Gelatin-biofermentative unsulfated glycosaminoglycans semi-interpenetrating hydrogels via microbial-transglutaminase crosslinking enhance osteogenic potential of dental pulp stem cells

Abstract Gelatin hydrogels by microbial-transglutaminase crosslinking are being increasingly exploited for tissue engineering, and proved high potential in bone regeneration. This study aimed to evaluate, for the first time, the combination of enzymatically crosslinked gelatin with hyaluronan and the newly developed biotechnological chondroitin in enhancing osteogenic potential. Gelatin enzymatic crosslinking was carried out in the presence of hyaluronan or of a hyaluronan–chondroitin mixture, obtaining semi-interpenetrating gels. The latter proved lower swelling extent and improved stiffness compared to the gelatin matrix alone, whilst maintaining high stability. The heteropolysaccharides were retained for 30 days in the hydrogels, thus influencing cell response over this period. To evaluate the effect of hydrogel composition on bone regeneration, materials were seeded with human dental pulp stem cells and osteogenic differentiation was assessed. The expression of osteocalcin (OC) and osteopontin (OPN), both at gene and protein level, was evaluated at 7, 15 and 30 days of culture. Scanning electron microscopy (SEM) and two-photon microscope observations were performed to assess bone-like extracellular matrix (ECM) deposition and to observe the cell penetration depth. In the presence of the heteropolysaccharides, OC and OPN expression was upregulated and a higher degree of calcified matrix formation was observed. Combination with hyaluronan and chondroitin improved both the biophysical properties and the biological response of enzymatically crosslinked gelatin, fastening bone deposition.


Introduction
Tissue engineering (TE) holds big promise for damaged or lost tissues. The demand for effective bone engineering has so far made the greatest gains; however, the ideal scaffold/cell construct is still to be found [1]. The most innovative constructs consist in the combination of stem cells with scaffolds closely mimicking the cell natural environment.
Gelatin (gel) and bioactive polysaccharides, such as hyaluronan (HA) and chondroitin sulfate, are widely employed in scaffold design to resemble the composition of the bio-organic part of bone extracellular matrix (ECM) (or, generally the ECM composition of other connective tissues) [2][3][4][5][6][7][8][9][10]. Gelatin, obtained from collagen I by hydrolytic processes, is successfully used in place of collagen since retaining the functional moieties recognized by cells while exhibiting lower antigenicity, ease of processing and availability at lower cost [1,[11][12][13][14]. Among polysaccharides, HA is widely employed due to its well-known structural and signaling role in the ECM of connective tissues affecting cell behavior, migration and differentiation. It creates a microenvironment that improves osteogenesis and mineralization; therefore, its incorporation in scaffolds is expected to positively affect this specific stem cells differentiation [2,[15][16][17][18].
It is well known that to obtain highly performing (ECM resembling) scaffolds, attention has to be paid not only to the polymeric composition but also to the crosslinking processes needed to obtain water-insoluble matrices [10].
In this respect, most of the gelatin-scaffolds proposed so far were produced using dialdehydes, polyepoxides, poly-isocyanates or genipin as crosslinkers or involving the use of carbodiimides [19][20][21][22]. Although interesting results have been achieved, toxicity may be associated to these crosslinking procedures. HA is mainly introduced into gelatin networks in chemically modified forms, crosslinked together with or independently from the protein chains. This reduces and may even hinder HA signaling role due to modification in the macromolecule conformation.
Safer crosslinking strategies, also allowing for the incorporation of biopolymers in their unmodified, bioactive form, are desirable and expected to better resemble ECM and to elicit a good biological response.
In the attempt to move a step forward, here we proposed new semi-interpenetrating (semi-IPN) gelatin-bioactive heteropolysaccharides matrices for bone regeneration. Specifically, we aimed to obtain gelatin networks using microbial-transglutaminase (m-TG), in the presence of naturally occurring hyaluronan and the novel unsulfated chondroitin. m-TG catalyzes the transamidation reaction between the amino-group of lysine residues and the carboxamide group of glutamine residues thus forming gelatin hydrogels under safe conditions. This crosslinking approach has only lately been exploited to produce gel matrices for tissue engineering [10][11][12][23][24][25][26][27][28][29][30][31][32]. Two recent reports demonstrated the ability of these hydrogels to stimulate osteogenic differentiation of primary murine mesenchymal stem cells from bone marrow and to promote human dental pulp stem cells differentiation into odontoblasts [31,32]. By carrying out gel crosslinking in the presence of linear HA, the latter should only be physically entrapped into the m-TG network. This would better resemble the state of the polysaccharide in natural ECM. Therefore, we expected to fully exploit HA bioactivity improving m-TG gel matrix osteogenic potential. Only chemically modified HA was combined to m-TG gel networks so far and no investigation for bone engineering was reported. We also aimed to evaluate the un-sulfated biotechnological chondroitin (BC) potential for bone engineering This peculiar molecule, which differs from the animal tissue-derived sulfated chondroitin, has been recently obtained, at pharmaceutical grade in our laboratories, exploiting a biotechnological process. Its bioactivity was shown in diverse in vitro models [33][34][35]. Specifically, it sustained human primary chondrocyte viability and held anti-inflammatory effects [33]. BC is for the first time employed, here, as a scaffold component to test for a potential synergic effect with HA in bone regeneration.
The work describes the development and the evaluation of the semi-IPNs. Specifically, the latter, containing HA and HA/BC, were characterized in comparison to the matrix containing only gelatin. The characterization studies aimed to verify semi-IPNs suitability for the intended application and to evaluate the effect of the heteropolysaccharides on the biophysical properties of the m-TG gel network as well as on its ability to promote bone regeneration, in combination with human dental pulp stem cells (hDPSCs).

Materials
Gelatin is a Nitta Gelatin Inc product (G-1865P, Japan). It is a type A, bovine gelatin, 146 gel strength. Microbial Transglutaminase, Activia WM (1% TG, 99% maltodextrin), 100 U/g was purchased from Ajinomoto (Japan). HA sodium salt was prepared by hydrolyzing an HA M w ¼1100 kDa (a gift from Altergon Srl, Italy) sample under heterogeneous acid conditions, as reported elsewhere [36]. Dulbecco's Phosphate Buffered Saline (DPBS) without calcium and magnesium was purchased from Gibco by Life Technologies. BC was obtained through extensive purification processes following fed-batch fermentation of Escherichia coli rfaH recombinant strain [37,38].

Methods
Biopolymer hydrodynamic analyses HA, BC and gel hydrodynamic characterization was performed using the size exclusion chromatography coupled with triple detector array (SEC-TDA) system by Viscotek (Malvern) under previously reported conditions [39]. The molecular weight (M w , M n , M w /M n ), molecular size (hydrodynamic radius-R h ) and intrinsic viscosity ([g]) distributions were derived. A dn/dc value equal to 0.186 was used for gel [40]. Each sample was analyzed in triplicate; results were reported as means 6 SD.

Gelation studies
Gelatin gelation due to m-TG action was studied by performing rheological measurements using a Physica MCR301 (Anton Paar, Germany) oscillatory rheometer equipped with a Double Gap geometry (DG26.7Q1-SN42960) with a measuring gap (internal) of 0.419 mm and a Peltier temperature control. Temperature was set at 37 C.
Measurements aimed at verifying proper conditions for gel matrix formation and, then, the occurrence and/or potential variation of the process in the presence of HA and HA/BC.
A gelatin solution 5.55 wt% was first prepared by dissolving the polymer in PBS, at 40 C, under stirring (800 rpm). Then, a m-TG solution in PBS was added to have 5 wt% gel and 2 U/mL enzyme final concentration.
Immediately after the addition of the enzyme, the sample was gently stirred and poured into the rheometer reservoir. The gelation process was monitored by recording the G 0 and G 00 values, at 10 rad/ s angular frequency (1.59 Hz) and 1% strain, as a function of time. The time needed to reach the sol/gel transition point (G 0 and G 00 crossover point, tan delta ¼ 1) was recorded. Attention was paid to start the registration of the moduli exactly 2 0 after the addition of the enzyme.
The process in the presence of HA and HA/BC was then studied. Gel/HA and gel/HA/BC solutions were obtained by dissolving gel at 5.55 wt%, HA and BC at 0.55 wt% in PBS as described above. A m-TG solution was added to have 5 wt%, 0.5 wt% and 2 U/mL final concentration for gel, HA and BC and the enzyme, respectively. The process was monitored as described above and the time needed for gelation was determined as described for gel solutions.
All the solutions were filtered on 0.22 mm before use. The experiment was performed at least three times for each specific polymeric composition and the gelation point was expressed as the mean 6SD.
Solutions containing (i) gel 5.55 wt%, (ii) gel 5.55 wt% and HA 0.55 wt% and (iii) gel 5.55 wt%, HA 0.55 wt% and BC 0.55 wt% in PBS were prepared and sterilized by filtration through 0.22 mm filters. A filter sterilized (0.22 mm) m-TG solution in PBS was added to each of the previous solutions to obtain a final gel concentration of 5 wt%, with HA at 0.5 wt%, HA at 0.5 wt% plus BC at 0.5 wt% and m-TG 2 wt%. The solutions were poured into cylindrical tubes (Teflon or PFA) to have 0.6 ml solution/cm 2 surface and gently stirred to homogeneously disperse the enzyme while avoiding bubble formation. The tubes were covered to prevent solvent evaporation and kept in an oven, at 37 C for 24 h. The resulting hydrogels, we will refer to as Xgel, Xgel/HA and Xgel/HA/BC, were punched out to obtain 25 mm disks. The latter were treated three times with ethanol/water 80/20 v/v to remove the PBS salts, then with ethanol (Absolute or 96%v/v) and, finally, dried under vacuum at 40 C.
The dried disks obtained constituted the three diverse scaffolds that were used for the characterization studies, and successively for the biological analyses.

FT-IR analyses
The hydrogels were examined by Fourier Transform Infrared spectroscopy (FT-IR) using a Perkin Elmer Spectrum One spectrometer in a transmission geometry. KBr pellets were prepared by mixing a small quantity of a dried material with KBr (at the ratio of 1/100). For every sample, two KBr pellets were prepared. All spectra were obtained using 64 scans in the range from 4000 to 450 cm À1 with a 4 cm À1 spectral resolution. Each measurement was performed in triplicate and the spectra reported are the average of three spectra acquired for each sample. The spectra were analyzed using the application routines provided by the software package ('Spectrum' Perkin Elmer Inc., Hopkinton, MA, USA) controlling the whole data acquisition system.

Swelling studies
The hydrogel swelling behavior was studied by means of gravimetric measurements. Briefly, the dried hydrogel disks were weighed (W d ), soaked in PBS and incubated at 37 C. At diverse time intervals up to 1 week, the disks were withdrawn from the medium, gently blotted with filter paper to remove excess water and then weighed (W s ). The swelling degree at each time point was calculated as follows: Experiments were performed in triplicate and the results are reported as the mean value 6 standard deviation. Kinetic curves were derived for each material by reporting the swelling degree as a function of time. The time needed for each material to reach the equilibrium was determined by analyzing the statistical significance of the values obtained over time (one-way ANOVA analysis with Holm post hoc correction for multiple comparison). Specifically, the time needed for the equilibrium swelling was identified as the time point at which a swelling degree significantly higher (P 0.05) than the values at shorter incubation time and comparable (P > 0.05) to the values at longer incubation was obtained. Finally, the equilibrium swelling degree for each material was calculated as average of all the equilibrium swelling degree data 6 standard deviation.

Rheological characterization
Hydrogels were swollen to equilibrium in PBS prior to the characterization. The minimum amount of PBS needed to reach the swelling equilibrium was added to each disk. Strain tests and frequency sweep tests were carried out using a parallel plate geometry, 25 mm plate diameter, as reported elsewhere, with slight modifications [41]. Profiled plates were used to prevent slipping during the measurements. The strain tests were performed at 10 rad/s angular frequency (1.59 Hz frequency) over a strain amplitude range from 0.1 to 100% [42][43][44]. The frequency sweep tests were carried out at 0.1% strain, over a frequency range from 0.159 to 10 Hz. All measurements were carried out at 37 C.
Strain tests were also carried out on the gels formed after 24 h m-TG action, before the purification process, for the determination of tan delta values.
Hydrogel stability Stability to hydrolysis under physiological conditions was investigated by soaking the materials in PBS or cell culture medium at 37 C, and by monitoring the appearance and the weight loss (%) as a function of the exposure time up to 21 days.
Hydrogel stability to collagenase action was also evaluated as previously described [45,46]. Briefly, comparable amounts of the materials were incubated in a collagenase solution (3 U/mL) in PBS, at 37 C, and the weight loss (%) was monitored over incubation time.
HA and CB release/retention studies To evaluate hydrogels' stability under physiological conditions, at least three samples for each gel were allowed to completely hydrate in cell culture medium and then incubated under cell culture conditions up to 30 days. Variation of hydrogel shape, dimension and weight was monitored.
To study glycosaminoglycan (GAGs) release from the matrices, hydrogel samples were soaked in PBS (30 ml/g) and incubated at 37 C. At diverse time intervals up to 30 days, the medium was entirely withdrawn and replaced by fresh PBS. The conditioned medium was filtered on 0.22 mm and then analyzed by size exclusion chromatography (SEC-high performance liquid chromatography (HPLC); SEC-TDA) to verify GAGs release, both qualitatively and quantitatively. The SEC-TDA analyses were performed as reported above. For SEC/HPLC analyses, a UPLC-Ultimate3000þ (Thermofisher, Italy) equipped with 2ÂTSK-Tosoh 3200 was used. Samples were eluted in 30 min run, at 30 C, using NaNO 3 0.1 M at 0.9 ml/min; a refractive index (RI) detector was used.
At each time point, the amount of HA or BC released (%) was calculated as indicated by the following: Released polysaccharideð%Þ ¼ released polysaccharide mass in the mediumðas quantified by chromatographic analysesÞ= total polysaccharide content in the initial sample Â 100: Cell culture and hDPSCs isolation Human dental pulps were extracted from teeth of healthy adults (aged 21-35 years) as described previously by Naddeo and collaborators [47]. All participants signed the Ethical Committee (University of Campania Internal Ethical Committee) consent form. This study was performed in line with the principles of the Declaration of Helsinki.
Briefly, the pulp was removed and immersed for 1 h at 37 C in a digestive solution of 3 mg/ml of type I collagenase and 4 mg/ml of dispase in PBS containing 40 mg/ml of gentamicin. Once digested, the solution was filtered through 70 lm Falcon strainers (Becton & Dickinson). Cells were cultured in standard medium consisting of Dulbecco's modified Eagle's medium (DMEM) with 100 units/ml of penicillin, 100 mg/ml of streptomycin and 200 mM l-glutamine (all purchased from Gibco), supplemented with10% fetal bovine serum (FBS) (Invitrogen). Cells were maintained in a humidified atmosphere under 5% CO 2 at 37 C and the media were changed twice a week.
At first passage of culture, cells were detached using trypsin-EDTA (GIBCO). At least 200 000 cells were incubated with fluorescent-conjugated antibodies for 30 min at 4 C, washed and resuspended in PBS. The antibodies used in this study were: anti-CD34 PE, anti-CD90 FITC and anti-CD45 APC-Cy7, all purchased from BD Pharmingen (Buccinasco, Milan, Italy). Isotypes were used as controls. Cells were analyzed with FACS ARIA III (BD Biosciences, San Jose, CA, USA) and data collected with Diva Software. hDPSCs were isolated using co-expression of CD90 and CD34 markers at FACS ARIA III (BD, Franklin Lakes, NJ, USA) as previously reported [47]. The purity of sorted populations was routinely 90%.

MTT analyses
In order to evaluate the viability and proliferation of hDPSCs on hydrogels, tetrazolium dye (3-4,5 dimethylthiazol-2-yl)2,5diphenyltetrazoliumbromide (MTT) analyses were performed. Materials were hydrated in osteogenic medium for 5 h, and hDPSCs were seeded at a density of 3 Â 10 4 cells/construct. After 1 h of incubation in 100 ml of culture medium to allow cell attachment, the constructs were cultured in osteogenic medium in a humidified atmosphere at 37 C and 5% CO 2 . Seeded hydrogels were collected after 24 h, 3 days and 7 days of culture. The medium was removed and the constructs (hDPSC-hydrogels) were incubated for 4 h in a solution of 5 mg/ml MTT. Hydrogels without hDPSCs were used as control. After medium removal, 1 ml of HCl 0.1 M in isopropanol was added to each well containing seeded hydrogels; after 40 min, supernatants were collected and absorbance at 550 nm was recorded. Absorbance values at 3 and 7 days of culture were normalized to the ones recorded at 24 h after seeding.
Hydrogel cell constructs preparation and osteogenic differentiation hDPSCs, at first passage of culture, were seeded on Xgel, Xgel/HA and Xgel/HA/BC at a density of 3 Â 10 5 cells/construct in DMEM standard medium at 10% FBS, and osteogenic medium for 7 days. At this time, osteopontin (OPN) and osteocalcin (OC) gene expression was evaluated by qRT-PCR.
Experiments on the same matrices in osteogenic medium were prolonged also at 15 and 30 days. Specifically, hydrogels were hydrated in osteogenic medium for 5 h, and hDPSCs were seeded. After 1 h of incubation in 100 ml of culture medium to allow cell attachment, the constructs were cultured in osteogenic medium in a humidified atmosphere at 37 C and 5% CO2. Osteogenic medium (OM) contained 100 nmol/l dexamethasone, 10 mmol/l betaglycerophosphate and 0.05 mmol/l L-ascorbic acid-2-phosphate. For osteogenic differentiation, the expression of bone related markers such as OPN and OC was evaluated at 7, 15 and 30 days.
SEM observation SEM analyses were performed on the hydrogels cultured with cells after 7, 15 and 30 days of culture. Samples were fixed in paraformaldehyde 4% v/v in PBS overnight, and dehydrated by incubating in ethanol/water mixtures at increasing ethanol concentration (ethanol 30-90 vol% for 5 min, absolute ethanol for 15 min for three times). Materials were then dried in a critical point dryer and sputtered with platinum-palladium (sputter coater Denton Vacuum Desk V). Fe-SEM Supra 40 Zeiss (5 KV, detector InLens) and Smart SEM Zeiss software were used for observation. SEM images were also acquired for not seeded materials, kept under the same culture conditions as above.
Finally, not seeded samples, dehydrated in ethanol and dried under vacuum, then frozen at À80 C and cut using a blade, were observed to acquire images of material sections.

Two-photon microscopy
Samples were fixed in paraformaldehyde 4% v/v in PBS for 30 min and then stained with Phalloidin (Sigma-Aldrich) (100 mM in PBS for 30 min at room temperature protected from light) and with Hoechst (Abnova) (2 mg/ml for 15 min at room temperature protected from light), and immediately observed, in their naturally hydrated form with two-photon microscopy (TPM).
TPM images were obtained using a modified Olympus Fluoview confocal laser scanning head (FV300) coupled to a fs-Titanium: Sapphire (Ti: Sa) laser (Chameleon Ultra, Coherent, Inc., Palo Alto, California, USA) and equipped with an upright Olympus BX51WI microscope. A water immersion objective (Olympus XLUMPLANFl20XW, WD: 2 mm, NA: 0.95) was used for focusing the laser beam and collecting the fluorescence signal from the samples. The laser pulse time width was estimated to be 150 fs with 76 MHz pulse repetition frequency. For TPM, excitation at 800 nm wavelength was adopted; the fluorescence emission was collected in the 450-550 nm range. The images were acquired with fixed excitation energy at a depth up to 200 lm below the sample surfaces with a 2 lm step. All of the acquired images were 230 Â 230 mm 2 in an area with a resolution of 512 Â 512 pixels and a pixel dwell time of 6.4 ms. For further details, see Lepore and collaborators [48].
Gene expression analyses of OC and OPN as osteogenic markers through qRT-PCR Total RNA was isolated from hDPSCs grown on the scaffolds for 7, 15 and 30 days and after homogenization through Tissue Ruptor Homogenizer (Qiagen, Hilden, Germany), the cells were lysed with TRIzol V R (Invitrogen, Milan, Italy). Following precipitation with isopropyl alcohol and washing with 75% ethanol, the RNA pellets were re-suspended in nuclease-free water. The concentration of the extracted RNA was determined through a Nanodrop spectrophotometer (Celbio, Milan, Italy) and 1 mg of DNase-digested total RNA was retro-transcripted in the cDNA using Reverse Transcription System Kit (Promega, Milan). Quantitative real time PCR was performed through iQTM SYBR V R Green Supermix (Bio-Rad Laboratories Srl, Milan, Italy) to analyze gene expression of some osteogenic markers such as OC (forward 5 0 -CTCA CACTCCTCGCCCTATTG-3 0 , reverse 5 0 -CTTGGACACAAAGGC TGCAC-3 0 ), and OPN (forward 5 0 -GCCGAGGTGATAGTGT GGTT-3 0 , reverse 5 0 -TGAGGTGATGTCCTCGTCTG-3 0 ). Samples were run in triplicates and the expression of specific mRNA relative to the control was determined after normalization with hypoxanthine guanine phosphoribosyl transferase (HPRT) (forward 5 0 -TGACCTTGATTTATTTTGCATACC-3 0 , reverse 5 0 -CGAGCAA GACGTTCAGTCCT-3 0 ) housekeeping gene (internal control). The fold-change of mRNA expression of the genes under evaluation was calculated by the 2 ÀDDCt comparative threshold method. The results were expressed as normalized fold expression versus Xgel sample, calculated by the ratio of crossing points of amplification curves of several genes and internal standard, through the Bio-Rad iQTM5 software (Bio-Rad Laboratories Srl) as previously reported [33,49].

Western blotting analyses
Cell pellets were lysate in radioimmunoprecipitation assay (RIPA) buffer and debris were separated by centrifugation. Protein concentrations were determined by protein assay reagent (Bio-Rad). Equal amounts of proteins (30 lg) were loaded by SDS-PAGE on 10% polyacrylamide gel. Then, the proteins were transferred to nitrocellulose membrane and blocked with milk (Santa Cruz Biotechnology, CA, USA) for 1 h. The filters were incubated with antibodies against OPN (mouse monoclonal sc-21742; 1:200 v/v), OC (mouse monoclonal sc-74495; 1:200 v/v) and actin (goat polyclonal sc-1616; 1:500) at room temperature for 2 h. Membranes were washed three times for 10 min and incubated with a 1:5000 dilution of horseradish peroxidase-conjugated anti-mouse antibodies and with a 1:10 000 dilution of horseradish peroxidase-conjugated anti-goat antibodies for 1 h, respectively. All primary antibodies were purchased from Santa Cruz Biotechnology (CA, USA), secondary antibodies were obtained from Bethyl Laboratories. Blots were developed using the ECL system according to the manufacturer's protocols (Amersham, Biosciences). Actin antibody was used as the gel loading control. The protein expression was quantified by using Image J software.

Statistical analyses
Each experiment was performed at least in triplicate and results are reported as the mean value 6 standard deviation. Data were statistically evaluated by performing one-way ANOVA tests followed by post hoc tests using Holm correction for multiple comparison. The level of significance was fixed at 0.05.

SEC-TDA analyses of the biopolymers
The hydrodynamic parameters for the biopolymers used, as derived from the SEC-TDA analyses, are reported in Table 1

Gelation studies
The rheological behavior recorded for gel and gel/GAGs solutions, in the presence of m-TG, proved the occurrence of gel crosslinking, even in the presence of GAGs (Fig. 1). In fact, as expected, the gel/ m-TG sample initially showed an essentially viscous behavior with G 00 higher than G 0 (Fig. 1a). A progressive increase in both the moduli was then observed, with G 0 increasing more markedly than G 00 , until a cross over was registered, indicating the transition to an essentially elastic behavior. A similar trend for the dynamic moduli was registered in the presence of GAGs (Fig. 1a). Further, no  Data are reported as the mean 6 SD of at least three replicates significant differences in the time needed for the sol/gel transition (G 0 and G 00 crossover) were observed for the diverse compositions indicating that GAGs did not interfere with gel gelation (Fig. 1b).

Chemical characterization
In Fig. 2, the FT-IR average spectra of the three hydrogels are reported. The typical main bands of gelatin were identified in the Xgel spectrum. Specifically, the large band at 3436 cm À1 (OH and NH stretching), the faint features in the 3100-2880 cm À1 (mainly due to NH stretching, CH 2 asymmetric stretching and CH 3 symmetric stretching, respectively), the band in the Amide I region (located at 1638 cm À1 ), the bands in the Amide II and Amide III regions (identified at 1545 and 1232 cm À1 , respectively) were detected [52][53][54]. The same pattern was observed in the presence of HA and HA and BC due to the overlap of the signals and the small amount of the polysaccharides within the samples. However, the amide I band shifted in the presence of HA and HA and Biotechnological chondroitin (CB) up to 1648 cm À1 that is an indication of the presence of the polysaccharides and of an interaction among the polymers. A table reporting the assignment of the main peaks is reported as supplementary material ( Supplementary Fig. S1)

Swelling behavior
Kinetic studies revealed that hydrogels reached the equilibrium swelling in PBS within 5 h (Supplementary Fig. S2). Images in Fig. 3a show the appearance of the gels, in their dried state and after hydration in PBS. The equilibrium swelling degree (%) values for the hydrogels are reported in Fig. 3b. Xgel increased its dry weight by about 1400%. The semi-interpenetrated matrix containing HA and BC exhibited significantly (P < 0.01) lower water up-take ability (swelling degree around 1200%). The swelling extent for Xgel/HA was slightly lower, compared to Xgel (P ¼ 0.05).

Hydrogel stability
The gels were highly stable under physiological conditions. In fact, no significant variation in the appearance/dimension of the hydrated matrices was detected in 21 days experiments. In addition, significant weight variation of the (hydrated) samples, beyond the release of linear HA and BC, that was specifically quantified (Section 'HA and BC release'), could not be detected within 4 weeks of incubation in PBS or cell culture medium.
When incubated in the presence of collagenase (Fig. 4), the hydrogels showed a comparable degradation profile with a residual mass of about 45% at 3 h of incubation while the weight loss was more than 90% at the longer time tested.

Rheological behavior
The strain sweep tests, on materials hydrated at the equilibrium, revealed, for all the hydrogels, G 0 highly exceeding (more than 100fold) G 00 in the linear viscoelastic range (LVR) (Supplementary Fig.   Figure 2. FT-IR average spectra of the hydrogels in the 4000-500 cm À1 range  ). The G 0 values for the matrices, measured within the LVR and at 1.59 Hz frequency, were in the range 1400-1900 Pa with Xgel behaving as the less rigid matrix (Fig. 5a). The presence of GAGs reflected in higher stiffness; however, no significant variation in rigidity was found between Xgel/HA and Xgel/HA/CB. The mechanical spectra indicated that G 0 /G 00 ratio was almost maintained over all the frequency range exploited (Fig. 5b); a flat dependence of the moduli on frequency, especially for G 00 , was detected in the region of high frequencies (Fig. 5b).
When strain sweep tests were performed on the gels immediately after 24 h m-TG action (prior to purification process and in hydration conditions other than the ones at equilibrium), tan delta values in the range 0.006-0.009 were found. No significant differences were detected among the formulations (P > 0.05).

HA and BC release
The chromatographic analyses of the polysaccharides released in PBS for prolonged incubation indicated a sustained release of HA and of both HA and BC from Xgel/HA and Xgel/HA/BC, respectively, up to 30 days. The quantitative evaluation (Fig. 6) indicated that less than 10 wt% HA was released from Xgel/HA at 24 h, about 30 wt% at 7 days, with a further increase up to about 50 wt% after 30 days of incubation (Fig. 6a). Therefore, about half of the amount of HA was still retained in the gel at the longer time tested.
Both HA and BC were released from Xgel/HA/BC (Fig. 6b). At each time point, the amount of HA (wt%) in the extracts was higher than the one registered for Xgel/HA (P < 0.05). The greatest amount of the polysaccharides was released in the first week; however, about 20 and 40% of BC and HA, respectively, were still retained at 7 days of incubation.

DPSCs isolation and cell culture
At first passage of culture, dental pulp cells were analyzed for the co-expression of CD34 and CD90 mesenchymal markers.   Cytometric analysis showed that all cells were positive for CD90, and negative for CD45, leucocyte marker. CD34 expression was about 20%. DPSCs were isolated by CD34 and CD90 coexpression.
When hDPSCs were cultured on the hydrogels in non-osteogenic medium, gene expression data for OC and OP showed no detectable difference for Xgel/HA and Xgel/HA/BC, compared to Xgel alone (data not shown).
Therefore, all experiments were performed using CD34 þ CD90 þ DPSCs at 1 passage of culture, seeded on Xgel, Xgel/HA and Xgel/ HA/BC and cultured in osteogenic medium.

MTT evaluation
To evaluate how the hydrogels affect viability and proliferation of hDPSCs, MTT analyses were performed. hDPSCs were cultured on Xgel, Xgel/HA, and Xgel/HA/BC for 24, 72 and 168 h (7 days). Cell viability at 3 and 7 days was reported normalized to the values obtained at 1 day of incubation (Fig. 7). Results showed that hydrogels were not cytotoxic. In addition, there were no changes in terms of proliferation up to 72 h of culture. At 7 days, proliferation was significantly different on the diverse hydrogels (P < 0.05). In particular, Xgel and Xgel/HA/BC induced the highest and lowest proliferation, respectively.

SEM observation
To better characterize the effect of the combination of gel with HA and HA/BC on ECM deposition, SEM observation of seeded hydrogels was carried out at 7, 15 and 30 days of culture (Fig. 8). Images in the upper right box of Fig. 8 represent not seeded materials. The results showed that already at 7 days of culture, all the hydrogels were highly colonized (Fig. 8A-C). In particular, Xgel induced DPSCs to form a homogeneous and continuous layer (Fig. 8D). Xgel/HA showed the same tendency of Xgel, with an initial deposition of ECM (arrows) (Fig. 8E). For Xgel/HA/BC, cell colonization was different compared to Xgel and Xgel/HA. A surface covered with calcification granules and cell clusters was observed (Fig. 8F). SEM observations indicated an increase in ECM deposition for all constructs at 15 and, above all, at 30 days, when all hydrogels induced an extensive ECM deposition with calcification granules covering the entire surface ( Fig. 8G and H).
Images of sections for not seeded materials ( Supplementary Fig.  S4) demonstrated material porosity.

Two-photon imaging
TPM investigation showed the presence of hDPSCs for all the examined hydrogels in different spatial regions whose localizations essentially depend on the time elapsed from the seeding (Fig. 9A-C). At 7 days of culture, TPM results clearly evidenced the presence of hDPSCs in all hydrogels at a depth equal to 60 mm indicating the potential of these gels to be colonized by cells (Fig. 9A-C). At 15 and 30 days, the cells penetrated hydrogels at greater depth. In particular, after 15 days of culture, hDPSCs penetration depth up to 120 mm was observed for Xgel/HA/BC.

Gene expression (qRT-PCR) analyses and western blotting
To evaluate the effect of hydrogels on osteogenic differentiation, the relative mRNA expression levels for OC and OPN, normalized to actin gene expression, were analyzed. At 7 days of culture (Fig. 10a) OC mRNA levels increased 36.5-fold in the presence Xgel/HA/BC (P < 0.05), and 4.7-fold in the presence of Xgel/HA. At longer times (15 and 30 days), the OC mRNA levels decreased for the materials containing biofermentative unsulfated GAGs still showing a slight overexpression compared to Xgel (Fig. 10b and c).
Regarding to OPN, a 3.8-fold higher expression was detectable at 7 days of culture for hDPSCs grown on Xgel/HA, compared to Xgel, while, at the same time, Xgel/HA/BC presented with a 2-fold increase (Fig. 10a). At 15 and 30 days, the OPN expression levels decreased and a down-regulation of this marker was found for Xgel/ HA/BC respect to Xgel at 30 days (Fig. 10b and c).
In order to better characterize the osteogenic differentiation, OC and OPN expression was also evaluated by western blotting (Fig. 11). OPN and OC, after 7 days of culture were overexpressed in hDPSCs seeded on Xgel/HA and Xgel/HA/CB compared to Xgel (P < 0.05) (Fig. 11), confirming molecular data. At 15 and 30 days, OC expression decreased compared to those obtained at 7 days. This decrease is related to the osteogenic differentiation that has already happened. However, a significant increase in OC expression was observed for Xgel/HA/BC with respect to Xgel and Xgel/HA (P < 0.05) (Fig. 11a). OPN expression, at 15 days, was higher for  HA and HA/BC-containing gels with the more marked increase for Xgel/HA/BC (Fig. 11b). On the other hand, at 30 days, OPN levels were almost comparable for Xgel/HA and Xgel/HA/BC (Fig. 11b).

Discussion
Considering that safe crosslinking strategies for scaffold production may exploit specific biocatalyst, in the present research work we aimed to incorporate unmodified biologically active macromolecules in enzymatically crosslinked gelatin matrices, for bone regeneration purposes.
The high potential of m-TG gelatin in bone engineering was already demonstrated [31,32]. Combination with HA and with HA/ BC in semi-IPN structures was here evaluated expecting improvements in the osteogenic potential, related to HA bioactivity, and to test, for the first time, the novel unsulfated chondroitin, in bone engineering.
The gel concentration selected for hydrogel production was previously reported for similar matrices and proved appropriate for cell proliferation [11]. In addition, the resulting viscosity allowed for concurrent dissolution of HA and BC. The latter were used in 1:10 weight ratio to gelatin according to the lower heteropolysaccharide content in the ECM, compared to collagen [2,8,55]. 2 U/mL m-TG was selected since allowing for homogeneous distribution of the enzyme within the polymeric solution before gelation, thus ensuring hydrogel homogeneity.
Hyaluronic acid used here, 280 kDa molecular weight, was selected to obtain a certain extent of entanglements with gel, whilst not impairing hydrogel formation due to excessive (high) viscosity. In addition, the medium-low size hyaluronan is known to prompt specific signaling [36]. Gelation studies demonstrated the occurrence of gelatin crosslinking under the selected conditions and, crucially, that HA and BC addition did not hinder neither modify the gelation process. Rheological data on the matrices, immediately after m-TG crosslinking, revealed comparable tan delta values thus suggesting the achievement of comparable gelatin crosslinking extent in the three formulations.
Hydrogels were then characterized supposing to use them as scaffolds to be implanted after crosslinking and cell seeding. However, the ease of handling of the polymeric mixture before crosslinking, the set temperature, the possibility to encapsulate cells during the safe crosslinking process and the gelation time recorded (2-5 fold shorter compared to the ones reported for previously developed similar matrices) are also compatible for a potential use of the matrices as injectable implants.
The biophysical characterization demonstrated that the introduction of HA and BC was responsible for reduced swelling and increased stiffness. Specifically, swelling extent decreased with the increase of the polysaccharide content, due to the semiinterpenetrating structures and the reduction in ionic osmotic pressure, expected in the presence of HA and BC. Matrices proved to absorb water up to 14-17 fold their dried weight, ensuring a pronounced matrix hydration which is beneficial for cell growth and proliferation. The hydrogel water up-take capability is consistent with the values reported for other, diversely crosslinked gelatin matrices for bone/tissue engineering (500-2000% swelling degree) [5,6,9,10]. The time needed for the materials to reach the maximum hydration is also in agreement with literature data [45,56].
Compared to Xgel, the semi-IPN containing HA showed a 40% increase in G 0 , which is coherent with the semi-interpenetrating structure, mechanically reinforcing the gel matrix. No further significant variation was recorded when also BC was added possibly due to its low molecular weight ($40 KDa). As expected for hydrogels, stiffness did not approach to that of native tissue but it was in the range reported for other gel or gel-polysaccharides matrices intended for the same use and obtained using various crosslinking chemistries [5][6][7]. In considering this aspect, we must also be aware that the developed materials are intended to resemble the organic part of the matrix tissue and that improvement can be obtained by incorporating an inorganic part. Further, a good balance between initial rigidity and permeability, more than mechanical properties directly resembling those of the native tissue, is often highly desirable in TE. In fact, even if exhibiting lower rigidity, the higher the scaffold permeability the more homogeneous the newly synthetized matrix distribution will be, thus possibly improving mechanical performance over time.
HA and BC addition did not affect the high chemical stability of m-TG gel under physiological conditions. The sound material stability is in agreement with data on similar m-TG gel networks and comparable to that reported for gelatin-HA scaffolds obtained through EDC-mediated crosslinking [9,12,46]. However, hydrogels proved more stable than glutaraldehyde-crosslinked gel/chitosan scaffold proposed for the same application (a 50% weight reduction was reported for these gels at 21 days of incubation in PBS) [6]. The studies carried out in the presence of collagenase suggest that the presence of the heteropolysaccharides into the hydrated gelatin matrix did not hinder enzymatic degradation thus ensuring resorbability.
Finally, release studies proved a prolonged retention of both HA and BC into the gels, this feature was never reported before. HA and BC persistence in the scaffold (matrices) has a pivotal role in beneficially affecting cell recognition, binding, migration, proliferation and differentiation, and, on the other side, their slow release may result in environmental signaling further guiding the regeneration process. Stem cell (hDPSCs) based in vitro experiments were accomplished to assess this potential effect of HA and HA/BC addition to m-TG gel.
hDPSCs were selected as a cell source for bone forming constructs. They are mesenchymal stem cells able to differentiate in osteoblasts. Our group demonstrated that these cells are able to form highly vascularized bone tissue when cultured on collagen sponges in human mandible bone defects [57,58]. In presence of human serum, they co-differentiate in osteoblasts and endotheliocytes [59]. DPSCs are easily established from dental pulp and have several advantages compared to other types of mesenchymal stem cells (MSCs): better proliferative potential, a simpler primary isolation method, and a higher success rate in long-term in vitro culture [60]. Moreover, DPSCs can be stored for a long time without losing their stemness and ability to differentiate [61,62].
In this study, we seeded DPSCs on the hydrogels obtaining three different constructs: Xgel (control construct), Xgel/HA and Xgel/ HA/BC. The constructs were cultured in osteogenic up to 30 days and their performance toward cell proliferation, ECM deposition and bone differentiation were evaluated. Preliminary experiments indicated that the culture in osteogenic medium was needed to highlight differences among hydrogels in promoting osteogenesis. In fact, many research reports investigated the stem cells osteogenic differentiation process, using scaffold and supplementing media with specific steogenic (or mineralizing) factors [63][64][65][66][67].
SEM observation demonstrated that all the investigated matrices were able to promote DPSCs attachment. In particular, the presence of HA resulted in highly colonized matrices consistently with the well-known HA bioactivity. The diverse cell organization on the matrix containing BC could suggest a diverse interaction of Xgel/ HA/BC gels with cells. In our constructs containing HA/BC, the final content of anionic macromolecules is double with respect to those containing only HA. This could influence interaction with cells as well as Ca 2þ deposition and ECM formation.
TPM observation showed that cells penetrated constructs, already at 7 days of cultures. With increased incubation time, cells were found at greater depth demonstrating the possibility for cells to colonize the bulk of the hydrogels. We supposed that a remodeling process combined with penetration of a porous material may be at the basis of the colonization process. In fact, the comparison of SEM images for seeded and not seeded materials (Fig. 8) suggests morphological extensive changes that are compatible with a remodeling process occurring in the presence of cells. Porosity was also demonstrated ( Supplementary Fig. S4), even if under conditions (materials dehydrated using ethanol) hardly resembling material 3D architecture in the physiological swollen state.
The presence of mineralized matrix observed already at 7 days of culture proved hydrogels' ability to induce early bone differentiation and this is in agreement with previously reported results for mTGgel [31]. Cell differentiation into osteogenic phenotype was corroborated by the expression data for OC and OPN, the major noncollagenous proteins involved in bone matrix organization and deposition, at both gene and protein level.
The semi-IPN hydrogels enhanced osteogenic differentiation with respect to Xgel when analyzed at SEM, considering cell clusters distribution and morphology. This was confirmed by the overexpression of both OPN and OC at 7 days, increased for the semi-IPNs constructs, compared to Xgel alone.
The relative OC and OPN expression is consistent with a more activated osteoblastic differentiation in matrices containing HA and HA/BC. Noteworthy, the hydrogel comprising BC showed higher OC and OPN expression levels within 15 days suggesting a potential role in initial (early) activation of differentiation in osteogenic medium.
When comparing hDPSCs viability on the three formulations, a lower proliferation index was found for the semi-IPNs, especially for the sample containing HA/BC. This finding is consistent with the earlier stem cell differentiation on these materials. It is known that, when cells start the differentiation, they stop the proliferation and exit from cell cycle [68,69].
It is worth underlining that the biological potential of the matrices may be even wider considering the possibility of incorporating cells during hydrogel formation. However, this potentiality was not explored in the present study.

Conclusions
In this study, a new combination of safely enzymatically crosslinked gelatin with unmodified HA and, especially, with the promising unsulfated chondroitin was proposed for bone regeneration. Results demonstrated that the semi-IPNs improved over the matrix containing gelatin alone since exhibiting comparable or higher biophysical performance and, especially the HA/BC-containing hydrogel, the ability to prompt earlier hDPSCs differentiation into osteoblasts. BC potential as a scaffold component in view of the specific application was, for the first time, demonstrated. The results achieved sustain the possibility to consider the developed hydrogels as a valid alternative to previously proposed gel-matrices, and give the basis for further pre-clinical studies for bone regeneration.

Author contribution
Annalisa La Gatta: Conceptualization, methodology, investigation, formal analysis, data curation and validation, visualization, supervision and project administration related to the development and biophysical characterization (hydration capacity, rheological behavior, stability, release studies) of the hydrogels; Writing -Review & Editing. Virginia Tirino: Conceptualization, investigation, formal analysis and data curation and validation, visualization, supervision and project administration related to the biological characterization of materials' interaction with stem cells, especially referring to hDPSCs isolation and characterization and to the preparation and the culture of the constructs and the analyses of the biomarker (OPN) expression by western blotting; Writing, Writing -Review & Editing. Marcella Cammarota: Investigation, data curation and validation, formal analysis and visualization related to the SEM observation; Writing. Marcella La Noce: Investigation (characterization of the interaction of hDPSCs with the hydrogels, especially referring to the preparation and the culture of the constructs and the analyses of the biomarker (OPN) expression by western blotting). Antonietta Stellavato: Investigation, data curation and validation, formal analysis and visualization for the biological characterization of the hydrogels/hDPSCs constructs, especially referring to the gene expression (OCN and OPN) experiments; writing. Anna Virginia Adriana Pirozzi: Investigation, data curation and validation, formal analysis and visualization related to the characterization of the interaction of hDPSCs with the hydrogels, especially referring to the analyses of the biomarker (OCN) and OPN (7 days) expression by western blotting; writing. Marianna Portaccio: investigation, data curation and validation, formal analysis and visualization related to the constructs' observation at the two-photon microscope and to the FT-IR analyses. Nadia Diano: Investigation, data curation and validation, formal analysis and visualization related to the constructs' observation at the two-photon microscope; funding acquisition. Luigi Laino: Resources (provision of dental pulps from selected patients according to the ethical committee for hDPSCs extraction). Gianpaolo Papaccio: Validation, supervision and project administration related to the biological characterization of materials' interaction with stem cells, especially referring to hDPSCs isolation and characterization and to the preparation and the culture of the constructs and the analyses of the biomarker (OPN) expression by western blotting; Writing -Review & Editing, funding acquisition. Chiara Schiraldi: Conceptualization, formal analysis, validation, supervision and project administration related to the development and biophysical characterization of the hydrogels and to the biological studies via SEM, gene expression (OCN, OPN) and protein expression (OCN); Writing -Review & Editing, funding acquisition.