In-vivo assessment of a tissue engineered vascular graft computationally optimized for target vessel compliance

Abstract

Tissue engineered vascular grafts (TEVGs) have the ability to be tuned to match a target vessel’s compliance, diameter, wall thickness, and thereby prevent compliance mismatch. In this work, TEVG compliance was manipulated by computationally tuning its layered composition or by manipulating a crosslinking agent (genipin). In particular, these three acelluluar TEVGs were compared: a compliance matched graft (CM_gel - high gelatin content); a hypocompliant PCL graft (HYPO_pcl - high polycaprolactone content); and a hypocompliant genipin graft (HYPO_gen - equivalent composition as CM_gel but hypocompliant via increased genipin crosslinking). All constructs were implanted interpositionally into the abdominal aorta of 21 Sprague Dawley rats (n=7, males=11, females=10) for 28 days, imaged in-vivo using ultrasound, explanted, and assessed for remodeling using immunofluorescence and two photon excitation fluorescence imaging. Compliance matched grafts remained compliance-matched in-vivo compared to the hypocompliant grafts through 4 weeks (p<0.05). Construct degradation and cellular infiltration was increased in the CM_gel and HYPO_gen TEVGs. Contractile smooth muscle cell markers in the proximal anastomosis of the graft were increased in the CM_gel group compared to the HYPO_pcl (p=0.007) and HYPO_gen grafts (p=0.04). Both hypocompliant grafts also had an increased pro-inflammatory response (increased ratio of CD163 to CD86 in the mid-axial location) compared to the CM_gel group. Our results suggest that compliance matching using a computational optimization approach leads to the improved acute (28 day) remodeling of TEVGs. To the authors’ knowledge, this is the first in-vivo rat study investigating TEVGs that have been computationally optimized for target vessel compliance.

Graphical Abstract

1. Introduction

Compliance mismatch is often cited as a cause of vascular graft failure, however the majority of clinical and preclinical grafts or tissue engineered vascular grafts (TEVGs) tend to overlook this critical design requirement [1, 2, 3]. Most TEVGs cellularized or acellular tend to be hypocompliant or less compliant than the target native vasculature [4, 5, 6, 7, 8, 9]. This decrease in compliance has been cited as the source for hemodynamic environmental changes or non-homestatic mechanical strains that result in the development of intimal hyperplasia, thrombus formation, and ultimately graft failure [10, 11, 12, 2]. Bypass surgeries in the coronary and peripheral arteries (< 5 mm diameter) are most affected by compliance mismatch and ultimately lead to high failure rates [10, 13]. The current gold standard for coronary bypass surgery is the autologous blood vessel (saphenous vein or mammary artery) which is less compliant, causes donor morbidity, is often unavailable due to pre-existing disease or prior harvesting, and fails in over 5% of patients each year [14, 15]. Synthetic grafts composed of Dacron and polytetrafluoroethylene used in the peripheral arteries are hypocompliant and also show slightly worse patency rates and a lower cellular infiltration than autologous grafts[16, 17, 18].

While it is clear that hypocompliant grafts are not ideal for bypass surgery, varying the compliance of the TEVG could be a way to help further increase patency rates and graft remodeling. Past studies have successfully varied compliance by modifying the TEVG’s composition or crosslinking agents (concentration or time) [1, 9, 19]. Generally, most in-vitro compliance studies have shown the positive effects of matching the TEVG’s compliance to the native vascular rather than using a hypocompliant graft. Specifically, compliance matched TEVGs have been shown to positively affect SMC phenotype, increase M2 macrophage infiltration, increase extracellular matrix formation, and promote proper cellular function [20]. While these in-vitro results are encouraging, preclinical studies often fail to isolate the effects that compliance has on TEVG design and remodeling [9, 21]. Specifically, these studies tend to look at a singular group or fail to incorporate the necessary experimental groups to test compliance as a major contributing factor.

TEVGs that can be tuned for biomechanical response by design offer a potential solution to test the effect of compliance on TEVG design and in-vivo remodeling. In this work, we utilize two approaches to tune the compliance of a TEVG: graft composition and crosslinking concentration. For composition, our laboratory has developed a computational/experimental approach that characterized the mechanical properties of gelatin:polycaprolactone blends and used them to develop a repeating multi-layer graft formulation that was electrospun and crosslinked with genipin to meet a target compliance [22]. Each graft contains an optimized percentage of gelatin and polycaprolactone (PCL) providing it improved biocompatibility and long-lasting mechanical strength, respectively [23]. The repeating multi-layer composition throughout the construct allows for varied infiltration/degradation, thickness, and structural support which can be tailored for compliance to match the host vasculature. The crosslinking agent genipin was selected to slow gelatin degradation and act as an alternative to glutaraldehyde, thus reducing the potential for calcification and cytotoxicity [24, 25, 26, 27, 28]. This approach was also able to create both compliance-matched (CM_gel - high gelatin content) and hypocompliant (HYPO_pcl - high PCL content) TEVGs [22]. As mentioned earlier, we also varied graft compliance by increasing the concentration of the genipin crosslinker (HYPO_gen - high genipin concentration) to match the compliance of HYPO_pcl constructs. The hypocompliant groups both had a similar compliance to synthetic grafts (Dacron and polytetrafluoroethylene) [29]. Our ability to tune the compliance of TEVGs in a controlled manner using either composition or crosslinking allows the direct investigation of compliance on TEVG functional performance.

For this study, the abdominal aorta of the Sprague Dawley rat was chosen for graft implantation as a cost effective, well documented, and high throughput model to test the acute (1 month) remodeling and performance of our compliance-matched TEVG [30, 31]. Our three acellular grafts (HYPO_pcl, CM_gel, & HYPO_gen) were implanted for 28 days, imaged in-vivo with ultrasound, and explanted for histological imaging. The explant was evaluated for smooth muscle cells, highly proliferative cells, macrophages (pan, M1, & M2), collagen, elastin, and endothelial cells using quantitative immunohistochemistry. Comparisons were made between graft type, location (proximal/middle), and region (lumen/graft/abluminal). The overall goal of this study was to evaluate how vascular graft compliance and composition affects acute remodeling, degradation, endothelialization, and cellular infiltration following 28 days in a rat aortic interpositional model.

2. Methods

2.1. TEVG Characterization and Fabrication

2.1.1. Preparation of Rat Aorta

Rat aorta was extracted and prepared for compliance testing. All tissues and protocols were approved by the University of Pittsburgh Institutional Animal Care and Use Committee. The infrarenal abdominal aorta of 13 male Sprague Dawley rats (175–225g) were extracted and any adventitial connective tissue was removed gently with forceps. The aorta, 1.5 cm in length, was then placed in phosphate-buffered saline (PBS) pH 7.4 (Thermo Fisher Scientific, Waltham, MA) with 1% (v/v) Gibco Penicillin-Streptomycin 10,000 U/mL (Thermo Fisher) and 1% Gibco Amphotercin B (Thermo Fisher) for 24–48 hours post mortem at 4°C before beginning of compliance testing to prevent contamination of the tissue.

2.1.2. Compliance Testing

The compliance of rat aorta was tested using our custom microbiaxial optomechanical device (MOD) [19, 32, 33, 34, 35, 36, 37]. This device has been extensively used in our laboratory to mechanically characterize tubular and planar vascular tissues and biomaterials. Briefly, each sample was cannulated on opposing ends using a 1 mm outer diameter (OD) glass capillary, tied with suture, and sealed using super glue (Loctite). The cannulated sample was then mounted inline within the MOD bath that was filled with saline, kept at 37°C. The arteries were preconditioned circumferentially (0 to 120 mmHg) and axially stretched (15% strain) seven times prior to the collection of final diameter and pressure. The outer diameter (OD) of each sample was measured using a digital camera as intraluminal pressure was slowly increased from 0 to 120 mmHg. Compliance was quantified as

$$\mathit{\text{Compliance}}=\frac{\left(O{D}_{120}-O{D}_{70}\right)/O{D}_{70}}{50\text{mmHg}}.$$ (1)

2.1.3. Compliance Matching and Computational Optimization Routine

Specific construct formulations were determined using our developed computational optimization routine [22, 38]. Previously, our research group has developed a finite element optimization scheme (Figure 1A) to predict the thickness and Gelatin:PCL (G:P) ratio of an alternating layered construct that matches the compliance and geometry of rat aorta [22, 38]. Briefly, tubular TEVGs consisting of either 20G:80P, 50G:50P, and 80G:20P were fabricated and tested using the MOD device. Each construct was pressurized from 0 to 120 mmHg at 0, 5, 10, and 15% axial strain. From this testing, the material constants were collected from a Fung-type constitutive model and assign to gelatin and PCL, in silico. Next, a computational optimization scheme was used to determine the TEVG’s formulation to match a specified compliance (hypocompliant or compliance matched). For this study, each construct consisted of 12 total layers with two alternating polymeric formulations. The optimization scheme uses Matlab (MathWorks Inc, USA) to generate an input file for the ABAQUS finite element solver (Dassault Systemes Simulia, France) which consisted of a four-node, reduced-integration, axisymmetric finite element model that simulated the compliance of the construct from 0 to 120 mmHg. For computational optimization, the constrained parameters were the construct’s inner diameter, number of layers, and total thickness. The open design parameters were the G:P ratio and the thickness of each of the paired layers. The optimization function modified the open design parameters until the ABAQUS simulation matched the specified target compliance to within 1%.

Figure 1:

A) Overview of the optimization routine; B) Electrospinning setup, demonstrating the two-nozzle setup of the device; C) Implantation of a CM_gel TEVG.

2.1.4. Electrospinning Materials and fabrication

All TEVGs were fabricated using an electrospinning process, crosslinked in genipin, and tested for compliance. Polycaprolactone with a molecular weight 80,000 (Sigma-Aldrich, St. Louis, MO) and gelatin extracted from porcine skin (Sigma-Aldrich) were dissolved in 1,1,1,3,3,3-hexafluoro-2-propanol (HFP) (Sigma-Aldrich) to create a 10% w/v solution. All constructs were electrospun using IME technologies electrospinning device (Waalre, Netherlands). Briefly, electrospinning solutions were loaded into 5 ml BD syringes (BD Franklin Lakes, NJ), placed in a computer-controlled syringe pump, and connected to 1mm PTFE tubing. This setup was connected to a translating stage, 300 mm/s, and was expelled through a 0.6 mm ID stainless steel tip at a working distance of 10 cm with a climate-controlled chamber at 25°C and 30% humidity. The polymeric solutions were dispensed at a rate of 50 μL/min onto a 1.1 mm diameter steel rod rotated at 300 rpm with a 15-kV voltage difference. The two polymer formulations were sequentially dispensed horizontally and vertically (Figure 1B) from the IME electrospinning device which allowed for a multi-layer composition. During the electrospinning process, the syringe with the higher gelatin formulation was spun from the vertical nozzle. It should also be noted that during the electrospinning process, the user waited 4 minutes before starting the next layer. This waiting period allowed the inline pressure to dissipate and extrude all the material. Once completed, HYPO_pcl and CM_gel were crosslinked in 0.5% (w/v) concentration of genipin (Wako Chemicals USA Inc, USA) in 200 proof ethanol for 24 hours at 37°C and shaken [22].

After crosslinking, half of the construct (1.5 cm) was compliance tested in our MOD device to confirm the construct met the desired compliance. Details regarding compliance testing are found in Section 2.1.2. Once confirmed, this half of the sample was embedded in O.C.T. Compound (Thermo Fisher Scientific, USA), frozen, cross-sectioned (10 μm) with Microm HM550 cryostat microtome (Thermo Fisher Scientific, USA), and imaged using bright field microscopy. Three locations along the length of the sample were averaged to determine the construct area, inner radius, and thickness using a Nikon 90i Eclipse fluorescence microscope (Nikon USA, USA). If the tested sample met the desired compliance, the other half of the sample (not tested) was implanted into the abdominal aorta of a Sprague Dawley rat.

The formulations of the compliance-matched and hypocompliant (stiff) grafts were determined using the optimization scheme detailed in Section 2.1.3. The compliance-matched (CM_gel) construct was compliance matched to rat aorta with a high gelatin content (layers composed of approximately 75G:25P and 50G:50P). The hypocompliant formation (HYPO_pcl) was composed of a high amount of PCL (layers composed of approximately 20G:80P and 35G:65P) and was targeted to 50% of the compliance value of the CM_gel grafts. This hypocompliant graft had a compliance similar to Dacron or ePTFE [29]. Both CM_gel and HYPO_pcl grafts were crosslinked using 24 hours of 0.5% genipin. To further isolate the effect of compliance in grafts with larger gelatin percentages, a third group was added to the study - HYPO_gen. For this formulation we initially tested the CM_gel graft formulation in 0.5%, 1%, and 10% w/v genipin concentrations crosslinked for 24 hours at 37°C and found that the CM_gel formulation had a compliance similar to HYPO_pcl when crosslinked in 10% w/v genipin. The hypocompliant construct was called HYPO_gen and acted as another control group with a hypocompliant mechanical response, but with a gelatin content similar to the CM_gel group. The detailed construct formulations are provided in Table 1. Layers 1 and 2 (L1 and L2) are the repeating units of the grafts which total the 12 layers.

Table 1:

Polymeric formulations, layer thickness, and crosslinking concentrations for the HYPO_pcl, CM_gel, and HYPO_gen graft formulations.

Experimental Group	Layer 1 Ratio G:PCL	Layer 2 Ratio G:PCL	Individual Layer Thickness (μm)				Genipin Conc. (w/v)
			L1	L2	Repeats (L1+L2)	Total
HYPOpcl	20:80	35:65	8	5	6	78	0.5%
CMgel	75:25	50:50	8	5	6	78	0.5%
HYPOgen	75:25	50:50	8	5	6	78	10%

2.1.5. Suture Retention

The suture retention of each scaffold was determined using an in house custom made uniaxial device. Electrospun tubular constructs (n=6) around 1 cm in length were clamped distally in our uniaxial setup. A single 10–0 nylon (AROSurgical) suture was used to create a single loop approximately 2 mm away from the scaffold’s proximal end. The free ends of the suture were secured to the upper clamp and any excess slack removed from the suture. The sample was then pulled at a rate of 2 mm/min until the suture tore out of the sample. The load displacement curve was used to determine the maximum force each sample held. Differences between groups were compared using a Kruskal-Wallis test.

2.1.6. Burst Pressure

Burst pressure of each scaffold (n=5) was quantified by increasing intraluminal pressure until catastrophic rupture occurred. For each test, the construct was cannulated on both sides with one end connected to a syringe pump (Harvard Apparatus, Holliston, MA) and the other clamped to form a closed loop. Saline was injected into the construct at 50 mL/min until failure. The maximum pressure (Weiss gauge, 0–60 psi) was recorded before failure and taken to be the burst pressure. Statistical differences between groups were determined using a Kruskal-Wallis test.

2.1.7. Cytotoxicity

Each scaffold’s degradation product was evaluated for cytotoxicity using rat aorta smooth muscle cells (RSMC (Lonza)) following closely the methods of Guan et al [39]. Cells were seeded at a density of 4.0 × 10⁴ onto a 24-well tissue culture plate. Eluent from each construct (n=5, 8 mm in length) soaked in PBS for 24 hours was added to each well at 1:10 ratio (eluent:media). The positive control was sterile PBS:media (no degradation products). Cell viability was measured with an MTS assay (Promega) at 48 and 96 hours with media changes occurring ever 48 hours. For the MTS assay, 20μL of MTS solution was added per 100 μL media to each well, incubated for 2.5 hours, and the absorbance at 490 nm was recorded to determine cell viability. Results were compared to the control using a two way ANOVA and normalized.

2.2. Rat Implantation

2.2.1. Rat Interpositional Abdominal Aorta Surgery

TEVGs were implanted into the abdominal aorta of 8-week-old male and female Sprague Dawley rats (total M and F 11 and 10, respectively) and weighed on average (203±37 g) with n=7 for each group. Male and female rats were chosen in order to present results more broadly applicable to the rat population. For each surgery, the rat was anesthetized with isoflurane (3% for induction and 1% for maintenance). An abdominal medial incision (3–4 cm) was made using No. 10 scalpel. The small intestine and the cecum were extracted from the abdominal cavity, retracted to right upper quadrant, and covered with wet gauze. The abdominal aorta was dissected, isolated (1 cm portion above the aortic bifurcation), and generally 1–2 branches of the aorta were ligated and transected. About 1 ml of heparin was administered on the small intestine and into the abdominal cavity to prevent clotting. Two 30 g clips were applied at the iliolumbar artery and the bifurcation to prevent blood flow. The aorta was transected at the center and each lumen was irrigated by saline. For the anastomosis, it was initiated on the proximal vessel with stay sutures, 10–0 nylon, placed at the 3 and 9 o’clock positions. After placing 3–4 stitches on the anterior wall, the vessel was rotated 180^o clockwise to expose the posterior wall and 3–4 stitches were placed on the posterior wall. This same procedure was performed on the distal location. Before the last suture was tied the lumen was irrigated and then the clips were removed. If oozing occurred, pressure was applied gently on the suture line which generally allowed for hemostasis. On some occasions an extra stitch was added when blood leakage was unable to be stopped. Each anastomosis had between 7–10 stitches per end with the average implant length consisting of 5.7±1.1 mm which is shown is Figure 1C. Blood flow was assessed by performing a patency test. Once complete, the small intestine and cecum were returned into the abdominal cavity. The rectus abdominis was closed with an interrupted 4–0 absorbable polyfilament suture, the skin was closed with buried 5–0 nylon suture, and the skin was stapled (to prevent the animals from chewing out their stitches). The average surgery time was around 1.75 hours with the anastomoses taking about 50 minutes. One animal in the CM_gel group was excluded from the study due to an unrelated death within the first three days. All other animals survived the duration of the study with little to no discomfort.

In days 1 to 3 post operatively, each rat received 0.5 mg/kg buprenorphine and 10 mg/kg cefuroxime intramuscularly twice a day to minimize pain and prevent bacterial infections, respectively. Carprofen (5 mg/kg) was also given at 24 and 48 hours post operatively to minimize inflammation. For solid medications, aspirin (pain reliever) and dipyridamole (blood thinner), were ground up with a mortar and pestle and mixed with jelly inside a miniature ice cream cone which also acts as another enrichment object for the animal. For the first week aspirin and dipyridamole were given to the rat at a concentration of 200 mg/kg and 250 mg/kg and decreased at weeks 2–4 to 100 mg/kg and 100 mg/kg, respectively. After 4 weeks, the rats were euthanized with carbon dioxide inhalation and cervical dislocation, the explant was recovered and place in 4% paraformaldehyde solution in PBS (Thermo Fisher Scientific, Waltham, MA), and later sectioned.

2.3. Graft Measurements and Quantification

2.3.1. In-vivo Ultrasound

High frequency Doppler ultrasound was used to evaluate graft patency, diameter, compliance, and velocity in-vivo. All longitudinal ultrasound data was collected using the VEVO 2100 (FUJIFILM VisualSonics Inc, Canada) with a 30 MHz probe. In-vivo measurements were taken for each animal up to a week before surgery and at one, two, and four weeks post-implantation. It should be noted, that the VEVO 2100 was located in another facility which prevented immediate imaging post implantation. To collect the ultrasound data, the rat was anesthetized with isoflurane (3% for induction and 1% for maintenance), abdominal fur was removed using Nair Hair removal (Church & Dwight), applied ultrasound gel, and imaged. Cross-sectional in-vivo imaging was used to determine the aorta’s location. Once located, the implant was imaged longitudinally. Patency was defined as an observable blood flow under ultrasound imaging. Any nonpatent vessel was excluded from subsequent statistical analysis. At each time point, the maximum and minimum diameter of the implant was recorded over one cardiac cycle to determine the implant’s average inner diameter and compliance using Equation 1. The average maximum velocity inside the implant was recorded using Doppler ultrasound and all measurements were taken inside the implant. All collected data was processed using the Vevo Lab software version 1.7.1 and adjusted for the probe angle using the approach described in Park et al. [40].

2.3.2. Preparation, Immunohistochemistry, and Fluorescent Imaging

The proximal and middle locations of each TEVG was evaluated for cell infiltration using immunohistochemistry. Each fixed sample was cut in the middle and separated into proximal and middle sections, frozen in O.C.T. medium, and kept at −80°C. All samples were sectioned at 10 μm using a Microm HM550 cryostat microtome. The middle samples were cut cross-sectionally while the proximal sections were cut longitudinally.

For immunohistochemistry, samples were rehydrated in PBS and soaked in 1% sodium dodecyl sulfate (Sigma-Aldrich, St. Louis, MO) in PBS for 15 minutes. The samples were next permeabilized in 0.3% triton X-100 (Sigma-Aldrich) for 15 minutes and blocked with 0.5% bovine serum albumin (Sigma-Aldrich), 0.3M glycine (Thermo Fisher Scientific), and 20% goat serum (Thermo Fisher Scientific) for one hour. Following blocking, all samples were incubated with a primary antibody over night at 4°C. α-Smooth muscle actin (αSMA, 1:100, Abcam), calponin (CAL, 1:100, Abcam) and smooth muscle myosin heavy chain (MHC, 1:100, Abcam) were used to detect smooth muscle cells. Ki67 was used to detect highly proliferative cells, thus acting as a surrogate for increased risk of intimal hyperplasia. Endothelial cells were detected using CD31 (1:100, Abcam) and macrophages using CD68 (1:100, Abcam), CD86 (1:100, abcam), and CD163 (1:100, Abcam), for inflammatory response, M1, and M2 macrophage phenotype, respectively. Elastin in the implant was detected using elastin alexa fluor antibody (Santa Cruz Biotechnology, Inc) which did not require a secondary antibody. After incubation and washing, the samples were incubated for 1 hour at room temperature with Alexa fluor 488 Goat Anti-Rabbit IgG (Abcam, 1:1000) except for αSMA which was incubated with Alexa Fluor 488 Donkey anti-goat IgG (Abcam, 1:1000). Each sample was then washed and counterstained with DAPI mounting medium (Abcam).

Each marker was imaged using the Nikon 90i Eclipse fluorescence microscope. Briefly, full mosaic images of each entire sample were taken in the DAPI (excitation at 359 nm; emission at 461), FITC (excitation at 495 nm; emission at 519 nm), and Cy5 (excitation at 647; emission at 665) emission channels using a Nikon Plan Apo 10x/0.45 objective. The DAPI channel was used to identify cell nuclei, Cy5 was autofluorescence from the genipin crosslinking/implant, and FITC was the marker of interest. It should be noted that the FITC channel exposure time was consistent between each marker so that quantitative analysis could be performed.

2.3.3. Multiphoton Imaging

Multiphoton microscopy was performed on the Pitt Advanced Intravital Microscope at the University of Pittsburgh and used to evaluate collagen deposition within the implant (cross-sectionally and longitudinally). In this setup an Olympus BX51 upright scanning microscope (Olympus, Tokyo, Japan) was coupled into a 120-fs tunable pulsed laser (INSIGHT DS+DUAL, Newport, Irving CA) and Olympus XLUMPLFL 20x water emersion objective (NA 0.9). The second harmonic generation (SHG) signal of fibrillar collagen was imaged using an excitation of 780 nm and a collection of 377/50 nm. DAPI and autofluorescence was also imaged at 780 nm with a power of 25–31 mW and collected with 525/50 nm and 620/60 nm bandpass filter, respectively. The signal was collected over multiple 499 × 499 μm fields of view with a 5 μm z-step-size through the thickness of the sample. Montage slices of each channel (377, 525, and 620 nm) were converted into maximum intensity projection and used for image quantification.

The percent construct remaining, cells signal inside the construct, and collagen throughout the construct were quantified from each image. Bright field images before implantation (see Section 2.1.4) and multiphoton images of the explant were used to calculate the percent construct remaining: ((T EV G_original−T EV G_explanted)/T EV G_original)* 100. For cells inside the construct, ImageJ (Fiji) was used to isolate the cells inside the construct which were counted using the analysis particle function. The percent collagen throughout the construct was determined similarly to antibody image processing and is described in Section 2.3.4.

2.3.4. Image Processing

Cellular and extracellular markers were evaluated using Matlab image processing to determine differences between groups. It should be noted that the FITC channel was changed to red and the Cy5 to green color. Following faux coloring, all data sets had 3 channels which were as follows: cell nuclei channel (blue), implant channel (green), and antibody or collagen channel (red). Each antibody channel was sampled and then binarized to identify the signal of interest and remove noise. The implant channel was used to separate antibody signal into three regions: lumen, graft, and abluminal. The lumen region was designated as the section from the inner boundary of the implant to the lumen. The graft and abluminal regions were composed of the inside or outside the implant, respectively. The area of the abluminal region was limited to a 25% increase of the exterior diameter of the implant. To calculate the amount of signal per antibody in a certain region, the total number of antibody pixels was divided by the total number of the section’s pixels (lumen, graft, or abluminal). For the cross-sectional samples, the whole implant was used for each sample and averaged among all rats in the specific group. The ratio of CD163 to CD68 signal was also evaluated during this process to help identify differences in macrophage phenotype. The cross-sectional samples were also used to calculate the percent endothelialization and luminal thickness. For percent endothelialization, the CD31 signal length was divided by the lumen circumference and multiplied by 100. For luminal thickness, each sample was measured at 4 locations and evaluated for significant differences using a Box Cox transformation and a mixed effect model.

In the longitudinal sections, the first proximal 150μm (lumen, graft, or abluminal) was processed as described in the paragraph above. The 150 μm portion was selected as this was the shortest segment of the longitudinal samples that were imaged for all groups. The sample size for the longitudinal samples was n=4 due to initial processing. Luminal thickness was measured three times along the length of the image and averaged.

2.3.5. Statistical Analysis

Gender differences were assessed using a paired t-test, however no differences were found and as such gender was not carried through subsequent statistical analysis. After processing, the samples were checked for normality and nonnormal data was transformed using a Box Cox transformation. A linear mixed effect model with a Tukey post hoc was used to statistically analyze how each outcome was influenced by graft type at a specific region in the proximal or middle location. The random effect was the sample and the fixed effects were graft type and region (lumen, graft, and abluminal). The total implant results were evaluated using an ANOVA or Krustal Wallis with a Tukey or Dunn post hoc.

Preliminary work using αSMA marker was used to determine the sample size. Specifically, a power analysis was calculated using an alpha of 0.05 and a power of 80%.

3. Results

3.1. Construct Characterization

Pre-implantation compliance, thickness, and inner diameter using bright field microscopy are shown in Figure 2. The HYPO_gen and HYPO_pcl constructs pre-implantation were found to be stiffer than those in the CM_gel (p<0.05) group and there was no gross delamination between electrospun layers in any construct. The average thickness of each construct was 104.2±15.9, 88.35±11.1, and 114.4±10.1 μm for the HYPO_pcl, CM_gel, and HYPO_gen groups, respectively. The HYPO_gen group was thicker than CM_gel,(p = 0.03) suggesting that genipin was the cause of the increase. The inner radius of the construct was not found to be statistically different across groups. The results of this testing confirmed the desired compliance of our constructs.

Figure 2:

Pre-implantation images, compliance, thickness, inner radius, suture retention, burst pressure, and cytotoxicity of the HYPO_pcl, CM_gel, and HYPO_gen TEVGs. The scale bar is 50 μm.*, p<0.05

The suture retention for HYPO_pcl, CM_gel, and HYPO_gen groups was 43.47, 10.83, and 13.22 g, respectively. The burst pressure for HYPO_pcl, CM_gel, and HYPO_gen groups were 1435, 504, and 497 mmHg, respectively. The suture retention and burst pressure for the HYPO_pcl group was greatly increased compared to CM_gel and HYPO_gen with only HYPO_gen to HYPO_pcl burst pressure not found to be significant, p<0.05.

The cytotoxicity from genipin in each construct was tested by evaluating the degradation products at 24 hours. The CM_gel and HYPO_gen showed a slight reduction in cell number at 48 hours and no differences occured between the experimental groups and the positive control and 96 hours.

3.2. Rat in-vivo Ultrasound

The in-vivo ultrasound results demonstrated that HYPO_pcl, CM_gel, and HYPO_gel grafts had a patency rate of 6/7, 5/7, and 6/7, respectively. These results were subsequently confirmed by histological analysis. The ultrasound sample size for each group was at least n=3 for each time point with an exception of the CM_gel group having an n=2 at the 2 and 4 weeks. The CM_gel group had a similar compliance to rat aorta throughout the study. In-vivo compliance measurements demonstrated that our mechanically optimized graft (CM_gel) remained compliance-matched throughout the implant period, while one or both of the HYPO (stiffer) grafts demonstrated significant reductions in compliance compared to CM_gel at 2 and HYPO_gel 4 weeks (Figure 3, p<0.01). Ex-vivo analysis of compliance was unable to be conducted due to the limited amount of sample remaing from IHC. The average diameter of the implants stayed relatively constant and similar to rat aorta throughout the study. The only notable difference in diameter between groups occurred at the 4-week time point where two HYPO_gen grafts showed an increase in construct diameter, however this result was not statistically significant.

Figure 3:

In-vivo ultrasound measurements of compliance, average inner diameter, and velocity in the HYPO_pcl, CM_gel, and HYPO_gen TEVGs.*, p<0.05,**, p<0.01

The average maximum velocities of the HYPO grafts at two weeks was different that rat aorta, p<0.05 Figure 3. There was also was a decrease in blood velocity for CM_gel and HYPO_gen grafts from 1 week to 2 and 4 weeks (p<0.05). We attributed the decrease in velocity in the HYPO_gen grafts to 2 graft dilations.

3.3. Multiphoton Imaging

The results of construct degradation, cellular infiltration, and luminal thickness using multiphoton imaging are shown in Figure 4. More degradation (~70%) occurred in the CM_gel and HYPO_gen compared to the HYPO_pcl constructs (~30%). This high degradation rate led to an increase in cellular infiltration with more cells in the HYPO_gen vs. HYPO_pcl, p<0.05. Luminal thickness was not statistically different between all grafts in the proximal and middle locations.

Figure 4:

Top) Cross-sectional multiphoton implant image of HYPO_pcl, CM_gel, and HYPO_gen TEVGs; Green, blue, and red represent the implant, cell nuclei, collagen, respectively; The scale bar is 50 μm. Bottom) Differences in the amount of construct remaining, cells inside our TEVGs, and luminal thickness. *, p<0.05

3.4. Quantitative Immunohistochemistry

Representative fluorescent images of both the cross-sectional middle and longitudinal proximal locations of our TEVG are shown in Figure 5, Figure 6 and Figure 7, respectively. The green and blue regions in these images represent the construct and cell nuclei, respectively. Figure 5 shows representative cross-sectional images of the image with the specific marker in red (αSMA, CAL, MHC, Ki67, CD68, CD163, CD86, and elastin). These markers were quantitatively assessed using image processing in Matlab and the results are shown in Figure 6 and Figure 7.

Figure 5:

Representative middle cross-sectional immunofluorescence images. Green, blue, and red represent the implant, cell nuclei, and marker, respectively. The scale bar is 50 μm.

Figure 6:

Representative proximal longitudinal immunofluorescence images (top). Immunohistochemistry quantification of signal in the lumen, graft, and abluminal regions for SMCs, Ki67 cells, pan macrophages (CD68), M2 (CD163), M1 (CD86), M2/Total (CD163/CD68), collagen, and elastin (bottom).* = p<0.05; ** = p<0.01; Y axis = Pixels of Interest/Total Pixels; Scale bar is 100 μm

Figure 7:

Representative middle cross-sectional immunofluorescence images (top). Immunohistochemistry quantification of signal in the lumen, graft, and abluminal regions for SMCs, Ki67 cells, pan macrophages (CD68), M2 (CD163), M1 (CD86), M2/Total (CD163/68), collagen, and elastin (bottom).* = p<0.05; ** = p<0.01; Y axis = Pixels of Interest/Total Pixels; Scale bar is 100 μm

In the middle location, all regions (lumen, graft, and abluminal) and the total αSMA signal were found to be higher in the HYPO_gen & CM_gel vs. HYPO_pcl groups (p<0.05). For the proximal location (Figure 6), there were no αSMA signal differences between the grafts regions or throughout the total implant.

The proximal location of the implants displayed an increase in the more phenotypic SMC markers CAL & MHC for compliance matched TEVGs compared to either of the HYPO constructs (Figure 6). In particular, we observed an elevation in total CAL signal in the CM_gel constructs compared to either of the HYPO formulations. This elevation was similar to an increase in MHC in our compliance matched TEVGs compared to both HYPO grafts when quantifying signal within the graft itself. These phenotypic SMC differences between graft type were not observed in the middle location of our TEVGs (Figure 7).

The total Ki67 signal in the middle location of all grafts was increased in HYPO_gen vs. HYPO_pcl TEVGs, p<0.01. Throughout the regions in the middle location, Ki67 was also increased for HYPO_gen vs. HYPO_pcl in the lumen and abluminal regions, p<0.05. In the proximal location, there was increased Ki67 in the graft region for CM_gel vs. HYPO_gen, p<0.05.

The middle location of all TEVGs displayed the only statistically significant differences in macrophage content between graft type (Figure 7). There was an elevation in the total CD163 signal in the CM_gel and HYPO_gen groups vs. HYPO_pcl group (p<0.05 for CM_gel vs. HYPO_pcl comparison). These results generated an increase in the total ratio of pro-remodeling macrophages to pan (CD163/CD68) in the CM_gel and HYPO_gen TEVGs (p<0.05). The CD86 signal was generally decreased in the HYPO_gen with differences when compared to lumen HYPO_pcl, abluminal CM_gel, and total CM_gel in the middle location.

Endothelialization of the constructs were evaluated using marker CD31 and only at the middle location of the construct. The percent endothelialization was found to be 48±30%, 15±10%, and 34±24% for the HYPO_pcl, CM_gel, and the HYPO_gen TEVGs, respectively. The total collagen content had no significant differences occurring in either location however there were regional differences. The lumen and graft regions in the middle location of CM_gel grafts had a significantly increased amount of collagen than either hypocompliant grafts, Figure 7. The lumen region of HYPO_pcl grafts had more collagen than the HYPO_gen TEVGs in the proximal location. The middle location generally was found to have more elastin in the high gelatin groups (CM_gel and HYPO_gen) with differences occurring in the HYPO_gen to HYPO_pcl in all regions/total and CM_gel to HYPO_pcl.

4. Discussion

Our results suggest that a TEVG computationally optimized for compliance matching displays improved acute (28 day) in-vivo graft remodeling. Overall, the high gelatin groups (HYPO_gen & CM_gel) degraded more quickly in-vivo thus allowing an increased cell infiltration into the constructs. Compliance matching the TEVG lead to quicker and more mature phenotypic SMCs in the proximal location. The mechanical stimulus provided by the compliance matched constructs also demonstrated a more favorable cellular environment with increased collagen production in the middle location in the lumen and graft regions. Hypocompliant TEVGs promote an elevated CD68 and CD163 response compared to compliance matched grafts, suggesting a more pro-inflammatory environment. These results along with desirable improvements in SMC phenotype demonstrate that tuning the compliance of a tissue engineered vascular graft improves the in-vivo remodeling of a layered TEVG over an acute 1-month implantation period.

From our immunohistochemistry results and the literature, compliance has consistently been shown to affect SMC phenotype [20, 41]. In this study, there were increases in CAL (mid-differentiated) and elevated trends in MHC signal (late differentiated) in the proximal location of our compliance matched grafts. This suggests that the mechanical environment provided by compliance matched grafts encourages SMCs to move towards a more contractile phenotype earlier. Past in-vitro work by Bingcheng et al. is in agreement with our work as they showed similar changes in SMC phenotype as the compliance of their electrospun grafts increased [41]. Additional time points and explant functional testing (e.g., vasoreactivity) will be necessary to determine if the observed results here continue to promote a more functional vessel as our TEVG continues to be replaced by native host cells and extracellular matrix.

Likely due to our limited implantation time point, the middle of our graft showed slower cell infiltration and a less differentiated SMC phenotype. This is consistent with the presence of an increased number of αSMA positive cells in this region [42]. These results suggest that gelatin is ideal for fast degradation and cellular infiltration which is similar to what other groups have shown in-vitro [43, 44]. For the rat model, a higher gelatin content graft appears to be ideal for promoting increased cell infiltration and migration, but as you move to other animals there may be a need to balance this aspect. Generally, other animal models may require varying lengths of time for cellularization and therefore may need a slower degrading graft. Finding the right balance between cell infiltration and degradation will be vital for any particular animal model. It should be noted that increasing the gelatin content beyond what was used in this manuscript (a hypercompliant group) led to an even lower suture retention than HYPO_gen & CM_gel, Figure 2, and an inability to implant the resulting graft. The HYPO_gen and CM_gel were found to be slightly more challenging to implant than the HYPO_pcl grafts as a result of the lower suture strength.

SMC phenotype and functional behavior is also influenced by the local mechanical environment and composition of the extracellular matrix surrounding these cells. Several groups have shown that SMCs tend to migrate faster on surface coatings with stiffer substrates [45, 46]. Substrate stiffness has also been cited as being able to change SMCs from a contractile to a more proliferative phenotype[47]. We initially attributed our SMC results in the HYPO_gen group in the middle location to substrate stiffness. However, the HYPO_gen group had more highly proliferative cells (Ki67) in the total region than both low genipin concentration groups (HYPO_pcl & CM_gel, p=0.09), suggesting that increased genipin concentration may be the cause rather than compliance. This increase in highly proliferative cells suggest that the 20x fold increase (22 to 440mM) of genipin between groups could be potentially disruptive in-vivo. Genipin has been cited as a less cytotoxic agent than glutaraldehyde in the literature, however increases in genipin media concentration increases (>10mM) can prevent cell proliferation and promote cell death [48, 49].

While our TEVGs were fabricated with genipin concentrations near or exceeding these levels, our preliminary data suggests that the degradation products of our graft have little to no affect on RSMC growth, Figure 2. This is likely due to the fact that our grafts were thoroughly washed prior to implantation and as such do not expose the cells to these higher concentrations of genipin. It also should be noted that we did see an increase of Ki67 signal in the proximal graft region for the CM_gel grafts compared to HYPO_gen, however the specific cause is unclear and will require future investigation.

General SMC alignment and distribution at either location was not able to be evaluated due to the acute nature of the study, however our histological images show a grouping of SMCs which is encouraging. We would expect to see increased TEVG remodeling and SMC alignment at longer implantation time points and hope to evaluate this in the future studies [50, 51, 52].

Macrophage infiltration and proliferation in the TEVG are vital to healthy graft remodeling and this varied in the proximal and middle location. Again, we saw signs suggesting early cellular infiltration in the proximal anastomosis and generalized macrophage infiltration occurring in the middle location. Specifically, there was a decrease in CD68 macrophages in the abluminal region for compliance matched constructs compared to HYPO_gen. For the higher gelatin grafts (HYPO_gen & CM_gel) there was an increasing trend of total CD163 which may suggest that the presence of gelatin leads to a more pro-ECM maintenance and healing response. This M2 response is generally correlated with extracellular matrix formation and collagen production, which we found to be significantly higher in the compliance-matched group in the middle location of in the lumen and graft regions (Figure 7). On the other hand, the stiff constructs generally showed a decreased trend in CD163/CD68 compared to compliance-matched grafts suggesting local mechanical response could be vital to tissue regeneration. This also demonstrated that matching compliance of the vessel can provide a favorable environment for extracellular matrix formation. Looking more specifically at M1 macrophage markers (CD86), the HYPO_gen group showed a general reduction in signal compared to HYPO_pcl and CM_gel. These results suggest that higher genipin concentrations could have an effect on reducing the M1 macrophage population. Genipin, specifically, has been shown to reduce M1 macrophages in liver crosslinked samples when compared to glutaraldehyde [53]. For CD68, it was generally increased in grafts with high PCL concentrations (similar to our HYPO_pcl group) [8, 54]. Here we demonstrate that similar levels of CD68 and CD163 are elicited in grafts with half as much PCL, suggesting that the local biomechanical environment may also contribute to this response.

Another important factor in the formation of a functional TEVG is the replacement of our graft materials with the production of extracellular matrix, in particular collagen and elastin. For collagen at the one month time point, the CM_gel grafts generally had an elevated signal in the middle location of the lumen and graft regions compared to the HYPO grafts. These results suggest that compliance matching a graft can promote constructive in-vivo remodeling. One unexpected result was the increase in collagen in the lumen of the proximal location of the HYPO_pcl grafts. The cause and implications of this result, along with the general elevation of M1 macrophages in that area seems to suggest the start of intimal hyperplasia [55, 56] but, this results will need to be studied in more depth at longer time points to confirm. For elastin, CM_gel and HYPO_gen both had elevated signal in the middle location of our grafts. These results suggest that higher gelatin content is more favorable for elastin formation than PCL. Finding the right balance between the production of collagen and elastin as our graft degrades will be important in maintaining the compliance of our graft as it continues to degrade over longer time points.

An interesting trend from our study was the endothelialization of our grafts. We found that both stiff grafts had an increased percentage of endothelialization compared to the compliance matched graft. This indicates that endothelial cells migrated more quickly onto our hypocompliant grafts [57], which is consistent with literature indicating cells migrate further and faster on stiffer substrates [58, 59]. This finding is contrary to our initial hypothesis (that compliance matching may promote endothelialization), however this result can be leveraged in future graft designs. For example, a graft with a very thin but stiff inner layer could help promote early endothelialization while still maintaining optimal compliance. In general, percent endothelialization in our study was similar to the literature at one month, as complete coverage of grafts in the literature typically occurs around 3 months post implantation [60, 52, 54, 61].

The primary goal of this study was to evaluate the influence of compliance and remodeling on TEVG design. Our ultrasound data (Figure 3) showed that the CM_gel constructs stayed compliance matched throughout the time course of the study. Maintaining compliance demonstrated that the mechanical stimulus provided by the environment on cellular infiltration was maintained and this has not been previously shown using a rat model. Most TEVGs tend to be hypocompliant before implantation and maintain their compliance in-vivo, however some groups have shown that compliance can vary over the course of remodeling. Wu and Allen have demonstrated that a stiff or hypocompliant graft can remodel in-vivo to increase compliance by 3 months and maintain compliance over time but still showed a decreased stiffness trend compared to rat aorta [62, 50]. Another group also demonstrated that a Poly(l-lactide-co-ε caprolactone) TEVG increased in compliance for up to 12 months [4]. From the quantification of cellular markers and extracellular matrix formation, these previous studies demonstrate that graft degradation/remodeling can occur and transition a degrading graft towards a functional vessel. In this study, the gelatin grafts (HYPO_gen & CM_gel) also had a high percent of degradation and evidence of remodeling (collagen content), however a unique feature of our graphs was their ability to maintain compliance throughout the study and increase cellular infiltration in the gelatin groups. Looking more specifically at degradation and cellular infiltration, our group has demonstrated that cellular infiltration and degradation can be varied depending on composition/cross-linking and this is useful for translating the results into larger animal models where varied cellular infiltration rates will be necessary.

The patency rate of our HYPO_pcl, CM_gel, and HYPO_gen grafts were 6/7, 5/7, and 6/7, respectively, leading to an overall patency rate of 81%. This is similar to several studies in the literature [7, 60, 63, 64, 65]. For example, Wu et al. has shown a similar patency rate in their acellular TEVG with 17/21 remaining patent out to 3 months [62]. Nieponice et al. showed lower patency rates for poly-(ester urethane) thermally-induced phase separation scaffolds with 53% at 8 weeks and Pektok showed 100% patency through 24 weeks using a PCL electrospun scaffold [66, 67]. While these results are encouraging, patency of our graft could be further improved in the future with the incorporation of cells and/or surface coating to reduce thrombogenicity [68].

The blood velocity results of our TEVGs were found to be relatively consistent with the current literature values, around 900 mm/s [63]. There were some graft dilations in HYPO_gen group at 2 and 4 weeks,(Figure 3), and these increases could be the onset of graft failure or aneurysm in future time points. An unexpected result was the lack of a decrease in blood velocity when comparing the HYPO to CM_gel groups. This may be attributed to the short implant period and the dilations seen in the a few of the HYPO_gen grafts.

There are several limitations in this study. Rat aortic interpositional implantations offer significant advantages including their use as a relatively high throughput model. However, this model is limited as it has major differences in the hemodynamics, regenerative ability, and physiology in comparison to target vessels in most human vascular diseases. In order to minimize some of these limitations, other groups have used the rat carotid which has more clinically relevant flow and a smaller diameter, but may still suffer from some physiological differences [8]. Future implantations should consider utilization of an animal model with a hemodynamic flow environment more similar to that in the target human vessel. The second major limitation of this study was the acute one-month implantation period which restricts the conclusions that can be made on long-term cellular infiltration and vessel functionality. The earlier time point used here did however serve to identify differences in the acute remodeling of our grafts. Our group is currently planning to investigate how the results and conclusions presented here are altered in longer (3 to 12 months) implantation periods. Another limitation was the range of materials selected to tune the compliance and thickness of our graft to that of a rat aorta. These constraints limited the ability to increase suture retention and burst pressure, as seen in the CM_gel and HYPO_pcl groups. Selecting different materials or increasing the thickness of the implant could help prevent these problems in the future. Finally, it should be noted that while compliance was controlled by manipulating both graft composition and genipin concentration, this approach does present a confounding factor within each group comparison. The authors of this manuscript want to emphasize that conclusions made about compliance should only be made when both HYPO groups have similar results and the compliance matched group, CM_gel, differs significantly from these groups. If the reader is only comparing the CM_gel and HYPO_pcl groups this illuminates the effect of not only compliance but also gelatin concentration. Similarly, the effect of compliance on differences between the CM_gel group and HYPO_gen groups is confounded by genipin concentration. Future studies will need to address and mitigate these confounding factors to help further elucidate the role compliance has in TEVG design.

In this study, the approach used to modify compliance was the composition and crosslinking of the TEVG, however there are alternative approaches to accomplish this same goal. Increasing crosslinking concentration is a common method to decrease graft compliance as demonstrated in our study and by that of the Abbott et al. group who altered the concentrations of glutaraldehyde [1] during crosslinking. Our group has used a similar approach in the past [19], however have since moved away from this approach due to the cytotoxic potential of glutaraldehyde in-vivo [69, 70, 71]. Other groups have shifted towards manipulating graft compositional changes to vary compliance [9, 21]. Nezarati el al. created synthetic polymer grafts with varying thickness, tortuosity, and fiber fusions that were able to exceed the compliance of the human saphenous vein [9]. Another group modified the outer PCL sheath thickness which created a TEVG with tunable compliance [21]. While several researchers have successfully tuned the compliance of TEVGs, none have been able to match the thickness and compliance of the target of the vessel simultaneously. In general, most of the afore-mentioned TEVGs and other researchers in the literature are at least several hundreds of microns thicker than that of rat aorta (approx. 80 μm) which could limit cellular infiltration, transport, and cellular migration. Soletti et al. showed promise with their graft thickness, however, their grafts were significantly less compliant than rat aorta [66]. Our unique method allowed us to create a compliance-matched TEVG that has a similar thickness to rat aorta with a unique and biomimetic multi-layer design which can be modified to more closely resemble the native vessel. This multi-layer approach could also be tuned to include naturally occurring biopolymers such as elastin and collagen.

To the authors’ knowledge, this study is the first to assess the functional performance of a multi-layered TEVG that has been compliance-matched using a computational optimization approach. The mechanically optimized graft is acutely remodeled to remain compliance matched in-vivo over a 1-month period. The CM_gel grafts also demonstrated an increase in collagen production within the implant that may emanate from a more desirable macrophage and SMC response. These results need to be further explored in future studies that thoroughly assess graft vasoreactivity, graft degradation, and ECM production in graft explants at longer term explant periods [72]. We believe the results of this study demonstrate the importance of careful control of graft design and how it may influence the in-vivo remodeling and functional performance of biodegradable TEVGs.