Gelatin methacryloyl is a slow degrading material allowing vascularization and long-term use in vivo

In situ tissue engineering is an emerging field aiming at the generation of ready-to-use three-dimensional tissues. One solution to supply a proper vascularization of larger tissues to provide oxygen and nutrients is the arteriovenous loop (AVL) model. However, for this model, suitable scaffold materials are needed that are biocompatible/non-immunogenic, slowly degradable, and allow vascularization. Here, we investigate the suitability of the known gelatin methacryloyl (GelMA)-based hydrogel for in-situ tissue engineering utilizing the AVL model. Rat AVLs are embedded by two layers of GelMA hydrogel in an inert PTFE chamber and implanted in the groin. Constructs were explanted after 2 or 4 weeks and analyzed. For this purpose, gross morphological, histological, and multiphoton microscopic analysis were performed. Immune response was analyzed based on anti-CD68 and anti-CD163 staining of immune cells. The occurrence of matrix degradation was assayed by anti-MMP3 staining. Vascularization was analyzed by anti-α-smooth muscle actin staining, multiphoton microscopy, as well as expression analysis of 53 angiogenesis-related proteins utilizing a proteome profiler angiogenesis array kit. Here we show that GelMA hydrogels are stable for at least 4 weeks in the rat AVL model. Furthermore, our data indicate that GelMA hydrogels are biocompatible. Finally, we provide evidence that GelMA hydrogels in the AVL model allow connective tissue formation, as well as vascularization, introducing multiphoton microscopy as a new methodology to visualize neovessel formation originating from the AVL. GelMA is a suitable material for in situ and in vivo TE in the AVL model.


Introduction
Large volume tissue defects or organ failure are treated with autologous tissue or allogenic organs such as the kidney or heart, respectively. Although the technical requirements are well established in reconstructive surgery and transplantation medicine, the required organs or tissues can cause problems due to donor site morbidity, tissue availability, and human leukocyte antigen matching. The generation of bioartificial tissues using the principles of tissue engineering (TE) is a promising approach to solving the aforementioned restrictions.
In situ TE is characterized as biomaterial-induced endogenous tissue formation directly at the site of injury or in situ, beginning with readily available resorbable grafts that are progressively transformed into autologous homeostatic replacement tissue with the ability to repair, reshape and grow. Grafts for in situ TE can be natural or synthetic; however, the prerequisite for this approach is first the recruitment of host cells in the transplantation site and its remodeling in order to achieve adaptive autologous tissue over time and second the biomaterial characteristics such as morphology, biochemical and biophysical properties which must be adapted to the tissue to be replaced [1,2]. For example, bioartificial bone tissue requires a matrix with solid parts [3][4][5], while fat tissue requires a rather soft matrix [6]. Furthermore, matrix degradation must also be adapted to the area of application so that the optimal degradation of scaffold materials corresponds to the formation of the required tissue. Another point to consider when choosing a matrix is biocompatibility. Some matrices have already been tested for biocompatibility, such as alginate di-aldehyde-gelatin (ADA-GEL) [7] and polycaprolactone [8], while some materials are already FDA approved and are in medical use, such as fibrin [9] and polymethyl methacrylate [9].
Another new potential field of in situ TE application is the establishment of a drug-producing tissue container. Therefore, an inert container, being used over a longer period (months or years), is implanted subcutaneously. Drug-producing cells are encapsulated into a non-degrading or slowly degrading matrix. The encapsulated cells can produce biological molecules, such as antibodies against tumor necrosis factor receptor II in the case of rheumatoid arthritis in vivo after being directly transferred into the organism. Therefore, the third prerequisite for in situ TE is applying a suitable carrier matrix that is stable for a long time, maintaining the concentration of active substances for years through continuous secretion. Such a system will have a potential application in cancer therapy or the treatment of autoimmune diseases.
Gelatin, a protein-based biopolymer obtained from the hydrolytic degradation of collagen, has shown high biocompatibility and angiogenic potential in situ as well as in vivo [10][11][12][13]. Gelatin is produced from different types of collagen (hydrolyzed collagen), which is easily degradable and does not cause a severe immune reaction, as can be the case with collagen [14,15]. Structurally, gelatin also contains integrin-binding motif (RGD) sequences improving cell adhesion. RGD motifs are also known to stimulate angiogenesis [16]. Furthermore, morphogenesis, wound healing and angiogenesis is supported by target sequences for matrix metallopeptidases (MMPs) contained in gelatin [17]. MMPs are jointly responsible for the degradation and rearrangement of gelatin [18]. The degradation of the matrix is quite fast with MMPs and proteases. Biodegradation of gelatin can be adjusted via a modification with methyl acrylate resulting in the production of gelatin methacryloyl (GelMA) [19]. The degree of methacrylation can be controlled during the synthesis procedure, and it has been shown that small amounts of MA group to gelatin <2 ml g −1 does not affect the encapsulated cells while preserving the physical and chemical properties show a high cytocompatibility [20]. The amino groups are not affected, and consequently, the RGD sequence remains intact within the gelatin without reacting with MA. Furthermore, GelMA can be optically crosslinked using a photoinitiator system, triggering the formation of free radicals, polymerizing the different methacrylamide, and methacrylate groups within the gelatin, enhancing the stability of the polymer network. Therefore, after preparing the GelMA solution at 37 • C in a liquid state, it can be easily used to fill any cavity before being crosslinked [21].
In order to ensure cell survival in vivo, adequate vascularization is required. Improvement of vascularization can be achieved by, e.g. growth factors such as vascular endothelial growth factor (VEGF) [22], endothelial cells, or surgically induced angiogenesis [23,24]. The latter is a powerful tool and can be applied in the form of an arteriovenous loop (AVL) model. An AVL between a femoral vein and an artery is micro-surgically anastomosed using a venous interponate. The matrix for in situ TE in the container can thus be tested regarding stability and angiogenic properties. Previous studies with materials like engineered spider silk [25], ADA-GEL and fibrin have already shown that angiogenesis begins after about 14 d [10,22].
This study targets to prove that GelMA is a biocompatible matrix with angiogenic properties and long-term stability in the rat AVL model.

GelMA production
GelMA was synthesized according to the protocol provided in Loessner et al [19]. Briefly, 6 g of Gelatin A (Porcine, Sigma-Aldrich, Missouri, USA) was dissolved in 50 ml of Dulbecco's Phosphate Buffered Saline (DPBS, Sigma-Aldrich) at 50 • C, followed by the addition of 12 ml of Methacrylic anhydride (MA, Sigma-Aldrich) stirred for 1 h. Thereafter, methacrylation process was terminated by adding warm DPBS. Subsequently, the solution was dialyzed at 37 • C for a week and lyophilized for another week. The final product was stored at −20 • C. To prepare a 10 wt% GelMA hydrogel as a matrix for AVL, the lyophilized GelMA was dissolved in sterilized DPBS at 37 • C while being continuously stirred for 30 min. Lithium phenyl-2,4,6trimethylbenzoylphosphinate (LAP, Sigma-Aldrich) was added to the dissolved GelMA with a concentration of 1 wt%, and the solution was protected from light. The gel was prepared freshly for each operation.

Degradation assay
A polytetrafluoroethylene (PTFE) chamber with a diameter of 10 mm and a height of 6 mm with 471 µl volume was filled with sterilized warm GelMA 10 wt% solution containing LAP as a photoinitiator and crosslinked for 30 s using a handheld UV-lamp (395-400 nm; 80-150 mcd) (EFL41UV UV, Perel, Gavere, Belgium). The chambers were then incubated in DPBS and DPBS containing 1.75 µg ml −1 collagenase (Sigma-Aldrich) in a humidified atmosphere (37 • C, 95% relative humidity, 5% CO 2 ) with a change of media three times a week. The weight of the chambers containing GelMA was taken over 14 d, where the empty chamber weight was subtracted and the degradation calculated via the mass loss.

Arteriovenous loop surgery
Ten male Lewis rats (Charles River Laboratories, Sulzfeld, Germany) with a weight range from 330 to 380 g underwent surgery. The surgeries were approved by the Animal Care Committee of the University of Erlangen and the Government of Mittelfranken (AZ 55.2-2532-2-763). The surgery was carried out under isoflurane general anesthesia (cp-pharma, Burgdorf, Germany) by one surgeon using an operative surgical microscope (Carl Zeiss, Oberkochen, Germany). A 2-3 cm incision in both groins was made, the femoral vessels were dissected, and a vein graft was gathered from the right leg. For the AVL, the vein graft was interposed between the femoral artery and vein on the left side with an 11-0 non-resorbable suture (Ethilon, Ethicon, New Jersey, USA) (figure 1). A PTFE chamber was half-filled with sterilized warm GelMA 10 wt% solution containing LAP and crosslinked as described before. After crosslinking a GelMA layer with a thickness of about 2.5 mm, the AVL was placed on top of it, and another layer of GelMA was cast and crosslinked. In total, approximately one milliliter of GelMA solution was used for each animal. Afterward, the chamber was fixed onto the thigh muscle, closed with a lid, and the skin was closed. Postoperatively, the animals received enoxaparin (10 mg kg −1 ) for 2 d. Five constructs were explanted 2 weeks after surgery and the other five after 4 weeks.

Explantation procedure
Vascularization of the constructs was visualized after intra-arterial perfusion with India ink. For this, a longitudinal laparotomy was performed, the descending aorta cannulated, and the inferior caval vein cut. The vascular system was flushed with a Ringer-Heparin solution (100 IU ml −1 ). Then, 20 ml of India ink solution, containing 50% (v/v) India ink (Lefranc-Bourgeois, London, England) in 5% gelatin (Carl Roth, Karlsruhe, Germany) and 4% mannit (Carl Roth) was applied into the aorta. The descending vein and artery were ligated, and the specimens were placed at −20 • C for 2 h before explantation. The constructs were cut in half, one half embedded in Roti ® -Histofix 4% (Carl Roth) for histological analysis and the other half frozen at −80 • C for proteome analysis.

Fluorescein isothiocyanate (FITC) perfusion and explantation
For multiphoton microscopy, 0.8 ml of a 5% fluorescein isothiocyanate (FITC) solution (Sigma-Aldrich) was injected intravenously. After 10 min, the vessels entering the PTFE chamber were ligated, and the chamber was explanted. The animal was euthanized and the construct was fixed in 4% paraformaldehyde (Sigma-Aldrich) for 4 h. Samples were dehydrated and optically cleared according to an established protocol [26] and stored in ethyl cinnamate (Sigma-Aldrich) under the absence of light until multiphoton microscopy.

Multiphoton microscopy
The samples were imaged while immersed in a clearing solution and contained in a polycarbonate chamber that was covered with a glass coverslip. An upright multiphoton microscope system (TriMScope II, LaVision BioTec, Bielefeld, Germany, described in [27]) with a Nikon Plan Fluor 10×/0.3NA objective was used. Images were recorded using an excitation wavelength of 810 nm and detecting the emission wavelengths 525 nm (525/50) and 450 nm (450/70). The images voxel size was 2.2 × 2.2 × 8 µm 3 with a 1.1 × 1.1 m 2 field of view. The images were acquired as 3D mosaic with 10% overlap and stitched using ImageJ to cover the whole sample volume

Histological staining and analysis
The constructs were embedded in paraffin, and 3 µm cross-sections perpendicular to the longitudinal axis of the AVL were cut with a microtome. Hematoxylin and eosin (H&E) and α smooth muscle actin (α-SMA) staining were carried out according to standard protocols. For macrophage detection, CD68 staining was carried out. Dewaxed sections were first treated with blocking solution (Zytomed Systems GmbH, Berlin, Germany) and then incubated with the primary anti-CD68 (BIO-RAD, Hercules, USA) antibody in a dilution of 1:300. For enzymatic detection, an alkaline phosphatase-labeled anti-mouse antibody (AP-Polymer) and Fast Red TR/Naphthol AS (Sigma-Aldrich) substrate were added for color reaction. Haemalaun was used for counterstaining. In order to further differentiate between pro-inflammatory and anti-inflammatory macrophages, CD86 (Abcam, Cambridge, USA) and CD163 (Leica Biosystems Inc., Illinois, USA) immunostainings were performed. The cross-sections were boiled before blocking and otherwise treated as described for CD68. The primary antibodies were diluted (1: 250 for CD86 and 1:500 for CD163) prior to application.
To highlight the matrix degradation of GelMA, staining for MMP3 was made. The histochemical staining of the deparaffinized cross-sections was performed with the anti-MMP3 antibody (1:50, Sigma-Aldrich) after incubation in citrate buffer (pH value: 6, at 121 • C, 1 min). After a first washing step, MMP3 was visualized using a second peroxidaselabeled antibody (Dako REAL EnVision Detection System, Agilent Technologies, Kalifornien, USA) and DAB substrate (3,3´-diaminobenzidine, Dako, California, USA).
The histological cross-sections were microphotographed with an Olympus IX81 microscope (Olympus, Hamburg, Germany). For obtaining highresolution overview images, 40-fold magnification images were merged together using the cellSens Dimension V1.5 Software of Olympus.
The construct and connective tissue area were analyzed by outlining the corresponding area with the intelligent-scissor-tool of GIMP 2.10 (GNU Image Manipulation Program) and measuring the enclosed area with ImageJ 1.52p (NIH, Bethesda, Maryland, USA). The vessels were detected based on their morphology in the α-SMA staining, counting α-SMApositive vessels with a clearly visible lumen or lumens filled with India ink. The vessel distribution was calculated using the distance to the closer main loop vessel measured with ImageJ.

Proteome analysis
Four frozen samples of the 4-week group and four untreated femoral veins were lysed with Triton X-100 (1%) and protease inhibitors (10 µl ml −1 Aprotinin, Sigma-Aldrich, 10 µl ml −1 Leupeptin and 10 µl ml −1 Pepstatin, both Bio-Techne GmbH, Wiesbaden, Germany) and centrifuged at 10 000 × g for 5 min. The proteome assay was performed according to the manufacturer's protocol (Proteome Profiler Mouse Angiogenesis Array Kit Bio-Techne GmbH). Briefly, the films were incubated with the lysed samples, blocked, and incubated with an antibody cocktail that specifically detects 53 angiogenesis-related proteins. Thereafter, streptavidin and a chemiluminescent detection mix consisting of hydrogen peroxide and luminol were applied, and images were taken with iBright (Thermo Fisher Scientific, Massachusetts, USA). The images were analyzed with the Protein Array Analyzer plugin for ImageJ. Thereby, the measured mean pixel density was normalized to the reference spots on each film. A heatmap of all analyzed and normalized protein values was created with R (PBC, Boston, USA).

Statistical analysis
Statistical analysis was performed with GraphPad Prism 8.00 (GraphPad Software, California, USA). The normal distribution of the samples was assessed with a Shapiro-Wilk test. An unpaired Student's ttest was used, and p-values ⩽0.05 were considered statistically significant. Data are shown as mean values ± standard deviation. A Wilcoxon test was done for the proteome analysis, and p-values ⩽0.05 were considered statistically significant.

GelMA hydrogels in the AVL model is stable for 4 weeks
In order to determine whether GelMA hydrogels are suitable for long-term in situ TE, the AVL model was utilized by filling the PTFE chamber with GelMA. All animals utilized in this study survived the operation. Furthermore, no surgical side effects or postoperative complications such as wound healing disorder or dislocation of the chamber occurred. Macroscopic analysis of the explanted in situ engineered tissues at 2 and 4 weeks after implantation revealed that all explants were in a stable form (figure 2). The GelMA hydrogel remained translucent, and the loop was clearly visible in the middle of all constructs. No gross-morphological difference was observed between samples at 2 and 4 weeks postimplantation (figures 2(A) and (B)). In accordance with this observation, the weight of the constructs at 2 and 4 weeks post-implantation was not significantly different (0.65 ± 0.04 g vs. 0.63 ± 0.09 g; n = 5 in each group) (figure 2(C)). These data suggest that GelMA hydrogels are stable under in vivo conditions in the AVL model. In the degradation assay, the GelMA was shown to degrade by the collagenase completely after 14 d (figure 2(D)). We measured a mass loss in the first 7 d of 65.50% and 100% until day 14. Whereas the control in PBS remained stable after an initial mass loss. With a weight loss of 19.56% (ns) after 7 d and 18.70% after 14 d (ns). The comparison with the degradation of the construct in vivo shows that the gel loses mass of 3.73% from 2 to 4 weeks.

GelMA hydrogels are biocompatible
In order to assess whether GelMA is biocompatible or induces an immune response, the presence of CD68-positive multinuclear giant cells as a sign of severe immunoreaction was determined. Anti-CD68 staining revealed only single mononuclear cells (macrophages) in both experimental groups, and no multinuclear giant cells were detected. There was no significant difference between the two groups regarding the number of macrophages per mm 2 (figure 3(E)). The macrophages tended to be located more peripheral in the interface between the newly formed tissue and the GelMA matrix (figures 3(A)-(D)).
Only a few pro-inflammatory (CD86, M1) and anti-inflammatory (CD163, M2) macrophages were detected in the corresponding staining, without a specific distribution pattern or predominance of one subtype (figures 4 and 3(F)). These data suggest that GelMA hydrogels are biocompatible.

GelMA hydrogels in the AVL model allow connective tissue formation
In order to assess the suitability of GelMA to promote tissue formation in the AVL model, we analyzed based on H&E staining the formation of connective tissue surrounding the AVL (figure 5). The histological evaluation revealed that all AVL were patent (10/10). In addition, most of the GelMA matrix was visually present after both 2 and 4 weeks post-implantation. Quantifying the cross-sections did not show any statistically significant differences between the two time points (29.53 ± 4.76 mm 2 vs. 33.01 ± 3.13 mm 2 ; n = 5 in each group; ns) ( figure 5(E)). In contrast, the amount of newly formed connective tissue was different. An area of 2.59% (0.76 ± 0.12 mm 2 ) and 3.40% (1.09 ± 0.31 mm 2 ) of the construct area was found at 2 and 4 weeks post-implantation, respectively (figure 5(F)).
Notably, we detected in both experimental groups MMP3 expression, a marker for matrix degradation, predominantly located in the transition zone between the newly formed connective tissue and the GelMA matrix as well as in the proximity of the newly formed vessels ( figure 6). Taken together, our data indicate that GelMA hydrogels are suitable for in situ TE as they allow the formation of novel tissues.

GelMA hydrogels allow the formation of vascularized tissue in the AVL model
One major issue in situ TE is the efficient vascularization of the newly formed tissue. Therefore, the explants were analyzed for new vessel formation surrounding the main AVL vessel (figure 7).
Newly formed vessels originating from the AVL were detected in both groups. Statistically, significant more vessels were identified after 4 weeks compared to the 2 weeks group (16 ± 9 vs. 4 ± 7; p ⩽ 0.05).
Most new vessels were located in the proximity of the AVL with a mean distance of 0.23 mm (figures 7(A) and (B)). The newly formed vessels were positive for α-SMA in the media layer (figures 7(C) and (D)). In addition to that, India ink was only intravascularly located as a sign for certain vessel patency and maturity, respectively. The evaluation of the lumen area of all vessels in the constructs showed no significant difference between 2 and 4 weeks group (0.06 ± 0.04 mm 2 vs. 0.08 ± 0.06 mm 2 ).
The formation of new vessels was further underlined by an increased expression of pro-angiogenic factors in the vein graft of the explants 4 weeks post-implantation compared to untreated femoral vein grafts explanted from untreated rats. The analysis of the expression of 53 angiogenesis-related proteins after setting the cutoff at 1.5 revealed that 28 proteins were upregulated, 22 were unchanged, and three were downregulated compared to the control (table 1).
The data is shown as absolute values plotted in a heatmap. The overview shows that the values of vein and loop are clearly clustered into two groups (figure 8), with the vein values being low and the loop values high. However, the individual values partly show a high variability within the groups of vein and loop. It is also possible to identify a clustered protein group where the vein expression values are at a similar level and upregulated to a similar extent after the implantation. This can be observed especially in the cluster starting with metallopeptidase with thrombospondin type 1 motif 1 (ADAMTS1) to serpin F1 (PEDF) ( figure 8).
Overall, of the 31 proteins with a cutoff at 1.5, 13 angiogenesis-relevant proteins were significantly upregulated compared to the control group, and one protein was significantly downregulated (table 2). The proteins with significant changes can be divided into different groups. These comprise paracrine acting molecules such as growth factors (HGF and IGFBPs) and cytokines (MCP-1 and SDF-1). Furthermore, extracellular matrix associated proteins (PAI1, THBS2, SPP1, SERPINEF1 and Endostatin) and proteolytic enzymes (ADAMTS1 and CD26) were upregulated in the loop group (table 2).
To validate the formation of new vessels sprouting from the AVL, FITC was perfused through the constructs, which were then analyzed by multiphoton microscopy ( figure 9). This technique allowed the detection of vessels with a diameter between 13 and 265 µm, and the analysis showed that the microvascular network was stained with FITC. This data indicates that at least the majority of newly formed vessels originated from the AVL.
Collectively, GelMA hydrogels allow the formation of mature vessels, further underlining their suitability for in situ TE.

Discussion
The in situ TE approach is based on the assumption that a temporary microenvironment is created by the resorbable immunomodulatory scaffold, which acts as an instructive road map for endogenous cells to invade and build fresh, living, and usable tissue. It is hypothesized that the scaffold and biocompatible biomaterial offers help for the development of mature tissue and sufficient mechanical properties to withstand hemodynamic loads upon implantation. The biomaterial should slowly degrade over time, eventually leading to a purely biological structure that has the capacity to repair, remodel and expand [1]. Proof-of-concept has been shown in our previous studies in the formation of small-diameter blood vessels and angiogenesis within the AVL using acellular scaffolds such as ADA-GEL and fibrin [10,29]. In this regard, hydrogels have thereby gained more and more importance. Hydrogels offer the advantage of a scaffold similar to an extracellular matrix with high porosity, a 3D network, and very high water content of at least 95% [30]. Hydrogels offer the possibility of tunable mechanical properties and a high volume to the surface ratio for safe cell encapsulation in biofabrication and TE applications. In addition, biocompatibility can also be enhanced by bioactive peptides, such as RGD motifs, thus leading to better adhesion, proliferation, and migration of encapsulated cells [31,32]. However, the matrix should be not only biocompatible but also immunocompatible without causing any severe immune reaction. Materials such as ADA-GEL and fibrin have already been tested in this respect, demonstrating good biocompatibility [10,29]. Fibrin is an interesting biomaterial already used in clinical practice [9,33,34]. Although fibrin and ADA-GEL have shown promising material characteristics with regard to cell transplantation and angiogenesis, the fast biodegradation within 4 weeks displays a major disadvantage [10,35]. For a long-term application, such as a drug-producing tissue container or bone TE, biomaterials supporting angiogenesis with slow biodegradation (from months to years) seem to be favorable. With regard to these properties, GelMA was investigated in this study as a hydrogel-based biomaterial that has already been tested in vitro for biocompatibility [36][37][38].
The GelMA crosslinked hydrogel was well tolerated by the animals without any complications. We were able to show that the constructs were stable even after 4 weeks, as indicated by the construct size and weight. The GelMA incubated in PBS showed a slow degradation rate, which reached 18.70% of mass loss after 14 d of incubation. However, by adding 1.75 µg ml −1 collagenase degradation rate increased, but still, it was only fully degraded after 14 d of incubation. This is slightly slower than previous results of degradation assays performed with 5% and 7% (w/v) GelMA, which can be explained by the higher percentage of the present gel [21]. Taking the degradation rate of GelMA in vivo into account the degradation with 3.73% from 2 to 4 weeks is more similar to the degradation rate of GelMA in vitro in PBS. It must be taken into consideration that the implantation of the gel in vivo is a closed system of the chamber with a lid. The only direct interaction between the matrix and the organism is achieved by means of the AVL. The AVL does not seem to deliver degradation products such as collagenase to the gel to the same extent as in the performed in vitro study. Therefore, it can be assumed that the degradation of the matrix is equal to new tissue formation in vivo. Comparing the cross-sectional area as a surrogate parameter for biodegradation, GelMA displayed very slow biodegradation. In former AVL studies using ADA-GEL or fibrin, significantly smaller cross-sectional areas were assessed with 14.6 or 5 mm 2 , respectively [10,39]. Especially considering that there was a substantial 75% reduction in construct size with fibrin (20 mm 2 after 2 weeks and 5 mm 2 after 4) [39]. Besides the well-known MMPs, like MMP1 and MMP2. MMP3 is also involved in the degradation of gelatin [40]. Using immune-histology, we were able to demonstrate MMP3 was involved in the biodegradation of GelMA. With increasing implantation time, we found an accumulation of MMP3 in the layer between GelMA and connective tissue. The slow biodegradation is a positive aspect of this hydrogel, especially with regard to long-term stability, which was demonstrated by the very low weight reduction of the constructs (0.65 ± 0.04 g after 2 weeks vs. 0.63 ± 0.09 g after 4 weeks).
With regard to angiogenesis, we found newly formed vessels originating from the AVL in both experimental groups. The number of newly formed vessels increased over time from 4 ± 7 per histological cross-section after 2 weeks to 16 ± 9 after 4 weeks. In comparison with other materials, such as recombinant electrospun spider silk or fibrin, with more newly formed blood vessels after 4 weeks (99 ± 52 and 107 ± 25), it is noticeable that the proportion was significantly lower in the GelMA [25,39]. In this respect, GelMA is more comparable with ADA-GEL with 28 ± 5 vs. 31 ± 24 newly formed blood vessels [10]. Nevertheless, the proangiogenic properties of the material have been shown in the production of proangiogenic signal molecules. Proteins with an influence on early stages of vascularization, such as serpin E1 (21.69-fold), osteopontin (8.03-fold), insulin-like growth factor binding proteins (IGFBP), and fibroblast growth factor (FGF) were significantly upregulated after 4 weeks [41][42][43][44]. Consistent with the accumulation of macrophages in the vascularized tissue parts, a significant upregulation of the signal molecule monocyte chemoattractant protein 1 (MCP-1) of 36.74 was detected [45]. In accordance     an important role in the first 2 weeks [29,39]. Among the downregulated proteins, there is stromal cellderived factor 1 (SDF-1), tissue inhibitor of metalloproteinase 1 (TIMP-1), and epidermal growth factor (EGF). TIMPs are known for their ability to inhibit MMPs, which in turn are responsible for the degradation of various extracellular matrix proteins [49]. EGF is a protein that, in combination with insulinlike growth factor-II, can contribute to increased angiogenesis in vivo [50]. The significantly downregulated SDF-1 (−1.64-fold) is a chemokine that plays a role in regeneration and angiogenesis in a paracrine manner. In hypoxia, the expression of SDF-1 is regulated by the transcription factor hypoxia-inducible factor-1 (HIF-1), which promotes the migration of cells in ischemic tissue [51]. The heatmap shows a good overview of the expression of the proteins, and it can be seen that the cluster with similar low values in the vein group and high values in the loop group are the proteins that have a significant change. For most of the other proteins, however, it can be observed that they show high variability. This variability may be due to the fact that the samples are from different individuals, and it cannot be ruled out that the animals react differently and that therefore different results are obtained in the proteome analysis. In future studies, it may be appropriate to focus on the proteins that are significantly different and investigate them in further detail. Multiphoton microscopy was used to visualize the AVL and neovessel formation originating from the latter one. Vessels with a diameter ranging from 13 µm to 265 µm were recorded with this technique. This methodology is new in the field of AVL studies and in situ TE and represents an advancement to describe tissue formation and vascularization more precisely. Using immune-histology, bioartificial tissues generated in the AVL model can be visualized, and the proximity to the newly formed vessels might be investigated. Also, the migration of previously fluorescence-labeled (e.g. GFP) cells through the scaffold towards the vascularized parts can be described.
In summary, with GelMA, we have found a biocompatible matrix suitable for long-term in vivo use due to its stability and due to its immunomodulatory properties allow for the recruitment of host cells and angiogenesis within the 3D construct. Another interesting approach might be to modify GelMA to promote angiogenesis further [21]. This can be performed using a lower concentration of GelMA as well as 3D bioprinting microchannels within the construct to enhance the perfusion of the nutrition and oxygen supply for earlier vascularization.

Conclusion
GelMA was successfully implanted in the rat AVL model. The constructs were found to be stable for 4 weeks. The degradation of GelMA is suitable for long-term implantation and in situ vascularization. The constructs could be examined by multiphoton microscopy with regard to angiogenesis, and the angiogenic potential could be demonstrated by proteomic analysis. In summary, GelMA offers a good approach for further in vivo implantations and in situ vascularization in the AVL model.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.