Cylindrical scaffolds were manufactured by Fin-Ceramica Faenza SpA (Faenza–Ravenna, Italy) starting from equine type I collagen gel (1 wt%) supplied in aqueous acetic buffer solution (pH = 3.5) (Opocrin SpA, Modena, Italy). The collagen gel was gently diluted in highly purified water and precipitated in fibers by drop-wise addition of 0.1 M NaOH solution up to the isoelectric point (pH = 5.5). In order to stabilize the scaffold whole structure, the molecular links between fibers was chemically optimized by using 1,4-butanediol diglycidyl ether (BDDGE) as crosslinking agent. The crosslinking reaction was achieved by 48 h-long immersion of the agglomerated fibers at 37°C in NaHCO3/Na2CO3 (Sigma Aldrich and Merck Millipore) buffered aqueous solution with a BDDGE solution ratio equals to 1 wt%.

The agglomerated fibers were then freeze-dried with a controlled freezing and heating ramp from 25°C to −35°C and from −35°C to 25°C achieving a porous 3D structure. The process was carried out over a period of 25 h under vacuum conditions (P = 0.29 mbar). Scaffold was then reshaped in cylinders of required diameters and height for physical characterization (Ø = 10–18 mm, h = 4 mm), in vitro and in vivo analysis (Ø = 8 mm, h = 5 mm). Scaffolds were gamma-sterilized at 25 kGy.

Scaffolds were characterized in porosity, swelling behavior and material density.

The morphological and microstructural analysis was executed by Scanning Electron Microscopy (SEM) performed on a SEM-LEO 438 VP (Carl Zeiss AG, Oberkochen, Germany). The samples were sputter-coated with gold prior to examination. 3 SEM micrographs were analyzed by image J software. The mean pore diameter was calculated as average of the major and minor axes of the ellipse representing pores cross-section. A total of 327 pores were analyzed obtaining a mean value of 67 ± 31 micron.

The swelling capacity of the material was evaluated on 20 cylindrical scaffolds (ø = 10 mm, h = 4 mm) of cartilage-like composition. Dry scaffolds were soaked at room temperature in PBS (Alchimia) until stabilization of scaffold dimension was reached. The swelling was then determined as percent increase in both dimensions and evaluating the weight increase as described by Ma et al. (2003). Outlier values (Huber test) were deleted from the data analysis.

The density and porosity of the collagen-based scaffold were evaluated with a glass pycnometer full of highly purified water on 20 scaffolds (d = 18 mm; h = 4 mm) (She et al., 2007).

The porosity of the scaffolds was then evaluated using the average value of the densities obtained and the geometric volume of the scaffolds.

where:

Human adipose derived stem cells (hADSCs) were derived from adipose tissue biopsies/lipoaspirates supplied by Mediterranean Institute of Oncology (IOM) (Viagrande, Italy) under an approved Institutional Review Board protocol (project ID code: 829_1 of 8 February 2013, IOM Institutional Review Board) and after informed consent. Isolation from adipose tissue, expansion and characterization by flow cytometry analysis using several MSCs surface markers of hADSCs was performed as previously reported (Calabrese et al., 2015, 2016a; Vicari et al., 2016). In addition, cells were characterized also by immunocytochemistry analysis using several positive (CD105, CD90, CD73, CD271) and negative (CD45, CD34, CD31, and GlycoforinA) MSCs surface markers.

Immunocytochemistry was performed on cells seeded in 8-well BD Falcon culture slides at a density of 5000 cells per cm² in MSC-GM (Lonza, Basel, Switzerland). The primary incubation was performed, overnight at 4°C, with the following anti-human antibodies: mouse CD105 (1:50, Novus Biologicals, Littleton, CO, USA), mouse CD90 (1:50, Santa Cruz Biotechnology, Dallas, TX, USA), mouse CD73 (1:25, Novus Biologicals), mouse CD271 (1:100, Santa Cruz Biotechnology), rabbit CD45 (1:100, Epitomics, Burlingame, CA, USA), rabbit CD34 (1:100, Epitomics), mouse CD31 (1:100, Santa Cruz Biotechnology), and goat polyclonal anti Glycoforin A (1:100, Novus Biologicals). After washing, slides were incubated with the appropriate secondary AlexaFluor 568 and 488 antibodies (Life Technologies Italia, Monza, Italy) at the dilution of 1:2000 for 1 h at RT. Nuclei were counterstained with DAPI (4',6-diamidino-2-phenylindole, 1:10,000). Finally, slides were mounted in fluorescent mounting medium Permafluor (Thermo Scientific, Waltham, MA, USA) and digital images were acquired using a Leica DMI4000B fluorescence microscope (Leica, Wetzlar, Germany). Control of immunostaining specificity was performed by omitting the primary antibody.

For flow cytometry analysis, cells were detached with 0.05% trypsin/EDTA and washed in PBS. 1 × 10⁴ cells/tube were stained with the following antibodies: CD45 FITC (Clone J.33), CD34 PE (Clone 581), Glycophorin A PE (Clone 11E4B-7-6), CD73 PE (Clone 581), CD90 FITC (Clone F15.42.1.5), CD105 PE (Clone 1G2), CD31PE (Clone 1F11), CD271 FITC (Clone ME20.4-1.H4) and corresponding isotypic controls according to manufacturer indications. All antibodies were purchased from Beckman Coulter (Milano,Italy), except CD271 that was provided by Miltenyi Biotec (Bologna, Italy). All tubes were incubated in the dark for 20 min at room temperature. Cells were then washed with PBS and finally analyzed by flow cytometry using an FC-500 five-color flow cytometer (Beckman Coulter, Pasadena, CA, USA). For each tube, 1000 events were acquired. CXP Analysis software (Beckman Coulter, Inc.) was used for data analysis.

hADSC chondrogenic differentiation was achieved as previously described (Calabrese et al., 2015). Briefly, 2.5 × 10⁵ cells were centrifuged to form a three-dimensional aggregate and resuspended in complete chondrogenic medium containing differentiation basal medium (Chondrogenic Basal Medium, Lonza) supplemented with chondrogenic differentiation inducing factors (hMSC Chondrogenic SingleQuots containing Dexamethasone, Ascorbate, ITS+supplement, GA-1000 Sodium Pyruvate, Proline and L-Glutamine, Lonza) and TGF-β3 (Lonza). Pellets were incubated at 37°C in a humidified atmosphere of 5% CO₂. The growth medium was replaced every 2–3 days. The chondrogenic differentiation was completed on day 28 after induction. Pellets were fixed in formalin at three different time points along differentiation (1, 2, and 4 weeks). Subsequently, pellets were paraffin embedded and cut into 3 μm-thick sections for immunohistological processing.

After deparaffinization and rehydration, sections were permeabilized with 0.4% Triton-X100, blocked with 4% BSA and then incubated overnight at 4°C with the following rabbit polyclonal primary antibodies: anti-Aggrecan (1:150), anti-Matrilin-1 (1:200), anti-Sox9 (1:200), anti-type II Collagen II (1:200), all purchased from LSBio (Seattle, WA, USA). The following day, sections were incubated for 1 h at RT with the Alexa Fluor anti-rabbit 568 secondary antibodies (1:2000, Life Technologies). Then slides were counterstained with DAPI (1:10,000) and mounted with Permafluor (Thermo Scientific). Control of immunostaining specificity was performed by omitting the primary antibody.

hADSC chondrogenic differentiation on scaffolds was performed as previously described (Calabrese et al., 2016a). Briefly, 2 × 10⁶ hADSCs at passage 3 were slowly drip seeded onto the scaffold and incubated in 24-well culture plates for 4 h at 37°C. ADSC-GM medium (2 ml) (Lonza, Basel, Switzerland) was added and 24 h later (day 0) the medium was replaced with chondrogenic or expansion medium. Both media were completely replaced twice a week. Each scaffold was analyzed on week 1, 2, 4, and 8 after chondrogenic induction.

Scaffolds seeded with hADSC were fixed in 4% PFA at the different time points analyzed (1, 2, 4, and 8 weeks), dehydrated, embedded in paraffin and cut into 3 μm-thick sections. Sections were mounted on slides and processed for immunohistochemical staining as above reported, using the same rabbit polyclonal primary antibodies: anti-Matrilin 1 (1:200, LSBio), anti-Aggrecan (1:150, LSBio), anti-Sox 9 (1:200, LSBio) and anti-Collagen II (1:200, LSBio). The following day, sections were incubated for 1 h at RT with the Alexa Fluor anti-rabbit 568 secondary antibodies (1:2000, Life Technologies Italia, Monza, Italy). Then slides were counterstained with DAPI and mounted with Permafluor. Control of immunostaining specificity was performed by omitting the primary antibody. Alternate sections were also labeled with Haematoxylin and Eosin (H&E) and with Alcian Blue staining (Panreac, Castellar del Valles, Barcellona, Spain). For the staining an Alcian Blue solution was prepared according to manufactured protocol. Slides were first deparaffinised in xylene and re-hydrated through passages in alcoholic solutions and, then, stained in Alcian Blue solution for 30 min. After incubation the staining solution was removed and the culture slides washed to get rid of excessive color. Slides were mounted and examined under light microscope.

Cell count analysis has been performed using Fiji image recognition software.

Data were analyzed as percentage of positive cells on total number of DAPI stained cells. Differences between experimental groups in histological data were evaluated by using two-way ANOVA for culture condition (scaffold in expansion medium, scaffold in chondrogenic medium o pellet differentiation) and timepoint (1, 2, and 4 weeks) for each of the assessed molecular endpoint (Sox9, Matrilin 1, Collagen II, and Aggrecan) followed by Tukey's HSD post-hoc test. For all experiments, a P < 0.05 was considered significant. All analyses were performed by means of Systat (Systat Software, USA).

The morphological and microstructural analyses of scaffolds were performed by SEM. The scaffolds images displayed porosity distribution from 16 to 180 μm, with larger channel (excluded from the analysis).

SEM images of the cartilage layer are displayed in Figure 1 with different magnification, 30x (Figure 1A), 50x (Figure 1B), and 150x (Figure 1C).

The cartilage-like layer shows a high level of porosity with three-dimensional interconnected pores without any preferential alignment of the collagen fibers. This kind of structures can help the cells function and guide it during their proliferation.

A qualitative analysis of the pores distribution has been performed, and it revealed a higher pores frequency, about the 65%, from 40 to 100 μm (Figure 1D).

Change of material structure was evaluated by swelling test. Each scaffold was weighed before and after the use of PBS solution to soak the scaffold. Swelling test clearly demonstrated that scaffolds are highly hydrophilic reaching the steady state in less than 1 min. The amount of PBS absorbed has been measured to evaluate the collagen scaffold ability to preserve liquid attributed to the maintenance of the three-dimensional structure. The absorption capability of the collagen scaffold resulted 2086 ± 238%.

Both diameter and thickness of the scaffold significantly increased after swelling (Figure 1E). The change in thickness (27 ± 9%) was much higher than in diameter (6 ± 1%).

The porosity of the cartilage in the biomaterial is crucial to assure the cell colonization through the whole scaffold. This parameter is related to the density of the biomaterial forming the scaffolds. The importance of the cell colonization directly influences scaffolds biocompatibility and cell adaptation toward the scaffold. The evaluation of the scaffolds porosity was determined using the medium of the densities obtained through a glass pycnometer and the geometric volume of the cylindrical shaped scaffolds. Applying the reported formula the density found is 0.89 ± 0.22 g/cm³ giving a porosity of 95.79 ± 0.29%.

All the experiments have been conducted with MSCs isolated from adipose tissue, and characterized by immunocytochemistry (Figure 2A) and flow cytometry (Figure 2B) analysis using several MSCs surface markers. Three different hADSC lines were used to study the expression of typically positive (CD73, CD90, CD105, CD271) and negative (CD31, CD34, CD45, glycoforin A) surface markers of stemness. In both analysis, all cell lines presented a strong positivity for CD73, CD90, CD105, and CD271, while no signal was detected for CD31, CD34, CD45, and glycoforin A (Figure 2).

Firstly, we wanted to evaluate the in vitro biological performance of a novel 3D porous scaffold, in terms of biocompatibility and ability to support hADSCs to differentiate toward a chondrocyte phenotype.

We have already demonstrated that hADSCs and bone-like scaffold have a high biocompatibility (Calabrese et al., 2016a). In this work, we found the same high biocompatibility of hADSCs with the collagen-based scaffold, as shown by haematoxylin and eosin (H&E) staining (Figure 3) that reveals the ability of cells to penetrate, adhere and proliferate into the scaffold. In particular, hADSCs loaded onto the scaffold were confined only on the surface during the first 2 weeks (Figures 3A–D), then, cells started to invade also the inner part of the scaffold, penetrating deeply inside (Figures 3E–H). This was observed in both experimental culture conditions: with expansion medium (Figures 3A,C,E,G) and with chondrogenic medium (Figures 3B,D,F,H), although a significant difference between the two experimental conditions was noted, in terms of cellularity and extracellular matrix quality. In fact H&E staining revealed not only the biocompatibility but also the differential proliferative and differentiation potential of the scaffold in the two different experimental conditions. Specifically, in expansion medium, at 1 week, cells are few and still confined on the surface of the scaffold, over the time, from 1 to 4 weeks, cells proliferate and migrate inside the scaffold, the scaffold results wrapped by a layer of spindled to stellate fibroblast-like cells that progressively invade the innermost part of material (Figures 3A,C,E,G). Anyway, at 4 weeks cells begin to appear suffering (Figure 3E), and, surprisingly, cellular necrosis and regressive phenomena occurred at 8 weeks (Figure 3G), as better shown in the squared image at higher magnification (Figure 3G square).

On the other hand, in presence of chondrogenic inducing factors (chondro medium) the increase in cellularity from first to eighth week is much more evident if compared with the same samples without inducing factors (expansion medium). In particular, starting from the second week (Figure 3D) it is also quite evident that the extracellular matrix assumes a chondromyxoid-like feature which progressively increases up to 8 weeks (Figures 3F,H). In fact, MSCs start to go toward condensation and differentiate in chondrocytes. The different nature of the cells is better highlighted in the images at higher magnification (Figures 3I–L) showing spindled and stellate cells during the first 2 weeks (Figures 3I,J). At 4 weeks the presence of pericellular lacunae (arrows) resembling chondrocytes, become quite clear (Figure 3K). At 8 weeks the chondromyxoid-like nature of the extracellular matrix, typical of cartilage tissue, with bigger and more numerous pericellular lacunae, is strongly evident (Figure 3L).

These observations were strongly confirmed by the Alcian Blue staining (Figure 4) that highlighted a striking difference between the samples in expansion medium (Figures 4A,C,E,G) and the samples in chondrogenic medium (Figures 4B,D,F,H), in terms of extracellular matrix nature. In expansion medium the blue staining did not reveal any chondromyxoid-like nature, whereas the presence of bioactive factors in the culture medium lead to the deposition of an extracellular matrix, starting to be visible already at 1 week, in which spindled and stellate mesenchymal cells are still present (Figures 4B,I). From the second week the amount of extracellular matrix deposition is bigger (Figure 4D) and become more and more consistent at 4 (Figure 4F) and 8 weeks (Figure 4H), where chondrocyte-like mesenchymal cells (Figures 4J,K) are embedded in a chondromyxoid-like matrix, presenting the typical cartilage pericellular lacunae, strongly evident in the higher magnification Figure 4L. The intense Alcian Blue staining confirms the cartilage-like nature of the neo-formed tissue.

In order to assess the chondro-inductive character of the scaffold in combination with the bioactive factors, we decided to compare the performance of the reference protocol defined by the inductive media manufacturer with the ones achievable by culturing of hADSCs on the collagen based 3D scaffold in both absence (expansion medium) or presence (chondro medium) of chondrogenic factors. At this aim we performed immunohistochemical analysis using markers typical of chondrogenic differentiation, including Sox9, type-II Collagen, Aggrecan and Matrilin-1.

Sox9 is a transcription factors that controls the expression of type II Collagen and Aggrecan and has a crucial role in chondrogenesis (Bi et al., 1999). The immunohistochemical analysis of Sox9 shows only few labeled nuclei during the first week in all experimental conditions (Figures 5A,D,G), at the second weeks a more evident fluorescent signals is present only in the pellet (Figures 5B,E,H), at 4 weeks a further increase is clearly visible in the pellet (Figure 5C), while the scaffold samples show a slight augment of Sox9 marked nuclei in both conditions, although in chondro medium (Figure 5F) such increase results higher compared to the expansion medium (Figure 5I).

In accordance with Sox9, Type-II Collagen immunohistochemical analysis shows a very weak staining during the first two weeks in all three conditions (Figures 5J,K,M,N,P,Q), although slightly more evident in the pellet (Figures 5J,K), whereas, starting from 4 weeks a more robust and extended signal becomes visible, although it results significantly higher in the pellet and in the scaffold samples in presence of inducing factors (Figures 5L,O), compared to the scaffold without bioactive factors (Figure 5R).

Aggrecan immunohistochemical analysis shows a protein expression pattern comparable to that of Type-II Collagen. Specifically, immunohistochemistry for Aggrecan reveals almost no signal during the first 2 weeks in both experimental conditions of the scaffold samples (Figures 6D,E,G,H), while in the pellet is already present a visible fluorescence at 1 week (Figure 6A) that becomes a little higher a 2 weeks (Figure 6B); at 4 weeks the pellet fluorescent staining is very strong (Figure 6C). At 4 weeks in scaffold, a modest fluorescent labeling is present in expansion medium (Figure 6I), while a quite stronger staining appears in chondrogenic medium (Figure 6F). Aggrecan is the principal proteoglycan (GAG) present in the ECM and forms, along with type-II collagen, the main structural element of the cartilage. As far as type-II Collagen, Aggrecan at 4 weeks presents a filamentous appearance (Figures 6C,F,I).

Matrilin-1 results very weakly expressed during the first 4 weeks of scaffold cultures in absence of differentiation inducing factors (Figures 6P–R). Conversely, in presence of chondrogenic inducing factors, already at 1 week a modest signal was visible (Figures 6L,O) both in pellet and scaffold samples (Figures 6J,M). Over time, the intensity and the extent of the fluorescent staining gradually increases until to be very strong at 4 weeks (Figures 6K,L,N,O), even if always a little higher in the pellet.

Results indicate that statistically significant differences in markers expression exist between all the differentiation methods assessed (p < 0.001), as graphically presented in Figures 5X, 6X.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.