A Decellularized Porcine Xenograft-Derived Bone Scaffold for Clinical Use as a Bone Graft Substitute: A Critical Evaluation of Processing and Structure

Background: Bone grafts are used in approximately one half of all musculoskeletal surgeries. Autograft bone is the historic gold standard but is limited in supply and its harvest imparts significant morbidity to the patient. Alternative sources of bone graft include allografts, synthetics and, less commonly, xenografts which are taken from animal species. Xenografts are available in unlimited supply from healthy animal donors with controlled biology, avoiding the risk of human disease transmission, and may satisfy current demand for bone graft products. Methods: In the current study, cancellous bone was harvested from porcine femurs and subjected to a novel decellularization protocol to derive a bone scaffold. Results: The scaffold was devoid of donor cellular material on histology and DNA sampling (p < 0.01). Microarchitectural properties important for osteoconductive potential were preserved after decellularization as shown by high resolution imaging modalities. Proteomics data demonstrated similar profiles when comparing the porcine bone scaffold against commercially available human demineralized bone matrix approved for clinical use. Conclusion: We are unaware of any porcine-derived bone graft products currently used in orthopaedic surgery practice. Results from the current study suggest that porcine-derived bone scaffolds warrant further consideration to serve as a potential bone graft substitute.


Introduction
A bone scaffold is a three dimensional matrix used to fill bone voids and support the body's intrinsic regenerative potential [1]. As osteogenic cells migrate and adhere to the scaffold, new bone is formed and the scaffold is eventually degraded and turned over [2]. The ideal bone graft substitute should mirror the properties of the "gold standard" autograft, and thus, be biocompatible, osteoconductive, osteoinductive, osteogenic, and also be readily available without risk of disease transmission [3]. Scaffolds can be constructed from synthetic materials that are sterile and non-immunogenic, allowing them to be safely implanted in a host without risk of disease transmission and minimal risk of an inflammatory reaction. However, to more closely approximate the morphology and biomechanical properties of autograft bone, scaffolds can also be derived from natural bone to optimize integration into surrounding host tissue [4]. Perhaps the most widely utilized bone scaffold is demineralized bone matrix (DBM), which is derived from human allograft bone. DBM serves as an osteoconductive matrix with osteoinductive potential imparted by the retained growth factors from the donor [5]. Biologic activity of DBM depends on donor health and bone processing techniques which ultimately results in considerable variability between DBM products [6]. The cost of one milliliter of DBM often exceeds $100, making their application cost prohibitive in the treatment of larger osseous defects. Scaffolds derived from alternative sources of natural bone, such as xenograft bone (non-human), could afford the same osteoconductive benefits as DBM at a lower cost, avoid the risk of human disease transmission and be immediately available in relatively unlimited quantities from a larger pool of healthy donors with controlled biology [7]. If a xenograft-derived scaffold also retained its donor's osteoinductive growth factors, the product could potentially serve the same clinical purpose as DBM.
Bovine bone has previously been transplanted into human patients, but orthopaedic clinical series showed poor results with frequent graft rejection and failure to integrate within host tissue [8][9][10][11][12][13]. Very few orthopaedic studies have considered bone graft substitutes derived from porcine xenograft sources [14][15][16][17][18][19][20][21]. To the best of the authors' knowledge, there are currently no commercially available porcine-derived bone graft products available for clinical use in orthopaedic surgery and there are no clinical reports of porcine bone grafting into human subjects in the orthopaedic literature. Porcine species share similar anatomy, physiology, and genetic makeup with human species [14,17] and historically have served as excellent xenograft donors to human recipients [22][23][24]. While orthopaedic literature has limited reports of xenograft use, facial reconstruction and periodontal literature has more experience with these products [25][26][27][28][29]. The biology of oral and facial grafting sites differs from sites of orthopaedic pathology treated with bone grafting but it is still important to recognize the recent successes seen with xenografts and various bone graft substitutes used in treatment of facial and periodontal pathology [30][31][32] with improved patient outcomes and quality of life [25,26]. This literature supports further consideration in orthopaedic applications as well.
A novel decellularization and chemical oxidation protocol developed in our laboratory has previously derived tendon, meniscus, and ligament scaffolds from both human and animal tissue [33][34][35][36][37]. Results from these studies showed that the decellularization technology can derive biocompatible and pathogen-free tissue scaffolds. We proposed that this protocol could be used to decellularize bone to derive scaffolds that could potentially serve as bone graft substitutes for clinical use. We hypothesize that cancellous bone in porcine femurs can be decellularized to produce a scaffold that: (1) is devoid of donor cellular material; (2) maintains the key ultrastructural properties of native bone critical for osteoconductive potential; (3) maintains similar biomechanical properties to donor bone; and (4) has a protein composition similar to DBM, an analogous human-derived bone graft product currently in clinical use with proven osteoinductive potential [5,38,39].

Macroscopic Field Images
Donor porcine bone ( Figure 1A) post-oxidation was devoid of adipose marrow contents and took on a bright white, uniform appearance. Post-lyophilization, the scaffolds ( Figure 1B) became more brittle with obvious reduction in density. Decellularized scaffolds were similar in appearance and texture to demineralized bone matrix (DBM) ( Figure 1C).

Scanning Electron Microscopy
Scanning electron micrographs highlighted the porous surface architecture present on both the donor bone and scaffold ( Figure 2). Overall microarchitecture of scaffolds was markedly similar to unprocessed donor bone, suggesting that the decellularization protocol did not grossly compromise the osteoconductive ultrastructure. Blade etchings from commercial production were visible on DBM surfaces ( Figure 2, DBM 30× magnification). Pores on DBM samples were notably larger than those on porcine bone and scaffold specimens. Comparative Scanning Electron Micrographs: Donor porcine cancellous bone ("Bone") was decellularized to derived bone scaffolds. Representative micrographs appreciate the porous surface architecture. Demineralized Bone Matrix (DBM) is a commercially available human cancellous bone graft product. DBM pores appeared larger but high magnification surface was similar. Bone appeared less porous with marrow contents filling pores.

Scanning Electron Microscopy
Scanning electron micrographs highlighted the porous surface architecture present on both the donor bone and scaffold ( Figure 2). Overall microarchitecture of scaffolds was markedly similar to unprocessed donor bone, suggesting that the decellularization protocol did not grossly compromise the osteoconductive ultrastructure. Blade etchings from commercial production were visible on DBM surfaces ( Figure 2, DBM 30× magnification). Pores on DBM samples were notably larger than those on porcine bone and scaffold specimens.

Scanning Electron Microscopy
Scanning electron micrographs highlighted the porous surface architecture present on both the donor bone and scaffold ( Figure 2). Overall microarchitecture of scaffolds was markedly similar to unprocessed donor bone, suggesting that the decellularization protocol did not grossly compromise the osteoconductive ultrastructure. Blade etchings from commercial production were visible on DBM surfaces ( Figure 2, DBM 30× magnification). Pores on DBM samples were notably larger than those on porcine bone and scaffold specimens.  Comparative Scanning Electron Micrographs: Donor porcine cancellous bone ("Bone") was decellularized to derived bone scaffolds. Representative micrographs appreciate the porous surface architecture. Demineralized Bone Matrix (DBM) is a commercially available human cancellous bone graft product. DBM pores appeared larger but high magnification surface was similar. Bone appeared less porous with marrow contents filling pores.

Micro-CT Imaging
Micro-CT imaging showed similar results for both the porcine bone and scaffold specimens with no change in gross architecture ( Figure 3). The decellularization protocol uses a weak acid (peracetic acid) but this did not demineralize the substrate enough to yield changes detectable by micro-CT imaging.

Micro-CT Imaging
Micro-CT imaging showed similar results for both the porcine bone and scaffold specimens with no change in gross architecture ( Figure 3). The decellularization protocol uses a weak acid (peracetic acid) but this did not demineralize the substrate enough to yield changes detectable by micro-CT imaging.

Histology
H&E stained sections showed that marrow contents ( Figure 4A,B) and osteocytes within lacunae of donor bone were removed after decellularization ( Figure 4C,D). Consistent with these findings, 4 ,6-diamidino-2-phenylindol (DAPI) stained sections showed an abundance of nuclear material in the source bone ( Figure 4F,G) that was removed after decellularization ( Figure 4H,I). DBM sections ( Figure 4E,J,O) were similar in appearance to scaffold sections.

Histology
H&E stained sections showed that marrow contents ( Figure 4A,B) and osteocytes within lacunae of donor bone were removed after decellularization ( Figure 4C,D). Consistent with these findings, 4′,6-diamidino-2-phenylindol (DAPI) stained sections showed an abundance of nuclear material in the source bone ( Figure 4F,G) that was removed after decellularization ( Figure 4H,I). DBM sections ( Figure 4E,J,O) were similar in appearance to scaffold sections.

Measurement of DNA Content
DNA concentration in donor bone was 14.52 ± 7.7 μg/mL and 0.244 ± 0.09 μg/mL in scaffolds (p = 0.001), representing a 98% reduction in DNA content achieved by the decellularization protocol.

Measurement of DNA Content
DNA concentration in donor bone was 14.52 ± 7.7 µg/mL and 0.244 ± 0.09 µg/mL in scaffolds (p = 0.001), representing a 98% reduction in DNA content achieved by the decellularization protocol.

Ultrastructure Measurements
The density of donor bone samples (Table 1) was significantly greater than decellularized scaffold samples (Bone density 1366 ± 19.82 mg/mL, Scaffold density 570.4 ± 91.32 mg/mL, p < 0.0001). Porosity was slightly greater in scaffolds (79.5 ± 9.1%) than donor bone (69.1 ± 11.1%) but this difference was insignificant (p = 0.20). Pore size was also similar (p = 0.77) between scaffolds (474.2 ± 76.2 µm) and bone specimens (458.5 ± 66.3 µm). Thickness of the bone struts in the trabecular architecture as well as anisotropy were not different between donor bone and scaffold specimens ( Table 1). Donor bone was decellularized to create scaffolds. Density and ultrastructure parameters are presented as the mean ± standard error and compared by independent samples t-tests.

Finite Element Analysis (FEA)
Finite element analysis (FEA) of bone and scaffold specimens ( Table 2) showed no significant differences in yield stress (von Mises), stiffness, or failure loads for the subvolumes of samples analyzed. There was significant variability in scaffold samples however (Table 1). Biomechanical properties of representative unprocessed bone and decellularized scaffolds are presented as the mean ± standard error and compared by independent samples t-tests. Parameters were measured by finite element analysis (FEA) modeling as well as mechanical compression testing on an Instron ® MTS.

Mechanical Compression Testing
Mechanical compression testing results supported the trends shown in the FEA analysis (Table 2). Significant differences were found between bone and scaffold specimens for maximum strain at failure (p < 0.0001), stiffness (p < 0.0001), and the elastic modulus (Young's modulus p < 0.0001). Load at failure was not different between scaffold and bone specimens.

Proteomics
Protein composition identified by mass spectrometry analysis of porcine bone scaffold and human DBM samples were similar (Table 3). A variety of structural proteins serving to support the extracellular matrix as well as proteins providing particular biologic functions were identified. Proteins supporting a variety of intra and extracellular signaling pathways (Chondroadherin, Lumican, Biglycan) were found in nearly all porcine and human specimens. Proteins found in blood (Annexin A5, Hemoglobin subunit beta) were identified in more pig than human specimens likely due to differences in specimen preparation and processing techniques between our methods and those of the commercial vendor.

Discussion
Our decellularization protocol effectively removed porcine cellular material, supported by the 98% reduction in DNA content observed. Removal of donor cellular material is critical in preventing host mediated macrophage response after transplantation [41]. In previous studies applying the same decellularization protocol to different soft tissues, we recorded a 67% reduction in DNA content from avian FDP tendons [37], a 55% reduction from ovine menisci [34], and a 61% reduction from porcine patellar tendons [33]. The relatively increased porosity of cancellous bone likely permitted deeper penetration of the treatment chemicals and ultimately resulted in more efficient decellularization in comparison to the other less porous soft tissues previously treated. Comparative histology of commercial DBM specimens was similar in appearance and also devoid of cellular material on DAPI sections. Decellularization did not appear to alter the micro-architecture of the bone scaffold critical for preservation of osteoconductive potential.
Density of porcine cancellous bone taken from the distal femoral metaphysis was~1.4 g per cubic centimeter, which is slightly greater than reports of human cancellous bone harvested from the calcaneus (1.135 g/cm 3 ) or iliac crest (1.098 g/cm 3 ) [4]. While decellularization did significantly reduce density of processed specimens, it did not significantly alter the micro-architecture of the donor bone. The significance of micro-architecture to osteoconductivity is well established [42], and bone graft substitutes should generally seek to mimic native bone architecture [43][44][45][46][47]. Of critical importance are the porosity and pore size of scaffolds, which affect efficiency of cell seeding, diffusion into the scaffold, and the mechanical strength of the scaffold [48]. Highly porous structures permit higher mass transport, vascularization, osteo-integration from adjacent bone, and infiltration of both osteoblasts and osteoclasts to mediate scaffold remodeling [49,50]. Less porous structures are stronger and have greater surface area for protein adherence and cell attachment, but with less porosity this is limited to the periphery of the structure and ultimately limits ingrowth and cell penetration [43,49,50]. Porosity of human cortical bone is 3-12% while porosity of cancellous trabecular bone is highly variable ranging from 50% to 90% depending on location and donor biology [4,42]. We saw a range of porosity in our porcine donor specimens that was similar to this human data. Decellularized scaffold porosity was 80% on average. This is comparable to the reported porosity of DBM which ranges from 62% to 88% [51,52]. Objective parameters for pore size are debated in the tissue engineering literature with a pore diameter of 100-500 µm considered ideal [43,53]. Pore diameter greater than 40 µm is required for osseous ingrowth, and osteoid formation requires a minimum pore diameter of 100 µm [4,43]. Pore size was highly variable in our specimens typically falling between 400 and 600 µm, which are defined as large pores [48,49]. By comparison, DBM pore size averages 200-500 µm [54], but is highly variable when derived from cancellous bone. Limited data on porcine demineralized cancellous bone is available, however, one study [55] reported findings similar to ours with 62% porosity and 520 µm average pore size in specimens derived from the proximal metaphysis of porcine femurs.
Similar to porosity, the biomechanical properties of cancellous bone are also highly variable and depend on bone location, as well as a number of donor variables such as specie, gender, and age [4,[56][57][58]. Young's modulus of human cancellous bone ranges from 50 to 389 MPa [16,59], with some reports in the GPa range. This depends on the orientation of specimen loading owing to the effects of anisotropy [42]. Similar to human cancellous bone, we found porcine cancellous bone average modulus in the range of 200-300 MPa. After decellularization, we consistently found that bone scaffolds were less stiff and had greater deformation at failure. Given that matrix density is directly proportional to its modulus [56,60,61], this result was partially expected knowing that bone specimens were significantly more dense than scaffold specimens. Chemical modifications made to the bone during decellularization also explain the reduction in elastic modulus. Although peracetic acid (PAA) is a weak acid, it does demineralize bone to a small extent which softens the structure, makes it tougher, and decreases elastic modulus [62]. Lyophilization of tissues has also been shown to compromise mechanical strength [62][63][64] and may have altered the elastic modulus in the final stage of scaffold production. When lyophilized specimens are rehydrated, as we did before destructive mechanical testing, they typically lose 20% stiffness and 18.9% ultimate strength [63]. Rehydration of specimens more closely mimics physiologic conditions, however, and was appropriate for testing.
Scaffolds derived from cancellous bone blocks are not intended to serve as load-bearing structures. However, the biomechanical properties of the structure still bear clinical significance in supporting bone regeneration. Bone grafts and scaffolds should mimic the fundamental mechanical properties of bone and closely match the surrounding tissue's compressive strength, toughness, and stiffness as closely as possible [65,66]. The scaffold cannot support large physiologic loads, but it must maintain its structure long enough to preserve the void it fills until remodeled by host ingrowth [65]. Without appropriate stiffness and strength at the implantation site, the scaffold may be resorbed by the host which will prevent progressive bone formation, remodeling, and osseous union with host ingrowth [66]. Stiffness of the scaffold has also been shown to direct stem cell fate with softer matrices directing cells to neurogenic lineages and stiffer matrices mimicking the cross-linked collagen structure of osteoid that pushes cells towards osteogenic lineages [67][68][69][70]. As the cells pull on the matrix through focal adhesions, they gauge the matrix's stiffness by the amount of resistance they sense [67]. Stiff matrices are defined by elastic moduli greater than 25 kPa and can have osteoinductive effects on stem cells [68]. Although significantly lower than donor bone, the modulus of our scaffold was on the order of MPa, thus making it a stiff matrix capable of differentiating stem cells down osteogenic pathways. The above biomechanical indices are likely affected by anatomic location of the donor bone. We harvested bone from the distal metaphysis of femurs because they are weight bearing bones we found to yield the largest quantity of cancellous bone for processing.
Very few studies use mass spectrometry to analyze protein composition of human DBM [71][72][73], and to our knowledge, no study has compared protein profiles of a xenograft derived bone scaffold against human DBM. The protein composition of the structural and biologic elements of our scaffold was remarkably similar to human DBM, a clinical product with proven osteoconductive and osteoinductive potential. Despite deriving our scaffold from a non-human species, this data is critical in demonstrating the structural and functional similarity of porcine and human-derived bone scaffolds, suggesting they may serve similar biologic as well as structural roles in bone grafting.
Several limitations exist in the current study. We were unable to control some porcine donor variables such as potential comorbidities, age, and specimen contamination at time of acquisition. All donor bone was harvested from the distal metaphysis of femurs, but we did not control for the variable density of the cancellous bone which has previously been shown to affect cell attachment and differentiation [52]. While histology showed convincing evidence of decellularization and DNA quantification supported this finding, it is unclear if DNA reduction is the most accurate assessment of effective decellularization. Cellular remnants and extracellular proteins not measured by DNA quantification may be as immunogenic as whole cells [41]. The extent of bone demineralization caused by peracetic acid treatment in our protocol was not defined in the current study. The biomechanical consequences of this cannot be distinguished from the effects of lyophilization or potential degradation of the organic matrix. Lastly, proteomics data is limited by the unknown biologic activity of detected proteins. Mass spectrometry identifies proteins by primary structure alone and does not characterize secondary, tertiary, or quarternary protein structures which are critical to protein function and biologic activity. Despite preserving proteins believed to impart osteoinductive potential, we cannot conclude that these porcine proteins will impart the same effects on human cells. Further in vivo study will be required to determine if our xenograft will yield clinical translation and those animal studies have not been performed. We can justify further in vivo study based on some of the results seen in the facial reconstruction and periodontal literature [25,26,28,29,31,[74][75][76][77]. A recent review by Cicciu et al. [26] highlighted the clinical and animal models in facial bone reconstruction using various xenograft derivatives with particular attention to marine origins of bone graft substitutes. A non-human primate study by Herford et al. [75] treated mandible defects with a combination of bovine bone and showed that addition of recombinant human BMP-2 increased bone and soft tissue regeneration with distraction osteogenesis. The additive effects of BMP are another possibility to explore with our porcine xenograft given the positive results seen in other animal models. In vivo animal model study with our porcine xenograft will be the next essential step in study of its osteoinductive potential and potential use in clinical practice.
Results presented in the current study show that our decellularization protocol can derive a scaffold from porcine bone that is devoid of donor cellular material while preserving the microarchitecture critical to osteoconduction, and, potentially, the growth factors thought to contribute to the osteoinductive potential of bone graft substitutes. While the overall structural parameters and yield strength were preserved after decellularization, the protocol did alter the biomechanical properties of the starting material as seen by significant decreases in bone stiffness.

Acquisition of Animal Tissue
The femurs from freshly sacrificed female pigs (Sus scrofa domesticus), aged 3-4 years old, weighing 450-650 lbs were obtained from City Packing Company (Burlington, NC, USA) with an educational and research permit granted by the North Carolina Department of Agriculture and Consumer Services. Pigs are commonly slaughtered at this age, therefore, we elected to study pigs of this age to be representative of the potential porcine donor pool. Femurs were transported to our laboratory on ice and immediately stripped of soft tissue before being frozen at −20 • C for later use.

Physical Processing of Specimens
Cross sections 6-10 mm thick were taken from the distal end of femurs using an upright stainless-steel autopsy grade band saw (Band Saw K220, KUGEL Medical, Regensburg, Germany). Only sections taken from the metaphysis were kept for processing, while specimens containing articular and subchondral elements were discarded. Sections were mounted to a drill press (DELTA Power Equipment Corporation, Milwaukee, WI, USA) and bone plugs were machined with a 3/8" plug cutter (Harbor Freight Tools, Calabasas, CA, USA). To prevent thermal necrosis of specimens, the bone sections were drilled while submerged in ice-chilled normal saline. Bone plugs were then washed at high pressure to flush out marrow contents.

Decellularization and Chemical Oxidation Protocol
Specimens were processed as previously described [37] using a patented decellularization and oxidation protocol [78]. Briefly, cancellous bone plugs were transferred into autoclaved screw-top glass flasks (Corning Inc., Corning, NY, USA) containing 1000 mL of DNase/RNase-free, distilled water (Gibco, Grand Island, NY, USA). Flasks were placed on a rotating shaker (Barnstead MaxQ400, Dubuque, IA, USA) at 200 rpm, 37 • C, for 72 h with fresh water exchanges every 12 h. Water was discarded and 500 mL of 0.05% trypsin-EDTA (Gibco, Grand Island, NY, USA) was added to flasks before returning them to the rotating shaker. After 2 h, the trypsin solution was discarded and 500 mL of Dulbecco's Modified Eagle's Medium (DMEM) high-glucose (Gibco, Grand Island, NY, USA) containing 10% fetal bovine serum (FBS) (Valley Labs, Winchester, VA, USA) and 100 I.U./mL Penicillin, 100 µg/mL Streptomycin, 0.25 µg/mL Amphotercin B (Gibco, Grand Island, NY, USA) was added to halt trypsin digestion and maintain an aseptic preparation. Flasks were returned to the shaker for 24 h before discarding the DMEM solution, adding 1000 mL of fresh DNasefree/RNase-free distilled water and returning it to the shaker for 24 h. Water was discarded, and a 1000 mL solution of 1.5% Peracetic Acid (Sigma, St. Louis, MO, USA), and 2.0% Triton X-100 (Sigma) prepared in distilled, deionized water was added, and flasks were returned to the rotary shaker for 3 h. The oxidizing solution was discarded and residual peracetic acid was removed by flushing the samples with 1000 mL autoclaved, deionized water and then repeated washes in 1-3 h cycles (each at 200 rpm, 37 • C).
Wash cycles were repeated until peracetic acid was no longer detectable by peracetic acid test strips (Sigma). Specimens were then transferred into conical tubes, frozen at −80 • C for 24 h, lyophilized (LabConco, Freeze Dry System, Kansas City, MO, USA) for 24 h, and stored at −80 • C until further use.

Scanning Electron Microscopy
Donor porcine bone, scaffolds, and DBM samples were mounted on aluminum stubs (Ted Pella, Redding, CA, USA) with carbon tape and gold sputter coated at 30 mTorr. Specimens were imaged on a Hitachi S-2600 scanning electron microscope (Hitachi, Tokyo, Japan).

Micro Computed Tomography
Scaffold and bone specimens were scanned in cross section at 10 µm voxel resolution and reconstructed into 3d images on a Scanco 80 micro-CT system (Scanco USA, Wayne, PA, USA). Images were inverted to visualize the porous domains so pore size and morphology could be quantified using methods previously described by Lin et al. [59]. Additional specimens were scanned on a nanotom micro-CT (phoenix nanotom ® m, General Electric, Fairfield, CT, USA) for high resolution images.

DNA Quantification
DNA content was measured in the donor cancellous bone (n = 6 femurs) and decellularized scaffold to assess efficacy of decellularization. DNA content was normalized to specimen volume to correct for changes in specimen density after decellularization. Volume was calculated by measuring the height and diameter of specimen plugs with a jeweler's caliper (Esslinger, St. Paul, MN, USA). Samples were individually cryomilled (6870 Freezer/Mill ® , Thomas Scientific, Swedesboro, NJ, USA) in liquid nitrogen, and lysed in 3× volume of mammalian protein extraction reagent (M-PER ® , Thermo Scientific, Rockford, IL, USA) with protease inhibitors (cOmplete ULTRA Tablets, Roche, Basel, Switzerland). DNA concentration of lysates was measured with Quant-iT™ PicoGreen ® dsDNA Assay Kit (Thermo Scientific) in polystyrene plates (Corning ® 96-well) on a fluorescence microplate reader (FLUOstar Optima Plate Reader, BMG Labtech, Ortenberg, Germany).

Density, Porosity, Microstructure
Uniform bone plugs were machined from distal femur metaphyseal cross sections, using the methods previously described. Bone plug volume was measured in cubic millimeters and converted into milliliters while dry weight was measured in milligrams. Microstructure measurements were derived from micro-CT imaging. Images were inverted to visualize porous domains and calculate average pore diameter. Porosity was calculated as percentage by subtracting specimen volume fraction from 1 [59]. Trabecular strut diameter, density, and spacing were measured and compared between bone and scaffold samples.

Finite Element Analysis
DICOM files (Digital Imaging and Communications in Medicine) from bone and scaffold specimens were imported into Scanco analysis software for non-destructive biomechanical testing using voxel-based finite element analysis (FEA) [79]. Testing was performed on 15 mm 3 sub-volumes of micro-CT specimens. Sample strength was defined by yield stress (von Mises stress). Because naturally-derived bone scaffolds undergo remodeling in the host until eventually being replaced, yield stress was of particular interest in biomechanical assessment [79,80].

Mechanical Compression Testing
Bone (n = 12) and scaffold (n = 12) specimens from the same individual femurs were machined into uniform cylinders using a drill press-mounted diamond hole saw bit (Lenox, East Longmeadow, MA) and stored at −80 • C. Specimens were soaked in normal saline for two h at room temperature prior to mounting them between stainless steel platens for axial compression testing on a material testing machine (MTS) (Instron ® , Norwood, MA, USA). Mounted specimens were preloaded to 5 N in a 2 kN load cell and compressed with a crosshead speed of 0.5 mm/s until fracture or 80% strain was reached [64,81,82]. Ultimate stress and strain were calculated from the first peak of the generated stress-strain curve (Bluehill ® Software, Instron, Norwood, MA, USA) while Young's modulus was derived from the slope of the stress-strain curve's linear portion.

Mass Spectrometry
The identities of proteins within the porcine bone scaffolds (n = 3) were analyzed using mass spectrometry (MS). For comparison against an analogous commercially available bone graft product, cancellous demineralized bone matrix (DBM) (Musculoskeletal Transplant Foundation, Edison, NJ, USA) samples (n = 3) were also analyzed. Specimens were cryomilled in liquid nitrogen, reduced with dithiothreitol, and alkylated with iodoacetamide before overnight digestion with trypsin (Trypsin Gold, Promega, Madison, WI, USA) at pH 8.0. Mass spectrometry analysis was performed using an electrospray ionization (ESI) ion trap LTQ Orbitrap XL hybrid FTMS (Fourier Transform Mass Spectrometer) (Thermo Fisher Scientific, Waltham, MA, USA) coupled to a nanoACQUITY UltraPerformance LC (UPLC) system (Waters, Milford, MA, USA). The Mascot version 2.2.07 (Matrix Science, Boston, MA, USA) search engine was used to search the protein sequence databases (Swiss Prot and National Center for Biotechnology Information (NCBI) NCBInr) with the following settings: MS/MS ion search, taxonomy filter = Sus scrofa (pig)/Homo sapiens (human) all entries, enzyme = trypsin, maximum missed cleavages = 1, fixed modifications = carbamidomethyl (C), variable modifications = deamidated (NQ)/formyl (N-term)/oxidation of Met and Pro (to detect hydroxyl proline in collagens). MS hits with protein scores ≥100 were reported [83,84]. Results were queried in ExPASy proteomics Bioinformatics Resource Protocol, as well as the UniProt Knowledgebase (UniProtKB) [40]. Functional information of queried protein sequences along with their referenced accession numbers was extracted from UniProtKB. Analysis of native donor porcine bone was deferred because protein density would preclude informative analysis.

Statistical Analysis
Values are reported as the mean ± standard error and compared with independent samples t-tests with α = 0.05 to determine significance.

Conclusions
Results presented in the current study show that our decellularization protocol can derive a scaffold from porcine bone that is devoid of donor cellular material while preserving the microarchitecture critical to osteoconduction, and potentially, the growth factors thought to contribute to the osteoinductive potential of bone graft substitutes.