Reinforcing Tissue‐Engineered Cartilage: Nanofibrillated Cellulose Enhances Mechanical Properties of Alginate Dialdehyde–Gelatin Hydrogel

Cartilage tissue engineering offers a promising option for treating osteochondral defects. Alginate dialdehyde–gelatin (ADA–GEL) hydrogel has been explored as promising material for soft tissue scaffolds; however, its low stiffness has posed a constraint to load bearing applications. Herein, this limitation is addressed by introducing nanofibrillated cellulose (NFC) into the ADA–GEL matrix. The effect of NFC on the physicochemical properties of hydrogels is evaluated. Fourier transform infrared spectra demonstrate no chemical interaction between NFC and ADA–GEL, while scanning electron microscopy pictures reveal NFC fibers embedded in the hydrogel matrix, thus confirming the fiber‐reinforced composite hypothesis. NFC‐reinforced ADA–GEL (AG‐N) composite hydrogels exhibit increased stiffness, with a maximum compressive effective modulus of 19.6 ± 3.0 kPa at 25% w/w NFC content. ATDC5 cell viability and proliferation as well as chondrogenic differentiation are assessed using immunohistochemical staining for sulfated glycosaminoglycans and collagen type II. A possible application of AG‐N hydrogels as an osteochondral plug is also proposed, with polyetheretherketone as the subchondral bone anchor part. The mechanical properties of the resulting osteochondral device highlight its potential as a promising biomaterial for treatment of osteochondral defects. These findings provide valuable insights into the development of AG‐N hydrogels for load‐bearing tissue engineering applications.


Introduction
Osteoarthritis (OA), a degenerative joint condition, is the major cause of chronic pain and immobility in elderly population.Knee joint OA alone was estimated to affect 22.9% of people aged 40 and above globally. [1]The disease typically affects knees, hips, and spine, leading to restricted mobility and comorbidities that worsen patients' well-being. [2,3]Osteochondral defect (OCD) is the most severe stage of OA in which cartilage has been irreversibly damaged and the lesion on subchondral bone occurred. [4,5]Substitution of the damaged tissue is performed on occasion, with the current process involving the harvesting of tissue from different cartilage in the patients' body; however, the multiple operation carries the risk of postoperative complications. [6]Tissue engineering (TE), in this light, presents a viable technique to restoring damaged joint tissues by replacing OCD with a synthetic material, hence eliminating intrusive interventions.Hydrogels based on biopolymers are often engineered to achieve such application, due to their biocompatibility, tunability, and ability to mimic soft tissue extracellular matrix (ECM).
Alginate is a natural polysaccharide made up of two uronic acids, namely, α-L-guluronate (G) and β-D-mannuronate (M).9] However, there are three limitations to using pure alginate as a scaffold: 1) the absence of cell attachment motifs; [10,11] 2) the lack of biodegradability in mammals; [12][13][14] and 3) inadequate mechanical properties for load bearing [8,15] To address these problems, alginate chains were often modified and/or combined with the other moieties such as polymers or particles.One form of oxidized alginate that is widely used in TE is alginate dialdehyde (ADA), which can be obtained by oxidation reaction with periodate.Unlike alginate, ADA is biodegradable in vivo due to the breakdown of uronic units at C-2 and C-3, where the two carbon atoms form an aldehyde functional group that allows additional chemical interaction to the other moieties. [12,16]These reactive aldehyde groups allow the formation of the Schiff base bond, a covalent cross-linking between the amino groups of gelatin and ADA.This paved the way for the application of ADA-GEL-based hydrogels in TE research.Physicochemical studies on ADA-GEL hydrogels dual cross-linked with Ca 2þ and microbial transglutaminase (mTG) demonstrated that it has a high potential for cartilage regeneration applications.By varying the concentration of mTG, the mechanical properties and degradation rate of ADA-GEL could be adjusted.Hydrogels with mTG of 5% w/v in cross-linking solution or higher remained intact for up to 30 days.19] Nanofibrillated cellulose (NFC), another polysaccharide that is derived from cell wall's cellulose fibers, consists of β-D-glucopyranose as subunit.The diameter of NFC is in the range of 5-60 nm, while the length is in the range of micrometer depending on the production technique. [20,21][24][25][26] Interestingly, alginate-NFC hydrogels were found to induce the expression of cartilage phenotype while having a compressive modulus in the range of 0.53 AE 0.59 MPa after 10 months of implantation in nude mice. [27]Another in vivo investigation using 3D printed alginate-NFC constructs preseeded with human primary nasal chondrocytes (hNCs) reported detection of sulfated glycosaminoglycans (GAGs) and collagen type II (Col II) after 90 days [28] Both studies confirmed that alginate-NFC composites had good structural integrity and can induce cartilage differentiation in vivo.However, the inability to degrade in mammalian bodies may be inconvenient for applications that require biodegradation of the scaffold in controllable manner.
The purpose of this study was to develop and validate a novel NFC-reinforced ADA-GEL composite hydrogel (AG-N) for cartilage regeneration and its potential use as an osteochondral plug.
The influence of NFC on physiochemical and compressive effective modulus was examined, as well as biocompatibility using ATDC5 cells, a murine chondrogenic cell line with cellular characteristics comparable to human chondrocytes. [29]The effect of NFC to the level of chondrogenic markers (GAGs and Col II) was also evaluated.In addition, we presented a biphasic osteochondral device that comprises the developed hydrogel as a scaffold for cartilage regeneration and polyetheretherketone (PEEK) as a subchondral bone substitute.[32][33] We would like to highlight, however, that our proposed model is not limited to PEEK only, as other materials with similar qualities might be used in future iterations.We found that the interface between the two phases has a considerable effect on the overall mechanical properties of the device.With the promising chondrogenic inductive property of AG-N, the proposed biphasic osteochondral device is a potential candidate for the treatment of OA.

ADA Synthesis
Oxidization of alginate was adapted from a protocol described by Sarker et al. [34] Per one batch of ADA, 10 g of sodium alginate (VIVAPHARM Alginate PH176, JRS Pharma GmbH & Co. KG, Germany) was dispersed in 50 mL ethanol (≥99.5%,Carl Roth) in a brown 1000 mL Duran bottle before the oxidation reaction.In the absence of light, the reaction was instigated by the addition of 1.34 g (6.507 mmol) of sodium (meta)periodate (NaIO 4 ) (≥99.8% ACS, Carl Roth) in 50 mL ultrapure water (Direct-Q, Merk Millipore).The mixture with a total volume of 100 mL was left stirring in a water bath, at low rpm and at room temperature for 6 h.Afterward, 10 mL of ethylene glycol (Spectrophotometric grade, ≥99.5%, Sigma-Aldrich) was added to the bottle and stirred for 30 min to ensure complete termination of the reaction.The mixture was left to sit so the oxidized alginate residual settle in the bottom, and then upper layer of ethanol was decanted.Subsequently, the rest of the mixture was dialyzed with dialysis membranes (Spectra/Por1 6-8 kD MWCO, Spectrum Labs) against 17 L of ultrapure water.The water was changed daily over the course of 5 days.The pure ADA solution from this step then underwent lyophilization, yielding dry and yellowish product.The degree of oxidation (DO) was estimated based on the reactions between aldehyde groups on ADA and hydroxylamine hydrochloride (HCl) (70 g L À1 in Methanol R, Bernd Kraft, Germany) which produce HCl as a byproduct; the protocol was adapted from Zhao et al. [35] 100 mg of ADA was dissolved in 7.5 mL ultrapure water (n = 3), and 100 mg of alginate was also dissolved in the same manner (n = 3) as a reference.For the reactant, hydroxylamine HCl was diluted to 0.5 M in ultrapure water and the pH was adjusted to 5 by NaOH (1 N, Carl Roth).7.5 mL of this reactant was then added to each beaker of the samples.All gel samples of ADA and reference alginate were stirred at room temperature for 15 min and then were left at 65 °C in an orbital shaking incubator at the speed of 550 rpm for 6 h to allow reactant to disperse and the reaction to occur homogeneously.We determined DO by titrating 0.1 N NaOH to the HCl released from the reaction, using the following equation: where N NaOH is the normality of NaOH used in titration, V NaOH,ADA and V NaOH,Alg are the volumes of NaOH consumed to adjust the solution back to pH 5 for samples containing ADA and reactant, as well as alginate and reactant, respectively, M w,Alg is the molecular weight of alginate dimer, and m ADA is the weight of ADA dissolved.

Hydrogel Constructs Preparation
All three polymeric components in the hydrogel system were dissolved separately before mixing.To prepare ADA-GEL, gelatin solution (%40 °C) was transferred to a beaker which already contained ADA with the same volume (ADA:GEL is 1:1); this resulted in 10% w/v ADA-GEL system with ADA contributing to 3.33% w/v and gelatin contributing to 6.67% w/v.ADA-GEL was stirred at low rpm, at 37 °C for 15 min to allow the formation of Schiff base bond between ADA and gelatin.NFC was added to the ADA-GEL hydrogel corresponding to volume ratio (ADA-GEL:NFC) of 90:10, 80:20, 70:30, with % w/w NFC/ADA-GEL of 11%, 25%, and 43% respectively.Subsequently, there are four groups of hydrogel samples, labeled according to this volume ratio as ADA-GEL, AG-N-9010, AG-N-8020, and AG-N-7030 (Table 1).Cross-linking solution was 5% w/v mTG (ACTIVA WM, Ajinomoto Co., Inc.) dissolved in 0.1 M calcium chloride (anhydrous, powder, ≥97%, Sigma-Aldrich) solution and was filtered through 0.22 μm sterile syringe filters.A sheet of silicone mold with multiple pits was placed in a polystyrene petri dish to prepare hydrogel constructs, followed by the casting of hydrogels into the pit.Hydrogels were handled using a positive displacement pipette (Microman E, Gilson).The petri dishes were then covered by their lids and wrapped with parafilm before transferring to a À20 °C freezer.The whole petri dishes containing hydrogel in mold were left in the freezer for at least 3 h to solidify the gel into the desired shape. [36]Afterward, the silicone mold was carefully peeled off the dish and the frozen hydrogel was immersed into the cross-linking solution at room temperature for 25 min with the ratio of 2 mL cross-linking solution per one hydrogel construct (8 mm diameter and 3 mm thickness).Finally, the cross-linked constructs were rinsed with Hank's Balanced Salt Solution (HBSS) , Gibco 14025-050, Thermo Fisher).

Chemical and Physical Characteristics
Chemical characteristics of ADA-GEL incorporated with NFC were assessed by attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) (IRAffinity-1S, Shimadzu).Samples of all groups were finely ground from the lyophilized constructs.FTIR was operated at absorbance mode, 40 scans per spectrum and at the 4 cm À1 resolution.The background signal was automatically subtracted by the machine's manufacturer complement software package (LabSolutions IR).Two sets of samples underwent different dehydration techniques lyophilization, and critical point drying (CPD), were gold sputtered and subjected to scanning electron microscopy (SEM) imaging using FIB-SEM (Auriga Cross Beam, Carl Zeiss AG) for the investigation of hydrogel microstructure.For CPD, samples were fixed with 2.5% glutaraldehyde in HBSS and a series of ethanol for water substitution before completely drying by a critical point drier (EM CPD300, Leica).

Stability Test In Vitro
The effect of NFC on shape and mass fidelity over time was investigated under conventional in vitro conditions.Six samples from all groups were placed in cell strainers (polypropylene, SPL Life Sciences, Korea) which were reshaped to fit in wells of 6-well plate and incubated in 7 mL of Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F-12 .438 g L À1 sodium bicarbonate, Gibco 11320-033, Thermo Fisher) supplemented with 5% v/v fetal bovine serum (FBS) and 1% v/v penicillin-streptomycin (Pen Strep) (Gibco 15 140 122, Thermo Fisher), at 37 °C and 5% CO 2 for 28 days.The culture medium was replaced 3 times a week to mimic the actual cell studies practice.At the time points of interest, each cell strainer containing a construct was lifted from the well-plate, and the excessed medium was discarded by sterile tissue papers.Subsequently, the cell strainer was placed in a sterile plastic beaker, and the whole assembly was weighed.The weight of individual hydrogel construct was obtained from the scale read subtracted weight of the cell strainer and the plastic beaker.The weighting of samples was done at 0, 3, 6, and 24 h and 2, 4, 7, 14, 21, and 28 days of incubation.The percentage weight change of the constructs was calculated by the following equation: where m t is the weight at time points of interest and m 0 is the weight before immersion in DMEM/F-12.

Mechanical Properties
The effective modulus of hydrogel constructs in compression before and after 24 h of incubation in culture medium was assessed by compression test using a universal testing machine (Instron 5967, Instron).Six samples from each group and conditions were tested in DMEM (Gibco 10566-016, Thermo Fisher) at room temperature for preincubated and at 37 °C for postincubated samples.We performed experiments at a compression rate of 10 mm min À1 using a 100 N load cell.The effective Young's moduli were the steepest slopes of the stress-strain curve within the range of 30% strain, calculated by Bluehill software, the materials testing software from Instron.In addition, the effect of AG-N on mechanical properties at the microlevel was also investigated using a nanoindenter (Piuma, Optics 11 Life, USA).A matrix of four indentations was performed twice on every six samples (6 mm diameter and 1.5 mm thickness) after crosslinking at room temperature while the samples were immersed in HBSS.The indentation was set up with a 0.25 N m À1 stiffness probe and a tip radius of 23.5 μm.

Cellular Interaction
The cell line used in this study was a murine chondrogenic cell line, ATDC5 (ECACC General Collection, Merk).The complete medium (540 mL) composed of DMEM/F-12 (DMEM/F-12 .438 g L À1 sodium bicarbonate, Gibco 11320-033, Thermo Fisher), 25 mL FBS, 5 mL PS, 10 μg human transferrin (T8158, Millipore Sigma, Merk), and 30 nM sodium selenite (bioreagent, ≥98%, S5261, Sigma-Aldrich).Cells maintenance routine includes changing the medium 3 times per week, and cell detaching for subculture was done by Trypsin-EDTA (0.25%) (Gibco 25200-056, Thermo Fisher), and incubation conditions of 37 °C, 5% CO 2 , and 95% humidity.Cytotoxicity was evaluated according to ISO10993-5, particularly the guideline for the direct test.Cells were seeded into 12-well plate at the density of 3 Â 10 5 cells cm À2 and were allowed to confluent for 24 h before introducing hydrogel constructs into the wells (n = 5).After another 24 h of incubation, the cellular metabolic level was assessed by 3% v/v WST-8 assay solution (CCK-8 kit, 96992, Sigma-Aldrich) in a complete medium.Cell proliferation was evaluated by direct cell seeding into samples (8 mm diameter, 3 mm thickness) in 48-well plate, and the seeding density on hydrogel constructs was 5 Â 10 5 cells cm À2 .After 3, 7, 14 days, metabolic activity was measured by 3% WST-8 assay in complete medium, while cells morphology and proliferation were investigated by fluorescence staining as follows; cell bodies were stained by Calcein AM (Invitrogen C1430, actin fibers were stained by Rhodamine Phalloidin (Invitrogen R415, Thermo Fisher)) and nuclei were stained by DAPI (Invitrogen D1306, Thermo Fisher).Fluorescence images were obtained by an inverted fluorescence microscope (Zeiss Axioscope.A1, Carl Zeiss AG), and images were processed using the manufacturer's compliment software package "Zen Blue" and ImageJ software.

Chondrogenic Marker
To evaluate the influence of NFC on chondrogenesis, samples were prepared as hydrogel films to allow imaging by optical microscope.The hydrogel solution of all groups at 40 °C was cast into 24-well plate (20 μL per well) and cross-linked in situ.Samples were incubated with ATDC5 cells (5 Â 10 5 cells cm À2 seeding density) for up to 21 days.Glycosaminoglycans (GAGs) deposition was evaluated at 14 days after incubation by histological staining using Alcian blue 8 GS (C.I. 74240, Carl Roth) and Safranin O (84120, Sigma-Aldrich) separately.
A stock solution of Alcian blue was prepared by dissolving 1% w/v in 1 N HCl, and the staining was done by incubating with a stock solution for 5 min at room temperature and rinsing 3 times with 1 N HCl to remove excess stain.Safranin O was also prepared as 1% w/v using HBSS as a solvent, the staining duration was 20 min, and samples were rinsed with 95% ethanol 3 times.The stock solution of both histological dyes was filtered using a 0.22 μm PVDF syringe filter.Images were obtained using optical microscopes (Axio Zoom and Axio Stemi305, Carl Zeiss AG).To determine Col II accumulation, samples of 21 days incubation were fixed with 3.7% formaldehyde in HBSS for 15 min before being incubated with 1% bovine serum albumin (BSA) (lyophilized powder, BioReagent, A9418, Sigma-Aldrich) in HBSS at 4 °C to block nonspecific protein binding sites.Col II was first labeled by incubation with Col II Monoclonal Antibody (6B3) (Invitrogen MA5-13026, Thermo Fisher) in HBSS (1:100) at room temperature for 2 h.Alexa Fluor 555 (anti-Mouse IgG, secondary antibody, A-21425, Thermo Fisher) with 10 μg mL À1 concentration in HBSS was used as secondary conjugation to fluorescence-tag the immunogen bounded collagen.Fluorescence imaging was done using a fluorescence microscope (Zeiss Axioscope.A1, Carl Zeiss AG).To assist with a qualitative comparison of Col II secretion, the red channel image was split from the original image and then the color was inverted for the ease of visualization.Later, black pixels representing fluorescence-tagged Col II were counted using Python programming.

Bone-Like Apatite Formation
Four samples from all groups were submerged in simulated body fluid solution (Kokubo's SBF [37] ) to evaluate whether the incorporation of NFC in ADA-GEL induces bone-like apatite deposition.Each sample was submerged in 16 mL SBF using a sterile plastic container.The incubation condition was 37 °C with 80 rpm orbital shaking.SBF was changed 3 times.All hydrogel constructs were collected and rinsed with ultrapure water after 14 days of incubation.The formation of bone-like apatite was assessed by X-ray diffraction technique (benchtop X-ray diffractometer, MiniFlex 600, Rigaku, Japan).

Preparation of an Osteochondral Plug
A potential use case as an osteochondral plug was explored by depositing ADA-GEL-NFC on surface-treated PEEK substrates.PEEKs substrates were prepared by machining a 1 m-long stockshaped PEEK rod (TECAPEEK MT classic white, Ensinger, Germany) of 8 mm diameter into PEEK discs with 3 mm thickness and the same diameter.The upper surface of PEEK discs was modified by two steps as follows.First, millimeter-scale grooves were created on the surface by laser engraving (iLaser4000, LTT Corp., Taiwan).The discs were then submerged in acetone, ethanol, and DI water using a sonicator bath to remove leftover impurities.Second, the same disc surface was immersed in concentrated sulfuric acid (98%, Carl Roth) for 90 s in a sonicator bath, followed by 2 h posttreatment in DI water at 60 °C, in a sonicator bath to discard the sulfuric residues.Deposition of the hydrogel layer was done by casting.Silicone molds were placed in PP petri dishes and then PEEK discs were placed into each pit of the molds, with the modified surface facing upward.AG-N-80 hydrogel heated to 40 °C was cast on top of the discs to fill the cavity.The whole casting assembled as a biphasic osteochondral plug was stored at À20 °C for 3 h to allow the total to the solidification of the hydrogel.Afterward, the bilayer constructs were removed from the mold and underwent cross-linking with 5% mTG in 0.1 M CaCl 2 for 25 min.The bilayer constructs were rinsed by HBSS to remove the excess cross-linking agents.The finish bilayer constructs' dimensions are 8 mm in diameter and 6 mm in height.The groove spacing and depth on the PEEK surface were varied, as shown in Table 2 2.5.

Mechanical Measurements
The effect of the hydrogel-PEEK interface on the mechanical properties of the bilayer constructs was assessed by a Discovery HR-3 rheometer (TA Instrument, Delaware) equipped with an 8 mm parallel geometry.After calibration of the testing device, two sandpapers (8 mm diameter) were attached to the bottom plate and geometry, followed by bonding the hydrogel side of bilayer constructs to the top geometry using cyanoacrylate glue (super glue gel, Pattex).Next, the geometry was slowly lowered to attach the PEEK side to the bottom plate with a preload<0.1 N, where the same glue was applied.A transparent immersion cup was placed before lowering the geometry, and after a few seconds of waiting time for drying the glue, HBSS was poured into the cup to avoid dehydration during testing.The temperature was set to 37 °C.The measurement was conducted on six samples from each group and included the following protocol: 1) cyclic compression-tension with three loading cycles, minimum and maximum stretches of 0.85 and 1.15 and a loading rate of 40 μm s À1 , 2) compressive and tensile stress relaxation with a holding time of 250 s at stretches of 0.85 and 1.15, 3) two sets of cyclic shear with maximum strains of 15% and 30% and a loading rate of 100 μm s À1 , and 4) shear stress relaxation with a holding time of 250 s at 30% shear strain.

Statistical Analysis
Quantitative data from experiments were analyzed using one-way ANOVA and Tukey test, where p-value <0.05 was considered significant for comparison among sample groups.Charts and bars represent mean value, and error bars represent the range of standard deviation.

Chemical and Physical Characteristics
Oxidization of alginate into ADA by periodate was confirmed.The degree of oxidation was calculated to be 11-12%, in consistent with the result from E. Karakaya et al. [38] The influence of NFC on the shift in covalent vibrational modes of ADA-GEL-NFC was determined by ATR-FTIR as shown in Figure 1B.
The spectra demonstrate gradual change in characteristic bands from pure ADA-GEL to pronounced NFC characteristic, especially at peaks around 1000-1100 cm À1 .The higher amount of NFC was observed to be correlated to an additional apex at 1055 cm À1 (strong C─O stretching), which is a prominent feature of cellulose's C─O bond at glycosidic C-6. [39]This peak is surrounded by two characteristic peaks of mTG-cross-linked ADA-GEL, 1022 and 1076 cm À1 , in which the former is an amide band from mTG's isopeptide bond. [40]The presence of C─H at the same C-6 position shifted the band's apex from 1410 to 1417 cm À1 . [41,42]The supplement of a small peak at 1320 cm À1 and the broadening of peaks in the region of 1230-1352 cm À1 depict the overlapping of NFC's C─O stretching band over the ADA's C─O stretching band.Apexes at 1543 and 1620 cm À1 , the characteristic band of imine ν(C═N) in Schiff base bond, were not shifted due to the presence of NFC. [43,44]Apart from the augmentation of the NFC characteristic peak to the spectra, there is no significant shifting of ADA-GEL characteristic bands.This implies that incorporating NFC into ADA-GEL system does not involve chemical interaction between both entities.The morphology of the hydrogel porous structure was influenced dramatically by the presence of NFC.The cross-sectioned images of the hydrogel structure (Figure 1C) prepared by lyophilization reveal that ADA-GEL has a honeycomb-like interior, in which the wall appears to orient in the same direction, and increasing amounts of NFC incorporated reduced such characteristics.AG-N-7030 exhibits the least organized interior structure, with irregular pore shape, while the walls were the thinnest.Images of hydrogels undergoing critical point drying provide information from a different perspective, albeit the persistent correlation between NFC contents and the decreasing pore size.Without NFC, the ADA-GEL interior comprises apparent pores with thick walls made of ADA-GEL matrix.In AG-N 9010, a similar characteristic could be observed, but the size of the pores was smaller while NFC fibers, which appeared to be in a darker shade, were embedded in the brighter color ADA-GEL matrix.In some spots, the fibers protruded from the hydrogel matrix and displayed nanoscaled dendritic morphology.It is ambiguous to distinguish between NFC fibers and the ADA-GEL matrix in the images of AG-N-8020 and AG-N-7030, though the porous structure was still visible in AG-N-8020 not in AG-N-7030.A notable impact of high-content NFC is that the characteristic of reinforced NFC fibers in the ADA-GEL hydrogel matrix, as appeared in AG-N-9010, could not be observed in AG-N-7030.Instead, the whole hydrogel system emerged as bundle of fibers bonded together into a construct.

Hydrogel Stability
The influence of NFC on weight stability of constructs is present in Figure 1A.The percentage weight change of each group became significantly different at 24 h.Compared to ADA-GEL-NFC, ADA-GEL weight change was drastic by 24 h while all ADA-GEL-NFC weights increased gradually up to 3 days.This implied that NFC reduced the rate of water absorption in the time frame of 24 h.After 48 h, the weight change of pure ADA-GEL decreased to þ66.26%, while AG-N-9010 sustained the increasing trend.AG-N-8020 and AG-N-7030 displayed a similar weight change at 24-48 h, suggesting the water absorption has reached the plateau.In the period between the 2nd day and the 7th day, ADA-GEL, AG-N-8020, and AG-N-7030 showed a similar reduction of swelling.On the contrary, AG-N-9010 maintained the swollen weight at approximately þ70% until the 7th day.AG-N-7030 reached the plateau of %wt earliest on the 14th day, and the weight slowly dropped from þ8.3% to þ5.8% from the 14th day to 28th day.No negative weight change was observed during the 28 days of this study.

Mechanical Properties
Compression tests by using the UTM machine equipped with 100 N load on cross-linked hydrogel constructs at room temperature displayed a positive correlation between the compressive effective modulus and the increasing content of NFC up to 25% (w/w) of ADA-GEL (AG-N-8020) (Figure 2A).However, the trend reverted for AG-N-7030, where the mean value dropped below that of ADA-GEL.Such mechanical behavior could also be observed at the micro level, where the nanoindentation tests were performed on the hydrogel constructs after cross-linking at room temperature.On the contrary, the effective modulus measured using the UTM after 24 h of incubation in cells growth medium at 37 °C exhibited the opposite trend, in which the modulus values decreased with higher content of NFC.In addition, all the hydrogel constructs containing NFC demonstrated lower moduli than before incubation, in contrast to pure ADA-GEL, for which the effective modulus increased from 12.5 kPa before the incubation to %15 kPa after incubation.AG-N-8020 exhibited the most drastic decline in effective modulus, which decreased by almost 80% of its initial value (Figure 2B).

Biocompatibility and Chondrogenic Capability
All hydrogel groups were tested to be noncytotoxic at 24 h after incubation, with viability over 70% (Figure 3).Absorbance at 450 nm depicts comparable cellular metabolic activity in all groups at days 3, 7, and 14, with the mean value increasing from 0.14 to 0.18 at 3 days to 0.58-0.60 at 14 days.One noteworthy pattern observed is that the absorbance of cells proliferated on TCP reverted at 14 days, incongruent with absorbance value from all hydrogel groups, which read to be approximately twofold of TCP.The proliferation of ATDC5 was visualized by fluorescence imaging of samples stained by Calcein AM and DAPI.The bright green spots represent the cell bodies, and nuclei were counterstained in blue.Cells were observed to spread entirely at 7 days, in which NFC content influenced the density of cells proliferated proportionally.Compared to cells proliferating in the constructs on day 3, which possess dendritic characteristics, cells at 14 days were notably smaller and appeared to have round morphology.Actin staining by Rhodamine Phalloidin presented the development of cell spreading and adhesion to the hydrogel structure throughout 14 days.While actins of ATDC5 appeared to be definitive to individual cells on day 3, actin appeared to be stretching to attach to the surrounding on day 7 and was found to be fibrous all over the hydrogel constructs on day 14.This implies that cells were confluent and had occupied the hydrogel matrices entirely by 2 weeks.XRD patterns of hydrogel constructs submerged in SBF for 14 days displayed peaks strongly matching with the 2θ pattern of Halite only.None of the hydrogels induced bone-like apatite formation in their matrices.
Comparative assessment of NFC influence on chondrogenic phenotype as performed by histological staining at 14 days after incubation with ATDC5 suggested a correlation between NFC content and GAGs secretion.Microscopic images of Alcian blue stained samples showed small blue spots sparsely dispersed in the hydrogel matrix of ADA-GEL; the density of blue dots increased with the portion of NFC.AG-N-8020 and AG-N-7030 displayed patches of blue spreading in the visual field, while the blue patch in AG-N-7030 has a notably more extensive area.Cells cultured in TCP presented only one blue spot, confirming that without a hydrogel matrix, there are negligible moieties stained by Alcian blue.Safranin O staining reassured the same trend of GAGs secretion.[47][48] As the staining solution was filtered by a 0.22 μm filter, it was confident that the red specks were not left-over insoluble Safranin O powder.ADA-GEL's staining showed a sparse distribution of small red spots, corresponding to the same characteristic as in Alcian blue stained ADA-GEL samples.In the same manner, AG-N-7030 displayed  some patches of red, with more red area covering the visual field.Samples stained for Col II were incubated with ATDC5 for 21 days.Figure 4, third row, presents fluorescence images of samples with orange dots representing Col II and Figure 4, fourth row, depicts the same images in inverted color to assist in visualization.Quantitative data from counting pixels confirms more significant amount of Col II was tagged in samples with higher NFC content.However, the structure of stained Col II appeared to be relatively small globular shape than fibrous as in native cartilage tissue.

Mechanical Behavior of Biphasic Constructs
The effect of groove dimensions on the PEEK surface on compressive stress relaxation and cyclic compression-tension responses of the biphasic constructs is depicted in Figure 5. PK-H-0505 shows similar stress relaxation behavior as pure hydrogel (AG-N-8020) reaching the relaxed state approximately after the same time.On the contrary, PK-H-0510, PK-H-1005, and PK-H-1010 exhibit significantly slower relaxation.All biphasic constructs displayed pronounced nonlinearity in cyclic compression-tension curves, while AG-N-8020 hydrogels only showed slight nonlinearity (Figure 5E).PK-H-0505 and PK-0510 showed a similar behavior with comparable nominal stresses.In contrast, PK-H-1005 and PK-H-1010 showed a characteristic S-shape curve and higher nominal stresses.PK-H-1005 exhibited the highest conditioning effect.Taken together, the groove depth seems to be a determinant factor for the cyclic compression-tension behavior of the plugs.Shear relaxation curves in Figure 6 indicate that increasing grooves width leads to less pronounced shear stress relaxation.Similar to the stress relaxation in compression, pure hydrogels also exhibit the fastest relaxation in shear.Cyclic torsional shear data also demonstrate differences in the behavior for different PEEK's surface grooves dimensions.Biphasic constructs with a groove depth of 0.5 mm (PK-H-0505 and PK-H-0510) showed similar behavior with lower maximum shear stresses at both 15% and 30% strain.In contrast, samples with a groove depth of 1.0 mm (PK-H-1005 and PK-H-1010) show significantly higher maximum stresses that also depend on the groove width.PK-H-1005 displayed a similar curve to AG-N-8020 at 30% strain with slightly more pronounced nonlinearity.PK-H-1010 overall exhibits the highest maximum shear stresses.While the curve is almost linear for lower strains up to 15%, it displays a strong S-shape at higher strains up to 30%.

Composite Behavior of NFC Incorporated in ADA-GEL and Effects on Stiffness
We have demonstrated in this work that incorporation of NFC into ADA-GEL hydrogel, cross-linked with CaCl 2 and mTG, did not involve chemical interaction between NFC and ADA-GEL, thus creating a composite system in which NFC took a role as reinforced fibers in the ADA-GEL matrix.As confirmed by FTIR spectra, no other chemical bond was detected, nor the change in Schiff base bond.Such characteristics could also be observed from SEM images (Figure 7c) where NFC fibers were embedded in ADA-GEL as separated entities.NFC certainly affected the swelling of hydrogel constructs in vitro, but it was not explicit whether NFC's water retention or gelatin release contributed more to this change.[51] On the other hand, few studies on the ADA-GEL system revealed significant gelatin released within the first 3 days of incubation. [17,34,52]Considering both factors, it can be assumed that NFC played a crucial role in accelerating noncross-linked gelatin to be released from the hydrogel matrix by pulling in the water, thus allowing gelatin to diffuse.As displayed in the SEM images, high NFC content, as in AG-N-7030, might be prone to losing its mass more rapidly than the other group as there is less of Schiff base bond to stabilize the matrix.
The increase of stiffness as the content of NFC rises also confirmed the nature of fibers reinforcement composite as reported in literature.Fu et al. constructed a model predicting modulus of short-fiber-reinforced polymers and indicated that the higher volume fraction increases modulus regardless of reinforced fibers' orientation. [53]Congruent to another model constructed by Facca et al., which predicted Young's modulus of a composite system composed of natural fiber reinforced in thermoplastics, [54] though it should be noted that both models were constructed based on solidified composite matrices.Other works with evidence assuring the influence of NFC volume fraction on hydrogel's Young's modulus could also be found.Borges et al. studied NFC reinforcement in Tween 20 trimethacrylate (T3) and demonstrated that the stiffness of their hydrogel system nonlinearly increased by higher NFC content. [55]A hydrogel system consisting of alginate and NFC in a study from Aarstad et al. exhibited a corresponding quality; Young's modulus of hydrogel constructed was found to be greater with the higher portion of NFC. [56]The AG-N-7030 composite shows a notable decrease in effective stiffness, which was even lower than the value of the ADA-GEL matrix.This underscores a common trait in composite materials, where an increase in strength tends to level off at a specific fiber volume fraction.In the case of AG-N-7030, the composite contains a high amount of NFC at 45% w/w within the ADA-GEL matrix.This excessive NFC content compromises load distribution among fibers.Furthermore, the reduced ADA-GEL matrix facilitates more fiber-fiber than fiber-matrix contact, leading to less effective fiber-matrix interaction, microstructural defects, and a subsequent decline in elastic constant.Thus, while the initial stiffness may improve, there is a limit in the NFC content that can be added to ADA-GEL to benefit the mechanical property of the composite.Furthermore, given the lack of chemical bonding between NFC and the ADA-GEL matrix, the diminished volume of the ADA-GEL matrix in AG-N-7030 likely results in reduced Schiff base bonds and calcium ion cross-links.Thus, the reduction in these primary mechanisms of cross-linking can significantly deteriorate further the strength and stiffness of the hydrogel.
However, an opposite trend was observed in samples incubated for 24 h at 37 °C in this study.This phenomenon may be attributed to several factors related to the thermoreversible property of gelatin.Gelatin exhibits a gelation temperature lower than 37 °C; thus, at physiological temperature, it loses its gel-like character and consequently its structural integrity, leading to a weaker hydrogel structure.Additionally, as the ADA-GEL matrix transitions to a less viscous state at 37 °C, the load distribution between the reinforced fibers (NFC) and the matrix becomes less effective, further contributing to the weakening of the hydrogel.Uncross-linked gelatin molecules can also be released into the environment at this temperature, a phenomenon previously observed by Distler et al. [17] and Schwarz et al. [18] This further weakens the structure of the hydrogel by diminishing its cohesion.Furthermore, as the NFC content in the hydrogel increases, the volume of ADA-GEL correspondingly decreases, leading to reduced structural support from three critical cross-linking mechanisms, as follows: 1) Schiff base bonds, which are linkages between the ε-amino groups of gelatin and aldehyde groups of ADA; 2) calcium ion cross-links in ADA, which stabilize the network; and 3) isopeptide bonds, formed between glutamine and lysine peptide units in gelatin.Together, these factors can explain the observed decline in mechanical properties at 37 °C.At room temperature, incorporation of NFC as fiber improved effective modulus of ADA-GEL up to twofold, which set the value closer to unconfined compressive modulus of human femoral cartilages (0.25 AE 0.15 MPa for superficial zone and 0.37 AE 0.18 MP for middle zone). [57]Nevertheless, further modification of the chemical bond between polymer chains is pivotal to maintaining ADA-GEL-NFC higher stiffness in biological settings.One possibility is to deploy surface-modified nanofibrillated cellulose.As explored in the literature, TEMPO-oxidized NFC offers a functional moiety for NFC to be bound by Ca 2þ , thus cross-linking with the alginate chain. [50,58,59]Another possibility is to strengthen the gelatin cross-link.Although increasing the amount of mTG can also enhance gelatin retention, a result from Distler et al. suggested that mTG content as high as 10% w/v of the cross-linking solution might suppress chondrogenic cell proliferation [17] In addition to our efforts to improve the stiffness to that of native cartilage through the investigation of alternate gelatin cross-linkers, we must also consider the impact on stress relaxation.Although stiffness and stress relaxation are different material properties, changes to increase stiffness may result in greater stress relaxation inadvertently.This, in turn, could have a negative effect on cellular proliferation and ECM secretion. [60,61]Consequently, it is essential to find an optimal balance between increasing stiffness and an adequate cellular response when exploring the different gelatin cross-linkers to improve the fidelity and degradability of the ADA-GEL-NFC hydrogel system.Figure 6.A,C,E,G,I) Shear stress-strain curves of osteochondral plugs at 15% strain (n = 6), B,C,F,H,J) shear stress-strain curve of osteochondral plugs at 30% strain (n = 6), K) image of measurement setup: at the upper surface, the hydrogel is attached to the geometry, and at the lower surface, PEEK is attached to the plate, and L) shear stress relaxation during 250 s of testing (n = 6).

Effect of NFC Reinforcement on Biocompatibility and the Chondrogenic Markers
The result from the WST assay indicates that NFC did not drastically affect ATDC5 proliferation.From the fluorescence microscope of samples 14 days after incubation, cell morphology in all groups appeared to be rounder and smaller, characteristics of chondrocytes in the superficial and middle zone.SEM images from Figure 1 revealed that the pores in all groups were partially filled after 7 days.Although it is ambiguous whether the aggregates were cell bodies or ECM proteins, it proved that cells migrated more profoundly into the hydrogel constructs, confirming the capability of the ADA-GEL and ADA-GEL-NFC systems as 3D scaffolds for tissue regeneration.Due to gelatin's arginineglycine-aspartic acid (RGD) units, a peptide sequence renowned for its role as a cell attach site in ECM, [62][63][64] cellular adhesion was inherently ensured in all groups.Yet, in the groups where the portion of ADA-GEL is less, the degree of proliferation was comparatively equivalent to pristine ADA-GEL, indicating that NFC also provides attachment motifs to cells.Nevertheless, the degree of proliferation does not dictate the material's potential to induce chondrogenesis.In this work, GAGs and Col II, considered chondrogenic markers, [65][66][67][68] were found to increase proportionally to the amount of NFC after 14 days of incubation.Samples stained by Alcian blue and Safranin O displayed the same tendency in which the stained entities increased with higher NFC content in hydrogel. [69]Atsumi et al., who successfully isolated the ATDC5 cell line from mouse teratocarcinoma, had established that without chondrogenesis promoter, the absorbance of Alcian blue was negligible at the 14th day of incubation whereas, with insulin as chondrogenesis promoter, cells appeared significantly stained at day 14. [70]A study from Shukunami et al., in which ATDC5 cells were maintained in a complete medium comparable to this study, also demonstrated that cells supplemented with insulin were stained by Alcian blue around the 15th day. [71]For Col II, protein secretion occurred in the later stage.Another study by Shukunami et al. addressed that significant gene expression of Col II was detected at day 21 of incubation in a medium supplemented with insulin. [72]erein, after 21 days of incubation, Col II was found to increase proportionally to the content of NFC.Substantially, incorporating NFC into ADA-GEL has the potency to induce cartilage differentiation without any supplementation.There is no significant evidence that pristine ADA-GEL stimulates GAG and Col II synthesis in chondrocytes.A study on ADA-GEL beads cross-linked with CaCl 2 and mTG displayed expression of Col II and proteoglycans after 14 days of incubation; however, the cells were supplemented with ascorbic acid, [18] which has been identified as a chondrogenesis inductor. [73,74]It is unlikely that either alginate or gelatin can mediate cartilage differentiation independently.Most studies on gelatin-based hydrogels for cartilage regeneration that have shown good cartilage phenotype expression involve the addition of growth factors or inductive differentiation substances.Although gelatin is a subunit of collagen, a naturally occurring cartilage protein that facilitates differentiation in chondroprogenitor cells, Fan et al. discovered that MSCs cultured on their pure gelatin pellet secreted negligible GAGs and Col II.Similarly, chondrocytes cultured in pure alginate hydrogel without a differentiation inducer released little to no GAGs and Col II. [75]The expression of early chondrogenic markers in ADA-GEL-NFC is undoubtedly attributed to cellular interaction with NFC; nevertheless, the effect of NFC on chondrogenesis is poorly understood.][78] An in vitro study on NFC-alginate bioink cultured with human-derived induced pluripotent stem cells (iPSCs) from Nguyen et al. indicated that their bioink with NFC/alginate of 80/20 expressed ACAN gene more than NFC/ alginate of 60/40 at week 3.This implies NFC promoted aggrecan accumulation, consistent with this study's outcome.[84][85] Further investigation on the interaction of NFC negative charges to cells and chondrogenic proteins is beneficial to deploying NFC composite hydrogel in cartilage applications.
Given that the hydroxyapatite (HA) pattern was absent in the XRD results from all groups, it can be concluded that the ADA-GEL system has a low tendency to precipitation of HA.This is particularly desirable in the aspect of articular cartilage surface because the apatite initiates hypertrophic differentiation.Hypertrophic chondrocytes are domestic in the calcified zone, the basal layer of cartilage, anchoring to the subchondral bone.Chondrocytes in this layer are larger and express substantially different phenotypes compared to the middle and superficial zone. [86,87][90] As the ultimate application of this composite hydrogel system is for cartilage generation in OA, the lack of apatite formation in the matrix is thus favorable.

Potential Application as Biphasic Osteochondral Plug
We designed and evaluated the mechanical properties of biphasic osteochondral plugs comprised of ADA-GEL-NFC as an articular cartilage scaffold and PEEK as a subchondral bone substitute.
The proposed device would provide an alternative treatment option for OCDs in which the articular cartilage was extensively injured and a subchondral lesion had occurred. [91,92]Aside from the biocompatibility of materials on both phases of the device, the mechanical response is critical to achieving the desired reaction from tissue surrounding the implant site because inappropriate mechanical properties of the osteochondral plug can lead to further degeneration of the adjacent healthy cartilage. [93,94]We demonstrated that the interface between the soft and hard parts influences the device's overall mechanical behavior.Data from cyclic compression-tension experiments suggested that shallow grooves (0.5 mm depth) on the PEEK surface had an insignificant effect on axial stress-strain behavior as the stress-strain curves of groups PK-H-0505 and PK-H-0510 resembled that of AG-N-8020.However, regardless of the width of the grooves, deeper grooves enhance the asymmetric characteristics of this compression-tension response and thus the difference between maximum stresses in compression and tension.We may conclude that the groove depth should be large enough to influence the mechanical properties of the plugs in compression and tension.Weizel et al. found concaved down curves skewed to the left in an experiment studying cartilage response to cyclic nominal stress at 15% strain with 500 μm s À1 loading speed, [95] similar to the curve from Figure 5A-D in the range of axial compressive loading.[98] A similar hysteresis shape was detected in biphasic osteochondral plugs, particularly in PK-H-1005 and PK-H-1010, implying that deeper groove dimensions yielded more robust strain-stiffening behavior (Figure 5).Similar properties were not observed in pure ADA-GEL samples, where the hysteresis loops appeared to be almost linear.This could be attributed to the fact that the osteochondral plugs imitate the biphasic feature of cartilage's zonal structure, which is hydrogel-like in the middle zone and solidified in the deep area.The grooves on the surface of PEEK may influence energy dissipation during shear loading in the same manner as the columnar arrangement of collagen fibers in the deep zone.Furthermore, because of its capability to induce hydroxyapatite formation on the surface, we hypothesized that the sulfonated PEEK surface would stimulate hypertrophic differentiation of chondrocytes cell, resulting in a gradient of chondrocytes that correspond to cartilage zones. [31]he mechanics of the interface between the cartilage phase and the subchondral phase must be considered while engineering a biphasic osteochondral plug.Such interface mechanics could be tailored for the device as conceived in this work by manipulating the dimension of grooves on the PEEK-hydrogel interface.Additional modification might be done to allow vascularization into the PEEK phase, further enhancing this biphasic osteochondral plug beyond its promising mechanical features.Instead of adopting a solid PEEK substrate as a subchondral bone substitute, porous PEEK may offer better osseointegration to the bone and a diffusion route for nutrients to nourish chondrocytes in the cartilage layer.However, more research on the effects of the sulfonate functional group on modified PEEK on cartilage regeneration is required to ensure that the device serves its objective as an OCD treatment.

Conclusion
The novel composite based on ADA-GEL hydrogel with NFC reinforcement presented in this study demonstrated increased stiffness at room temperature as well as promising chondrogenic inductivity for cartilage regeneration applications.In terms of stability, AG-N structures remained intact throughout 28 days in this study whereas NFC water absorption and retention significantly affected hydrogel swelling in vitro.Thermal stability, however, remains an issue for the composite ADA-GEL hydrogel, as the constructs lost stiffness when incubated at 37 °C due to gelatin's reverse gelation property.To produce sufficient AG-N's stiffness for load bearing tissues, the addition technique to cross-link gelatin is thus crucial.The greater NFC content had no negative effect on cellular response, as cell proliferation and cell adhesion were comparable across all sample groups.The presence of NFC had a remarkable impact on cartilage differentiation.Without any chondrogenesis supplement, we found that the amount of GAGs and Col II secreted from ATDC5 was proportional to NFC content.We also presented a feasible orthopedic device made from the hydrogel by combining it with a PEEK substrate as a biphasic osteochondral plug.We evaluated the mechanical properties of the devices and confirmed that the interface between the two phases had a considerable impact on their overall mechanical behavior.Although we adopted solid PEEK substrates in the conceptualization of the osteochondral plug, porous substrates may serve the purpose better as the plug's anchor site.Additional in vivo examination of the whole device is required to determine its practicality.Overall, AG-N has the potential as a cartilage scaffold, and when paired with a bone replacement material or scaffold, it can be used as a biphasic osteochondral device for the treatment of OCDs.

Figure 1 .
Figure 1.A) Weight change of hydrogel constructs when incubated in cell culture medium at 37 °C over 28 days.B) FTIR spectra of ADA-GEL, AG-N-9010, AG-N-8020, AG-N-9010 hydrogels after cross-linking with 0.1 M CaCl 2 and 5% w/v mTG comparing to dry NFC powder.C) SEM images of immaculate hydrogels constructs after processed by freeze drying and critical point drying, together with critical point dried hydrogels incubated with ATDC5 cells for 7 days on the lowest row.

Figure 2 .
Figure 2. A-C) Compressive effective modulus of the hydrogel as investigated by UTM at room temperature (preincubated), at 37 °C after incubation for 24 h, and by nanoindenter at room temperature (preincubated), respectively.(*) indicates statistic significant level p < 0.05.D) XRD patterns of freeze-dried hydrogel after 14 days immersion in SBF.

Figure 3 .
Figure 3. Viability test result with ATDC3 cells after 24 h of direct test according to ISO10993-5 and WST assay result of absorbance at 450 nm after 3, 7, 14 days incubation with ATDC5 cells in comparison with TCPs.

Figure 4 .
Figure 4. Microscopic images represent early chondrogenic marker assessments on GAGs by Alcian blue and Safranin O staining (first two rows).The third row represents fluorescence images of hydrogel with Col II appearing in orange pixels.The fourth row is the invert of the red channel of the third row to assist in pixel counting, and visualization as the magnification power was low to evaluate the prevalence of Col II.The chart represents average pixel counts (n = 3) from inverted images of Col II staining.

Figure 5 .
Figure 5. A-E) Nominal stress-stretch curves of cyclic compression-tension loading (n = 6), F) compressive stress relaxation of each group during 250 s of testing (n = 6), and G) images of osteochondral plugs and schematic illustrating the dimensions of grooves on the PEEK surface.

Figure 8 .
Figure 8.In vitro proliferation and cell adhesion; assessment of ATDC5 on hydrogel constructs after 3, 7, 14 days of incubation; green, blue, and red fluorescence stained are Calcein AM, DAPI, and Rhodamine Phalloidin positive, respectively.

Figure 7 .
Figure 7. a,b) The higher magnification power of hydrogels incubated with ATDC5 for 21 days, green, blue, and orange represent cell cytosol, nuclei, and Col II, respectively.c) SEM images of critical point dried AG-N-9010 construct without cells, emphasizing the area where NFC fibers protrude from the ADA-GEL matrix.This image confirms the NFC dimension, with diameter in the nanometer range, d) ATDC5 cells proliferate on AG-N-7030 constructs after 3 days of incubation.Noting cells morphology conforming with microscopic fluorescence image in Figure 8.

Table 2 .
Dimension of grooves engraved on PEEK's upper surface where the deposition of AG-N-8020 takes place.