Injectable thiol-ene hydrogel of galactoglucomannan and cellulose nanocrystals in delivery of therapeutic inorganic ions with embedded bioactive glass nanoparticles

We propose an injectable nanocomposite hydrogel that is photo-curable via light-induced thiol-ene addition between methacrylate modified O -acetyl-galactoglucomannan (GGMMA) and thiolated cellulose nanocrystal (CNC-SH). Compared to free-radical chain polymerization, the orthogonal step-growth of thiol-ene addition al- lows a less heterogeneous hydrogel network and more rapid crosslinking kinetics. CNC-SH reinforced the GGMMA hydrogel as both a nanofiller and a crosslinker to GGMMA resulting in an interpenetrating network via thiol-ene addition. Importantly, the mechanical stiffness of the GGMMA/CNC-SH hydrogel is mainly determined by the stoichiometric ratio between the thiol groups on CNC-SH and the methacrylate groups in GGMMA. Meanwhile, the bioactive glass nanoparticle (BaGNP)-laden hydrogels of GGMMA/CNC-SH showed a sustained release of therapeutic ions in simulated body fluid in vitro , which extended the bioactive function of hydrogel matrix. Furthermore, the suitability of the GGMMA/CNC-SH formulation as biomaterial resin to fabricate digitally designed hydrogel constructs via digital light processing (DLP) lithography printing was evaluated.


Introduction
Injectable and in situ crosslinkable hydrogels have shown immense promise to function as delivery vehicles of biotherapeutic agents for onsite therapy Cheng et al., 2020;Wu et al., 2019;Wu et al., 2020). In the past decade, injectable hydrogels prepared by natural-origin polymers like gelatin, chitosan, or alginate, have attracted arising attention due to their biocompatibility and outstanding matrix properties mimicking the native extracellular matrix (Bidarra et al., 2014;Malafaya et al., 2007;Nawaz et al., 2021;Wang et al., 2021). In the family of biopolymers derived from lignocellulosic biomasses of large availability, cellulosic nanomaterials and hemicellulose biopolymers both have been widely used as building blocks in constructing hydrogels, highlighting competitive niches such as the chemical versatility with ease of modifications, high water retention properties, and non-cytotoxicity (Hynninen et al., 2018;Markstedt et al., 2017).
In the fabrication of injectable hydrogels, the crosslinking method plays a core role in resulting mechanically strong and robust hydrogels. Physical crosslinking strategies by adopting ionic-, pH-or thermalstimulus responsive polymers, as well as chemical crosslinking strategies including Schiff's base formation, Michael addition, enzymatic crosslinking, or photo-induced polymerization, are commonly engaged depending on the actual application scenarios (Balakrishnan et al., 2014;Hou et al., 2018;Jabeen et al., 2017;Jin et al., 2010;Lin et al., 2015;Park et al., 2014;Zhang et al., 2014). As an external stimulusresponsive fabrication strategy, the photo-induced crosslinking, via a mechanism of either free-radical chain polymerization or orthogonal step-growth of thiol-ene 'click' addition, has been well accepted as an approach with great convenience in fabricating injectable hydrogels thanks to its rapid polymerization kinetics with minimal heat generation (Hu et al., 2012;Liu et al., 2017). In this context, natural polymers of various origins have been chemically modified with such a photo-reactive moiety to facilitate their derivatives as photo-crosslinkable biomaterials, e.g., gelatin, hyaluronic acid, sericin, or ulvan methacryloyl (Le et al., 2018;Ning et al., 2019;Qi et al., 2018). In the category of biomass-derived hemicelluloses, we have earlier reported a facile synthesis of methacrylated galactoglucomannan (GGMMA) that showed great UV-crosslinking ability through free-radical chain polymerization, as well as the formulation of cellulose nanofiber with GGMMA as an auxiliary biopolymer for curing the hydrogels in lightassisted, hydrogel-extrusion 3D printing of the as-formulated biomaterial inks (Xu et al., 2019). Due to its reaction kinetics, free-radical chain polymerization is challenged by oxygen inhibition and in general would lead to heterogeneity of local network structures of GGMMA, resulting in a mismatch between bulk and local mechanical property of the hydrogel (Ligon et al., 2013;Lim et al., 2016Lim et al., , 2020Seiffert, 2017b;Sunyer et al., 2012). In this perspective, the photo-induced thiol-ene 'click' chemistry outperforms as it provides advantages such as high conversion rate and selectivity, less oxygen inhibition, and formation of homogeneous hydrogel networks with manipulating mechanical properties (Hoyle and Bowman, 2010;Seiffert, 2017b;Yilmaz and Yagci, 2020).
Therefore, a thiolated crosslinker containing thiol moieties is needed to form a homogenous network with GGMMA through thiol-ene addition. Cellulose nanocrystals (CNC) are unique rod-like cellulosic nanomaterials that have received significant interest due to their mechanical, optical, chemical, and rheological properties (Eyley and Thielemans, 2014;Thomas et al., 2018). Previously, the synthesis of thiolated CNC (CNC-SH) was reported via the engraftment of L-cysteine to oxidized CNC (CNC-CHO) by reductive amination (Ruan et al., 2016). In addition, CNC has been popularly exploited as a high-performance reinforcement nanofiller for interpenetrating the polymer networks (De France et al., 2016;Domingues et al., 2014;Hynninen et al., 2018). Inspired by these peer studies, we proposed the fabrication of an injectable hydrogel with GGMMA and CNC-SH as building blocks through thiol-ene addition with the advantages of rapid photo-crosslinking kinetic and homogenous hydrogel structure. Within this initiative to develop all-polysaccharide nanocomposite hydrogels, the CNC-SH would function as both a crosslinker and a reinforcing nanofiller in the polymer network of GGMMA. By adjusting the stoichiometric ratio between thiol and ene moieties in respective CNC-SH and GGMMA, the mechanical properties of the hydrogels would be greatly adjusted. Meanwhile, the injectable GGMMA+CNC-SH hydrogel is applicable to establish the localized and sustained release of the therapeutic agents in situ. To extend the biofunctionality of the GGMMA+CNC-SH hydrogel, bioactive glass nanoparticles (BaGNP) were further encapsulated in the injectable hydrogel formulation to investigate the release of therapeutic ions of Si, Ca, or Cu through the hydrogel matrix. In addition, the formulated photocrosslinkable GGMMA+CNC-SH inks could also meet the requirements for the digital light processing (DLP) 3D printing of hydrogel. The feasibility to fabricate the GGMMA+CNC-SH hydrogel via DLP printing offers great potential for future exploiting in various biomedical applications ranging from wound dressing to tissue engineering scaffolds.

Synthesis of CNC-SH, GGMMA and BaGNP
CNC was prepared by H 2 SO 4 (64 wt%) hydrolysis of MCC with a solid/liquid ratio of 1 g/10 mL at 45 • C for 1 h. After dialysis against deionized water (cut-off 12-14 kDa), aldehyde groups were introduced to CNC by NaIO 4 oxidation according to Sun's method (Sun et al., 2015). Briefly, NaIO 4 was added into CNC suspension (0.5 wt%) with a mass ratio of 4:1 (NaIO 4 : CNC). The pH of the suspension was adjusted to 3.5 using acetic acid followed by reaction at 45 • C for 4 h in a dark place. The aldehyde group content of CNC-CHO was determined by titration of the HCl release during the oxime reaction with NH 2 OH⋅HCl following the procedure reported by Alam et al. (Alam and Christopher, 2018). Lcysteine was further grafted onto the CNC-CHO through a reductive amination reaction. In short, L-cysteine and NaBH 3 CN were added into a CNC-CHO suspension (0.58 wt%) with a mass ratio of 8.47: 3.5: 1 (Lcysteine: NaBH 3 CN: CNC-CHO). The pH of the suspension was adjusted to 4.5 using acetic acid followed by reaction at 45 • C for 24 h in a dark place. The resulting L-cysteine grafted CNC-SH was further dialyzed and stored under a nitrogen atmosphere. The as-synthesized CNC-SH was characterized by TEM and its degree of substitution (DS) was further quantitatively agreed with elemental analysis and liquid-state 13 C NMR, as detailed in Supplementary Materials. GGMMA was synthesized by reacting methacrylic anhydride with GGM according to a method reported by Xu et al. (2019). The degree of methacryloylation (DM: 0.9 mmol/g) and molecular weight (Mn = 16 kDa) of the obtained GGMMA was quantified using 1 H NMR and HPLC-SEC, respectively, as displayed in Supplementary Materials. The BaGNP samples (BaGNP with a nominal composition: 70SiO 2 -30CaO-0CuO in mol% and the copper-doped Cu-BaGNP with a nominal composition of 70SiO 2 -25CaO-5CuO in mol %) were synthesized according to a modified protocol reported by Zheng et al. (2017). The as-prepared BaGNP samples were characterized by TEM and SEM-EDXA (EDXA, LEO Gemini 1530 with a Thermo Scientific UltraDry Silicon Drift Detector, X-ray detector by Thermo Scientific).

Formulation of UV crosslinkable GGMMA/CNC based hydrogel precursors and hydrogel fabrication
The UV crosslinkable hydrogel precursors were prepared by dissolving GGMMA (2 wt%) and photoinitiator (0.1-0.5 wt% Irgacure 2959 or 0.25 wt% LAP) into the CNC suspensions (1 and 2 wt% CNC-CHO or 1, 2, and 3 wt% CNC-SH). The hydrogel precursors were thoroughly mixed by a vortex for 5 min. The hydrogel discs were fabricated through photopolymerization by transferring hydrogel precursors into transparent cylindrical moulds (diameter: 8 mm and height: 4.6 mm) and curing by a UV-LED (120 mW cm −2 , 365 nm, bluepoint LED eco, The Hönle Group) for 300 s. The fabricated GGMMA/CNC hydrogels were kept in PBS buffer prior to testing (Scheme 1).

Rheological behaviors of the formulated hydrogel precursors
The rheological profiles of hydrogel precursors of GGMMA+CNC-SH were registered by an Anton Paar Physica MCR 702 rheometer (Anton Parr GmbH) using a plate-plate geometry (25 mm diameter) with a gap distance of 0.5 mm (a coaxial double gap geometry DG26.7 was used to measure the 2% GGMMA solution) at 25 • C. The viscosity curves of the hydrogel precursors were recorded by shear flow measurement with a shear rate of 0.1 to 1000 s −1 with 1 s per data point. Oscillatory amplitude sweep was performed under a strain range from 0.1 to 500% with a constant frequency of 1 Hz. Photo-rheology profiles were measured under oscillation mode with a gap distance of 0.2 mm at a constant oscillatory strain and frequency of 0.1% and 1 Hz, respectively. The tested samples were irradiated upon a light source (365 nm or 405 nm) starting at 60 s of the measurements. The change in storage modulus was recorded. The measurements were carried out in triplicate.

Mechanical properties of the GGMMA+CNC-SH hydrogels
Compression measurements of hydrogel discs were performed by a universal tester Instron 4204 (Instron). Young's moduli of the GGMMA+CNC-SH hydrogels were calculated based on extrapolating and linear fitting of the elastic region of the stress-strain curves (Xu et al., 2019). 3 hydrogel discs of each formulation were prepared for the compression measurement. Statistical analysis was performed using the GraphPad Prism 9 software by a one-way ANOVA analysis. A Tukey test with significance level of 0.05 was apllied for the analysis.

In vitro enzymatic degradation study
In vitro enzymatic degradation of the fabricated hydrogels was performed in an air bath shaker (Boekel Scientific) at 37 • C. Briefly, hydrogels (40 mg) of different compositions (2% GGMMA, 2% GGMMA+1% CNC-CHO, and 2% GGMMA+1/2/3% CNC-SH) were immersed in a digestion liquid containing a mixture of 475 μL of PBS buffer and 25 μL of endo-1,4-β-Mannanase (5000 U/mL) in sealed bottles. The bottles were taken out at time points of 0.5, 1, 2, 3, 5, and 7 days and boiling for 10 min to deactivate the enzyme. The soluble carbohydrate content of the supernatant was analyzed to indicate the degradation of GGMMA (Sundberg et al., 1996). 3*6 parallel samples of each hydrogel formulation were prepared for the experiments, and 3 parallel samples were analyzed for soluble carbohydrate content at each time point.

Therapeutic ion dissolution from the BaGNP-laden GGMMA+CNC-SH hydrogel
Both BaGNP and Cu-BaGNP in weight percentages of 0.4, 1, and 2 were doped into the hydrogel precursors of 2% GGMMA+2% CNC-SH, respectively. The BaGNP-laden GGMMA+CNC-SH hydrogels were fabricated using the same protocol as described in the above section.
For the ion dissolution test, BaGNP-laden hydrogels (225 mg) were immersed in SBF (15 mL) in airtight polyethylene containers followed by placing in an incubating orbital shaker at 37 • C with agitation at 100 rpm. The samples were incubated for a total period of 7 or 14 days and 1 mL of immersion solution was sampled at 1 d (day), 3 d, 5 d, 7 d, 11 d and 14.5 d. Afterwards, 1 mL fresh SBF was replenished for consecutive immersion. The ionic concentrations of Ca and Si ions in the sampled solution were analyzed with an inductively coupled plasma optical emission spectrometer (ICP-OES) (Optima 5300 DV, Perkin Elmer, Shelton, CT). At the end of the immersion test, the hydrogels were collected from the SBF, washed extensively with deionized water, frozen in liquid nitrogen and eventually lyophilized to obtain the corresponding cryogels. The surface morphology and elemental analysis of the scaffold were characterized with SEM-EDXA. All experiments were carried out in triplicate.

DLP printing of honeycomb structure hydrogels with the BaGNPladen hydrogel precursor of 2% GGMMA+1% CNC-SH
The honeycomb structure of hydrogels was demonstrated by a DLP 3D printer (M-One Pro 30, wavelength of 405 nm) equipped with a digital micromirror device (resolution: 1920 × 1080). The CAD model of the honeycomb construct was designed by Fusion 360 software and transfer into the digital pattern by the XMaker V2.7.1 software. Hydrogel precursors of 1 wt% CNC-SH, 2 wt% of GGMMA and 0.25 wt% LAP with or without 0.4 wt% 5Cu-BaGNP and 0.4 mM tartrazine were loaded onto the printing bed and fabricated into a hexagonal structure under light exposure. The layer height of the printer construct was set at 35 μm. Scheme 1. Illustration of the hydrogel fabrication. CNC was isolated from the MCC and oxidized to introduce aldehyde followed by reductive amination to graft SH moiety; GGM was isolated from the tree and esterification was performed to introduce MA moiety. Hydrogel was obtained through light-induced thiol-ene addition.

Synthesis and characterizations on CNC-SH and GGMMA
The CNC-SH was synthesized via the route illustrated in Fig. 1a. After being surface modified with pendant cysteine groups, CNC-SH maintained the rod-like nanomorphology with an average length of 145 nm and diameter of 5 nm, as observed in the TEM image. The chemical modification in the molecular structure of the CNC was quantitatively determined by the liquid-state quantitative 13 C NMR (King et al., 2018). As shown in Fig. 1b, signals of anomeric carbon (C1, 101 to 105 ppm) and C2 to C6 (60 to 82 ppm) of CNC samples can be attributed to the featured signals of cellulose. After the NaIO 4 oxidation, the bond between C2 and C3 was selectively cleaved in formation of dialdehyde (C2 ′ and C3 ′ at 165 to 167 ppm) in sites. Meanwhile, the signals of C2 ′′ and C3 ′′ were detected at 93 and 98 ppm, respectively, attributed to the hemiacetal formation of the aldehyde group (Amer et al., 2016;Münster et al., 2017;Nypelö et al., 2021). The DS of the aldehyde group (DS = 0.26) in CNC-CHO was computed by the comparison of sum signal integration of C2 ′ , C3 ′ , C2 ′′ , and C3 ′′ to C1 as displayed in Fig. 1b, which is in line with the DS of 1.45 mmol/g calculated by titration. As shown in Fig. 1b, the grafting of L-cysteine to CNC-CHO was confirmed by the appearance of a new signal at 172 to 173 ppm, which is attributed to the carboxyl carbon (C9) of L-cysteine. The DS of L-cysteine (DS = 0.20) in CNC-SH was determined by the integral comparison of signal of the C9 to that of C1 and C9 as displayed in Fig. 1b, which is in line with the DS value (DS = 1.04 mmol/g) that is calculated from the nitrogen content in CNC-SH by elemental analysis as shown in Table S1 and S2. It is noted that the DS of L-cysteine appears smaller than the DS of aldehyde in CNC-CHO. The signals attributed to the dialdehyde completely disappeared in the 13 C NMR spectra of CNC-SH, as the remaining aldehydes were further reduced into the hydroxyls by NaBH 3 CN in the reductive amination. Compared with the solid-state 13 C NMR analysis on the Lcysteine grafted CNC carried by Li et al., the employment of a mixture of ionic liquid tetrabutylphosphonium acetate ([P4444][OAc]) and DMSO-d 6 (1:4 w/w) as in liquid-status 13 C NMR facilitates the quantitative analysis to the chemical modifications that are induced in the CNC samples . This is critical information to register for precise formulation control in realizing on/off stoichiometric thiol-ene chemistry between CNC-SH and GGMMA. The chemical structure and 1 H and 13 C NMR spectra of GGMMA are presented in Fig. 1c. The GGMMA with such a DM (0.9 mmol/g) was chosen to fabricate hydrogels with CNC-SH through thiol-ene chemistry, taking into balance between the decent DM and good solubility of the biopolymer under consideration.

Rheological properties and photo-crosslinking kinetics of hydrogel precursors of GGMMA+CNC-SH and mechanical property of photocured hydrogels
In making high-performance injectable hydrogels, the hydrogel precursors of GGMMA+CNC-SH are critically expected to show outstanding injectability (shear-thinning behavior) and rapid crosslinking kinetics. The rheological properties and photo-crosslinking kinetics of the hydrogel precursors of GGMMA+CNC-SH were promptly assessed. As a soluble polysaccharide, 2% GGMMA solution showed a low viscosity and behaved like a Newtonian liquid at high shear rates, as shown in Fig. 2a. The addition of CNC-SH into 2% GGMMA solution drastically increased the viscosity of the resulted hydrogel precursors. The viscosity also increased with the increase of the CNC-SH concentration. Meanwhile, in comparison to the pristine CNC-SH solution, the incorporation of GGMMA increased the zero-shear viscosity of the hydrogel precursors of GGMMA+CNC-SH. All the hydrogel precursors presented a characteristic shear-thinning behavior exhibiting a viscosity reduction as they flow upon shear. Viscoelastic properties of the GGMMA+CNC-SH hydrogel precursors were analyzed through amplitude sweep, the storage and loss modulus (G ′ and G ′′ ) verse shear stress were piloted as displayed in Fig. 2b. The hydrogel precursors showed rather weak viscoelasity with the low G ′ and the flow stress (shear stress at the crossover point of G ′ and G ′′ ) were lower than 10 Pa. The G ′ and flow stress of the hydrogel precursors increased with the increase of CNC-SH concentration. The low viscosity and viscoelasticity indicated a good flowability of the hydrogel precursors, facilitating their extrusion process for injectable hydrogel or recoating process in DLP printing (Bertsch et al., 2019;Lim et al., 2020).
Nevertheless, the injectable hydrogel is required to instantly form gel conforming to the desired geometry at the application site and to supply mechanical support after injection (Bertsch et al., 2019). Sufficiently rapid gelation kinetics are imperative for gelation of the hydrogel precursors. Here, the GGMMA+CNC-SH based hydrogel precursors were subjected to photo-initiated crosslinking via the high-efficacy thiol-ene chemistry, which is expected to present a faster crosslinking speed and to result in a more homogenous microscopic structure within the gel, in comparison to the free-radical chain polymerization that is the case for the photo-initiated crosslinking of GGMMA . The crosslinking kinetics of the formulations was assessed using photorheology. As shown in Fig. 2c, the crosslinking of all the hydrogel precursors of GGMMA+CNC-SH was initiated immediately upon UV irradiation as indicated by the dramatic increase in G ′ . Meanwhile, the hydrogel precursors of GGMMA+CNC-SH showed a rapid crosslinking kinetics with the G ′ value levelling off within 30 s to reach the maximum value of G ′ max . In contrast to the G ′ max of GGMMA after crosslinking, the G ′ max of the hydrogel precursors of GGMMA+CHC-CHO with different CNC-CHO content further increased. It is noteworthy that the G ′ tends to be even higher when the same content of CNC-CHO was systemically replaced by CNC-SH. It is indicative that CNC-CHO only acts as a reinforcing component instead of contributing crosslinking efforts (Sampath et al., 2017). The result is consistent with the previous observation that G ′ of hydrogels reinforced with chemically bound CNC was higher than neat CNC reinforced hydrogels at the same loading (Yang et al., 2013). This result is likely attributed to CNC-SH being both physically entrapped within and chemically bound to the hydrogel network, thus serving as a reinforcing agent and a crosslinker. The dosage of photoinitiator plays a vital role in crosslinking kinetics, where the G ′ of 2% GGMMA+2% CNC-SH hydrogel increased more rapidly with the increase of the Irgacure 2959 concentration, as shown in Fig. S1.
After gelation, the hydrogel is expected to provide adequate mechanical support and robustness as demanded at the specific application site. The compressive Young's moduli of the hydrogel discs are displayed in Fig. 2d. The rod-like CNC-CHO as an effective nanofiller significantly enhanced the mechanical property of the GGMMA hydrogel, where the compressive strength increased drastically after adding 1% of CNC-CHO. In addition, the physically entrapped and chemically bound CNC-SH exhibited even better mechanical reinforcement performance than CNC-CHO. Compared with the GGMMA+CNC-CHO hydrogels that were photo-polymerized through the free-radical chain polymerization to crosslink the GGMMA, the GGMMA+CNC-SH hydrogels would provide a better spatial network homogeneity within the gel through the orthogonal step-growth mechanism of thiol-ene addition (Grigoryan et al., 2019). The relatively homogeneous network structure shall consequently result in a better match between bulk and local property, and thus influence the mechanical properties (Sunyer et al., 2012). By varying the compositional ratio between GGMMA and CNC-SH, Young's moduli of hydrogels could be tailored in the spectrum of 1.43 to 12.35 kPa. These as-measured stiffness values fall into the range (5-40 kPa) of the mechanical stiffness of hydrogel matrix suitable for cell culture study of different types, including pre-osteoblast, fibroblast, and cardiovascular cells (Nemir and West, 2010). Potentially, the GGMMA+CNC-SH hydrogels could function as a candidate biomaterial system for in vitro cell culture studies. Not surprisingly, the 2% GGMMA+2% CNC-SH hydrogel with an on-stoichiometry molar ratio of MA: SH close to 1:1, showed the highest Young's modulus among all the hydrogels. As a Q. Wang et al. comparison, the 2% GGMMA+3% CNC-SH with a higher MA: SH ratio of 1:1.7 showed higher viscoelasticity (Fig. 2c) with higher CNC-SH loading, but showed a decreased mechanical property no matter of the excess amount of CNC-SH as reinforcement. This could be attributed to an almost equal amount of thiol and methacrylate groups resulting in no excess off-stoichiometry and maximizing polymer mechanical properties (Carlborg et al., 2011).

Mannanase mediated degradation of the GGMMA/CNC hydrogel in PBS buffer
Considering the in stimuli-responsive or controlled therapeutic delivery, it is imperative to regulate the degradation of the construct hydrogel systems. However, human body lacks the enzymes that can degrade lignocellulosic biopolymers. Thus, enzyme immobilization in or secretion to the matrix is typically required. To address this perspective, the enzymatic degradation of GGMMA+CNC-SH hydrogels was evaluated in vitro with the endo-1,4-β-mannanase from Cellvibrio japonicas in PBS buffer. The applied mannanase could actively and randomly hydrolyze (1, 4)-β-D-mannosidic linkages at pH 7.0. Owing to the hygroscopic property and porosity structure of the hydrogels, mass transfer of enzyme is easily undergoing between the hydrogel and the surrounding aqueous media facilitating the degradation process (Gorgieva and Kokol, 2012). The degradation kinetics of GGMMA was revealed by quantifying the soluble carbohydrate contents by gas chromatography, as shown in Fig. 3. Overall, GGMMA/CNC hydrogels presented faster degradation kinetics than the pristine GGMMA hydrogels. The physically entrapped CNC-CHO in 2% GGMMA+1% CNC-CHO hydrogel might sterically block the crosslinking of GGMMA to a certain extent (Yang et al., 2013). It is speculated that the steric spacing of CNC in the matrix would result in a relatively loose structure of GGMMA, which presumably facilitated faster mannanase diffusion in/out the hydrogel system. The hydrolysis kinetics of GGMMA was further enhanced in GGMMA+CNC-SH hydrogel, which might be attributed to the improved local homogeneity resulting from orthogonal step-growth polymerization of thiol-ene addition (Seiffert, 2017a). The on-stoichiometry thiol-  ene hydrogel of 2% GGMMA+2% CNC-SH presented the fastest hydrolysis kinetic, reaching around 29% GGMMA degradation in half-a-day and around 50% of GGMMA after 7 days. Meanwhile, the degradation kinetics of GGMMA in 2% GGMMA+3% CNC-SH hydrogel fell off the track of 2% GGMMA+2% CNC-SH after half-a-day hydrolysis. It is suspected that excess CNC-SH covering the GGMMA surface and blocking the enzyme binding site due to the intrinsic interaction between cellulose and hemicellulose (Lucenius et al., 2019). Nevertheless, the addition of CNC-CHO or covalent-bound CNC-SH could tailor the hydrogel degradation for potentially achieving a controlled therapeutic delivery manipulation.

BaGNP laden GGMMA+CNC-SH hydrogels and in vitro sustained release of therapeutics ions in SBF
Supported by the sustained ion release profiles and the confirmed none cytotoxicity in the culture of various cell lines, BaGNP has been suggested as promising nano-sized fillers to develop nanocomposites both for bone regeneration and wound healing, especially Cu-BaGNP (Wang et al., 2016;Weng et al., 2017). The two sol-gel-derived BaGNP samples, BaGNP and Cu-BaGNP, are highly dispersive in water and display as mono-dispersed solid spheres with a diameter around 400 nm as determined from TEM and SEM imaging in Fig. 4a. The addition of Cu precursor resulted in no distinct variations in surface morphology between the BaGNP and Cu-BaGNP, as the comparatively low doping content of Cu in Cu-BaGNP. Semi-quantitative surface element analysis was carried out using the EDXA in conjugation with SEM imaging to determine the composition of the BaGNP samples, as presented in Fig. 4a. Both BaGNP and Cu-BaGNP showed a significant deviation from their respective nominal composition, and the actual contents of CaO and CuO were significantly lower than the nominal values: 4.53 mol% CaO was confirmed in the final composition of BaGNP, and 0.41 mol% CuO was further introduced as a competing dopant with 4.17 mol% CaO for Cu-BaGNP. The result is in line with the previous study, owing to the limited active sites in the silicate particles, only a certain amount of Ca or/and Cu ions are adsorbed in the Stöber process and eventually integrated into the network structure as dopants in the calcinated BaGNPs (Zheng et al., 2017). In line with the strategy mentioned above, the hydrogel precursors of GGMMA+CNC-SH were then proposed to construct a UV-curable nanocomposite hydrogel of polysaccharides and BaGNP as a delivery system aiming to provide a sustained release of therapeutic ions including Si, Ca, or/and Cu ions. The hydrogel of 2% GGMMA+2% CNC-SH was selected as the carrier system matrix due to its outstanding mechanical strength and relatively homogenous structure. BaGNP or Cu-BaGNP was introduced at a dosage of 0.4%, 1%, or 2% to prepare the nanocomposite hydrogels of GGMMA+CNC-SH+BaGNP. The release profiles of Si, Ca, and Cu ions/ species from these nanocomposite hydrogels were registered in SBF for up to 7 or 14 days, as shown in Fig. 5. BaGNP and Cu-BaGNP laden hydrogels showed a similar Si release profile, and the released Si concentration is mainly related to the BaGNP loading in the hydrogel. A sustained release of Si ions/species was observed during the evaluation, in which an almost linear increase within 7 days, as shown in Fig. 5a and c. To be noticed, BaGNP and Cu-BaGNP alone showed rapid Si release profiles in the first 3 days, then a slow release in the next 11 days (Zheng et al., 2017). This indicates that the GGMMA+CNC-SH hydrogel, as a delivery matrix, could impact the release of Si and enable a sustainable release profile. For both BaGNP laden hydrogels, a depletion of Ca ions was initially observed within the time point of 2 days and continued to increase in the late dissolution stage. This might be owing to the released Ca and P formed CaP-rich species. However, no obvious apatiteformation (typically as needle-like crystals) was observed under SEM investigation to the cross-sectioned lyophilized cryogels after immersion of dissolution test. The GGMMA+CNC-SH cryogel showed a highly porous network, as displayed in Fig. 4b (IV and VII). Still, a relatively homogeneous and spatial embedding of BaGNP in the matrix of GGMMA+CNC-SH was suggested in Fig. 4b (V and VIII for 2% BaGNPladen cryogel; and VI and IX for 2% Cu-BaGNP-laden cyrogel). Besides, no detectable concentration of Cu ion was found in SBF by the ICP-OES analysis (or the concentration of Cu ion is lower than the detection limit). However, this did not imply that no Cu ions were dissolved from the embedded Cu-BaGNP in the hydrogel matrix. It is speculated that the dissolved Cu ions were adsorbed onto the GGMMA and CNC-SH matrix through the ionic complexation with the presence of sulfate half ester groups on the surface of CNC and carbonyl groups in grafted L-cysteine.

Fabrication of GGMMA+CNC-SH hydrogels with DLP lithography printing
DLP additive manufacturing (AM) creates models in a layer-by-layer manner through photo-polymerization via UV or visible light. The technological dimensions of DLP 3D printing are highlighted with excellent spatial resolution in pattern fidelity and rapid fabrication speed, which has made it popular in fabricating custom-designed hydrogel constructs with biomaterial resins of different kinds (Hong et al., 2020;Shen et al., 2020;Ye et al., 2020). Preliminary, we further investigated the applicability of the hydrogel precursors of GGMMA+CNC-SH as the biomaterial resin in DLP printing. Considering the requirement on resins suitable for DLP in terms of flowability (ideally low-viscosity Newtonian fluids in recoating process), the less viscous hydrogel precursor of 2% GGMMA+1% CNC-SH was chosen as the resin to print a honeycomb structure in 1 mm height, as digitally designed in a model depicted in Fig. 6a.
In this AM technique, apart from the pixel size as defined by the photonics in the DLP printer, the printing resolution is majorly determined by the kinetics of photo-polymerization. When printing the resin of 2% GGMMA+1% CNC-SH, blurred projected pattern (over-curing layers beyond the focus plane, indicated by arrows) and excess crosslinking were observed in the honeycomb. This was caused by the 'light trespassing' associated with the weak light absorption of optically clear GGMMA+CNC-SH. Here, tunable crosslinking kinetics is necessary to improve the shape fidelity of the printed hydrogel. Commonly, a photoabsorber that functions as a light-attenuating additive is added in resin formulation to absorb excess light, e.g. water-soluble dyes such as tartrazine, curcumin, or anthocyanin that has strong absorbance in the near-UV to visible blue light (Grigoryan et al., 2019;Yu et al., 2020). Herein, tartrazine was incorporated as a photoabsorber in 2% GGMMA+1% CNC-SH. As shown in Fig. 6e, the crosslinking kinetics of the resin was gradually inhibited with increasing the tartrazine concentration. With optimizing this parameter in the printing of the honeycomb hydrogel, 0.4 mM tartrazine was found to significantly preventing the over-curing of resins and improved the shape fidelity of the hydrogel, as shown in Fig. 6(b, c, f and g). When 0.4% Cu-BaGNP was further encapsulated in the formulation, the printed honeycomb hydrogel with sharp edges also shows good shape fidelity and a clear X-Y resolution could be observed from the microscopy images from Fig. 6h.

Conclusion
Through photo-clickable thiol-ene crosslinking, the methacrylated derivative of woody polysaccharide (GGMMA) together with the thiolgrafted CNC (CNC-SH) are high-performance building blocks to form an injectable and rapidly photocurable nanocomposite hydrogel. Based on on/off-stoichiometry content control of thiol:ene, the mechanical stiffness of these all wood-derived polysaccharide hydrogels was tunable within the range of 1.43 to 12.35 kPa. The on-stoichiometry thiol-ene addition guaranteed a homogenous network within the gel, which supports strong mechanical properties and a fast enzymatic degradation kinetics when incubated with mannanase. As an extended therapeutic delivery function, a sustained release of Si and Ca ions/species was achieved by embedding the BaGNP in the hydrogel of GGMMA+CNC-SH. Moreover, the GGMMA+CNC-SH formulation is suitable as biomaterial resin in DLP lithography printing to fabricate digitally designed hydrogel constructs.

Declaration of competing interest
The authors declare no conflicts of interest. Q. Wang et al.