Bacterial nanocellulose-reinforced gelatin methacryloyl hydrogel enhances biomechanical property and glycosaminoglycan content of 3D-bioprinted cartilage

Tissue-engineered ear cartilage scaffold based on three-dimensional (3D) bioprinting technology presents a new strategy for ear reconstruction in individuals with microtia. Natural hydrogel is a promising material due to its excellent biocompatibility and low immunogenicity. However, insufficient mechanical property required for cartilage is one of the major issues pending to be solved. In this study, the gelatin methacryloyl (GelMA) hydrogel reinforced with bacterial nanocellulose (BNC) was developed to enhance the biomechanical properties and printability of the hydrogel. The results revealed that the addition of 0.375% BNC significantly increased the mechanical properties of the hydrogel and promoted cell migration in the BNC-reinforced hydrogel. Constructs bioprinted with chondrocyte-laden BNC/GelMA hydrogel bio-ink formed mature cartilage in nude mice with higher Young’s modulus and glycosaminoglycan content. Finally, an auricle equivalent with a precise shape, high mechanics, and abundant cartilage-specific matrix was developed in vivo. In this study, we developed a potentially useful hydrogel for the manufacture of auricular cartilage grafts for microtia patients.


Introduction
In clinical practice, microtia is one of the most common congenital malformations. At present, autologous costal cartilage transplantation has become the most common treatment for microtia [1][2][3][4][5] . However, the complications caused by this operation, such as pneumothorax, post-operative pain, and chest wall deformity, are hardly inevitable. Moreover, the manually carved stent relies more on the technique of the surgeon, and there is no accurate shape for a stent [6,7] .
International Journal of Bioprinting applied tissue-engineered auricular cartilage based on polyglycolic acid/polylactic acid and chondrocytes to clinical practice [9] . However, it was found that the implants collapsed at varying degrees after some time. Polymer materials can easily cause aseptic inflammation [10] , and it is challenging to achieve uniform distribution of chondrocytes in the scaffolds using the traditional seeding cell method [11] ; these justify the unsatisfactory results. Due to their low aseptic inflammation, natural hydrogel materials such as alginate, gelatin, and hyaluronic acid have attracted attention recently [12][13][14][15][16][17] . The three-dimensional (3D) bioprinting technology has significant advantages in the construction of accurate anatomical contours because it can precisely define the spatial distribution of cells and materials [18,19] . Therefore, 3D bioprinting based on natural hydrogel materials is a promising approach to overcoming the bottleneck of cartilage tissue construction [15,20,21] . However, the insufficient mechanical stability of hydrogel materials makes it difficult to ensure the shape of the scaffold in vivo.
Gelatin methacryloyl (GelMA) is one of the commonly used hydrogels in 3D bioprinting, and it has been effectively applied to cartilage tissue engineering [22][23][24] . However, due to low mechanical properties, its printing fidelity is limited, and it is difficult to produce large-scale functional tissue constructs only using GelMA as the scaffold [25,26] .
To improve the mechanical properties of the scaffold with natural materials, we propose adding bacterial nanocellulose (BNC) to GelMA to enhance the mechanical stability and printability of hydrogels for cartilage bioprinting. BNC is a type of naturally occurring highmolecular-weight polymer with good biocompatibility, high Young's modulus, excellent water retention, and good flexibility [27,28] . In addition, it has good printability and shear thinning characteristics, which make it suitable for 3D bioprinting [29] . In this study, we explored the appropriate proportion of BNC in the composite hydrogel and the feasibility of using this bio-ink to regenerate auricular cartilage with high mechanical performance in vivo.

Animals
Japanese white rabbits (female, 2 months old) and nude mice (male and female, 6 weeks old) were bought from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). All animal experiments were approved by the Institutional Animal Care and Use Committee of the Plastic Surgery Hospital (Approval ID: 2017 -37).

Preparation of hydrogel materials
BNC was produced using BC as previously described [30] . In brief, BC was purified with 2% (w/v) NaOH at 80°C for 1 h and then washed repeatedly with distilled water until a neutral pH was obtained, was repeatedly hydrolyzed and centrifuged under acidic conditions to obtain BNC suspension, which was then dialyzed and freeze-dried to produce BNC powder. GelMA lyophilized powder and LAP were dissolved in Dulbecco phosphate-buffered saline (DPBS; Gibco, Grand Island, NY, USA) to form 10% w/v GelMA and 0.25% w/v LAP-based solutions, which were used in the control group. A certain proportion of irradiated BNC was added to the GelMA solution to form BNC/GelMA hydrogels with different concentrations (Table 1).

Compression test on hydrogels
Silicone elastomer bases (Slygard TM 184, Dowsil, USA) were heated at 70°C for 3 h to make a 10 × 4 mm disk mold. A 300 μl hydrogels were added into the mold and irradiated with 405 nm, 30 mW/cm 2 ultraviolet (UV) light for 10 s to make it completely cross-linked to form a cylindrical-shaped scaffold (10 × 4 mm) for compression test (n = 7). A biomechanical analyzer (Instron 5967, USA) with a 100 N pressure sensor was used for the compression tests. A constant compressive strain rate was maintained at 1 mm/min until 80% of maximal deformation had been reached. The hydrogels' Young's modulus was measured with the slope of the stress-strain curve set between 10% and 20% strain. A humidifier was used to maintain the surrounding air humid throughout the tests.

Rheological analysis
The rheological tests were performed with an Anton-Paar MCR 302 rheometer (Anton-Paar GmbH, Austria) using a 25 mm diameter parallel plate (PP25, d = 1.0 mm). Temperature sweep test was conducted to evaluate the hydrogels' behavior at various temperatures by setting a temperature increase at a rate of 2°C/min in the range of 0 -40°C, and the values of G' (storage modulus) and G" (loss modulus) were recorded for each temperature. Viscosity was measured as a function of shear rate (0.1 -100 s -1 ) at 21°C. All measurements were performed at 1 Hz and 1% strain [31] .

Scanning electron microscopy (SEM) examination
Following lyophilization and gold sputter coating, samples were analyzed using a Quanta 2000 scanning electron microscope (FEI Co., The Netherlands) at 15 kV. ImageJ software (ImageJ Software Inc., USA) was used to analyze photomicrographs.

Printability test
3D printing was performed with the 3D-Bioplotter printer (EnvisionTec, Germany). The hydrogel was placed into the bioprinter barrel and incubated for 10 min at 37°C before extrusion. The extrusion test was carried out at 21°C using various nozzles with inner diameters ranging from 150 μm to 600 μm.
The nozzle with a 400 μm inner diameter was used to print various models to test the printing resolution and precision of the composite hydrogel. The stereolithography (STL) files of the cube and the acronym PSH characters for plastic surgery hospital were built by AutoCAD software (Autodesk, San Rafael, CA). The human nose construction of the STL file was downloaded from an open-source website at https://www.thingiverse.com/, and the mandibular model bracket was provided by Envision-Tec. Subsequently, the STL files were imported into the slicing software Perfactory RP (EnvisionTec, Germany). The layer height was set as 320 μm and sliced in the model. The sliced model was imported into the 3D bioprinting system visual machine (EnvisionTec, Germany). In the printing system, the internal structure of the scaffolds was set as a cross grid, and the line spacing was set as 800 μm. The applied extrusion pressure for the composite hydrogel was 0.4 -0.8 bar, and the nozzle speed was 3.5 -4.5 mm/s ( Table 2). The composite hydrogel was placed in the barrel of the 3D bio-printer and incubated at 21°C for 15 min before printing. Each stack of two layers was completely cross-linked by a 405 nm, 30 mW/cm 2 UV source for 10 s. During printing, images of each layer were collected using the inbuilt camera.

Isolation and cultivation of auricular chondrocytes
As previously described [32] , the ear cartilage of Japanese white rabbits was extracted and minced into 1 mm 3 pieces under sterile conditions. The cartilage fragments were digested with 0.2% type II collagenase solution for 8 h at 37°C, then filtered through a 100 μm filter screen. The chondrocytes were collected, cultivated, and expanded in culture medium containing high-glucose DMEM, 10% FBS, and 1% PSN with 5% carbon dioxide (CO 2 ) and 95% humidity at 37°C. Cells at second passage were collected and used for the subsequent experiments.

3D bioprinting of cell-laden constructs
For building cell-laden constructs, 0.375% BNC, 10% GelMA, and 0.25% LAP were dissolved in culture medium as described above. The chondrocytes were collected and mixed into the hydrogels to make bio-ink with a concentration of 1 × 10 7 cells/mL. During printing, the extrusion pressure and printing speed were adjusted according to the material drawing state. After printing, the scaffolds were immersed in the culture medium and placed in the CO 2 incubator for culture.

Cell viability and migration assays
After being cultured in vitro for 1, 4, and 7 days, the Calcein-AM/PI Double Staining Kit (DOJINDO, Japan) was used to evaluate the viability of cells in hydrogels. The results were examined using Leica TCS SP8 CARS confocal microscope. ImageJ software was used to measure the cell viability in three randomly chosen visual fields.
To evaluate cell migration, BNC/GelMA and GelMA hydrogels were cross-linked on one side of 15 mm confocal Petri dish loaded with chondrocytes (1 × 10 6 cells/mL). Cell-free hydrogels were used to cover the opposite side of the Petri dish [33] . Following in vitro culture for 7 days, the Calcein-AM was used to stain the living cells, and a confocal microscope was used to observe cell migration. International Journal of Bioprinting

Cartilage regeneration in vivo
Cell-laden constructs were cultivated in vitro for 1 day and then implanted into nude mice subcutaneously to observe cartilage formation in vivo. The samples were taken out at 4,8,12, and 24 weeks after implantation for biomechanical test, glycosaminoglycan (GAG) content test, total collagen content test, and histological examination.

Biomechanical and biochemical analysis of the regenerated cartilage
The compression tests of regenerated cartilage (n = 3) were performed by Instron 5967 with a 100 N pressure sensor as described above, and Young's modulus of the samples was measured based on the slope of the stress-strain curve set between 10% and 20% strain.
The dimethylmethylene blue GAG test kit (GenMed Scientifics Inc., Shanghai, China) was used to detect the GAG content in the samples (n = 3). Being a unique amino acid in collagen, hydroxyproline accounts for about 13.4% of the total content of collagen. The hydroxyproline detection kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) was used to quantify the content of hydroxyproline in samples (n = 3). All operations were carried out according to previously established protocols [34] .

Histological examination
Regenerated cartilages were collected and subjected to histological examinations. The samples were fixed in 4% paraformaldehyde, dehydrated, and embedded in paraffin. Hematoxylin and eosin (H&E), Alcian blue (Solarbio, China), and Safranine-O (Solarbio, China) staining were performed, as previously described, to evaluate the cartilage extracellular matrix deposition and histological structure in the regenerated cartilage [32] .

Statistical analyses
GraphPad Prism 8.0 software was utilized for statistical analyses. All experimental data were obtained from at least three repeated experiments, and data are reported as means ± standard deviation. Statistical significance was evaluated by Student's t-test or one-way analysis of variance. Values of P < 0.05 were considered statistically significant. Data distribution should meet the normal distribution requirements.

Selection of BNC proportion in the composite hydrogels
This study showed that BNC existed in suspension in 10% GelMA solutions after heating and mixing in the water bath because BNC is insoluble in water, and it would form a colloid or suspension in hydrogel [35] . With the increase of BNC ratio, the composite hydrogel became more turbid ( Figure 1A). The stress-strain curves of each composite hydrogel sample showed that with the increase of BNC content, the slope of the curve gradually increased ( Figure 1B). By calculating the slope of 0.1 -0.2 mm/mm strain, Young's moduli of 10% GelMA, 0.075% BNC, 0.225% BNC, 0.375% BNC, and 0.525% BNC were 28.19 ± 3.058 kpa, 28.72 ± 2.066 kpa, 35.34 ± 1.430 kpa, 49.94 ± 2.775 kpa, and 53.74 ± 3.844 kpa, respectively ( Figure 1C). From the 0.225% BNC group, Young's modulus of the hydrogels was significantly higher than that of 10% GelMA (P < 0.001) and increased with the increment of BNC concentration. However, there was no significant difference in Young's modulus between 0.375% BNC and 0.525% BNC (P > 0.05). In our opinion, when the BNC content is too high, its distribution in the solution becomes uneven, which may affect the full crosslinking of GelMA. Fourati et al. found that when nanocellulose was used as the additive phase, the mechanical properties of the composites would initially increase, but decreased after exceeding a certain amount [36] . In another study, Fan et al. also found that when cellulose nanocrystals (CNCs) were added to 15%, there was no significant difference between 10% CNC and 15% CNC when CNCs were used to strengthen GelMA/HAMA hydrogels [37] . Moreover, we found that the hydrogel with 0.525% BNC was difficult to be extruded and the nozzle could be blocked easily. Therefore, in the subsequent experiment, 0.375% BNC was selected as the experimental group (BNC/GelMA group), and 10% GelMA without BNC was set as the control group (GelMA group).

Rheological test
We detected the gelatinization with temperature change and shear thinning behavior of the BNC/GelMA hydrogel by rheological analysis. We observed that at 0 -40°C, G' (storage modulus) and G" (loss modulus) of both hydrogels decreased with the increase in temperature. At 0 -23.5°C, when the storage modulus (G') dominated (G' > G"), both materials had gel-like properties. Conversely, at 23.5 -40°C, when the loss modulus (G") dominated (G' < G"), the materials had liquid-like properties. Thus, 23.5°C was the common gel-forming point of these two materials ( Figure 2A and B). The gelling temperature of the hydrogel did not change after BNC was added. In addition, it could be seen that both the BNC/GelMA group and the GelMA group have obvious shear thinning behavior at 21°C ( Figure 2C). These results suggested that the BNC/GelMA hydrogel was suitable for 3D printing.

SEM examination
Both hydrogels in BNC/GelMA group and GelMA group had loose pore structures ( Figure 2D). Through further analysis of the pore size, it was found that the average pore size of the hydrogels in BNC/GelMA group and GelMA group was 172.8 ± 54.19 μm and 126.0 ± 35.21 μm, respectively ( Figure 2E). The scaffold's pore size of the BNC group was significantly larger than that of the GelMA group (P < 0.0001). The loose pore structure may be conducive to the proliferation and migration of cells in the scaffold [38] .

Printability of composite hydrogel
We tested the printability of the hydrogel with nozzles of different inner diameters ranging from 150 μm to 600 μm ( Figure 3). The nozzles with inner diameters of 150 μm-210 μm could not be extruded completely, and the nozzles with inner diameters of 250 μm-300 μm were occasionally blocked. When a 400 μm nozzle was used, the hydrogel could be smoothly extruded under pressures above 0.4 bar. Taking the cell viability and printing accuracy into consideration, a 400 μm nozzle and 0.4 -0.8 bar pressure were used as printing parameters for subsequent experiments.
Different models were used to test the printing formability of the composite hydrogel. The presence of BNC led to the extrusion of a stable filament, which, therefore, contributes to a homogenous diameter distribution alongside the length ( Figure 4A). When stacked up to 18 layers, the grid structure formed by the hydrogel could still be seen ( Figure 4B). Then, we printed 0.3 times the size of the human jaw model (3.1 × 2.2 × 0.896 cm) ( Figure 4C) and 0.4 times the size of the human nose model (2.4 × 1.6 × 0.576 cm) ( Figure 4D). The addition of BNC helped obtain uniform lines, complete printing structure, clear surface outline, and high fidelity; these characteristics are essential for the construction of scaffolds with precise morphology.

Cell viability
The Calcein AM/PI double staining kit was used to stain the live and dead cells on the 1 st , 4 th , and 7 th days after printing to evaluate the cell viability. Most of the cells in the scaffolds were dyed green fluorescence, and only a few cells showed red fluorescence, suggesting high cell viability in the hydrogels ( Figure 5A). ImageJ software was used for quantitative analysis of the number of live and dead cells. The percentages of cell viability of the BNC/GelMA group on the 1 st , 4 th , and 7 th days were 96.81 ± 1.541%, 96.12 ± 0.6627%, and 97.34 ± 1.450%, respectively ( Figure 5B). There was no significant difference in terms of cell viability between the BNC/GelMA group and the GelMA group (P > 0.05).

Cell migration
To explore whether larger internal pores of the BNC/GelMA hydrogel affected the migration of cells in the scaffold, we   Figure 5C). Cells in the scaffold of the BNC/GelMA group had obviously migrated to the acellular side on the 7 th day, while there was no cell migrating in the GelMA group ( Figure 5D). These results suggested that the cells in the BNC/GelMA scaffold had a better migration ability than the cells in the GelMA group. The scaffold of appropriate pore size may justify a higher number of migrating cells in the BNC-containing scaffold [39] . In addition, the previous studies showed that the fiber structure of BNC could promote movement of cells along fiber surfaces [40] .

Mechanical properties of the regenerated cartilage
To explore the cartilage regeneration ability in vivo of the composite hydrogel, we implanted the scaffolds into nude mice and took the samples out at the 4 th , 8 th , 12 th , and 24 th weeks after implantation. It could be seen from the gross view that white cartilage-like tissue began to form in both BNC/GelMA and GelMA scaffolds from the 8 th week after implantation ( Figure 6A). At the  week, Young's modulus of the BNC/GelMA group was significantly higher than that of the GelMA group (P < 0.01). By the 24 th week, Young's modulus of the BNC/GelMA group was more than 2 times higher than that of the GelMA group ( Figure 6B), which is close to the elastic modulus of the human ear helix (1.41 ± 0.67 MPa) [41] .

Quantitative and histological performance
To analyze the extracellular matrix secretion of the regenerated cartilage, we tested the GAG content ( Figure 6C) and total collagen content ( Figure 6D) of the samples. At the 4 th , 8 th , 12 th , and 24 th week, the GAG contents in BNC/GelMA group were 5.349 ± 0.2706 mg/g, 6.928 ± 0.3651 mg/g, 11.90 ± 0.4840 mg/g, and 31.99 ± 0.6753 mg/g, respectively, whereas the GAG contents in GelMA group were 5.268 ± 0.2009 mg/g, 6.564 ± 0.3038 mg/g, 10.04 ± 0.8624 mg/g, and 24.10 ± 0.7412 mg/g, respectively. The GAG content of the regenerated cartilage in the two groups gradually increased from the 4 th week. By the 12 th and 24 th weeks, the GAG content in the regenerated cartilage of the BNC/GelMA group was significantly higher than that of the GelMA group (P < 0.001), which was closer to the natural rabbit ear cartilage (44.31 ± 0.4858 mg/g). We measured the content of hydroxyproline to estimate the total collagen content. The results showed that the total collagen content also gradually increased over time and attained the expression level of natural rabbit ear cartilage collagen at the 24 th week after implantation. However, there was no significant difference between the two groups at all the tested time points (P > 0.05).
The mechanical results were in line with the GAG content results. A previous study has shown that the compressive capacity in cartilage tissue is mainly attributed to the GAG [42] , which was consistent with our results.
We further performed H&E, Alcian blue, and Safranine-O staining for histological evaluation. It could be seen from the H&E staining that from the 4 th week, cartilage lacunae had formed in both BNC/GelMA and GelMA scaffolds, and with the extension of time, the staining of cartilage extracellular matrix in the scaffolds became deeper in color, as shown in Alcian blue and Safranine-O staining results ( Figure 7). It could be seen in the gross specimens and histological staining results that the hydrogels were still present in the samples at 24 weeks. A previous study on the degradation of GelMA showed that GelMA could remain in collagenase solution for 10 days [43] , but no relevant reports on the degradation of GelMA and BNC in vivo were found. However, the hydrogel degrades more slowly in nude mice due to International Journal of Bioprinting the lack of immunity. Therefore, experiments using hydrogels with fluorescent labeling and large animals with immunity are needed to evaluate the degradation of hydrogels in vivo.

3D bioprinting and regeneration of ear-shaped cartilage constructs
To detect the feasibility of morphological maintenance of the composite hydrogel, we constructed a scaffold with 0.4 times the size of a normal human ear by 3D bioprinting using the composite hydrogel and rabbit auricular chondrocytes (Video clip 1). The cartilage scaffolds were superimposed layer by layer under the 3D printer, and finally, 22 layers were printed ( Figure 8A). The printed human ear-shaped scaffold was basically consistent with the 3D model, and there was no material collapse during the printing process ( Figure 8B). The live/dead staining results demonstrated that the cells embedded in hydrogels had good viability ( Figure 8C). Subsequently, we implanted the scaffold subcutaneously in nude mice ( Figure 8D) and collected it at 24 weeks after implantation. The regenerated cartilage, which was well maintained and in condition similar to before implantation ( Figure 8E), showed a milky white cartilage appearance ( Figure 8F) and had good elasticity (Supplement 2). Histological examination revealed the formation of a considerable amount of cartilage-specific extracellular matrix indicated by Safranine-O and Alcian blue staining and the formation of typical lacunae structure indicated by H&E staining ( Figure 8G).
In this study, BNC/GelMA composite hydrogel was used to regenerate ear cartilage. We confirmed that a small amount of BNC could greatly improve the mechanical strength of GelMA hydrogel materials. In addition, the cell viability in the hydrogels was still above 95% on day 7, which was higher than in a previous study, in which Markstedt et al. constructed ear cartilage scaffolds with alginate/nano-cellulose hydrogel through 3D bioprinting, and the cell viability on the 7 th day was only 85.7% [44] . In another study, Martínez et al. proposed the use of a nanocellulose hydrogel for 3D bioprinting https://doi.org/10.18063/ijb.v9i1.631 BNC-reinforced GelMa enhances property of bioprinted cartilage International Journal of Bioprinting to regenerate auricular cartilage, and the cell viability after 3D bioprinting was only 70.9 ± 7.2% [45] . In addition, it should be noted that the BNC-containing hydrogel promoted cell migration, which is an important cellular feature for tissue morphogenesis. The most important criterion for testing whether a biomaterial is suitable for tissue engineering is to evaluate tissue regeneration in vivo [11] . We implanted the chondrocyte-laden 3D-bioprinted scaffolds into nude mice and cultured them for 24 weeks in vivo. The results demonstrated that the composite hydrogel was beneficial to the GAG secretion in regenerated cartilage, and its biomechanical properties were significantly enhanced, which was close to the Young's modulus of human ear cartilage. Moreover, a precise ear-shaped construct was bioprinted, and the cartilage with 3D morphology was regenerated successfully in vivo, which further verified the feasibility of applying BNC-reinforced GelMA hydrogels in auricular cartilage tissue engineering. Certainly, future studies in large animal models are required to verify its potential in preclinical applications.

Conclusions
A BNC/GelMA composite hydrogel was prepared in this study. Compared with 10% GelMA, the 0.375% BNC composite hydrogel has superior mechanical properties as well as better printability and cell migration ability. Auricular cartilage was regenerated in nude mice using chondrocyte-laden BNC/GelMA hydrogel. The cartilage tissue regenerated by the hydrogel had higher GAG content and better biomechanical properties. Finally, the ear-shaped construct was bioprinted with composite hydrogels, and the cartilage tissue was successfully regenerated in vivo. Although the superiority of the composite hydrogel needs to be further verified in large animal experiments, this study offers insights into using an alternative material coupled with detailed technical parameters in the construction of precise-shaped cartilage.