2.1 Materials

Recombinant human bone morphogenetic protein 2 (rhBMP-2), sterile water, and collagen sponges were obtained from Medtronic (Minneapolis, MN) as part of an XX Small Infuse Synthetic Bone Graft kit. Keratin in both the keratose (KOS) and kerateine (KTN) forms was obtained from KeraNetics, LLC and used without further modification. DyLight800 (DL800), AlexaFluor488 NHS ester, and Nalgene 890 FEP tubing were obtained from ThermoFisher Scientific.

2.2 Keratin scaffold preparation

Keratin scaffolds were prepared by first formulating 15% w/v keratin hydrogels with varying ratios of KOS and KTN by methods similar to those previously described [14, 15]. Powders of keratose, kerateine or mixtures of keratose and kerateine were weighed in sterile fashion into 50 mL conical tubes. Then, 500 μL of sterile water with or without rhBMP-2 was added to the keratin powders and then thoroughly mixed with a sterile spatula followed by vortexing to ensure even loading of the rhBMP-2 within the keratin. Each of the keratin formulations was prepared in the same manner. In some experiments, the rhBMP-2 was labeled with AlexaFluor 488 (AF488) or DL800 (see “Fluorescent Labeling of rhBMP-2” section, below). For in vitro studies, 100 μL of the keratin hydrogels were used instead of scaffolds due to absorption of water into the scaffolds that impacted the accuracy of initial time points. For in vivo studies, the keratin hydrogels were packed into 1.5 cm long × 4.8 mm inside diameter FEP tubes with a small amount of air pressure applied by syringes attached to each end to ensure complete packing of keratin in the tubes. The keratins were then allowed to gel overnight. The resulting gels were frozen at − 80 °C for at least 24 h and then lyophilized on a Labconco Freezone lyophilizer (Labconco, Kansas City, MO). We refer to these freeze-dried keratins as keratin scaffolds. After lyophilization, the scaffolds for in vitro or in vivo studies were stored in sterile containers under vacuum until use. Typical preparations for the keratin scaffolds are shown in Table 1. For the in vitro studies, the starting amount of rhBMP-2 was 10 μg (in 100 μL volume of keratin). As described below, the actual amount of rhBMP-2 implanted at the femur defect site for the in vivo studies was 14.2 μg.

Table 1 Typical preparation of keratin hydrogels with different KOS and KTN masses2.3 Rat femur defect model and implantation of biomaterials

Femur defect surgeries were conducted with the approval of the Miami University Institutional Animal Care and Use Committee (IACUC). Approximately 3-month old Sprague–Dawley male rats (average weight of 371.5 g) were randomly assigned to experimental groups. Rats were induced with isoflurane (3–5%) and then maintained with isoflurane via nosecone at 1–3%. The left hindlimb was shaved over the femur. The left femur was exposed by making incisions through the skin and then the muscle along the longitudinal axis of the femur. A custom-built internal fixator device was designed and implanted as described previously [16, 17]. In brief, a pilot hole was drilled into the left femur with a wire. The fixator device was placed midway between the proximal and distal ends of the femur. Two gold-plated stainless steel (grade 303) screws were used to secure the fixator to the bone. An 8 mm segment of bone was removed by using a pneumatic reciprocating saw. The bone tissue was removed from the defect site and the defect site was rinsed with sterile saline to remove any bone fragments. The size of the defect was made with guides in the fixator set at 8 mm.

Collagen or keratin biomaterials were then implanted at the site of the defect, with the exception that negative controls received no biomaterial implants (empty). Keratin scaffolds were prepared as described above. The freeze-dried scaffolds were removed from the FEP tubes and the scaffolds were then cut to length in the surgical field at a length slightly longer than 8 mm. We assumed the the keratin with rhBMP-2 was well-mixed. Based on the volume of the keratin scaffolds implanted (142 mm3 or 0.142cm3) and the rhBMP-2 concentration used to prepare the hydrogels prior to freeze-drying (100 μg/mL), the implanted mass of rhBMP-2 was approximately 14.2 µg.

For collagen, an 8 mm segment of collagen was cut by scissors from the XX Small Infuse kit. 14.2 μg of rhBMP-2 in water (142 μL of 100 μg/mL rhBMP-2) was then added to ensure that the mass of rhBMP-2 was the same for collagen as for keratin. The rhBMP-2 was allowed to adsorb to the collagen for 15 min according to the manufacturer’s recommendation. The swollen collagen sponge was then implanted at the defect site.

No additional means were used to secure collagen sponges or keratin scaffolds at the implant site. The site was then closed in layers (muscle followed by skin) and the skin was stapled. Sutures and staples were removed within 14 days following surgery.

In the retention study described below, the rhBMP-2 was labeled with DL800 before being added to keratin or collagen biomaterials. All other methods remained as described above.

2.4 μ-CT image collection

For μ-CT imaging, animals from Miami University were transported to the University of Cincinnati Vontz Imaging Center at 4 and 8 weeks after implantation as approved on the Miami University IACUC protocol. Upon arrival at the Vontz Center, rats were temporarily placed on an imaging protocol approved by the University of Cincinnati IACUC. Rats were anesthetized with isoflurane (3–5% to induce, then maintained at 1—1.5% via nosecone). Animals were placed in the prone position with their left hindlimb extended. Rats were imaged on a Siemens μ-CT unit with a single rotation about the axis with 80 kV voltage, 500 microAmp current, 450 ms exposure at 360 steps to achieve voxel resolution of ~ 52 cubic microns. After imaging, rats were transferred back to the Miami University protocol and returned to Miami University for housing (4-week time point) or humane euthanasia (8-week time point).

2.5 μ-computed tomography image analysis of bone regeneration

DICOM images from the μ-CT scans were reconstructed in Osirix or Horos DICOM-viewers with 3D volume rendering. Bridging of the defects was determined by visualization of the resulting scans. Regions of interest (ROI) were drawn manually in Osirix in the defect region from every tenth image and interpolated to calculate the volume of new bone growth. Bone mineral density (BMD) was determined by gray scale standards run with each μ-CT scan and determined by gray-scale analysis in ImageJ (NIH, USA). Each of these processes were conducted by a blinded investigator.

2.6 Bone harvest and mechanical testing

After humane euthanasia, left and right (uninjured contralateral) femurs from all rats except one were immediately dissected and measured by caliper at hip, knee, and central diaphysis. In total, 5 left and right femurs were used for mechanical testing (N = 5) with the other femur (N = 1) used for histology (see below). Mechanical testing was conducted by the four-point bend method in which the superior side of the bone was placed in the downward direction. The bottom fixtures were set at L = 26 mm apart and top fixtures at (L-2a) = 10 mm apart. The hip was placed to the right side. Specimens were measured on an Instron Mechanical tester apparatus equipped with a 1000 N load cell. The bones were then pre-loaded in compression to 0.5N from the top fixture. The bones were compressed at a crosshead speed of 1 mm/min until fracture. Load and displacement were measured and converted to engineering stress and strain measurements as described previously [18].

In brief, \(\sigma =\frac \text \varepsilon =\frac\)where σ sigma is engineering stress, ε is engineering strain, F is the measured force in compression, L is the length between the bottom fixtures, c is the maximum distance from the cross-sectional center of mass (taken to be the minor axis of the bone) to the outside edge of the bone, d is the displacement of the top supports relative to the bottom supports, a is the distance from one top fixture to one bottom fixture (8 mm), and I is the second moment of the cross-sectional area relative to the horizontal axis. Modulus was determined as the maximum slope of the stress vs. strain curves. A running average was used to obtain the maximum slope.

Due to ectopic growth in some animals, we assumed that the diameter of the specimen at the sites of load application was the average of the knee and hip widths and diameters in an oval shape. Further, we could not reliably measure the inner and outer bone diameters and therefore assumed a solid cylinder for modulus calculations. This assumption as well as the use of a four-point bend likely underestimates the true modulus [18], but we report the modulus as the percentage of the contralateral limb in order to normalize the data.

2.7 Histology

One bone sample was used for histological imaging. The tissue was fixed for three days in 10% Neutral Buffered Formalin (NBF) and then decalcified for three days in Immunocal (Decal Chemical Corporation, Tallman, NY). Decalcification was stopped with Cal-Arrest (Decal Chemical Corporation, Tallman, NY). Bone tissue was soaked in 30% sucrose overnight, and then snap frozen in OCT Compound for frozen sectioning. Tissue was sectioned in the longitudinal direction on a Leica cryostat. Sections were collected on positively-charged slides, dried overnight, stained with Masson’s Trichrome (Master Tech, Lodi, CA) or hematoxylin and eosin (H&E), and imaged at the Miami University Center for Advanced Microscopy and Imaging (CAMI) on a Nikon 300 microscope equipped with a 2X objective. Images of individual sections were stitched together into mosaics by using GNU Image Manipulation Program (GIMP).

2.8 Fluorescent labeling of rhBMP-2

1.35 mL of 1 mg/mL of rhBMP-2 was prepared in DI water. 37.8 μL of DyLight800 (DL800) NHS ester in DMF was added and the solution was mixed in the dark by magnetic stirring at room temperature for 1 h. Two (2) μL of 0.5 M HCl was then added to stop the reaction. The solution was then transferred to a 3500 Da MWCO Slide-A-Lyzer membrane (ThermoFisher Scientific) and dialyzed exhaustively with three changes of DI water. The DL800-labeled rhBMP-2 was removed from the dialysis cassette, measured at 280 nm for determining rhBMP-2 concentration against a standard curve. The resulting DL800-rhBMP-2 was then diluted to 1 mg/mL in water. The DL800 concentration was determined by absorbance readings at 777 nm. A mole dye:mole protein ratio of 2 was typical (i.e., average of 2 dye molecules per rhBMP-2 molecule). This solution of DL800-rhBMP-2 was stored at − 20 °C until use. The DL800-rhBMP-2 was added (“spiked”) to unlabeled rhBMP-2 at 1 part in 15 (1:15) ratio and then used for preparation of collagen sponges or keratin scaffolds for implantation into the femur defect model, as described above.

For an in vitro study, the same method was used to label rhBMP-2 with the exception that AF488 NHS ester was used as the fluorophore.

2.9 In vitro release of AF488-labeled rhBMP-2

Keratin hydrogels were prepared as described in the “Keratin Scaffold Preparation” section where 100 μg/mL AF488-labeled rhBMP-2 was used and 100 μL the keratins hydrogels were placed into 1.5 mL tubes. 150 μL of PBS was placed on top of each keratin hydrogel formulation and removed at specified time points (1.5 h, 3 h, 6 h, 12 h, 24 h and then daily through 7 days). Keratin hydrogels that did not contain AF488-labeled rhBMP-2 were used as negative controls. Fluorescence was detected at excitation/emission wavelengths of 485 nm/528 nm. Autofluorescence from keratin proteins was determined from the negative controls (no rhBMP-2) and this amount of signal was subtracted from samples with AF488-labeled rhBMP-2. The concentration was determined by comparison to a standard curve of known concentrations of AF488-labeled rhBMP-2.

2.9.1 Femur retention and biodistribution of rhBMP-2 after implantation

To determine the amount of DL800-rhBMP-2 retained at the defect site, we first developed a standard curve for known amounts of DL800-rhBMP-2 at the defect site. 15% (w/v) KOS scaffolds were used for creating the standard curve for DL800 fluorescence. The DL800-rhBMP-2 was prepared at rhBMP-2 concentrations of 100, 50, 25, 12.5, 6.25, 3.125, 1.5625 or 0 μg/mL, which correspond to 14.2, 7.25, 3.625, 1.813, 0.906, 0.453, 0.227, and 0 μg implanted mass of rhBMP-2. The scaffolds were implanted as described in the femur defect model studies, above. Rats were randomly assigned to one of the DL800-rhBMP-2 concentrations for the standard curve and 3 rats (N = 3) were imaged for each DL800-rhBMP-2 concentration. After closure of the implant site, rats were humanely euthanized and frozen until transport to the University of Cincinnati Vontz Imaging center. The incision site was opened to expose the defect site and implanted materials. The rats were placed on a Bruker in vivo Multispectral Imaging System. One X-ray and one optical image were taken to confirm location of the femur defect and then the rats were imaged in the near-IR range by using a 730/790 Ex/Em filter set. Images were collected and pixel intensity was integrated with the Magnetic Lasso Tool in Photoshop to correlate DL800-rhBMP-2 concentration to fluorescence intensity. A standard curve was then created relating DL800-rhBMP-2 concentration with the measured pixel intensity, where it was assumed that the DL800-rhBMP-2 concentration was the same as initially implanted.

To determine the retention (where it was assumed that non-retained rhBMP-2 had been released from the implant site) of DL800-rhBMP-2 from keratin or collagen carriers at the implant site, rats were implanted with KOS, KTN, 50:50 KOS:KTN, or collagen containing 14.2 μg DL800-rhBMP-2. Rats were randomly assigned to each treatment group and four rats (N = 4 at Days 1 and 3, N = 5 at Day 7) were implanted for each formulation at each time point. Preparation of collagen and keratin as well as implantation in the femur defect model as described above. Rats were humanely euthanized and transported to the University of Cincinnati Vontz Imaging Center at 1, 3, or 7 days after surgical implantation of the scaffolds with rhBMP-2 (spiked at 1:15 with DL800-rhBMP-2). The incision site was opened to expose the defect site and imaged as described above for the standard curve. Pixel intensity was determined in Photoshop and the DL800-rhBMP-2 concentration was determined from the standard curve generated above. After imaging the femur defect site, the brain, liver, spleen, kidneys, heart and lungs were removed and imaged on the Bruker in vivo multispectral imaging system.

2.9.2 Statistical methods

Animal studies were randomized. Imaging and data analysis were conducted by blinded observers. We confirmed previous studies showing that keratin scaffolds lacking rhBMP-2 do not lead to bridging in this critically-sized defected model. To reduce the number of animals in the study, we combined results from empty (no treatment), keratose only (no rhBMP-2) and kerateine only (no rhBMP-2) and treated these as near replicates for statistical comparisons. These are shown as “Combined Controls” in the figures and for statistical comparisons, but where appropriate we have also shown the results for the Combined Control groups. Comparisons between groups were conducted with analysis of variance followed by Tukey’s honestly significant different post-hoc test.