Design and development of a hepatic lyo-dECM powder as a biomimetic component for 3D-printable hybrid hydrogels

Bioprinting offers new opportunities to obtain reliable 3D in vitro models of the liver for testing new drugs and studying pathophysiological mechanisms, thanks to its main feature in controlling the spatial deposition of cell-laden hydrogels. In this context, decellularized extracellular matrix (dECM)-based hydrogels have caught more and more attention over the last years because of their characteristic to closely mimic the tissue-specific microenvironment from a biological point of view. In this work, we describe a new concept of designing dECM-based hydrogels; in particular, we set up an alternative and more practical protocol to develop a hepatic lyophilized dECM (lyo-dECM) powder as an ‘off-the-shelf’ and free soluble product to be incorporated as a biomimetic component in the design of 3D-printable hybrid hydrogels. To this aim, the powder was first characterized in terms of cytocompatibility on human and porcine mesenchymal stem cells (MSCs), and the optimal powder concentration (i.e. 3.75 mg ml−1) to use in the hydrogel formulation was identified. Moreover, its non-immunogenicity and capacity to reactivate the elastase enzyme potency was proved. Afterward, as a proof-of-concept, the powder was added to a sodium alginate/gelatin blend, and the so-defined multi-component hydrogel was studied from a rheological point of view, demonstrating that adding the lyo-dECM powder at the selected concentration did not alter the viscoelastic properties of the original material. Then, a printing assessment was performed with the support of computational simulations, which were useful to define a priori the hydrogel printing parameters as window of printability and its post-printing mechanical collapse. Finally, the proposed multi-component hydrogel was bioprinted with cells inside, and its post-printing cell viability for up to 7 d was successfully demonstrated.


Introduction
Nowadays, 3D printing is one of the most promising methods to develop advanced 3D in vitro models (Molander et al 2022) for the liver both to study pathophysiological disease mechanisms related to this organ (Ma et al 2018) and to test new drugs (Mazzocchi et al 2019, Bouwmeester et al 2021).Indeed, this technique enables the spatial deposition of one or more hydrogels containing different cell types in a precise and controlled manner, mimicking in this way its intrinsic complexity in terms of To reproduce its unique biochemical composition, a broad spectrum of hydrogel formulations has been investigated in liver bioprinting (Ye et al 2019).In particular, the formulation of hybrid hydrogels, i.e. based on more than one single component, that can be natural or synthetic polymers and other functional moieties for mediated cell response, has become a smart strategy for designing more performant bioinks as it allows to overcome the main limitations of a single material by exploiting the advantages of the other elements (Cai et al 2021, Kim et al 2022).For instance, combining sodium alginate with gelatin results in a printable hydrogel with good mechanical, rheological, and biological properties.However, these two biomaterials are not fully representative of the complex micro-environment of the liver, mainly because of the absence of native bioactive ECM factors.For this reason, decellularized extracellular matrix (dECM) has been introduced as a key ingredient in the design of new hydrogels to fill this biological lack.
Indeed, decellularization is a technique that allows the removal of the cellular and nuclear components by preserving the ECM, a complex network formed by different functional and structural biomolecules, such as proteins (e.g.collagen, laminin, and fibronectin), glycosaminoglycans (GAGs), glycoproteins, and growth-factors that deeply influence cell behavior in vivo (Choudhury et al 2018, Kim et al 2020a).
Once the decellularized tissue is obtained, it must be processed to be printed, and currently, in literature, there are two different ways to develop dECM-based hydrogels (figure 1(a)).
The first and more utilized protocol is based on the solubilization of decellularized tissues by achieving, in the end, a hydrogel (Pati et al 2014).However, this protocol has the main limitation of denaturing the biochemical structure of the tissue when the digestion is over-prolonged in time (Pouliot et al 2020) and leading to the formation of hydrogels with low viscosity and slow crosslinking rate, two nonideal features especially if the hydrogel is conceived for the extrusion-based 3D printer.Regarding this protocol, the first work applied to the liver is described by Lee et al (2017), which obtained a bioink with enhanced stem cell differentiation and HepG2 cell functions in comparison to those of commercial collagen bioink.However, because of its low viscosity and mechanical properties, it was printed together with a polycaprolactone (PCL) framework.Multicomponent approach and similar improved biological results with respect to the considered controls were shown by other studies where the liver dECM hydrogel was mixed with other natural biomaterials, for example with gelatin and silk-fibroin (Sharma et al 2021), increasing the general viscosity of the final bioink, or with gelatin methacrylate (GelMA) (Ma et al 2018), obtaining in this way a photo-crosslinkable bioink whose mechanical properties could be tuned depending on the UV exposure time.
Differently, the second approach consists of mechanically crushing the decellularized tissue, preventing the degradation of ECM biological components and obtaining microparticles, that, however, show as the main drawback to be insoluble and thus not able to fully merge the 3D fibrous network when loaded within another hydrogel (Kara et al 2022).In the field of liver bioprinting, Kim et al (2020b) were the first to employ this protocol, adding the liver microparticles inside a gelatin-based hydrogel.Besides improving the shape fidelity of the 3Dprinted constructs compared with the conventional liver dECM bioink obtained with protocol A and with only gelatin, even in this case, the presence of the microparticles allowed the maintenance of high cell viability and enhanced cell function.
In this work, we propose an alternative and more practical protocol (figure 1(b)) for the development of a lyophilized dECM (lyo-dECM) powder derived from decellularized porcine liver as a ready-to-use and free soluble product, suitable to be incorporated as a biomimetic component in the design of 3D-printable hydrogels.First, a biological characterization of this product was performed regarding cytocompatibility, immunogenicity, and efficacy in sustaining the enzymatic activity.Then, as a proof-of-concept, the lyo-dECM powder was added to a sodium alginate/gelatin blend, a well-established formulation in the literature.The obtained multicomponent hydrogel was characterized from a rheological point of view, and its printing parameters, printability and mechanical stability were studied with the support of computational models.Finally, the cell-embedded hydrogel was printed, and cell viability was assessed.

Lyo-dECM powder: development
The development of lyo-dECM powder consisted of three main steps: decellularization, digestion and lyophilization, as shown in figure 1(b).
All procedures were performed at room temperature in sterility.Regular changes of decellularization detergents were performed to ensure the effective removal of cellular material.Acellular liver scaffolds were stored in 1% amoxicillin/clavulanic solution at 4 • C.

Protocol for lyo-dECM powder preparation
Decellularization tissue pieces were partially dried on gauze and then weighed.The tissue was subsequently suspended in 0.1 M hydrochloric acid (HCl) solution (Sigma-Aldrich, United States) with a final concentration equal to 2% w/v and digested with pepsin (Sigma-Aldrich, United States) for 24 h using a magnetic stirrer set at 37 • C. Two different amounts of pepsin were evaluated, equal to 10% w/w and 20% w/w of the total weight of the wet liver tissue to be processed.Then, the pH of the digested solution was adjusted to 7.4 with 1 N sodium hydroxide (NaOH) solution (Sigma-Aldrich, United States) and freezedried after adding mannitol (0.5% w/v) as a cryoprotectant (Bari et al 2018).The solution was frozen at −80 • C and freeze-dried (Christ Epsilon 2-16D LSCplus) at 8 × 10 −1 mbar and −50 • C for 72 h.The obtained powder was stored at −20 • C until use.

Lyo-dECM powder: biological characterization
The developed lyo-dECM powder was characterized in terms of cytocompatibility, in vitro immunogenicity, and anti-elastase activity.

Cytocompatibility test
The cytocompatibility and proliferation ability of dECM powder were evaluated on porcine bone marrow mesenchymal stem cells (pMSCs, passage 4) and human adipose mesenchymal stem cells (hMSCs, passage 10).
Both these cell types were seeded in a 96-well plate (5000 cells cm −2 ) and cultured with DMEM/F12 supplemented with 10% v/v fetal bovine serum (FBS), 100 U ml −1 penicillin, 100 µg ml −1 streptomycin, 0.25 µg ml −1 amphotericin, 4 mm glutamine, 1 mm sodium pyruvate.After 3 d from cell seeding, the supernatants were discarded and replaced with 100 µl of culture medium (not supplemented with FBS) containing dECM powder at seven different concentrations (0.23, 0.5, 0.9, 1.9, 3.7, 7.5, and 15 mg ml −1 ).Untreated cells (i.e.cultured without dECM powder) were considered as control, i.e. accounting for 100% of metabolic activity.Qualitative analyses based on cell morphology, confluence and eosin counting were performed after 24 and 48 h of incubation.An MTT assay was performed after 3 and 7 d of incubation.At each time, supernatants were discarded, and 100 µl of MTT solution (0.5 mg ml −1 ) was added to each well.After 3 h of incubation, the MTT solution was removed, and 100 µl of DMSO was added.The absorbance was measured by a microplate reader (Synergy HT, BioTek, Swindon, UK) at 570 nm and 670 nm (reference wavelength).The cell metabolic activity % was calculated as follows (equation (1)): Metabolic activity (%) = 100 × ( Abs sample /Abs ctr ) (1) where Abs sample is the mean value of the measured optical density of the tested samples and Abs ctr is the mean value of the measured optical density of cells not incubated with dECM powder.Each condition was tested in triplicate.

Anti-elastase activity evaluation
The anti-elastase activity was determined in conditions of inhibition of the enzyme activity induced by the lyophilized secretome from MSCs (lyosecretome), which contains potent inhibitors of this activity (Bari et al 2019).In detail, the anti-elastase activity was evaluated considering different final concentrations of dECM powder in the final reaction mix (20, 10, 5, 2.5 mg ml −1 ).This activity was determined in the presence and absence of lyo-secretome at concentrations of 20, 10, 5, and 2.5 mg ml −1 .Briefly, porcine pancreatic elastase (PPE) was solubilized in phosphate buffer pH 6.8 (0.5 IU ml −1 ).The substrate N-Succinyl-Ala-Ala-Ala-Ala-p-Nitroanilide was diluted in TRIS buffer to obtain a final concentration of 0.41 mm.Each sample was incubated for 20 min with the enzyme, and subsequently, the substrate was added.Their kinetic reaction was monitored by spectrophotometric analysis (Victor Nivo, PerkinElmer, Waltham, MA, USA) at 410 nm for 60 min (one measurement each minute).The reaction mix without the sample was used as a negative control, while EGCG was considered a positive control (concentration tested: 7.2 mg ml −1 ).All analyses were performed in triplicate, and the results are reported as anti-elastase activity percentage, calculated using the following equation ( 2 where A ctr is the absorbance of the negative control and A sample is the absorbance of the sample.

Hydrogel preparation
Sodium alginate (Alg; viscosity 5-40 cps, product number W201502) and gelatin (Gel; derived from bovine skin, product number G9391) powder were purchased from Sigma Aldrich.The hydrogels listed in table 1 were prepared.The blend of Alg and Gel was prepared according to protocols previously implemented (Delgrosso et al 2023), with a final concentration equal to 80 mg ml −1 and 40 mg ml −1 , respectively.Gel was first dissolved into a sterile phosphatebuffered solution (PBS, Euroclone) previously heated at +72 • C. Alg was then added to the Gel solution, and the following blend was mixed and homogenized on a rotational shaker for 5 min and pasteurized in a thermostatic water bath at +72 • C for 1 h.In the case of the presence of lyo-dECM powder within the hydrogel (GEL3 and GEL30), this was added to the Alg/Gel blend after the pasteurization process and following cooling.Finally, they were transferred into a syringe and kept in the fridge at 4 • C overnight before its use.C, in which the shear rate was kept constant at 0.1 s −1 for 60 s (step I), at 500 s −1 for 5 s (step II), and again at 0.1 s −1 for 60 s (step III).Each test for each hydrogel was repeated twice.

3D printing system
The 3D printing process was realized using a pneumatic-based 3D bioprinter (INKREDIBLE+, CELLINK, SE).The printing process starts with a 3D virtual geometry translated into machine instructions (i.e.g-code) through a slicing software (Slic3r), which generates the coordinates and the printing speed of the printing head in each layer.Moreover, in the g-code, other parameters, such as layer height, perimeter, infill percentage, and pattern, are defined.The printing head positioning system has a 10 micron resolution in the three axes.Briefly, the process involves three steps: (a) XYZ homing axes to position the printhead in the middle of the print bed; (b) Z axis calibration to tune the distance between the nozzle and the printing bed properly; (c) the pressure calibration, which is manually set up through a lateral knob present in the bioprinter to find the optimal pressure value that enables the proper material extrusion.

Computational analysis as support for 3D printing assessment
Computational models were implemented using the commercial software Multiphysics ® software (COMSOL Inc., US) to simulate the hydrogel when extruded through the bioprinting system and its mechanical deformation under the effect of gravity force.

Computational modeling of extrusion bioprinting process
Computational fluid dynamics (CFD) simulations were conducted to simulate the hydrogel extrusion throughout the bioprinting process through an extrusion-based 3D bioprinter (figure 2(a)).All the geometries were designed with Inventor ® (Autodesk, US) and imported into COMSOL for computational analyses.In particular, the whole extrusion system, including cylindrical and conical nozzles with inner diameters equal to 200 µm, 250 µm, and 400 µm, were reproduced as shown in figure 2(b).To reduce the computation time by exploiting the axisymmetric property of the extrusion system, a 2D axisymmetric space was used in the simulation (figure 2(c)).As boundary conditions, constant pressure was imposed as inlet, the boundaries on the axis of symmetry were set as 'axial symmetry,' the tip of the nozzle R1 was set as an outlet with no pressure, and the remaining external boundaries were provided as 'no slip' condition.An automatic extra-fine triangular mesh, with refined mesh (8 layers mesh type) close to nozzle walls to capture the sharp velocity gradients near the wall, was used in all the simulations.Due to its non-Newtonian behavior, the constitutive model used to define the hydrogel behavior in the simulations was described by the power law of Ostwald-de Waele (equation ( 3)): where η is the viscosity (Pa s), γ is the shear rate (s −1 ), K is the flow consistency factor (Pa s), and n is the flow behavior index (-).For each composition, K and n were derived from the logarithm curve obtained from the shear rate ramp (i.e.shear rate vs. viscosity) by applying a linear fitting.
Pressure distribution, flow velocity field, and the associated shear stress profile in each geometry were computed.The average velocity at the nozzle outlet (v out ) was calculated as half of the maximum computed velocity value.

Computational modeling of 3D-printed structure deformation
The computational analysis was implemented to evaluate the collapse over time of a 3D-bioprinted strand with a thickness equal to 0.9 mm along different gap lengths (i.e. 6, 8, 10 mm) due to gravity.The structural mechanical module was used, implementing a 2D model and performing a transient analysis between 0 and 100 s.The hydrogel was described as a viscoelastic material with the Kelvin-Voigt equation, which is suitable for modeling the creep response, i.e. when a deformation under a constant load, such as gravity, is considered.Kelvin-Voigt model consists of a Hookean elastic spring and a Newtonian dumper placed in parallel and whose stress (σ)-strain (ε) relation can be written as follows (equation ( 4)): where ε (s −1 ) is the deformation rate, E (Pa) is the elastic modulus describing the Hookean elastic spring, and η 0 (Pa s) is the viscosity at rest, describing the Newtonian dumper.The parameters E and η 0 were obtained from the rheological characterization of the hydrogel: η 0 was set equal to the hydrogel viscosity at low shear strain obtained from the shear rate ramp test; E was estimated with the following equation ( 5): where υ is the Poisson ratio set equal to 0.49 due to the high content of water in the hydrogels and G ′0 was derived from the storage modulus of the hydrogel recorded in the frequency sweep test by fitting the log-log curve with a linear regression and considering its intersection with the Y-axis.The density of the material was set equal to 1000 kg m −3 due to its high water content.The gravity force was applied as a volumetric load in the domains.Finally, an automatic extra-fine triangular mesh was used.

Shape evaluation of the 3D-printed geometries
All printings were performed at room temperature (25 • C) using a conical nozzle of 0.41 mm.A Petri dish was used as printing support.Three samples for each structure were printed.The images were processed using ImageJ software (NIH, USA).Firstly, the printing of a single-layer and continuous strand of material following a path with decreasing inter-filament distance (i.e.2.5, 2, 1.5, and 1 mm) was performed to analyze the combination of the printing velocity and pressure in the strand size and inter-filament resolution.
Once these values were defined, grid structures with two different infills (i.e.10% and 15%) were printed to quantify the shape fidelity of the pores.Printability factor (Pr) was used as a parameter to describe the precision in printing structures with porosity, and it is calculated as a function of the pore perimeter (P) and pore area (A) (equation ( 6)): Pr was calculated for each pore within the same grid structure and then averaged.
Finally, to evaluate the strand collapse, we printed a continuous strand over a supporting structure composed of increasing gap distances (i.e. 6, 8, and 10 mm).

Printing cell-embedded GEL3 hydrogel
hMSCs were detached with Trypsin/EDTA, counted, centrifuged at 1200 rpm for 10 min, resuspended in medium respecting the 1:10 (medium: hydrogel) ratio, and gently mixed with GEL3 hydrogel by using two syringes joined by a luer-lock to form the bioink.A final cell density equal to 2 × 10 6 cells ml −1 of hydrogel was used.The bioink was loaded in a cartridge, and short-run centrifugation was performed to compact the bioink.The cartridge was connected to a 0.41 mm conical nozzle and placed in the bioprinter.After the printing, the 3D-printed constructs were crosslinked for 5 min with a CaCl 2 (0.5% w/v, Sigma Aldrich) solution.Then, the crosslinking solution was removed and replaced with DMEM/F12 supplemented with 10% v/v fetal bovine serum (FBS), 100 U ml −1 penicillin, 100 µg ml −1 streptomycin, 0.25 µg ml −1 amphotericin, 4 mM glutamine, 1 mm sodium pyruvate.The medium was changed every 3 d.All procedures were performed in sterility.

Cell viability in 3D-printed constructs
The viability of hMSCs cultured inside GEL3 hydrogel was assessed at 1, 3, and 7 d after the 3D printing process by applying a live/dead cell viability kit (LIVE/DEAD™ Cell Imaging Kit (488/570), ThermoFisher Scientific, Milano, Italy) on 3 printed samples for 15 min at 25 • C. Live and dead cells were visualized with a fluorescence microscope (Leica DM IL LED, Leica Microsystems, Milano, Italy).The images were processed using ImageJ software (NIH, USA).

Statistical analysis
Raw data were processed using Statgraphics XVII (Statpoint Technologies, Inc., Warrenton, VA, USA).A generalized linear analysis of variance (ANOVA) model was used, followed by Fisher's least significant difference (LSD) procedure to estimate the differences between the means (n = 3).In detail, regarding the raw data about cell metabolic activity, for each cell line, the sample concentration and time were considered as fixed factors, and the cell metabolic activity (%) was the response variable.The anti-elastase activity data were elaborated considering the sample, concentration, and time as fixed factors and the activity (%) as the response variable.The statistical significance was set up at p < 0.05.

Lyo-dECM powder development
Starting from our previous work (Croce et al 2022) in which porcine liver tissue was successfully decellularized and its influence on MSC differentiation was demonstrated, we set up a new protocol for the final obtainment of a dECM-based hydrogel to be used in the field of the bioprinting.As summarized in figure 1(b), the protocol consists of the digestion of the decellularized tissue using a pepsin-HCl solution for 24 h until a solution is obtained.Initially, two different amounts of pepsin were assessed, i.e. 10% w/w and 20% w/w, with respect to the tissue weight to be digested.Both concentrations could digest the tissue into a solution; however, the 20% w/w pepsin was selected as it allowed a faster digestion, thus with better chances of preserving ECM components.The digested solution was then frozen and lyophilized, achieving a white and water-soluble powder (lyo-dECM powder) at the end of the process.The dECM powder, when dissolved in water, medium or PBS at concentration up to 30 mg ml −1 , resulted in being too liquid to be printable alone and, for this reason, from this evidence, we considered its application as component to be added within other hydrogels.

Lyo-dECM powder biological characterization
The lyo-dECM powder was at first characterized in terms of cytocompatibility.To this end, different concentrations (15, 7.5, 3.7, 1.9, 0.9, 0.5, 0.23, 0 mg ml −1 ) of lyo-dECM were tested on MSCs from human or porcine origin also to collect indications about xenogenic use.After 24 h, a qualitative analysis regarding cell morphology and confluence degree was performed: both hMSCs and pMSCs morphology resulted in adherent and with the typical fusiform shape along all the concentrations, meaning there is not  3).For pMSCs, a dose-dependent decrease in cell metabolic activity was observed for both the times considered (p < 0.001).The cell metabolic activity was above 70% up to 1.9 mg ml −1 and 3.75 mg ml −1 after 3 and 7 d of treatment, respectively.For hMSCs, the effect of both lyo-dECM concentration and treatment time was significant (p < 0.001).After 3 d of treatment, the cell metabolic activity increased for concentrations up to 1.9 mg ml −1 and then decreased, becoming cytotoxic at 15 mg ml −1 .After 7 d of treatment, instead, the increase in cell metabolic activity was less marked, and a cytotoxic effect was even in this case observed only for the highest concentration tested (15 mg ml −1 ).Considering the increased metabolic activity observed on hMSCs, investigations were conducted to rule out any potential immunogenic effect of lyo-dECM.As shown in table 2, the proliferation of human lymphocytes was not stimulated by the presence of lyo-dECM at different concentrations (7.5, 15, and 30 mg ml −1 ), but it can be noted that, as the concentrations increase, SI tends to decrease indicating cytotoxicity.The powder is, therefore, not immunogenic but toxic at high concentrations between 15 and 30 mg ml −1 , as also demonstrated with cell viability tests on MSCs.
Finally, the ability of lyo-dECM to support the enzymatic activity was assessed.Firstly, the antielastase activity of lyo-dECM was determined at different final concentrations in the reaction mix (20, 10, 5, and 2.5 mg ml −1 ); at none of the tested concentrations, lyo-dECM inhibited the elastase enzyme activity.Subsequently, the anti-elastase activity of lyosecretome (the secretome of human MSCs containing alpha-1-antitrypsin and other potent protease inhibitors) was assessed at the same concentrations but in the absence of lyo-dECM.According to what was reported in our previous work (Bari et al 2018), the lyo-secretome alone exhibited a dose-dependent antielastase activity: 0% for 2.5 mg ml −1 , 10% ± 0.985 for 5 mg ml −1 , 38% ± 2.965 for 10 mg ml −1 , and 79% ± 3.236 for 20 mg ml −1 , which was not significantly different from the positive control.However, in the presence of lyo-dECM (even at 2.5 mg ml −1 ), the anti-elastase activity of lyo-secretome was completely inhibited.

Rheological characterization of hydrogels containing lyo-dECM powder
We analyzed the influence of lyo-dECM powder in the rheological behavior of the hydrogel in which it was included, which was, in our case, a blend of Alg (80 mg ml −1 ) and Gel (40 mg ml −1 ).The lyo-dECM powder was evaluated at two different concentrations: at 3.75 mg ml −1 (GEL3), which is the tolerated threshold found from cell toxicity tests, and at 30 mg ml −1 (GEL30) to evaluate an eight times higher amount.As a control, Alg/Gel was considered.
First, the viscoelastic behavior (i.e.storage modulus G ′ and loss modulus G ′′ ) of these formulations was studied in relation to different variables (i.e.frequency, temperature, and crosslinking time).However, before evaluating this, defining the linear viscoelastic region (LVR) in which G ′ and G ′′ are independent of imposed strain values was necessary.So, a strain sweep test was conducted for each formulation, and a common linear trend was observed until 10% of strain, which was chosen as extreme of the LVR and set for the following sweep tests (figures 4(a), (e) and (i)).
The frequency sweep test gives information about liquid-like and solid-like behavior.Alg/Gel hydrogel presented a solid-like behavior as shown by the phase angle values lower than 45 • and G ′ consistently higher than G ′′ at increasing frequency, thus demonstrating a dependency on the frequency values (figure 4(b)).The same behavior was presented by GEL3 hydrogel (figure 4(f)).The formulation GEL30 hydrogel has the same trend (i.e. the two moduli increased at the increasing frequency) but with G ′′ always greater than G ′ , describing, in this case, a liquid-like behavior, confirmed by phase angle oscillating from about 65 • to 55 • for increasing frequency (figure 4(l)).
Afterward, G ′ and G ′′ were recorded as a function of temperature (figures 4(c), (g) and (m)), allowing to assess possible physical crosslinking given by temperature change and, consequently, to evaluate any possible gelation point (i.e.G ′ = G ′′ ), at which the transition from the initial dominant liquid-like behavior to gel-like behavior (or vice versa) happens.Because of the presence of the gelatin in the composition, Alg/Gel hydrogel presents at low temperatures the storage modulus G ′ higher than the loss modulus G ′′ , with a switch in these moduli after the gelation point (equal to ∼32 • C), passing in this way to a liquidlike behavior.As in the previous test, the same trend of Alg/Gel was observed in GEL3.Differently, GEL30 showed the loss modulus G ′′ consistently higher than the elastic modulus G ′ within the entire temperature range, and consequently, no gelation point was detected.Similarly to the temperature sweep analysis, the time sweep test allows studying the gelification point but as a function of time due, for example, to a chemical crosslinking mechanism, like in our case with the addition of a CaCl 2 -based solution at 60 s after the start of the test.Switching from a liquidlike to a solid-like state happened in a few seconds (∼15 s) in the Alg/Gel and GEL3 hydrogel, as visible in figures 4(d) and (h).Differently, the gelification process was slowed down about ten times when the dECM powder was present at a high concentration (figure 4(n)).
Finally, since the hydrogel is meant to be printed, the viscosity of each formulation was characterized as a function of the shear rate, which is a characteristic variable of the bioprinting process, through two types of analyses, i.e. shear rate ramp and thixotropic test.In the shear rate ramp test (figure 5(a)), GEL30 hydrogel (green curve) resulted in being slightly less viscous than Alg/Gel and GEL3 (red and blue curves, respectively) between 0.1 and 10 s −1 .Moreover, all the formulations exhibited shear-thinning behavior as the viscosity decreased at increasing shear rate values.The degree of shear-thinning behavior, expressed as a power-law index (n), was quantified by fitting viscosity-shear rate data with Power law model: Alg/Gel, GEL3, GEL30 hydrogels showed values equal to 0.256, 0.206, and 0.327, respectively, with a good linear fit (R 2 = 99% for all the formulations), indicating that all hydrogels possessed high shear-thinning attributes.
The thixotropy test (figure 5(b)) was divided into three steps to mimic the inks in the cartridge before the printing (step I, low shear stress), the ink during the extrusion process (step II, high shear stress), and the ink at the steady state after the extrusion (step III, low shear stress).Viscosity was measured at all steps: in accordance with the shear rate ramp test, at the low shear rate, the viscosity has higher values than at a high shear rate because of the hydrogel shear-thinning property.However, for Alg/Gel and GEL3, in the final step, it takes a long time to recover the initial viscosity shown in step I. Indeed, after 60 s, they reached values around 400-450 Pa s −1 , lower with respect to the plateau value (2800-3000 Pa s −1 ) of step I. Similarly to what was observed in the shear rate ramp, GEL30 (green curve) presents a slower viscosity, mainly in step I, but with a complete recovery of the initial viscosity in step III.

Printing assessment with the support of computational analyses
Since 3.75 mg ml −1 was selected as the optimal concentration for the final bioink formulation from the biological characterization, we studied the printing parameters and performance of GEL3 hydrogel.This was done without following a classic trial-and-error approach but implementing computational simulations not to waste lyo-dECM powder.Table 3 summarizes material parameters derived from rheological tests (shear rate ramp and frequency sweep tests) and utilized as input for computational models.
We simulated the hydrogel extrusion process through CFD simulation for a range of nozzle sizes (i.e. 200, 250, and 410 µm) and geometries (i.e.conical and cylindrical), and printing pressures (i.e. 10,20,30,40,and 50 kPa) to control the hydrogel extrusion and quantitatively estimate the associated shear stress values developing within the nozzle during the printing process.Indeed, from the computed flow velocity field and shear stress distribution within the extrusion system (figure 6(a)), we derived the average extrusion velocity and the maximum shear stress at the nozzle outlet (v out and τ wall , respectively) for each of the printing conditions (figure 6(b)).Based on our expertise, we highlighted in blue and yellow the regions in which the printing velocity was lower than 1 mm s −1 or higher than 60 mm s −1 , respectively, and in red where τ wall was closed or exceeding 2500-3000 Pa, critical values for cell viability, particularly for MSCs, as reported in the literature (Koch et al 2020).The remaining white sections define the window of printability that led us to exclude cylindrical nozzles for GEL3 hydrogel printing and to limit the printing pressure between 30 and 50 kPa.
As validation of the so-defined printing window, we printed serpentine structures by selecting, as an example, the blue conical nozzle (i.e.0.41 mm).In the g-code, we set up a printing velocity equal to 5.8 mm s −1 and tested three different pressures: 30 kPa (i.e. the target value), and 10 kPa and 50 kPa, corresponding to a lower and higher value than the target value, respectively.As expected, the best resolution of the printed line was obtained with a pressure equal to 30 kPa (thickness = 0.76 mm ± 0.08) compared with 50 kPa (thickness = 1.19 mm ± 0.1), while with a pressure of 10 kPa, it was not possible to extrude the hydrogel properly (figure 6(c)).Then, by applying the selected printing parameters (pressure = 30 kPa, feed rate = 5.8 mm s −1 ), we evaluated the geometric accuracy of the pores of two different structures with an infill equal to 10% and 15% in terms of printability index (Pr), which was equal to 0.95 ± 0.03 and 0.92 ± 0.04, respectively (figure 6(d)).
Moreover, in the aim of printing assessment, we studied the structural collapse of 3D-printed lines due to the gravity force, considering different length gaps (i.e. 6 mm, 8 mm, and 10 mm) (figures 7(a) and (b)).As expected from the viscous nature of the material, the collapse is more marked, passing from lower to higher distances, with displacement equal to 1.25 mm ± 0.07, 2.3 mm ± 0.1, and 3.7 mm ± 0.3, respectively.As shown in figure 7(c), the following experimental measurements resulted in line with the results obtained from computational simulations, which were also useful to better visualize the time range, equal to 10-20 s after its deposition, in which the deformation results to be stable at a plateau value (figure 7(d)).

Cell viability in 3D-printed constructs
After defining the printing parameters and the collapse entity, we printed GEL3 hydrogel with cells embedded inside.A 2-layers grid structure (figure 8(a)) was designed to allow a better diffusion of nutrients and oxygen inside the cell-laden constructs.Cell viability was qualitatively estimated by applying a live/dead assay on day 1, day 3, and day 7 after the printing process (figure 8(b)).From this observation, high cell viability was observed at each time point, leading us to conclude that the developed bioink was cytocompatible.

Discussion
Decellularized tissues are, to date, the golden standard to closely mimic the complex biochemical microenvironment experienced by cells in vivo.Following this consideration, we aimed to develop a dECMbased hydrogel for liver bioprinting applications.We started from our previous work (Croce et al 2022), in which porcine liver tissue was demonstrated to be successfully decellularized and positively influence MSC differentiation, and we defined an alternative protocol able to overcome the main limitations of the ones currently reported in the literature.
Indeed, in the literature, there are currently two ways to develop dECM-based hydrogels (figure 1(a)).The first and more utilized protocol consists of dECM solubilization, i.e.ECM is lyophilized, mechanically crushed into powder form, and enzymatically digested at low pH by achieving a hydrogel (Pati et al 2014, Lee et al 2017, Ma et al 2018).Differently, the second approach consists of producing dECM microparticles using mechanical methods (e.g. by freeze-milling) without solubilizing the tissue; the microparticles are then loaded within another hydrogel (Kim et al 2020b, Kara et al 2022, Guagliano et al 2023).However, the first approach leads to hydrogels with low viscosity, slow crosslinking speed, and low mechanical properties, and, as it requires a long time (normally 48-72 h), some biochemical components of dECM could be denatured (Pouliot et al 2020).Furthermore, we found it tricky to replicate it for the hepatic tissue since, after the lyophilization step, the decellularized tissue resulted being very rubbery, probably due to its high elastic content (Carter et al 2001, Skardal et al 2015, Abaci and Guvendiren 2020).With the second approach, instead, the microparticles obtained are insoluble and thus unable to mimic the complex biochemical microenvironment because they do not fully merge the 3D network of the hydrogel in which they are incorporated; furthermore, they can potentially block the correct extrusion  of the bioink through very thin nozzles during the printing process.
Based on these listed drawbacks, we designed an easy and short protocol that did not imply any specific instruments and that was scalable and practical to use.We optimized the pepsin amount (i.e.20% w/v) to speed the digestion of the decellularized tissue inside the HCl solution to only 24 h, and consequently, to prevent possible denaturation of the dECM.The achieved solution was subsequently lyophilized, following a procedure and using cryo/lioprotectants selected from our previous studies in which it was demonstrated they were able to preserve the biological components (proteins and lipids) of complex mixtures, such as the secretome from MSCs (Bari et al 2018).This protocol enabled us to obtain a soluble and 'ready-off-the-shelf ' dECM powder, i.e. a powder that can be stored at −20 • C and added within other hydrogels, as biomimetic component, just before the printing process, and with the additional feature to form the network in which cells are embedded.
To characterize lyo-dECM powder and to optimize the protocol, the effect on cell viability in response to the presence of the dECM powder dissolved in culture medium was evaluated by performing qualitative (morphological evaluations) and quantitative (MTT) tests on human and porcine MSCs, to have a view of its effect on different cell origins.In general, the results confirmed a cytotoxic effect of the lyo-dECM powder at high concentrations (i.e.>15 mg ml −1 ) and allowed us to determine the concentrations that are most beneficial in terms of improving the biocompatibility of both kinds of cells to the matrix by defining at the end a tolerance threshold equal to 3.75 mg ml −1 to use in the final hydrogel formulation.This concentration value is in line with the dECM amount used in other scientific works present in literature to develop dECM-based hydrogels (Hiller et al 2018, Giobbe et al 2019, Kort-Mascort et al 2021, Sharma et al 2021).Furthermore, we demonstrated that lyo-dECM powder was not immunogenic as it did not induce in vitro proliferation of human PBMCs because of the absence of cell components in the final product, highlighting the efficiency of the decellularization step.We also investigated the potential presence of in vitro biological activity of the lyo-dECM powder in terms of anti-elastase, which is the ability of the extracellular matrix to reactivate elastase activity under conditions of inhibition.As far as we know, no studies investigating these properties on dECMderived powder/hydrogel have been published until now.We suppose that this is due to a conformational change of the enzyme, induced by the proteins of the matrix, which reduces the binding affinity with the inhibitor (i.e.lyo-secretome) but not with the substrate.Indeed, the hepatic extracellular matrix is a complex network of macromolecules that provides an extracellular scaffold and plays an important role in regulating cellular and enzymatic activities (Hui and Friedman 2005).Especially in the liver, enzymes in the endoplasmic reticulum of hepatic cells protect the body from accumulating fat-soluble exogenous and endogenous compounds, converting them into water-soluble metabolites that kidneys can efficiently excrete.It follows that the extracellular matrix of the liver tissue could contribute to creating a microenvironment in which enzymes, including probably also elastase, can function at their maximum activity, with a certain tolerance to inhibitory mechanisms.
Once the concentration value tolerated by cells was established, we defined an example of hydrogel formulation in which the lyo-dECM powder could be incorporated.We selected a blend of Alg/Gel as proof-of-concept; this choice was made according to the literature, as these two biomaterials are widely used in the bioprinting field and for the development of in vitro models (Hiller et al 2018, Fantini et al 2019, Kort-Mascort et al 2021, Roche et al 2021, Delgrosso et al 2023), and from our preliminary data obtained from a comparison between hydrogels formed by only Alg (80 mg ml −1 ) and by combining Alg (80 mg ml −1 ) and Gel (40 mg ml −1 ), where the latter formulation resulted in being better in terms of cell response and distribution (see figure S1).
A deep rheological assessment was conducted on Alg/Gel hydrogel without and with lyo-dECM powder (GEL3 and GEL30) to study the viscoelastic behavior typical of this biomaterial characterized by a molecular structure.In particular, the trend of storage modulus (G ′ ) and loss modulus (G ′′ ) was recorded by varying different parameters (strain, frequency, temperature, and time).G ′ and G ′′ are the critical parameters in representing the viscoelastic behavior of a bioink, with G ′ describing the elastic part of the bioink and G ′′ the viscous component of the system.In general, it was possible to observe that lyo-dECM powder at 3.75 mg ml −1 did not alter the rheological properties of the basic hydrogel in any of the performed tests.Differently, a higher concentration, such as 30 mg ml −1 , caused differences from the original hydrogel.Indeed, in the frequency sweep test, Alg/Gel and GEL3 presented a solid-like behavior, resulting in line with the literature (Mondal et al 2019), while GEL30 had a liquid-like behavior with a G ′′ slighter higher than G ′ , as also found in other work with a similar bioink composition (Potere et al 2022).We explained this difference due to the fact that ECM is characterized by a net negative charge owing to the presence of GAG chains, which are rich in negatively charged functional groups (Augustine et al 2020).These negative charges, when present in high concentrations, create an electrostatic repulsion with the negatively charged polymeric chains present in the Alg/Gel blend, generating more space along the macromolecules and thus making the final material liquid-like.Probably, the following feature also affected the chemical crosslinking mechanism by means of a CaCl 2 solution.Indeed, Ca 2+ ions can interact with the alginate network, leading to an irreversible change from a liquid-like to a solid-like state (Gregory et al 2022).This switch happened in a few seconds in the Alg/Gel and GEL3 hydrogel.On the contrary, the gelification process was slowed when the lyo-dECM powder was present at higher concentrations.Such effect, also observed in other works (Guagliano et al 2023), can be justified by the presence of GAGs in the lyo-dECM that, due to their negative charges, are able to interact with Ca 2+ ions, thus competing with alginate and interfering with crosslinking kinetics.
Further discrepancy was observed in the physical crosslinking mechanism due to the temperature.Alg/Gel and GEL3 hydrogels presented a temperature-dependent behavior, given by the presence of the gelatin, whose behavior is reversibly influenced by temperature (i.e.solid-like below 37 • C and liquid-like above) (Chang et al 2009).Differently, in GEL30, a liquid-like behavior was detected within the entire temperature range, with no gelation point.We hypothesized that the presence of lyo-dECM powder at high concentration could 'switch off ' the dependence effect on the temperature given by the gelatin in the blend and, in general, it leads to a liquid-like behavior, as demonstrated by the frequency sweep test.Indeed, differently from gelatin, hydrogels formed only by dECM derived through protocol A are typically temperature dependent but with an opposite trend: liquid-like at low temperatures, starting gelation at a temperature beyond 15 • C and forming a crosslinked gel when incubating at 37 • C for 30 min (Pati et al 2014, Lee et al 2017).However, all these hypotheses proposed need to be subject to further investigations.
Finally, material viscosity was characterized as the function of shear rate, a key parameter during the printing process.From the shear rate ramp test, an optimal degree of shear-thinning behavior was observed for all the formulations.This feature facilitates bioink extrusion along the nozzle where high levels of shear rate are reached.At the same time, it allows keeping the 3D-bioprinted shape once extruded from the nozzle, as higher viscosity values are present at low shear rates.Nevertheless, from thixotropy tests, it was also observed that Alg/Gel and GEL3 take a long time to recover the initial viscosity values at a low shear rate after being extruded, negatively affecting the shape fidelity of printed constructs.
Generally, from these preliminary biological and rheological characterizations, we concluded that working with a concentration of lyo-dECM equal to 3.75 mg ml −1 was beneficial in terms of cell viability, and it did not alter the starting hydrogel's viscoelastic properties.For this reason, we selected GEL3 hydrogel for further analyses in terms of printing and cell viability.
As far as bioink design for 3D bioprinting is concerned, tuning printing parameters to obtain the optimal shape fidelity still represents a significantly material-consuming stage, as the most widespread method is uniquely based on a trial-anderror approach.To avoid the useless waste of lyo-dECM powder, we integrated computational analysis, which has been demonstrated to be a powerful tool for bypassing this approach (Chiesa et al 2020, Scocozza et al 2023).In particular, we implemented computational models for GEL3 hydrogel both to simulate the micro-extrusion process through a pressure-driven 3D bioprinter and the collapse of 3D-printed structures.From the rheological characterization, we extrapolated the inputs necessary to set up reliable computational models.Indeed, by fitting the shear rate ramp curve, we derived the two parameters (K and n) necessary to define the power law model, which is the simplest mathematical approach to describe the viscous nature of hydrogels.Simulating the material flow in the extrusion system allowed us to easily match different parameters (nozzle geometries and sizes, printing pressure, feed rate, shear stress, and material properties) and, therefore, straightforwardly optimize the printing process.From these results, a window of printability was defined based on the maximum shear stress value tolerated by cells (Koch et al 2020) and suitable feed rates (Paxton et al 2017).It was validated by printing serpentine structures at a selected printing velocity and imposing different pressure values.By evaluating the strand size, it was demonstrated that the target pressure obtained from CFD simulations gave the most optimal material extrusion.After that, we printed grid structures and assessed the shape fidelity of the pores, which resulted in being very close to the ideal value of 1, underscoring the good printing precision and shape fidelity of the multi-component hydrogel, even if with a certain degree of collapse, which is typical in viscous materials like the hydrogels.
Furthermore, the mechanical stability of the hydrogel was assessed following the experimental protocol established in the literature (Ribeiro et al 2018) and with the support of computational analyses.In this case, to describe the creep response of our viscoelastic material, we chose to use the Kelvin-Voigt equation, which is a simple model defined by two parameters, E and η 0 , corresponding to the elastic and viscous part, and that were extrapolated from the shear rate and frequency sweep tests.The hydrogel proved to be collapsible at higher gap lengths, and the implemented computational model resulted in agreement with the experimental measurements, which is promising in view of optimizing more complex shapes for 3D-printed constructs.
Finally, the multi-component hydrogel was printed with cells inside, and it was revealed to be suitable for culturing MSCs for up to 7 d, as no dead cells were observed.This final result demonstrates the biocompatibility of lyo-dECM even in combination with common but not tissue-specific materials used in the bioprinting field (e.g.Alg/Gel blend), highlighting its potential use as a biomimetic component in the design of new hybrid hydrogels.

Conclusions and future directions
The design of the 'perfect' bioink is an essential step toward developing functional and reliable 3D-printed in vitro models.dECM-based bioinks are gaining more importance thanks to their ability to represent the native biochemical microenvironment experienced by cells.In this context, we defined an alternative protocol for developing dECM-based hydrogels.Indeed, we converged the advantages of the two protocols already existing in literature, with the additional trait of being easier to carry out (also for laboratories non-specialized in this expertise) and shorter in timing.Throughout this protocol, we obtained lyo-dECM powder, an off-the-shelf product characterized by two main features: (i) ready-to-use before the printing process, an essential aspect in the perspective of product sale and scale-up of the whole process; (ii) water-soluble, meaning that the powder can be added as a biomimetic component to widely used, but non-tissue specific, hydrogels such as Alg/Gel blend.Its suitability for cell culturing and printing without altering the rheological properties of the Alg/Gel blend was demonstrated, making, in this way, possible its application for future perspectives.Indeed, our final aim is the development of a hepatic 3Dprinted in-vitro platform, capable to conjugate biological aspects with tunable mechanical properties, to mimic different stiffness values characteristic of specific conditions of liver (e.g.healthy or cirrhotic) and hosted in a perfusion device to recreate the vascular element, important feature in the study of this organ.Thus, this work can be considered as a piece of a bigger puzzle focused on the development of lyo-dECM powder to reproduce the biological features of the hepatic microenvironment.However, in this context, many aspects, and consequently many challenges, have to be still faced and studied, such as the culture of hepatocytes inside the so defined hydrogels, their coculture with non-parenchymal cells (e.g.fibroblasts and endothelial cells), and, most importantly, the evaluation of the biological functionality of the model.Connected to this last aspect, the stability of the 3D-printed constructs has to be evaluated over time as longer culture timing (>14 d) will be necessary to have a complete assessment of the hepatic function and of its response to drug treatments.

Figure 2 .
Figure 2. Computational modeling of extrusion bioprinting process: (a) bioprinting process based on the use of an extrusion-based system; (b) CAD-based reconstructions of the cylindrical (left) and conical (right) nozzles; (c) 2D axisymmetric geometries used for the CFD simulations.

Figure 3 .
Figure 3. Cell metabolic activity of pMSCs (a) and hMSCs (b) treated with different concentrations of lyo-dECM powder after 3 and 7 d.Multifactor ANOVA, mean values ± least significant difference (LSD), n = 3. Different letters (a, ab, b, c, d, e, f) indicate significant differences between the means (p < 0.05), whereas the same letter indicates no significant difference (p > 0.05).

Figure 6 .
Figure 6.Definition and optimization of bioprinting parameters: (a) example of velocity field and shear stress profile inside a cylindrical and conical blue (i.e.D = 0.41 mm) nozzle at a pressure equal to 30 kPa; (b) window of printability for GEL3 hydrogel: each box shows a couple of average extrusion velocity and maximum shear stress value, calculated at the nozzle outlet, associated to its nozzle sizes and geometries and printing pressures: blue area; (c) validation of CFD based window printability for GEL3 hydrogel: on the left the 3D-printed serpentine structure (scale bar 5 mm); on the right the relative measurements of the stand thickness; (d) shape fidelity assessment performed on 10% and 15% grids (scale bar 5 mm) in terms of Pr.

Figure 7 .
Figure 7. Analysis of the post-printing mechanical collapse: representation of the hydrogel deformation from the experimental tests (a) and from computational simulations (b); (c) quantification of the displacement obtained from the experimental tests and from computational simulations; (d) trend of the deformation over time of the 3D-printed lines obtained from the computational analyses: the final deformation is reached after 10-20 s (red line to indicate the time point for each curve) after its deposition.

Figure 8 .
Figure 8. 3D printing cell-embedded GEL3 hydrogel: (a) design of the 3D-printed constructs; (b) cell viability using live/dead assay at 1, 3, 7 d after the printing process (merge of green channel for live cells and red channel for dead cells); scale bar of 100 µm.

Table 3 .
Material parameters used as inputs for computational models derived from rheological tests performed on GEL3 hydrogel.