Tissue Engineering of Axially Vascularized Soft-Tissue Flaps with a Poly-(ɛ-Caprolactone) Nanofiber-Hydrogel Composite

Abstract

Objective: To develop a novel approach for tissue engineering of soft-tissue flaps suitable for free microsurgical transfer, using an injectable nanofiber hydrogel composite (NHC) vascularized by an arteriovenous (AV) loop.

Approach: A rat AV loop model was used for tissue engineering of vascularized soft-tissue flaps. NHC or collagen-elastin (CE) scaffolds were implanted into isolation chambers together with an AV loop and explanted after 15 days. Saphenous veins were implanted into the scaffolds as controls. Neoangiogenesis, ultrastructure, and protein expression of SYNJ2BP, EPHA2, and FOXC1 were analyzed by immunohistochemistry and compared between the groups. Rheological properties were compared between the two scaffolds and native human adipose tissue.

Results: A functional neovascularization was evident in NHC flaps with its amount being comparable with CE flaps. Scanning electron microscopy revealed a strong mononuclear cell infiltration along the nanofibers in NHC flaps and a trend toward higher fiber alignment compared with CE flaps. SYNJ2BP and EPHA2 expression in endothelial cells (ECs) was lower in NHC flaps compared with CE flaps, whereas FOXC1 expression was increased in NHC flaps. Compared with the stiffer CE flaps, the NHC flaps showed similar rheological properties to native human adipose tissue.

Innovation: This is the first study to demonstrate the feasibility of tissue engineering of soft-tissue flaps with similar rheological properties as human fat, suitable for microsurgical transfer using an injectable nanofiber hydrogel composite.

Conclusions: The injectable NHC scaffold is suitable for tissue engineering of axially vascularized soft-tissue flaps with a solid neovascularization, strong cellular infiltration, and biomechanical properties similar to human fat. Our data indicate that SYNJ2BP, EPHA2, and FOXC1 are involved in AV loop-associated angiogenesis and that the scaffold material has an impact on protein expression in ECs.

Volker J. Schmidt, MD

Justin M. Sacks, MD, MBA, FACS

Introduction

Biologically derived acellular dermal substitutes have been used for reconstruction of superficial soft-tissue defects, burn injuries, and chronic wounds with promising results.^1–3 A variety of biological scaffolds are commercially available, which allow native tissue integration, recellularization, and blood vessel ingrowth, thereby enabling reconstruction of soft-tissue defects in conjunction with skin grafts sometimes even in challenging cases with exposed tendons or bone.^4,5 The incorporation of nanofibers into acellular scaffolds has recently gained popularity in tissue engineering because they may improve the biomimetic and mechanical properties of a given scaffold. Nanofibers can mimic extracellular matrix (ECM) fibers and thus promote cellular attachment⁶ and protein adsorption,⁷ increase stability,⁸ or serve as vehicles for drug delivery.⁹ However, soft-tissue reconstruction with acellular scaffolds depends on extrinsic vascularization from the wound bed and hence is insufficient for coverage of large, deep, or irradiated wounds with poor perfusion. In these situations, reconstruction with intrinsically vascularized tissue, that is, autologous flaps is required.

Arteriovenous (AV) loops induce angiogenesis through upregulation of shear-responsive signaling pathways in the vascular endothelium.^10,11 Since the prevascularization of a skin graft with an AV loop was first shown by Melvin Spira's group in 1980,¹² various types of scaffolds have been used in AV loop-based tissue engineering with mixed results.^13,14 Our group has previously demonstrated the free microsurgical transfer and successful coverage of complex wounds with tissue-engineered flaps based on an acellular collagen-elastin (CE) scaffold, which had been prevascularized by an AV loop, in a rat model.¹⁵ A disadvantage of commercially available dermal substitutes is their stiffness that does not allow for a customized three-dimensional flap design specifically adapted to given defect geometries.

Using microarrays, we have previously shown that proangiogenic transcriptomic signatures are significantly upregulated in vascular tissue from AV loops in rats as well as human patients compared with vein graft controls.^11,16 Moreover, we have shown that the deregulated gene expression in AV loops is related to distinct microRNA (miRNA) signatures that particularly influence the expression levels of the mitochondrial membrane protein synaptojanin-2 binding protein (SYNJ2BP), ephrin receptor kinase 2 (EPHA2), and the transcription factor forkhead box C1 (FOXC1),¹⁶ which are downregulated in AV loop venous tissue and seem to be key regulators of AV loop-induced angiogenesis. In this study, we aim to investigate the impact of different scaffold materials on the expression of these proteins.

Clinical Problem Addressed

Autologous tissue transfer is limited by donor site morbidity, which can cause a significant burden and sometimes requires secondary defect coverage.^17,18 In patients with major burns, severe trauma, or a low body mass index, donor sites for autologous soft-tissue flaps are often scarce, making defect reconstruction in these patients especially challenging. An approach to overcome this dilemma is a combination of intrinsic vascularization and scaffold-based soft-tissue reconstruction in the form of intrinsically vascularized bioengineered flaps. In this study, we investigate a novel injectable hydrogel composed of hyaluronic acid (HA) cross-linked with electrospun poly-(ɛ-caprolactone) (PCL) nanofibers in AV loop-based flap engineering and compare neoangiogenesis, ultrastructure, and protein expression of SYNJ2BP, EPHA2, and FOXC1 with a commercially available collagen-based dermal substitute. Moreover, we determine the rheological properties of the tissue-engineered flaps and compare them with native human adipose tissue.

Materials and Methods

Fabrication of nanofiber hydrogel composite

HA was modified with acrylate groups plus polyethylene glycol (PEG)-dithiol. Electrospun PCL nanofibers were grafted with polyacrylic acid (PAA) and surface-modified with maleimide groups. The nanofibers were ground with a cryomilling system and uniformly dispersed inside the HA hydrogel to create a nanofiber hydrogel composite (NHC) with interfacial bonding. The fabrication process of the nanofiber hydrogel was described elsewhere in detail.¹⁹

Animals

All animal experiments were performed at the research laboratory of the BG Trauma Center Ludwigshafen according to the German Animal Welfare Act and were approved by the Institutional Animal Care and Use Committee of the local governmental authorities (Landesuntersuchungsamt Rheinland-Pfalz G-13-7-008). Surgeries were performed on 34 female Sprague Dawley rats (Charles River Laboratories, Sulzfeld, Germany) weighing 280–320 g (ages 10–14 months). Animals were kept at a 12 h dark/light cycle and had free access to food and water. All surgeries were performed with a surgical microscope under 16× magnification (OPMI pico; Carl Zeiss, Germany) under inhalation anesthesia with isoflurane (5% and 2.5%) in pure oxygen. Until POD 2, animals were treated with heparin (80 IU/kg i.v.) and buprenorphin (0.05 mg/kg s.c.; Bayer Vital GmbH, Leverkusen, Germany). At the end of the experiments, euthanasia was achieved by intracardial injection of pentobarbital under deep anesthesia.

Microsurgical AV loop creation

The saphenous artery and vein were dissected and exposed along the medial thighs of the rats. A 20-mm-long saphenous vein graft was harvested on the left leg and transferred to the right leg, where an AV loop was created by anastomosis of the vein graft between the right saphenous artery and vein in an end-to-end manner with 11-0 Ethilon sutures (Ethicon, Somerville, NJ) (Fig. 1A). Patency of the microanastomoses was assessed by observation of pulsatility and a double occlusion test. AV loops were placed into round polytetrafluoroethylene (PTFE) isolation chambers (height 6 mm × inner diameter 10 mm), which were closed with a lid (2 mm × 14 mm) (Harhaus Medical Devices, Remscheid, Germany) and sutured onto the underlying muscle fascia (Prolene 6-0; Ethicon) (Fig. 1B, C) The chamber design is described elsewhere in detail.²⁰ Wound closure was performed with running subcutaneous (Vicryl 3-0; Ethicon) and cutaneous sutures (Vicryl 4-0; Ethicon).

Figure 1.

(A) Experimental setup: transfer of a saphenous vein graft (light blue) to the contralateral leg and microsurgical end-to-end anastomosis between saphenous artery (red) and vein (dark blue). (B) Intraoperative image after microsurgical AV loop creation and placement into the isolation chamber with its bottom covered by the nanofiber hydrogel composite (NHC) scaffold. (C) After embedding the AV loop into the NHC scaffold. (D) Comparison of perfused blood vessel count in CE and NHC AV loop flaps. (E) CE and (F) NHC AV loop constructs after perfusion with black ink and explantation on postoperative day 15. Vascular pedicle facing downward. (G) Cross sections of the prefabricated constructs composed of NHC and CE after explantation on postoperative day 15. Areas of neoangiogenesis with differentiation into arterioles (#) and venules (*) surrounding the main AV loop pedicle vessels (+) are magnified. AV, arteriovenous; CE, collagen-elastin; SA, saphenous artery; SV, saphenous vein; VG, vein graft from the contralateral saphenous vein. Color images are available online.

Experimental groups

The chambers were filled with 0.5 mL of the NHC (n = 10 animals) or a HA hydrogel without nanofibers (HA-only, n = 4 animals). The injectable hydrogels were evenly distributed below and above the AV loop (Fig. 1B, C). In another experimental group (n = 10 animals) the chambers were filled with a commercially available form-stable acellular dermal substitute consisting of bovine CE with an average pore volume of 20–140 μm (MatriDerm^®; MedSkin Solutions Dr. Suwelack, Billerbeck, Germany). The AV loops were placed between two sheets of the CE scaffold with a thickness of 2 mm each. To investigate the impact of AV shunt flow on vascularization and protein expression in the two types of scaffolds, two control groups were analyzed, in which the native saphenous vein was placed into a chamber with two opposing openings filled with either the CE scaffold (CE-Co, n = 5) or the NHC (NHC-Co, n = 5). The isolation chamber served to create a defined space and to largely prevent the ingrowth of unwanted neoangiogenesis from the wound bed into the scaffold.

Histological analysis of neovascularization

To visualize functional neoangiogenesis within the scaffolds on POD 15 after AV loop creation, the descending aorta of the rats was cannulated with a 24-gauge catheter and flushed with heparin solution (100 IU/mL) followed by 30 mL of warm (37°C) India Ink solution (50% v/v India Ink [Windsor & Newton, London, England] in 5% gelatin and 4% mannitol). After explantation from the chambers the constructs were fixed in 4% paraformaldehyde, dehydrated and embedded into paraffin. Histological cross sections of 4 μm thickness were performed perpendicularly to the longitudinal AV loop axis in the central area of the chamber between the fixation pins. Hematoxylin and eosin staining was performed according to standard protocols. Cross sections were visualized with a Zeiss Axio Vision microscope and recorded with the Axio Vision 4 software (Carl Zeiss Microscopy, Jena, Germany) (Fig. 1G).

Immunohistochemical staining

Serial sections were immunostained for SYNJ2BP (unconjugated rabbit polyclonal anti-synaptojanin 2 binding protein antibody, CSB-PA023018GA01HU-150; Cusabio, Houston, TX), EPHA2 (unconjugated rabbit polyclonal anti-ephrin type-A receptor 2 antibody, GTX32587-100; Gene Tex, Irvine, CA), and FOXC1 (unconjugated rabbit polyclonal anti-forkhead box C1 antibody, BYT-ORB87648-50; Biorbyt, San Francisco, CA). Staining was performed with a BenchMark XT immunostainer (Roche Diagnostics, Germany) using the UltraView 3,3′-diaminobenzidine (DAB) detection and amplification kit (Roche Diagnostics). The slides were counterstained with hematoxylin for 4 min and postcounterstained with bluing agent for 4 min. Afterward the slides were washed and dehydrated in 70–100% alcohol baths and xylene baths before applying coverslips.

Scanning electron microscopy

Sections were deparaffinized with xylole and washed with 100% ethanol. Then, sections were treated with hexamethyldisilazane (HMDS) and ethanol (1:1) for 10 min. and with HMDS (twice for 10 min). After drying, sections were coated with carbon using an SCD 030 Sputtering Coater (Bal-Tec AG, Balzers, Liechtenstein). Images were acquired at 500 × , 1,000 × , and 2,000× magnification with an XL 30 Environmental Scanning Electron Microscope (SEM) (FEI Philips, Eindhoven, Netherlands).

Rheological analysis

A TA Instrument ARES-G2 was used to perform rheological testing on AV loop flaps (NHC and CE) and control scaffolds (NHC-Co and CE-Co) immediately after explantation. Rheological testing was also performed on nonimplanted CE and NHC scaffolds soaked in phosphate-buffered saline as well as four adipose tissue samples collected from a human abdominal pannus specimen from a patient undergoing abdominoplasty at Stanford University Hospital (n = 3 or 4 for each group), Human tissue collection had been approved by the Institutional Review Board at Stanford University (IRB 54225).

Samples were placed on the bottom plate of the rheometer, and an 8 mm flat plate was brought down upon the sample. All samples were kept at 37°C within a Peltier Solvent Trap and Evaporation Blocker to prevent them from drying out. Rheological testing was performed at a frequency of 1 Hz across a range of strains from 0.01% to 20%. The storage modulus, G′, represents the elastic properties of the scaffold and was calculated by taking the plateau value at the linear viscoelastic region at low strains (between 0.01% and 1%). Storage modulus was compared between groups (Fig. 5).

Figure 5.

Rheological testing on nanofiber hydrogel composite AV loop flaps (NHC AVL) (A) and collagen-elastin AV loop flaps (CE AVL) as well as implanted control scaffolds (NHC Co and CE Co) and nonimplanted controls (NHC, CE) immediately after explantation. Rheological testing was also performed on four adipose tissue samples collected from a human abdominal pannus specimen from a patient undergoing abdominoplasty (B) Comparison of storage modulus (G′) between the groups (C). Color images are available online.

Quantitative image analysis

Quantification of ink-filled blood vessels within the scaffolds around the AV loop main vessels was performed in three sections per animal using an observer-independent algorithm especially developed for this analysis (previously described in detail²¹) using the MATLAB software (Mathworks, Natick, MA).

Assessment of endothelial cell (EC) protein expression was performed with the open source image analysis software ImageJ (NIH, Bethesda, MD). Images covering the entire AV loop vessel cross sections were obtained at 63 × magnification under standardized conditions for white balance and exposure time (161.2 ms) (Axio Vision 4; Carl Zeiss Microscopy). Images were then cropped to multiple nonoverlapping areas of 165 × 220 μm covering the entire endothelium of one vessel cross section. After conversion of the RGB images into grayscale, 24 ECs per cross section were manually selected as regions of interest (ROI). Integrated Density (IntDen, defined as the product of area [μm²] and mean gray value) was determined in each of the 24 ROIs and means were calculated. Average IntDen of all animals per group were compared between the groups.

Protein expression within the scaffold was analyzed using a code written in MATLAB (Mathworks) by the authors, adapted from previous image analysis studies by one of the authors (K.C.).²² Images of 20× magnification were acquired. Four ROIs of 400 × 400 μm were defined on four locations around the main vessel cross section. First, the RGB images of the immunostained slides were separated into DAB- and hematoxylin channels with color deconvolution using optical density vectors for DAB and hematoxylin (DAB red, green, blue [RGB] vector of [0.27, 0.57, 0.78] and HE RGB vector of [0.18; 0.20; 0.08]). The DAB image was converted to grayscale. To remove any ink within the vessel cross sections from the image, pixel coordinates with values <18 (out of a maximum of 255) were deemed “ink” and removed from the analysis. Finally, the average pixel intensity was calculated by summing the total pixel gray value across the image by the number of pixels within the image (Fig. 3B, C).

Figure 3.

(A) Protein expression of SYNJ2BP, EPHA2, and FOXC1 in the scaffolds expressed as mean integrated density (μm² × pixels). (B) Examples of MATLAB image analysis showing color deconvolution into DAB and hematoxylin channels as well as (C) removal of black ink within vessel cross sections. CE, collagen-elastin dermal substitute; CE Co, control of collagen-elastin dermal substitute without AV loop; DAB, 3,3′-diaminobenzidine; NHC, nanofiber hydrogel composite; NHC Co, control of nanofiber hydrogel composite without AV loop. Color images are available online.

Analysis of fiber alignment within the scaffolds of CE and NHC flaps was performed on scanning electron microscope images of 2,000× magnification using the custom software MatFiber that is based on an intensity-gradient-detection algorithm originally developed by Karlon et al.²³ and previously used by the authors.²² In brief, we used this algorithm to analyze the angle of alignment and strength of alignment (mean vector length, or MVL) of subregions (90 pixels in area) across the image. The MVL ranges from a value of 0, representing completely random alignment, to 1, representing completely aligned fibers. The overall strength of alignment of the fibers were calculated as mentioned previously.²²

Statistical analysis

Statistical analysis was performed using Prism 8 (GraphPad, San Diego, CA). Continuous variables were compared using an unpaired Student's t-test or one-way analysis of variance with Tukey's multiple comparisons test. Normal distribution of the data was confirmed using the D'Agostino–Pearson omnibus K² test. Data are presented as means ± standard error of the mean. p < 0.05 was considered statistically significant.

Results

Postoperative outcomes

All surgeries were completed without any complications by anesthesia. All animals of the HA-only group developed AV loop thrombosis and exhibited macroscopically visible shrinking and degradation of the scaffold compared with NHC and CE scaffolds and were not used for further analysis. NHC and CE flaps preserved their shape for a period of 15 days without gross evidence of shrinking (Fig. 1E). One animal of the NHC group was excluded due to AV loop thrombosis. In the CE group, one animal was excluded due to a wound infection triggered by automutilation.

Axial vascularization of tissue-engineered flaps

Functional neovascularization within the scaffolds detected by black ink perfusion was visible in both NHC and CE scaffolds on POD 15 after AV loop creation. Vascularization of the scaffolds defined by mean count of ink-perfused vessel cross sections did not show statistically significant differences between NHC and CE flaps (CE: 375.24 ± 54.71 vs. NHC: 298.0 ± 65.49; p = 0.32) (Fig. 1D).

Angiogenesis-related protein expression in vascular endothelium and scaffolds

Immunohistochemical analysis of SYNJ2BP expression showed a significantly lower expression in AV loop ECs in the NHC group compared with the CE group (mean IntDen: 4134.0 ± 351.44 vs. 5525.67 ± 469.52, p < 0.05) (Fig. 2, Table 1). SYNJ2BP expression trended to be lower in NHC and CE scaffolds from AV loop constructs compared with the control constructs without AV loops (Co-NHC, CE-Co); however, statistical significance was not reached (Fig. 3A).

Figure 2.

Comparison of protein expression of SYNJ2BP, EPHA2, and FOXC1 in AV loop ECs (white arrows). IntDen = integrated density (μm² × pixels). The black ink appears detached from the luminal surface due to processing artifacts on some images. *p < 0.05. ECs, endothelial cells; EPHA2, ephrin receptor kinase 2; FOXC1, forkhead box C1; SYNJ2BP, synaptojanin-2 binding protein. Color images are available online.

Table 1.

Protein expression in endothelial cells

Protein	Integrated Density (μm2 × pixels) AV Loop Endothelium
	NHC AVL		CE AVL		p
	Mean	SEM	Mean	SEM
SYNJ2BP	4134.0	351.44	5525.67	469.52	0.03
EPHA2	33331	368.65	5317.67	792.07	0.03
FOXC1	7527.20	1632.67	3197.56	590.73	0.03

Protein expression is shown as mean integrated density ± SEM.

The bold p values are statistically significant.

AVL, arteriovenous loop; CE, collagen-elastin scaffold; EPHA2, ephrin receptor kinase 2; FOXC1, forkhead box C1 (FOXC1); NHC, nanofiber hydrogel composite; SEM, standard error of the mean; SYNJ2BP, synaptojanin-2 binding protein.

EPHA2 expression was significantly downregulated in ECs from NHC constructs compared with CE constructs (mean IntDen: 33,331 ± 368.65 vs. 5317.67 ± 792.07; p < 0.05) (Fig. 2, Table 1). Within the scaffold no significant differences for EPHA2 expression were found (Fig. 3A).

A significant upregulation of FOXC1 was found in ECs of NHC constructs compared with CE constructs (7527.20 ± 1632.67 vs. 3197.56 ± 590.73; p < 0.05) (Fig. 2, Table 1). In the scaffold, the highest expression of FOXC1 was found in CE controls without AV loop (CE-Co; mean IntDen: 1.44 × 10⁷), whereas FOXC1 expression was significantly lower in both groups of AV loop constructs (NHC: 5.58 × 10⁶, CE: 3.74 × 10⁶; p < 0.01; Fig. 3) (Tables 2 and 3).

Table 2.

Protein expression within the scaffolds

Protein	Mean Integrated Density (IntDen)
	NHC AVL	CE AVL	NHC Co	CE Co	p (NHC AVL vs. NHC-Co)	p (NHC AVL vs. CE AVL)	p (NHC AVL vs. CE Co)	p (NHC Co vs. CE AVL)	p (NHC Co vs. CE Co)	p (CE AVL vs. CE Co)
SYNJ2BP	6.10 × 106 (1.41 × 106)	3.39 × 106 (7.69 × 105)	8.42 × 106 (1.85 × 106)	6.53 × 106 (1.22 × 106)	0.62	0.36	>0.99	0.06	0.82	0.38
EPHA2	8.23 × 106 (1.9 × 106)	8.12 × 106 (2.14 × 106)	1.08 × 107 (9.32 × 105)	1.06 × 107 (1.55 × 106)	0.81	>0.9	0.84	0.8	>0.9	0.83
FOXC1	5.58 × 106 (1.43 × 106)	3.74 × 106 (9.79 × 105)	9.0 × 106 (1.79 × 10)	1.44 × 107 (2.07 × 106)	0.40	0.76	0.005	0.12	0.20	0.001

Protein expression is shown as mean integrated density ± SEM.

The bold p values are statistically significant.

CE Co, control of collagen-elastin scaffold without arteriovenous loop; NHC Co, control of nanofiber hydrogel composite without arteriovenous loop.

Table 3.

Storage moduli (Pa) from rheological testing

Storage Modulus G′(Pa)
	NHC	NHC Co	NHC AVL	CE	CE AVL	CE Co	Human Fat
Mean	1156.75	313.50	571.69	1076.33	1164.29	446.50	387.43
SEM	194.64	96.62	43.03	203.12	219.17	49.43	230.64
p vs. NHC		0.005	0.07	1.0	1.0	0.02	0.04
p vs. NHC Co	0.005		0.69	0.01	0.005	0.97	1.0
p vs. NHC AVL	0.07	0.69		0.14	0.06	0.98	0.96
p vs. CE	1.0	0.01	0.14		1.0	0.04	0.07
p vs. CE Co	0.02	0.97	0.98	0.04	0.02		1.0
p vs. CE AVL	1.0	0.005	0.06	1.0		0.02	0.03

The bold p values are statistically significant.

Data are shown as mean integrated density ± SEM.

Ultrastructure of tissue-engineered flaps

Scanning electron microscope images of the native scaffolds have been previously published by us (NHC)¹⁹ and others (CE).²⁴ Scanning electron microscope of the tissue-engineered flaps revealed a strong infiltration of mononuclear cells into the NHC AV loop constructs. Mononuclear cells were found to be attached to the fibrillary nanofibers. The ultrastructure of CE scaffolds showed a mesh-like appearance with sparse mononuclear cell infiltration (Fig. 4). Analysis of fiber alignment of the 2,000× images using the MatFiber algorithm showed that both constructs demonstrated modest alignment (MVL: NHC AV loop: 0.41 ± 0.06; CE AV loop: 0.28 ± 0.05). The NHC scaffolds trended toward higher alignment, but statistical significance was not reached (p = 0.12) (Fig. 4).

Figure 4.

(A) Ultrastructure of the prefabricated AV loop constructs composed of nanofiber hydrogel composite (NHC) and CE dermal substitute in 1,000 × , 2,000 × , and 5,000 × magnification determined by scanning electron microscopy. MC infiltration along the nanofibers (arrows) is visible in the NHC constructs. (B) Analysis of fiber alignment of the 2,000 × images using the MatFiber algorithm. MC, mononuclear cell. Color images are available online.

Rheological analysis

Rheological analysis of the elastic properties of the scaffolds showed that the nonimplanted CE and NHC scaffolds had a storage modulus (G′) at around 1,000 Pa, which was significantly higher compared with both implanted control scaffolds with saphenous vein grafts (NHC-Co, CE-Co) and human fat specimens that demonstrated G′ of around 500 Pa (p < 0.05). Implantation of an AV loop into the CE scaffold preserved the mechanical properties of the nonimplanted CE scaffold. By contrast, implantation of an AV loop into the NHC showed a significantly lower G′ compared with the nonimplanted NHC and did not significantly differ from the implanted control scaffold (NHC-Co). Interestingly, the NHC AV loop had mechanical properties closely matching native human adipose tissue (∼500 Pa) (Table 3 and Fig. 5).

Discussion

Currently available options for reconstruction of large soft-tissue defects have considerable limitations. Autologous free tissue transfer can cause significant donor site morbidity, whereas a combination of acellular dermal substitutes in conjunction with skin grafts is only suitable for small and superficial defects. Therefore, the need for new approaches for soft-tissue reconstruction of large defects from the field of tissue engineering is clearly evident. Prevascularization of acellular scaffolds by an AV loop has shown to be a promising technique for the creation of tissue units that can be autologously transplanted through microvascular anastomosis of the AV loop pedicle to recipient vessels at the defect site, thereby allowing reconstruction of defects with exposed bone in rat models.¹⁵

The ideal scaffold for soft-tissue engineering should mimic the characteristics of the ECM and fulfill several criteria: It should provide a suitable microenvironment for sustaining a sufficient neovascularization, allowing it to be transplanted as a tissue unit; it must maintain its size and form for a period of several weeks during the prevascularization period, and it should exhibit biomechanical properties similar to native soft tissue, allowing it to be molded to create three-dimensional soft-tissue flaps that specifically fit different defect geometries. A substantial limitation of commercially available dermal substitutes is their form stability and the fact that they are produced as rather rigid sheets with limited pliability.

Electrospun nanofibers have demonstrated a multitude of benefits for applications in tissue engineering. They can mimic ECM fibers and thereby improve biomechanical properties of scaffolds and act as delivery vehicles for bioactive agents with minimal impact on drug activity, while maintaining a sustained drug release.²⁵ In this study, we use an established in vivo vascularization model to investigate the biomechanical properties, neovascularization, and angiogenesis-related protein expression within a novel nanofiber hydrogel composite (NHC).

The incorporation of PCL nanofibers into HA hydrogels enabled us to create a scaffold with favorable biomechanical properties for flap fabrication. In contrast to HA-only scaffolds that exhibited volume loss and shrinking, the NHC constructs remained stable for a period of 15 days. A solid neovascularization developed within NHC scaffolds that extended to the margins of the tissue units and was noninferior to CE scaffolds that are the current standard scaffold for flap tissue engineering in our laboratory. We previously compared neovascularization in bioengineered AV loop flaps based on CE scaffolds (MatriDerm) and bovine collagen and chondroitin-6-sulfate (Integra; Integra LifeSciences, Plainsboro Township, NJ) and demonstrated that the collagen/chondroitin-6-sulfate constructs exhibited significantly less vascularization after 14 days.¹⁴ Fibrin has also been studied as a scaffold for AV-based tissue engineering; however, it was proven to be less suitable since it leads to contraction and volume loss within a period of 15 days.²⁶ Although Matrigel has demonstrated superior integration and stability into native tissue compared with fibrin, it cannot be translated into clinical applications, due to oncological safety concerns since it is derived from mouse sarcoma tissue.²⁶

Tissue-engineered flaps based on the novel NHC scaffold exhibited favorable mechanical properties with a storage modulus similar to human fat. By contrast, flaps created using the CE scaffold were significantly stiffer than NHC flaps and the implanted control scaffolds, with a storage modulus similar to nonimplanted CE scaffolds. The presence of the AV loop and subsequent angiogenesis as well as cell migration into the CE scaffold likely maintains the base mechanical properties of the scaffold, whereas the scaffold surrounding the nonangiogenic saphenous vein decreased in mechanical strength over time. Interestingly, the stiffness of the NHC AV loop flaps only marginally increased after implantation compared with their controls without an AV loop, although the flaps exhibited a comparable amount of neoangiogenesis and a higher cell infiltration along the nanofiber compared with the CE scaffold. This highlights the favorable biomechanical properties of the NHC scaffold for soft-tissue engineering.

We previously investigated miRNA and gene expression profiles by microarray analysis in venous tissue samples from rat AV loops and found deregulations of several genes related to angiogenesis, including SYNJ2BP, EPHA2, and FOXC1. These genes showed strong inverse correlations and predicted target interactions with distinct miRNAs, thus seem to be of paramount importance in the process of flow-induced angiogenesis.¹⁶ SYNJ2BP enhances the stability of the Notch ligands Delta-like 1 and 4 and promotes Notch signaling by increasing the expression of Notch targets, such as HEY1, thereby inhibiting EC migration, proliferation, and angiogenic sprouting.²⁷ In tissue samples from rat and also from human AV loops we previously found a significantly reduced SYNJ2BP gene expression compared with native venous tissue.^13,22 Immunohistochemical analysis of the prefabricated flap constructs in this study demonstrated a trend toward lower SYNJ2BP protein expression in the scaffold of AV loop constructs compared with their controls without an AV loop. Statistical significance was marginally missed, presumably due to the sample size. These data indicate that the previously observed reduction in SYNJ2BP gene expression in AV loops also extends to the protein level and indicates its importance for AV loop-associated angiogenesis. SYNJ2BP expression was significantly downregulated in AV loop ECs from NHC constructs compared with CE constructs. These findings suggest that the composition of the scaffold material has an impact on protein expression in ECs and the development of angiogenesis.

EPHA2 belongs to the Ephrin family of receptor tyrosine kinases and is an epithelial cell receptor involved in the regulation of angiogenesis in various tumors²⁸ and moreover increases vascular permeability.²⁹ EPHA2 has been shown to affect the interaction between pericytes and ECs and EPHA2 deficiency in mice leads to a reduced number of airway capillaries.³⁰ However, EPHA2 deficiency can also have proangiogenic effects as shown by Okazaki et al. for certain inflammatory conditions.³⁰ Moreover, Eph receptors have been reported to enhance the phosphorylation of focal adhesion kinase, a key regulator in mechanotransduction.³¹

Our previous data on gene expression from rat AV loop vascular tissue indicated a downregulation of EPHA2 expression¹⁶ compared with venous control tissue. EPHA2 expression in ECs from NHC constructs was significantly downregulated compared with CE constructs, indicating the contribution of the scaffold material to angiogenic vascular protein expression. No significant differences were evident within the scaffolds between the groups, which is consistent with EPHA2 mainly being an epithelial cell marker.

Constitutional loss-of-function mutations in the transcription factor FOXC1 cause Axenfeld–Rieger syndrome, which is characterized by pathological corneal vessel growth. The crucial role FOXC1 plays in vascular development depends on the regulation of two competing molecular mechanisms: It inhibits the expression of matrix-metalloproteinases (MMPs), thereby limiting the bioavailability of vascular endothelial growth factor (VEGF), which is cleaved from the ECM by MMPs. The proangiogenic effect of FOXC1 is mediated by an inhibition of the antiangiogenic soluble VEGF receptor 1 (sVEGFR1).³²

In our previous study that investigated vascular gene expression in AV loops using microarrays, we found a significantly reduced expression of FOXC1 compared with venous control tissue.¹⁶ Thus, these data are in line with our current immunohistochemical data, showing a reduced FOXC1 expression in NHC and CE compared with the control groups without AV loops, although statistical significance was only reached in the CE group. With regard to FOXC1 expression in ECs, we found a significantly higher FOXC1 expression in AV loop ECs in NHC compared with CE scaffolds. As we have previously shown that FOXC1 gene expression within AV loop vessels is related to the expression of miRNA-511-3p,¹⁶ miRNA-mediated post-transcriptional silencing of FOXC1-mRNA might be the cause of the observed differences between gene and protein expression in the AV loop vessels.

Data from murine models indicate cell type-specific effects of FOXC1 on angiogenesis: Mice with FOXC1 null mutations show an excessive corneal blood vessel growth, whereas specific deletions of FOXC1 in murine vascular ECs lead to corneal avascularity.³³ Our data show that the scaffold composition and likely the presence of nanofibers has an impact on FOXC1 expression in AV loop ECs; however, the exact role FOXC1 plays in flow-induced angiogenesis has to be further investigated.

In a rabbit soft-tissue defect model, our group has previously shown that the NHC promotes a stronger infiltration of macrophages and polarization toward a proregenerative M2-phenotype compared with HA-hydrogels without nanofibers.¹⁹ These findings are reflected in our scanning electron microscope analysis that demonstrates a stronger mononuclear cell infiltration along the nanofibers as well as a trend toward a higher fiber alignment in the NHC compared with the CE scaffold.

Limitations of this study are mainly related to its sample size, which, however, is comparable with previously published studies in this field and was in accordance with the Animal Welfare Act. We show a comparably solid neovascularization in both examined scaffold types after 15 days of implantation; however, differences in vascularization at later stages still have to be determined in future studies. We specifically decided to explant the constructs on POD 15, since vascularization within AV loop constructs has proven to be sufficient for free transplantation and integration into the defect area in previous studies on a rat defect model.¹⁵

Innovation

This study demonstrates the feasibility of tissue engineering of axially vascularized soft-tissue free flaps using an injectable NHC together with an AV loop, which, in contrast to commercially available collagen-based scaffolds, recapitulates the biomechanical properties of native human fat. The NHC examined in this study has the great benefit of being moldable, while also showing sufficient stability and vascularization similar to CE scaffolds. The consistency of the NHC is especially suitable for custom-made flap engineering for variable defect geometries with the use of three-dimensional-printed isolation chambers, which we are currently investigating in our laboratory. Owing to their favorable biomimetic and biomechanical properties, nanofiber scaffolds, in combination with intrinsic vascularization and a vascular pedicle for microsurgical anastomosis provided by an AV loop, constitute a promising therapeutic strategy for reconstruction of soft-tissue defects in patients with limited donor site availability. Future confirmatory studies in large animal models are required to achieve translation into future clinical applications.

KEY FINDINGS

• A combination of HA hydrogels cross-linked with PCL nanofibers and axial vascularization induced by an AV loop constitutes a promising approach for tissue engineering of soft-tissue free flaps.

• Nanofiber hydrogel-based flaps develop a functional neovascularization, remain stable for a period of 15 days after implantation, and exhibit similar rheological properties as native human fat.

• The scaffold material affects the protein expression of SYNJ2BP, EPHA2, and FOXC1 in ECs and within the scaffold of AV loop-based flaps, thus indicating its importance for tissue engineering.

Abbreviations and Acronyms

AV

arteriovenous

CE

collagen-elastin

DAB

3,3′-diaminobenzidine

EC

endothelial cell

ECM

extracellular matrix

EPHA2

ephrin receptor kinase 2

FOXC1

forkhead box C1

HA

hyaluronic acid

HMDS

hexamethyldisilazane

miRNA

microRNA (micro ribonucleic acid)

MMPs

matrix-metalloproteinases

MVL

mean vector length

NHC

nanofiber hydrogel composite

PEG

polyethylene glycol

PTFE

polytetrafluoroethylene

ROI

region of interest

SEM

standard error of the mean

SYNJ2BP

synaptojanin-2 binding protein

VEGF

vascular endothelial growth factor

Author Disclosure and Ghostwriting

S.K.R., J.M.S., and H.Q.M. are inventors on intellectual property covering the technology that has been filed by Johns Hopkins and have equity positions in LifeSprout, a privately held company that is seeking to bring these advances into clinical development. The content of this article was written by the authors listed. No ghostwriters were used to write this article.

About the Authors

Dominic Henn, MD, is a plastic surgery resident at BG Trauma Center Ludwigshafen and a postdoc at the Hagey Laboratory for Pediatric and Regenerative Medicine at Stanford University. Kellen Chen, PhD, is postdoc at the Hagey Laboratory for Pediatric and Regenerative Medicine at Stanford University. Annika Rauh, MD, is a plastic surgeon at BG Trauma Center Ludwigshafen. Katharina Fischer and Patricia Niedoba are medical students. Yoo-Jin Kim, MD, is a professor of pathology. Matthias Hannig, DMD, is a professor of dentistry with a research focus on electron microscopy. Russell A. Martin, PhD, is a postdoc at the Department of Materials Science and Engineering at Johns Hopkins University. Hai-Quan Mao, PhD, is a professor of materials science and engineering at Johns Hopkins University and associate director of the Institute of NanoBioTechnology (INBT). Sashank K. Reddy, MD, PhD, is an assistant professor of plastic surgery at Johns Hopkins University. Ulrich Kneser, MD, is the chairman of the department of hand, plastic, and reconstructive surgery, BG Trauma Center Ludwigshafen. Geoffrey C. Gurtner, MD, is the Johnson and Johnson distinguished professor of surgery and professor (by courtesy) of bioengineering and materials science at Stanford University. Justin M. Sacks, MD, MBA, FACS, is an associate professor of plastic surgery at Johns Hopkins University. Volker Schmidt, MD, is the head of the department of plastic and breast surgery at Zealand University Hospital, Roskilde, Denmark.