Injectable cell-laden hydrogels fabricated with cellulose and chitosan nanofibers for bioprinted liver tissues

Bio-based hydrogels as three-dimensional (3D) constructs have attracted attention in advanced tissue engineering. Compared with conventional two-dimensional (2D) cell culture, cells grown in 3D scaffolds are expected to demonstrate the inherent behavior of living organisms of cellular spheroids. Herein, we constructed cell-laden nanofiber-based hydrogels in combination with 2,2,6,6-tetramethylpiperidine 1-oxyl-oxidized cellulose nanofiber (TOCNF) and chitosan nanofiber (CsNF) for bioadaptive liver tissue engineering. The carboxylates of TOCNF and amines of CsNF were directly crosslinked via EDC/NHS chemistry. The rheological properties of the solutions for the nanofibers and hydrogels revealed sufficient physical properties for the injection, printing, and plotting process, as well as significant encapsulation of living cells. As-designed hydrogels exhibited excellent viscoelastic properties with typical shear-thinning behavior, and had a storage modulus of 1234 Pa ± 68 Pa, suitable for cell culture. Non-cytotoxicity was confirmed using a live/dead assay with mouse-derived fibroblast NIH/3T3 cells. Human hepatocellular carcinoma HepG2 cells could be cultured on a gel surface (2D environment) and encapsulated in the gel structure (3D environment), which enabled 10 d growth with high gene expression level of albumin of HepG2 spheroids in the 3D gels. The biodegradable cell-laden hydrogels are expected to mimic the cellular microenvironment and provide potential for bioadaptive 3D cell cultures in biomedical applications.


Introduction
As reported by Vantage Market Research, the global three-dimensional (3D) cell culture market size was valued at approximately USD 905.8 million in 2021 [1]. Supposing a compound annual growth rate of 15.1%, the market is estimated to reach USD 2051.5 million by 2028 [1]. The potential of the 3D cell culture market is mostly based on the increasing focus on developing alternatives to animal testing. Additive manufacturing, also known as 3D printing, is one of the most revolutionary technologies of 21st and has driven innovations in multidisciplinary areas [2]. Since 2004, there has been a growing trend of using 3D printing in tissue engineering and regeneration medicines, which is known as 3D bioprinting.
A precise 3D cell culture system can contribute to mimicking the typical morphology and microarchitecture of organs [3,4]. Three-dimensional tissueengineered models for COVID-19 [5], cancer [6], and other clinical disorders [7] are promising alternatives to traditional cell culture techniques. The models also have the potential to serve as a relatively simple in vitro tumor − host environment compared with cost-effective two-dimensional (2D) techniques [8,9].
Injectable inks for bioprinting, called bioinks, are key biomaterials that can be printed into different shapes with cultured cells on the surface or inside of the constructs. Natural polymer-derived hydrogels are common materials with tunable viscosity that are used in extrusion bioprinting systems. The most widely used, nature-derived polymer materials are collagen, gelatin derivatives such as gelatin methacryloyl (GelMA), alginate, acid-soluble chitosan, hyaluronan, and fibrin/fibrinogen [10][11][12][13][14]. Alternatively, synthetic polymer-based bioinks, such as polyethylene glycol and polylactic acid, have advantageous properties, such as control of the mechanical stability, like elastic modulus and tensile strength [15,16]. Current natural or synthetic biomaterials used in the production of bioinks may fulfill either the biomaterial or biological requirements, although typically not both. Researchers have attempted to improve the properties of bioinks by adding supporting materials or involving secondary crosslinking [12,17]. However, complex formulations or extra crosslinking may lead to an increase in cost and time, as well as creating new issues.
Bio-based polysaccharide nanofibers, such as 2,2,6,6-tetramethylpiperidine 1-oxyl-oxidized cellulose nanofiber (TOCNF) and chitosan nanofiber (CsNF), have been investigated to design novel bioadaptive cell culture scaffolds, contributing to the cell attachment, proliferation, and migration [18,19]. However, in 2D cell culture substrates like films or casted hydrogels, the adhered cells generally have a flat shape with a single layer, which is significantly different from the actual 3D environments in living bodies. In fact, 3D cell cultures provide suitable microenvironments much closer to in vivo cell-cell interactions, cell differentiation potential, and drug metabolism than 2D cultures [7][8][9]. Our recent work also confirmed that the porous 3D cell culture scaffolds comprising TOCNF and CsNF fabricated via Pickering emulsion templating could be applied as a suitable extracellular matrix (ECM)-mimicking structure upon which hepatocytes could grow inside of the 3D solid foams [20]; however, the scaffold had limited mechanical strength for long-term incubation. Furthermore, relatively easy-to-handle tissues, such as skins, blood vessels, and myocardia, have been reproduced, while functional complex tissues, such as liver and kidney, are still limited in bioprinting applications.
In this work, our strategy was to construct an injectable cell-laden hydrogel comprising TOCNF and CsNF, aiming to mimic living liver tissue. By regulating the weight ratio of TOCNF and CsNF under different crosslinking conditions, human hepatocellular carcinoma HepG2 cell-laden hydrogels with excellent viscoelastic properties and high cell viability were successfully fabricated. Compared with HepG2 cells cultured on the surface of bioink constructs, the cells printed within the bioink constructs successfully formed spheroidal structures with increasing sizes over two weeks, indicating the possibility of culturing functional liver cell spheroids in the hydrogels. In addition, the cells cultured within bioink exhibited 2.5-fold higher gene expression level of albumin than those cultured on convential 2D substrates. This work further investigated the potential of biodegradable, non-cytotoxic, and injectable nanofiber-based hydrogels for 3D cell culture and biomedical applications.
The mouse fibroblast-like cell line (NIH/3T3) was purchased from RIKEN CELL BANK, Japan. Human hepatocellular carcinoma HepG2 cells were provided by the Japanese Collection of Research Bioresources Cell Bank, National Institutes of Biomedical Innovation, Health and Nutrition, Japan. Dulbecco's modified Eagle's medium (DMEM; high glucose), Lglutamine, penicillin, streptomycin, sodium pyruvate solution, and trypsin−ethylenediaminetetraacetic acid were obtained from Invitrogen Co., USA. Fetal bovine serum (FBS) was obtained from Biowest Co., Ltd, France. Tissue culture polystyrene (TCPS) plates (4-well and 24-well, Thermo Scientific Co., Ltd, Japan) and 35 mm Petri dishes (Sumitomo Bakelite Co., Ltd, Japan) were used as cell culture substrates. Calcein AM (4 mM DMSO solution) for green fluorescence staining of live cells and ethidium homodimer III (2 mM DMSO/H 2 O 1:4 (v/v) solution) for red fluorescence staining of dead cells were purchased from PromoCell GmbH, Germany (Live/Dead Cell Staining Kit II).

Hydrogel preparation
In accordance with a previously reported protocol [21], the CsNF suspension (2.0 wt%) was treated with an ultrahigh pressure water jet system equipped with a dual nozzle (StarBurst Labo, Sugino Machines Limited, Uozu, Japan) at 200 MPa for ten passes to untangle commercial CsNF samples. For preparing the TOCNF/CsNF preformed gel (T/C pre-gel), TOCNF (1.5 wt%) and CsNF (0.3 wt%) suspensions were used, and the concentration of NaCl in each nanofiber suspension was fixed at 10 mM. These mixtures were further treated by an ultrasonic homogenizer (VP-300N, Taitec Corp., Saitama, Japan) for 30 s. The weight ratio of T/C in the pre-gel was fixed at 20:1 or 5:1, coded as T 20 C 1 or T 5 C 1 , respectively.
For amide coupling of TOCNF and CsNF, 50 mM EDC with 30 mM NHS was added to each 10 ml of pre-gel, followed by homogenization for 30 s, and then incubation at 37 • C for ca. 10 min until complete gelation. The obtained hydrogels were defined as GEL-T 20 C 1 or GEL-T 5 C 1 , corresponding to the weight ratio of the two nanofibers.

Characterization of nanofibers and pre-gels
Transmission electron microscopy (TEM) observation of the morphology of TOCNF and CsNF was performed using a JEM-2100HCKM microscope (JEOL Ltd, Tokyo, Japan) under an accelerating voltage of 200 kV at the Ultramicroscopy Research Center, Kyushu University. The suspension of TOCNF or CsNF was dropped onto a copper grid and dyed with 1% sodium phosphotungstate before TEM observation. Atomic force microscopy (AFM) observation of the surface nanotopography of TOCNF, CsNF, T 20 C 1 , and T 5 C 1 was performed using a dimension icon AFM (Bruker Japan, Tokyo, Japan) in tapping mode with a RTESP-300 probe (k = 40 N m −1 , F 0 = 300 kHz) under ambient conditions. The zeta potential of the nanofiber suspensions and pre-gel suspensions was determined by a Zetasizer Nano ZS (Malvern Panalytical, Tokyo, Japan). Xray diffraction (XRD) patterns were recorded using a Rigaku SmartLab diffractometer (Rigaku Corp., Tokyo, Japan) at the Center of Advanced Instrumental Analysis, Kyushu University. Freeze-dried samples were pressed to form a disk pellet. The XRD apparatus was operated at 40 kV and 20 mA with Ni-filtered Cu K α radiation (λ = 0.1528 nm) in reflection mode. Scattered radiation was detected from 5 • to 40 • at a scan rate of 2 • min −1 with 0.01 • intervals.

Hydrogel characterizations
The Fourier transform infrared (FTIR) spectra of the samples were recorded in the range of 400 cm −1 -4000 cm −1 using an FT/IR-620 spectrometer (JASCO Corp., Tokyo, Japan). Approximately 3 mg of the dried samples were mixed with KBr at a ratio 1/100 and pressed into a pellet under vacuum.
The morphology of the hydrogels was determined using a scanning electron microscope (SEM; SU3500, Hitachi High-Tech Corp., Tokyo, Japan) operating at 5 kV. In the preparation of the samples for SEM observation, the pre-gel suspensions and hydrogels, which were frozen in liquid nitrogen and then lyophilized, were coated using an osmium coater (HPC-1SW, Vacuum Device, Ibaraki, Japan) operated for 2 s.
To perform rheological tests on the samples, 3 ml of pre-gel suspensions including crosslinkers were poured into Petri dishes, and heated at 37 • C until complete gelation. After fully crosslinking, the gelled samples were taken out of the dishes, and measured using a rheometer (Rheosol-G2000, UBM, Tokyo, Japan). A typical oscillatory mode in 37 • C was incorporated for rheological tests. The storage modulus (G ′ ) and loss modulus (G ′′ ) were determined in the linear viscoelastic region. Furthermore, a steady shear rate sweep was performed to investigate the shearthinning behavior of the hydrogels as a function of the shear rate from 0.01 s −1 to 100 s −1 . Each sample was replicated at least three times, and each analysis was repeated at least three times.
For the printability test, the pre-gels with crosslinking agents were transferred to 3 ml syringes and allowed to gel completely at 37 • C. The syringe was then fitted with 18 G (inner diameter: 0.8 mm), 20 G (inner diameter: 0.6 mm), or 24 G (inner diameter: 0.3 mm) needles to print various alphabets and Chinese characters.

2D seeding of NIH/3T3 cells and HepG2 cells
To evaluate the cytotoxicity of the prepared hydrogels, cells were seeded on flat gels and cultured. Hydrogels prepared in Petri dishes were hollowed out to the same size with a 14.5 mm cork borer and washed with deionized water to remove residual chemicals. The hydrogel samples for 2D cell seeding were sterilized with UV light for 60 min and immersed in ethanol for 15 min. Gels were then washed with phosphate buffered saline (PBS) to removed residual ethanol, and immersed overnight in DMEM supplemented with FBS (10% v/v), penicillin (100 U ml −1 ), and streptomycin (100 µg ml −1 ) overnight under ambient temperature. After removal of DMEM, 0.5 ml of the cell suspension (cell density of 25 000 cells/well for NIH/3T3 cells; 50 000 cells/well for HepG2 cells) was seeded on each sample, and incubated at 37 • C under a humidified atmosphere of 5% CO 2 and 95% air for the designated time. For cells cultured for 3 d or more, the DMEM medium was refreshed every 2 d. Cells seeded on the TCPS plates at the same density without hydrogel materials were used as the control. The cell survival on each sample was assessed using a Live/Dead Cell Staining Kit II. Microscopic observations of cell adhesion, morphological behavior, and fluorescent images were performed with a Leica DMI 4000B microscope (Wetzlar, Germany).

3D bioprinting of hydrogel bioinks encapsulating HepG2 cells
Cell-embedded hydrogel samples were subjected to 3D cell culture. The pre-gels T 5 C 1 were sterilized by autoclaving at 121 • C for 20 min. The pre-gels were then cooled to room temperature and stirred until achieving a liquid-like suspension. The 3D printing process is shown in figure S5, where the yellow area indicates room temperature, and the green area indicates 37 • C. The solutions containing EDC (174 mM) and NHS (232 mM) were sterilized using a 0.22 µm filter. Sterilized mixture of EDC with NHS (0.3 ml) was added to a 15 ml centrifuge tube with the pregels (3.5 ml). After sufficiently mixing by gentle pipetting and then incubating, GEL-T 5 C 1 was formed at 37 • C in the tube. At the same time, the cells cultured in a flask were collected, and a predetermined number of cells were re-suspended in DMEM (0.2 ml). The crossliking was triggered by EDC/NHS chemistry, and heated at 37 • C just before mixing with cell suspension at 37 • C, followed by mild mixing of the corsslinked hydrogel with the cell suspension. The mixtures were transferred to a 1 ml syringe, and then injected using a 20 G needle to fabricate the desired constructs.
Bioinks encapsulating different densities of HepG2 cells were prepared, and a circular construct (0.1 ml) was printed in a 35 mm diameter dish. After printing, 1.9 ml of DMEM was added. Samples were cultured at 37 • C under a humidified atmosphere of 5% CO 2 and 95% air for the designated time, ranging from 1 d to 14 d. For long-term cell culture, the DMEM medium was refreshed every three days. The survival of the cells grown on each sample was assessed using Live/Dead Cell Staining Kit II. After staining, Z-stack images were acquired using an inverted research microscope (ECLIPSE TE2000-U; Nikon Corp., Japan).

RNA extraction and quantitative RT-PCR analysis
Total RNA was isolated from HepG2 cells cultured for 10 d using ISOGEN II (Nippon Gene Co., Ltd, Tokyo, Japan), followed by genomic DNA removal and reverse transcription using ReverTra Ace ® qPCR RT Master Mix with gDNA Remover (Toyobo Co., Ltd, Osaka, Japan). The qRT-PCR was carried out using Brilliant III Ultra-Fast SYBR ® Green QPCR Master Mix with Low ROX and the AriaMx Real-time PCR System (Agilent Technologies Japan, Ltd, Tokyo, Japan), in accordance with the manufacturer's protocols. PCR primer sequences are listed in supplementary table S1, and primer sequences for hepatocyte nuclear factor 4 alpha (HNF4ɑ) and albumin were those reported by Sasaki et al [22]. Relative expression levels of the respective mRNAs were determined by the comparative Ct method; the expression levels of individual mRNAs were normalized to the levels of the reference gene HPRT1, which encodes hypoxanthine phosphoribosyltransferase 1.

Statistical analysis
The quantitative data are presented as the mean ± standard deviation using Origin 9.0 (OriginLab Corporation, Northampton, Massachusetts, USA). The sample sizes (n) reflect the number of independent replicates and are indicated in figure legends.

Characteristics of the polysaccharide nanofiber-based hydrogels
Based on our previous studies, the morphological structures formed by different types of nanofibers strongly influence the cell culture behavior [18][19][20]. In this work, the nanofiber structure of TOCNF and CsNF was confirmed using XRD analysis and AFM, as shown in figure 1. Both nanofibers possessed the respective characteristic crystalline peaks, while maintaining the corresponding natural crystalline structure ( figure 1(A)). The crystallinity of TOCNF and CsNF obtained by the Segal method was approximately 58.1% and 60.9%, respectively [23]. The T 5 C 1 hydrogel, which was fabricated with TOCNF and CsNF at a weight ratio of 5:1, exhibited a similar XRD pattern as TOCNF. AFM images displayed the difference in the two nanofiber morphologies, in which TOCNF had a shorter nanofiber structure with an average width of 9.8 nm ± 1.7 nm; CsNF had longer fibers and a larger aspect ratio with an average width of 38.0 nm ± 6.7 nm, which was consistent with our previous work [20]. Because of the larger CsNF, sediment and a slight phase separation occurred in its suspension (figure S1). Additionally, because of the opposite charges of negatively charged TOCNF and positively charged CsNF (table  S2), aggregates easily formed in the T/C mixtures, as observed for sample 8 in figure S1. Thus, to promote homogenization, we treated the CsNF using a water jet system according to the reported protocol [21]. The morphology of CsNF treated before and after homogenization was investigated by TEM, as shown in figure S2. The TEM images further confirmed that TOCNF was much shorter than CsNF. Moreover, after homogenization by the water jet system, CsNF was broken into smaller nanofibers; however, also along with the longer fibers ( figure S2(C)). After physical homogenization of the CsNF samples, the T/C mixture remained clearly translucent, and exhibited better stability, as shown by sample 9 in figure S1.
Considering the influence of ionic strength on the stability of nanofiber suspensions, the zeta-potential values of the two nanofibers and pre-gels with or without NaCl were determined, as listed in table S2. In this study, there was little influence of NaCl on the charges of the nanofibers used. The entire gelation behavior via EDC chemical crosslinking is illustrated in figure 2(A). Direct interfiber amidation was expected to form between the carboxylate of TOCNF and amine of CsNF. As shown in figure 2(B), the successful binding of TOCNF and CsNF was confirmed After the crosslinking reaction, porous structures were confirmed for GEL-T 20 C 1 and GEL-T 5 C 1 , which were significantly different than the pre-gel samples. After complete gelation, no flow was observed in figure 2(D). These results indicated the successful bridging of TOCNF and CsNF, which greatly contributed to the formation of rigid frameworks using the polysaccharide nanofibers.

Mechanical properties of the hydrogels
For practical bioink products, the rheological properties of hydrogels influence their printability and injectability [24]. As shown in figure 3(A), the hydrogels exhibited typical shear-thinning behavior, which was similar to the inherent TOCNF suspension. The viscoelastic behavior of the storage modulus (G ′ ) and loss modulus (G ′′ ) of GEL-T 5 C 1 and GEL-T 20 C 1 were determined, as shown in figure 3(B), and the values are listed in table S2. For both hydrogels, G ′ was much higher than G ′′ , indicating a solid-like gel behavior. Moreover, the G ′ value, which was a significant factor to mimic the ECM environment for cell culture, could be regulated by simply controlling the T/C ratios. Although GEL-T 20 C 1 had a higher content of TOCNF, which might contribute to higher hydrophilicity, GEL-T 5 C 1 , with a relatively higher CsNF content, had much stronger crosslinking density between nanofibers, which also contributed to the higher G ′ value ( figure 3(B)). GEL-T 5 C 1 , with a higher G ′ value of approximately 1234 Pa ± 68 Pa, also revealed a desirable printability and injectability, as shown in figure 3(C). Movies S1 and S2 display the videos of the bioinks being manually extruded as a preliminary trial, although the actual printing parameters require further study. Different sizes of needles were used to determine the print compatibility, and the results demonstrated that GEL-T 5 C 1 had excellent compatibility with different size needles, enabling the use of commercial bioprinters. Additionally, as shown in figure S3, different Chinese letters, Latin alphabet B characters, and lattice-like shapes could be printed using the GEL-T 5 C 1 gel bioink, confirming its considerable printability. The printed constructs did not require secondary crosslinking, e.g. with calcium ions or UV light irradiation, to enhance the as-constructed structures. As no washing processes for the hydrogels were required, DMEM could be directly used after printing for further cell culture. A color change from cloudy white to reddish pink was observed when phenol-red DMEM was added to the hydrogels; as a result, water in the hydrogels could be easily exchanged with medium by immersion ( figure 3(D)). Therefore, as-designed hydrogels were medium-friendly and able to maintain their original shapes in DMEM for a week ( figure 3(D)).
As shown in figure 4(A), the swelling rates of the two hydrogels were determined by monitoring for 24 h in distilled water, PBS solution, or DMEM. Both samples had high water uptake capacity because of the hydrophilic nature of both nanofibers and porous structures after freeze-drying. Interestingly, the determined swelling rate of GEL-T 5 C 1 was much greater than previous works using cellulose/chitosan hydrogels [25][26][27], indicating its potential application in bio-printing 3D cell culture. The weight loss of the two hydrogels was measured over 14 d, as shown in figure 4(B). Both hydrogels demonstrated limited weight loss over 7 d because of the strong and successful crosslinking that contributed to the integrity of the network structures comprising crystalline polysaccharide nanofibers, as shown in figure 2(C). Thus, as-prepared hydrogels fulfilled the physical requirements of cell culture scaffolds.

Cytotoxicity of hydrogels for 2D culture
The cytotoxicity of the two hydrogels was evaluated by direct contact of mouse-derived fibroblast NIH/3T3 cells and human hepatocellular carcinoma HepG2 cells. Each cell was cultured on hydrogel-loaded TCPS, which exhibited continuous cell proliferation. As shown in figure 5(A), NIH/3T3 cells were well spread on the surface of hydrogels after 3 d of incubation without dead cells, indicating non-cytotoxicity. Furthermore, the proliferation rate of cells was greater on the hydrogels than on the control substrate, TCPS. As shown in figure 5(B), most of the HepG2 cells proliferated well on the surface of the hydrogels, while a few dead cells were observed on the TCPS substrate. Additionally, the cells grown on the hydrogels tended to form spheroid-like aggregates, which were very different to the cells that proliferated on the plastic TCPS substrate. Depending on the cell type, such as fibroblasts and epithelial cells for NIH/3T3 and HepG2 cells, respectively, the cell culture results were different. The nanofiber morphology and viscoelasticity of the polysaccharide nanofiber-based hydrogels have influenced the cell shape and cellular activity. However, at day 7 in figure S4, there was nearly no space for cells to grow because they reached confluency, and the stack of cells caused more dead cells. As a result, it  was necessary to develop a 3D cell culture system that mimicked the ECM structures and functions in vitro.

Cell behavior of HepG2 cells for 3D culture
GEL-T 5 C 1 with satisfactory mechanical properties, exchangeable capacity, and excellent stability for long-term incubation was selected for the 3D cell culture experiments. Figure S5 illustrates the procedure for bioinks comprising HepG2 cells and pre-gel comprising TOCNF and CsNF. The procedure is also detailed in section 2.6. The confocal laser scanning microscopy images of HepG2 cells, with different seeding density, are shown in figure 6 after 1, 6, 10, and 14 d incubation. Because the 3D culture of HepG2 cells via GEL-T 5 C 1 did not involve further washing after printing, residual crosslinking agents, such as EDC and NHS, might have an adverse effect on the cell growth and proliferation. However, the HepG2-laden hydrogels showed high cell viability, and few dead cells were observed, suggesting that the possible effect of the chemical crosslinking agent on the cells could be ignored, as previously reported [28][29][30]. Few dead cells were observed at day 1, and thus there was nearly no adverse effect on the cells during the printing process of extruding the bioinks through a fine needle. Moreover, when the spherical construct was observed under a microscope, the cells were well dispersed in the hydrogels. No significant decrease in viability was confirmed throughout the 14 d incubation.
During long-term 3D cell culture, HepG2 cells tended to form spheroids, instead of large aggregations, which was significantly different than the 2D cell culture. As shown in figure 6(A), the diameter of the spheroids increased with increasing incubation periods. Moreover, under different seeding densities, all of the samples exhibited excellent cell proliferation because of the larger space for cell growth in the 3D direction that was created via the robust network structure of the crystalline nanofibers, TOCNF and CsNF. As shown in figure 6(B), by observing different locations in GEL-T 5 C 1 , no leakage of HepG2 cells was found, confirming its remarkable encapsulation capacity. Furthermore, the cells that were grown around the edge of the hydrogel constructs had better proliferation than the central part, which might indicate the influence of mechanical properties on the cell culture. As shown in figure 6(C), the Z-stacking images of hydrogels revealed that the HepG2 cells were well encapsulated in the whole hydrogel, and no significant difference was observed in different stacks. Moreover, numerous dead cells were observed in the TCPS culture. Comparing with the single cell layer of TCPS, cells are proliferated in the whole hydrogel, which contributed to the formation of spheroids. As-designed polysaccharide nanofiber-based hydrogels must have a high encapsulation capacity for 3D cell growth.
As shown in figure 7(A), the clustered cell area distribution was determined after 1, 6, and 10 d incubation under two different seeding densities (5 × 10 5 and 10 × 10 5 cells ml −1 of hydrogel). After 1 d of incubation, most of the cells, approximately 80%, were smaller than 100 µm 2 , indicating the cells were individual and well distributed inside the 3D constructs. With increasing incubation time, the cell size increased from 100 µm 2 to 300 µm 2 at day 6 for both seeding densities. However, after 10 d, the cell size distribution of the higher seeding density was obviously different to the lower density. The cell size was evenly distributed from 100 µm 2 to 2000 µm 2 under lower seeding density, while for the higher seeding density, the cell size was focused on 100-300 µm 2 . It was difficult to detect the accurate cell number; however, a high cell viability greater than 90% was approximated in all periods of the 10 d culture shown in figure 7(B). On day 6, the samples shared the highest viability of approximately 94%, which indicated the highest stability of the samples. As shown in figure 7(C), the mean lengths of clustered cells were calculated by digital analysis. With increasing incubation from day 1 to day 10, the mean length of the cell was gradually increased, which also indicated a greater proliferation rate. To assess the liver function, total RNA was extracted from HepG2 cells cultured in the hydrogels to measure the expression levels of albumin and HNF4ɑ genes, as shown in figure 7(D), suggesting higher gene expression level of albumin of HepG2 cells in the 3D gels than that on the 2D TCPS substrate.

Discussion
The morphology and crystallinity of the two polysaccharide nanofibers were determined by AFM, TEM, and XRD analyses. Each polysaccharide nanofiber in the GEL-T 5 C 1 sample maintained their inherent crystalline structures as estimated by the XRD patterns. Our previous studies revealed that the nanofiber structures were significant in cell culture and for cell behavior [17][18][19]. Our research strategy focused on the flexible polysaccharide nanofibers and their combination, although various previous studies investigated the combination of other soluble biopolymers [31][32][33]. Thus, the hydrogel samples were expected to exhibit different microenvironments for cell growth, and preferable cell behavior for 3D printing. Because of the opposite charges of the two nanofibers (TOCNF, COO − = 1.59 mmol g −1 ; CsNF, NH 3 + = 4.35 mmol g −1 ), some aggregations of the bioinks occurred, possibly adversely influencing the pre-gel stability. According to previous works, the stability of nanofiber suspensions can be improved by adding an appropriate amount of NaCl [34,35]. As a result, 10 mM NaCl was added to the two nanofiber suspensions before gelation. Although the zeta-potential values showed little change, the storage modulus of the TOCNF suspensions and pre-gel (T 5 C 1 and T 20 C 1 ) were markedly enhanced, as shown in table S2. Therefore, the 3D printing process of cell-laden hydrogels illustrated in figure S5 could be applied to making hydrogel constructs containing live HepG2 cells. The pre-gels for 3D cell culture before crosslinking could be stored in a fridge at 4 • C for over a month, indicating the long-term storage capacity and the ability to be transported (date not shown). Moreover, the pre-gels could be stored without a sterile environment, and a typical autoclave treatment could be applied before cell culture, which would also contribute to the convenience of storage and handling. EDC crosslinking is simpler and lower cost than other crosslinking methods [36,37]. Only 1.31 µmol of EDC and 1.74 µmol NHS were required for one sample in the following test. As shown in figure 2(C), significant changes in the gel section morphology were confirmed. Different from the single-layer porous structure of GEL-T 20 C 1 , a multi-porous structure was observed in GEL-T 5 C 1 due to higher content of CsNF, which possibly resulted in higher crosslinking degree. The rheological properties of the two gel samples confirmed that the G ′ value could be regulated by controlling the T/C ratios, and the higher content of CsNF contributed to the higher covalent crosslinking intensity, which would have a higher G ′ value, as shown in figure 3(B). In addition to the rheological properties, the crosslinking mechanism also contributed to the printability and stability of the hydrogels. The typical shear-thinning behavior of hydrogels further improved the printing process, in which the hydrogels could be easily injected from the syringes without blockage, as shown in figures 3(C), (D), S3 and movies S1, S2. Notably, the hydrogels could be plotted without a secondary crosslinking process or support materials, which were required in previous studies with other bioinks [38,39]. Moreover, the hydrogels exhibited a typical shear-thinning behavior and a considerable storage modulus of 1234 Pa ± 68 Pa that would provide a suitable microenvironment for cell culture [40][41][42]. GEL-T 5 C 1 could be easily printed to a width less than 1 mm, which would be required to mimic the structure of human liver lobules with a size of 1.0-2.4 mm [43,44]. In addition, the as-designed hydrogel biomaterials showed medium-friendly properties, such as high swelling capacity, and the DMEM could freely enter the samples (figures 3(D) and 4(A)). During long-term incubation, the hydrogels exhibited limited weight loss, which indicated high stability ( figure 4(B)). As used in the bioprinting process, GEL-T 5 C 1 with better mechanical performance was involved.
For 2D cell seeding of mouse fibroblast NIH/3T3 cells, the cells grown on the surface of the hydrogels exhibited much faster proliferation than cells grown on TCPS, as shown in figure 5(A). Moreover, almost no dead cells were observed on the hydrogels, which confirmed the non-cytotoxicity of hydrogels. However, when HepG2 cells were used, the cells grown on the surface of the hydrogels tended to induce aggregation, as shown in figure 5(B). Compared with figure  S4, the increasing incubation periods create more dead cells on both TCPS and hydrogels because of the limited growing space, which is a common disadvantage of 2D culture. As a result, 3D cell culture is a promising approach to investigate the cell-to-cell or cell-to-matrix interaction behavior over time.
For 3D cell culture test, human hepatocellular carcinoma HepG2 cells were embedded in the EDC/NHS-treated pre-gels, and immediately injected into different shapes for long-term incubation from 1 d to 14 d. We did not include a step to remove residual chemicals because the concentration in the gels was quite low, and estimated to be negligible due to further dilution by large amounts of a culture medium. As shown in figure 6(A), after 1 d of incubation, cells grown inside the constructs exhibited a high viability with uniform cell size and distribution. This phenomenon confirmed that limited damage occurred during the printing process. Moreover, with increasing incubation periods, the cell size also increased, eventually forming large spheroids on day 14. By using different cell-seeding densities, varying from 1 × 10 5 cells ml −1 to 1 × 10 6 cells ml −1 , sufficient space for cell proliferation could be maintained for long-term cell culture, which demonstrated a significant advantage over conventional 2D cell cultures. An H 2 O 2 -enzymatic dynamic system was previously fabricated for supporting cell growth post-treatment [45]. However, in our work, the porous structure enabled the long-term cell growth without complete collapse of the hydrogels. The rigid, crystalline polysaccharide nanofibers provided robust frameworks for cell proliferation inside the hydrogels. The spheroidal structures of HepG2 cells are expected to exhibit the complex liver functions observed in vivo. In addition, larger spheroids were located on the edge of the gels in figure 6(B). In previous studies, a 3D emulsion-based scaffold [20] and an alginate-GelMA-CNC bioink [46] had similar results, which may indicate characteristic cell behaviors. Albumin production is a representative liver-specific function. HNF4a is involved in the regulation of the constitutive expression of genes that are responsible for the specialized functions of the liver and in the regulation of metabolism [47]. As compared to the 2D TCPS culture, albumin gene expression drastically increased about 2.5-fold, suggesting that the liver function of HepG2 cells inside the hydrogels was clearly improved over 2D culture systems, although there was no difference in HNF4ɑ expression level, albeit preliminary consideration. Previous studies have already revealed that 2D layer and 3D spheroid cultures of HepG2 cells result in different gene expressions and functional phenotypes, with enhanced P450 activity and albumin synthesis in the spheroids [48,49]. Our encapsulation technique using polysaccharide hydrogels also exhibited the similar result for albumin production during 10 d 3D spheroid culture, confirming once again that 3D cultures can significantly improve the metabolic functions of liver cells. As one of possible approaches, our novel strategy for the combination of two natural crystalline polysaccharide nanofibers would provide new insight into advancing 3D bioprinting in tissue engineering.

Conclusion
Injectable hydrogels, with excellent viscoelastic properties and printability, were successfully fabricated using two natural polysaccharide nanofibers, woodderived TOCNF and crab-derived CsNF, via chemical crosslinking. Different ratios of TOCNF and CsNF (T/C as 5/1 or 20/1) were used to prepare the hydrogels named GEL-T 5 C 1 and GEL-T 20 C 1 . The EDC/NHS crosslinker reacted with carboxylates of TOCNF and amines of CsNF, contributing to the enhanced storage modulus and viscoelastic property. By using a 2D cell culture of mouse fibroblast NIH/3T3 cells, the hydrogels exhibited noncytotoxicity and better proliferation than TCPS. By using a 3D cell culture of human hepatocellular carcinoma HepG2 cells, different sizes of cell spheroids that are typically observed in vivo were obtained together with high gene expression level of albumin of HepG2 spheroids, under long-term incubation. Considering that cell-laden hydrogels are widely suitable for 3D cell culture and drug loading and delivery, the use of bio-based hydrogels is expected to have promising applications as biomedical materials, especially for advanced tissue engineering. The cellladen hydrogels, which are constructed by using natural polysaccharide nanofibers, are also expected to advance the development of multicellular culture systems via a renewable and sustainable way.

Data availability statement
The data cannot be made publicly available upon publication because no suitable repository exists for hosting data in this field of study. The data that support the findings of this study are available upon reasonable request from the authors.