ACh-induced NO production. The fact that the ACh-mediated relaxation of vascular spheroidal SMCs was not completely inhibited by L-NAME suggested an eNOS-independent component. This interpretation is consistent with findings from studies of mice deficient in eNOS, which also revealed an eNOS-independent vasodilatory mechanism (Chataigneau et al.,1999). Similar to what we observed, Chataigneau et al. (1999) found that ACh-induced SMC relaxation of mesenteric arteries was only partially blocked by L-NAME. The eNOS-independent relaxation process in these arteries appeared to involve a product of cyclo-oxygenase (Chataigneau et al.,1999). These findings, together with ours, lead us to conclude that vascular spheroidal endothelium stimulates SMC relaxation in response to ACh by means of mechanisms similar to those used by the endothelium of “true” blood vessels.
Another indication that spheroids display physiological vasodilatory responsiveness is our finding that HDL, a vasodilatory agent which acts through eNOS (Nofer et al.,2004) caused relaxation of precontracted spheroids. The spheroid relaxation response to HDL was also dependent on S1P3 signaling by means of eNOS (data not shown). These observations agree with findings of Nofer et al. (2004) that HDL-induced relaxation in aortic rings was dependent on S1P3 activation of eNOS.
Critical to the generation of vascular spheroids was the use of VEGF. In the absence of exogenously added VEGF, spheroids formed with a core network of small diameter blood vessels. This is in contrast to the large diameter EC-lined cavity that formed when the spheroids were cultured in VEGF-supplemented medium. Based on previous studies in which elevated VEGF levels were shown to promote the formation of large diameter vessels and vascular sinuses by means of the fusion of small diameter vessels (Drake and Little,1995; Drake and Fleming,2000; Argraves et al.,2002; Dor et al.,2002; Dominguez et al.,2007), we speculate that exogenously added VEGF acted to promote vascular fusion in the spheroids. Noteworthy in this regard is the fact that the level of serum-derived VEGF in spheroid culture medium supplemented with fetal bovine serum (10% final concentration) was insufficient to promote vascular fusion.
There are several possible mechanisms by which exogenously added VEGF could induce vascular fusion (Kowanetz and Ferrara,2006). One possibility is that exogenously added VEGF augments signaling through the low affinity VEGF receptor, Flk-1. Mechanistically, the availability of VEGF for binding to Flk-1 (KD = 760 pmol/L; Waltenberger et al.,1994) is limited by high affinity VEGF binding to the high affinity VEGF receptor, Flt1 (KD = 16 pmol/L; Hornig et al.,2000; Shibuya,2001). In the spheroid culture, elevated VEGF levels may overcome Flt-1–mediated suppression of VEGF signaling and thereby mediate vascular fusion. This is consistent with the vascular fusion-like phenotype observed in mice lacking Flt1 (Fong et al.,1995).
The vascular spheroids described herein provide a novel tool for evaluating the role of EC–SMC interactions in the morphogenesis/maturation of embryonic blood vessels. We also demonstrate the utility of the spheroids to evaluate the significance of genetic mutation (i.e., S1P3 knockout) on vascular tone. Vascular spheroids derived from embryos bearing targeted gene mutations may prove useful for analyzing the function(s) of genes implicated in blood vessel maturation and/or vascular tone, and particularly for those in which the mouse mutants die early in development and, thus, preclude the use of the aortic ring assay.
We also recognize the potential utility of vascular spheroids to act as modules to engineer blood vessels. This possibility is strengthened by evidence that other types of tissue spheroids can fuse. For example, spheroids generated from proepicardial tissue, which is a source of EC progenitor cells that contribute to coronary vessel formation, can fuse to form linear structures (Perez-Pomares et al.,2006). We also believe that the use of exogenous VEGF to transform networks of vessels into large vessels, as we have shown here, will be translatable to the process of engineering blood vessel-like spheroids from either adult human SMCs and ECs or from SMCs and ECs derived from stem cells.
Rat antibodies to mouse PECAM-1 and VE-cadherin were obtained from BD PharMingen (San Diego, CA). Rat antibodies to mouse CD34 were purchased from Research Diagnostics Inc. (Flanders, NJ). Cy3 conjugated anti-smooth muscle alpha actin (SMαA, clone 1A4) was purchased from Sigma-Aldrich (St. Louis, MO). Rabbit anti-mouse SM22-α was purchased from Abcam (Cambridge, MA) and Rabbit anti-human smooth muscle myosin II heavy chain was purchased from Biomedical Technologies, Inc. (Stoughton, MA). Fluorescently conjugated secondary antibodies were purchased from Jackson ImmunoResearch Labs, Inc. (West Grove, PA). Recombinant human VEGF165 was obtained from R&D Systems (Minneapolis, MN). Potassium chloride (KCl), ACh, L-NAME, and Hoechst 33342 nuclear stain were purchased from Sigma-Aldrich. Quantum dot 655 nanocrystals were obtained from Invitrogen (Carlsbad, CA) as part of a Qtracker 655 Kit (catalog no. Q25021MP). HDL (1.063< d<1.21 g/ml) were isolated from the blood of normolipemic volunteers not receiving prescription medication for any acute or chronic condition and without family history of coronary artery disease, peripheral vascular disease, or stroke.
Embryonic tissues used for this study were obtained from approximately 80 wild-type ICR timed-pregnant female mice purchased from Harlan (Indianapolis, IN). Mice with a targeted deletion of the S1P3 gene (Kono et al.,2004) were obtained from Dr. Richard L. Proia (National Institutes of Health, Bethesda, MD). Ten timed-pregnant S1P3 null mice were used in this study.
Generation of Spheroids From Allantoic Tissue
To generate spheroids, E8.5 mouse allantoic tissue was first isolated as previously described (Drake and Fleming,2000; Argraves et al.,2002). Individual allantoides were then suspended in 30 μl of complete culture medium (Dulbecco's Modified Eagles Medium [DMEM], Invitrogen) supplemented with 10% fetal bovine serum (Atlanta Biologicals, Lawrenceville, GA) and 1% penicillin, streptomycin/L-glutamine (Invitrogen), and spotted onto the underside of a lid of a 60-mm plastic bacterial culture dish (Fisher Scientific, Sewanee, GA). The lids were then inverted and placed onto culture dishes to create hanging drops. The hanging drop cultures were placed in a humidified container and cultured for 18 hr (37°C, 5% CO2). After 18 hr of culture, the resulting spheroidal-shaped tissues were resuspended in 30 μl of complete culture medium containing 0.3–1 μg/ml VEGF and spotted onto the underside of a 60-mm Petri dish lid. Again, the lid was inverted, placed onto the culture dish, and the spheroids were incubated for an additional 18 hr (37°C, 5% CO2). Spheroidal diameter measurements were made on nonfixed spheroids suspended in medium.
Fixation and Immunolabeling of Spheroids
Spheroids were fixed in 4% paraformaldehyde (Sigma-Aldrich; 30 min, 22°C) and then rinsed for 5 min in phosphate buffered saline containing 0.01% sodium azide (PBSA). Fixed spheroids were permeabilized in 0.02% Triton X-100/PBSA (25 min, 22°C), washed in PBSA, and then incubated in blocking solution (BS) containing 5% donkey serum (Jackson ImmunoResearch) and 3% bovine serum albumin (BSA, Sigma-Aldrich) in PBSA (30 min, 22°C). Spheroids were then incubated in primary antibodies (10 μg/ml) in BS (2 hr, 22°C). Following three washes (10 min each, 22°C) in PBSA, spheroids were incubated in secondary antibodies (10μg/ml) in BS (2 hr, 22°C). Spheroids were then washed in PBSA (3 × 10 min, 22°C) and mounted on glass slides under number 1 cover slips (Fisher Scientific) using Gelmount mounting medium (Biomedia Corp., Foster City, CA).
Microscopic Analysis of Spheroids
Conventional fluorescence or differential interference contrast (DIC) images of allantoic spheroids were obtained at 22°C using a Leica DMR research grade microscope equipped with Leica objectives (×5, ×10, ×20/0.7, ×40/0.85) and a SPOT-RT camera (Vashaw Scientific, Raleigh, NC). Images were acquired using SPOT-RT 3.5.7 software. Laser scanning confocal microscopic images were acquired using Zeiss ×5/0.15 and ×10/0.30 objectives on a Bio-Rad MRC 1024 laser-scanning confocal microscope (Bio-Rad, Microscopy Division, Cambridge, MA) equipped with Lasersharp 2000 software (Bio-Rad Cell Science Division, Hemel Hempstead, UK) or using a ×20/0.70 HC PL APO objective on a Leica TCS SP2 AOBS equipped with Leica confocal software (Leica Microsystems, Inc., Exton, PA). Confocal z-series were projected using Image J 1.31v (National Institutes of Health, Bethesda, MD). Montages of microscopic images were generated using Adobe Photoshop 7.0 software (Adobe Systems, Inc., San Jose, CA).
Microinjection of VEGF-Treated Spheroids
Spheroids generated in hanging drop culture were placed into a 60-mm Petri dish containing Dulbecco's phosphate buffered saline supplemented with 0.7 mmol/L CaCl2 and 0.5 mmol/L MgCl2 (EPBS; 22°C). The tip of a glass micropipette containing 1 ml of Qdots 655 was inserted into each spheroid and the Qdots microinjected into the central cavity using a pneumatically driven Pico-Injector (Medical Systems, Greenvale, NY). Microinjected spheroids were fixed in 4% paraformaldehyde (30 min, 22°C) and immunolabeled as described above.
Immunohistological Analysis of Spheroids
Spheroids were fixed and immunolabeled as described above, and then processed for histological analysis as follows. Spheroids were first dehydrated through a series of ethanol solutions (70%, 95%, 100%, 100%; 20 min each) and infiltrated with Histoclear II clearing solvent (National Diagnostics, Atlanta, GA) for 20 min (22°C). Spheroids were then infiltrated with paraffin using Histoclear:paraffin mixtures (3:1, 1:1, 1:3; 20 min each, 60°C), and finally infiltrated with 100% paraffin (20 min, 60°C). Spheroids were then embedded in paraffin and 3-μm sections were cut using a Leica RM 2125 microtome (Vashaw Scientific, Raleigh, NC), mounted on glass slides, cleared, and cover-slipped for microscopic analysis.
Physiological Analysis of Uniluminal Vascular Spheroids
Spheroids were first embedded in gels composed of type I collagen. The collagen solution (1.3 mg of type I collagen/ml) was prepared at 4°C by combining 0.8 ml of a 3.13 mg/ml solution of rat-tail collagen type I (BD Biosciences, Bedford, MA) with 1 ml of complete medium and 0.1 ml of sterile water (pH 7.5). A drop (80 μl) of chilled collagen solution was then pipetted onto the surface of a bacterial culture dish and individual spheroids, which had also been transferred into chilled collagen solution, were then pipetted into the drops. After a 15-min incubation period (37°C, 5% CO2) to allow for polymerization of the collagen, serum-free medium was added to the cultures. The cultures were then incubated in serum-free medium containing either KCl (100 mmol/L), acetylcholine (0.001 nmol/L − 1 mmol/L), L-NAME (3 mmol/L), or HDL (500 mg/ml). After addition of each agent, sequential images of the spheroids were acquired using a Leica MZ-12 microscope equipped with a Dage MTI CDD-300 camera. Morphometric measurements (diameter) were made from the images using Image J 1.31v software (National Institutes of Health, USA). The percentage of the maximal possible spheroid relaxation induced by HDL was calculated as the quotient of Δ1 (the diameter of HDL-treated spheroid-1 minus its KCl-contracted diameter) and Δ2 (the diameter of untreated spheroid-1 minus its KCl-contracted diameter) ×100.
The authors thank Dr. Tim McQuinn for advice and critical suggestions, and Joyce Edmunds for assistance with histology. We also thank Dr. Richard L. Proia (National Institutes of Health, Bethesda, MD) for providing us with the mouse strain carrying a targeted deletion of the S1P3 gene and Dr. Samar M. Hammad (Medical University of South Carolina) for providing us with HDL. C.J.D, W.S.A, and K.M.A. were funded by the NIH, V.M was funded by NSF.