Nanocellulose Reinforced Hyaluronan-Based Bioinks

Bioprinting of hydrogel-based bioinks can allow for the fabrication of elaborate, cell-laden 3D structures. In addition to providing an adequate extracellular matrix mimetic environment and high cell viability, the hydrogels must offer facile extrusion through the printing nozzle and retain the shape of the printed structure. We demonstrate a strategy to incorporate cellulose oxalate nanofibrils in hyaluronan-based hydrogels to generate shear thinning bioinks that allowed for printing of free-standing multilayer structures, covalently cross-linked after bioprinting, yielding long-term stability. The storage modulus of the hydrogels was tunable between 0.5 and 1.5 kPa. The nanocellulose containing hydrogels showed good biocompatibility, with viability of primary human dermal fibroblasts above 80% at day 7 after seeding. The cells were also shown to tolerate the printing process well, with viability above 80% 24 h after printing. We anticipate that this hydrogel system can find broad use as a bioink to produce complex geometries that can support cell growth.


■ INTRODUCTION
Hydrogel-based three-dimensional (3D) cell culture has garnered steadily increasing interest as it affords the possibility to grow cells in a geometry more similar to their native microenvironment compared to conventional two-dimensional (2D) systems. 2D cell culture substrates are typically rigid coated plastic surfaces, which lack the ques driving many aspects of cellular behavior, such as proliferation, 1 differentiation, 2 and migration. 3,4 An optimized 3D cell culture scaffold can mitigate many of these drawbacks. In addition to a favorable geometry, hydrogels used for a 3D cell culture can provide the structural and chemical components that are integral for cells to thrive. 1 Hydrogels based on harvested extracellular matrices (ECMs) 5,6 feature the full range of ECM proteins and fibrils. However, due to the complex composition and biological origin of these materials, they are very challenging to characterize and tailor to specific needs. Furthermore, they tend to suffer from large batch-to-batch variation, 7 leading to poor experimental reproducibility. Not to mention that they are expensive to produce. 1,7 Engineered ECM-mimetic hydrogels are interesting alternatives to biosourced ECM, since they allow full control of the included components and high reproducibility with respect to both chemical and mechanical properties. Engineered ECM inspired hydrogels will always have a less complex composition than native ECM, but their properties can be tailored to promote biologically relevant cellular behavior. 8 The advances in 3D cell culture have promoted the use of hydrogel-based scaffolds in several emerging applications, such as drug screening, 9,10 regenerative medicine, 11,12 and 3D bioprinting. 13 For bioprinting, the hydrogels not only must provide a biologically relevant microenvironment but also exhibit rheological properties that support the printing process. Polymer solutions with low viscosity may not maintain an even distribution of cells within the bioink cartridge during the printing process. K. Na et al. found the cell sedimentation to reduce significantly when comparing bioinks with viscosities of 0.003 and 60 Pa·s. 14 Another challenge with low viscosity bioinks is that they may not retain the printed structure, resulting in a poor printing resolution. One approach to address this issue is to increase the viscosity of the bioinks to mitigate cell sedimentation and improve the resolution. However, the high viscosity requires higher extrusion pressure, which can potentially be harmful to cells. 15 Shear thinning hydrogels are attractive components in extrusion bioinks, since this facilitates the dispensing procedure while allowing for printing with high shape fidelity. 16 As the name implies, the viscosity of a shear thinning material will decrease under shear stress, which can reduce the impact of shear forces generated during the extrusion process on the cells, while the higher viscosity at rest benefits shape fidelity and resolution of the printed structures after extrusion. A range of macromolecules, of both biological and synthetic origin, have been explored as structural components in ECM mimetic hydrogels for bioprinting, such as collagen, 17 alginate, 18 hyaluronic acid (HA), 19 polyethylene glycol (PEG), 20 silk fibroin, 18 and cellulose. 21 Cellulose is an abundant biopolymer with excellent cytocompatibility 22 that can be tailored for a wide range of applications. 23−26 Nanocellulose, where the cellulose fibers have been separated into structures with at least one dimension in the nanometer range, are explored for numerous applications, ranging from energy storage devices 25 to reinforcement in composites 23 to biomedicine. 24 In addition to rheological properties that support cell viability throughout the printing process and shape fidelity immediately after printing, the printed structure must maintain high structural stability throughout the time frame in which the printed structure is intended to be used. This is commonly achieved by some mode of cross-linking. The cross-linking procedure should ideally impede neither the extrusion process nor cell viability. Inclusion of a cross-linker in the printing cartridge restricts both the printing time and size since the fully cross-linked material will not be able to flow. A common method of postdeposition cross-linking is to use photoinitiators in combination with UV light, which can have cytotoxic effects. 27 In contrast, noncovalent strategies to form linkages in the gel, such as relying on hydrophobic interactions or hydrogen bonding, is often well tolerated by cells but tends to result in hydrogels with relatively low stability over time and under varying conditions. 27 An attractive option is covalent cross-linking post extrusion through a bioorthogonal reaction.
We have recently developed ECM mimetic hydrogels based on cyclooctyne-functionalized HA (HA-BCN) cross-linked through a strain-promoted azide−alkyne cycloaddition (SPAAC) reaction using multiarm azide-functionalized PEG (PEG-Az 8 ). 28,29 In addition to allow for bioorthogonal crosslinking of the hydrogels, unreacted BCN groups can be utilized for functionalization of the HA with cell adhesion motifs or other functional groups. 8 This versatile hydrogel system has been demonstrated to support the proliferation of several cell lines such as fibroblasts, 28 hepatocytes, 29 and neural cells 28 and has been utilized as a bioink for 3D bioprinting. 28 However, since the cross-linking commences immediately after the crosslinker (PEG-Az 8 ) has been added to HA-BCN, the printing window, i.e. when the hydrogel can be extruded, is limited to a few minutes. Printing too early, before the hydrogel is sufficiently viscous, results in low printing resolution. Waiting too long, on the other hand, results in too extensive crosslinking and obstruction of the extrusion nozzle. This is an issue seen with many hydrogel systems where cross-linking is initiated upon mixing of the components, 30 and which can significantly complicate bioprinting. In this paper, we present a strategy to enhance the printability of this hydrogel system by the incorporation of cellulose nanofibrils (Figure 1), generating high zero shear viscosity hydrogels with a large printing window that both support cell proliferation and allow for convenient bioprinting.

■ EXPERIMENTAL SECTION
Nanocellulose dispersions: Nanocellulose dispersion prepared as previously described 31 was kindly provided by the company FineCell Sweden AB (Linkoping, Sweden). For the sample named NC 0 , the cellulose oxalate powder was dialyzed against Milli-Q water before homogenization. For the samples NC 15 and NC 60 , the cellulose dispersion was diluted 10× by Milli-Q, placed in a round-bottom flask equipped with magnetic stirrer, which was placed in an oil bath preheated to 115°C, and left there with stirring for 15 and 60 min, respectively. Then, NC 15 and NC 60 were again dialyzed against Milli-Q water and centrifuged at 4500 rpm for 60 min at 10°C, after which the supernatant was discarded. The dry content of the nanocellulose dispersions was determined by placing a preweighed sample in an oven at 90°C overnight and then comparing the dry weight to the initial sample weight. Charge density of the nanocellulose dispersion was determined through conductometric titration, as previously described. 32 Synthesis of HA-BCN: The synthesis of HA-BCN was performed as previously described. 33 In short, HA (100−150 kDa, Lifecore Biomedical) was dissolved in MES buffer (100 mM, pH 7) and N- propyl]-carbodiimide and 1-hydroxybenzotriazolehydrate. This solution was then added to the HA. The carbodiimide cross-linking reaction was allowed to proceed for 24 h, followed by dialysis in acetonitrile/water (1:10 v/v) for 24 h, followed by Milli-Q water for 3 days before lyophilization. Hydrogel formation: HA-BCN, PEG-Az 8 (8-armed poly(ethylene glycol) aizde, 10 kg/mol, from Creative PEGworks, Chapel Hill, NC, U.S.A.), and nanocellulose dispersions were UV sterilized (60 kJ/cm 2 ) for 1 min. HA-BCN and PEG-Az 8 were dissolved in PBS or a cell culture medium. The cellulose dispersion was diluted as required by the requisite amount of Milli-Q water to achieve the desired concentration. The HA-BCN solution was mixed with nanocellulose dispersion (or Milli-Q water when preparing a gel without nanocellulose). This HA-BCN:nanocellulose solution was then mixed with the PEG-Az 8 solution, aiming for a BCN/N 3 ratio of 10:1, with the exception of material use for bioprinting. In the latter case, the PEG-Az 8 solution was added after printing a 3D structure of the HA-BCN/nanocellulose mixture. Hydrogel characterization: The mechanical properties of the hydrogels were evaluated through oscillatory rheology using a Discovery HR-2 Biomacromolecules pubs.acs.org/Biomac Article rheometer, TA Instruments. Preformed hydrogels swollen in PBS buffer overnight were evaluated for a minimum of quadruplicate samples at room temperature using an 8 mm plate geometry with frequency sweeps at a fixed oscillatory strain of 1%, and amplitude sweeps at a fixed frequency of 1 Hz. Gelation kinetics was evaluated for at least duplicate samples with a 20 mm 1°cone−plate geometry at 0.1% strain and an oscillatory frequency of 1 Hz. The components of the hydrogel were mixed and immediately placed on the sample holder at 4°C. After bringing the geometry into position and applying a solvent trap to avoid dehydration of the hydrogel, the temperature of the sample holder was increased to 37°C and the measurement started. Creep-recovery tests were performed in a manner similar to the gelation kinetics experiments, but at room temperature. When the samples had cross-linked fully, a constant shear of 50 Pa was applied for 60 min, and strain changes recorded. Then, the shear was released, and strain was recorded for another 75 min. SEM: Samples were prepared for scanning electron microscopy (SEM) by immersion in N 2 (l) followed by lyophilization, mounted on a carbon grid, and finally sputtering with Pt. SEM measurements were performed on a LEO 1550 Gemini (Zeiss) operating at 5 kV. ■ RESULTS AND DISCUSSION Nanocellulose Dispersions. The nanocellulose dispersions used in this study consisted of cellulose oxalate. To study the influence of the charge density on the rheological properties and cell viability of the resulting hydrogels, samples of three different charge densities were prepared by removing oxalate groups from the initial nanocellulose sample (NC 0 ) through hydrolysis. This yielded samples NC 15 and NC 60 , where the subscripts represent the duration of the hydrolysis step. The charge densities of the three cellulose dispersions

Biomacromolecules pubs.acs.org/Biomac
Article were evaluated through conductometric titration on duplicate samples for each dispersion and measured 107, 127, and 207 μequiv/g for dispersions NC 60 , NC 15 , and NC 0 , respectively. These numbers corroborate a difference in degree of functionalization regarding oxalate side groups and that the hydrolysis of NC 15 and NC 60 was successful. The measured charge density for NC 0 is in agreement with previous measurements by FineCell on their cellulose oxalate dispersions (unpublished data). Nanocellulose Reinforced Hydrogels. HA-PEG hydrogels were obtained by cross-linking HA-BCN by PEG-Az 8 through SPAAC as previously described. 28,29 To evaluate the effect of nanocellulose charge density on the rheological properties of the hydrogels, 1% (w/v) HA-PEG was prepared, containing 0.5% (w/v) of NC 0 , NC 15 , or NC 60 and evaluated by oscillatory rheology (Figure 2). The difference in rheological properties between these hydrogels was very small. There was a slight but significant (p < 0.05) increase in loss modulus (G′′) for the hydrogel containing NC 60 compared to NC 0 and NC 15 . Given the very small difference in the rheological properties, subsequent rheological measurements were performed only with NC 0 . The choice fell on NC 0 since the nanocellulose was well dispersed during the homogenization procedure, and the less it was processed after homogenization, the lower was the tendency of the dispersed fibrils to aggregate.
Comparing the rheological properties of 2% (w/v) HA-PEG hydrogels containing between 0 and 0.7% (w/v) of NC 0 (Figure 3a), the addition of 0.1−0.3 w/w% NC 0 had little to no effect on the storage modulus, whereas there was an increase in the storage modulus of the resulting hydrogel for NC 0 concentration above 0.3% (w/v). The tan delta and loss modulus increased with approximately 1 and 2 orders of magnitude, respectively, when increasing the NC 0 concentration from 0 to 0.7% (w/v) NC 0 ( Figure S1). The effect of different concentrations of HA-PEG on the rheological properties was also evaluated. Figure 3b shows an increase in the storage modulus with increasing HA-PEG content. The tan delta was somewhat higher for the sample with the lowest HA-PEG content compared to that for the other three compositions ( Figure S2), which can be a result of the cellulose having a larger influence on the viscoelastic properties of this sample given the low HA-PEG content. These observations indicate that the dispersed nanocellulose forms an entangled network within the covalent HA-based hydrogel that efficiently dissipates energy, resulting in shear thinning hydrogels.
These findings show that the rheological properties of nanocellulose reinforced hydrogels can be fine-tuned by changing the concentrations of either HA-PEG or nanocellulose. In addition, the mechanical properties of the hydrogels could also be modified either by addition of Ca 2+ to increase G′ or cellulase to decrease G′. The oxalate side groups on the cellulose can coordinate with Ca 2+ ions, resulting in ionic cross-linking of the nanocellulose. Both the storage modulus and viscosity of the nanocellulose containing hydrogels increased by approximately 12% when immersed in CaCl 2 , whereas this change was not observed for hydrogels without nanocellulose ( Figure S3). These hydrogels contained   Cellulases are a group of enzymes with the ability to catalyze the hydrolysis of cellulose macromolecules. 34 Immersion of a nanocellulose containing HA-PEG hydrogel in a 3.3 mg/mL cellulase solution resulted in a reduction of the hydrogel storage modulus with more than 50% (Figure 4). The difference between the storage modulus values before and after immersion in cellulase solution was statistically significant for the sample containing cellulose (t(4) = 4.72, p = 0.009), but not for the sample without cellulose. The addition of nanocellulose to the HA-PEG hydrogels had a pronounced effect on the gelation kinetics ( Figure 5). The rate of gelation for the nanocellulose containing hydrogels was markedly higher during the first 20 min compared to hydrogels without nanocellulose. The cellulose fibrils likely become physically trapped within the HA-PEG network during the cross-linking reaction, resulting in a faster increase in storage modulus and viscosity despite the actual covalent cross-linking taking place at a similar rate in the samples with and without nanocellulose. The storage and loss moduli reached higher values during the gelation kinetics measurements than during measurements on preformed hydrogels at the same dry content of nanocellulose and HA-PEG, since the preformed hydrogels were evaluated after swelling in PBS overnight. The effect of swelling in PBS on storage and loss moduli was more pronounced for the nanocellulose containing hydrogels than for hydrogels consisting of HA-PEG alone. The viscosity of the nanocellulose relies on short-range interactions such as van der Waals forces and hydrogen bonding between the cellulose molecules. 35 It is likely that a certain degree of swelling of the hydrogel therefore has a higher effect on the rheological properties of the cellulose component than on the covalently cross-linked HA-PEG.
Creep-recovery tests ( Figure 6) showed that a higher nanocellulose content in the hydrogels increased the permanent deformation of the hydrogels after stress had been applied and released due to rearrangement of the nanocellulose fibrils under stress. A higher permanent deformation corresponds to a material with a higher plasticity,    36,37 To investigate the microarchitecture of the hydrogels, samples composed of 1% (w/v) HA-PEG and 0.5% (w/v) NC 0 were imaged by SEM (Figure 7). The samples were frozen by immersion in N 2 (l) and lyophilized prior to imaging. The hexagonally shaped pores are likely an effect of the freezing procedure. The morphology on SEM was similar for hydrogels with and without nanocellulose ( Figure S4).
Cell Viability. To assess the cytocompatibility of the hydrogels, we encapsulated and cultured both a neuroblastoma cell line (SH-SY5Y) and primary human dermal fibroblasts in the hydrogels. The relatively delicate cell line SH-SY5Y was used to evaluate any possible differences in cell viability when cultured in hydrogels containing nanocelluloses of different charge densities. No significant difference in cell viability were seen for SH-SY5Y cultured in hydrogels containing either NC 0 , NC 15 , or NC 60 ( Figure S5), for which reason subsequent experiments were performed only with NC 0 . Fibroblasts were seeded in 2% (w/v) HA-PEG with and without 0.5% (w/v) NC 0 . Live/dead assay indicated high viability for both conditions at day 7 ( Figure 8) with no significant difference with and without nanocellulose (t(6) = 1.85, p = 0.11). The rounded fibroblast morphology is likely a result of the lack of cell-adhesion motifs in the hydrogels. We have previously demonstrated that cell adhesions motifs, such as the fibronectin and laminin derived peptides RGD and IKVAV, with terminal azide groups can be easily conjugated to the HA-BCN backbone, resulting in improved cell−hydrogel interactions. 8 Printability. Shear thinning hydrogels are attractive as bioinks for extrusion-based 3D bioprinting since this facilitates the dispensing procedure while allowing for printing with high shape fidelity. Addition of nanocellulose (1.3% w/v) to HA- Biomacromolecules pubs.acs.org/Biomac Article BCN (without PEG-Az 8 ) resulted in an increase in storage modulus and viscosity of several orders of magnitude, from ∼3 to ∼5700 Pa and ∼0.5 to ∼900 Pa·s, respectively ( Figure 9, Table 1). Thanks to the high storage modulus and shear thinning behavior of this bioink, printing of self-supporting structures was possible without addition of cross-linker prior to extrusion (Figure 10a−d). The cross-linker (PEG-Az 8 ) was instead added after printing by applying PBS containing 1% (w/v) PEG-Az 8 . After careful optimization of printing parameters, high aspect ratio self-supported multilayer structures could be printed (Figure 10a). After the addition of a cross-linker, printed structures were stable and immersed in PBS (Figure 10b,c), while printed structures that were not cross-linked disintegrated under the same conditions (Figure 10d,e). The shear thinning properties combined with the biorthogonal postprinting cross-linking further ensured that cells retained high viabilities. The viability of the primary human fibroblast printed with this bioink after 24 h was estimated to be 83% (Figure 10f,g).

■ CONCLUSIONS
This study demonstrates the successful incorporation of a nanocellulose oxalate dispersion into a hyaluronan-based bioink. The incorporation of nanocellulose resulted in a hydrogel with excellent printability that could be cross-linked post extrusion using a bioorthogonal SPAAC reaction. The nanocellulose further offered the option of either cross-linking the cellulose oxalate with Ca 2+ to increase storage modulus, or to decrease storage modulus by addition of cellulase which degraded the cellulose component of the hydrogels. The charge density of the cellulose oxalate had little to no effect on rheological properties or cell viability. The rheological properties of the hydrogels could be tuned by adjusting the concentration of either HA-PEG or nanocellulose. We anticipate that this hydrogel system can find broad use as a bioink to produce complex geometries which can support cell growth.
Frequency sweeps for data presented in Figure 3; Rheological properties of hydrogels before and after addition of CaCl 2 ; SEM micrographs of lyophilized HA-PEG hydrogels; Viability of SH-SY5Y cultured in hydrogels containing either NC 0 , NC 15 or NC 60 ; AlamarBlue data related to Figure 8; Quantification of the swelling of hydrogels after storage in PBS overnight compared with immediately after cross-linking (PDF)