Engineering elastic sealants based on gelatin and elastin‐like polypeptides for endovascular anastomosis

Abstract Cerebrovascular ischemia from intracranial atherosclerosis remains difficult to treat. Although current revascularization procedures, including intraluminal stents and extracranial to intracranial bypass, have shown some benefit, they suffer from perioperative and postoperative morbidity. To address these limitations, here we developed a novel approach that involves gluing of arteries and subsequent transmural anastomosis from the healthy donor into the ischemic recipient. This approach required an elastic vascular sealant with distinct mechanical properties and adhesion to facilitate anastomosis. We engineered two hydrogel‐based glues: an elastic composite hydrogel based on methacryloyl elastin‐like polypeptide (mELP) combined with gelatin methacryloyl (GelMA) and a stiff glue based on pure GelMA. Two formulations with distinct mechanical characteristics were necessary to achieve stable anastomosis. The elastic GelMA/mELP composite glue attained desirable mechanical properties (elastic modulus: 288 ± 19 kPa, extensibility: 34.5 ± 13.4%) and adhesion (shear strength: 26.7 ± 5.4 kPa) to the blood vessel, while the pure GelMA glue exhibited superior adhesion (shear strength: 49.4 ± 7.0 kPa) at the cost of increased stiffness (elastic modulus: 581 ± 51 kPa) and reduced extensibility (13.6 ± 2.5%). The in vitro biocompatibility tests confirmed that the glues were not cytotoxic and were biodegradable. In addition, an ex vivo porcine anastomosis model showed high arterial burst pressure resistance of 34.0 ± 7.5 kPa, which is well over normal (16 kPa), elevated (17.3 kPa), and hypertensive crisis (24 kPa) systolic blood pressures in humans. Finally, an in vivo swine model was used to assess the feasibility of using the newly developed two‐glue system for an endovascular anastomosis. X‐ray imaging confirmed that the anastomosis was made successfully without postoperative bleeding complications and the procedure was well tolerated. In the future, more studies are required to evaluate the performance of the developed sealants under various temperature and humidity ranges.


| INTRODUCTION
When blood flow fails to satisfy metabolic demand, resultant ischemia leads to eventual cell death. The most frequent clinical scenario demonstrating this pathophysiology is arterial stenosis upstream of the tissue capillary bed due to atherosclerosis (lipid deposition into the vessel lumen). Any organ supplied by such an arteriovenous circuit is susceptible, as evidenced by common ailments like stroke, myocardial infarction, and distal extremity gangrene. Current revascularization procedures, including intraluminal stents and artery-to-artery bypass, have shown some benefits, but are not without limitations. [1][2][3] Cerebral ischemia from intracranial atherosclerosis, in particular, remains difficult to treat. Stenting a diseased cerebral artery risks apposing plaque against small perforating branches and occluding them. In addition, intracranial bypass is technically challenging and carries high preoperative morbidity, in part due to infarcts occurring while the recipient vessel is clamped and the anastomosis sutured in. To overcome these limitations, one treatment option is to use a bypass approach whereby clamp time can be eliminated.
Identifying the optimal medical device to secure successful anastomosis requires a detailed evaluation of the required characteristics.
Wound closure devices are utilized as physical barriers in order to decrease the tension on wound edges and to hold them together for a better sealing. 4,5 Proper sealing is essential to prevent postoperative complications such as infection and dehiscence. 6,7 The ideal wound closure devices should maintain their structural integrity, mechanical strength, and tissue adhesion under applied stress and/or changes in temperature and humidity. 8,9 In addition, biocompatibility and biodegradability are additional important characteristics of wound closure devices. 10 Sutures, staples, strips, and hydrogel-based sealants are examples of medical devices that have been used so far for wound closure with distinct advantages and disadvantages. 5,11 Despite the emergence and rapid growth of hydrogel-based surgical sealants since early 2000s, sutures and staples are projected to continue dominating the wound closure market for the upcoming decade. This is largely due to their relatively lower cost and ease of use as compared to the hydrogel-based sealants. However, sutures and staples are not suitable for all types of tissues or applications. For example, they can cause tissue damage upon insertion, leading to infection and further complications. Furthermore, internal organs such as the lungs, bladder, and blood vessels undergo changes in volume and pressure throughout their normal functions. Sutures and staples are not compatible with these organs as they may limit the natural tissue movement and function, and cause stress (damage) to the tissues. 6,12 To overcome these limitations, several formulations of sealants have been developed, specifically for vascular applications, such as poly(glycerol sebacate) acrylate based sealant, SETALIUM, and bovine serum albumin-glutaraldehyde-based sealant, BioGlue. These sealants have shown promising results in carotid artery and jugular vein defects in a porcine model without thrombus formation or stenosis. 13,14 In addition, there are a number of fibrin-based sealants used to strengthen vascular suture lines. 15 Although these sealants are designed to promote coagulation while maintaining biocompatibility and biodegradability, they suffer from low stiffness and elasticity and limited adhesion to the wet biological surfaces.
To address the above-mentioned limitations, in this study, we developed a new approach for endovascular anastomosis based on using two types of surgical sealants with high tissue adherence and specific mechanical properties to tolerate high pressure and/or stress.
The first glue was composed of gelatin methacryloyl (GelMA), a functionalized derivative of collagen, and the second sealant was made of GelMA and methacryloyl elastin-like polypeptide (mELP), a 365-amino acid (aa) recombinant elastomer designed to mimic the properties of natural elastin. In our previous work, we formed highly elastic ELP hydrogels through disulfide bond formation upon exposure to UV light. 16 In this study, we functionalized the lysine, serine, and tyrosine residues on ELP with methacrylamide and methacrylate groups, respectively, to obtain mELP, an elastomer capable of forming a stable and elastic photocrosslinkable hydrogel standalone upon exposure to LED light at 450-500 nm wavelength. However, pure mELP-based hydrogels had insufficient mechanical characteristics to serve as a surgical sealant. Therefore, we introduced a photocrosslinkable secondary hydrogel (GelMA) in order to provide the necessary stiffness for endovascular anastomosis. The mechanical properties of the engineered photocrosslinkable composite glue were optimized to mimic the native vascular tissue. In addition, its adhesive properties were optimized to obtain high adhesion to vascular tissue while retaining biocompatibility in vitro. The mechanical properties and adhesion of the sealants were evaluated through experimental protocols established by the American Society for Testing and Materials (ASTM). Finally, an ex vivo experiment and two in vivo tests using rat and swine models were performed to evaluate the vascular retention and adhesion of the developed system for the proposed application of anastomosis in cerebrovascular ischemia.
2 | RESULTS AND DISCUSSION 2.1 | Synthesis and structural characterization of GelMA/mELP hydrogels In this study, we developed new formulations of elastic sealants to be used in an endovascular anastomosis procedure for the treatment of cerebrovascular ischemia. These hydrogel-based sealants were engineered using two modified biopolymers: mELP (Figure 1(a)) and GelMA (Figure 1(b)). mELP is a photocrosslinkable, recombinant elastomer produced by genetically modified Escherichia coli; it serves as an elastic peptide that provides penetrability and extensibility. 16 On the other hand, GelMA is a photocrosslinkable biopolymer comprised of modified gelatin and it provides physiological cell binding motifs and protease-sensitive degradation sites as well as high mechanical strength and adhesion. 17 Here, we incorporated both mELP and GelMA into a polymeric network, enabling the modulation of several features such as degradation rate, mechanical properties, and tissue adhesion of the resulting composite glues. In addition, we used a visible light activated photoinitiator system to minimize the biosafety concerns associated with UV light such as DNA damages. 18 In particular, we utilized the Type 2 initiator Eosin Y, the co-initiator triethanolamine (TEA), and the co-monomer N-vinylcaprolactam (VC) for the photocrosslinking (Figure 1(c)). Briefly, visible light excites dye molecules of Eosin Y into a triplet state, which abstracts hydrogen atoms from TEOA. The deprotonated radicals initiate vinyl-bond crosslinking with VC via chain polymerization reactions, which leads to accelerated gelation. 19 To verify the degree of methacryloyl substitution of each biopolymer, proton nuclear magnetic resonance ( 1 H-NMR, 400 MHz) analysis was used. Results showed the emergence of the methacrylate (ɑ/β) and the methacrylamide (γ/δ) proton peaks for mELP ( Figure 1(d,e)) within 5.2-5.7 ppm range. Knowing the stoichiometric amounts of lysine, methionine, serine, and threonine residues, we used a reference molecule (PEG 2000 ) to determine the percentages of modified amino acids. 1 H-NMR analysis of mELP showed 40% degree of methacryloyl functionalization of lysine and terminal methionine amines to form methacrylamide groups and 8% degree of methacryloyl functionalization of serine and threonine residues to form methacrylate groups. These values are in agreement with the degree of methacryloyl functionalization of tropoelastin to yield methacrylated tropoelastin (MeTro) with 44% methacrylation, following the same synthesis method. 20 The engineered ELP sequence is shown in Figure 1 The lysine (light green) and the N-terminal methionine (dark green) residues contain primary amine ( NH 2 ) groups that allow methacrylamide functionalization. The serine and threonine residues (red) have hydroxyl ( OH) groups that allow methacrylate functionalization. The C-terminal serine (dark red) has two potential methacrylation sites 82%, which is in agreement with the previously published results following similar synthesis protocol. 17

| Mechanical characterization of the engineered hydrogel-based sealants
Mechanical properties of the hydrogel-based sealants were characterized through tensile and cyclic compression tests. As shown in Figure 2(a), the unconfined compressive moduli of the 15% GelMA/15% mELP composite sealant (30 ± 9 kPa) was between the pure mELP hydrogel (4 ± 3 kPa) and the pristine GelMA hydrogel GelMA and pure mELP hydrogels were 581 ± 51 and 218 ± 1 kPa, respectively. Although the Young's modulus of the engineered composite hydrogels decreased with increasing the mELP concentrations, their extensibility exhibited an opposite trend, which was later determined to be a critical factor for the anastomosis application. The pure GelMA hydrogel could be extended up to 13.6 ± 2.5% before rupture while the pure mELP hydrogel had an extensibility of 172 ± 17%. Prior studies also demonstrated 163 ± 11% extensibility for hydrogels containing 15% (w/v) ELP. 21 It was notable that four of the five prepolymer solutions could be easily handle; however, the pure mELP prepolymer solution was too viscous to be pipetted/injected and thus was considered impractical for clinical settings. Yet, incorporation of mELP biopolymer in the composite hydrogel significantly improved the elasticity (i.e., extensibility) after exposure to visible light and crosslinking.

| In vitro swelling and degradation of the engineered hydrogel-based sealants
Another benefit for the use of hydrogels as surgical sealants is controlled degradation in wet environments. Therefore, we aimed to investigate the in vitro degradation profiles of the engineered hydrogels in collagenase Type II solution. Results demonstrated that the in vitro degradation was consistently higher for the composite hydrogels compared to that of pure GelMA or mELP alone which corresponded to 46 ± 5% and 14 ± 3%, respectively, after 7 days of incubation in the collagenase Type II solution ( Figure 2(e)).
The 15% GelMA/15% mELP composite had the highest 7-day degradation at 72 ± 3%. The tunable degradation rate of the engineered sealants, based on GelMA and mELP concentrations, make them suitable for a wide range of surgical applications.
We also determined the swelling ratios of the resulting hydrogels at various time points, throughout their incubation in Dulbecco's phosphate-buffered saline (DPBS) at 37 C. In general, the results for all samples showed initial decrease in swelling (i.e., deswelling) ( Figure 2(f)). This was the expected behavior for both types of biopolymers. Previous works have shown that the swelling decreased with an increase in total concentration of GelMA, and the swelling ratio approached zero at 20% total polymer concentration. 22 Our results demonstrated that hydrogels formed by using pure GelMA had minimal deswelling compared to the hydrogel compositions containing higher mELP concentration, while the 15% GelMA/15% mELP formulation had a deswelling ratio of 0.88 ± 0.02 at 24 h and remained The results showed that the burst pressure significantly increased from 1.8 ± 0.6 kPa ($13 mmHg) for the pure 30% mELP hydrogels to 25 ± 1 kPa ($187 mmHg) for the pure 30% GelMA hydrogel ( Figure 3 (a)), which is above human systolic blood pressure during hypertensive crisis (180 mmHg). In addition, no significant difference in adhesive properties of different composite hydrogels was observed, via wound closure test using porcine skin as the biological substrate (Figure 3(b)).
The interactions between the polymer chains and tissue before and after photocrosslinking are shown in Figure 3 Next, an ex vivo anastomosis model was developed to further evaluate the adhesive strength and sealing functionality of the hydrogels (Figure 4(a,b)). The results showed that the pure GelMA bioadhesives failed to resist stress at an intra-arterial pressure of 12.7 ± 2.6 kPa. We hypothesized that this was due to the previously determined low extensibility (brittleness) of the GelMA hydrogel; the anastomosis resulted in cracks within the sealant that led to leakage.
Next, we applied the 15% GelMA/15% mELP composite hydrogel, which showed optimal adhesion and elasticity with a burst pressure value of 110 kPa ($825 mmHg). The rationale behind choosing this formulation is due to its mechanics (higher elasticity relative to 20% GelMA/10% mELP, Figure 2(a-d)) and adhesion (higher adhesion relative to 10% GelMA/20% mELP, Figure 3(a)). However, the fast rate of in vitro degradation of the 15% GelMA/15% mELP composite bioadhesive, in conjunction with the slow degradation rates of the pure GelMA adhesive brought us the idea of the circumferential application of the pure GelMA formulation. Therefore, we introduced a two gluebased system in which we first applied 15% GelMA/15% mELP composite bioadhesive (i.e., Glue II) between two vessels to allow for penetration, followed by the circumferential application of the stronger and highly adhesive pure GelMA hydrogel (Glue I) for further fortification and sealing (Figure 4(b,c)). The application of two hydrogel-based glues on the same endovascular anastomosis model achieved an intraarterial pressure of 34.0 ± 7.5 kPa, which is well over normal, elevated, and hypertensive crisis systolic blood pressures in humans at 16, 17.3, and 24 kPa, respectively (Figure 4(e)). In addition, the lap shear strengths of the pure 30% GelMA (Glue I) and the composite 15% GelMA/15% mELP bioadhesives (Glue II) were calculated. Shear adhesive strength of glue II was significantly lower (26.7 ± 5.4 kPa) than that of glue I (49.4 ± 7.0 kPa) (Figure 4(d)).

| In vitro cytocompatibility of the engineered hydrogel-based sealants
The optimal bioadhesive for anastomosis should be cytocompatible. It should also permit cell proliferation within the injured tissue for faster integration and healing. Therefore, we aimed to evaluate the in vitro cytocompatibility of the engineered 15% GelMA/15% mELP composite bioadhesives (Glue II) via live/dead and PrestoBlue assays, as well as actin/DAPI and CD31/DAPI staining of human umbilical vein endothelial cells (HUVECs) seeded on the surface of the bioadhesives.
Cytotoxicity of hydrogels at various concentrations of GelMA has previously been assessed by our group using various cell types, and GelMA hydrogel has been shown to be cytocompatible. 23 Figure 5(a,b)). The cells were also adhered and spread on the bioadhesive over 7 days of culture ( Figure 5(c)). In addition,

| Synthesis of GelMA
GelMA was synthesized as explained previously. 17 Briefly, gelatin from cold water fish skin was dissolved in DPBS (10% w/v). Then, methacrylic anhydride (Sigma-Aldrich) was added dropwise (8% v/v) at 60 C and the mixture was allowed to react for 3 h under continuous stirring. The reaction was then stopped by 1:4 dilution with DPBS. Finally, the solution was dialyzed against deionized water for 7 days, frozen at À80 C for 2 h, and desiccated for 5 days to yield high GelMA. After the fifth cold spin, the solution was pipetted into dialysis membranes and dialyzed against milli-Q water (changed twice per day) at 4 C for 4 days. 16, 21 The purified solution was frozen at À80 C and lyophilized to yield ELP.

| Synthesis of mELP
Purified ELP was then dissolved in DPBS (10% w/v) at 4 C and methacrylic anhydride was added dropwise to a 15% v/v final concentration. The mixture was continuously stirred in an ice bath and was allowed to react for 16 h. The mixture was then diluted into 4x volume with cold DPBS and dialyzed in a dialysis cassette against milli-Q water (changed twice per day) at 4 C for 4 days. The purified solution was frozen at À80 C and lyophilized to yield mELP.

| 1 H-NMR characterization of GelMA and mELP polymers
There are well-established methods to determine the degree of Here, the H* and H* 0 are the two terminal alkene protons of a methacryloyl group that present as two distinct singlets of exactly the same intensity at slightly different chemical shifts due to stereochemistry around the alkene. The terms n PEG and n m-ELP are the controlled number of moles of PEG 2000 and mELP, respectively, in the spectra. Ψ is the number of relevant residues per ELP chain; for methacrylamide functionalization, Ψ = 2 (2 lysine residues in sequence) and for methacrylate functionalization, Ψ = 9 (5 serine and 4 tyrosine residues in sequence). PEG 2000 integral at 3.47 ppm was nominally integrated for 179 aliphatic protons to obtain the total methacryloyl group content (i.e., aforementioned H* and H* 0 peaks). For GelMA, the same calculation was carried out using residue per unit mass (i.e., 2.5 Â 10 À4 mollysine/mg-gelatin) instead of residue per chain.

| Mechanical characterization
The prepolymer solutions were pipetted into polydimethylsiloxane molds of rectangular geometry ( In vivo feasibility of the two hydrogel-based glues using a nonsurvival anastomosis pig model. The arteries were tied together with sutures and glued in accordance with the described model for the two hydrogel-based glues. The needle device was inserted into the bloodstream from the femoral artery and advanced toward the anastomosis site within the donor carotid artery. A successful anastomosis was performed through the composite glue into the recipient artery cycle. The energy loss percentage was calculated by the area between the loading and unloading for each cycle.

| In vitro swelling ratio and degradation test
For the swelling test, cylindrical samples were prepared as described (w 0 -w ti ) to initial weight (w 0 ).

| Rheology test
Oscillatory rheology measurements were carried out on Anton Paar (MCR 302) by using a cone plate (radius 8 mm, cone angle 2 ). A solvent trap was used to minimize water evaporation during the measurements. Temperature sweeps were performed from 5 to 40 C at a heating rate of 1 C/min. For all measurements a frequency of 1 Hz and a strain of 1% were applied. This strain and frequency were previously determined to be within the linear viscoelastic region of these polymer solutions.

| Burst pressure test
The sealing capability of engineered sealants was measured according to a modified ASTM standard (F2392-04) for burst pressure as described previously. 5,11 Briefly, 40 μl of prepolymer solution was injected and photocrosslinked on a 1 mm in diameter hole made on a collagen sheet. Air was pumped at a rate of 10 ml/min using a syringe pump and the pressure inside the seal was measured using a PASCO wireless pressure sensor and software until burst (Supp. Figure 1).

| Wound closure test
The adhesion strength of the engineered sealants was measured using a modified ASTM standard test (F2458-05) according to previously published methods. 29,30 Porcine skin was used as the biological substrate in order to evaluate the relative adhesion strength of various formulations. The tissue was cut in 3 Â 1 Â 0.5 cm pieces and fixed onto two pieces of glass slides by superglue with a 0.5 cm overhang.
Two opposing pieces were then placed next together and 100 μl of prepolymer solution was pipetted and photocrosslinked on 1 Â 1 cm surface area via exposure to visible light (Supp. Figure 2). The adhesive strengths of the sealants were measured at the detachment point using an Instron 5542 mechanical tester. Tensile loading was conducted at strain rate of 1 mm/min. Adhesive strength was reported as the maximum stress on the stress-strain curve, corresponding to the breaking point.

| Lap shear test
The shear strength of the bioadhesives was measured using a modified lap shear test based on ASTM standard (F2255-05) according to previously published protocol. 31 As a biological substrate, porcine arteries (5 mm collapsed width) were cut into 20 mm long segments and fixed on glass slides by superglue. Prepolymer solution was then applied on half of one segment (10 Â 5 mm), over which the second segment was placed (Supp. Figure 3). After photocrosslinking with visible light, the thickness of the hydrogel was measured with a digital caliper and the glass slides were loaded to Instron 5542 mechanical tester and pulled apart at a rate of 1 mm/min. Shear stress (kPa) was measured at the maximum stress where the two artery segments were separated using a BlueHill Universal software.

| Ex vivo test using a porcine anastomosis model
Porcine carotid arteries were prepared by removal of the tunica adventitia. Segments with approximately 5 cm in length without branching were cut from parent vessel. To achieve the best outcome for sealing and anastomosis, we used a two-step sealing procedure using two different formulations of the engineered glue: an elastic and soft formulation based on 15% GelMA/15% mELP composite, named Glue II, and a stiffer formation based on pure GelMA, 30% (w/v), named as Glue I. This procedure allowed for the penetration of the elastic glue (Glue II) without cracks to achieve a leak-free anastomosis, and the subsequent reinforcement of the anastomosis site with the stiff glue (Glue I) to protect it against external stress.
In the first step of gluing, the composite prepolymer solution
The samples were then imaged using an inverted fluorescence microscope (Zeiss Axio Observer Z7).

| Pig nonsurvival anastomosis model
A newly developed in vivo swine model was used to assess the feasibility of applying our two-step sealing procedure for the endovascular anastomosis. The endovascular bypass was performed in swine (50-140 lbs, n = 4) under general anesthesia. The right common carotid and ascending cervical arteries were then surgically exposed, cleaned of adventitia, and loosely looped with 3-0 silk through a para-midline linear skin incision. Topical papaverine hydrochloride was applied to the exposed arteries for vasodilation. The two arteries were placed closely in parallel and gently unmoistened with gauze.

| Statistical analysis
Unpaired, one-tailed Welch's t test was applied with a 95% confidence interval with a minimum of three replicates per sample. Error bars represent mean ± SD of measurements (*p < .05, **p < .01, and ***p < .001).

| CONCLUSION
In this work, we introduced a new class of composite hydrogel sealants primarily for a novel, arterial anastomosis procedure, which can be fine-tuned to be utilized for a broader range of vascular applications. Composite hydrogels were fabricated from two bio- The former is to ensure sufficient blood flow to both arteries; the latter is critical to maintain the structural integrity of the glue, securing the anastomosis.