Introduction. Articular cartilage damage is a frequent occurrence that can lead to osteoarthritis, the most prevalent joint disease and leading cause of disability in the United States and other developed nations 1,2 . Tissue engineering, a prospective alternative treatment for this musculoskeletal disorder, aims to repair, maintain, or regenerate these damaged tissues; however, human tissues are complex in function, structural hierarchy, and scale, making it difficult to synthesize functional tissue in a lab that can be used for clinical treatments [3][4][5] .
Although there have been advances in tissue engineering, significant barriers remain regarding the ability to generate functional articular cartilage that imitates native cartilage both in structure and mechanical function 6,7 .
Over the last decade, articular cartilage tissue engineering has evolved from two-dimensional (2D) cell cultures grown on planar surfaces to culturing cells in complex three-dimensional (3D) architectures 5,8 . The goal with next-generation bioscaffolds is to closely mimic the native environment of articular cartilage as it is known that the material properties of bioscaffolds can drive specific cell behaviors such as proliferation, differentiation, and extracellular matrix (ECM) production leading to tissue formation [9][10][11] . In addition, bioscaffold properties can be used to deliver localized physical cues needed to stimulate tissue growth for engineering articular cartilage 12 .
Although biological materials such as collagen and alginate can simulate native extracellular matrix (ECM), they lack mechanical durability with a limited potential for improved functionality 13,14 . Novel engineered biomaterials introduce a better mechanical integrity, mimicry of complex structures, and a high degree of control over material properties 15 . New generation composite scaffolds such as polyvinyl alcohol-based hydrogels and polymeric/alginate composites exhibit more favorable mechanical properties however they do not address the limit of scalability and tunable control over cell behavior 16,17. Graphene and its derivatives have been established as superb scaffolds for cell culture with the dexterity to undergo long-term in vitro tissue engineering required for the growth of cartilage [16][17][18][19][20] .
Graphene foam (GF) is a porous 3D biocompatible substrate that is easily produced via chemical vapor deposition (CVD) on a nickel template 20,21 . Its unique material properties, such as high electron mobility, excellent thermal conductivity, and high mechanical strength, can be harnessed to drive cell behavior, while synthesis via the CVD process lends to its scalability [22][23][24] . Further, the microporous structure of GF facilitates nutrient exchange, and the high surface to volume ratio provides a favorable environment for long-term cell attachment and growth, making it an optimal candidate for next-generation 3D biomaterials 25,26 .
Although 3D environments are more conducive to functional tissue formation, characterization of cell proliferation and migration in these systems remains a challenge 27 . Unlike 2D cell cultures, analyzing a single cell plane is not sufficient when working with 3D systems as it is important to assess proliferation and migration of cells within the bioscaffold to determine the correlation between porosity, structure thickness and pore interconnectivity 28,29 . A high cell density and an even spatial distribution are associated with functional tissue formation; therefore, it is important to accurately evaluate cell attachment as well as cell distribution and density after seeding 11 .
Common characterization and analysis protocols such as transmission electron microscopy, scanning electron microscopy, and confocal fluorescence microscopy are tailored to analyzing cells in a 2D format, but these methods are not fully adapted to 3D analysis of the bioscaffolds internal structure and are limited by the bioscaffold's opacity 30 . Microcomputed tomography (MicroCT) techniques have been developed to study bioscaffold architecture without sample damage, however, direct imaging of the cells in a 3D environment is difficult due to their low contrast 31 . This limits understanding of the mobility of cells in optically opaque bioscaffolds like GF, and to our knowledge, a technique to assess cellular migration in the interior of GF bioscaffolds has not yet been realized. A previous study comparing the effect of different cell seeding methods within opaque poly(L-lactide-co-ε-caprolactone) based composite bioscaffolds utilized iron oxide nanoparticles to label cells in an attempt to investigate cellular infiltration, however the technique proved difficult to quantify through MicroCT as the iron particles were a similar density as the ceramic components of the scaffold. Further, the efficacy of the labeling technique varied between cell donor lines, and the intracellular particle uptake was nonexistent three days after labeling32.
Here we have demonstrated a method for labeling cells grown on GF bioscaffolds to quantify the effect of GF properties on cellular spatial distribution in a 3D environment. We also highlight the limitations of planar analysis in 3D environments, emphasizing the importance of characterizing cell behavior throughout the entire bioscaffold. The novelty of this research lies in the development of a labeling and correlative imaging techniques using a conjugated fluorophore to study cellular spatial distribution on GF bioscaffolds in conjunction with MicroCT techniques. The proposed labeling technique can be applied to other bioscaffold materials, offering a promising method for quantifying cell migration in various opaque 3D architectures.
An open-source CVD furnace was utilized to synthesize GF bioscaffolds using a nickel foam template (Figure 1A) 21 . Following synthesis, the graphene/nickel foam substrates are etched in 3 M hydrochloric acid until complete dissociation of the nickel. After etching, scanning electron microscopy (SEM) was used to characterize the superficial microstructure of GF (FEI Teneo Field Emission Scanning Electron Microscope). 1B show the macroporous structure and wrinkled topography of GF with increasing magnification as well as microcracks in the branches due to the dissociation and internal removal of the nickel foam template. GF was further analyzed with Raman spectroscopy to quantify the graphitic nature of our CVD GF (Figure 1C). Raman spectra were compared at three separate locations across a single sample, further confirming that the quality of GF is consistent throughout the bulk. Each of the separate Raman spectra exhibit the characteristic G (~1585 cm -1 ) and 2D (~2700 cm -1 ) peaks typical of graphitic materials, whereas the absence or low intensity of the characteristic D (~1350 cm -1 ) peak indicates the low defect density of the GF 32 . Additional analysis was done using X-ray photoelectron spectroscopy (XPS) which verified complete dissociation of the nickel foam template (Figure 1D).
diagram showing the process for the indirect labeling of ATDC5 cells is shown in Figure 2A. After 7 days of cell growth, fixation and permeabilization of the cells was performed. Samples were first stained with a primary polyclonal antibody (BS-0061R) targeting beta actin, followed by staining with a secondary antibody that is conjugated to 10 nm colloidal gold and Alexa Fluor™ 488 (AB_2536179). In Figure 2C and D samples were imaged using circular polarized lightdifferential interference contrast (C-DIC), a reflected light technique, which converts gradients in the specimen optical path into sample amplitude differences and allows for the visualization of the wrinkled topography of GF bioscaffolds 33 . Fluorescently labeled actin allowed for the quantification of anchorage dependent ATDC5 cell behavior on the superficial GF surface, where cells spanning GF pores can be visualized with both reflected light and fluorescence imaging.
Cell attachment was further analyzed through confocal immunofluorescent imaging, as demonstrated in Figure 3. The reconstruction of z-stacks using the maximum intensity projections (MIP) fusion method makes clear the constraints that arise from GF thickness when imaging using this standard optical technique. In Figure 3D and H, heatmaps with color-coded representations illustrate the attainable thickness range using a 10x objective while managing extraneous out of focus illumination from adjacent planes in the sample. Evaluation of GF and cell-GF interactions using scanning electron microscopy. Scanning electron micrographs allowed for the quantification of cell attachment and morphology on the surface of GF without any further processing due to the gold nanoparticle-antibody conjugates used to label actin. Traditionally, cells are sputter-coated with a layer of conductive material for this type of analysis 26,34 . Although the sputtering technique allows for the visualization of cells spanning GF pores and some surface interactions, it is limited in its ability to resolve singular cellular interactions with the rough GF surface. The indirect labeling technique with colloidal gold resulted in high resolution micrographs where several cell-graphene interactions can be visualized as seen in Figure 2. Cells wrap around GF branch structures (Figure 4A & B), exhibit both bipolar (Figure 4D & F) and multipolar (Figure 4E) morphologies indicative of fibroblastic cells.
bioscaffolds. MicroCT (Bruker, Skyscan 1172, Belgium) was performed on bare GF as well as GF with ATDC5 cells labeled with the antibody-gold nanoparticle conjugates to determine the internal structure of GF and spatial distribution of the cells within the 3D environment (Figure 5A). NRecon software was used to reconstruct the angular projections into cross sectional slices for 3D reconstruction and volumetric analysis using the same attenuation range for all samples.
The reconstructed images were then processed using Skyscan's CT Analyzer (CTan) to binarize the 2D images for 3D reconstruction and volumetric analysis (Figure 5A). The binarized 3D model of bare GF was calculated to have an average structure thickness (St.Th) of 8.625 m ±2.54 m, surface area to volume ratio of 368.55 mm -1 , and an object volume to total volume (Obj. V/TV) ratio 2.561% corresponding to a porosity of 97.439%. Several studies have shown that cellular behavior is influenced by fluid flow and nutrient diffusion in 3D culture environments, and the high porosity of our GF is advantageous such that it facilitates the exchange of waste products for fresh nutrients during longer culture periods 34 . Visual analysis was done using CTVox software, which takes the reconstructed 2D images from the scans and projects them in three dimensions, and CTVol software which takes the binarized 2D cross sectional slices and renders a complete 3D reconstruction. In both the projection and 3D reconstruction, the conjugated gold nanoparticles enabled the optical segmentation and false coloring of the cells (green) and the scaffold (grey). In Figure 5B, the CTVox projections enables visualization of how the cells are distributed throughout the bulk of the GF and reveals the cell density is higher at the edges of the scaffold. The 3D reconstruction in Figure 5C allows for the quantification of the spatial distribution of cells within the internal structure of the GF. Cells can be visualized more clearly by decreasing GF opacity, and although a side view with the GF set to 0% opacity indicates that cells are evenly dispersed throughout the bulk of the scaffold, a top view indicates that cell density is higher towards the outer edge, which is indicative of GF's hydrophobicity, and a characteristic in agreement with suspected observations of fluorescence micrographs and confirmed CTVox projections. In addition, our CVD synthesis method for GF bioscaffolds results in microcracks in the branch sidewall as seen in SEM (Figure 1B), leaving the internal branch structure open for cell migration, attachment, and proliferation, but difficult to confirm. Traditional methods of sputtering for electron microscopy would not allow for characterization of that behavior, whereas antibody staining with colloidal gold enabled visualization of cell migration within the branch structure using MicroCT techniques (Figure 6).
Discovery of this behavior indicates that with CVD GF bioscaffolds, surface area is not necessarily sacrificed for increased porosity.
Discussion. Indirect labeling with the conjugated fluorophore allowed us to utilize fluorescence microscopy and the conjugated gold enabled characterization of cell attachment and morphology on the superficial plane of our GF bioscaffolds using SEM. However, it is apparent that planar analysis alone is not sufficient in 3D environments. Common 3D scaffolds typically rely on seeding via the 'drop-on' method, with most cells remaining on the superficial seeded surface of the scaffold, resulting in poor cellular penetration and spatial distribution 36,37 . This study shows that the external surface alone does not accurately represent cell activity through the bulk. We have demonstrated a simple labeling technique that can be utilized for each of the aforementioned characterization modes, as well as MicroCT without any additional processing to study this behavior and its correlation to the volumetric and material properties of the GF. MicroCT scans and 3D rendering of porous structures are computationally expensive and time consuming.
However, we have demonstrated this staining protocol can be verified prior to MicroCT scans by using fluorescence microscopy or SEM. Additionally, the labeling can be verified visually in the MicroCT before scanning by using the real-time display utilizing the same X-ray settings for scanning bare GF (Figure S1A).
Through this work we have found that cellular distribution within GF bioscaffolds is not limited to the superficial seeding surface, despite its hydrophobicity, and that cellular distribution, while limited, takes place throughout the bulk of the scaffold without any external induction such as dynamic flow or rotational seeding. In addition, this is the first instance where cellular attachment within the internal structure of the graphene branch has been visualized. Confirmation of this activity indicates that GF bioscaffolds can be further engineered to utilize this feature to better suit certain cell types and tissue organization. Through the CVD process, the GF can be tailored to achieve the desired porosity and structure thickness of the bioscaffolds by varying the geometrical parameters of the nickel foam template. Furthermore, since cells can migrate to the internal branch structures of the GF, this finding demonstrates that cells may attach to either side of the GF branches which increases the effective surface area of the bioscaffolds creating a more robust tissue coverage. By determining how the structure of GF bioscaffolds affects cell behavior, bioscaffolds can be designed to facilitate cell organization that corresponds to articular cartilage tissue engineering.
The ability to quantify cell migration in opaque bioscaffolds is a major gap in 3D tissue engineering, as scaffolds with increased mechanical strength generally have increased opaqueness.
There is no one way to ensure a certain seeding procedure is optimal across different scaffolds materials as it is widely unique to the scaffold properties and architecture, however, the imaging protocol utilized is not limited to GF, and could be adapted to any bioscaffold with an optical density that can be segmented from gold.
behavior, making GF an ideal candidate for next-generation 3D biomaterials. However, characterizing cell proliferation and migration within such 3D systems remains a challenge. To address this gap, we developed a labeling technique using a conjugated fluorophore to study cellular spatial distribution on GF bioscaffolds using MicroCT techniques. This approach allowed us to visualize cells within the internal branch structures of GF, providing insights into cell migration, attachment, and proliferation within the 3D environment, the feasibility of which would be unattainable through conventional optical methods of characterization.
The results demonstrate that cells exhibit even spatial distribution throughout the bulk of the GF bioscaffold, despite using the "drop-on" method of cellular seeding. This finding opens up new possibilities for engineering GF bioscaffolds to utilize this feature for specific cell types and tissue organization. The CVD synthesis method for GF offers the potential to tailor GF physical properties in order to achieve high porosity and structure thickness for tissue engineering applications.
Additionally, this study highlights the limitations of relying solely on planar analysis in 3D
environments and emphasizes the importance of characterizing cell behavior throughout the entire bioscaffold. The presented labeling technique can be applied to other bioscaffold materials, offering a promising method for quantifying cell migration in various opaque 3D architectures.
Overall, our results offer a new tool to probe the fundamental role of GF's 3D structure on cell behavior. The combination of GF's unique material properties and the proposed labeling technique holds great potential for future advancements in tissue engineering and regenerative medicine applications. The findings presented in this manuscript pave the way for further studies in the field, aiming to refine tissue engineering strategies and ultimately improve the treatment options available for patients suffering from articular cartilage damage and osteoarthritis.
An open-source CVD furnace was used to synthesize GF bioscaffolds using a 1.2 mm thick nickel (Ni) foam template 21 . The Ni foam cut was cut (3 cm x 8 cm) and placed inside a 2-inch quartz tube. The Ni foam was annealed for 30 min at 1000C and graphene grown under CH4 flow at 1000C for 50 minutes before undergoing a cooling cycle to room temperature.
The Ni/graphene foam composite was coated with polymethyl methacrylate (PMMA) and dried for 24 hours to maintain the structural integrity of the foam during post-processing. Coated
Ni/graphene foam substrates were etched in 3M HCl on a 60C hotplate until nickel foam template was completely dissociated at which point the PMMA was dissolved with acetone. The resulting GF was rinsed with Millipore water, dried, and cut into circles with an 8 mm diameter before characterization.
Bare graphene foam characterization. The superficial microstructure and surface topography of the GF bioscaffolds were evaluated via scanning electron microscopy (FEI Teneo Field Emission Scanning Electron Microscope). SEM samples were attached to the SEM post with double sided carbon tape, and electron micrographs were collected at 5.00 kV and 25 pA utilizing an Everhart-Thornley detector (ETD). Micrographs from SEM were used to manually measure average pore size with ImageJ software for comparison with MicroCT volumetric analysis (Figure S2). Raman spectroscopy on the (Horiba Scientific LabRAM HR Evolution Raman Microscope) and x-ray photoelectron spectroscopy (XPS) were performed to determine the graphitic nature of the GF and confirm complete dissociation of the nickel foam template. X-Ray Photoelectron Spectroscopy (XPS) measurements were performed with a Physical Electronics ESCA 5600 in the Atomic Films Laboratory at Boise State University using an Al kα excitation source. Low-resolution survey scans of the surface were performed initially to measure relative atomic concentrations. The sample was found to contain 91.04% Carbon (C1s) and 8.96% Oxygen (O1s). Data were analyzed with MultiPak 9.6 software, and all spectra were referenced to the C 1s peak (284.8 eV) for adventitious carbon. The survey region ranged from 0-1400 eV with a step size of (0.400 eV).
Elemental peaks in the survey spectra were initially identified using the software's automatic peak identification feature and verified using the Handbook of X-ray Photoelectron Spectroscopy 38 .
Peak fitting on the high-resolution spectra utilized a Gaussian-Lorentzian fit with a Smart Microcomputed tomography of GF. Bulk structural characterization was evaluated via microcomputed tomography (SkyScan 1172 Xray MicroCT). Briefly, GF samples were mounted onto a porous polyethylene pipette filter with double sided tape where a drop of 70% ethanol was placed atop the GF to ensure mounting to the tape without needing to add pressure to the top of sample. After drying, the GF/filter was placed upright on the sample holder and secured into place with double sided tape to eliminate scan artifacts due to random movement 32 . Scan acquisition on bare and labeled GF bioscaffolds was conducted with a 26 kV source voltage, 145 A current, and a 2650 ms exposure time. Scan parameters were defined with a step size of 0.25 degrees, ten-frame averaging, and 2.24 m pixel size. NRecon software was used to reconstruct the angular projections into cross sectional slices for 3D reconstruction and volumetric analysis with an attenuation value for all samples of 0 to 0.5000. Bruker Skyscan CT Analyzer (CTan) software was used to binarize the 2D images; bare GF scans were binarized with a threshold value of 18-115 range on the contrast scale, while labeled samples were segmented into two contrast scales:
(1) the gold nanoparticle labeled cells (115-255), and (2) for the GF (18-115), for 3D reconstruction and volumetric analysis. GF structure thickness, surface area to volume ratio, object to total volume ratio and the corresponding porosity were calculated using CTan software. 3D reconstruction of the GF environment was qualitatively analyzed using CTVol software, and shadow projections were visualized using CTVox software. To highlight the cells versus the GF, cells were false colored to green with the transfer function editor using the linear interpolation method in CTVox 39 (Figure S1B) and by overlaying the falsely colored 3D models in CTVol.
Preparing GF for cell culture. The GF bioscaffolds used for cell culture were cut from the same sheet of Ni foam and synthesized in the same batch to ensure consistency across the substrates. To prepare our GF bioscaffolds for ATDC5 (Sigma Aldrich, St. Louis, MO, U.S.A.) cell culture, they were sterilized with 70% ethanol prior, rinsed with DPBS to conditioning them in growth media (F12/Dulbecco's Modified Eagle Medium (DMEM:F12), 5% (v/v) fetal bovine serum (FBS), and 1% (v/v) penicillin/streptomycin) for 24 hours before seeding them with cells. Additionally, an anti-adherence rinsing solution (STEMCELL technologies) was used in the well plates to prevent cell growth on the cultureware containing the GF and to promote cell growth on the scaffold.
Cell culture. Conditioned GF bioscaffolds were seeded with ATDC5 chondrocyte progenitor cells by pipetting 500 L of cell suspension (5x10 6 cells) to the topside of the GF. They were then incubated for 7-days in growth media (GM) at 37C and 5% CO2. Cell growth was monitored with transmitted light microscopy and GM changed daily.
Cells were fixed on GF bioscaffolds with 0.2% paraformaldehyde, permeabilized with 0.1% Triton-X, and directly labeled with beta-Actin polyclonal antibody (Thermo Fisher Scientific) at a concentration of 1g/mL before incubation at 37°C for 30 minutes. Bioscaffolds were rinsed 10 times with PBS diluted in nanopure water (10:1), then stained with a 30 g/mL concentration of Goat anti-Rabbit IgG (H+L) Secondary Antibody Alexa Fluor TM 488-10 nm colloidal gold, incubated at 20ºC in the dark for 30 minutes, and then rinsed 10 times in diluted PBS and dried.
The diluted PBS rinsing steps ensure that salt crystals that form from drying PBS do not affect SEM or MicroCT acquisition (Figure S3). Samples were imaged with a Zeiss Axio Imager.M2 upright microscope fitted with a Zeiss Colbri 5 LED light source, fluorescence filter cube, C-DIC slider, and an Axiocam 305 color digital camera (Carl Zeiss, Inc). Z-stack images using reflected light, fluorescence, and C-DIC were acquired the EC Epiplan 50x/0.7, 20x/0.4, 10x/0.25 HD M27 objectives. Samples were then imaged with the Zeiss LSM 900 confocal system combined with a Zeiss Axio Observer.Z1. Confocal Z-stack micrographs were acquired using the Plan-Apochromat 10x/0.45 objective with laser wavelengths of 405 nm and 488 nm at a laser power of 1.2% and 1% respectively. Image processing was performed with ZEN imaging software with the exception of the maximum intensity projections for the Z-stacks, which were acquired with FIJI software. GFcell interactions with the GF surface were analyzed with SEM and MicroCT with the same mounting and scanning methods used for the bare GF samples.
MicroCT video of ATDC5 cell distribution in 3D space allows for full quantification of how cells are spatially arranged when the scaffold is set at 0% opacity, confirming that cells migrate through the bulk of the scaffold despite using the "drop-on" method of seeding on the superficial surface.