Synthesis of graphene oxide (GO)
GO was synthesized using a method we have previously reported24,25. Specifically, GO was synthesized using a modified Hummers’ method85. Here, 10 g of graphite flakes (graphite flake, natural, − 325 mesh, 99.8% metal basis; Alfa Aesar, Ward Hill, MA, USA) were dispersed in 250 mL of concentrated sulfuric acid (Fisher Scientific, Pittsburgh, PA, USA) in a 2 L Erlenmeyer flask. The graphite dispersion was placed over ice and cooled. Then, 20 g of KMnO4 (Sigma-Aldrich, St. Louis, MO, USA) was slowly added in small aliquots over 20–30 min with stirring while maintaining the ice bath. The ice bath was removed, and the mixture was allowed to warm to room temperature and stirred for 2 h, followed by gentle heating to 35 °C and an additional 2 h of stirring. The heat was then removed, and the reaction was quenched by quickly adding 1400 mL of deionized (DI) water followed by the slow addition of 20 mL of 30% hydrogen peroxide (Fisher Scientific). The remainder of the DI water (450 mL) was slowly added at first until bubbling ceased and then added in one aliquot. The mixture was left to stir at room temperature overnight.
GO was purified via vacuum filtration through a Buchner funnel with large cellulose filter paper. Following filtration, the filtrate was discarded, and the GO was carefully removed from the Buchner funnel without scraping the cellulose filter paper. GO was directly added into 3500 molecular weight cutoff dialysis tubing (SnakeSkin dialysis tubing, Thermo Scientific, Waltham, MA, USA). Then, GO was dialyzed against DI water for 3–5 days until the water was clear. On the first day, the DI water was changed twice and then changed once each day thereafter. The GO was transferred from the dialysis tubing into 50 mL conical centrifuge tubes, with no more than half the tube filled, and frozen to − 80 °C and lyophilized for 3–5 days until dry.
Synthesis of calcium phosphate graphene (CaPG)
CaPG was synthesized from GO using a modified Arbuzov method that we have reported previously24,25. Here, 500 mg of GO, 500 mg of magnesium bromide diethyl etherate (Alfa Aesar, Haverhill, MA, USA), and 500 mL of triethyl phosphite (Sigma Aldrich, St. Louis, MO, USA) were added in sequential order to a 1 L, flame dried round bottom flask under nitrogen. The mixture was bath sonicated (240 W, 42 kHz, ultrasonic cleaner, Kendal) for 1 h to ensure homogeneous dispersion of reagents followed by the addition of 2.5 g of anhydrous calcium bromide (Alfa Aesar, Haverhill, MA, USA). The mixture was then sonicated for an additional 30 min. After sonication, the reaction was heated to reflux (156 °C) under stirring and nitrogen. After a 72 h reflux, the heat was removed, and the reaction was cooled to room temperature and filtered via vacuum filtration through a Buchner funnel containing large cellulose filter paper. The filtrate was discarded and the CaPG filter puck was carefully removed from the funnel without scraping the filter paper and placed into 50 mL conical centrifuge tubes.
CaPG was purified by adding ~ 45 mL of fresh solvent to the centrifuge tube containing the material followed by vortexing and then centrifugation at 3600×g for 5 min. Centrifugation pelleted the CaPG to the bottom of the tube, and the supernatant was discarded. The CaPG pellet was re-dispersed in fresh solvent, vortexed, and centrifuged again for the next wash step. Several wash steps were repeated to purify CaPG that included 2 acetone, 1 ethanol, 1 DI water, and 2 acetone washes. The resulting CaPG pellet was then dried under vacuum for 24–48 h until dry.
Fourier-transform infrared (FTIR) spectroscopy
FTIR spectra were collected on a PerkinElmer Frontier FT-IR Spectrometer with an attenuated total reflectance (ATR) attachment, where the ATR attachment contained a germanium crystal. Raw spectra were recorded in percent transmittance from 4000 to 700 cm–1 with a 4 cm–1 resolution. ATR and baseline corrections of the raw spectra were performed in Spectrum software (Spectrum version 10, PerkinElmer, https://www.perkinelmer.com/product/software-kit-spectrum-10-lx108873). The data was smoothed with a boxcar of 10 and then offset for clarity.
Thermogravimetric analysis (TGA)
TGA was conducted on a PerkinElmer TGA 4000 using PerkinElmer ceramic TGA pans. The pans were cleaned, and flame dried prior to all TGA measurements. Further, TGA was performed from 50 to 800 °C with a heating rate of 10 °C/min under a nitrogen atmosphere with a 20 mL min–1 flow rate.
X-ray photoelectron spectroscopy (XPS)
All XPS analysis was conducted on a Thermo Fisher ESCALAB 250 Xi instrument using an Al K-Alpha source gun and a flood gun in charge compensation standard mode. Spectra were acquired using the standard lens mode (angle and field of view of 32,000 steps), Constant Analyzer Energy (CAE) scan mode, and a 200 μm analysis spot size. The graphenic samples were prepared by adhering the powders on double-sided copper tape. Care was taken to ensure that there was no loose powder and that the substrate was completely covered. Then, the double-sided copper tape containing the samples was mounted onto a sample boat for analysis.
Survey spectra were collected for graphenic materials using 5 cumulative scans per spectrum. Further, the survey spectra were acquired over a binding energy range of 1350 to − 10 eV, using a pass energy of 150 eV, an energy step size of 1.0 eV, and a dwell time of 10 ms. The elemental composition of each graphenic material was quantified from the survey spectrum by integrating the area under peaks unique to each element. Quantification was performed using CasaXPS software (CasaXPS Version 2.3.15, Casa Software Ltd, http://www.casaxps.com/) using a smart background and standard peak type.
High resolution XPS spectra were acquired and smoothed in OriginPro (version 2019b) using second order polynomial Savitzky-Golay smoothing. The C1s, P2p, and Ca2p spectra were smoothed with a 15, 25, and 35 points of window, respectively. Further, the C1s, P2p, and Ca2p XPS spectra were Shirley baseline subtracted in Fityk (version 1.3.1).
The C1s spectra were also peak fit in Fityk using procedures previously described24,25. Specifically, C1s spectra were fit using Gaussian peak shapes with a fixed full-width-at-half-maximum of 1.4 eV. The peaks for carbon-containing functional groups consisted of C–P, C–C/C=C, C–O, C=O, and O–C=O that were centered at 283.5, 284.8, 286.5, 287.4, and 289.0 eV, respectively with a variance of ± 0.2 eV. The area under the fitted peaks in the high resolution C1s spectra is reported in atomic percent.
Matrix printing
Matrices were printed by Dimension Inx (Chicago, IL). CaPG powder was sent to Dimension Inx, and Dimension Inx used its patented 3D printing technology to print high CaPG content matrices. Details of the ink formulation and print process have been reported in the literature64,65. The final product was a 3D matrix of 90 wt% CaPG and 10% poly(lactic-co-glycolic acid) (PLGA).
Through its research and development, Dimension Inx has identified matrix print parameters that are well suited for in vitro and in vivo applications. For this study, we purchased large sheets of each type (Supplementary Fig. 3) from which we were able to punch out matrices of desired size for experiments.
The in vitro matrix sheet was 3 cm × 3 cm × 0.1 cm. It consisted of 4 layers, of which each layer was 170 μm thick. The layers were 90° offset. The strut-to-strut distance was 700 μm. The spacing between adjacent fibers in same layer was 500 μm, and the strut diameter was 200 μm.
The in vivo matrix sheet was 3 cm × 3 cm × 0.05 cm. It consisted of 3 layers, of which each layer was 170 μm thick. The layers were 120° offset. The strut-to-strut distance was 700 μm. The spacing between adjacent fibers in same layer was 500 μm, and the strut diameter was 200 μm.
Matrix bulk density
The bulk density (ρbulk) of 3DP-CaPG matrices was determined by the matrix mass (mmatrix) by volume (Vmatrix):
$${rho }_{bulk}={m}_{matrix}/{V}_{matrix},$$
(1)
where mmatrix was measured on an analytical balance. The Vmatrix was calculated using the thickness and diameter of the cylindrical matrices that were measured with calipers.
Matrix total porosity
The total porosity (ϕtotal) of 3DP-CaPG matrices, expressed as a percent, was calculated using the following:
$${phi }_{total}=frac{{rho }_{T}-{rho }_{bulk}}{{rho }_{T}}times 100%,$$
(2)
where ρT is the theoretical density. The ρT was calculated using the rule of mixtures, which is a weighted average based on the composition of the material. In the case of 3DP-CaPG matrices, 90% is comprised of CaPG and 10% is PLGA that have theoretical densities of 2.26 g cm–3 and 1.30 g cm–3, respectively. Thus, the theoretical density of 3DP-CaPG matrices was calculated by the following:
$${rho }_{T} = 0.9{rho }_{graphite}+0.1{rho }_{PLGA}=2.16, left[text{g cm}^{-3}right].$$
(3)
Dynamic mechanical testing
All mechanical testing was conducted on a Discovery Hybrid Rheometer (TA Instruments, New Castle, DE) using a disposable, aluminum Peltier cylinder with a diameter of 25 mm as the bottom geometry and a sandblasted Peltier cylinder with a diameter of 8 mm as the top geometry. All measurements were acquired after applying a compressive 0.1 N pre-force.
Frequency sweeps were acquired in compression at room temperature with an axial strain of 0.3%. Data was acquired in triplicate for each disk over a frequency of 0.1–11.0 Hz with 10 points per decade. The storage and loss moduli of 3DP-CaPG matrices were reported at 1.0 Hz.
The stress–strain curves of the 3DP CaPG disks were measured in compression with a constant linear rate of 0.01 mm/s. The acquired data was not processed with a correction formula and is reported as-acquired. The ultimate compressive strengths (UCS) and maximum compressive strains (σF) were determined from the peak inflection on the stress–strain curves indicative of material failure. Toughness (UT) was calculated from the area under the stress–strain curves.
Cell culture conditions
Adipose–derived human mesenchymal stem cells (hMSCs) were purchased from Thermo Fisher Scientific (#R7788–115). The hMSCs were cultured in a humidified, 37 °C, 5% CO2 incubator (HERAcell 150i CO2 incubator with copper chamber, Thermo Fisher Scientific, #51026283) in filter cap 25 cm2 flasks (Greiner Bio-One CELLSTAR, #690175). “Growth media” is a commercially available formulation (Thermo Fisher Scientific) for the expansion and preservation of potency of hMSCs. Growth media consists of reduced-serum MesenPRO RS Medium (#12746012) that is supplemented with l-glutamine (#25030081) at a final concentration of 2 mM. Growth media was also spiked with penicillin/streptomycin (#15140122) at a final concentration of 100 U mL–1. “Osteogenic media” is a commercially available formulation (Thermo Fisher Scientific): StemPro Osteogenesis Differentiation Kit (#A1007201). Osteogenic media was also spiked with gentamicin (#15710064) diluted to 5 µg mL–1. For subculture, hMSCs were detached using TrypLE Express without phenol red (#12604013), as it participates in fewer non–specific reactions, decreasing cellular damage. No hMSC reagents contained phenol red since it affects osteogenic differentiation86. To ensure potency, hMSCs were not used beyond passage seven.
hMSCs on matrices
For experiments, we chose to use cylindrical matrices of a 3.5 mm diameter. This size ensured the ability to acquire reliable, valid data while not requiring an excessive amount of material or hMSCs. Further, it matched the diameter of the calvarial defects used in the mouse model. Cylindrical matrices were punched out of a large (3 cm × 3 cm × 0.1 cm) sheet that was 3D printed using a biopsy punch. The matrices were placed into individual wells of 48 well tissue culture plates (Greiner Bio-One CELLSTAR, #677180) and sterilized based on methods described by the manufacturer and reported in literature64,65. The matrices were bathed in 70% ethanol for 1 h. Then, they were washed three times with cell-culture grade phosphate buffered saline (PBS, Thermo Fisher Scientific, #10010049) for at least 5 min per wash. Matrices were maintained in PBS until the hMSCs were ready for seeding.
Based on literature reports for similarly 3D-printed matrices65 and manufacturer (Dimension Inx) recommendations, we targeted a seeding density of ~ 5000 cells mm–3 to be administered in a seeding volume of 10 μL. This seeding density corresponded to ~ 50,000 cells matrix–1. To seed the cells onto the matrices, the matrices had the PBS aspirated, yielding matrices that were moist but not too wet. Then, 5 μL of 25,000 cells 5 µL–1 were directly pipetted on top of the matrices. After 25 min, the matrices were carefully inverted, and another 5 µL of 25,000 cells 5 µL–1 were directly pipetted onto the current tops. After a total of 50 min from the initial seeding (25 min from the second seeding), the wells containing the matrices were flooded with 0.500 mL of growth media.
Since it was inevitable that some cells would not adhere to the matrix but instead pass through the matrix and adhere to the tissue culture plate, after 1 day the matrices were carefully transferred using a sterile spatula to a pristine well with fresh media. This prevented any interaction and confounding effects from cells initially seeded on the tissue culture plate. Also, at this point, osteogenic media was first given to matrices that were designated to be cultured in it. Cells were cultured on the matrices for 10 days, and media changes were provided no longer than every 3.5 days.
Cytocompatibility
After 10 days of the hMSCs being cultured on the matrices, the cell culture media was aspirated, the matrices gently washed with PBS, and then the cells were exposed to staining solution. Staining solution was added at 0.500 mL well–1 and consisted of PBS spiked with Hoechst 33342 at 20 µM (Thermo Fisher Scientific, #62249); Calcein AM at 5 µM (PromoKine, #PK–CA707-80011-2); and propidium iodide at 2 µM of propidium iodide (Alfa Aesar, #J66584). Cells were exposed to the staining solution for 15 min in the incubator. Then, the staining solution was removed, being replaced with fresh PBS.
Since the matrix is not only 3D but also highly absorbing due to the graphenic component, fluorescent microplate measurements would be inaccurate. Thus, to more accurately determine cytocompatibility, the matrices were directly imaged (Thermo Fisher Scientific, EVOS FL Auto Cell Imaging System, #AMAFD1000) with a 10×, 0.30 numerical aperture objective (Thermo Fisher Scientific, #AMEP 4681) and a 20×, 0.40 numerical aperture objective (Thermo Fisher Scientific, #AMEP4682). Hoechst 33342 labels the DNA of all cells (imaged with the DAPI light cube; Ex: 357/44 nm, Em: 447/60 nm; Thermo Fisher Scientific, #AMEP4650), and propidium iodide labels the DNA of dying cells whose membrane integrity are compromised (imaged with the RFP light cube; Ex: 531/40 nm, Em: 593/40 nm; Thermo Fisher Scientific, #AMEP4652). Imaging can be used to identify and quantify labeled nuclei without relying on quantification directly from fluorescence intensity. Thus, Hoechst 33342 and propidium iodide-labeled nuclei (> 740 per sample) were quantified from multiple, representative images of each matrix. Calcein AM is converted to a fluorescence form inside metabolically active cells (imaged with the GFP light cube; Ex: 470/22 nm, Em: 510/42 nm; Thermo Fisher Scientific, #AMEP4651). Average cellular viability (live cells as a percentage of all cells) were calculated for each matrix, and the individual matrices were then averaged together (n = 4).
To create images for display purposes, the as acquired fluorescence images were processed in Leica Application Suite Advanced Fluorescence Lite software 2.6.0 build 7266 (Leica Microsystems, https://www.leica-microsystems.com/). Individual images were overlaid and colormap ranges were adjusted to best visualize labeling.
Alkaline phosphatase (ALP) expression
ALP expression was determined using the ImmPACT Vector Red Alkaline Phosphatase kit (Vector Laboratories, Inc., #SK-5105) based on the manufacturer’s protocol. After 10 days of hMSCs being cultured on the 3D printed matrices, the medias were aspirated, and the matrices were washed with PBS. Next, the hMSCs on the matrices were fixed by exposure to 3.7% formaldehyde in PBS v/v for 10 min. Then, the formaldehyde solution was aspirated, and the cells were washed with PBS. To each well, 500 µL of staining solution was added. The staining solution consisted of ImmPACT Vector Red substrate working solution spiked with Hoechst 33342 at a final concentration of 20 µM. The hMSCs were exposed to the staining solution for 1 h. Then, the samples were washed with PBS for 5 min and were then maintained in excess (500 µL) PBS.
Since the 3D printed matrices contain a significant amount of CaPG, the matrices are highly absorbing due to the graphenic backbone of CaPG. Thus, while the ALP assay can usually be quantified using a fluorescence microplate reader, these graphenic matrices would not yield reliable data. Thus, instead, the matrices were subjected to fluorescence and light microscopy. Using an automatic microscope (EVOS FL Auto Cell Imaging System) with a 10× objective, whole-matrix images of Hoechst 33342 fluorescence, ImmPACT Vector Red-labeled ALP fluorescence, and transmission bright field images were acquired. To probe the cellular/sub–cellular distribution of ALP expression at higher resolution, higher-magnification imaging was performed using a 20× objective and a long–working distance 40×, 0.65 numerical aperture objective (Thermo Fisher Scientific, #AMEP4683). Overlays of the different channels were created using Leica Application Suite Advanced Fluorescence Lite software. As-acquired large images that extended beyond the matrices (to ensure that entire matrices were imaged) were cropped to the size of the matrices in ImageJ (US National Institutes of Health).
To quantify ALP expression, the whole-matrix images of fluorescently labeled ALP were background intensity thresholded in ImageJ. Then, the images were read into MATLAB (The MathWorks, Inc.), which was used to calculate the sum of all the pixel intensity values. Multiple matrices were averaged together (n = 3), and quantified results are reported as sample mean plus and minus sample standard error of the mean.
Alizarin red S (ARS) labeling
After 10 days of hMSCs being cultured on matrices, calcium deposits were labeled using ARS. A staining solution was prepared by diluting ARS (VWR, #97062-616) to 40 mM in DI water. To label the hMSCs, the cell culture media was aspirated; the matrices were washed with calcium-free PBS, the samples were fixed in 3.7% formaldehyde for 10 min; the formaldehyde solution was aspirated; the samples washed twice with PBS; and then the matrices were exposed to the ARS staining solution for 1.5 h at room temperature. After labeling, the samples were washed 7 times with PBS and then maintained in excess (~ 500 µL) PBS.
Whole matrix images were acquired using an automated microscope (EVOS FL Auto Cell Imaging System) with a 10× objective and a color camera. To quantify ARS labeling, the color whole-matrix images were read into MATLAB. The RGB format images were converted to CIELAB color space (L*a*b*). In L*a*b* color space, the a* channel represents the green/red opponent colors, where 0 is neutral, increasingly negative is increasingly green, and increasingly positive is increasingly red. Thus, the greater the value of a*, the more red. The mean value of a* was calculated for each matrix. Multiple matrices were averaged together (n = 3), and quantified results are reported as sample mean plus and minus sample standard error of the mean.
Gene expression
To quantify osteogenic gene expression, reverse transcription quantitative polymerase chain reaction (RT-qPCR) experiments were designed and performed. Details are described below in accordance with the Minimum Information for Publication of Quantitative Real-Time PCR Experiments (i.e., MIQE) guidelines87.
To do so, hMSCs were cultured on 3D printed matrices—2 matrices for each type of media—for 10 days. Then, RNA was immediately extracted using a commercially available RNA purification kit (RNeasy Mini Kit, Qiagen, #74104). During the process, the lysates from the same media type were pooled together. Once the RNA was purified, it was assessed with a microvolume UV–Vis spectrophotometer (NanoDrop One, Thermo Fisher Scientific, #ND-ONE-W). Spectra from 190 to 800 nm with a 0.5 nm step size were acquired from 2 µL of RNA solution. Analysis of spectra indicated a strong absorption of guanidine HCl that was a leftover reagent from the RNA extraction/purification kit (Supplementary Fig. 5a). In our other studies, attempts at RNA cleanup to remove the guanidine HCl were unsuccessful but yielded viable RNA for RT-qPCR experiments. Thus, spectroscopic analysis of RNA purity was not feasible. To obtain a rough estimate of the concentration of RNA, we used the generally accepted RNA extinction coefficient of 40 ng cm µL–1 and calculated concentration using the Beer-Lambert equation, as absorption from guanidine HCl was relatively low at 260 nm. RNA solution was stored at − 20 °C until use.
All PCR plastic consumables were certified DNase, RNase, and pyrogen free, and the pipette tips contained aerosol filters and were purchased sterilized via γ-irradiation. All PCR procedures were performed in a sterile environment. RT-qPCR experiments were setup on ice in 96–well PCR plates (VWR, #82006-644) and sealed (VWR, #60941-070) before beginning RT.
TaqMan hydrolysis probes (Thermo Fisher Scientific) were used to fluorescently report amplification in real time. All probes spanned exon junctions, negating the need for a DNA digestion step. All probes used were commercially available through Thermo Fisher Scientific: bone morphogenetic protein 2 (BMP-2, #Hs01055564_m1); collagen type I alpha 1 (COL1A1, #Hs00164004_m1), runt-related transcription factor 2 (RUNX–2, #Hs00231692_m1); small nuclear ribonucleoprotein D3 (SNRPD3, #Hs00188207_m1); and proteasome subunit beta 2 (PSMB2, Hs00267650_m1). Amplicon lengths were 84, 66, 116, 68, and 80 base pairs for BMP–2, COL1A1, RUNX–2, SNRPD3, and PSMB2, respectively. SNRPD3 and PSMB2 were selected as reference genes as they recently have been shown to be exceptionally uniformly expressed88.
To perform RT-qPCR on the RNA, a one-step assay (TaqMan RNA-to-CT 1-Step Kit, #4392938, ThermoFisher Scientific) was used. Master mixes were made for all components as appropriate, and the reaction volume was 10 µL. For a typical individual reaction, there was 5.00 µL of TaqMan RT-PCR Mix (2×), 0.50 µL of the TaqMan hydrolysis probe (20× mix diluted to final concentrations of the forward primer, reverse primer, and probe of 900, 900, and 250 nM, respectively), 0.25 µL of TaqMan RT enzyme mix (40×), 2 µL of RNA solution (~ 10 ng RNA), and 2.25 µL of DPEC-treated water (Thermo Fisher Scientific, #AM9906). All samples were run in quadruplicate. To minimize competition for PCR resources, there was no multiplexing.
Reverse transcription/cDNA synthesis and real-time PCR was performed on an Applied Biosystems 7300 instrument controlled by Sequence Detection Software Version 1.4 (https://www.thermofisher.com/order/catalog/product/4379633). The reverse transcription step occurred at 48 °C for 15 min and was followed by activation of AmpliTaq Gold DNA Polymerase Ultra Pure at 95 °C for 10 min. Then 60 cycles were performed of denaturing at 95 °C for 15 s and anneal/extend at 60 °C for 1 min. Sequence Detection Software was used to calculate and plot ΔRn and to automatically calculate a baseline and threshold to determine the threshold cycle, also more generally referred to as quantification cycle (Cq).
To determine PCR efficiency, serial tenfold dilutions of the RNAs were performed, and the samples were subjected to RT-qPCR for SNRPD3. The Cq values were identified, and for those more than 3 Cq units smaller than the no template control, they were plotted versus the logarithm of RNA concentration (Supplementary Fig. 5b). When plotted in this manner, the slope of the line is a measure of efficiency: (Efficiency={10}^{left(-1/sloperight)}-1). A PCR efficiency of 100% corresponds to a slope of − 3.32. To determine PCR efficiency of the experimental samples, a line was fit to the data, and the slope was used to calculate PCR efficiency. PCR efficiencies for the undiluted RNA samples were acceptable and had well-resolved Cq values. Thus, these were the amounts of RNA used for the RT-qPCR experiments.
The Cq values for the RNA samples for BMP-2, RUNX-2, and PSMB2 were not less than 3 away from their corresponding Cq values of the no template control. Thus, those samples were not valid. However, the COL1A1 and SNRPD3 yielded accurate data as the no template controls either had a much later amplification (i.e., larger Cq value) or had no detectable amplification at all.
To analyze and report gene expression, the ({2}^{-Delta Delta {C}_{T}}) method was used89. COL1A1 was the target; SNRPD3 was the reference; a PCR efficiency of 100% was assumed for all experiments, and the growth media sample is the calibrator. Four measurements were made (from the pooled RNA of two matrices) for each sample. The sample means and standard errors of the means were calculated, and then the uncertainties were propagated using the derivative method of error propagation for calculations on the data.
Mechanical properties
After 10 days of hMSCs being cultured on the matrices, the mechanical properties of the matrices were analyzed using axial compressive dynamic mechanical analysis (DMA) on a TA Instruments Discovery Hybrid 2 rheometer with a DMA attachment. Individual matrices were delicately removed from the cell culture media by gently scooping with a spatula and placing them on the bottom of a parallel plate geometry. The top plate was lowered until an axial force of ~ 0.1 N, corresponding to a prestress of ~ 10 kPa, was achieved. Then, the samples were subjected to axial compressive DMA using a strain of 0.3% and a frequency sweep from 0.1 to 11 Hz. A total of four different matrices for each media condition were analyzed. Averaged data is presented as both the entire frequency sweep, as well as bar plots of storage (E’) and loss (E”) moduli at 1 Hz for ease of comparison. Data is sample mean, and error bars are standard error of the mean.
Animal study
The objective of this study was to evaluate the bone regenerative capacity of 3DP-CaPG in a mouse non-union calvarial defect model. This study was a randomized, blinded, and controlled laboratory experiment. All aspects of the animal protocol were approved by UConn Health Institutional Animal Care and Use Committee (IACUC), animal welfare assurance number D16-00295, and all methods were performed in accordance with the relevant guidelines and regulations. The study was conducted in accordance with ARRIVE guidelines. All mice were constructed in the laboratory of Dr. David Rowe at UConn Health and were generated, bred, and maintained at the Center for Laboratory Animal Care of UConn Health. The animals had free access to sterile water and standard rodent chow ad libitum.
CD-1 transgenic mice containing the 3.6-kb fragment of the rat collagen type 1 promoter fused to a cyan fluorescent protein (Col3.6Cyan) were used as the donor mice to isolate bone marrow stromal cells. NOD.Cg-Prkdcscid Il2rgtm1Wj1/SzJ (NOD scid gamma, NSG) immunodeficient mice containing the 3.6-kb fragment of the rat collagen type 1 promoter fused to a topaz fluorescent protein (NSG/Col3.6Topaz) were used as host mice. The mice were divided into three groups: 3DP-G, 3DP-CaPG, and OsteoWrap demineralized cortical plate membrane (demineralized bone matrix, DBM). There were no inclusion or exclusion criteria beyond mortality resulting from infectious disease, as described below. During the allocation and conduction of the experiment, the authors were blinded. The authors only became aware of each group’s experimental conditions during analysis. The matrices for implantation were prepared using disposable biopsy punches of 3.5 mm diameter and sterilized as described for in vitro experiments.
Since these mice are immunocompromised, some occasionally become sick and die. These mortality events are due to their immunocompromised state and not the experimental conditions. Thus, if and when mice became sick and died during the course of the experiment, subsequent experiments were conducted to ensure that at least 4 valid replicates were obtained for each condition to ensure significant results. Ultimately, the final sample sizes were as follows: DBM (+) BMSCs: n = 6; DBM (−) BMSCs: n = 4; 3DP-G (+) BMSCs: n = 6; 3DP-G (−) BMSCs: n = 5; 3DP-CaPG (+) BMSCs: n = 5; and 3DP-CaPG (−) BMSCs: n = 5.
Bone marrow stromal cell isolation and culture
Col3.6Cyan (female, 8–10 weeks old) were used to derive bone marrow stromal cells. The mice were sacrificed by CO2 asphyxiation followed by cervical dislocation. The femurs and tibias were carefully isolated and dissected from the surrounding soft tissue. The two ends were cut and the bone marrow was collected by flushing complete media consisting of high glucose DMEM with l-glutamine, 10% FBS, 1% penicillin/streptomycin with a 25 gauge needle. When all the marrows were obtained, the suspension was passed through an 18.5 gauge needle. The cells were then counted and plated in a 100 mm dish at a density of approximately 6 × 107 cells per dish. The cells were kept in a Sanyo incubator under low oxygen conditions. At day 3 and 6, the media was replaced with fresh complete media.
At day 7 of BMSC culture, the cells were washed with PBS, detached with Accutase, and resuspended in complete media at a concentration of 1 × 106 cells mL–1. The cells were centrifuged at 1200 rpm for 5 min, after which the media was removed completely. The cells were resuspended in 10 μL of media, from which 5 μL was taken and seeded onto the matrices that were placed onto glass slides. After 1 h, the matrices were flipped, and the remainder of the cell solution was seeded onto the other side. The cells were allowed to attach for another hour, prior to implantation.
Surgical procedure
NSG/Col3.6Topaz (male, 11–13 week old) were used for calvarial surgeries. The mice were anesthetized with an intraperitoneal injection of Ketamine (135 mg kg–1)/Xylazine (15 mg kg–1). The head was shaved and the surgical site was cleaned with 75% ethanol. An incision was made just off the sagittal midline to expose the parietal bone. A 3.5 mm defect was created on one side of non-suture associated parietal bone using a trephine drill. The calvarial disk was carefully removed to avoid any injury to the underlying dura mater and the matrices were implanted into the defect. The skin was sutured with 5-0 vicryl, and the mice were subcutaneously injected with 0.08 mg kg–1 buprenorphine for analgesia. An additional dose of buprenorphine was given within 24 h of surgery.
Gross morphology and radiology
One day prior to sacrifice, Alizarin Red (AC) bone label at a dose of 30 mg kg–1 body weight was injected intraperitoneally to mark areas of newly deposited mineral within 24 h of sacrifice.
Animals were sacrificed at 8 weeks post-implantation by CO2 asphyxiation followed by cervical dislocation. The intact calvaria were carefully dissected from the skull and surrounding tissue and fixed in 10% formalin at 4 °C. After 2–3 days, the calvaria were imaged digitally and next radiographically (6 s at 26 kVp) using a digital capture X-ray cabinet (Faxitron LX-60).
Histological analysis
After X-ray imaging, the calvaria were placed in a 30% sucrose solution in PBS (pH 7.4) for 1 day. The tissues were then positioned in Richard-Allan Scientific™ Neg-50™ frozen section medium (Thermo Scientific). Cryosections (5 μm) through the non-decalcified calvaria were obtained on a Leica CM3050-S cryostat (Leica, Wetzlar) using a disposable steel blade (Thermo Scientific) and tape transfer process (cryofilm type IIC (10), Section-Lab Co. LTD) as previously described. The slides were then prepared with 50% glycerin in PBS as the mounting medium. The sections were initially imaged for differential interference contrast (DIC) using the Zeiss Axio Scan.Z1 (Carl Zeiss Microscopy). Next, the endogenous fluorescence of the Col3.6Topaz and Col3.6Cyan fluorescent reporters, and the AC mineralization label were imaged. The sections were then sequentially stained and imaged for TRAP enzymatic activity, ALP, DAPI, and toluidine blue O, as previously reported24,78,90. This sequence was possible because the cryofilm tape adheres to the tissue and allows for the coverslip to be removed between the imaging steps without damaging the section.
Statistics
Data is reported as mean ± standard error of the mean, unless otherwise indicated. Differences are labeled as statistically significant if the two-tailed p-value, calculated using an unpaired t-test (Prism, GraphPad Software, https://www.graphpad.com/), is less than 0.05. The p-values for growth media compared to osteogenic media for cellular viability is 1, for ALP expression is 0.22, for ARS labeling is 0.0070, for COL1A1 expression is 0.030, for E’ is 0.21, and for E” is 0.17. The p-values for E’ and E” for the pristine matrices compared to matrices with hMSCs cultured on them for 10 days are 0.75 and < 1 × 10–4, respectively.
