Short-term post-implantation dynamics of in vitro engineered human microvascularized adipose tissues

Tissue engineering is now recognized as a promising approach for tissue reconstruction and replacement. To ensure the survival, the integration and the maintenance of the functions of engineered substitutes, a fast and adequate vascularization must be achieved upon grafting [1, 2]. Angiogenesis can be stimulated by the use of various in vitro strategies that have been addressed in many reviews [3–8]. In particular, cells forming engineered tissues can be genetically modified to express angiogenic factors. Alternatively, angiogenic factors can be incorporated into the scaffold or encapsulated into microspheres that are combined with the constructs. Encapsulation strategies with biomaterials can also improve the angiogenic capacity of cells used for vascular regeneration. Hypoxic preconditioning on substitutes is also performed to stimulate angiogenesis. Another approach relies on the incorporation of endothelial cells within the constructs, a technique called prevascularization. Both human umbilical vein endothelial cells (HUVECs) or adult microvascular cells (from the endothelium of adult, postnatal small-caliber vessels), are able to undergo tubulogenesis and create a network of capillaries in vitro. After implantation, the preformed network of capillaries is expected to connect with the host vasculature to ensure a fast perfusion of the grafts [3–6]. This strategy can be applied to a multitude of substitutes such as reconstructed skin containing a microvascularized dermis [9, 10], a field in which the Laboratoire d'organogénèse expérimentale center has been a pioneer [11–13]. Our team also developed an adipose tissue model featuring a capillary network in vitro [14, 15], since the field of adipose tissue engineering would also greatly benefit from enhanced graft vascularization upon implantation.

Autologous fat grafting or lipotransfer is currently one of the most employed technique to fill soft tissue defects for patients affected by accidents, deep burns, tumor removal or congenital diseases. However, this approach is associated with a limited long-term volume retention ranging from 20%–80% of the initial volume after one year [16, 17]. This would be attributable to the fragility of mature lipid-containing adipocytes that can rupture during fat aspiration and injection, but also to the disruption of the native vascular network during these procedures [18]. Adipose tissue is now recognized as an active producer of many different hormones, growth factors and cytokines acting locally and systemically to impact key biological processes such as food intake, energy balance, blood pressure, angiogenesis, to name a few [19–21]. For example, leptin secretion is indicative of the adipocyte metabolic status, while the balance between secreted angiopoietin (Ang)-1 and Ang-2 modulates angiogenic processes [22–25]. The engineering of adipose tissue is therefore important as an alternative to fat lipotransfer and should aim at producing substitutes that will procure long-term structural and functional enhancement after grafting.

Nowadays, adipose tissue is known as a source of stromal/stem cells possessing an important multi-differentiation potential [26, 27], which is very useful for the engineering of various tissue types including adipose tissue itself [28–32], in addition to secretory functions beneficial in regenerative medicine [33]. Our method for adipose tissue engineering is the self-assembly technique [30, 34, 35]. This approach is based on the property of mesenchymal cells such as human adipose-derived stromal/stem cells (ASCs) to secrete and efficiently assemble their own extracellular matrix components upon stimulation with ascorbic acid and serum [36–38]. This results in the formation of matrix-rich cell sheets, without the use of thermoresponsive cultureware, that are superposed in the desired number to obtain thicker tissues. In parallel to matrix accumulation, an adipogenic induction step allows ASC differentiation into adipocytes and results in the production of adipose sheets [30, 36]. The seeding of human microvascular endothelial cells (hMVECs) onto adipose sheets during the reconstruction enables the production of human reconstructed adipose tissue substitutes containing a preformed network of capillary-like structures [14]. Indeed, the natural human cells and matrix provide a biologically relevant environment conducive to the migration, proliferation, and tubule formation by adult hMVECs, leading to a capillary network that can be modulated by external angiogenic modulators [15].

In this study, we report on the fate of human engineered adipose tissue substitutes, preseeded or not with hMVECs in vitro, after their implantation in vivo. Extensive imaging analyses have been performed in order to describe the early kinetics of vascularization of the grafted tissues in nude mice (0–14 d). We hypothesized that preseeding hMVECs in vitro during the engineering of adipose tissue substitutes would favor anastomosis events upon implantation in vivo, and procure benefits to the grafted tissues by improving the extent of vascularization. Globally, our short-term in vivo experiments established the high compatibility of our constructs to integrate at the site of implantation, both in the presence and in the absence of hMVECs. The reorganization of the hMVECs in vivo led to an extended human capillary network that connected with the host vessels, as supported by the immunolabelings for human and murine CD31 and by the presence of murine red blood cells (mRBCs) within the human capillaries. Finally, the evaluation of the global extent of graft vascularization indicated a tendency towards an increase of the total capillary network surface and volume in the presence of hMVECs within the grafted adipose tissues substitutes.

Cell isolation

Reconstruction of human adipose tissue substitutes

In vitro contraction and thickness determination

Secretion profile analysis

In vivo implantation

Macroscopic and histological analyses

Immunological stainings

Global microvascularization image analysis

Statistical analyses

Adipose tissue substitutes composed only of human cells and matrix elements were produced in vitro by tissue engineering. Macroscopic and phase contrast microscopy images of these human adipose tissue substitutes revealed a semi-opaque whitish aspect (figures 1(a) and (b)) and the presence of adipocytes (figures 1(c) and (d)) for both types of tissues. Masson's trichrome staining on histological cross-sections also confirmed the presence of adipocytes (void spaces) and the presence of an abundant extracellular matrix (in blue) in both experimental groups (figures 1(e) and (f)). Quantification of the number of adipocytes on these histological sections indicated an average of 377 adipocytes per mm² of human adipose tissue substitutes. Before grafting, the substitutes were detached from their peripheral anchorage device (figures 1(a) and (b), anc = anchorage device). The resulting tissue contraction (figure 1(g)) was not significantly different between experimental groups. The thickness of the substitutes, resulting from the superposition of three cell sheets, was also evaluated and no significant differences were measured as a function of hMVECs addition (figure 1(h)).

Zoom In Zoom Out Reset image size

Next, immunolabelings were performed on whole-mount samples of the human adipose tissue substitutes before grafting to assess the adipocyte and endothelial content of the tissues. It revealed numerous perilipin-positive adipocytes within each adipose tissue substitutes (figure 2(a)), with the additional presence of hCD31-positive structures for the substitutes preseeded with hMVECs only (figure 2(a)). The hMVECs structures were also positive for the endothelial marker vWF (figure 2(a)). The endothelial cell-secreted protein Ang-2 was detected in conditioned media from substitutes enriched with hMVECs while few to no expression was detected in absence of endothelial cells (figure 2(b)). Similar leptin levels were measured in conditioned media from both tissue types independently of the presence of endothelial cells (figure 2(c)).

Zoom In Zoom Out Reset image size

The reconstructed adipose tissues were then implanted subcutaneously on the muscle of both flanks of adult male athymic mice (figure 3). Grafts were harvested after 3, 7 and 14 d of implantation for short-term behavior analysis. The gross appearance of the grafted adipose tissue substitutes was shiny and smooth (lipid-filled adipocytes), while the recipients bed pink color at all time-points suggests progressive neovascularization (figure 4(a)). Tissue remodeling occurred after implantation leading to a 38.7%–57.0% decrease in graft surface after 3 d when compared to the initial graft size at day 0 (figure 4(b), p < 0.0001). The initial graft surfaces were 130.4 ± 20.6 mm² and 132.3 ± 23.1 mm² (ASC population #1) and 198.7 ± 43.1 mm² and 194.5 ± 38.5 mm² (ASC population #2) for the adipose tissue substitutes preseeded or not with endothelial cells, respectively. For both experiments, graft surface measured on macroscopic images decreased between day 3 and 14 but no significant differences were associated with endothelial cell incorporation (figure 4(b)).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The persistence of adipocytes (void spaces) over 14 d and the presence of blood vessels (arrows) within the grafts were first evaluated on histological cross-sections (figures 5(a) and S2). As part of the tissue remodeling process, a gradual thickening of the grafted tissues was observed (figures 5(a) and (b)). The graft thickness values from day 3, 7 and 14 (figure 5(b)) were significantly different (p < 0.001 or p < 0.0001) from the day of grafting (figure 1(h)), but were not significantly different between experimental groups at the three time-points considered (figure 5(b)). Based on surface and thickness analyses, we calculated an estimated graft volume index (figure 5(c)), which indicates a rather stable graft maintenance over time for all series. Similarly, quantification of adipocyte numbers in the grafts was performed, supporting the persistence of the adipocytes for at least 14 d after implantation, for both types of human adipose tissue substitutes (figure S2).

Zoom In Zoom Out Reset image size

In terms of microvasculature development, a fast reorganization of the hMVECs was observed in vivo. Indeed, hCD31 staining revealed the presence of an organized human capillary network within the grafted adipose tissue substitutes containing hMVECs at day 3 of implantation (figures 6(a) and (a')). Connections between the human network and the host vessels were visible at the interface between the graft and it's recipient bed as early as 3 d after implantation (figure 6(b)) and were also observed after 14 d (figure 6(b')). Blood circulation within the human capillaries was also demonstrated following immunolabelings showing the presence of mRBCs within the capillary structures (figures 6(c) and (c')) after 14 d.

Zoom In Zoom Out Reset image size

A more detailed assessment of the murine and human capillary networks evolution within the grafts after 3, 7 and 14 d of implantation revealed a gradual sprouting of the host network (in green) from the recipient bed towards the grafted tissues for both experimental groups (figure 7(a)). Of note is the presence of the human capillary network (in red) within the microvascularized grafts (figures 6(c) and 7(a)). Indeed, confocal images of whole-mount samples showed that the human capillary network was robustly present with well-defined capillary structures even at 14 d after grafting (figures 8(a) and (a')). This is confirmed by the quantification of the graft volume occupied by the human capillary network showing a significant increase over time (figure 8(b)).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In terms of global vascularization (human and murine networks), the evaluation of the graft surface occupied by the total capillary network on tissue cross-sections revealed an increase over time for both experimental groups with a tendency towards a higher percentage for human adipose tissue substitutes preseeded with hMVECs (8.0 fold, 2.6 fold and 1.6 fold for day 3, 7, and 14, respectively) (figure 7(b)). The quantification of the graft volume occupied by the total network (human and murine) was next performed using confocal images of whole-mount samples (figure 7(c)). Similarly to the surface analysis, when expressed as tissue volume occupied by the networks, a higher percentage of capillary network coverage was measured for human adipose tissue substitutes preseeded with hMVECs after 3, 7 and 14 d of implantation (11.3 fold, 4.0 fold, 2.4 fold, respectively), although it did not reach statistical significance (figure 7(c)).

In order to achieve a successful implantation of engineered tissues, a timely perfusion of the grafts is required. The incorporation of an endothelial component to engineered tissues has become a widely studied strategy to favor rapid vascularization after grafting [3–6, 41, 42]. In the field of soft tissue engineering in particular, various models of prevascularized substitutes are being developed, comprising either differentiated adipocytes (adipose tissue substitutes) [14, 43–49], or adipogenic-induced or non-induced human ASCs (connective tissue substitutes) [50–55]. These human cell-based reconstructed tissues, which differ in cell types, scaffolding elements, formulations, and size, are used as tools for in vitro studies or as potential tissue fillers in preclinical studies.

In this study, we evaluated the impact of the in vitro incorporation of adult microvascular endothelial cells on the vascularization dynamics of our adipose tissue model after transplantation. These engineered adipose tissues represent highly relevant substitutes as they are densely populated with adipocytes but are still easily manipulated (using or not a peripheral anchorage device) and then sutured in vivo. They are engineered using ascorbic acid supplementation to promote matrix production and assembly by the ASCs while supporting adipogenic differentiation, leading to adipose cell sheets that are then combined into thicker tissues [36]. The production of these adipose tissue substitutes does not rely on the use of thermoresponsive plates or exogenous scaffolds/supporting devices also used for the engineering of cell sheet-based substitutes [56, 57]. Under our culture conditions, the extent of adipogenic differentiation was also similar in presence or absence of hMVECs, as observed in our previous work [14]. Perilipin staining revealed the presence of numerous viable adipocytes in the substitutes. Perilipin is a lipid-droplet associated protein and loss of perilipin content is associated with adipocyte cell death [58–61]. Quantification of leptin secretion confirmed the presence of functional adipocytes in our human adipose tissue substitutes. The incorporation of hMVECs did not impact leptin secretion under these engineering conditions, which were previously optimized to favor both adipogenesis and angiogenesis [14]. Maintaining functional adipocytes in vascularized adipose tissue models is important and can be a challenge considering that the timing of endothelial cell incorporation, the seeding density and the choice of cell culture media and supplements can clearly negatively impact adipose tissue engineering [14, 62–64]. This explains why many soft tissue models do not actually comprise differentiated adipocytes at the time of implantation, and rely on in vivo adipogenesis for adipose tissue formation. Our human adipose tissue substitutes are mainly designed to provide a living and functional adipose layer after grafting. They are not injectable and could find applications for various reconstructive surgery procedures such as maxillofacial defects and injuries in order to promote healing and restore contour and function. They could be used as an alternative or a complementary approach to autologous fat grafting. It would also be interesting to evaluate their capacity to stimulate the in situ formation of adipose depots during long-term grafting experiments.

Immunolabelings for hCD31 and vWF confirmed the endothelial nature of the structures formed within the human adipose tissue substitutes when hMVECs were seeded on adipose sheets featuring already differentiated adipocytes (14–15 d). The detection in conditioned media of Ang-2, secreted mainly by endothelial cells [65], confirmed the presence of metabolically active hMVECs in the tissues. The survival and proliferation of hMVECs are well-supported within our engineered tissues. This is likely due to the endogenous human matricial elements that are produced and assembled by ASCs in our self-assembly model [14, 15]. Indeed, the importance of extracellular matrix composition, quality and density on the modulation of angiogenic processes is now recognized [66, 67]. The stability of the capillary networks in our tissues is particularly valuable for the in vitro evaluation of angiomodulatory molecules over many days [15].

After implantation, the grafted human adipose tissue substitutes presented a smooth appearance with the presence of blood vessels surrounding the grafts, independently of their prevascularization status. For both groups, the graft surface area decreased sharply during the first three days after implantation but slowed down afterwards, as evaluated from macroscopic pictures. This was accompanied by a thickening of the grafts measured on histological cross-sections, supporting the expected tissue contraction and remodeling. The reorganization of the hMVECs into fine capillaries observed three days after grafting confirmed their capacity to form an extended network that persisted and even developed throughout the 14 d period we analyzed. The human capillary network was able to connect with the host microvasculature as soon as three days after grafting and was also able to convey blood supply to grafted adipose substitutes, as supported by mRBCs found inside human capillaries.

In our study assessing the early dynamics of the reconstructed adipose grafts, no impact of the presence of hMVECs was observed on their estimated volume over 14 d. The incorporation of hMVECs translated into the presence of a developed and functional network of capillaries leading to an increased global vascularization after implantation, although statistical significance was not reached when compared to non-microvascularized tissues. Indeed, the vascularization analyses were performed on a limited number of samples. However, we used two approaches to quantify the capillary networks, namely on frozen cross-sections (% surface) and on whole-mount samples (% volume), which independently support the conclusions.

In a study conducting a side-by-side comparison of constructs devoid or preseeded with HUVECs, Verseijden and collaborators evaluated their model 7 d after transplantation into nude mice [54]. Their prevascularized fibrin constructs contained a 1:5 ratio of human ASCs (induced towards adipogenesis for 7 d) and HUVECs. Similarly to our results, no differences were seen for the total graft vascularization (human and murine) after 7 d. However, the contribution of the human-specific vascularization seemed much lower for these constructs than for our human adipose tissue substitutes [54]. In our study, the absence of a significant impact on graft volume and the extent of global vascularization in presence of hMVECs is likely due to the rapid vascularization of the tissues averaging 120 μm in thickness on histological sections before grafting. Despite the large surface of our constructs (130–200 mm²), the incorporation of hMVECs might not provide a robust stimulation to an already effective neovascularization of the non-microvascularized tissues. It will therefore be important to evaluate the impact of hMVECs on the vascularization of thicker tissues in which the oxygen rate is limiting and to extend the follow-up period after implantation. The challenge of accurately determining the volume and shape of grafted adipose substitutes is always present when performing preclinical studies. In our previous work, thicker human adipose tissue substitutes that were not microvascularized in vitro (∼1 mm-thick) were successfully implanted into nude mice and followed in vivo using magnetic resonance imaging (MRI). Dynamic contrast-enhanced (DCE)-MRI studies were also performed, showing signal enhancement indicative of the graft perfusion after 14 and 21 d. Graft volume retention was followed up to 6 weeks, showing a gradual resorption to 44% of the substitutes initial volume on average [39]. Therefore, determining the influence of prevascularization on thicker adipose tissue substitutes will be particularly informative using these non-invasive imaging modalities in longitudinal comparative studies.

Other teams have evaluated the impact of HUVEC co-transplantation over a longer period of time after grafting and found a rather limited impact of prevascularization. In their study, Frerich and collaborators compared two different types of constructs, namely microparticles (on collagenous carriers) and larger constructs (13 mm in diameter) for which the microparticles were embedded in a fibrin gel [55]. After 12 days, 4 weeks or 4 months post-implantation, the mean explanted weight of the fibrin constructs was similar whether or not they were preseeded with HUVECs. Quantifications performed on histological cross-sections revealed a significantly smaller number of acellular/necrotic areas after four weeks and four months for HUVECs-enriched experimental groups. However, no long-term benefits could be measured in terms of adipose tissue formation over four months [55]. Globally, this highlights that more research is needed to understand the potential benefits of prevascularization and the dynamics of fat grafting, both for native and reconstructed substitutes since many parameters can influence the outcomes of grafting such as the anatomic site of the recipient bed, the size and delivery of the constructs, as well as the choice of an animal model (immunocompromised or not) [68].

Successful reconstruction of adipose tissue requires efficient strategies, such as in vitro prevascularization, to stimulate angiogenesis after transplantation. The short-term behavior of the entirely human adipose tissue substitutes we produced, preseeded with human microvascular endothelial cells, indicated the formation of human capillary networks that developed and even expanded after transplantation in mice. Our model displays many advantageous features: our biomimetic, scaffold-free human substitutes are built from adipose-derived stem cells, which efficiently produced and assembled extracellular matrix in addition to their recognized capacity to secrete therapeutic molecules. The lack of exogenous biomaterials could reduce the risks of immunogenic reactions in some patients. We used adult microvascular cells that are more representative of autologous therapy. Therefore, it will be relevant to evaluate the potential benefits of preformed human capillary networks in thicker human adipose tissue substitutes in order to engineer optimal replacement tissues for patients suffering from adipose tissue loss or defects.

This work was supported by the Canadian Institutes of Health Research (CIHR) grants #84368 and #111233. We acknowledge the support of the CHU de Québec-Université Laval Research Center of the Fonds de Recherche du Québec-Santé (FRQS) and of the Quebec network for cell, tissue and gene therapy-ThéCell (a thematic network supported by the FRQS). MP received studentships from National Science and Engineering Research Council of Canada (NSERC), FRQS and Canadian Federation of University Women (CFUW). JF received a FRQS career award. The confocal imaging system was obtained through the Fonds des leaders program from Canada Foundation for Innovation (CFI) to JF. The authors are grateful to Amandine Maux for technical assistance and experimental input. We thank Stéphane Chabaud and Véronique Moulin for providing human skin microvascular endothelial cells as well as Anne-Marie Moisan and Michel Fortin for excellent assistance with the animal studies.

No conflicts of interest.