Silk fibroin vascular graft: a promising tissue-engineered scaffold material for abdominal venous system replacement

No alternative tissue-engineered vascular grafts for the abdominal venous system are reported. The present study focused on the development of new tissue-engineered vascular graft using a silk-based scaffold material for abdominal venous system replacement. A rat vein, the inferior vena cava, was replaced by a silk fibroin (SF, a biocompatible natural insoluble protein present in silk thread), tissue-engineered vascular graft (10 mm long, 3 mm diameter, n = 19, SF group). The 1 and 4 -week patency rates and histologic reactions were compared with those of expanded polytetrafluoroethylene vascular grafts (n = 10, ePTFE group). The patency rate at 1 and 4 weeks after replacement in the SF group was 100.0% and 94.7%, and that in the ePTFE group was 100.0% and 80.0%, respectively. There was no significant difference between groups (p = 0.36). Unlike the ePTFE graft, CD31-positive endothelial cells covered the whole luminal surface of the SF vascular graft at 4 weeks, indicating better endothelialization. SF vascular grafts may be a promising tissue-engineered scaffold material for abdominal venous system replacement.

Silk fiber is a natural protein fiber and silk thread has long been used in surgery for suturing and ligature 13 . Silk fibers comprise silk fibroin (SF) and silk sericin 14 . Silk sericin is an antigenic gum-like protein that surrounds the SF core fibers 13 and can be removed through a degumming process 15 . SF biomaterial has biologic advantages, such as better biocompatibility, high affinity for cells, and susceptibility to proteolytic degradation in vivo without antigenicity 16,17 . Recently, experimental artery replacement using double-raschel knitted SF vascular grafts coated with an SF sponge was reported in animal models 18,19 . Enomoto et al. first reported rat aorta replacement using an SF graft with a 1-year patency rate of 85%. Anti-CD31 and anti-smooth muscle actin immunostaining revealed that endothelial cells and smooth muscle cells migrate into the SF graft early after implantation and become organized into endothelial and medial layers 18 . No venous replacement model has yet been reported.
We evaluated the patency rate of double-raschel knitted SF grafts coated with an SF sponge as an abdominal venous system replacement for the IVC in a rat model. In addition, we investigated the histologic reaction to the SF graft. We hypothesized that SF grafts would have a better patency rate than ePTFE grafts.

Results
Scanning electron microscopy observation and mechanical properties of SF grafts. A schematic figure of the double-raschel knitting pattern is shown in Fig. 1A. Scanning electron microscopy images, including the surfaces and cross-sections of SF grafts coated with SF sponges are shown in Fig. 1B,C. The longitudinal suture retention strength, circumferential tensile strength, and circumferential compressive elastic modulus of SF grafts coated with SF sponges were 6.4 ± 0.6 N, 51.0 ± 3.0 N, and 0.013 ± 0.002 N mm 2 , respectively. Porosity was 82%.
Animal model. The number of rats, age, and mean body weight were as follows; SF group: n = 19, 13 weeks, 450 g, and the ePTFE group: n = 10, 12 weeks, 405 g, respectively. The surgical time did not differ significantly between groups (SF group, 43 min vs ePTFE group, 48 min; Table 1). The SF graft was soft and flexible, similar to the ePTFE graft. An SF vascular graft before and immediate after IVC replacement shown in Fig. 2A,C. The SF graft quickly absorbed blood after recanalization and became reddish in color without bleeding. The ePTFE graft was still whitish in color after recanalization (Fig. 2B,D). No rats died before we evaluated the graft patency.
Graft patency. A representative Doppler ultrasonography result showing colored flow from the distal to proximal sides and ~ 5 cm/s steady shear flow in the SF group is shown in Fig. 3. The patency rate at 1 and 4 weeks after replacement in the SF group was 100.0% (n = 19/19, solid line) and 94.7% (18/19), respectively, whereas in the ePTFE group, it was 100.0% (10/10, dotted line) and 80.0% (8/10), respectively. The difference between groups was not significant (p = 0.36, Fig. 4). One SF vascular graft occluded at 3 weeks 5 days after  Histologic analysis. Hematoxylin and eosin staining of each graft is shown in Fig. 5. The lumens of both the ePTFE and SF vascular grafts remained patent, but the lumen of the ePTFE graft was circumferentially narrowed (Fig. 5A,B). In the SF vascular graft, cellular proliferation was observed around the SF fibers, and the luminal and outer surfaces were covered by flat cells (Fig. 5B, arrowhead). In the ePTFE vascular graft, fibrin fibers accumulated, but no flat cells were observed on the luminal surface (Fig. 5D, arrowhead). Elastica van Gieson staining of the SF vascular graft revealed collagen fibers, red, around the SF fibers, but no elastic fibers, black, (Fig. 6A, EVG, arrowhead). CD31 was expressed on the luminal surface of the SF vascular graft (Fig. 6A, CD31, arrowhead). Anti-alpha smooth muscle actin (αSMA) antibody staining of the SF vascular graft was positive and backing CD31 positive cells (Fig. 6A, αSMA, arrowhead). Podoplanin was weakly expressed on the outer surface and partially in the wall of the SF vascular graft (Fig. 6A, Podoplanin, arrowhead). The SF vessel graft wall was filled with collagen fibers. In one example of a cross section, the areas of the SF fibers and infiltrated native cells were calculated to be 48% and 52% of the total cross-sectional area of the SF vessel graft (Fig. 7).  www.nature.com/scientificreports/ In the ePTFE graft, Elastica van Gieson staining showed thick fibrin inside the lumen (Fig. 6B, EVG, arrowhead). No CD31 positive cells were observed on the luminal surface of the ePTFE vascular graft (Fig. 6B, CD31, arrowhead). Anti-αSMA antibody staining was positive inside the lumen in thick fibrin, which may indicate intimal thickening (Fig. 6B, αSMA, arrowhead). Podoplanin-positive cells were found not only inside the lumen, but weak positive staining was also observed on the outer surface of the ePTFE vascular graft (Fig. 6B, Podoplanin, arrowhead).
As a positive control of the rat small artery, positive CD31 staining of endothelial cells was observed in an inner surface with a thin layer of lumen (Fig. 6C, CD31, single arrowhead). In the negative control of CD31, Figure 4. The patency rates at 1 and 4 weeks after replacement in the SF group (solid line) were 100.0% and 94.7%, respectively, whereas in the ePTFE group (dotted line), the patency rates were 100.0% and 80.0%, respectively. The difference between the 2 groups was not statistically significant (p = 0.36). ePTFE; Expanded polytetrafluoroethylene, SF; silk fibroin. An image was made by a software, IBM SPSS Statistics, Ver. 25.0, https ://www.ibm.com/jp-ja/produ cts/spss-stati stics .

Discussion
This is the first report of venous replacement using an SF vascular graft in a rat model. We compared the patency rate and histologic reaction between SF vascular grafts and ePTFE grafts. The SF grafts had a favorable patency rate, and better endothelialization and collagen fiber infiltration than the ePTFE grafts at 4 weeks after venous replacement. SF may thus be an ideal scaffold material for grafting in the abdominal venous system, such as for replacing the IVC.
In present study, the SF graft had a 95% patency rate at 4 weeks, whereas the ePTFE graft had an 80% patency rate, although the difference was not statistically significant. Unlike reports on arterial replacement, there are few reports of animal models of abdominal venous replacement. In a canine IVC replacement model, an ePTFE graft had a 100% patency rate at 4 weeks, but more than half of each ePTFE graft lumen was filled with a thrombus 20 . Several studies have reported the patency rate of venous replacements in the clinical setting. For hepatic vein replacement, the 4-week patency rate of autologous vein grafts is 100%, whereas that of cryopreserved homologous veins is 95% 10 . In contrast, the 4-week patency rates of ePTFE grafts is only 81% 21 . Our findings revealed . Anti-αSMA antibody staining was positive inside the lumen in thick fibrin, which may indicate intimal thickening (B, αSMA, arrowhead). Podoplanin-positive cells were found not only inside the lumen, but weak positive staining was also observed on the outer surface of the ePTFE vascular graft (B, Podoplanin, arrowhead). As a positive control of the rat small artery, positive CD31 staining of endothelial cells was observed in an inner surface with a thin layer of lumen (C, CD31, single arrowhead). In the negative control of CD31, there was no CD31 staining in the thick smooth muscle layer (C, CD31, double arrowheads), in contrast to αSMA positive thick smooth muscle layer (C, αSMA, arrowhead). An example of native rat IVC showed lining endothelial cells at inner surface in HE staining (C, HE, arrowhead) and smooth muscle and collagen fibers layer around lumen in Elastica van Gieson staining (C, EVG, arrowhead). An occluded ePTFE vascular graft (C, HE × 40) at 3 weeks 3 days showed neutrophils infiltrated inside the graft aggregating around erythrocytes (C, HE × 100, single arrowhead) and lymphocytes (C, HE × 100, double arrowhead), indicating infection and inflammation, respectively. αSMA; α-smooth muscle actin, SF; silk fibroin, IVC; inferior vena cava, HE; Hematoxylin and eosin, ePTFE; expanded polytetrafluoroethylene. All images were taken by a software, Nikon DS-L2, Ver. 4.61, https ://www.nikon .com/ produ cts/micro scope -solut ions/suppo rt/downl oad/softw are/camer asfor /ds_l2_v461u .htm.

Scientific Reports
| (2020) 10:21041 | https://doi.org/10.1038/s41598-020-78020-y www.nature.com/scientificreports/ similar patency rates between SF and ePTFE grafts. Moreover, the long-term 1-year patency rates of autologous and cryopreserved homologous veins are 68% and 55%, respectively, for hepatic vein replacement 10 . The 4-month patency rate of ePTFE grafts is 39% and ePTFE grafts tend to occlude earlier 21 . As we only observed graft patency for a relatively short period ( 4 weeks ), additional studies are needed to evaluate the long-term patency rate of SF grafts for venous replacement in an animal model. The remodeling processes is key to developing tissue-engineered vascular graft using various scaffold materials. Of these, SF is a biodegradable protein derived from silk that provides an ideal scaffold for various cell types in tissue engineering because of their favorable biocompatibility 22 . Endothelialization of remodeling processes contributes antithrombogenic and antiatherogenic properties for better patency 23 , and was reported for an SF vascular graft model of artery replacement 16 . Endothelial cells covered most of the luminal surface of the SF graft walls at 12 weeks after artery replacement 18 . Haga and colleagues reported the early completion of endothelialization at 3 weeks after rat aorta replacement with an SF graft 16 . Furthermore, endothelialization protects the graft from infection 23 , but endothelialization is not observed in ePTFE grafts 24 . In present study, CD31-positive endothelial cells covered the whole luminal surface of the SF vascular grafts at 4 weeks. These findings suggest the potential application of the SF vascular graft for abdominal venous replacement in a contaminated surgical field, such as in hepato-biliary-pancreatic surgery. ePTFE vascular grafts cannot absorb blood because the ePTFE membrane is resistant to liquid infiltration 24 . Although SF fibers are coated with glycerin to prevent blood leakage between the fibers during replacement 25 , in the present study, the SF vascular grafts quickly absorbed blood after recanalization without bleeding (Fig. 1C). This absorption ability may make SF fibers biocompatible. According to several reports, SF fibers provide a durable scaffold as a vessel wall, but they are degradable, decompose in vivo, and are absorbed slowly 26 . In an artery replacement model, the amount of collagen tissue fibers increased while the amount of SF fibers decreased 18 . We observed collagen fiber proliferation around SF fibers. This may also contribute to their biologic compatibility 27,28 , and sufficient collagen accumulation may enhance the strength as a framework of the vessel wall 18 . Additionally, in the present study, some αSMA positive cells were observed, indicating that smooth muscle cells become established on the SF vascular graft 29 . Although this remodeling process corresponds to the artery replacement SF graft model, it seemed to be thinner than the artery 18 .
In the present study, podoplanin-positive cells were observed on the outer surface of the SF vascular graft, indicating that mesothelial cells were established. Podoplanin is a mesothelial marker 30 and outer side of rat native IVC is covered by mesothelium (retroperitoneum) in general. If the outer surface of the SF graft was framed by mesothelial-like cells, the SF graft would have tolerance against infection because mesothelial cells may provide a non-adhesive and protective abilities for the replacement SF graft 31 . In the present study, however, the inside wall of the SF and the fibrin-like tissue of the ePTFE graft also showed podoplanin staining. The significance of podoplanin for venous remodeling in the present experiment model requires further investigation.
Although the wall structure of the rat native aorta is similar to that of the inferior vena cava, the elastic fibers and smooth muscle are thicker 32 . Endothelialization and cell infiltration were similar in the SF vascular graft Figure 7. Semi-quantitative analysis using Image J 35 was used to determine the ratio of the SF fibers and infiltrated tissue area of the representative cross-section of the SF graft wall. In one example of a cross section, the areas of the SF fibers and infiltrated native cells were calculated to be 48% and 52% of the total crosssectional area of the SF vessel graft. SF; silk fibroin. An image was made by a software, Image J, Ver. 1.8.0_172., https ://image j.nih.gov/ij/index .html.
Scientific Reports | (2020) 10:21041 | https://doi.org/10.1038/s41598-020-78020-y www.nature.com/scientificreports/ compared with a previously reported artery replacement model 18 , but the thickness of the elastic fibers and smooth muscle, and the remodeling speed may differ. Further studies are needed to elucidate the differences. A limitation of the present study is that we did not assess infection tolerability and inflammatory response to the SF graft. Further experiments with not only a 4-week time-point, but also a shorter and longer observation periods are needed to clarify the remodeling processes, the infection tolerability and the differences between venous and artery replacement animal models.
In conclusion, SF vascular grafts may be a promising tissue-engineered scaffold material for replacement in the abdominal venous system.

Methods
SF vascular grafts coated with an SF sponge. The SF double-raschel knit tube made of Bombyx mori SF threads was prepared using a double needle bar raschel with ten 30-guage needles and 20 courses on a computer-controlled double-raschel knit machine (Fukui Wrap Knitting Co. Ltd. Fukui, Japan). In our previous report 33 , the permeability of SF grafts with a double bar-cord knitting pattern based on the international organization for standardization (ISO) 7198 procedure was 4 ml/cm 2 /min and its density was 10.96 mg/cm 2 . The double bar-cord knitting pattern produced by a double-raschel machine has better strength to avoid coming apart at the seams and a thicker structure to allow cell infiltration inside the knitting pattern as a scaffold.
The inner diameter of the SF tube was 3 mm. To remove silk sericin, the knitted SF tube was degummed in a mixture of sodium carbonate (0.08% w/v) and Marseille soap (0.12% w/v) solution at 95 °C for 2 h, as previously reported 19,33 . To make the SF sponge, an aqueous solution of SF was prepared. The remaining degummed SF fibers were dissolved by adding CaCl2-H2O-EtOH solution (molar ratio: 1:8:2) to SF at a concentration of 10% w/v and boiling them at 70 °C for 1 h. This solution was filtered to remove residual solid components, and then dialyzed against a cellulose dialysis membrane (36/32, MWCO 14,000, Viskase Companies, Inc., Lombard, IL, USA) at 4 °C for 3 days. An approximately 4% (w/v) SF aqueous solution was obtained and centrifuged for 30 min at 4 °C at 18,000 rpm to remove impurities. The knitted SF tube was then coated with the SF sponge was as follows. The SF tube was immersed into a pipe filled with the mixed aqueous solution at a ratio of 1 : 1 (w/w) SF and glycerin, which was used as a porogen for coating. The pipe was placed in a desiccator where the interior was kept under reduced pressure of 100 hPa, and the SF graft was removed from desiccator and frozen at − 20 °C overnight. The graft was then immersed in distilled water for 3 days with several water changes to remove the glycerin. The graft was placed a bag with distilled water and sterilized in an autoclave at 120 °C for 20 min.

Scanning electron microscopy observation and mechanical properties of SF grafts. The mor-
phology of the outer surface and the cross-section of the knit SF grafts coated with SF sponges were observed by scanning electron microscopy (VE-7800, Keyence Corp., Osaka, Japan).
The longitudinal suture retention strength, circumferential tensile strength, and circumferential compressive elastic modulus of the SF grafts coated with SF sponges were determined using a tensile testing machine, table-top material tester (EZ Graph , Shimadzu Corp., Kyoto, Japan) according to the previously reported methods 18,19,33,34 . For the longitudinal suture retention test, a sample tube was cut to a length of 20 mm and then the suture was passed through 2 mm from the end., pulled by a clamp (3 mm/min.) using a 100 N operator cell until the breaking point, and analyzed. To determine the circumferential tensile strength and circumferential compressive elastic modulus, we prepared dry, short, ring-shaped specimens with an axial length of 10 mm for each tube. The load cell was set to 5 N, and the rate of stretching was 2 mm/min. The tensile strength was measured as a function of the stroke distance and the circumferential compressive elastic modulus was measured when the specimen was compressed using 10% of the inner diameter. All specimens were hydrated for 1 h before testing. Porosity, ε (%), was calculated by following the formula, ε (%) = (V2 − V1)/(V3 − V1) × 100. An SF graft was immersed in hexane solution (V1) for 10 min (hexane with SF Graft volume, V2) under reduced pressure (0.05 MPa). The volume of the hexane solution (V3) was measured again after removing the SF graft filled with hexane.

Animal model. The study protocol (I-P16-034) was approved by the University of Tokyo Animal Ethics
Committee in accordance with the Japanese and ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. Male Sprague-Dawley rats (CLEA Japan, Inc., Tokyo, Japan) were used. All rats were kept for 1 to 4 weeks in micro-isolator cages with a 12-h light/dark cycle and fed a certified diet (CRF-1, Charles River Laboratories Japan, Inc., Yokohama, Japan). The rats were fasted overnight before undergoing the surgical procedure.
Surgical procedure. All surgical procedures were performed by hepato-biliary-pancreatic surgeons (S.K., D.I.). The rats were anesthetized by intraperitoneal injection of pentobarbital (30 mg/kg body weight). The IVC was exposed from the left renal vein at the cranial side to the bifurcation on the caudal side, and all branches of the IVC were ligated and divided with 6-0 nylon (Keisei Medical Industrial CO., LTD., Tokyo, Japan) knotted sutures using an electric scalpel. After intravenous injection of unfractionated heparin (100 IU/kg), the proximal and distal portions of the infra-renal IVC were clamped with vascular clips. A 10-mm segment of the IVC was removed and replaced with an SF vascular graft (10 mm long, 3 mm in diameter, n = 19 ) by end-to-end anastomosis using 10-0 nylon (Keisei Med.) sutures, starting with 2 stay sutures at 180° apart at both the cranial and caudal sides, then suturing the front wall, followed by the back wall. The cranial and caudal side anastomosis each required 12 stitches. The caudal and cranial sides of the vascular clamps were slowly removed, and vessel flow was restored through the SF vascular graft. Graft patency was confirmed by gross inspection. In another group of rats, the IVC was replaced with an ePTFE graft (10 mm long, 3 mm in diameter, W.L. Gore & Associates, Inc., Newark, DE, USA, n = 10 ) in the same manner. No anticoagulant, antiplatelet, or antibiotic agents were administered postoperatively.

Scientific Reports
| (2020) 10:21041 | https://doi.org/10.1038/s41598-020-78020-y www.nature.com/scientificreports/ Patency assessment. The patency of the grafts in the SF and ePTFE groups was monitored by Doppler ultrasonography with a diagnostic ultrasound system (8-MHz sector probe, TUS-A300, Toshiba Corp., Tokyo, Japan) at 1 day and every 3 days for 1 to 4 weeks after venous replacement. Graft occlusion was defined as the absence of a color Doppler signal. When we found the absence of a color Doppler signal, the rat was anesthetized with an intraperitoneal injection of 30 mg/kg body weight pentobarbital and the graft was grossly and pathologically evaluated.
Histologic analysis. At 4-weeks after replacement, rats underwent a general physical examination to evaluate their condition. Following induction of general anesthesia with an intraperitoneal injection of pentobarbital (50 mg/kg body weight), the rats were perfused with 0.9% saline solution through the left ventricle. The grafts were carefully removed together with the surrounding tissue. Tissue cross-sections were prepared at the middle of the SF and ePTFE vascular grafts and fixed in 10% formalin and snap-frozen in Tissue Tek O.C.T. compound (Sakura Finetek Japan Co., Ltd., Tokyo, Japan) for histologic analysis. The tissues were fixed in 10% formalin, embedded in paraffin, sectioned (4-µm thick, Tissue-Tek Auto Section), and then processed for hematoxylin and eosin staining. Elastica van Gieson stain was also applied to detect elastic and collagen fibers. Semi-quantitative analysis was used to determine the ratio of the SF fibers and infiltrated tissue area of the representative cross-section of the SF graft wall. SF fibers, seen as aggregations of transparent dots on a representative cut surface were encircled by yellow lines on a histologic image. At the same time, infiltrated native cells, the other area of the whole SF graft wall, was encircled with a blue line. The ratio of the remaining SF fiber area (yellow) to native cells (blue) was determined to calculate the area of one cut surface of the SF graft that was replaced by native cells at 4 weeks. The analysis was performed using Image J software (version 1.44; National Institute of Mental Health Bethesda, MD, USA) 35 .

Statistical analysis.
Factors of both groups were compared using the Mann-Whitney U test. The log rank test was used to assess the patency rate in both groups. A p-value less than 0.05 was considered statistically significant. IBM SPSS Statistics for Windows, version 25.0 (IBM Corp, Armonk, NY, USA) was used for the analysis.