Adipose stem cell‐derived extracellular matrix represents a promising biomaterial by inducing spontaneous formation of prevascular‐like structures by mvECs

Tissue constructs of physiologically relevant scale require a vascular system to maintain cell viability. However, in vitro vascularization of engineered tissues is still a major challenge. Successful approaches are based on a feeder layer (FL) to support vascularization. Here, we investigated whether the supporting effect on the self‐assembled formation of prevascular‐like structures by microvascular endothelial cells (mvECs) originates from the FL itself or from its extracellular matrix (ECM). Therefore, we compared the influence of ECM, either derived from adipose‐derived stem cells (ASCs) or adipogenically differentiated ASCs, with the classical cell‐based FL. All cell‐derived ECM (cdECM) substrates enabled mvEC growth with high viability. Prevascular‐like structures were visualized by immunofluorescence staining of endothelial surface protein CD31 and could be observed on all cdECM and FL substrates but not on control substrate collagen I. On adipogenically differentiated ECM, longer and higher branched structures could be found compared with stem cell cdECM. An increased concentration of proangiogenic factors was found in cdECM substrates and FL approaches compared with controls. Finally, the expression of proteins associated with tube formation (E‐selectin and thrombomodulin) was confirmed. These results highlight cdECM as promising biomaterial for adipose tissue engineering by inducing the spontaneous formation of prevascular‐like structures by mvECs.

vascular system would be desirable to allow constructs of a larger size and to maintain comprehensive cell behavior. Furthermore, such vascularized tissue constructs would allow in vitro investigations regarding the development and therapy of vascular diseases.
The inclusion of a functional vascular system remains one of the biggest challenges in three-dimensional (3D) tissue engineering. To date, there are several strategies to vascularize engineered 3D tissue constructs, for example, functionalized scaffolds, perfusion bioreactors, cocultures, and in vivo approaches (Laschke & Menger, 2016). Proangiogenic factors immobilized in the scaffold material were found to enhance vascularization (Laschke et al., 2008;Yoon, Chung, Lee, & Park, 2006). For example, vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) are known to induce vessel formation and platelet-derived growth factor (PDGF) β supports stabilization of the newly formed vessels (Gaengel, Genove, Armulik, & Betsholtz, 2009). Different coculture systems, including monolayer or spheroid cultures, demonstrated spontaneous formation of vascular-like structures (Walser et al., 2013;Wenz, Tjoeng, Schneider, Kluger, & Borchers, 2018). In particular, the coculture of endothelial cells (ECs) with adipose-derived stem cells (ASCs) showed a beneficial effect on prevascular-like structure formation (Verseijden et al., 2012;Volz, Hack, Atzinger, & Kluger, 2018). Abovementioned techniques help to prevascularize a construct, but the complete vascularization is only achieved post implantation in vivo (Laschke, Strohe et al., 2009;Laschke, Vollmar, & Menger, 2009). So far, there is no successful in vitro approach to create a physiological and functional vascular system, which ensures adequate stability and reproducibility. In most approaches, some type of feeder cells are used to support the formation of vascular-like structures by ECs.
This living cellular part impedes a commercial application due to the difficulties in handling and storage. In contrast, lyophilized acellular biomaterials can be stored for long periods. Moreover, they evoke far fewer concerns regarding their application in regenerative medicine compared with the approaches including living cells. However, changes in structure and composition may occur during dehydration of natural materials. Thus, it has to be clarified if the processing of a biomaterial affects its ability to influence cellular behavior. To address this issue, next to the effect of the wet hydrogel-like form, the effect of the dehydrated materials on cellular behavior should be investigated.
A critical requirement for engineering tissue constructs is the use of a suitable scaffold that provides appropriate biological and physicochemical properties. The cell surrounding material also plays an important role in vascularization. There are several synthetic and natural scaffold materials used for vascularized tissue engineering approaches, for example, polylactic acid, polyethylene glycol, collagen, or hyaluronic acid. However, the extracellular matrix (ECM) as the natural environment of the cells in vivo represents the most physiological biomaterial. A variety of ECM-hybrid materials and pure decellularized ECM were investigated towards their ability to support stem cell differentiation and (neo)vascularization in vivo and in vitro (Adam Young, Bajaj, & Christman, 2014;Badylak, Freytes, & Gilbert, 2009;Flynn, 2010;Flynn, Prestwich, Semple, & Woodhouse, 2009). All these studies were performed with decellularized ECM derived from native tissue. For the past years, another source of natural ECM moves to the fore. In vitro generated cell-derived ECM (cdECM) was isolated from different cell-types (e.g., fibroblasts and ASCs) and used as a biomaterial in a variety of applications Lu, Hoshiba, Kawazoe, Koda et al., 2011;Sart et al., 2016;Schenke-Layland et al., 2009;Wolchok & Tresco, 2010). Several studies show that cdECMs, obtained from different cell-types, can induce adipogenic, chondrogenic, and osteogenic differentiation of ASCs indicating its influence on cell fate (Dzobo et al., 2016;Guneta et al., 2017;Guneta, Loh, & Choong, 2016;Guo et al., 2013).
Our previous study revealed a spontaneous formation of prevascular-like structures by microvascular endothelial cells (mvECs) in a coculture with adipogenically differentiated ASCs (Volz et al., 2018). In the following the term "prevascular-like structures" defines the aggregation/alignment of CD31-positive mvECs to fiber-or network-like structures, which stand out from the rest of the cellular monolayer. This term was previously used by  and Verseijden, Posthumus-van Sluijs, Pavljasevic et al. (2010) to describe the alignment of ECs in spheroids without lumen formation. The formation of vacuoles within ECs and subsequent lumenogenesis (tube formation) of prevascular-like structures requires the activation of cellular pathways and the transcription of different genes (Bayless & Davis, 2002). Furthermore, the expression of E-selectin and thrombomodulin was shown to contribute to tube formation (Oh et al., 2007;Pan et al., 2017). In this study, we aimed to analyze whether the formation of prevascular-like structures by mvECs has to be attributed to cell-cell or cell-matrix interactions. The maintenance of the biological impact after processing and storage represents an important feature regarding the commercial application of biomaterials. The most common processing method for the preservation of biomaterials is drying. Consequently, we directly compared the effect of the hydrogel-like, wet cdECM and the dried cdECM as a coating regarding its ability to support the formation of prevascular-like structures by mvECs. The formation of vascular structures is rather a developmental process than a maintenance phenomenon as it can, for example, be found during (adipogenic) differentiation rather than in the stem cell niche. Thus, we tested whether there is a difference between cdECM derived from stem cells and adipogenically differentiated cells regarding their capability to induce prevascular-like structure formation.

| MATERIALS AND METHODS
All research was carried out in accordance with the rules for the investigation of human subjects as defined in the Declaration of Helsinki. Patients provided written agreement in compliance with the Landesärztekammer Baden-Württemberg (F-2012-078, for normal skin from elective surgeries).

| Cell isolation and expansion
ASCs were isolated from human tissue samples obtained from patients undergoing plastic surgery (Dr. Ziegler; Klinik Charlottenhaus, NELLINGER ET AL.

| Macroscopic pictures and degree of swelling
F I G U R E 1 Schematic overview of the study procedure. The proangiogenic potential of different acellular and cellular substrates was analyzed. ASCs were cultured in growth and adipogenic differentiation medium respectively for 7 days. For acellular ECM substrates (cdECM), ASCs were removed and the remaining cdECM was dried or stored under wet conditions. For cellular substrates (FL), ASCs were not removed. MvECs were seeded onto the different substrates. Cytocompatibility was determined at Day 3 and prevascular-like structure formation was determined at Day 14 of cell culture.

| Seeding of mvECs on cell-derived ECM and feeder layer
Isolated dry and wet cdECM substrates were reseeded with mvECs at a density of 1 × 10 4 cells/cm 2 in a defined mvEC adipocyte coculture medium (Volz et al., 2018). For FL approaches, mvECs were directly seeded on top of adipogenically differentiated and undifferentiated ASCs at a density of 1 × 10 4 cells/cm 2 in defined coculture medium, developed by us earlier (Volz et al., 2018). Cells were cultured for 14 days and the medium was changed every other day ( Figure 1). As a control, all experiments were performed on Collagen I (COL I; rat tail; 250 μg/ml in 0.1% acetic acid) coated tissue culture polystyrene (COL I) and uncoated tissue culture polystyrene (TC). All media were supplemented with 1% penicillin/streptomycin.

| Immunofluorescence staining of cell-specific proteins
For IF staining of cell-specific proteins, cells were fixed in 4% paraformaldehyde for 10 min and permeabilized for 10 min with 0.1%

| Enzyme-linked immunosorbent assay
For characterization of cdECM substrates regarding growth factors composition, substrates were washed 3 days in culture medium. For the characterization of FL, medium from Day 3 was collected.
Quantification of growth factors VEGF, bFGF, and PDGFβ was performed using enzyme-linked immunosorbent assays (ELISA) (all PE-PROTech, Germany) according to the manufacturer's instructions.

| Statistical analysis
All experiments were performed at least three times, using cells from at least three different biological donors of ECs. The obtained data were compared by a one-way analysis of variance with repetitive measurement and a Bonferroni post hoc test using OriginPro 2018b.
Statistical significances were stated as *p ≤ .05, very significant as **p ≤ .01, and highly significant as ***p ≤ .001.   F I G U R E 3 Cytocompatibility of the acellular and feeder layer substrates. About 1 × 10 4 cells/cm 2 mvECs were seeded in a defined medium onto the different substrates. (a) Relative LDH release was measured at Day 3 after seeding with mvECs. For acellular substrates, values were normalized to TC. No significant increase in LDH release could be observed on COL I coating or dry and wet cdECM for both, scdECM and acdECM. For FL approaches, values were normalized to stem cell FL without mvECs (FL stem cell). None of the FL approaches (stem cell and adipogenic differentiated) exhibited a significant increase of released LDH after seeding of mvECs (FL + mvECs). (b) Live-dead staining (FDA, indicating alive cells displayed in green/PI, indicating dead cells, displayed in red) was performed on Day 14 after seeding with mvECs. A confluent layer of viable cells was observed in all approaches. For each approach, the percentage of dead cells was quantified using ImageJ. The analysis revealed an amount of <1% of dead cells for all acellular substrates with no significant differences between the different substrates. In FL approaches, a higher number of dead cells could be found in adipogenic approaches compared to stem cell approaches. F I G U R E 4 Formation of prevascular-like structures by mvECs on cellular and acellular ECM substrates. About 1 × 10 4 mvECs /cm 2 were seeded in defined coculture medium onto the different substrates and were cultured for 14 days. Medium was changed three times a week. For determination of newly formed prevascular-like structures, IF staining of CD31 (indicated in red) was performed at Day 14 after seeding with mvECs. On controls (TC and COL I), a confluent layer of mvECs could be observed without any structure formation. On cdECM substrates, the formation of prevascular-like structures could be observed with the strongest manifestation on wet acdECM. The highest degree of structure formation could be observed on adipogenic FL. On stem cell FL cluster formation of mvECs could be found and a considerably lower degree of structure formation compared to the adipogenic approach was detected. For each representative overview image, a magnified section of the cellular monolayer or the prevascular-like structures is pictured in the upper right corner to show the localization of CD31 at cell-cell contacts. Length per structure and number of formed nodes was quantified using ImageJ. Analysis revealed a significantly higher structure length of mvECs on wet acdECM substrate compared to dry acdECM and stem cell approaches (dry and wet scdECM), comparable with FL approaches. Structure length on adipogenic FL was significantly higher compared with all approaches except wet acdECM.

| Quantification of proangiogenic factors on substrates
To confine which cdECM components are responsible for its proan- On FL substrates, a higher concentration of PDGFβ could be found compared with acellular substrates but no difference between stem cell and adipogenically differentiated approach was observed.

| Expression of proteins associated with tube formation in newly formed prevascular-like structures
Studies showed that the expression of adhesion molecules E-selectin and thrombomodulin in ECs is associated with the tube formation of new blood vessels (Oh et al., 2007;Pan et al., 2017). To get an indication if lumenogenesis occurs to any extent, we investigated the expression of these proteins in the newly formed prevascular-like structures ( Figure 6). approaches. E-selectin and thrombomodulin staining corresponded to the CD31 staining pattern of the prevascular-like structures ( Figure S1).

| DISCUSSION
The implementation of a functional vascular system into an engineered tissue construct would address one of the major F I G U R E 5 Proangiogenic factor concentrations on cellular and acellular ECM substrates. For the determination of VEGF, bFGF, and PDGFβ from the different substrates, supernatant from Day 3 was investigated regarding the concentration of the growth factors using ELISA. For statistical analysis, values on TC were set as 1 and data were normalized to TC. For VEGF, a significantly higher amount could be found in acdECM substrates (dry and wet) compared with all other acellular substrates including controls. FL substrates exhibited a 10-fold higher concentration of VEGF compared with acellular substrates. For bFGF, a higher concentration could be found in cdECM substrates compared with controls. On FL approaches, a higher concentration could be found compared with all other approaches. For PDGFβ, a significantly higher (three-fold) amount could be found in all cdECM substrates compared with the controls TC and COL I. Between the individual cdECM substrates no difference in remaining PDGFβ could be found. FL substrates exhibited significantly higher PDGFβ concentrations compared with acellular substrates. acdECM, adipogenic cell-derived extracellular matrix; bFGF, basic fibroblast growth factor; cdECM, cell-derived extracellular matrix; COL I, collagen I; ELISA, enzyme-linked immunosorbent assay; FL, feeder layer; PDGFβ, platelet-derived growth factor beta; scdECM, stem cell-derived extracellular matrix; TC, tissue culture; VEGF, vascular endothelial growth factor. (*p ≤ .05; **p ≤ .01; ***p ≤ .001) bottlenecks in tissue engineering and regenerative medicine. In the present study, we aimed to investigate the supportive effect of cdECM on the self-assembled formation of prevascular-like structures by mvECs for its use as a biomaterial for adipose tissue en- which results in a lower total cell number in adipogenic approaches (Fajas, 2003). Thus, the amount of cdECM in stem cell approaches may be higher which leads to more densely packed collagen fibers and smaller pores. The larger pores found in adipogenic ECM might also be able to enhance the degree of prevascular-like structure Visualization of mvECs on Day 14 after seeding by staining of the specific surface protein CD31 showed the self-assembled formation of prevascular-like structures on all substrates except for the controls COL I and TC. Structure formation on the adipogenic FL approach was in line with our previous study (Volz et al., 2018) as a lower degree of structure formation was found on the stem cell FL approach. In addition, on dry and wet cdECM approaches, the degree of prevascular-like structure formation on adipogenic ECM substrates was higher compared with the corresponding stem cell approach, which is reflected by longer structures and a higher number of nodes. The effect of enhanced structure formation on adipogenic substrates could be explained by the different secretomes of ASCs and (pre-)adipocytes (Kapur & Katz, 2013). It is well known that ASCs secrete a broad spectrum of proangiogenic proteins and they were often used as a delivery system of growth factors and cytokines in vascularization approaches (Kondo et al., 2009;Liu et al., 2011;Moon et al., 2006;Nakagami, Maeda, Kaneda, Ogihara, & Morishita, 2005;Rehman et al., 2004). For example, Matsuda et al. (2013) showed that conditioned cell culture medium of ASCs positively influenced EC proliferation and the formation of new vessels in vivo. During adipogenic differentiation, ASCs secret further proangiogenic factors like leptin. Leptin is known to be upregulated during adipogenic differentiation and was shown to exhibit a proangiogenic effect itself but also upregulates the secretion of VEGF (Cao, Brakenhielm, Wahlestedt, Thyberg, & Cao, 2001 Tomanek, Hansen, & Christensen, 2008). Therefore, in our study, the induction of the formation of prevascular-like structures might among other events, be attributed to the synergistic effect of available VEGF and bFGF. We further investigated the amount of the proangiogenic factor PDGFβ from cdECM substrates.
It is secreted by ECs during angiogenesis to attract perivascular cells, which stabilize the newly formed vessels (Gaengel et al., 2009).
Nevertheless, it also plays a role in the homing of endothelial progenitor cells (EPCs) and therefore promotes neovascularization.
Studies showed that E-selectin potentiates angiogenesis in ischemic tissue, by mediating EPC-endothelial interactions (Oh et al., 2007).
During this process of neovascularization, EPCs are mobilized from the bone marrow into the circulation and recruited to new sites of vascularization, using cues that resemble an inflammatory response.
Therefore, E-selectin plays a crucial role in EPC homing and following neovascularization and tube formation. Thrombomodulin is a transmembrane protein expressed on ECs acting as an anticoagulant (Dahlback & Villoutreix, 2005;Dittman & Majerus, 1990). The fourth and fifth region of an EGF-like region of thrombomodulin (TME45) was shown to stimulate the proliferation of human umbilical vein ECs and to promote tube formation and angiogenesis .
In this study, we use these proteins as indicators for the development of the prevascular-like structure towards a tubular vascular structure with a lumen. IF staining of E-selectin and thrombomodulin revealed specific expression of E-selectin almost exclusively on the newly formed prevascular-like structures. Thus, we suggest that newly autologous tissue and the generation of ECM from different developmental stages. This study confirms a supportive effect of cdECM on the spontaneous formation of prevascular-like structures by mvECs. Furthermore, it could be shown that dry cdECM partly maintains its biological properties regarding the induction of selfassembled prevascular-like structure formation of mvECs with some restrictions. Drying of the cdECM would be a convenient method to improve storage possibilities when necessary. Due to the relatively low amounts of cdECM, which can be produced with current methods, this study is limited to 2D approaches, which insufficiently reflect physiological conditions. Further studies should focus on the upscaling of the generation of cdECM to enable the setup of continuative experiments in 3D constructs consisting of cdECM, which would better reflect the situation in vivo.

| CONCLUSION
In the present study, we demonstrated that cdECM (as a dry coating and as a wet hydrogel-like form) is able to induce the self-assembled formation of prevascular-like structures by mvECs and helps to support their maintenance. Mainly acdECM was confirmed as a promising material for adipose tissue engineering by supporting the formation of prevascular-like structures. In addition, scdECM also provides the ability to induce prevascular-like structure formation and can be used for approaches addressing other tissues. In future investigations regarding other lineage-specific cdECMs, the upscaling of cdECM generation and the transfer from 2D cell culture to 3D cell culture should be pursued.

ACKNOWLEDGMENTS
This study was financially supported by the Landesgraduiertenförderung by the Ministry of Science, Research, and the Arts (Baden-Württemberg, Germany) under the program "Intelligent Process and Material Development in Biomateriomics" (University of Tuebingen and Reutlingen University).