New Insight on 2D In Vitro Angiogenesis Models: All That Stretches Is Not a Tube

Highlights Tube formation on MatrigelTM and tube formation in co-culture with MSCs are two different stages of angiogenesis. uPA, uPAR, Jagged1, and Notch2 are common upregulated genes for ECs on MatrigelTM, in co-culture and in dividing/migrating cells. EndMT activated at a much greater extent in ECs in a co-culture model than in a MatrigelTM assay. Only in the MatrigelTM assay are the Notch and Hippo pathway-related genes upregulated. Abstract A Matrigel-based tube formation assay is a simple and widely accepted 2D angiogenesis model in vitro. Extracellular matrix (EM) proteins and growth factors (GFs) from MatrigelTM exclusively trigger endothelial cell (EC) tubular network (ETN) formation. Co-culture of ECs with mesenchymal stromal cells (MSCs) is another and more reliable in vitro angiogenesis assay. MSCs modulate ETN formation through intercellular interactions and as a supplier of EM and GFs. The aim of the present study was to compare the expression profile of ECs in both models. We revealed upregulation of the uPA, uPAR, Jagged1, and Notch2 genes in dividing/migrating ECs and for ECs in both experimental models at 19 h. The expression of endothelial–mesenchymal transition genes largely increased in co-cultured ECs whereas Notch and Hippo signaling pathway genes were upregulated in ECs on MatrigelTM. We showed that in the co-culture model, basement membrane (BM) deposition is limited only to cell-to-cell contacts in contrast to MatrigelTM, which represents by itself fully pre-assembled BM matrix. We suggest that ETN in a co-culture model is still in a dynamic process due to immature BM whereas ECs in the MatrigelTM assay seem to be at the final stage of ETN formation.


Introduction
Endothelium in healthy organisms is a cellular monolayer of tightly contacted cells. Various external influences such as hypoxia, tissue damage, or inflammatory factor release could cause the disruption of tight cell contacts and a subsequent increase in vascular permeability and initiation of angiogenesis, the regrowth of new blood vessels from already existing ones. While new vessels grow, endothelial cells communicate with mural cells and with the extracellular matrix (EM), remodeling it.
Although much is already known about angiogenesis, there are still a lot of blank spots, which researchers are trying to explore in different in vitro models.

Cell Isolation and Culture
The MSC cell line ASC52Telo was obtained from the collection of human biomaterials of the Institute for Regenerative Medicine (Lomonosov Moscow State University, collection ID: MSU_MSC_AD; repository catalogue at www.human.depo.msu.ru). MSCs were cultured in DMEM-GlutaMAX™ (ThermoFisher Scientific, Waltham, MA, USA) with 10% fetal bovine serum (FBS) (Cytiva, Marlborough, MA, USA) and penicillin/streptomycin (ThermoFisher Scientific, Waltham, MA, USA). Umbilical cords from healthy donors were collected in the Obstetric Department of the V. I. Kulakov National Medical Research Center for Obstetrics, Gynecology, and Perinatology after written informed consent was obtained from all women. Endothelial cells from the umbilical vein (HUVECs) were isolated as previously described [6] and cultured in complete EGM-2 medium (Lonza, Basel, Swiss). MSCs and HUVECs were grown until 70% confluence in a humidified chamber incubator at 37 • C and 5% CO 2 . Before the co-culture experiments, cells were detached with 0.05% trypsin/EDTA (ThermoFisher Scientific, Waltham, MA, USA). In all experiments, HUVECs were used at passage 3 to 5. In each experiment, we used at least three biological repeats (one donor per repeat).

Tube Formation Assay on Matrigel TM In Vitro
The in vitro endothelial cell tube formation assay was performed and quantified as described by us previously [7]. Briefly, HUVECs were seeded onto the 60 cm 2 Petri dish covered with a thick layer of Matrigel TM (Corning, NY, USA) at a density of 5 × 10 4 cells/cm 2 and cultured for 19 h in the EGM-2 media (Lonza, Basel, Swiss). Cells were utilized for mRNA expression analysis or fixed with 4% formaldehyde and visualized using a wide-field fluorescent Axiovert 200 M Microscope (Zeiss, Oberkochen, Germany).

HUVEC-MSC Two-Dimensional (2D) Co-Culture Model
A co-culture model was performed as described by us previously [4]. Briefly, the mixtures of HUVEC:MSC at a ratio of 1:3 and at a total density of 6 × 10 4 cells/cm 2 were seeded onto a 100 cm 2 Petri dish and co-cultured for 19 h in EGM-2 for mRNA expression analysis or for 48 h for immunofluorescent staining.

The 2D Co-Culture Model of HUVECs with MSCs and Its 48 h Synthesized Extracellular Matrix (EM)
MSCs were grown as a monoculture for 48 h ("48 hold MSCs") to allow synthesis/secretion/assembly of extracellular matrix. HUVECs were then seeded on the top of "48-h-old MSCs" and co-cultured for an additional 19 h in EGM-2 for mRNA expression analysis.

4 Condition of HUVECs Growth
HUVECs were seeded in four growth conditions: (1) Onto a 60 cm 2 Petri dish covered with Matrigel TM at a density of 5 × 10 4 cells/cm 2 .
HUVECs were cultivated for 19 h in EGM-2.

RNA Isolation, Reverse Transcription, and Real-Time Quantitative PCR
HUVECs were grown for mRNA expression analysis on Matrigel TM , in co-culture with MSCs, as a sparsely populated monoculture (3.6 × 10 4 cells/cm 2 ) or densely populated (7 × 10 4 cells/cm 2 ) for 19 h in EGM-2. Cells were detached with trypsin/EDTA and washed free of trypsin. Cells from co-cultures were separated using a flow cell sorter MoFlo (Agilent, Santa Clara, CA, USA) as described by us previously [4].

Data Analysis
The mRNA expression data were analyzed using the online program Heatmapper [8]. The data are presented in three repeats. Clustering Method: Average linkage. Distance Measurement Method: Pearson.
Functional protein association network analysis was performed in STRING [9].

Statistical Analysis
Data are expressed as the mean ± standard deviation (SD). The statistical significance of the difference between values was determined using Student's t-tests. p < 0.05 was considered significant.

Results
In this work, we compared the classical Matrigel-based tube formation assay [1] and 2D co-culture model [4].
Endothelial cells were able to form tubular-like structures on Matrigel TM (Figure 1a). At 16 h, HUVECs on Matrigel TM already form tubular-like structures but after 24 h, these structures become unstable and disintegrate [10]. tubular-like structures that became evident by 14 h [4]. By 48 h of co-culturing with MSCs, HUVECs organized into an extensive capillary network (Figure 1a). MSCs contributed to HUVEC tube formation by the synthesis and assembly of the extracellular matrix. By 48 h of pre-culturing alone, MSCs created an organized extracellular matrix. This matrix increased HUVEC tube formation by more than 6-fold in 24 h in comparison with the coculture of HUVECs and MSCs without pre-cultivation [4].  However, HUVECs formed tubes on Matrigel TM ( Figure 1a) and in co-culture with MSCs ( Figure 1a) and these tubes differed in their form. HUVECs on Matrigel TM formed clusters of cells with a round shape (Figure 1b1) connected by "tubes" (Figure 1b2). HU-VECs in co-culture with MSCs formed a tubular network with "tip"-like cells (Figure 1b3,b4). HUVEC tubes in co-culture were in a looser state compared to the highly strained tubes on Matrigel TM . Cellular clusters on Matrigel TM (Figure 1b1) were in a more compact form compared to the HUVEC static monolayer (Figure 1b7). Proliferating/motile HUVECs in spare culture also formed some "tubes" (Figure 1b5) and occupied more space (Figure 1b6) compared to the static monolayer ( Figure 1b7).
Previously, we have shown that endothelial cells in co-culture with MSCs formed tubular-like structures that became evident by 14 h [4]. By 48 h of co-culturing with MSCs, HUVECs organized into an extensive capillary network (Figure 1a). MSCs contributed to HUVEC tube formation by the synthesis and assembly of the extracellular matrix. By 48 h of pre-culturing alone, MSCs created an organized extracellular matrix. This matrix increased HUVEC tube formation by more than 6-fold in 24 h in comparison with the co-culture of HUVECs and MSCs without pre-cultivation [4].
We chose 19 h as the optimal time point. We analyzed the mRNA expression of endothelial gene markers (Table 2) related to angiogenesis at this time point for all culture conditions: (1) HUVECs on Matrigel TM , (2) HUVECs in co-culture with MSCs, (3) sparsely populated monoculture "proliferating/motile", and (4) densely populated monoculture "static monolayer". We found that HUVECs on Matrigel TM , HUVECs from co-culture, and proliferating/motile HUVECs (sparse culture) had four common reliably (compared to monolayer HUVECs) upregulating genes ( Figure 2a): PLAU (uPA), PLAUR (uPAR), JAG1 (Table 1. ligand Jagged1), and NOTCH2 (Notch receptor Notch2) and no common reliably downregulated genes (compared to monolayer HUVECs).  Platelet-derived growth factor subunit B  (a) Venn diagram summarizes the number of significantly changed genes (p < 0.05) that are differentially upregulated or downregulated between experimental groups compared to the HUVEC static monolayer. ECs from direct co-culture were separated using anti CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". (b) Hierarchical clustering dendrograms of differentially expressed mRNA expression patterns in HUVECs on Matrigel TM , in co-culture with MSCs, and proliferating/motile HUVECs analyzed using online software Heatmapper [8].  (a) Venn diagram summarizes the number of significantly changed genes (p < 0.05) that are differentially upregulated or downregulated between experimental groups compared to the HUVEC static monolayer. ECs from direct co-culture were separated using anti CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". (b) Hierarchical clustering dendrograms of differentially expressed mRNA expression patterns in HUVECs on Matrigel TM , in co-culture with MSCs, and proliferating/motile HUVECs analyzed using online software Heatmapper [8]. The data are presented in three repeats. Hierarchical cluster analysis revealed that HUVECs from co-culture and proliferating/motile HUVECs had a similar pattern of gene expression in contrast to HUVECs on Matrigel TM (Figure 2b). Moreover, HUVECs on Matrigel TM and proliferating/motile HUVECs had a direct opposite pattern of gene expression (Figure 2c). This means that cells in these conditions undergo different processes. This data poses many questions and led us to a more precise analysis of groups of markers attributed to different processes engaged in angiogenesis.

Endothelium Activation
External stimuli such as infection, hypoxia, or injury lead to "activation" of the endothelium [11] (Figure 3a). This activation leads to the recruitment of inflammatory cells by upregulation of vascular cell adhesion molecule 1 (VCAM1), intercellular adhesion molecule 1 (ICAM1), and monocyte chemoattractant protein 1 (MCP-1; also known as CCL2).
We can see that HUVECs on Matrigel TM had an "activation" mode with VCAM-1 and ICAM-1 expression upregulation in contrast to HUVECs from co-culture and proliferating/motile HUVECs that had downregulated VCAM1 (Figure 3b). CCL2 expression was significantly downregulated in HUVECs on MSCs.
One of the angiogenic stimuli produced by ECs themselves is angiopoetin 2 (ANGPT2). ANGPT2 was upregulated only in HUVECs on Matrigel TM (Figure 3b).
The loosening of intercellular contacts leads to actin rearrangement and upregulation of TGFb, EDN-1, and CAV1, proteins associated with EndMT, thus activating a positive feedback loop. Remarkably, EDN1 was downregulated up to 25-fold in HUVECs in the Matrigel TM assay.
We did not observe a clear picture of the gene expression pattern in all three culture conditions. We can only talk about partial EndMT. However, ECs in co-culture are involved in EndMT to a higher extent.

Tip Cell
During sprouting angiogenesis, the leading (tip) cell of the growing vessel changes its morphology and undergoes partial EndMT, which enables the tip cell to be more mobile and quickly respond to external stimuli.
Tip cells are different from stalk cells, forming the body of the newly formed vascular outgrowth. Tip-cell-enriched genes include PLAU, PLAUR, JAG1, DLL4, KDR, NRP1, EFNB2, ITGB1, PDGFB, ESM1, APLN, CCN1, and YAP1, and stalk-cell-enriched genes include DLL1, JAG1, and FLT1 ( Figure 4a). Additionally, stalk cells have active Notch signaling accompanied with HES and HEY upregulation in contrast to tip cells, which have restricted Notch signaling [15][16][17].  HUVECs from the Matrigel TM assay, co-culture with MSCs, and proliferating/motile HUVECs compared to the static confluent monolayer. Cells in direct co-culture were separated using anti CD31 magnetic beads after 19 h. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of cells on Matrigel TM , in co-culture with MSCs, and proliferating/motile HUVECs compared to the mRNA levels for the same transcript in the confluent monolayer, "*": p < 0.05, n ≥ 3.

Tip Cell
During sprouting angiogenesis, the leading (tip) cell of the growing vessel changes its morphology and undergoes partial EndMT, which enables the tip cell to be more mobile and quickly respond to external stimuli.
Analysis of tip-or stalk-enriched genes did not show a clear picture. Most of tip and stalk genes were upregulated in HUVECs in the Matrigel TM assay; however, APLN, ESM1, and CCN1 were downregulated. HUVECs were grown on Matrigel TM , in co-culture with MSCs, or as proliferating/motile HUVECs or a confluent monolayer for 19 h. Cells in direct co-culture were separated using anti CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of cells on Matrigel TM , in co-culture with MSCs, and proliferating/motile HUVECs compared to the mRNA levels for the same transcript in the confluent monolayer, "*": p < 0.05, n ≥ 3.
In the co-culture model, PLAU, PLAUR, and JAG1 were upregulated and DLL1, DLL4, EFNB2, and HEY1 were downregulated. In the proliferating/motile HUVECs, PLAU, PLAUR, JAG1, PDGFB, ESM1, APLN, and CCN1 were upregulated and DLL4, KDR, NRP1, EFNB2, ITGB1, YAP1, DLL1, FLT1, HES1, and HEY1 were downregulated (Figure 4b). HUVECs were grown on Matrigel TM , in co-culture with MSCs, or as proliferating/motile HUVECs or a confluent monolayer for 19 h. Cells in direct co-culture were separated using anti CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of cells on Matrigel TM , in co-culture with MSCs, and proliferating/motile HUVECs compared to the mRNA levels for the same transcript in the confluent monolayer, "*": p < 0.05, n ≥ 3.
Analysis of tip-or stalk-enriched genes did not show a clear picture. Most of tip and stalk genes were upregulated in HUVECs in the Matrigel TM assay; however, APLN, ESM1, and CCN1 were downregulated.
In Figure 1, we can see that HUVECs formed tip-and cell-like structure only in co-culture. On Matrigel TM , cells tended to form clusters (Figure 1b).

Notch
One of the important regulators of intracellular communication and cause of angiogenesis is Notch signaling (Figure 5a) [18]. We analyzed the expression of four Notch ligands: JAG1, JAG2, DLL1, and DLL4; three receptors: NOTCH1, NOTCH2, and NOTCH4; and transcription factors whose expression is initiated by the Notch signaling pathway: HES1 and HEY1. Interestingly, in HUVECs on Matrigel TM , all genes were upregulated. This indicates active Notch signaling. In proliferating/motile HUVECs, only JAG1 and NOTCH2 were upregulated and DLL1, DLL4, NOTCH1, NOTCH4, HES1, and HEY1 were downregulated, which indicates a decrease in the Notch signaling activity. In co-cultured HUVECs, JAG1, JAG2, and NOTCH2 were upregulated and DLL1, DLL4, and HEY1 were downregulated, which means a decrease in the Notch signaling activity. We also analyzed the expression of Notch genes in MSCs co-cultured with those HUVECs. In MSCs, JAG1, JAG2, NOTCH1, NOTCH2, NOTCH3, HES1, and HEY1 were upregulated. It is worth mentioning that JAG1 and NOTCH3 were upregulated more than 6-fold and JAG2 more than 20-fold (Figure 5b). ands: JAG1, JAG2, DLL1, and DLL4; three receptors: NOTCH1, NOTCH2, and NOTCH4; and transcription factors whose expression is initiated by the Notch signaling pathway: HES1 and HEY1. Interestingly, in HUVECs on Matrigel TM , all genes were upregulated. This indicates active Notch signaling. In proliferating/motile HUVECs, only JAG1 and NOTCH2 were upregulated and DLL1, DLL4, NOTCH1, NOTCH4, HES1, and HEY1 were downregulated, which indicates a decrease in the Notch signaling activity. In cocultured HUVECs, JAG1, JAG2, and NOTCH2 were upregulated and DLL1, DLL4, and HEY1 were downregulated, which means a decrease in the Notch signaling activity. We also analyzed the expression of Notch genes in MSCs co-cultured with those HUVECs. In MSCs, JAG1, JAG2, NOTCH1, NOTCH2, NOTCH3, HES1, and HEY1 were upregulated. It is worth mentioning that JAG1 and NOTCH3 were upregulated more than 6-fold and JAG2 more than 20-fold (Figure 5b). were grown on Matrigel TM , in co-culture with MSCs, or as proliferating/motile HUVECs or a confluent monolayer for 19 h. Cells in direct co-culture were separated using anti CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of HUVECs on Matrigel TM , in co-culture with MSCs, proliferating/motile HUVECs compared to the mRNA levels for the same transcript in confluent monolayer, or to a monoculture for MSCs "*": p < 0.05, n ≥ 3.

Ephrin-Eph Signaling Pathway
Another important angiogenic regulator that also works in tandem with Notch is the Ephrin-Eph signaling pathway (Figure 6a) [19,20].
EphA receptors mainly to bind ephrinA ligands and EphB receptors to ephrinB ligands; however, some receptor-ligand interactions also occur between opposite classes (Figure 6b) [19]. There are also high binding interactions (less than 1nM) of ephrin ligands and EphB receptors within the same class [19] (Figure 6c). were grown on Matrigel TM , in co-culture with MSCs, or as proliferating/motile HUVECs or a confluent monolayer for 19 h. Cells in direct co-culture were separated using anti CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of HUVECs on Matrigel TM , in co-culture with MSCs, proliferating/motile HUVECs compared to the mRNA levels for the same transcript in confluent monolayer, or to a monoculture for MSCs "*": p < 0.05, n ≥ 3.

Ephrin-Eph Signaling Pathway
Another important angiogenic regulator that also works in tandem with Notch is the Ephrin-Eph signaling pathway (Figure 6a) [19,20].
EphA receptors mainly to bind ephrinA ligands and EphB receptors to ephrinB ligands; however, some receptor-ligand interactions also occur between opposite classes (Figure 6b) [19]. There are also high binding interactions (less than 1 nM) of ephrin ligands and EphB receptors within the same class [19] (Figure 6c).
We analyzed seven ephrin ligands: EFNA1, EFNA2, EFNA3, EFNA5, EFNB1, EFNB2, and EFNB3, and six ephrin receptors: EPHA2, EPHA4, EPHB1, EPHB2, EPHB4, and EPHB6. In HUVECs on Matrigel TM , no genes were downregulated and EFNA3, EFNA5, EFNB2, and EFNB3 were upregulated (Figure 6e). In proliferating/motile HUVECs, most genes were downregulated and only EPHA2 was upregulated (Figure 6f). In co-cultured HUVECs, EPHA2 and EFNB1 were upregulated and EFNA1, EFNB1, EFNB2, EFNB3, and EPHB4 were downregulated. In the corresponding co-cultured MSCs, EFNA1, EFNA2, EFNB1, EFNB2, EPHB1, EPHB2, and EPHB4 were upregulated and EFNA5, EFNB3, EPHA2, EPHA4, and EPHB6 were downregulated (Figure 6d). Cells 2022, 11, x FOR PEER REVIEW 16 of 28 Known interactions between EPH receptors and EFN ligands ( [21]). (c) High binding interactions, Kd < 1nM ( [19]). HUVECs were grown in co-culture with MSCs (d), on Matrigel TM (e), proliferating/motile HUVECs (f), or a confluent monolayer for 19 h. Cells in direct co-culture were separated using anti CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of HUVECs on Matrigel TM , in co-culture with MSCs, and proliferating/motile HUVECs compared to the mRNA levels for the same transcript in a confluent monolayer or to a monoculture for MSCs "*": p < 0.05, n ≥ 3. Hypothetical scheme of EPH-EFN interactions in a co-culture of HUVECs and  [21]. (c) High binding interactions, Kd <1 nM [19]. HUVECs were grown in co-culture with MSCs (d), on Matrigel TM (e), proliferating/motile HUVECs (f), or a confluent monolayer for 19 h. Cells in direct co-culture were separated using anti CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of HUVECs on Matrigel TM , in co-culture with MSCs, and proliferating/motile HUVECs compared to the mRNA levels for the same transcript in a confluent monolayer or to a monoculture for MSCs "*": p < 0.05, n ≥ 3. Hypothetical scheme of EPH-EFN interactions in a co-culture of HUVECs and MSCs (g), HUVECs on Matrigel TM (h), and proliferating/motile HUVECs (i). Bold font indicates fold of upregulation >2.3.

b) Known interactions between EPH receptors and EFN ligands
We constructed a scheme of the proposed Ephrin-Eph interaction between HUVECs and MSCs in the co-culture model (Figure 6g), between HUVECs in the Matrigel TM assay (Figure 6h), or between proliferating/motile HUVECs (Figure 6i) taking into account genes that were upregulated or changed insignificantly.
The scheme of proliferating/motile HUVECs (Figure 6i) is the simplest and represents EPHA2-EFNA2 interaction. The richest pattern of interactions corresponds to HUVECs in the Matrigel TM assay (Figure 6h).

Extracellular Matrix
Vessel growth is accompanied by extracellular matrix remodeling; the most significant step is basement membrane assembly [3].
In proliferating/motile HUVECs, extracellular matrix proteins FN1, COL4A1, LAMA4, and LAMB2 were downregulated and SDC2, SDC4, CCN1, and CCN2 were upregulated (Figure 7a). CCN1 and CCN2 upregulation may indicate turning off the Hippo pathway and this makes sense since cells lose contact and proliferate.
In HUVECs in the Matrigel TM assay, FN1, LAMA2, LAMA5, LAMB1, LAMB2, LAMC1, SDC1, SDC3, CCN3, ITGA5, ITGAV, and ITGB1 were upregulated and only CCN1, CCN2, and ITGB3 were downregulated (Figure 7a). This means that HUVECs in the Matrigel TM assay probably remodel the extracellular matrix by fibronectin synthesis and organization by integrin a5b1. The combination of upregulated laminin subunits indicates probable synthesis of basement membrane laminin 511 (a5b1c1). This poses many questions since Matrigel TM is thought to be compositionally similar to the basement membrane. CCN1 and CCN2 downregulation may indicate an active Hippo pathway and it means that cells are in the final stage of ETN formation.
Next, we analyzed the impact of the extracellular matrix assembled by MSCs. Previously, we have shown that MSCs provide the basis for extracellular matrix assembly in co-culture with HUVECs [4]. We cultivated MSCs as a monoculture for 48 h ("48 hold MSCs") to allow synthesis/secretion/assembly of the extracellular matrix. HUVECs were then seeded on top of the "48 h-old MSCs" and co-cultured for an additional 24 h. A comparison was made with the 24 h co-culture of HUVECs and MSCs ("0 hold MSCs").
Pre-synthesized EM leads to the downregulation of CDH1, EFNB2, ITGAV, and SER-PINE1 and upregulation of VWF, TIE1, DLL1, OCLN, APLN, EFNB1, ICAM1, and CCN2 (Figure 8a). , or as a confluent monolayer for 19 h. Cells in direct co-culture were separated using anti CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of HUVECs on Matrigel TM , in co-culture with MSCs, proliferating/motile HUVECs compared to the mRNA levels for the same transcript in confluent monolayer, or to a monoculture for MSCs "*": p < 0.05, n ≥ 3. (c) HUVECs were preloaded with CMFDA before being placed in direct co-culture. Cells were grown for 48 h, washed, fixed with 4% PFA, and stained with anti-fibronectin, collagen I, collagen IV, and laminin antibody conjugated with Alexa Fluor 594 (red). The nuclei were counterstained with DAPI (blue). Images were taken using an Axiovert 200 M microscope (Zeiss, Oberkochen, Germany).
co-culture. Cells were grown for 48 h, washed, fixed with 4% PFA, and stained with anti-fibronectin, collagen I, collagen IV, and laminin antibody conjugated with Alexa Fluor 594 (red). The nuclei were counterstained with DAPI (blue). Images were taken using an Axiovert 200 M microscope (Zeiss, Oberkochen, Germany).
Next, we analyzed the impact of the extracellular matrix assembled by MSCs. Previously, we have shown that MSCs provide the basis for extracellular matrix assembly in co-culture with HUVECs [4]. We cultivated MSCs as a monoculture for 48 h ("48 hold MSCs") to allow synthesis/secretion/assembly of the extracellular matrix. HUVECs were then seeded on top of the "48 h-old MSCs" and co-cultured for an additional 24 h. A comparison was made with the 24 h co-culture of HUVECs and MSCs ("0 hold MSCs").
Pre-synthesized EM leads to the downregulation of CDH1, EFNB2, ITGAV, and SER-PINE1 and upregulation of VWF, TIE1, DLL1, OCLN, APLN, EFNB1, ICAM1, and CCN2 (Figure 8a). Figure 8. Input of EM synthesized by MSCs as a monoculture on gene expression of further coculture with HUVECs. MSCs were grown as a monoculture for 48 h ("48 hold MSCs") to allow synthesis/secretion/assembly of FN fibrils. HUVECs preloaded with CMFDA (green) were then seeded on top of "48 h-old MSCs" and co-cultured for an additional 19 h. Cells in direct co-culture were separated using anti-CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of HUVECs in co-culture with "48 h-old MSCs" MSCs compared to the mRNA levels for the HUVECs in 19 h co-culture with "0 h-old MSCs". (a) Fold changes in the expression significantly altered mRNA presented as log2. p < 0.05, n ≥ 3 experiments. (b) Functional protein association networks analysis was performed in STRING [9]. MSCs were grown as a monoculture for 48 h ("48 hold MSCs") to allow synthesis/secretion/assembly of FN fibrils. HUVECs preloaded with CMFDA (green) were then seeded on top of "48 h-old MSCs" and co-cultured for an additional 19 h. Cells in direct co-culture were separated using anti-CD31 magnetic beads. Total RNA was isolated and specific mRNA levels were quantified by qRT PCR as described in "Methods". The data are presented as fold changes in the mRNA levels of HUVECs in co-culture with "48 h-old MSCs" MSCs compared to the mRNA levels for the HUVECs in 19 h co-culture with "0 h-old MSCs". (a) Fold changes in the expression significantly altered mRNA presented as log2. p < 0.05, n ≥ 3 experiments. (b) Functional protein association networks analysis was performed in STRING [9].
We analyzed these genes using functional protein association network analysis and attributed them to four groups with a strength of more than 1.15: extracellular matrix organization, regulation of cell-cell adhesion, angiogenesis, and cell-cell junction organization (Figure 8b). The upregulated gene set was enriched by angiogenesis genes.

Discussion
We found that at a time point of 19 h, the tube formation on Matrigel TM and tube formation in co-culture could be attributed to two different stages of angiogenesis. As a reference point, we chose a confluent monolayer.
Hierarchical cluster analysis revealed that HUVECs in co-culture with MSCs and dividing/migrating HUVECs had more in common than HUVECs on Matrigel TM .
We found that although most of the analyzed genes were differentially expressed, some genes (PLAU, PLAUR, JAGd1, and NOTCH2) were upregulated in all experimental settings used.
It is known that urokinase system members, uPA and uPAR, are attributed to EM remodeling and are expressed by activated endothelial cells but not in the quiescent vessels (or expressed at a low level) [23,24].
For Notch signaling, it is not all that clear-cut. In a growing vessel, in tip cells, Notch signaling is blocked while in stalk cells, it is active [25]. The suppression of Notch signaling in active angiogenesis leads to increased endothelial proliferation with excessive vascular branching [26]. Single-cell RNA sequencing in mice showed that Notch signaling is upregulated in mature capillary and arterial ECs [27]. At the same time, active Notch1 was detected by immunofluorescence in adult tissue only locally in vasculature, adipose tissue, kidney, brain, and lung [26]. Previously, we showed that inhibition of Notch signaling caused a decrease in the length of tube formation in a 2D co-culture model [4] and 3D co-culture model [28]. Here, we observed upregulation of Jagged1 and Notch2 in all three EC conditions.
The activation of Notch signaling in HUVECs in Matrigel TM assay was accompanied by upregulation of all Notch ligands and receptors. In dividing/migrating HUVECs, the downregulation of receptors and ligands was accompanied by suppression of Notch signaling. Moreover, it looks like only Jagged1, Jagged2, and Notch2 were active in intercellular communication. A similar picture was observed for co-cultured ECs, but in the co-cultured MSCs, we observed activation of Notch signaling and strong upregulation of Notch ligands and receptors, excluding Notch2 ( Figure 5).
Intracellular interactions in co-culture are thought to be carried out through NOTCH1 and Jagged1 on ECs and NOTCH2, NOTCH3, and Jagged1 on vascular smooth muscle cells [29,30]. We found that Notch signaling in co-culture was performed mostly through Notch1/Notch3 on MSCs and Jagged1/Jagged2 on ECs ( Figure 5). We can propose another important interaction through NOTCH2 on ECs and Jagged1/Jagged2 on MSCs. However, the intercellular domains of Notch1 and Notch2 are thought to be functionally equivalent [31] and the Notch2 extracellular domain appeared to be more efficient in inducing ligand-mediated receptor activation in kidney cells [32]. These data give a lot of food for thought and further experiments, especially in regard to Alagille syndrome, a dominant, multisystem disorder caused by mutations in the genes encoding Jagged1 and Notch2 [33], should be carried out.
Others showed that JAG1 was upregulated in ECs at least 3 h after cells were plated on Matrigel TM and remained upregulated at 6 h after being plated on Matrigel TM [34].
Another process that goes hand in hand with Notch signaling is EPH-EFN signaling. EPH-EFN signaling is quite complex and intricate. It involves forward signaling through EPH receptors and reverse signaling through EFN ligands [19,20].
The EPHA2 receptor was upregulated and its ligand EFNA1 was downregulated in dividing/migrating HUVECs and in HUVECs in co-culture (Figure 6d,f). The results from other studies indicate that inversely to the expression of the GPI-ligand EFNA1, its receptor EPHA2 showed the highest expression at low cell densities [35]. As EFNA1 was downregulated, the only ligand that did not change its expression was EFNA2. Therefore, in dividing/migrating ECs, we can assume interaction occurs through EFNA2-EPHA2.
Here, we observed an increase in EPHB1 expression in HUVECs in co-culture (Figure 6d). Others showed that EphB1 and EphA2 activation leads to an increase in cell migration and angiogenic sprouting [37]. These data agree with ours. However, the possibility of intracellular interaction of ECs with MSCs through ephrins requires additional research.
For HUVECs in the Matrigel TM assay, the picture was much more complicated. No downregulation of EFN ligands or EPH receptors was observed. As EFNA1 and EPHA2 did not change compared to the confluent monolayer, it appears that HUVECs on Matrigel TM reached a certain shape/structure. It is interesting that there was also upregulation of EPHB4 and EFNB2, which resembles the interactions of veins and arteries to stimulate remodeling and assembly of new vessels [38,39]. Of interest, Notch acts upstream of EPHB4/EFNB2 [26]. Figure 6g-i illustrates a hypothetical scheme of the EPH-EFN interaction of ECs in all three assays.
We observed that HUVECs formed tip cells only in co-culture with MSCs (Figure 1b3,b4). However, alterations in gene expression indicated a possible tip-stalk pattern. In dividing/migrating HUVECs, we noticed downregulation of tip and stalk genes.
Activated endothelium loosens its contacts and tip cells undergo partial EndMT, becoming more motile. Key transcriptional regulators of EndMT are Snail (SNAI1), Slug (SNAI2), Zeb1 (ZEB1), Zeb2 (ZEB2), and Twist (TWIST1). The shared characteristic of these proteins is the transcription repression of E-cadherin (gene CDH1) [12]. However, these proteins differ in their functions and their influence on each other is very complicate. In cancer, Snail can induce endothelial-to-mesenchymal transition (EMT) in epithelial cells while Zeb1/2 and Twist maintain the invasive mesenchymal phenotype. Slug is the primary initiator of sprouting angiogenesis while the induction of Snail occurs at a much later time [12]. Snail can upregulate Zeb1 and Zeb2, Slug can activate Zeb1, and Twist1 can regulate the expression level of Snail and Slug. Slug and Snail negatively regulate each another's expression [12]. Snail plays a major role in inducing EMT while Zeb1/2 and Twist are mainly involved in maintaining the invasive mesenchymal phenotype [12]. Slug-dependent pathways include regulators of cell morphology, junctional and matrix adhesions, proliferation, and TGFβ signaling [40].
ECs in co-culture with MSCs underwent EndMT to a greater extent than ECs on Matrigel TM . Surprisingly, dividing/migrating ECs did not undergo EndMT (Figure 3b).
In this context, we should not forget about the different stiffness of Matrigel TM and the plastic dish.
Polystyrene has an elastic modulus of approximately 3 GPa. In comparison, the elastic modulus experienced by cells in situ in most tissues is four to six orders of magnitude lower [41]. The average modulus value for Matrigel TM was found to be approximately 450 Pa for samples maintained at 37 • C [42]. Cells have been found to spread less on softer surfaces and those with a low density of EM protein at the surface while cells spread more on stiffer surfaces and those with a higher density of EM protein [43]. At a stiffness of 4 or 5 kPa of polymer, HUVECs spread as a monolayer. In contrast, at stiffness values between 0.5 and 1.5 kPa, HUVECs formed a tubular network comparable with that on the Matrigel TM [44]. The blocking of laminin and integrins α1, α2, α3, and α6 inhibits tube formation and cells form an endothelial monolayer [44], underling the role of laminins in tubular network formation.
Previously, we have shown that HUVECs placed on MSCs that were pre-grown for 48 h and synthesized an EM scaffold formed ETN faster than those placed into a co-culture with MSCs on plastic [4].
However, we found that extracellular matrix pre-synthesized by MSCs as a monoculture was adequate for ECs to create ETN similar to on Matrigel TM [4].
We think that in the absence of endothelial cells, MSCs do not form elements of the basement membrane as part of the extracellular matrix. The major laminin isoforms of the vascular BM are laminin-411 (α4:β1:γ1) and laminin-511 (α5:β1:γ1) [45]. According to the PCR data, MSCs contributed the most to laminin deposition (Figure 7b) but only in co-culture with ECs ( Figure 7b). As for collagen type IV, Yamamoto showed that in HUVEC or MSC monocultures, collagen type IV deposition was not detected and was deposited around capillary structures covered by MSC-derived pericytes [46]. It is supposed that in the absence of mural cells, ECs could not also form the basement membrane. Both cell types, ECs and MSCs, are involved in endothelial basement membrane assembly in co-culture.
This prompted us to switch to the idea of EC interplay with the basement membrane. In Matrigel TM assay, it represents the basement membrane matrix by itself [3] while in the co-culture model, basement membrane deposition is limited to only the cell-to-cell contact area.
It was shown that mature vessel formation accomplished with basement membrane formation leads to the arrest of division, migration, and stabilization of the vessel structure [47,48].
Other EM proteins include syndecans. The syndecan family consists of syndecan-1, syndecan-2, syndecan-2, and syndecan-4 [49]. These proteins act like molecular glue to stabilize signaling complexes on the cell surface (including integrins, growth factors, and others), acting as co-receptors to modify the activity of integrins and interact with heparin-binding domains in extracellular matrix molecules and growth factors. Syndecan-1 is a key regulator of angiogenesis and syndecan-2 impairs angiogenesis [49]. Syndecan-1 is thought to be involved in EMT, syndecan-2 in angiogenesis, syndecan-3 in mitogenic signal transduction, and syndecan-4 in the formation of focal contacts and in the shear stress-sensing complexis [50]. Our data indicate that syndecan-4 was upregulated only in ECs in the co-culture model. However, at the same time, syndecan-2, which reduces angiogenesis, was upregulated in dividing/migrating ECs and in ECs in co-culture. This controversial data needs additional experiments.
The CCN1 (CYR61) protein and the CCN2 protein (CTGF) are positive regulators of angiogenesis through the engagement of integrin αvβ3. The CCN3 (NOV) protein displays a divergent function, providing protection from aberrant excessive vessel growth [49]. Downregulation of the CCN1 and CCN2 Hippo target genes in ECs on Matrigel TM suggests activation of the Hippo signaling pathway while in dividing/migrating ECs, this pathway was probably inactive. Taking this into account, we can conclude that ECs in Matrigel TM assay at 19 h display the final stage of tube formation, with upregulation of CCN3 and downregulation of CCN1 and CCN2 (Figure 7a).
Others showed that CCN1 was downregulated in HUVECs at least 3 h after being placed on the Matrigel TM , the same time point that ECs began forming tubes [34]. For ECs in co-culture, 18 h is the time that ECs began forming tubes [4].
Therefore, we suggest that HUVECs in co-culture and proliferating/motile ECs are still in a dynamic process at 19 h. In contrast, HUVECs in the Matrigel TM assay seem to be in the final stage of ETN formation due to the fully pre-assembled basement membrane (Matrigel TM itself) ( Figure 9).
We can conclude that the mechanisms underlying ETN formation in these models are completely different due to ECs-EM interplay.
The final stage of endothelium stabilization can be monitored by endothelium activation factors, for example, ANGPT2, VCAM1, ICAM-1, and MCP1. ANGPT1 is expressed in mural cells as an agonistic ligand for the angiopoietin 1 receptor (TIE2) expressed on the surface of ECs whereas ANGPT2 is predominantly expressed as an antagonistic ligand for TIE2 and stored by ECs [51]. During inflammation or hypoxia, ECs deactivate the quiescence signal of TIE2 by expressing ANG2, which competes with ANGT1 for binding to TIE2. This leads to a loss of TIE2 signaling and activation of the endothelium by weakening endothelial cell-cell junctions and induction of the expression of the pro-inflammatory adhesion molecules, ICAM1 and VCAM1 [51]. VCAM1 and ICAM1 are also known to be upregulated in low, oscillatory flow conditions [52].
We observed upregulation of ANGPT2 and VCAM1 and ICAM-1 and the same downregulation of MCP1 in ECs on Matrigel TM (Figure 3b). In dividing/migrating ECs and in ECs in co-culture, VCAM1 was downregulated. On the one hand, this means that ECs on Matrigel TM became activated, but on the other hand, they could already complete the process of tubular network formation. Cells 2022, 11, x FOR PEER REVIEW 23 of 28 We can conclude that the mechanisms underlying ETN formation in these models are completely different due to ECs-EM interplay.
The final stage of endothelium stabilization can be monitored by endothelium activation factors, for example, ANGPT2, VCAM1, ICAM-1, and MCP1. ANGPT1 is expressed in mural cells as an agonistic ligand for the angiopoietin 1 receptor (TIE2) expressed on the surface of ECs whereas ANGPT2 is predominantly expressed as an antagonistic ligand for TIE2 and stored by ECs [51]. During inflammation or hypoxia, ECs deactivate the quiescence signal of TIE2 by expressing ANG2, which competes with ANGT1 for binding to TIE2. This leads to a loss of TIE2 signaling and activation of the endothelium by weakening endothelial cell-cell junctions and induction of the expression of the proinflammatory adhesion molecules, ICAM1 and VCAM1 [51]. VCAM1 and ICAM1 are also known to be upregulated in low, oscillatory flow conditions [52].
We observed upregulation of ANGPT2 and VCAM1 and ICAM-1 and the same downregulation of MCP1 in ECs on Matrigel TM (Figure 3b). In dividing/migrating ECs and in ECs in co-culture, VCAM1 was downregulated. On the one hand, this means that ECs on Matrigel TM became activated, but on the other hand, they could already complete the process of tubular network formation.
Other works showed an increase in VCAM1 as early as 1 h after the beginning of tube formation in the Matrigel TM assay in human microvascular endothelial cells (HMVECs) [53]. This indicates that this process is very quick.
Other arguments in support of our hypothesis regarding the Matrigel TM assay as a representation of the final stage of tissue formation are: 1. Apelin, which is thought to promote cell growth [54], was downregulated in ECs on Matrigel TM and was upregulated in dividing/migrating ECs. Other works showed an increase in VCAM1 as early as 1 h after the beginning of tube formation in the Matrigel TM assay in human microvascular endothelial cells (HMVECs) [53]. This indicates that this process is very quick.
Other arguments in support of our hypothesis regarding the Matrigel TM assay as a representation of the final stage of tissue formation are: Thirty-four years ago, Kubota et al. had already shown that ECs in tubular-like structures on Matrigel TM did not proliferate, formed tight cell-cell contacts, and took up acetylated LDL, which is a marker of differentiation for these cells, and this was not observed for cells cultured in a monolayer on either plastic or collagen I substrates [1]. Moreover, a higher cell density leads to more concentrated clusters but not to greater tube formation [56].
The above-mentioned evidence and our new data indicate that EC tube formation in the Matrigel TM assay and in a co-culture model rely on quite different processes despite being morphologically similar. We suppose that ECs in the Matrigel TM assay are trying to set up the vessel lining, spreading over the basement membrane (Matrigel TM ). ECs form tight contacts and attach to Matrigel TM through laminin receptors that prevent monolayer formation, as in the case of more rigid surfaces such as plastic dishes.
Differences in the expression levels of key genes involved in the EndMT, Hippo, and Notch auxiliary angiogenic pathways are presented in a summarizing scheme (Figure 9).
On the other hand, in the MSC co-culture model, ECs form tubular-like structures through interaction with stromal cells that alters both the cell expression profiles and causes basement membrane assembly. ECs, in this case, form ETN only in places covered by basement membrane proteins such as laminins and collagen IV.
One particularly interesting finding of our work was the absence of EndMT in dividing/migrating ECs, but at the same time, they had upregulated tip cell markers (PLAU, PLAUR, ESM1, APLN, CCN1) and decreased Notch signaling. This unexpected combination of altered genes at this time point requires additional comprehensive research.
Undoubtedly, one cell type cannot approximate the behavior of endothelial cells from other organs and tissues. In this work, we used HUVECs because of several benefits. HUVECs are the most commonly used type of endothelial cells in the published research on biomaterials. HUVECs are easy to isolate without contamination by other cell types, and umbilical cord is readily available as a discarded biological waste after childbirth. In contrast, the isolation of human microvascular endothelial cells has a number of disadvantages, such as the small isolation yield, admixture of mesenchyme-like cells, and low proliferative activity. In simple 2D in vitro models, HUVECs and microvascular endothelial cells show a similar response to external stimuli. Sieminski et al. showed that HUVECs and human dermal microvascular endothelial cells have much in common in simple in vitro models: vascular network formation, gel contraction, cell elongation, survival, and inhibition of network formation by blocking antibodies to a2b1 but not avb3 integrins [57]. Lidington et al. showed that HUVECs and the immortal human endothelial cell line HMEC-1 (human dermal microvascular endothelial cells) have a similar pattern of expression for PECAM-1, ICAM-1, and VCAM-1 [58].
However, despite all their benefits, HUVECs do not necessarily represent a universal model of endothelial cells for every application. For more complex 3D models of angiogenesis or for the production of a functional vasculature within in vitro engineered tissues, whether the endothelial cells belong to microvessels or large vessels is more crucial.
We hope that our results will provide new insights for the understanding of angiogenesis mechanisms.