CHANGES IN OSTEOGENESIS BY HUMAN DENTAL PULP STEM CELLS ON PORCINE DECELLULARISED ADIPOSE TISSUE SOLID FOAMS OF DIFFERENT POROSITY

The extracellular matrix of white adipose tissue (AT) is a very promising biomaterial for tissue engineering, due to its abundance, easy accessibility, malleability, and proven biological activity. Decellularised AT (DAT) can be processed by freeze-drying in acetic acid solutions, and changing the DAT concentration in the solution gives rise to three-dimensional (3D) scaffolds of different stiffness and porosity. In a previous report, we demonstrated that human Dental Pulp Stem Cells (hDPSCs) could differentiate to osteoblasts and generate mineralised bone on 3D solid foams of porcine Decellularised Adipose Tissue (pDAT) at a concentration of 0.5 % (w/v). In this research work, we assessed whether and how osteogenesis by hDPSCs would be regulated by testing pDAT solid foams formulated at three different concentrations: 0.25 % (w/v), 0.5 % (w/v), and 1 % (w/v), which showed different stiffness, porosity and water retention properties. As a control condition we tested solid foams formulated with 0.5 % bovine Collagen-I. Thus, we performed Alkaline Phosphatase and Alizarin Red staining, together with Transmission Electron Microscopy and the detection of osteoblastic differentiation markers Osterix and Osteocalcin at both protein and transcript level, to compare the osteogenesis mediated by hDPSCs grown on all these 3D scaffolds, in the presence or absence of osteoblastic induction media. Our results demonstrate that pDAT at 0.25 % supported osteogenesis better than the rest of tested scaffolds, including bovine Collagen-I, in 3D hDPSC cultures. This enhanced osteogenesis could be attributed to the formulation of 0.25 % pDAT solid foams, which presented a higher porosity.


Introduction
Three dimensional (3D) human stem cell culture systems relying on biologic scaffold materials constitute an effective strategy to generate humanised tissues with very important applications in drug screening and tissue bioengineering.Accordingly, the interactions that take place between stem and progenitor cells and their surrounding extracellular matrix (ECM), critically affect their differentiation outcome to adult cell lineages (Nowwarote et al., 2022;Rao et al., 2022).Understanding how physical and biochemical cues from the ECM affect stem cell lineage commitment and differentiation is thus of paramount importance to tissue bioengineering.
In the particular case of bone tissue, the stiffness and porosity of the ECM scaffold are regarded as two of the main factors that regulate differentiation of multipotent Mesenchymal Stem Cells (MSCs) to bone producing osteoblastic cells (Steward and Kelly, 2015;Xu et al., 2017).Thus, there is ample evidence demonstrating that osteogenic differentiation of MSCs is promoted by growing these cells on stiff substrates above 25-40 KPa (El-Rashidy et al., 2021;Engler et al., 2006;Lee et al., 2014a;Lee et al., 2014b), whereas it is reduced by growing them on soft substrates below 1 KPa (Engler et al., 2006;Lee et al., 2014b).
Regarding the effect of ECM porosity on osteogenesis, scaffolds with a higher degree of porosity have been shown to support bone formation better than less porous ones (Gupte et al., 2018;Jodati et al., 2020;Krieghoff et al., 2019).Accordingly, macropore sizes above 300 µ m were considered as optimal to facilitate angiogenesis and promote deposition of calcified bone ECM by MSCs or osteoblast cells seeded on different types of biomaterials (Haugen et al., 2019;Karageorgiou and Kaplan, 2005;Murphy et al., 2010).However, other studies have reported that smaller macropores of up to 50 µ m may also sustain osteogenesis to a similar extent (Itälä et al., 2001).So, the optimal ECM pore size for osteogenesis may be context-dependent, and not only their size but other factors like how these matrix pores are interconnected may play a pivotal role as well.The produced bone matrix, once mature and fully calcified, may also hamper nutrient and gas exchange.This "barrier effect" could also compromise the metabolic balance of the seeded cells in MSC 3D cultures once the process of osteogenesis and matrix mineralisation has been engaged (Salgado et al., 2004).
Human Dental Pulp Stem Cells (hDPSCs) may be regarded as some of the most promising stem cell types for cell therapy, because of their extremely high versatility for multilineage differentiation, including the possibility to give rise to many different types of mature adult cells like osteoblasts, adipocytes, chondrocytes, muscle cells, vascular cells and even neural cells (Aurrekoetxea et al., 2015;Luzuriaga et al., 2021).These stem cells also possess additional advantages that make them a very attractive option for cellular therapy in human subjects, such as their easiness of extraction in aseptic conditions, their immunomodulatory properties, high metabolic adaptability, and good chances of implementation into autologous transplant www.ecmjournal.orgtherapy (Ibarretxe et al., 2012;Luzuriaga et al., 2021;Tirino et al., 2011).
In particular, the osteo/odontogenic differentiation potential of hDPSCs is especially powerful, compared to other types of osteogenic human MSCs (Alksne et al., 2022;Huang et al., 2009;Kotova et al., 2021;Ma et al., 2019;Pisciotta et al., 2012).Thus, hDPSCs may readily differentiate to osteoblast/odontoblast lineage cells in standard foetal serum containing cell culture media, and similarly to other MSCs, this process can also be significantly accelerated by the addition of several soluble differentiation inducing factors like dexamethasone, betaglycerol phosphate, and ascorbate (Aurrekoetxea et al., 2015;Langenbach and Handschel, 2013).Therefore, hDPSCs are a particularly well suited stem cell type for bone tissue engineering.
In this research work we cultured hDPSCs on porcine Decellularised Adipose Tissue (pDAT) solid foams, obtained with different concentration.pDAT constitutes a very interesting biomaterial for tissue engineering applications, due to is complete biodegradability, biological activity, and low immunogenicity (Cicuéndez et al., 2021;Yang et al., 2020).In a previous report, we demonstrated that pDAT solid foams were optimal scaffolds for osteogenesis, with performance levels comparable to the gold standard bovine Collagen I (bCOL-I) (Luzuriaga et al., 2022).However, the hDPSC-mediated osteogenesis on pDAT solid foams could be further enhanced by optimising their structural properties such as stiffness and porosity.pDAT solid foams are relatively soft scaffolds with a stiffness usually lower than 1 KPa (Garcia-Urkia et al., 2022;Luzuriaga et al., 2022).However, both the stiffness and the porosity of the solid foams can be finely tuned by changing the pDAT concentration in the solution during their preparation.Thus, in this work we tested pDAT solid foams generated from solutions at 0.25 %, 0.5 %, and 1 % concentration (w/v), to search for the conditions that best supported osteogenic differentiation and matrix mineralisation by hDPSCs in 3D culture.

Adipose tissue decellularisation and processing as solid foams
The methodology of adipose tissue (AT) decellularisation and processing has been thoroughly described elsewhere (Madarieta et al., 2017;Madarieta et al., 2023).Briefly, Porcine AT was cleaned, creamed with a beater, and homogenised on ice using a Polytron (PT3100; Kinematica AG, Malters, Switzerland) with two different rods (PT-DA3012/EC and PT-DA-12/2EC-F154) at 12,000 rpm for 5 min.The homogenised tissue (25 mL) was centrifuged at 900 g for 5 min with 25 mL of ultrapure water to produce the phase separation of lipids, which were discarded, and protein pellets were taken and treated with isopropanol  Spain) for 36 h in an orbital shaker at room temperature.Then, the pDAT material was cleaned and lyophilised.Finally, the material was milled using a mixer mill (Restch MM400, Hann, Germany) to generate a fine-grainy powder (Fig. 1a) which could be kept at 4 °C in a vacuum desiccator until use.Milled pDAT met the decellularisation criteria as previously described (Cicuéndez et al., 2021;Madarieta et al., 2017;Madarieta et al., 2023).
Different solid foams were prepared by the freeze-drying method all the while maintaining the freezing temperature at -20 °C.Thus, 0.25 % (w/v), 0.5 % (w/v) and 1 % (w/v) milled pDAT, or alternatively 0.5 % (w/v) bovine collagen I (bCOL-I; ref C9879, Sigma, St. Louis, MO, USA) were added to a 0.5 M acetic acid solution and homogenised by magnetic stirring for 48 h.After that, different moulds were used to prepare the solid foams.For Scanning Electron Microscopy (SEM) and mechanical and swelling testing, 20 mm in diameter and 3 mm thickness Teflon moulds were used with a 1 mL solution.For in vitro cell culture assays, pDAT www.ecmjournal.organd collagen solutions were cast into conventional 48-and 24-well polystyrene tissue culture treated plates (approximately 200 µ L/cm 2 ).All the solid foams were obtained by freezing the solutions in the molds at -20 °C overnight, and then freeze-drying them (0.63 mbar and -20 °C) for 72 h.Freeze-dried solid foams were sterilised using ethylene oxide (Esterilizacion SL, Barcelona, Spain) for 270 min at 38 °C and 40 % relative humidity and kept at 4 °C in a vacuum desiccator until use.

Assessment of swelling and mechanical properties of solid foams
Mechanical properties of the pDAT and collagen solid foams were assessed by oscillatory shear rheology, with parallel-plate geometry (20-mm diameter steel with a gap of 1 mm) on a TA instruments HR20 rheometer (Waters, New Castle, DE, USA), as previously described (Garcia-Urkia et al., 2022;Luzuriaga et al., 2022).Stress amplitude sweeps were performed at a constant frequency of 0.1 Hz to establish the amplitude parameter for each sample and to ensure that all data were collected in a linear viscoelastic regime.All the measurements were made in constant deformation control mode over a frequency range from 0.01 to 10 Hz.
Swelling properties of the solid foams were determined by a water absorption assay.The dry weight (Wd) was recorded from lyophilized foams prior to bringing them to maximum hydration in distilled water for 24 h.Thereafter, the foams were carefully blotted to remove the excess liquid and the wet weight (Ws) was recorded.The equilibrium water content (% ECW) and mass swelling ratio (S) were calculated as previously described (Garcia-Urkia et al., 2022;Luzuriaga et al., 2022).

Assessment of porosity of solid foams
The International Union of Pure and Applied Chemistry (IUPAC) defined three classes of porous materials (Sing, 1985), namely www.ecmjournal.orgmicroporous (pore size < 2 nm), mesoporous (> 2 nm and < 50 nm) and macroporous (> 50 nm).The presence of macropores was analysed using Scanning Electron Microscopy (SEM) images of solid foams by Fiji-ImageJ free software (v1.54f.National Institute of Health, Bethesda, Maryland, USA).A total of 3 images and a 0.8 mm 2 surface area per image were analysed, counting only the pores that were completely visible in the image and taking their widest diameter.Results are shown as number of pores per image and diameter.The presence of micro and mesopores was assessed by using Barrett-Joyner-Halenda (BJH) analysis based on capillary condensation (Barrett et al., 1951).About 35 replicas of each sample type were individually incorporated into the flask of the gas porosimeter measurement system (Accelerated Surface Area and Porosimetry System, ASAP2420, Micromeritics, Norcross, GA, USA) and liquid N2 at 77K was applied as a gas, the pressure was measured at each moment of the analysis, and the amount of gas absorbed was determined.The porosity was quantified by the Brunauer-Emmett-Teller (BET) model to determine the total area (Surface area, g/cm 3 ).The Density Functional Theory (DFT) was used as the model to determine the distribution of the porosity.The pore volume (cm 3 /g) was calculated by extrapolating the values of each point of the isotherm.

3D culture of hDPSCs on solid foams
Primary cultures of hDPSCs from third molar teeth were established following a wellestablished protocol, as thoroughly described elsewhere (Gronthos et al., 2002;Pineda et al., 2022;Uribe-Etxebarria et al., 2017).The hDPSCs were cultured in standard Dulbecco's modified Eagle's medium (DMEM) supplemented with 10 % fetal bovine serum (FBS), L-glutamine (1 mM, ref.G7513, Sigma, St. Louis, MO, United States) and the antibiotics penicillin (100 U/mL) and streptomycin (150 µ g/mL) (Ref.15140-122,Gibco, Waltham, MA, USA).After the first culture passage (P1), cells were transferred to 3D culture plates where the bottom of the wells was filled with a layer of solid foams of 0.25 %, 0.5 % or 1 % (w/v) pDAT, or control 0.5 % (w/v) bCOL-I.3D cultures of hDPSCs on solid foams were kept for a period of 2 to 4 weeks to carry out the different experiments.Briefly, hDPSCs were cultured on solid foams in standard medium or alternatively with osteogenic medium supplemented with osteodifferentiation factors, as previously described (Luzuriaga et al., 2022).The osteogenic differentiation medium consisted of regular culture medium supplemented with 50 µ M ascorbic acid (ref.

Histochemical and immunohistochemical assessment of osteogenic differentiation and solid foam mineralisation
These protocols have been thoroughly described elsewhere (Madarieta et al., 2023).The in situ differentiation of hDPSCs to osteoblastic cells on solid foams was assessed by Alkaline Phosphatase (ALP) staining and immunohistochemistry (IHC) against BGLAP/osteocalcin after 2 weeks in culture.For ALP staining, solid foams were fixed for 1 min with 4 % paraformaldehyde (PFA) and then washed with 0.05 % Tween-20 in PBS buffer.ALP staining was performed in Tris-HCl buffer at pH = 9.5, using 5-Bromo-4-chloro-3indolyl phosphate/Nitro Blue tetrazolium (ref. B3804; Sigma, MO, USA) as substrate.The staining progress was checked every 3 min, and the reaction was stopped by washing foams with PBS three times for 5 min.Then, ALP absorbance was measured at 420 nm using a Synergy HT Multi-Mode Microplate Reader (Biotek, Winooski, VT, USA).For the acquisition of microscopy images, solid foams were counterstained with 4′,6-diamino-2-phenilindol (DAPI, 1:1000; Invitrogen, Camarillo, CA, USA), and then detached from the culture plates to be mounted onto conventional glass slides.Images of mounted foams were taken with an epifluorescence and transmitted light OLYMPUS IX71 microscope, coupled to an OLYMPUS DP71 digital camera (Olympus Iberia, L'Hospitalet de Llobregat, Barcelona, Spain).
For IHC experiments, pDAT and bCOL-I solid foams were fixed with 4 % PFA, and the blocking performed with 10 % goat serum (ref.A32731, ThermoFisher Scientific, Waltham, MA, USA) for 1 h at room temperature.Cell nuclei were counterstained with DAPI (1:1000; Invitrogen, CA, USA), and detached from the culture plates to be mounted onto glass microscopy slides.Samples were observed with a confocal microscope Zeiss LSM800 (Zeiss, Overkochen, Germany) coupled to an Axiocam 305 (Zeiss, Overkochen, Germany) color camera.Relative BGLAP immunostaining was calculated by dividing the overall IF labelling intensity by the number of nuclei present in the confocal orthogonal image projection, using Fiji-Image J free software.
Mineralisation/calcification of solid foams was assessed by Alizarin Red (AR) staining.hDPSCs cultured for 4 weeks in pDAT or bCOL-I solid foams were fixed with 4 % PFA for 10 min.Samples were then washed with distilled water and stained using 2 % AR (Acros organics, Sigma, MO, USA) at pH 4.1-4.3 for 45 min on the dark at room temperature.The staining solution was removed and samples were washed with distilled water until no AR could be found dissolved on water.AR absorbance was measured at 405 nm using a Synergy HT Multi-Mode Microplate Reader (Biotek, Winooski, VT, USA).Thereafter, foams were detached from culture plates and mounted onto microscopy slides to obtain higher resolution images.Cell nuclei were counterstained with DAPI (1:1000; Invitrogen, CA, USA).Images of mounted foams were taken with an epifluorescence and transmitted light OLYMPUS IX71 microscope, coupled to an OLYMPUS DP71 digital camera.

Scanning and Transmission Electron Microscopy (SEM; TEM)
The microarchitecture and porosity of solid foams was assessed by SEM.Samples were mounted, and pulse coated with gold and visualised with a JEOL JSM-5910 LV SEM (JEOL, Tokyo, Japan) with an accelerating voltage of 10 kV.The cell and matrix ultrastructure in solid foams were assessed by TEM at 4 weeks after hDPSC seeding.Foams were fixed with 2 % glutaraldehyde (ref.

RNA extraction, reverse transcription and quantitative real-time (RT-qPCR)
Differentiation of hDPSCs to osteoblastic cells was also assessed at transcript level, as previously described (Luzuriaga et al., 2022).RNA was extracted by a 4:1 solution of Trizol (ref 10296010, Invitrogen, Waltham, USA) and chloroform (ref C0549-1PT, Sigma, MO, USA).Samples were then centrifuged at 12,000 g and 4 °C to separate organic and aqueous phases.The latter were collected and RNA precipitation was made by incubating for 10 min with isopropanol at 1:1 ratio.Then, samples were centrifuged again at 12,000 g for 5 min at 4 °C, to form a gellike pellet containing the RNA, which was washed with 75 % ethanol, dried, and finally resuspended in 20 µ L of elution buffer (ref 12183020, RNA mini-kit, Invitrogen, Waltham, USA).RNA purity and concentration were calculated by measuring the 260/280 nm absorbance ratio, using a Synergy HT spectrophotometer (Biotek, Winooski, VT, USA).Reverse transcription was made using using the iScript cDNA kit (Ref.1708890;BioRad, Hercules, CA, USA).Finally, qPCR experiments were conducted using an iCycler My iQ™ Single-Color Real-Time PCR Detection System (BioRad, Hercules, CA, USA), using 4.5 µ L of Power SYBR® Green PCR Master Mix 2× (ref 4367659, Applied Biosystems™, Waktham, MA, USA), 0.5 µ L of primers (0.3125 µ mol/L), 0.3 µ L of cDNA (1.5 ng/µ L) and nuclease-free water for a total volume reaction of 10 µ L. Data were processed by CFX Manager™ Software (BioRad, Hercules, CA, USA).All RT-qPCR reactions yielded only one amplicon as assessed by the melting curve www.ecmjournal.orgmethod.BGLAP and OSTERIX were taken as representative gene markers of ongoing and terminal osteoblastic differentiation respectively, and their expression was normalised to the average of three housekeeping genes GAPDH, β-ACTIN, and RPS18.The sequences of primers are listed in Table 1.

Characterisation of solid foams made with different pDAT concentrations
The AT decellularisation procedure was carried out as previously described (Cicuéndez et al., 2021;Luzuriaga et al., 2022).Once the fine-grainy powder of pDAT was obtained (Fig. 1a), this was reconstituted at different concentrations (0.25 %, 0.5 %, and 1 % w/v) in acetic acid solution to obtain 3D solid foams with different porosity and stiffness after freeze-drying.These ECM biomaterials were incorporated to conventional polystyrene cell culture plates for hDPSC culture.The viscoelastic behaviour of the solid foams showed higher G´ (storage modulus) values than G´´ (loss modulus) during the entire range of frequencies, indicating the predominance of the elastic behaviour over the viscous one for all assayed samples.The solid foams showed higher average storage modules (G´) according to the increase of pDAT concentration in the formulation: 68.7 ± 10.6 Pa, 176.4 ± 26.4 Pa and 633.8 ± 75.7 Pa for pDAT 0.25 %, 0.5 % and 1 % respectively.Thus, there was approximately an order of magnitude of stiffness difference (G´and G´´) between the pDAT foams at the upper and lower concentration ends of 0.25 % pDAT, and 1 % pDAT.0.5 % bCOL-I solid foams used as control scaffold showed a higher storage and viscous modulus than its counterpart pDAT at 0.5 % (Fig. 1b, Table 2).These results are in agreement with water absorption and microstructural analysis of the solid foams, because the water contribution to the overall resistance of the materials to deformation was higher in pDAT solid foams at 0.5% than in bCOL-I at the same concentration.
The microstructural SEM analysis of the solid foams showed that they all presented homogeneous porous structures, with substantial differences in shape and number.0.25 % pDAT solid foams showed more irregular pores. 1 % pDAT and 0.5 % bCOL-I solid foams showed much better defined pores with lower connectivity.0.5 % pDAT foams showed well defined and interconnected pores (Fig. 1c).

www.ecmjournal.org
Overall, the highest levels of porosity were found on pDAT 0.25 % solid foams, which showed a much higher density of macropores, particularly of those with ≤ 50 µm diameter (Fig. 2a).In this regard, all pDAT solid foams were found to be far more porous than bCOL-I ones, where the presence of macropores on pDAT 0.5 % and pDAT 1 % foams approximately doubled that of bCOL-I foams (Fig. 2a).pDAT 0.25 % was the most porous solid foam, with a total density of macropores that almost doubled that of the other two pDAT foams (Fig. 2b).BET analysis also showed the presence of mesopores within the solid foams, with an average pore size in the range of 44 to 53 Å (Table 3).The total area and volume of mesopores was higher in pDAT 0.5 % and bCOL-I 0.5 %, especially compared to pDAT 1 %, which was the least porous material on this respect.pDAT 0.25 % solid foams showed intermediate mesopore levels (Table 3).No micropores were detected in any of the assayed samples (not shown).We sought to address whether changes in pDAT solid foam stiffness would reflect in changes of hDPSC differentiation efficiency to osteoblastic cells, by IHC and RT-qPCR analysis.The immunolabeling of the bone matrix protein BGLAP was found to be much higher in 0.25 % pDAT foams, than in 1 % pDAT foams (316 ± 61 % vs 68 ± 10 % respectively, with respect to 100 ± 30 % of bCOL-I; Fig. 3a,b; p < 0.01).BGLAP is a bone matrix protein whose expression is indicative of an osteoblastic cell commitment but does not necessarily imply a terminally concluded osteoblastic differentiation (El-Rashidy et al., 2021;Irastorza et al., 2019).The transcription factor OSTERIX (SP7) was thus chosen as an alternative gene marker more specific of terminally differentiated osteoblastic cells (Liu et al., 2020).Immunolabeling against OSTERIX was also found to be much higher in the 0.25 % pDAT foams (202.91 ± 12 % vs 100 ± 6 % of bCOL-I).This difference was even larger when DPSCs were cultured under osteodifferentiation induction conditions (Fig. 4a-c).Interestingly, a decreasing intensity of BGLAP and OSTERIX immunofluoresence in 3D hDPSC cultures was observed as the pDAT concentration increased, with lowest staining levels for 1 % pDAT, intermediate levels for 0.5 % pDAT, and highest levels for 0.25 % pDAT (Fig. 3a,b and Fig. 4a,b).This pattern was also maintained at transcript level, where qPCR data showed a much larger BGLAP and OSTERIX expression in the 0.25 % pDAT matrix condition, with respect to 0.5 % pDAT and 1 % pDAT solid foams (Fig. 4d; p < 0.001).hDPSCs grown on 0.25 % pDAT mineralised the solid foam matrix with greater efficiency than on 0.5 % pDAT, 1 % pDAT, and 0.5 % bCOL-I Immunostaining and RT-PCR experiments suggested a more efficient osteoblastic differentiation on 0.25 % solid foams, compared to the rest of conditions.Thus, it could be expected that this enhanced differentiation would also translate into differences at histochemical level (i.e., enhanced matrix mineralisation).Two types of histochemical labelling were performed: ALP staining, and Alizarin Red staining.ALP histochemistry at 2 weeks post-seeding showed higher labelling levels when hDPSCs were grown on 0.25 % pDAT solid foams, in the presence of osteoblastic induction media (Fig. 5a-c).This was despite a high background absorbance by solid foams at 420 nm, which partly overshadowed the specific ALP signal (not shown).The final proof of matrix mineralisation is the presence of hydroxyapatite mineral in the solid foams, which can be stained with AR, but hydroxiapatite crystal accumulation and matrix mineralisation is a relatively slow process which can only be assessed in the long-term.Thus, we performed AR staining to solid foams at 4 weeks after hDPSC seeding and 3D culture in both control and osteoinductive conditions (Fig. 6).Again, we found some background absorbance in the solid foams, which was higher in 1 % pDAT and 0.5 % b-COL I foams, intermediate in 0.5 % pDAT foams, and lower in the case of 0.25 % pDAT foams (Fig. 6c).We did not find any differences in AR staining on hDPSC-containing solid foams in control conditions (Fig. 6a-c).However, a very intense AR staining was found www.ecmjournal.org in solid foams containing hDPSCs grown under osteoinductive conditions, which was clearly distinguishable from the background absorbance (Fig. 6b,c).Measuring AR absorbance levels on the foams cultured in osteoinductive media showed statistically significant differences, with 0.25 % pDAT solid foams displaying significantly higher mineralisation levels than the other three conditions (Fig. 6b-d; p < 0.01).Once again a decreasing gradation of AR absorbance was found to be in accordance with the pDAT concentration on the solid foams, with highest staining levels found at 0.25 % pDAT (168 ± 19 %; p < 0.01), and lowest levels at 1 % pDAT (102 ± 8 %).Control solid foams of 0.5 % bCOL-I showed mineralisation levels (100 ± 18 %) which were smaller than those of 0.25 % pDAT, but not significantly different from those of 0.5 % pDAT and 1 % pDAT (Fig. 6b-d).
To assess in detail the cellular phenotypes and the calcified matrix ultrastructure in the 3D solid foam scaffolds, we performed a TEM analysis.Thus, we found that mineralised matrix regions could be distinguished from the rest of non-mineralised regions in the solid foam by their higher electron density (Fig. 7 and Fig. 8a).We found out that many of the seeded cells in the foams had effectively differentiated to osteoblastic cells, which were actively secreting highly intertwined collagen fibers that underwent a rapid calcification, as assessed by TEM (Fig. 7).In contrast, some other collagen containing regions that were initially part of the solid foams remained non-calcified (Fig. 8a).Consistently with the rest of the assays, the de novo produced electrodense calcified matrix regions were far more abundant in pDAT 0.25 % solid foams than in the rest of conditions (Fig. 7).The cross-section diameter of these newly synthesised collagen fibrils was about 50-100 nm, and was much smaller than those present on bovine collagen solid foams (bCOL-I).The cell morphology of in situ osteodifferentiated DPSCs included a well-developed endoplasmic reticulum consistent with a high protein secretion activity, also showing fine cellular processes which were getting progressively embedded in the electrodense calcified collagen matrix (Fig. 7; dashed yellow line).Secretory cells with a prominent Rough Endoplasmic Reticulum were detected (showed for 0.5 % pDAT).The pDAT solid foam scaffold at 0.25 % underwent a more efficient mineralization, featured www.ecmjournal.orgby a more abundant presence of electrodense fiber bundles, and fine cellular processes getting embebbed on the electrodense calcified bone matrix (dashed yellow line).Note the bigger diameter of control bovine collagen fibers (bCOL-I; left panel), compared to de novo secreted human collagen bundles by osteoblast-differentiated cells in 0.25 % pDAT and 1 % pDAT.Scale bars: 1 µ m (top), 500 nm (bottom).

Cell phenotypes in 3D solid foams after 4 weeks of DPSC-seeding include osteoblastic cells, ECM degrading cells, and multilocular adipocytes
In addition to cellular morphologies characteristic of secretory osteoblasts, we also identified cells displaying variable amounts of endolysosomes, which suggested an ECM digestion activity (Fig. 8b, green arrows).Thus, it appeared that the seeded hDPSCs were undertaking a true ECM remodelling within the solid foam, secreting new ECM, but also degrading some of it.Consistent with this interpretation, both exocytic and endocytic vesicles could be observed in the plasma membrane of some of the seeded cells (Fig. 8b, blue arrows).Some cells were also identified as storing large amounts of glycogen granules (Fig. 8b, red arrows).Finally, some of the seeded hDPSCs also showed a characteristic multilocular adipogenic differentiation phenotype after 4 weeks, which could be recognised by the formation of numerous cytosolic lipid droplets of homogeneous content, with moderate to high electron density (Fig. 8b, yellow arrows).These brown fat-like cell morphologies could be observed in all solid foams except bCOL-I, with no apparent www.ecmjournal.orgdistinction according to the different pDAT concentrations.

Discussion
Decellularised ECM biomaterials obtained from adipose tissue constitute very promising tissue engineering scaffolds for 3D cell culture and personalised regenerative therapies (Badylak, 2007;Yang et al., 2020).In a previous report, we found that solid foams derived from pDAT could support osteogenesis after hDPSC seeding, to similar effectiveness to the so far widely employed gold standard bCOL-I (Luzuriaga et 2022; Rico-Llanos et al., 2021).However, pDAT solid foams can be tailored according to a variety of factors.An obvious one is the pDAT concentration in the starting freeze-dried solution, which would affect both the stiffness and porosity of the resulting scaffold.In this study, we generated solid foams of different pDAT concentrations, which resulted in scaffolds with different stiffness and water retention properties.Our initial assumption was that the stiffer (i.e., more concentrated) pDAT solid foams obtained at 1 % might have been better scaffolds to support hDPSC osteodifferentiation and osteogenesis.It had been already amply reported that the osteogenic differentiation efficiency of MSCs depended on the substrate characteristics, with stiffer surfaces promoting osteogenic differentiation, and softer substrates promoting adipogenic and/or neural differentiation (Engler et al., 2006;Lee et al., 2014a).However, even though significant differences were present between conditions, increasing the pDAT concentration from 0.25 % to 1 % did not increase the stiffness of the solid foams above 1KPa (Fig. 1), which meant that even in the most concentrated condition (1 % pDAT), the stiffness of our pDAT biomaterial would still be comparatively very low, with respect to previous studies where an enhanced osteodifferentation was observed at stiffness levels ranging over 25-30 KPa (Engler et al., 2006;Lee et al., 2014b).
Besides stiffness, ECM porosity is regarded as another fundamental factor governing osteogenesis (Gupte et al., 2018;Jodati et al., 2020;Krieghoff et al., 2019).It was proposed that pore diameters of at least 100 µm and optimally 300-400 µ m were necessary to allow an adequate cell infiltration, metabolic exchange, and angiogenesis, to support bone matrix formation (Haugen et al., 2019;Hulbert et al., 1970;Murphy et al., 2010).Nevertheless, other reports have come to challenge that idea, suggesting that smaller pore sizes up to 50 µm may be effective in promoting osteogenesis as well (Bobyn et al., 1980;Itälä et al., 2001).Apart from the mere pore size, the level of interconnectivity of those pores also matters, and so does the presence of micro and mesopores which could facilitate gas and nutrient exchange to the matrix-infiltrating cells.
In our system of osteogenesis in relatively soft 3D solid foams of pDAT and bCOL-I, we found that porosity was a crucial characteristic of the scaffold that supported its mineralisation by hDPSCs.We found that the most effective solid foam to support hDPSCs osteogenesis was the 0.25 % pDAT, which was shown to be largely more porous, particularly to what concerned the presence of macropores in a size range around 50 µ m, as assessed by SEM (Fig. 1c and Fig. 2a).It should be emphasised that throughout this work we have followed the porosity classification by IUPAC (Sing, 1985), which is not universally adopted (Jodati et al., 2020).The greater porosity of pDAT 0.25 % solid foams correlated well with an enhanced osteogenesis of hDPSCs in these conditions, which was corroborated by different techniques including: IHC, RT-PCR, ALP, AR, and TEM, assessing not only osteoblastic differentiation, but also the biomaterial mineralisation.We thus attribute the better ability of 0.25 % pDAT solid foams to support osteogenesis to a higher material porosity.It is to be remarked that the principal difference of 0.25 % pDAT to the rest of solid foams is the much greater abundance of small macropores up to 50 µ m.No increases in pore density were observed regarding big macropores > 50 µ m, where 0.5 % pDAT and 1 % pDAT foams showed higher relative values.As for the presence of mesopores between 2 nm and 50 nm, the least porous material was 1 % pDAT, and the most porous was 0.5 % bCOL-I (Table 3).
Therefore, we found a clear correlation between solid foam porosity and osteogenesis in our 3D hDPSC cultures.0.25 % pDAT solid foams have the highest density of macropores, which explains the better results obtained in the mineralisation and osteoblastic detection assays, compared to the rest of formulations.www.ecmjournal.orgComparing bCOL-I and pDAT at the same concentration of 0.5 %, it was found that pDAT had a higher macroporosity than bCOL-I, which could explain the better performance of the former in some of the assays.As for 1 % pDAT, this was the material that showed the lowest mesoporosity, and arguably the worst osteogenic capability, even though that was not significantly different from that of bCOL-I.Thus, in our 3D culture system in the absence of microfluidic perfusion, pores of different sizes may all play an important role in supporting osteogenesis.
As assessed by confocal microscopy, in our 3D culture conditions, most cells apparently preferred not to migrate very deep inside the solid foams and stayed within the first 200-400 µ m below the foam surface.To allow a deeper cell penetration in the absence of an in vivo vasculature, the solid foam would have needed some sort of nutrient and/or gas exchange system, which is especially important in the case of bone tissue engineering, due to the additional substance diffusion restrictions imposed by the presence of a calcified ECM (Salgado et al., 2004).Thus, according to this view, the more porous microstructure of 0.25 % pDAT solid foams would provide for a better and more efficient metabolic exchange, which would allow for a higher rate of bone matrix deposition than in the rest of conditions.
The pDAT ECM-derived biomaterial is deemed to be non-immunogenic due to its processing and effective decellularisation (Cicuéndez et al., 2021;Yang et al., 2020).Compared to other osteo-permissive biomaterials like bCOL-I, the procurement of pDAT can be much cheaper and easier (Luzuriaga et al., 2022;Madarieta et al., 2023).Our group has developed and patented both solvent-related and enzyme-related methodologies for pDAT extraction, and each can be adapted to the application of choice (Brown et al., 2011;Cicuéndez et al., 2021).More compatibility experiments will still be necessary to assess the translatability of pDAT to regenerative therapies in human patients, but the experiences with other biological products of porcine origin have been so far substantially positive (Badylak, 2004;Mosala Nezhad et al., 2016;Rana et al., 2017).Moreover, pDAT showed a remarkable capacity for processing, and can be used to obtain a large variety of scaffolds, including hydrogels for cell encapsulation and injection, and bio-inks for 3D bioprinting (Dzobo et al., 2019;Kabirian and Mozafari, 2020;Zhang et al., 2017), allowing an extraordinary versatility for tissue engineering.
Apart from xenogenic sources, our group is also investigating on human derived DAT (hDAT) from liposuction surgery in collaboration with hospitals, thus allowing a fully personalised clinical application of this biomaterial, whether for the design of in vitro screening platforms including autologous 3D cell cultures, or in vivo grafts for tissue regeneration.In a previous study, we found that hDAT was much more refractory to mineralisation by hDPSCs, likely owing to the higher human specificity of the different biomodulatory signals naturally present in the adipose tissue ECM (Luzuriaga et al., 2022).However, this resistance to osteogenesis can also be beneficial in some circumstances, such as the reconstruction of bone-adjacent soft tissues like tendons and ligaments (Ibarretxe, 2022;Screen et al., 2015).Even though pDAT had rather different bioinductive characteristics to hDAT, we found that a proportion of the seeded hDPSCs still differentiated to adipocyte cells in the presence of pDAT (Fig. 8b).The adipogenic phenotype of these cells was multilocular, corresponding to brown fat-like adipocytes.This brown-fat bioinduction is also of interest from the point of view of regenerative therapies of joints or ligaments, which can benefit from a locally increased thermogenesis in the grafted area, to accelerate endogenous healing mechanisms (Diaz et al., 2014).Finally, it could be very interesting to combine these pDAT and hDAT scaffolds with hDPSCs grown in vasculogenic serum-free media, as recently described (Luzuriaga et al., 2019;Luzuriaga et al., 2020;Mattei et al., 2021).The development of autologous micro-tissues harbouring endogenous vasculature could constitute a real leap forward in the development of more effective regenerative cell therapies, and for all their versatility and fine-tuning possibilities, these decellularised adipose tissue ECM materials appear to be very attractive tools for this purpose.www.ecmjournal.org

Conclusions
The fine customisation of pDAT solid foams as a scaffold to support osteogenesis in 3D cell culture can significantly enhance the performance of this biomaterial for bone tissue engineering.We report here that significant changes in hDPSC-mediated osteogenesis take place when culturing these cells in 3D pDAT solid foams of different concentration: 0.25 %, 0.5 %, and 1 % (w/v).These tailored pDAT biomaterials presented different stiffness, porosity and water retention properties.The softest foams of pDAT at 0.25 % were the most effective in driving osteoblastic differentiation of hDPSCs and ECM mineralisation, owing to their much higher porosity.The wide-ranging possibilities of pDAT and hDAT formulations offer very positive prospects for an optimisation of tissue engineering constructs for personalised medicine.

Fig. 1 .
Fig. 1.Viscoelastic behaviour and ultrastructual morphology of the 3D solid foams.(a) pDAT powder of different concentration obtained after decellularisation.(b) Logarithmic quantification of the viscoelastic behaviour showing storage and loss modulus.(c) Ultrastructural images of the different solid foams.Scale bars = 100 nm.

Fig. 2 .
Fig. 2. Relative density and sizes of macropores in solid foams.(a) SEM images of the different solid foams were analysed and macropores were classified according to their diameter.(b) Quantification of relative macropore density as total number of macropores per SEM image area.Data shown as mean ± SD.

Fig. 5 .
Fig. 5. ALP staining of hDPSCs seeded on 3D pDAT and b-COL I solid foams after 2 weeks.(a) Control conditions.(b) Osteodifferentiation conditions.Brightfield images were taken with the same light intensity on the different solid foams.Cell nuclei are counterstained with DAPI.Scale bar (a,b) =

Fig. 6 .
Fig. 6.AR staining of hDPSCs seeded on pDAT and b-COL I solid foams after 4 weeks.(a) Control conditions.(b) Osteodifferentiation conditions.Brightfield images were taken with the same light intensity for each of the different solid foams.Cell nuclei are counterstained with DAPI.Scale bar (a,b) = 50 µ m.(c) Brightfield images showing background absorbance of solid foams with seeded cells.(d) Quantification of AR absorbance by plate photometry in different foams after background subtraction.Data shown as mean ± SEM. **: p < 0.01 (n = 4).

Fig. 7 .
Fig. 7. TEM images of hDPSCs-seeded solid foams after 4 weeks, showing osteoblastic cell phenotypes and matrix mineralisation.Top panel: Lower magnification; Bottom panel: higher magnification.Newly secreted collagen fibers calcify and form compact electrodense bundles.Secretory cells with a prominent Rough Endoplasmic Reticulum were detected (showed for 0.5 % pDAT).The pDAT solid foam scaffold at 0.25 % underwent a more efficient mineralization, featured

Fig. 8 .
Fig. 8. TEM images of hDPSCs-seeded solid foams after 4 weeks, showing ECM calcification, ECM degrading cells, and multilocular adipocytes.(a) Sharp transitions in calcification of solid foam areas featuring de novo secreted electrodense collagen fibers by osteoblast differentiated hDPSCs on 0.25 % pDAT solid foams.Limits of highly calcified foam areas are marked by a yellow line.(b) The seeded hDPSCs digest and remodeled the solid foams, as assessed by the presence of endolysosomes (b; top left; 0.5 % pDAT, green arrows), and endocytic vesicles (b; top right; 1 % pDAT, blue arrows).Some cells showed abundant glycogen granules (b; top right; 1 % pDAT, red arrows).Some cells also underwent a brown multilocular adipocyte-like differentiation, as assessed by the presence of numerous lipid droplets together with mitochondria in the cell cytoplasm.Lipid droplets showed a characteristic spherical and homogeneous appearance that distinguished them from the rest of cell organelles (b; bottom panels; 0.25 % and 0.5 % pDAT, yellow arrows).Scale bars: 2 µ m for all images, except top right (200 nm).