Repairing Volumetric Muscle Loss in the Ovine Peroneus Tertius Following a 3-Month Recovery

Abstract

Much effort has been made to fabricate engineered tissues on a scale that is clinically relevant to humans; however, scale-up remains one of the most significant technological challenges of tissue engineering to date. To address this limitation, our laboratory has developed tissue-engineered skeletal muscle units (SMUs) and engineered neural conduits (ENCs), and modularly scaled them to clinically relevant sizes for the treatment of volumetric muscle loss (VML). The goal of this study was to evaluate the SMUs and ENCs in vitro, and to test the efficacy of our SMUs and ENCs in restoring muscle function in a clinically relevant large animal (sheep) model. The animals received a 30% VML injury to the peroneus tertius muscle and were allowed to recover for 3 months. The animals were divided into three experimental groups: VML injury without a repair (VML only), repair with an SMU (VML+SMU), or repair with an SMU and ENC (VML+SMU+ENC). We evaluated the SMUs before implantation and found that our single scaled-up SMUs were characterized by the presence of contracting myotubes, linearly aligned extracellular matrix proteins, and Pax7⁺ satellite cells. Three months after implantation, we found that the repair groups (VML+SMU and VML+SMU+ENC) had restored muscle mass and tetanic force production to a level that was statistically indistinguishable from the uninjured contralateral muscle after 3 months in vivo. Furthermore, we demonstrated the ability of our ENCs to effectively bridge the gap between native nerve and the repair site by eliciting a muscle contraction through direct electrical stimulation of the re-routed nerve.

Impact statement

The fabrication of tissues of clinically relevant sizes is one of the largest obstacles preventing engineered tissues from achieving widespread use in the clinic. This study aimed to combat this limitation by developing a fabrication method to scale-up tissue-engineered skeletal muscle for the treatment of volumetric muscle loss in a large animal (sheep) model and evaluating the efficacy of the tissue-engineered constructs after a 3-month recovery.

Introduction

Musculoskeletal disorders are a major problem in regenerative medicine that accounts for $800 billion dollars in annual health care costs in the United States alone.¹ Volumetric muscle loss (VML) is one of these disorders and significantly affects both military and civilian trauma patients.^2–5 Specifically, VML accounts for 60–65% of military disability patients and results in an estimated lifetime disability cost of $340,000 to $440,000 per patient.^2,3 VML is defined as the traumatic or surgical loss of a large volume of skeletal muscle, resulting in sustained functional impairment and, which in many cases, is accompanied by peripheral nerve injuries or physical deformity.⁴ While skeletal muscle has an innate capacity for regeneration, that capacity is overwhelmed in the case of VML, and instead, fibrotic scar tissue develops in the defect site.^3,6 The current standard of care for VML injury involves replacing the lost muscle by translocating autogenic tissue from a donor site into the wound. This technique is limited by donor site morbidity and graft tissue availability.⁷ Other treatment techniques use specialized powered bracing in combination with physical therapy.^7,8 Multiple factors may prevent complete recovery, including the severity of the injury, poor vascularization, denervation, and fatty infiltration. Therefore, despite promising advancements in surgical techniques, results are variable and often fail to restore full functionality in patients,^7–9 and in severe cases, repair failure can lead to amputation.¹⁰

Thus, the restoration of function is critical to improving clinical outcomes for patients with VML injuries. As such, various tissue engineering and regenerative medicine strategies are being developed to address the limitations of current treatment options and promote functional healing. These strategies include the delivery of myogenic cells, the implantation of acellular scaffolds, and the implantation of tissue-engineered skeletal muscle, including both scaffold-based and scaffold-free approaches. A living tissue-engineered muscle construct has the ability to restore muscle function, donate new muscle fibers to the repair site, recruit native regenerative cells, and integrate with native tissue, while addressing the issue of morbidity associated with current surgical treatments.^8,9 The strengths of scaffold-free engineered tissues make them one of the most appealing options for the future of VML treatment. Scaffold-free technologies possess specific advantages over scaffold-based and cell therapy techniques, in that they do not incur the foreign body response associated with synthetic scaffolds, the potential immunogenicity associated with naturally derived scaffolds, or the low cell viability common to cell therapy techniques. Previous work in our laboratory has demonstrated the success of the scaffold-free technique: using our engineered skeletal muscle units (SMUs) to repair a 30% VML injury in a rat tibialis anterior resulted in significantly greater force production compared to unrepaired negative controls after a 28-day recovery, although these forces were still significantly lower than the uninjured contralateral muscles.⁸ Our results have also demonstrated the ability of our SMUs to donate new muscle fibers to the repair site and integrate with host tissue, as evidenced by vascularization and innervation throughout the construct.^9,11,12

The success of these studies has prompted our laboratory to scale up our SMUs for use in a more clinically relevant large animal (sheep) model of VML repair. One of the most significant obstacles preventing tissue-engineered technologies from achieving widespread use in the clinic is the fabrication of tissues of clinically relevant sizes. This is because, in an avascular engineered tissue, nutrient availability is limited to diffusion, and prolonged time in vitro can result in the formation of a necrotic core. Currently, in vitro prevascularization is the primary approach taken to address the limitations of tissue scale-up, but this technique is limited by inefficiency and length of time required to prevascularize an engineered tissue.¹³ To circumvent these challenges, herein we designed a modular scale-up of our existing SMU technology with vascularization driven in vivo by the host tissue. In this approach, before implantation, we place single SMUs side-by-side, allowing lateral fusion and scale-up of the tissue width, while still maintaining the original thickness of a single SMU, such that nutrient diffusion distance is unchanged in that dimension and necrotic cores are not an issue. We have previously demonstrated the effectiveness of the modular approach in a rat VML model¹² and in sheep models of tendon¹⁴ and ligament repair.^15,16 Similarly, in this study, we expected to be able to modularly fuse larger SMUs into a single larger graft tissue and apply it to a sheep model of VML.

Treatments for traumatic musculoskeletal injuries often involve rerouting native nerve to the injury site; however, this is impossible when there is not enough native nerve available to span the distance to the repair site. To address this limitation, we have developed an engineered neural conduit (ENC) to bridge this spatial gap and direct neuronal growth to the repair site. In this study, we simulated this situation by dissecting 1 cm of a re-routed native nerve and used an ENC to bridge the 1 cm gap between the native nerve and the repair site.

ENCs are primarily composed of collagen and are devitalized before implantation. In general, they serve as a scaffold or guide for regenerating axons to migrate from a damaged nerve toward the targeted muscle and to prevent neuroma formation, which can manifest when axon regrowth occurs in an unorganized pattern leading to poorly vascularized and dense fibrotic formations.^17–20 The neural influence on skeletal muscle is well evidenced by the occurrence of muscular atrophy in instances of denervation due to nerve damage.²¹ Consequently, ensuring timely innervation of the SMUs is essential to prevent atrophy of SMU myofibers and increase the likelihood of functional recovery.

Thus, the goal of this study was to develop a fabrication method for scaled-up, allogeneic SMUs and ENCs, evaluate the SMUs and ENCs in vitro, and to test the efficacy of our SMUs and ENCs in restoring muscle function in a clinically relevant large animal model: sheep receiving a 30% VML injury in a load-bearing hindlimb, specifically the peroneus tertius (PT) muscle, following a 3-month recovery period.

Methods

Animal care

All animal care procedures followed The Guide for Care and Use of Laboratory Animals,²² according to a protocol approved by the University's Institutional Animal Care & Use Committee. In all instances, animals were first sedated through the administration of intramuscular xylazine (0.2 mg/kg) and then anesthetized through the administration of intravenous propofol (8 mg/kg) and gaseous isoflurane at concentrations between 2% and 5% to maintain a deep plane of anesthesia. For survival procedures, the animals were fasted and a fentanyl patch (75 mcg/h) was administered 24 h before surgery. Perioperatively, an intravenous dose of cefazolin (20 mg/kg) was administered. The animals received an intramuscular dose of flunixin (2.2 mg/kg) immediately after surgery with a subsequent dose administered 24 h postoperatively as supplementary analgesia. The fentanyl patch was removed 48 h after surgery. The animals were monitored daily for 10–14 days after surgery by University of Michigan veterinary staff. As a part of this daily health monitoring, the animals' gait was monitored to see if the animals were favoring the surgical leg or exhibiting “toe-touching,” a sign of pain. Surgical staples were removed 10–14 days after surgery. For terminal procedures, all animals were euthanized through the administration of a lethal dose of sodium pentobarbital (195 mg/kg) and subsequent bilateral pneumothorax.

Muscle biopsy collection

A total of three 4-month-old female Polypay sheep weighing ∼35 kg (Oswalt Farms, Vicksburg, MI) were anesthetized and whole muscle dissections of the semimembranosus muscles were completed under aseptic conditions. After the dissection was complete, animals were subsequently euthanized. The biopsies were transported to the laboratory in chilled Dulbecco's phosphate-buffered saline (DPBS) supplemented with 2% antibiotic-antimycotic (ABAM; cat. no. 15240-062; Gibco).

Muscle progenitor cell isolation

Muscle progenitor cells were isolated as described previously.^9,12,23–27 Briefly, 3.5 g muscle biopsies were sanitized in 70% ethanol and finely minced with a razor blade. The muscle then underwent enzymatic digestion in a solution composed of 2.3 mg/mL dispase and 0.3 mg/mL collagenase type IV at 37°C with constant agitation for 2.5 h. Following enzymatic digestion, the resultant suspension was filtered through a 100 μm mesh filter followed by a 40 μm mesh filter and centrifuged. The supernatant was discarded, and the cells were resuspended in freezing medium (70% Dulbecco's modified Eagle's medium (DMEM, cat. no. 11995-065; Gibco), 20% horse serum, and 10% dimethyl sulfoxide, and supplemented with 1% ABAM), frozen to −80°C at a rate of −1°C/min, and stored in liquid nitrogen until plating. Alternatively, a subset of the cell isolates was resuspended in muscle growth media (MGM) (60% F-12 Kaighn's modification nutrient mixture [cat. no. 21127-022; Gibco], 24% DMEM, 15% fetal bovine serum [FBS, cat. no. 10437-028; Gibco], 2.4 ng/mL basic fibroblast growth factor [cat. no. 100-18B; PeproTech, Rocky Hill, NJ], and 1% ABAM, and supplemented with an additional 10 μL/mL 1 μM dexamethasone [DEX, cat. No. D4902; Sigma])^9,11,23–26 and seeded immediately for SMU fabrication.

Modular skeletal muscle units fabrication

Cells were plated in MGM at a density of 10,000 cells/cm² onto 500 cm² tissue culture plates. After seeding, the cells were left undisturbed for 5 days and subsequently fed MGM every 2 days. The media were replaced with muscle differentiation media (MDM) (70% M199 [cat. no. 11150-059; Gibco], 23% DMEM, 6% FBS, 1% ABAM, 10 μL/mL 1 μM DEX, 1 μL/mL insulin–transferrin–selenium-X [cat. no. I1884; Sigma], and 0.72 μL/mL 50 mM ascorbic acid 2-phosphate [cat. no. A8960; Sigma])^9,11,23–26 when the plates were 100% confluent and elongating myotubes began to form a network (∼7 days postseeding). Light microscopy images of representative monolayers were taken 10 days after initial plating and before monolayer delamination. After 5–7 days on MDM, the monolayers delaminated off the cell culture surface and were transferred to Sylgard-coated plates, which were fabricated as described previously.^{9,11,12,23–26} Minutien pins were used to pin the monolayer into a 3D cylindrical construct at lengths of 14 ± 1 cm and ∼1 cm in diameter. Henceforth, this construct is referred to as a “single scaled-up SMU.” To achieve the desired size to fill the VML defect, a modular approach was used in which 2–4 single scaled-up SMUs were placed side-by-side and allowed to fuse. These constructs are referred to as “modularly fused SMUs.” It should be noted that after delamination of the monolayer, the SMUs (both single and modularly fused) are not attached to the culture substrate and are actually suspended in the culture media by the pins. After ∼1 week in vitro, the modularly fused SMU was ready for either implantation or in vitro characterization. An overview of this fabrication process is shown in Figure 1. Notably, the SMUs were allografts and were not devitalized or decellularized before implantation.

FIG. 1.

Construct fabrication process. Semimembranosus muscle and bone marrow biopsies were collected from a donor animal. Muscle progenitor cells were isolated from the semimembranosus biopsies and driven down a muscle lineage in culture. After 10 days in culture, the monolayer spontaneously delaminated to form a 3D skeletal muscle unit (SMU). Several SMUs were then combined in a modular manner and allowed to fuse together before they were implanted into an animal. Similarly, mononuclear cells were isolated from a bone marrow biopsy and driven down a tendon lineage in culture. Several monolayers were wrapped around a piece of nylon tubing to create a lumen. The monolayers were then allowed to fuse in culture to produce a 3D ENC. The ENC was then devitalized by freezing and then implanted into an animal. ENC, engineered neural conduit; SMUs, skeletal muscle unit.

In vitro characterization of SMUs

A subset of each cohort of the SMUs fabricated was reserved for in vitro characterization, including biomechanical testing and histology. Approximately 24–48 h after 3D formation, contractile properties of the SMUs were measured as described previously.^9,11,23–28 Briefly, contractions were elicited through field stimulation with a platinum electrode and measured by an optical force transducer (cat. no. SI-KG7A; World Precision Instruments) secured to one end of the construct. Tetanic forces were elicited using a 1 s train of 2.5 ms pulses at 600 and 800 mA and 60, 80, 120, and 150 Hz, and measured using custom LabVIEW 2012 software (National Instruments). Following contractile testing, SMUs were coated in tissue freezing medium (cat. no. 15-183-13; Fisher Scientific) and frozen in dry ice-chilled isopentane and subsequently cryosectioned at 10 μm. Cryosections of SMUs were stained with hematoxylin and eosin (H&E) and Masson's trichrome (cat. no. 25088-1; Polysciences, Inc.) to examine morphological characteristics of the SMUs, as well as immunohistochemically stained to identify the presence of myosin heavy chain (5 μg/mL dilution, cat. no. MF-20c; DSHB), laminin (5 μg/mL dilution, cat. no. ab7463; Abcam), α-actinin (15 μg/mL dilution, cat. no. ab18061; Abcam), and Pax7 (cat. no. Pax7c; 1 μg/mL dilution, DHSB) as described previously.^{9,11,12,27,29}

Engineered neural conduit fabrication

Bone marrow stromal cells (BMSCs) were harvested from an iliac crest marrow aspiration of one adult female sheep and mononuclear BMSCs were then isolated from the bone marrow as described previously.^{14–16,30–33} Following an established protocol,^{14–16,30,31} isolated cells were induced to a tendon lineage and expanded for five passages. The cells were then trypsinized and seeded at a density of 21,000 cells/cm² onto tissue culture dishes with embedded constraint pins. As a confluent monolayer began to spontaneously delaminate off the cell culture surface, the monolayer was wrapped around a 5 cm long piece of super soft nylon 11 tubing, creating a lumen (outer diameter 5/32,” cat. no. 1J-262-01; Freelin-Wade ). The tubing was adhered to the dish by pinning it to the embedded constraint pins. A total of four monolayers were wrapped around the tubing and allowed to fuse. The resultant nerve conduits were ∼4 mm in diameter and 7 cm in length. Seven days after the last monolayer was added, media were removed, and ENC was frozen at −80°C to devitalize the cells in the construct and allow for the preservation of the extracellular matrix (ECM) until implantation or in vitro characterization. Just before implantation, the silicone tubing was removed from the ENC. A subset of the ENCs was coated in Tissue Freezing Medium and frozen in dry ice-chilled isopentane for histology. Cryosections of the ENCs were stained with H&E and Picrosirius red (cat. no. 24901-500; Polysciences, Inc.) to visualize their morphology. An overview of the ENC fabrication process is shown in Figure 1.

Surgical procedures

Animals used for the surgical implant procedures were 7-month-old Polypay wethers (castrated males) weighing 45–55 kg. The animals were randomly divided into three experimental groups: VML only (n = 15), VML+SMU (n = 15), and VML+SMU+ENC (n = 15) (Fig. 2). On the day of surgery, the animals were weighed and then placed under a deep plane of anesthesia. A 15 cm incision was made along the front of the lower left leg (surgical side) to expose the PT muscle and peroneal nerve. The experimental and control limbs were not randomized; the surgical leg was always the left leg. Gross measurements of the muscle were taken, and a custom algorithm using a 3D model of the PT muscle was used to calculate the muscle mass constituting 30% of the total muscle volume. Subsequently, a full-thickness longitudinal portion of the PT constituting the calculated muscle mass was dissected. The VML-only animals (negative control) received the injury without a repair; the fascia and skin were closed with suture and the skin was stapled along the incision. In the VML+SMU group, the injury was immediately repaired by suturing an SMU within the defect. In addition, the distal branch of the peroneal nerve was transected and re-routed to the SMU to aid the process of reinnervation. In the VML+SMU+ENC group, the VML injury was also repaired with an SMU; however, in this group, 1 cm of the re-routed peroneal nerve was dissected to simulate a nerve injury, which often accompanies VML. The gap between the SMU and the peroneal nerve was then bridged with an ENC. 4-0 prolene was used to suture the re-routed nerve and ENC, while 4-0 PDS II was used to suture the SMU, fascia, and skin. All animals were monitored daily for 10–14 days after surgery and then returned to herd housing. The animals were allowed to recover for 3 months before we assessed functional recovery and performed muscle histological analyses.

FIG. 2.

Experimental groups. In all groups, a full-thickness longitudinal portion of the PT constituting 30% of the total muscle volume was dissected. (A) The VML-only animals (negative control) received the injury without a repair. (B) In the VML+SMU group, the injury was immediately repaired by suturing an SMU within the defect. In addition, the distal branch of the peroneal nerve was transected and re-routed to the SMU to aid the process of reinnervation (black arrow). (C) In the VML+SMU+ENC group, the VML injury was also repaired with an SMU. Also, 1 cm of the re-routed peroneal nerve was dissected to simulate a nerve injury. The gap between the SMU and the peroneal nerve was then bridged with an ENC (black arrow). PT, peroneus tertius; VML, volumetric muscle loss.

In situ biomechanical testing

We conducted in situ biomechanical testing of both the contralateral and surgical PTs using a custom biomechanical testing system. We chose to perform in situ biomechanical testing as opposed to in vitro testing because, in an initial subset of animals, we were unable to get accurate results through in vitro testing. This was because the muscle became hypoxic before completion of testing. Indeed, this has been evidenced in the literature, as the size of the sheep PT exceeds the recommended size of muscles for which in vitro biomechanical testing would be acceptable (i.e., hypoxia would not occur if appropriately perfused).^34–36 Furthermore, it has been observed that absolute forces of muscles tested in vitro can be significantly lower than those tested in situ.³⁶ Thus, we performed in situ biomechanical testing of the PT muscles and excluded the data collected through in vitro testing. For this reason, there is a reduced n-number in the analyses involving force measurements.

Following the 3-month recovery, animals were weighed and then placed under anesthesia. Both the contralateral and surgical PTs were dissected leaving the proximal origin intact. To secure the knee, a metal rod was inserted through the femoral epicondyles and secured to a rig mounted to the surgical table. In addition, adjacent nerves and musculature were severed so as not to interfere with testing. The distal tendon of the PT was secured to a custom strain gauge force transducer (Vishay Precision Group, Malvern, PA) to measure the force of the muscle contractions. Contractions were elicited by stimulating the nerve innervating the PT (i.e., a branch of the deep peroneal nerve) with a bipolar platinum wire electrode. Biomechanical testing of the muscles was conducted as described previously.^9,37,38 Briefly, the muscle was placed in the slack position and single 0.1 ms pulses of increasing current amplitudes (i.e., 60, 70, 80, 90, 100, 110, and 120 mA) were delivered until peak twitch force was reached. Maintaining the current, the muscle length was subsequently adjusted to the length at which twitch force was maximal. The length of the muscle at which twitch force was maximal was defined as the optimal length (Lo). The stimulus was then switched to a tetanus in which a 600 ms train of 0.1 ms pulses were delivered. The frequency of these pulses was increased (i.e., 60, 80, 100, and 120 Hz) until isometric tetanic force was maximal. Data were recorded using custom LabVIEW 2018 software. This process was then repeated on the contralateral (uninjured) PT muscle. Immediately after biomechanical testing, both the contralateral and surgical PTs were fully dissected, weighed, and prepared for histology. The animals were subsequently euthanized.

Histology and collagen content

After dissection, the muscles were weighed, and gross measurements were taken. The muscles were then divided into segments and prepared for histology in one of two ways; the sample was either coated in tissue freezing medium and frozen in dry ice-chilled isopentane or fixed in 10% formalin for 24 h and then embedded in paraffin. Frozen samples were cryosectioned at 10 μm and then immunohistochemically stained to identify myosin heavy chain (5 μg/mL dilution, cat. no. MF-20c; DSHB) and laminin (5 μg/mL dilution, cat. no. ab7463; Abcam ) as described previously.^9,11,12,27 Longitudinal samples were immunohistochemically stained for acetylcholine receptors (α-bungarotoxin, 1:2000 dilution, cat. no. B1601; Life Technologies), synaptic vesicle protein-2 (2 μg/mL dilution, cat. no. SV2c; DSHB), and neurofilament (0.5 μg/mL dilution, cat. no. 837904; BioLegend ) to identify the presence of neuromuscular junctions.

Paraffin-embedded cross-sectional samples were sectioned at 5 μm and stained with H&E, Picrosirius red, and Masson's trichrome. To quantitatively evaluate the collagen content of the muscles, a cross-section from the midbelly at the widest portion of the contralateral and surgical PT muscles was stained with Picrosirius red. These samples were imaged and the percentage of positive staining relative to the total cross-sectional area (percent collagen) was measured using ImageJ/Fiji (NIH).

Statistical analysis

Statistical analyses were performed using GraphPad Prism 7 software. Statistical differences between groups in which the contralateral muscle was compared to the surgical PT muscle were assessed with a two-way analysis of variance (ANOVA) with repeated measures (two-way repeated measures ANOVA [RM ANOVA]) and Sidak's multiple comparisons (Sidak's MC) tests. Statistical differences between groups in which the values were normalized to the contralateral side were assessed with a one-way ANOVA with Tukey's multiple comparisons (Tukey's MC) test. Results were significant at p < 0.05. Bars on graphs indicate mean ± standard deviation.

Results

SMU graft scale-up

We successfully scaled-up SMUs grown from allogeneic ovine muscle progenitor cells by using a larger cell culture surface to produce single scaled-up SMUs, which were then combined in a modular manner to produce modularly fused SMUs. Single scaled-up SMUs were increased in size by ∼350% when fabricated on 245 × 245 mm tissue culture plates compared to the 100 mm diameter tissue culture plates used in our previous study.¹¹ By day 10 in the fabrication process, monolayers exhibited 100% confluence and comprised mainly highly aligned myotubes (Fig. 3B). It is important to note that the scaled-up monolayers did not exhibit an overgrowth of fibroblasts, demonstrating similar morphology to SMUs fabricated on 100 mm dishes. Utilization of our scaled-up fabrication method resulted in 3D cylindrical constructs that were 14 ± 1 cm in length and ∼1 cm in diameter (Fig. 3A). These single scaled-up SMUs were then modularly fused for implantation.

FIG. 3.

In vitro SMU characterization. (A) Grossly, modularly fused SMUs are about 14 cm long before implantation. (B) Images of representative monolayers were taken 10 days after initial plating, before the delamination of the monolayer. Arrow indicates a fused myotube. Masson's trichrome staining of a single scaled-up SMU cross-section (C) and longitudinal section (G) shows that the extracellular matrix is composed of collagen (blue) and collagen fibers are oriented parallel to the longitudinal axis. Immunostaining for myosin heavy chain (MF20, green) and laminin (red) of an SMU cross-section (D) and longitudinal section (H) reveals the presence of muscle fibers dotted throughout the SMU. (E) Nuclear staining with DAPI in cross-section shows the absence of a necrotic core. (F) We noted the presence of satellite cells (Pax7, green), laminin (red), and nuclei (DAPI, blue) in the single scaled-up SMUs in vitro. (I) Additional staining for α-actinin (green) and laminin (red) reveals the presence of an organized sarcomeric structure and striations characteristic of native muscle.

Histological characterization of SMUs

To examine general morphological characteristics of the engineered constructs in vitro, staining with H&E and Masson's Trichrome was performed. Histological analyses of stained longitudinal sections revealed that single scaled-up SMUs were composed of linearly aligned muscle fibers surrounded by a collagen ECM (Fig. 3C, G). In addition, Figure 3E shows that the core of the single scaled-up SMUs was not necrotic, as indicated by the presence of DAPI-stained cells throughout the entire thickness of construct. The presence of cells throughout the construct suggests that there was adequate nutrient delivery and waste removal during the construct fabrication and maturation process.

To characterize the SMU structural composition, immunohistochemical (IHC) staining was performed with antibodies specific to myosin heavy chain and laminin (Fig. 3D, H). Staining shows that SMU fibers were surrounded by a distinct laminin-rich ECM, much like the basal lamina surrounding native muscle fibers. In addition, IHC was performed with antibodies specific to α-actinin to visualize the organization of Z-discs. Staining revealed sarcomeric organization and striations characteristic of native skeletal muscle (Fig. 3I). Also, we performed IHC staining for Pax7⁺ satellite cells and found that the SMUs contained a population of muscle progenitor cells before implantation (Fig. 3F).

Biomechanical testing of SMUs

For each of the modularly fused SMUs that were implanted, it was necessary to fabricate sentinel constructs using smaller 60 mm diameter dishes and measure force production in vitro to ensure the quality of our cell isolation and culture conditions. Ultimately, we were limited by our in vitro force testing equipment, since the construct length and size of the scaled-up SMUs exceeded our in vitro testing capabilities. Thus, we fabricated smaller sentinel SMUs using the same protocol and media supplies for in vitro mechanical testing.

After the monolayers delaminated and were organized in 3D form, the maximum isometric tetanic force production of each sentinel SMU was measured (n = 22 total). Force testing indicated an average isometric tetanic force of 992 μN with the minimum and maximum tetanic force produced by a single SMU fabricated on a 60 mm dish ranging from 132 to 4587 μN, respectively. For SMUs of this size, our release criterion stipulates that SMUs need to produce a tetanic force of at least 100 μN to be suitable for implantation. Based on years of historical data from our laboratory, we have observed that SMUs that do not meet the 100 μN requirement are not successful when implanted and show little muscle regeneration. Notably, all of the constructs met this implantation requirement. We then used these data to estimate the force production of our single scaled-up SMUs. We first calculated the force per cross-sectional area of each of these sentinel SMUs and found that the average force per cross-sectional area for the sentinel SMUs was 1180 μN/mm² with the minimum and maximum being 22.4 and 5840 μN/mm², respectively. We then multiplied this value by the average cross-sectional area of the single scaled-up SMUs. Using this estimation, we found that the average tetanic force for the single scaled-up SMUs was predicted to be 8370 μN with a minimum and maximum value of 158 and 41,300 μN, respectively, and the 95% confidence interval for the tetanic force of a single scaled-up SMU was predicted to be 3090–13,600 mN.

Histological characterization of ENCs

H&E staining was performed for qualitative morphological assessment of the ENCs and indicates that the wall thickness of the ENCs was ∼1.5–2 mm and contained densely packed nuclei (Fig. 4A, B). Picrosirius red staining was used to characterize the content and alignment of collagen within the engineered ENC before implantation (Fig. 4C, D). When imaged under polarized light, the collagen birefringence showed that the ENCs demonstrated a semiorganized collagen fiber framework with regions of aligned collagen bands similar to a native-like collagen crimp pattern (Fig. 4E). Grossly, the ENCs are ∼7 cm in length before implantation (Fig. 4F).

FIG. 4.

In vitro ENC characterization. H&E staining of (A) longitudinal sections and (B) cross-sections of ENCs shows general morphology, particularly the hollow lumen (B) resulting from placement of the nylon tubing. Picrosirius red staining of (C) longitudinal sections and (D) cross-sections shows the ENCs are primarily composed of collagen, and under polarized light (E), the collagen displays parallel structure. (F) Grossly, the constructs are ∼7 cm in length before implantation. H&E, hematoxylin and eosin.

Surgical procedures

Across all groups, the average percentage of VML that was created was 27.92% ± 3.26% and the actual mass removed was 5.26 ± 0.76 g. The variability in the weights removed is due to differences in size of the PT muscles between individuals. There was no significant difference in the magnitude of the VML injury between groups (one-way ANOVA: p = 0.1909, n = 15 per group). Specifically, the VML injury was equal to 26.95% ± 3.08% in the VML-only group, 27.72% ± 3.26% in the VML+SMU group, and 29.11% ± 3.38% in the VML+SMU+ENC group (Fig. 5A).

FIG. 5.

VML injury and mass recovery. (A) Across all animals, the average percentage of VML was 27.92% ± 3.26%. There was no significant difference in the magnitude of the VML injury between groups (p = 0.1909). (B) The majority of the animals gained weight normally during the 3-month recovery period. The surgical group did not significantly affect the animals' body weight over time (p = 0.2528, n = 45). (C) We assessed the ability of our grafts to restore lost muscle mass by comparing the weights of the contralateral and surgical PTs at the time of explant. The VML-only group exhibited a significant difference in muscle mass between the contralateral and surgical sides (p = 0.0003, n = 15), indicating a lack of mass recovery. Conversely, there was no significant difference in muscle mass observed between the contralateral and surgical PTs of the VML+SMU and VML+SMU+ENC groups (p = 0.1406 and p = 0.0555, respectively). (D) Expressed as a percentage of the contralateral, the mass deficit in the VML-only group is more apparent, although not significant.

Body mass and muscle mass recovery

Animals were awake, alert, and weight-bearing within 2 h of completing surgery. None of the animals was observed to have abnormal gait after surgery; however, n = 2 animals did exhibit “toe-touching” up to 24 h following surgery, but subsequently showed no signs of “toe-touching.” In addition, animals experienced mild to moderate swelling at the injury site for up to 2 weeks postoperation. Around 89% of animals gained weight normally during the 3-month recovery period and there was no significant difference in the animals' body weight between experimental groups (two-way RM ANOVA: experimental group: p = 0.2528, n = 45) (Fig. 5B). This indicates that the experimental group did not significantly impact the health of the animals during this 3-month recovery period. Also, no signs of rejection were observed in the sheep. Bloodwork taken at the time of explant revealed no sign of a chronic immune response, although one animal did exhibit monocytosis.

Gross observations at the time of explant revealed larger amounts of connective tissue in the surgical PTs of all animals relative to what was observed in the uninjured contralateral. No muscle unit tethering was noted. At the time of explant, we compared the weights of the surgical and graft PTs. Notably, there were no significant differences in muscle mass between the contralateral and surgical PTs in the VML+SMU and the VML+SMU+ENC groups (Sidak's MC: p = 0.1406 and p = 0.0555, respectively, n = 15 per group) (Fig. 5C). In contrast, there was a significant muscle mass deficit in the VML-only group (Sidak's MC: p = 0.0003, n = 15) indicating a lack of muscle mass recovery. The mass deficit experienced by the VML-only group was expected, as the defect in the VML-only group was not repaired with a construct or any other filler. Specifically, the mean difference in muscle weight between the contralateral and surgical PTs in the VML-only, VML+SMU, and VML+SMU+ENC groups was 2.82, 1.34, and 1.62 g, respectively. We also compared groups after normalizing the weight of each surgical PT muscle to the contralateral muscle. Represented as a percentage of the contralateral, there were no significant differences in PT weight between groups (one-way ANOVA: p = 0.2613) (Fig. 5D). On average, the VML-only, VML+SMU, and VML+SMU+ENC groups experienced an 8.24%, 4.10%, and 5.28% muscle mass deficit, respectively.

Force recovery

To assess force recovery, we compared the maximum tetanic force of the surgical PT to that of the contralateral PT (Fig. 6A). Notably, only the VML-only group had a significant difference in maximum force production between the surgical and contralateral PTs (Sidak's MC: p < 0.0001, n = 11), indicating a lack of force recovery. In contrast, the VML+SMU and VML+SMU+ENC groups produced forces that were not significantly different from the contralateral side (Sidak's MC: p = 0.0613, n = 9, and p = 0.5755, n = 11, respectively). Specifically, the mean difference in maximum tetanic force between the contralateral and surgical PTs in the VML-only, VML+SMU, and VML+SMU+ENC groups was 71.40N, 39.04N, and 17.00N, respectively. Represented as a percentage of the contralateral, the VML+SMU+ENC group produced significantly higher forces than the VML-only group (Sidak's MC: p = 0.0320) (Fig. 6B). On average, the VML-only, VML+SMU, and VML+SMU+ENC groups experienced a 30.01%, 17.51%, and 7.78% force deficit, respectively. In addition, the data show no significant variability in the frequency required to produce a maximum tetanus. For the uninjured contralateral, maximum tetanus was achieved at 82.6 ± 13.2 Hz and for the surgical legs, maximum tetanus was achieved at 85.2 ± 11.3 Hz. Specifically, the frequency was 84.5 ± 9.9 Hz, 91.1 ± 12.9 Hz, and 80.9 ± 9.0 Hz for the VML-only, VML+SMU, and VML+SMU+ENC groups, respectively. Statistically, there is no significant difference between these groups (one-way ANOVA: p = 0.2623).

FIG. 6.

Restoration of force production. Tetanic isometric forces elicited by an electrical stimulus were measured from both the surgical and contralateral PT muscles. (A) In the VML-only group, the maximum force production of the surgical PT was significantly lower than the contralateral PT (p < 0.0001, n = 11). In contrast, there was no distinguishable difference in force production in the VML+SMU group (p = 0.0613, n = 9) or the VML+SMU+ENC group (p = 0.5755, n = 11). (B) When the force capability of the injured muscles is expressed as a percentage of the contralateral, it is apparent that the VML+SMU+ENC group experienced the greatest restoration of force production and was significantly higher than the VML-only group (p = 0.0320). Maximum isometric tetanic forces were also normalized to muscle mass. (C) In the VML-only group, the force per gram of muscle of the surgical PT was significantly lower than the contralateral PT (p = 0.0018, n = 11). In contrast, there was no distinguishable difference in force production in the VML+SMU group (p = 0.1225, n = 9) or the VML+SMU+ENC group (p = 0.9596, n = 11). (D) When the force per gram of muscle of the surgical PTs is expressed as a percentage of the contralateral, it is apparent that the VML+SMU+ENC group experienced a greater normalized force than the VML-only group, but this was not significantly different (p = 0.0783). (E) There was a significant difference in the optimal length of the muscles between the contralateral and surgical PTs of the VML-only group (p = 0.0025, n = 10) and in the VML+SMU+ENC group (p = 0.0222, n = 11), but not in the VML+SMU group (p = 0.0868, n = 9). (F) Represented as a percentage of the contralateral, there was no significant difference in the percentage of optimal length between the groups (p = 0.6939).

Because of the complex muscle architecture of the PT muscle, we were not able to calculate specific force in the traditional sense, by dividing the maximum force by the physiological cross-sectional area. Instead, we normalized the maximum force to the muscle weight (Fig. 6C, D). The trends noted previously were again present in these normalized data. Once again, the VML-only group experienced a significant deficit in force per gram of muscle between the surgical and contralateral PTs (Sidak's MC: p = 0.0018, n = 11), while the VML+SMU and the VML+SMU+ENC groups had no significant differences between the contralateral and surgical PTs (Sidak's MC: p = 0.1225, n = 9 and p = 0.9596, n = 11, respectively) (Fig. 6C). Specifically, the mean difference in force per gram of muscle between the contralateral and surgically repaired PTs in the VML-only, VML+SMU, and VML+SMU+ENC groups was 1.91, 1.16, and 0.22 N/g, respectively. Represented as a percentage of the contralateral, there were no significant differences in the force per gram of muscle between groups (one-way ANOVA: p = 0.0783) (Fig. 6D). On average, the VML-only, VML+SMU, and VML+SMU+ENC groups experienced a 22.55%, 15.88%, and 3.17% normalized force deficit, respectively.

Although we did not notice any muscle unit tethering, we did observe changes in optimal length (Lo) of the muscle when comparing the surgical to the contralateral PT. A Sidak's multiple comparisons test revealed that there was a significant difference in the optimal length of the muscles between the contralateral and surgical PTs of the VML-only group (p = 0.0025, n = 10) and in the VML+SMU+ENC group (p = 0.0222, n = 11), but not in the VML+SMU group (p = 0.0868, n = 9) (Fig. 6E). Represented as a percentage of the contralateral, there was no significant difference in the percentage of optimal length between the groups (one-way ANOVA: p = 0.6939) (Fig. 6F). This reduction in Lo suggests a change in the tissue's mechanical properties and the gross architecture of the muscle.

In addition to stimulating the nerve innervating the PT, we also sought to elicit a muscle contraction by stimulating the nerve that was transected and re-routed to the surgical site. Direct stimulation of the re-routed nerve resulting in a muscle contraction occurred in 25% (3 out of 12) animals in the VML+SMU group and 75% (9 out of 12) animals in the VML+SMU+ENC group. This direct stimulation produced a contraction through the center of the muscle at the site of the initial injury. This suggests that the re-routed nerve was able to successfully form neuromuscular junctions in the injury site, and notably, the ENC was able to effectively bridge the gap between the re-routed nerve and the surgical site in the majority of animals.

Histological analysis of muscles

We performed qualitative and quantitative histological analyses on midbelly PT cross-sections of both contralateral and surgical PT muscles. In all surgical groups, the repair site was characterized by a fibrotic region, as evidenced by the H&E staining (Fig. 7A–D) and Masson's trichrome staining (Fig. 7E–H). The blue regions in the sections stained with Masson's trichrome demonstrate the collagen deposition in the repair site. Immunostaining for myosin heavy chain and laminin (Fig. 7I–L) showed that there was also laminin in the repair site. We did not qualitatively observe an obvious difference in the abundance of positive MF20 staining between groups when observing the muscle as a whole.

FIG. 7.

Gross morphology of explanted muscle. Cross-sections of explants from contralateral (A, E, I), VML-only (B, F, J), VML+SMU (C, G, K), and VML+SMU+ENC (D, H, L) groups were taken from the midbelly of the muscle. Tissues stained with H&E (A–D) show that the surgical site is characterized by the presence of disorganized, hypercellular tissue. Serial sections stained with Masson's trichrome (E–H) indicate that there are large fibrotic regions, evidenced by positive collagen staining (blue) that spans between the native muscle (red) in all surgical groups. (I–L) Immunostaining for myosin heavy chain (MF20, red) and laminin (green) show that muscle is absent in a large portion of the surgical site. Scale bars = 2 mm.

At higher magnifications, we noticed the presence of vasculature in the repair site of the H&E-stained sections of all experimental groups (Fig. 8A–D). We also observed the presence of intramuscular fat near the repair site of the surgical groups, but not in the uninjured contralateral muscle (Fig. 8A–D). IHC staining for myosin heavy chain (MF20) and laminin revealed the presence of small muscle fibers within the repair site in the VML-only, VML+SMU, and VML+SMU+ENC groups (Fig. 8E–H). The presence of these small muscle fibers suggests that our SMUs are promoting muscle regeneration; however, the origin of these fibers has not been determined. We also noted the presence of neuromuscular junction formation in all experimental groups as evidenced by positive staining for acetylcholine receptors, synaptic vesicle protein-2, and neurofilament in longitudinal sections of the repair site (Fig. 8I–L).

FIG. 8.

Microstructures of explanted muscle. H&E-stained cross-sections of the repair site demonstrated the presence of fat (black arrows) in all surgical groups (B–D) compared to the contralateral (A). Furthermore, tissues stained with H&E (A–D) show that the repair site is characterized by what appears to be an increase in vasculature (asterisks) in all experimental groups compared to the contralateral muscle, although this was not quantified. Immunohistochemical staining for myosin heavy chain (MF20, red) and laminin (green) (E–H) revealed the presence of small muscle fibers within the repair site in the VML-only (F), VML+SMU (G), and VML+SMU+ENC (H) groups as indicated by the white circle. (I–L) We noted the presence of neuromuscular junctions in the contralateral muscle as well as the surgical site of all experimental groups through immunostaining longitudinal sections for acetylcholine receptors (red), synaptic vesicle protein-2 (red), and neurofilament (green). The inserts show zoomed-in images of the neuromuscular junctions. Scale bars = 200 μm.

To quantitatively evaluate the collagen content of the muscles, we measured the percentage of positive Picrosirius red staining (percent collagen) in midbelly cross-sections of the PT muscles (Fig. 9). We first verified this method by evaluating the collagen content of the uninjured contralateral muscles and found that they have very little variability in the percent collagen. Specifically, the average percent collagen in the uninjured contralateral PTs was 17.63% ± 2.76% (n = 23). In contrast, the average percent collagen in the surgical PTs was 46.01% ± 8.53%, 36.46% ± 4.54%, and 39.59% ± 9.18% for the VML-only (n = 7), VML+SMU (n = 6), and VML+SMU+ENC (n = 10) groups, respectively. The percent collagen of the surgical PT of the VML-only group was significantly higher than the VML+SMU group (Sidak's MC: p < 0.0323), which indicates that there was significantly more connective tissue deposition in the VML-only group compared to the VML+SMU groups (Fig. 9E).

FIG. 9.

Analysis of collagen content in explanted muscle. Midbelly PT cross-sections of the surgical sites of muscles in the (B) VML-only, (C) VML +SMU, and (D) VML+SMU+ENC groups were stained with Picrosirius red to evaluate the collagen content in the surgical site relative to (A) the uninjured contralateral muscle. (E) The percentage of positive Picrosirius red staining relative to the total cross-sectional area (percentage of collagen) was measured using ImageJ. The surgical PT of the VML-only group had a significantly greater percentage of collagen than the VML+SMU group (p = 0.0323). *p < 0.05. Scale bars = 2 mm.

Discussion

Currently, there is a lack of large animal models of VML. Large animal models are especially important for both clinical relevance and to address the limitations of rodent models. Specifically, rodents do not typically present the clinical manifestations of fibrosis seen in humans, and the small sizes of rodent models do not pose a significant challenge to vascular and neural regeneration and growth into the injury site.^3,39,40 Furthermore, the FDA recommends testing in both large and small animal models when seeking regulatory approval for a new technology.⁴¹ Currently, the only large animal VML models that exist are pigs^42–44 and dogs.⁴⁵ This study is novel in that, it contributes a sheep model of VML.

Much effort has been made to fabricate tissues on a scale that is clinically relevant to humans; however, scale-up remains one of the most significant technological challenges of tissue engineering to date.⁷ The primary obstacle to scale-up is the nutrient requirements of the tissue. In larger, avascular engineered tissues, nutrient availability is limited by diffusion, which can result in the formation of a necrotic core if nutrients cannot adequately penetrate the entire thickness of the construct.^46–48 Our approach utilizes a modular tissue scale-up in which individual engineered tissue units are combined just before implantation and allowed to fuse together, which we have shown prevents the formation of a necrotic core in vitro. We combine this strategy with the concept of the “body as a bioreactor” to drive vascularization of our constructs in vivo, to eliminate the need for prevascularization. Overall, this study constitutes a significant step toward development of a VML treatment that overcomes many of the obstacles associated with the scale-up of engineered tissues and that restores both muscle mass and force production to a level that is statistically indistinguishable from the uninjured contralateral muscles after only 3 months.

Specifically, we were able to use a larger culture surface and our modular fusion approach to scale-up and fabricate engineered muscles that were 14 ± 1 cm long and ∼1 cm in diameter. Qualitatively, the monolayers showed a highly aligned and dense myotube network. Elongated and networking myotubes indicate native-like skeletal muscle cell differentiation. Immunohistochemistry revealed that in vitro, the SMUs contained a population of satellite cells after 3D formation. The presence of these Pax7⁺ cells may have improved the myogenic potential of our SMUs in vivo and contributed to the regeneration observed in these groups.

In addition, the presence of positively stained DAPI cells throughout the in vitro construct indicated that the construct core was not necrotic, which supports the use of the larger culture surface to fabricate larger single scaled-up SMUs. Evaluation of the contractile properties of our SMUs before implantation showed that the isometric force production consistently met our release criteria. In fact, many of the constructs far exceeded our expected release criteria for force production by 20- to 40-fold. Potential explanations for the considerable variation in tetanic forces may be due to the alignment of the myofibers, since contractile function of the construct is determined by the muscle architecture. We are currently developing techniques to predict the functional potential our constructs noninvasively using methods previously described by Syverud et al.²⁷

Notably, despite the allogeneic nature of the SMUs and ENCs used in this experiment, the animals receiving engineered tissues exhibited no signs of rejection. Furthermore, the in vivo results demonstrate the ability of our SMUs to restore both muscle mass and force production to a level that was statistically indistinguishable from the uninjured contralateral muscle after only 3 months. As a percentage of the contralateral, there was no significant difference in the percentage of mass recovery between the experimental groups. In contrast, as a percentage of the contralateral, the VML+SMU+ENC group experienced a significantly higher percent force recovery than the VML-only group, which suggests that the mass recovery experienced in all experimental groups cannot solely be attributed to connective tissue deposition in the defect site. This is further supported by the IHC analysis, which revealed the presence of small muscle fibers in the injury site of VML+SMU and VML+SMU+ENC groups to a higher degree than what was qualitatively observed in the VML-only group.

The animals did experience skeletal growth during the 3-month recovery period as evidenced by changes in muscle dimensions, including increased muscle length, between the time of implantation and the time of explant. As the majority of sheep gained weight during the recovery period, this increase in muscle size may also be attributable to hypertrophy of the muscles as a result of increased weight load. The growth of the muscle may have influenced its recovery from the VML injury; however, the effect of the animals' growth would be equivalent in all experimental groups.

We also found that the VML-only group had a significantly greater percentage of collagen within the muscle cross-section than the VML+SMU group. This suggests that the quality of the repair was lower in the VML-only group and corroborates the greater force deficit observed in the VML-only group compared to the VML+SMU and VML+SMU+ENC groups. Thus, when accounting for the expected mass deficit of the VML-only group (as this group received no mass to fill the defect), the quality of the repair of the VML+SMU and VML+SMU+ENC groups is highlighted.

Notably, our results also demonstrate the ability of the ENCs to effectively treat a concomitant peripheral nerve injury by bridging the gap between the re-routed nerve and the repair site, as evidenced by direct stimulation of the re-routed nerve, which elicited an action potential to the injury site in majority of animals in the VML+SMU+ENC group. In particular, we were able to stimulate the re-routed nerve in a greater percentage of animals in the VML+SMU+ENC group (75% vs. 25% in the VML+SMU group). We speculate that the reason the VML+SMU+ENC animals experienced greater success could be due to the fact that the re-routed nerve was isolated from the inflammatory environment of the VML injury for some period of time compared to the VML+SMU group. In the VML+SMU group, it is possible that immediately placing the re-routed nerve in the inflammatory environment of the injury site may have promoted neuroma formation, which occurs when neuronal regeneration takes place in the presence of fibrotic tissue formation.²⁰ In contrast, the additional time it took for the re-routed nerve in the VML+SMU+ENC group to grow into the injury site may have kept it sufficiently isolated from the inflammatory microenvironment of the VML injury and contributed to its success.

In conclusion, these results demonstrate a significant advancement to the field of skeletal muscle tissue engineering. We successfully scaled-up our technology by fabricating tissues of clinically relevant sizes and demonstrated their potential for mass and force recovery in an ovine model of VML. In future experiments, we would like to explore the inclusion of additional functional assessments, such as gait analysis, range of motion assessments, and functional benchmarking before surgery. Any change in gait, although not noted in this study, could have influenced the force-generating capacity of the PTs in both the uninjured contralateral muscle as well as the surgical PT, thereby skewing functional results. Future experiments will also aim to better understand the mechanism by which our constructs are aiding the repair process by tracking the migration of endogenous and exogenous cells in the repair site.

Disclosure Statement

No competing financial interests exist.

Funding Information

The research in this manuscript was funded by the Department of Defense (W81XWH-16-1-0752). Additionally, NIH P30 support was provided by the Michigan Integrative Musculoskeletal Health Core Center.