Biofabrication     

PAPER

Biofabrication of human articular cartilage: a pathtowards the development of a clinical treatmentTo cite this article: Carmine Onofrillo et al 2018 Biofabrication 10 045006

 

View the article online for updates and enhancements.

Related contentDevelopment of the Biopen: a handhelddevice for surgical printing of adipose stemcells at a chondral wound siteCathal D O’Connell, Claudia Di Bella,Fletcher Thompson et al.

-

A comparison of different bioinks for 3Dbioprinting of fibrocartilage and hyalinecartilageAndrew C Daly, Susan E Critchley, EmilyM Rencsok et al.

-

3D bioprinting of BM-MSCs-loaded ECMbiomimetic hydrogels for in vitroneocartilage formationMarco Costantini, Joanna Idaszek,Krisztina Szöke et al.

-

This content was downloaded from IP address 128.250.0.99 on 23/08/2018 at 01:29

Biofabrication 10 (2018) 045006 https://doi.org/10.1088/1758-5090/aad8d9

PAPER

Biofabrication of human articular cartilage: a path towards thedevelopment of a clinical treatment

CarmineOnofrillo1,2,3,8 , SerenaDuchi1,2,3,8, Cathal DO’Connell3, RomaneBlanchard1,3,Andrea JO’Connor4,Mark Scott5, GordonGWallace2,3, Peter FMChoong1,2,3,6 andClaudiaDi Bella1,2,3,6,7

1 Department of Surgery, St Vincent’sHospital, University ofMelbourne, Clinical Sciences Building, 29Regent Street, 3065 Fitzroy, VIC,Australia

2 ARCCentre of Excellence for Electromaterials Science, Intelligent Polymer Research Institute, InnovationCampus, University ofWollongong, NSW,Australia

3 BioFab3D, AikenheadCentre forMedical Discovery, St Vincent’sHospital,Melbourne, Australia4 Department of Biomedical Engineering, University ofMelbourne, Parkville, VIC, Australia5 Imaging Laboratory, TheWalter andElizaHall Institute, 1GRoyal Parade, 3052 Parkville, VIC, Australia6 Department ofOrthopaedics, St Vincent’sHospital, Fitzroy, Victoria, Australia7 Author towhomany correspondence should be addressed.8 Duchi andOnofrillo contributed equally to this work.

E-mail: carmine.onofrillo@unimelb.edu.au, c.onofrillo@unimelb.edu.au, sduchi@unimelb.edu.au, cathal.d.oconnell@gmail.com,romane.blanchard@unimelb.edu.au, a.oconnor@unimelb.edu.au, scott.m@wehi.edu.au, gwallace@uow.edu.au, pchoong@unimelb.edu.au and claudia.dibella@unimelb.edu.au

Keywords: biofabrication, osteoarthritis, cartilage, stem cells, Biopen, hydrogel, bioscaffold

Supplementarymaterial for this article is available online

AbstractCartilage injuries cause pain and loss of function, and if severemay result in osteoarthritis (OA). 3Dbioprinting is now a tangible option for the delivery of bioscaffolds capable of regenerating thedeficient cartilage tissue. Our teamhas developed a handheld device, the Biopen, to allow in situadditivemanufacturing during surgery. Given its ability to extrude in a core/shellmanner, the Biopencan preserve cell viability during the biofabrication process, and it is currently the only biofabricationtool tested as a surgical instrument in a sheepmodel using homologous stem cells. As a necessary steptoward the development of a clinically relevant protocol, we aimed to demonstrate that our handheldextrusion device can successfully be used for the biofabrication of human cartilage. Therefore, thisstudy is a required step for the development of a surgical treatment in human patients. In this workwespecifically used human adipose derivedmesenchymal stem cells (hADSCs), harvested from the infra-patellar fat pad of donor patients affected byOA, to also prove that they can be utilized as the source ofcells for the future clinical application.With the Biopen, we generated bioscaffoldsmade of hADSCsladen in gelatinmethacrylate, hyaluronic acidmethacrylate and cultured in the presence ofchondrogenic stimuli for eight weeks in vitro. A comprehensive characterisation including gene andprotein expression analyses, immunohistology, confocalmicroscopy, second harmonic generation,light sheet imaging, atomic forcemycroscopy andmechanical unconfined compression demonstratedthat our strategy resulted in human hyaline-like cartilage formation. Our in situ biofabricationapproach represents an innovationwith important implications for customizing cartilage repair inpatients with cartilage injuries andOA.

1. Introduction

Hyaline cartilage tissue regeneration remains a chal-lenge for musculoskeletal tissue engineering. As ahighly specialised tissue, it is characterised by low cellnumbers, an avascular, non neural and aliphatic

environment, a distinct three-dimensional collagenfibre structure, aggregated proteoglycans and matrixorganisation [1]. All these characteristics confer spe-cific mechanical properties [2, 3] and make articularcartilage a challenging tissue to regenerate in vitro andin vivo. Promising surgical treatments include

RECEIVED

20March 2018

REVISED

21 July 2018

ACCEPTED FOR PUBLICATION

8August 2018

PUBLISHED

21August 2018

© 2018 IOPPublishing Ltd

Autologous Chondrocyte implantation (ACI) andmatrix-induced ACI (MACI) that some have reportedto be superior to self-repair inducing techniques suchas microfractures [4, 5]. Nevertheless, MACI and ACIare multistage, complex procedures. These techniquesrequire a double operation: the surgical excision ofnative cartilage tissue and the expansion of adultchondrocytes in vitro before implantation [5]. Pre-vious studies have reported that 68% of patientstreated with the clinical product approved by the USFood and Drug Administration (Carticel®) had graftfailure, delamination or tissue hypertrophy [6]. Themajor drawbacks include the need for two operations,and that adult chondrocytes have the propensity toundergo de-differentiation [7], thus losing the cap-ability to produce hyaline cartilage extracellularmatrix(ECM) and giving rise to reparative tissue that lacksdurability [8, 9]. The lack of durability of repairedcartilage leads to the development of osteoarthritis(OA) [10, 11], which severely limits the quality of lifeof affected individuals and represents one of thegreatest contributors to healthcare expenditure world-wide [12].

Regenerating healthy and long lasting articularhyaline-like cartilage is therefore considered a key aimin musculoskeletal tissue engineering [13]. In thisregard, mesenchymal stem cells (MSCs) have beenheralded as an appropriate candidate for cartilagerepair owing to several specific characteristics such asinherent chondrogenic property, ready availability,and cell homing potential [14]. Despite these qualities,their use is still limited, with only 16% of reported celltherapy procedures for cartilage repair registered athttps://clinicaltrial.gov/ usingMSCs [15].

3D bioprinting has added considerably to standardtissue engineering techniques, thanks to its precisionand control [16]. This promising technique can beapplied to regenerate biological tissues such as boneand cartilage [17, 18]. A key challenge is now to bringcumbersome bench-based technology to the operatingroom for real-time application.

3D printing of scaffolds for chondral and osteo-chondral repair has been tried, using a mosaicplasty-type technique, based on the implantation of scaffoldsin defects that are pre-made in the laboratory beforesurgery [19–22]. 3D printing has the advantage ofbeing coupled with the use of ‘custom-designed’implants modelled on pre-operative MRI and/or CTscans. For ‘custom designed’ bench-based 3D printingtechniques a highly detailed pre-operative image of thecartilage defect needs to be obtained, so that the 3Dprinter can create the desired bioscaffold to match thespecific defect. Despite these strategies, however, amismatch between the bioscaffold and the host defectwill likely remain because surgical debridement beforeimplantation alters the anatomy previously imaged,and even the most advanced imaging techniques stillfail to identify small gaps and defects. These issuesinevitably impede perfectlymatching off-the-shelf and

custom-designed prefabricated bioscaffolds. Themechanical mismatch between implanted scaffoldsand surrounding tissues may inhibit cartilage regen-eration at defect sites leading to implant failure [23]. Intheir literature review, Mollon et al summarized thelimitations associated with current cartilage repairtechniques and noted that graft-site mismatch andsize-depthmismatchwere themost frequent [24].

To tackle this challenge, we propose in situ additivemanufacturing as an alternative to computer aided 3Dprinting. The advantages of in situ biofabrication forcartilage regeneration include the possibility of per-fectly match the defect geometry without specific pre-liminary image analysis, shaping the bioscaffoldwithin the defect and achieving the best possible con-tact between the bioscaffold and the host tissue. In situbioprinting has been proposed as a valid option todeliver at the time and point of need a bioscaffold cap-able of regenerating deficient tissue [25, 26] and tolimit the graft-site mismatch. Also a number of ex vivoapproaches have been proposed [27, 28] to test theeffective use of cell based therapies within a defineddefect, but to our knowledge our extrusion device, theBiopen [29], remains the only hydrogel injection sys-tem that has been tested as a surgical instrument in alarge animal model of cartilage regeneration [30]. TheBiopen is an advanced handheld co-axial extrusiondevice that allows the deposition of cells embedded ina hydrogel material in the surgical setting. Throughthis device, stem cell laden hydrogel scaffolds can beextruded in a co-axial core/shell distribution, whichfavourably affects cell survival while maintaining ade-quate mechanical properties for cartilage tissue engi-neering purposes. This result is achieved thanks to thesegregation of the cells in the core away from the pho-tocrosslinking reaction confined to the shell compart-ment [31].

Our in vivo study was performed using allogenicovine adipose derived stem cells [30]. To progresstowards translating our biofabrication tool into clin-ical practice, we aim to demonstrate that the advan-tages of the Biopen are pivotal for the biofabrication ofthe human cartilage. Therefore, this study is a requiredstep in the development of a clinically relevant proto-col for the application of this biofabrication process inthe clinical practice (figure 1).

In this study we specifically used human adiposederived mesenchymal stem cells (hADSCs), harvestedfrom the infra-patellar fat pad (IPFP) of donor patientsaffected by OA, envisioning them as the source of cellsfor the future clinical application. We screened threedifferent hADSCs from three different patients to takein account also the inter-patient variability. Thus, foreach cell line we evaluated stemness and differentia-tion potential.

We used the Biopen to generate bioscaffolds inwhich hADSCs were laden in gelatin methacrylate(GelMa) and hyaluronic acid methacrylate (HAMa)[29]. We cultured the bioscaffolds in chondrogenic

2

Biofabrication 10 (2018) 045006 COnofrillo et al

media for eight weeks and we analysed their ability toform hyaline-like cartilage using molecular, imagingandmechanical analyses.

To our knowledge, we are the first to describe thecapacity of a surgical device to biofabricate in situ aregenerative stem cells niche. The impact of this tech-nology has the potential to change the clinical approa-ches in the surgical practice for the regeneration oftissues and organs.

2.Materials andmethods

2.1. Experimental procedureHuman stem cells were harvested and isolated fromthe IPFP of patients undergoing total joint kneereplacement. A comprehensive profile analysis ofimmunophenotypic characterization and evaluationof the differentiation potential was used to assess thestemness of three different cell lines from threedifferent donors to be used for the bioprinting step.We then utilized the Biopen to biofabricate thescaffolds. To control the size and shape of our samples,we used PDMS cylindrical moulds of 1 cm diameterand 1 mm thickness with a total volume of approxi-mately 80 μl, so<2 cm2, which is considered a criticalsize defect in human cartilage injuries [32]. Thelimitations of the PDMS material, which is morecompliant and flexible than native cartilage, thisbioprinting approach guarantees reproducibility ofstructural organization and cell number, with suitablegeometry for mechanical testing, and was used for theentire set of experiments. We then extruded multiplelayers until the mould was filled to recreate a 3Dstructure.

After the photopolymerization reaction, wemoved the samples to static culture conditions and wecultivated the samples for 8 weeks in control or chon-drogenic differentiation media. The samples, threebiological replicates [n=3] for each of the threepatients derived cell lines, were collected and analysedevery 4 weeks for a total of three time points. We then

analysed the capability of our biomanufactoring strat-egy to produce hyaline-like neocartilage. The cellularresponse to the chondrogenic stimuli was determinedby gene and protein expression analysis. The buildingof new tissue was defined by protein localization andorganization of the main components of hyaline carti-lage. The function of the neocartilage was then relatedto the acquisition of mechanical properties over time(figure 2).

2.2. Synthesis of GelMa/HAMaGelatin-methacryloyl/hyaluronic acid methacryloyl(GelMa/HAMa) was synthesized as previouslydescribed [31]. Briefly, the materials were dissolved toa final concentration of 100 mgml−1 GelMa and20 mgml−1 HAMa (10%GelMa-2%HAMa) in sterilePBS (Sigma-Aldrich), containing 100 Uml−1 penicil-lin and 100 μgml−1 of streptomycin (Gibco).

2.3. Stem cells isolationhADSCs were isolated from human IPFP obtainedintraoperatively from total knee arthroplasty withinformed consent and approval. Patients with mild/severeOAwere recruited. A total of three patients wereselected for IPFP harvest. Use of all human samplesand procedures in this study was approved by theHumanResearch Ethics Committee ResearchGovern-ance Unit of St. Vincent’s Hospital, Melbourne,Australia [HREC/16/SVHM/186] and all the experi-ments were performed in accordance with relevantguidelines and regulations.

hADSCs were isolated and expanded as previouslydescribed [33]. Briefly, the fat was diced using a sterilescalpel and digested with 0.1% collagenase type I(Worthington Biochemical) for 3 h at 37 °C underconstant agitation, filtered through 100 μm cell strai-ner nylon (BD Falcon) and centrifuged at 400 g atroom temperature for 5 min to separate the stromalfraction from the floating adipocytes. The supernatantwas discarded and the cell pellet was resuspended inRed Cell Lysis Buffer (160 mM NH4Cl; Sigma-

Figure 1.The path towards the development of a clinical treatment. The flow chart lists the steps achieved (in green) and the stepsahead (in red). The current study is indicated as the neocartilage biofabrication step and represents the last in vitro analysis before theex vivo and in vivo animal studies.

3

Biofabrication 10 (2018) 045006 COnofrillo et al

Aldrich) and incubated at room temperature for10 min. The lysate was centrifuged at 400 g at roomtemperature for 5 min and filtered through a 40 μmnylon cell strainer (BD Falcon). The isolated cells werethen plated in monolayer culture and allowed toadhere for 48 h in culture media containing low glu-cose DMEM (St. Louis, LA, USA) supplemented with10%FBS (GIBCO, Thermo Fisher Scientific Inc.,Wal-tham, MA, USA), 100 Uml−1 Penicillin and100 μg ml−1 Streptomycin solution (GIBCO), 2 mML-Glutamine (GIBCO), and 15 mMHEPES (GIBCO),20 ng ml−1 epidermal growth factor and 1 ng ml−1

fibroblast growth factor (R&D Systems Inc., Minnea-polis, MN, USA). Non-adherent cells were thenremoved and the medium was replaced with freshmedium. Cells were further cultivated and expandedfor cell surface epitope immunophenotypic character-ization (flow cytometry) and multilineagedifferentiation.

2.4. Stemness immunophenotypic characterizationanddifferentiation potentialImmunophenotypic characterization of hADSCs wasperformed by fluorescence flow cytometry analysis ofcell–surface markers at passage 2 (P2). Cells werelabelled with monoclonal antibodies against CD14,CD31, CD34, CD45, CD73, CD90, CD106, CD146–FITC conjugates and CD105, HLA-DR, CD29, CD44,CD49e—APC conjugated (Affimetrix eBiosceince,ThermoFisher Scientific). Control samples werelabelled with isotype-matched control antibodiesIgG1K-FITC and IgG1K-APC (Affimetrix

eBiosceince). In brief, cells were trypsinized andaliquoted at a concentration of 2.5×105 cells ml−1,fixed in 0.5% paraformaldehyde for 30 min at 4 °C,and washed once in PBS 1X. Next, samples wereincubated with either conjugated specific antibodiesor isotype-matched control at the manufacturer’srecommended concentrations, diluted in PBS 1Xsupplemented with 5% FBS (FACS buffer). Labelledcells were washed twice, suspended in FACS buffer,and analysed using a FC500 flow cytometer (BeckmanCoulter).

Chondrogenic differentiation was induced usingthe micromass pellet culture technique as describedpreviously [34, 35]. Briefly, 2.5×105 confluent P03cells ml−1 were placed in 1.5 ml screw cap tubes (Sar-stedt), centrifuged at 400 g for 5 min to form the pel-lets and cultivated with either non chondrogeniccontrol medium consisting of DMEM high-glucose(Lonza), 100 Uml−1 penicillin and 100 μg ml−1 ofstreptomycin (Gibco), 1X Glutamax (Gibco), and15 mM HEPES (Gibco), or chondrogenic mediumconsisting of DMEM high-glucose (Lonza),100 Uml−1 penicillin and 100 μg ml−1 of streptomy-cin (Gibco), 1XGlutamax (Gibco), and 15 mMHEPES(Gibco), 1% insulin-transferring-selenium (Sigma-Aldrich), 100 nM dexamethasone (Sigma-Aldrich),50 mg ml−1 ascorbate-2-phosphate (Sigma-Aldrich),10 ng ml−1 TGFβ3 (Prepotech), and 10 ng ml−1

BMP6 (R&DSystems).Osteogenic differentiation was induced at P03 by

seeding hADSCs in α-MEM supplemented with 2%FBS in six-well plates at 5×105 cells per well. The

Figure 2.Graphical representation of the experimental procedure and the corresponding time line. Three different cell lines isolatedfrom three patients were analysed in triplicate.

4

Biofabrication 10 (2018) 045006 COnofrillo et al

next day, an osteogenic inducing cocktail composed of10 mM β-glycerophosphate (Sigma, St. Louis, MO,USA), 50 μg ml−1 ascorbic acid (Sigma) and 100 nMdexamethasone (Sigma)was added.

As a negative control, cells seeded under the sameconditions were maintained in a non-inducingmedium.

The medium was changed every 2 d and three bio-logical replicates [n=3] per time point and type ofanalysis at 0 and 4 weeks of culture/group were pro-cessed. Immuno/histochemistry and reverse tran-scription-quantitative polymerase chain reaction (RT-qPCR)were performed as described below.

2.5. Generation of bioscaffolds with the BiopenCo-axial extrusion was performed using the handheldextrusion system (Biopen) using the same parameterspreviously described [31]. Briefly, both Biopen cham-bers were loaded with 10% GelMa-2% HAMa.hADSCs were mixed to a final concentration of10×106 cells ml−1 and carefully loaded in the corechamber, while for the shell chamber 0.1% w/vLithium-acylphosphinate (Tokyo Chemical IndustryCo., Tokyo, Japan) was mixed through the GelMa/HAMa. The temperature of the solutions was stabi-lized at 37 °C prior to preparation. Samples wereBiopen extruded into PDMS moulds to produce disc-like shaped bioscaffolds (height=1 mm,diameter=10 mm). Immediately after extrusion thesamples were then UV irradiated at room temperaturefor 10 s, using a 365 nm UV source (Omnicure LX400+, Lumen DynamixLDGI) fitted with a 12 mm lens(25 mm focal distance) with a light intensity of700 mW cm−2.

2.6. In vitro culture of bioscaffoldsThe generated bioscaffolds were then transferred to a24 well plastic plate, washed three times in PBS 1X and1 ml of chondrogenic or controlmediumwas added toeach well. Pictures of the entire samples were takenwith a 7megapixel camera.

Chondrogenic differentiation was induced as pre-viously described for the pelleted culture cells (para-graph 2.3.2.) and three biological replicates [n=3]per time point and type of analysis at 0, 4 and 8 weeksof culture/group (differentiated versus undiffer-entiated)were processed.

2.7.Histological analysisFor chondrogenic analysis, pelleted cells and Biopenmanufactured samples were fixed in 1% paraformal-dehyde (Santa Cruz Biotechnology, Dallas, TX, USA)for 4 h at room temperature, embedded in O.C.T. TMCompound (Tissue-Tek, Sakura, Leiden, Nether-lands) and flash frozen in liquid nitrogen. Cryosec-tions of 10 μm thickness were mounted onto glassslides and stained with Safranin O (Sigma-Aldrich) for10 min, dipped in 95% and 100% EtOH, cleared three

times for 1 min each in Xylene (Chem-Supply, GILL-MAN, SA, Australia) and then mounted in Pertexmedium (GraleHDS, Ringwood, VIC, Australia).

For osteogenic differentiation analysis, culturedcells were stained with Alizarin Red-S (AR-S) (Sigma)to reveal the deposition of a calcium-rich mineralizedmatrix [36]. Quantification of Alizarin was performedby Cetylpyridinium chloride extraction and spectro-photometricmeasurement at 562 nm.

Samples were imaged using an epifluorescentinverted NikonTiE microscope equipped with a DS-Ri2 and NIS-Elements software using a Plan FluorELWD 20X DIC L NA 0.45 objective. Figure panelswere assembled using Photoshop software (Adobe).

2.8. RNA extraction andRT-qPCRTotal RNA from hADSCs, cultured, pelleted orbioprinted, were harvested at indicated time pointsusing Tri Reagent (Ambion, Austin, TX, USA) accord-ing to the manufacturer’s protocol. DNA contamina-tion were digested by DNAse I (Sigma). Reversetranscription (RT) was performed using OmniscriptRT kit (Quiagen) following the manufacturer’s proto-col. The relative amounts of COl2A1, COL1A2,ACAN, SOX9, RUNX2, COL10A1, RPL13 andGAPDH RNAs were evaluated with TaqMan Geneexpression assay (Applied Biosystems, Foster City, CA,USA) using the following probes:

SOX9 (Hs00165814_m1), ACAN(Hs00153936_m1), COL2A1 (Hs00264051_m1) andCOL1A2 (Hs00164004_m1) RUNX2(Hs01047973_m1), COL10A1 (Hs00166657_m1) astarget genes; GAPDH (Hs02786624_g1) and RPL13a(Hs04194366_g1) as housekeeping genes. qPCR wasperformed on a QuantStudio 6 Flex Real-Time PCRSystem (Thermo Fisher Scientific) and relative quanti-fication was calculated with the 2E−ΔΔCT method.The mean ΔCT value of the control sample was usedin each experiment to calculate the ΔΔCT value ofsample replicates by using an average of two differenthousekeeping genes (RPL13 and GAPDH). Statisticalanalysis on the data obtained was performed by usingunpaired T-Test on three experimental replicates byconsidering significative data with a P value<0.05.Graphs were designed using Prism 5 software (Graph-Pad) and figure panels were assembled using Photo-shop software (Adobe).

2.9. Protein extraction andWestern blot analysisBiopen printed samples were collected at the indicatedtime points, washed three times for 5 min each withPBS 1X on a rotating platform at room temperature,transferred in a 1.5 ml Eppendorf tube, then flashfrozen in liquid Nitrogen and stored at−80°C.Wholeprotein content was extracted on ice by manuallybreak with a pestle in 30 μl/samples of 2X RIPA buffer(750 mM NaCl, 5% Triton X100; 2.5% SodiumDeoxycolate; 10% SDS; 1M Tris, pH 8) with the

5

Biofabrication 10 (2018) 045006 COnofrillo et al

addition of protease inhibitor cocktail (Roche). Sam-ples were incubated on ice for 20 min and cleared bycentrifugation at 14 000 rcf for 20 min at 4 °C. Thesupernatant was recovered and carefully transferred ina new labelled tubes. Protein extract was quantifiedspectrophotometrically with the BCA Protein Assay(Thermo scientific). The same amount of proteins wasseparated in loading buffer (Thermo scientific) in 4%–

12% Bolt Bis-Tris precast Gel (Thermo scientific) inMES buffer (Life technologies) for 30 min at constantvoltage (180 V). Spectra Multicolor Broad rangeprotein ladder (Thermo scientific) was used as aprotein standard. After the electrophoretic separation,proteins were blotted to PVDFmembrane with iBlot 2system (Life Technologies), following manufacturer’sinstructions. Protein transfer was assessed by Ponceau(sigma) staining. The membrane was blocked againstnon specific bindings with 5% non-fat dry milk(Coles) in TBS-T (50 mM Tris-Cl, pH 7.6; 150 mMNaCl; 0.1% Tween 20) for 1 h at room temperature.After threewashes inTBS-T for 10 min, themembranewas incubated overnight at 4 °C with rabbit mono-clonal anti human SOX 9 antibody (#EPR14335,Abcam) or mouse monoclonal anti human β-Actin(#A5316, Sigma-Aldrich) in 3.5% Bovine SerumAlbumin in TBS-T. After three washes in TBS-T for10 min, the membrane was incubated for 1 h at roomtemperature with anti-rabbit secondary antibodyHRPconjugated (Dako). Finally, after three washes in TBS-T for 10 min, the signal was revealedwithWestern BlotLightning ECL pro kit (PerkinElmer) and imaged withChemidocMP imaging system (Biorad). Figure panelswere assembled using Photoshop software (Adobe).

2.10. Immunostaining analyses andfluorescentimagingFor fluorescence analysis, 10 μm thickness slices fromboth pelleted cell culture and Biopen printed, werewashed three times in PBS1X and permeabilized for15 min in PBS 1X-0.1% TritonX-100 (PBT). Antigenretrieval was performed only for hydrogel printedsamples by adding 1 mgml−1 Hyaluronidase (SIGMA,#H6254) diluted in PBS 1X and incubating 30 min atroom temperature. After three washes 5 min each inPBS 1X, samples were dropped in blocking solution(10% goat serum diluted in PBT) for 60 min at roomtemperature and then incubated overnight at 4 °Cwith mouse anti-human Collagen II (#II6B3, DSHB),goat anti-human Collagen I (#sc8784, Santa Cruz),mouse anti-human Proteoglycan (#MAB2015,MerckMillipore) and mouse anti-human Collagen X(#ab49945, Abcam), diluted 1:250 in blocking solu-tion. The day after, samples were washed three timesfor 10 min each and secondary antibodies both diluted1:100 in blocking solution were added, respectivelyanti-mouse IgG Alexa Fluor-647 (#715-605-151,Jackson Immuno Research) and anti-goat IgG AlexaFluor-546 (#A11056, Thermo Fisher Scientific Inc.),

and incubated for 2 h at room temperature. After threewashes 5 min each in PBS 1X, actin was labelled withTexas Red-X Phalloidin (#T7471, Thermo FisherScientific Inc.) 1:100 diluted in PBS 1X for 60 min atroom temperature and nuclei were stained by incuba-tion with 5 μg ml−1 DAPI (Thermo Fisher ScientificInc.) diluted in PBS 1X for extra 60 min at roomtemperature.

The sections were washed three times 5 min eachin PBS 1X, mounted with Fluoromount-G (SouthernBiotech, Birmingham, AL, USA) onto glass slides. Pel-let sections were imaged using an epifluorescent inver-ted NikonTiE microscope equipped with a DS-Ri2and NIS-Elements software using a Plan Fluor ELWD20X DIC L NA 0.45 objective. Biopen printed sampleswere imaged with NikonA1R confocal microscopeusing aNikon Plan

VC 20×DICN2N.A. 0.75 objective lens and ‘Scanlarge image’ fromNIS-Elements software tool was usedto image a largerfield of view.

Digital images were processed using NIS-Elementssoftware (Nikon, Amsterdam, Netherlands) and Pho-toshop software (Adobe) was used to assemble thefigure panels. All the images shown in this study arerepresentative of at least three independentexperiments.

The fluorescence measurement analysis offigure 6(C) was performed with the ‘Annotations andMeasurements’ feature ofNIS-Element software.

The intensity of the Collagen type II corresp-onding channel was calculated by normalizing the sig-nal for the identified positive area. The percentage ofthe positive area was calculated in relation to the totalarea of the field of view, which was the same for all theimages acquired and that corresponded to 3 mm2.

Those analyses were performed on three differentfields of view for each sample.

2.11. Second harmonic generation (SHG)microscopyRegions of interest images (5 mm×5 mm) wereacquired on a fully automated, programmable, multi-photon imaging platform (Genesis® 200, HistoIndexPte Ltd, Singapore) following sectioning and mount-ing of 10 μm thick cryosections washed out fromOCT. Laser excitation occurred at 780 nm. Forward-scatter two-photon excitation (TPE) and SHG signalswere detected using dedicated photomultiplier tubesfor each channel. Magnification was set to 20×. TPEsensitivity was set to 0.7, and SHG sensitivity to 0.6. Abandpass filter with center wavelength at 550 nm andbandwidth of 88 nm was set in front of the TPEphotodetector. Laser baseline power was set at 0.5, andthen stepped down by 60% using a set of 2 opticalattenuators (0.1OD, 0.3OD—‘Low’ laser power con-figuration). No change wasmade to the polarization ofthe laser source, and no polarization was used on the

6

Biofabrication 10 (2018) 045006 COnofrillo et al

SHG detection channel. Tissue areas were scannedoncewith no frame averaging feature.

2.12.Whole sample staining and light sheet imagingEntire bioscaffolds were transferred to clean 12 wellplates, washed 3 times 10 min each in PBS 1X and fixedin Paraformaldehyde 4% for 30 min. After washingthree times for 10 min each in PBS 1X, samples werepermeabilized in PBT (0.1% Triton in PBS 1X) for60 min. Actinwas labeledwith Texas Red-Xphalloidin(#T7471, Thermo Fisher Scientific Inc.) 1:100 dilutedin PBS for 2 h. The excess of phalloidin was removedby washing 3 times 10 min each in PBS 1X. Nucleiwere stained by incubation with 5 μg ml−1 DAPI(Thermo Fisher Scientific Inc.) diluted in PBS 1X forextra 60 min. Samples were finally washed 3 times10 min each in PBS1X and left at 4 °Cuntil imaging.

For light sheet imaging, samples were hooked andloaded in a glass capillary (size 2 black, inner diameterof capillary ∼1 mm, #701932, BRAND GmbH). Theglass capillary was mounted in the chamber, whichwas filled PBS 1X. A Lightsheet Z.1 microscope (CarlZeiss Microscopy GmbH, Jena, Germany), with 20×/

1.0 NA water dipping detection objective and two-sided 10×/0.2 illumination optics, equipped with twoPCOEDGE 4.2 cameras (sCMOS sensor, square pixelsof 6.5×6.5 μm side length, 2048×2048 pixel reso-lution, 3-channel images, 16 bit dynamic range) (PCOAG, Kelheim, Germany)were used for light sheet ima-ging. For DAPI and Phalloidin-actin 1.5%–405 nmand 8.6%–561 nm laser powers respectively and 50msexposure time were used. Pivot scanning mode (CarlZeiss) was used to deliver homogeneous illuminationand prevent shadows along the illumination axis. Forall 3D datasets, a z-interval of 2 μm with a zoom of0.50 was used. The volumetric images were1131×1131×1000 μm (1920×1920×500 pix-els) in size, with 0.589×0.589×2.000 μm3 resolu-tion. To counteract the degradation of the light sheetby the high amount of scattering, the specimen wassequentially illuminated through each of the twoopposing illumination objectives, generating pairs ofilluminated single-side images, and then combinedinto fused single illumination optical sections with aconsiderably improved penetration depth. Allacquired LSFM raw data were processed using ZEN2011 imaging software (Carl Zeiss, Germany). Opticalsections were merged and combined into a singlethree-dimensional data set and reconstructed usingIMARIS 9.0 software.

2.13. Compression testingThe tests were performed at room temperature using aTA Electroforce 5500 mechanical loading device (TAInstruments, New Castle, USA) fitted with a calibrated22 N load cell. The contact point between the twoplates was recorded. Then, the sample was placedbetween two 4.2 cm diameter compression plates, in

an unconfined setting as previously described [37].The displacement of the upper plate was controlled bya ramp function, at a rate of 0.01 mm s−1, until a totaldisplacement of 25%of the sample height. The contactarea of the sample with the plate was measured bymicroscopy imaging before the test. Additionally, thepoint of inflexion of the load versus time curve showedthe contact point between the sample surface and thecompression plate to give the sample height. Load anddisplacement measurements were converted intostress (σ) and strain (ε) data using the sample surfacearea and height. The compressive modulus was thencomputed using stress data between 10% and 15%strain as follows: Ec=(σ15−σ10)/(ε15−ε10). Threebiological replicates [n=3] for each of the threepatients per time point and type of analysis at 0, 4 and8 weeks of culture/group (differentiated versus undif-ferentiated) were processed. Data graphs weredesigned using Prism5 software (GraphPad).

2.14. Atomic forcemicroscopyAFM nanoindentation was performed on an MFP-3D(AsylumResearch) using a contact modeMLCT probe(Bruker Nano Inc.) with a stiffness calibrated as0.135 Nm−1. Force curves were performed to anindentation force of 5 nN at an approach rate of2 m s−1. The Young’sModulus was obtained using theHertz indentation model approximating the tip-shapeas a 19.2° cone. Bioprinted hydrogel samples weresubmerged in PBS 1Xduringmeasurement.

Sixteen indentations across 5 regions were per-formed in astandardized fashion, making 80 forcecurves per sample. Three samples for each of the con-trol week 8 and chondrogenic week 8 groups weremeasured. The same AFM probe was used for allmeasurements.

2.15. Statistical analysisAll statistical analyses were performed using Prism 5(GraphPad) software. Differences between the exper-imental groups were determined using unpaired t testor one-way Anova tests with Dunnett’s corrections asindicated. Significancewas represented as follow:

*=p�0.05; **=p�0.01; ***=p�0.001;not significant (n.s.)=p>0.05.

3. Results

3.1. Comprehensive profile analysis of hADSCsidentifies the IPFP as a source of stem cells for in situbiofabricationIn order to account for any possible inter-patientvariability, we decided to test our biofabricationstrategy on different stem cells lines derived from threedonors. Thus, we isolated infra-patellar hADSCs fromthree patients undergoing total joint knee replacementand characterized their stemness and differentiationpotential.

7

Biofabrication 10 (2018) 045006 COnofrillo et al

We first evaluated the cell morphology, plasticadherence after isolation and expression of typicalstemness surface markers. The three lines displayedthe typical spindle-shaped appearance at early pas-sages and exhibited a smoothened morphology afterlong-term expansion, developing a larger and moregranular cytoplasm (data not shown). Flow cytometryanalysis documented that>90%of hADSCs expressedCD90, CD73, CD105 CD29, CD49.More than 90% ofcells were also positive to the hyaluronan receptorCD44, which role is important in the early stages ofcartilage development [38]. More than 60% ofhADSCs were positive for HLA-DR, whereas theexpression of the hematopoietic markers CD14,CD31, CD34, CD45 was <3%. The markers CD106and CD146 were expressed in less than 10% ofhADSCs as previously shown [39]. No significant dif-ference among the three patients was scored, confirm-ing that with our established laboratory procedures weare able to isolate and expand a homogenous stem cellpopulation despite the donor tissue variability (sup-plementary figure 1 is available online at stacks.iop.org/BF/10/045006/mmedia). The chondrogenicpotential was then tested by pellet culture in vitro dur-ing a period of 4 weeks, and production of chondralECMwas confirmed (figure 3).

The initial assessment of chondrogenesis wasbased on the size and appearance of the pellets. Aftertheir generation, pellets grew in size and displayed aconsistent white opaque macroscopic appearance(figure 3(A)). Safranin O staining showed accumula-tion of glycosaminoglycans (GAGs) deposition

(figure 3(A)) and changes in cell morphology fromround shapes at week 0 to elongated cells encapsulatedby an abundant ECM in the entire pellet structure(figure 3(A)). Gene expression data confirm theincrease expression of Collagen type II (COL2A1) andtype I (COL1A2), Aggrecan (ACAN) and Sox-9(SOX9) markers during the differentiation process(figure 3(B)). COL2A1 and ACANwere not detectablein both 2D and pellet cultured hADSCs at steady state,but were activated only during the chondrogenesisprocess as a consequence of SOX9 increased expres-sion. Finally, the immunostaining analyses revealed anincreased expression of both Collagen type II and Iduring time, with a much stronger intercellular accu-mulation of Col II (figure 3(C)). As previously shown[34, 35], hADSCs from fat pad display the tendency toexpress higher level of collagen type II RNA and pro-tein respect to Collagen type I during in vitro chon-drogenesis under stimulation of TGFβ3 andBMP6.

To confirm further the differentiation potential ofthe isolated populations, we also tested the osteogenicpotential through Alizarin red staining and geneexpression (supp. figure 2).

3.2. Neocartilage generation in vitro3.2.1. GAG accumulation during chondrogenicdifferentiation of hADASCs in GelMa/HAMabioscaffoldsWe used the Biopen to generate bioscaffolds in whichhADSCs were laden in GelMa and HAMa. BothGelatin (denatured collagen) and hyaluronic acidretain relevant cell-binding sites, while the addition of

Figure 3.Comprehensive profile analysis of infrapatellar fat pad derived hADSCs. (A)Histological analysis on 10 μmcryosectionsstainedwith SafraninO to detect accumulation of GAG in the ECM. In the bottom left, the insets show the brightfield images of thewhole pellets. The small panel on the right upper corner at week 4 shows a highermagnification image of cells encapsulated andsurrounded by the ECM. (B)The graphs represent the fold changes calculatedwith 2^ΔΔCΤmethod of Collagen type II (COL2A1)and type I (COL1A2), Aggrecan (ACAN) and Sox-9 (SOX9)markers. Fold Changes were normalized for RPL13a andGAPDH. Errorbars represents standard deviation between three biological replicates and significant activity was calculatedwith unpaired t-test.These analyses have been performed on the hADSCs lines derived from three different patients and statistical analyses were performedas described inMaterials andmethods section. (C) Immunostaining analysis on 10 μmcryosections frompellet at weeks 0 and 4 ofchondrogenesis, performed using antibodies against Collagen type II (Col II) and I (Col I) shown in green, andDAPI shown in violetto detect cells nuclei. Note: the images inA andC are themost representative selected from the immunohistology performed on thethree hADSCs patient derived cell lines.

8

Biofabrication 10 (2018) 045006 COnofrillo et al

methacrylate groups renders the material amenable torapid crosslinking using light, creating a hydrogelstructure which is stable at physiological temperaturesand achieves mechanical properties approaching thatof native hyaline cartilage. As described in ourprevious paper [31] the 10 s crosslinking time at700 mW cm−2 with a 365 nm UV light source do notcause proliferation inhibition to hADSCs in the samecore/shell bioprinting conditions.

We produced tridimensional structures of the sizeof a critical cartilage injury, where a critical size impliesa defect whichwill not heal spontaneously without anyintervention [40, 41].

The neocartilage development was first qualita-tively evaluated via the appearance of the bioscaffolds.The macroscopic brightfield images showed a markeddifference in the opacity of the printed samples with aconsistent white opaque macroscopic appearancedeveloping over time only in the chondrogenic stimu-lated samples and not in the control group(figure 4(A)).

The deposition of the ECM in the hydrogel matrixwas then verified histologically, by staining thesections derived from the same samples with SafraninO, as a reporter of GAG content [42] (figure 4(B)). Inthe acellular scaffold, Safranin O stains a non-homo-genous material. On the other hand, in the cell–ladenbioscaffolds, the stem cells occupied defined lacunaein the surrounding hydrogel and exhibited a roundshaped morphology when maintained in a controlnon-chondrogenicmedia.

Upon chondrogenesis, the bioscaffolds showed aprogressive formation of a compact and homogeneousmatrix at weeks 4 and 8 that is positive to Safranin O,demonstrating the accumulation of GAG(figure 4(B)). The cells appeared elongated throughouttheGAG forming a continuous structure. This analysis

indicated a cell response to chondrogenic stimuli inGelMa/HAMa bioscaffolds and the consequent accu-mulation ofGAGover time.

3.2.2. Gene expression analysis evidences chondrogenicdifferentiation of hADSCs inGelMa/HAMabioscaffoldsTo elucidate the capability of hADSCs to undergochondrogenesis in the bioscaffolds, we analysed thegene expression of chondrogenic markers over aperiod of 8 weeks by RT-qPCR and Western blotanalyses.

The three hADSCs lines showed a progressiveincrease of the master transcriptional regulator of thechondrogenic program SOX9 both at transcriptionaland translational level [43, 44] (figures 5(A), (B)).

The intracellular accumulation of SOX9 led inturn to a progressive activation of its target genesCOL2A1 and ACAN (figure 5(A)). The three hADSCslines did not exert any basal expression of the ECMcomponents before the induction of chondrogenesis,as also confirmed by electrophoretic analysis of PCRamplified cDNA (figure 5(A), lower panels). ACANand COL2A1 expression was detectable 4 weeks afterthe start of the chondrogenesis differentiation, exert-ing an average fold change of 20 and of 140 respec-tively during the following 4 weeks. hADSCs alreadyexpressed COL1A2 before the chondrogenesis differ-entiation and its level increase during time. Never-theless, COL1A2 showed only a 10 fold increase fromweek 0, while Collagen type II, which was not detect-able at week 0, increases up to 200 folds at week 8 com-pared to week 4. Thus, the ratio between COL2A1 andCOL1A2 indicated the generation of hyaline-like car-tilage rather than fibrocartilage, as described in pre-vious works [45, 46] and in agreement with pelletculture data (figure 3).

Figure 4.Neocartilage building and evaluation of GAG accumulation in Biopen extruded bioscaffolds. (A)Photographs showbioscaffolds from the control and the chondrogenic groups at the indicated time points. Thewhite boxes outline the areas imaged inthe histological analysis shown in the corresponding panels below. (B)Histological analysis from the corresponding samples of 10 μmcryosections, stainedwith SafraninOdye. Imageswere acquiredwith a 20× objective lens. Note: the images inA andB are themostrepresentative selected from the analyses performed on bioscaffold constituted by three hADSCs lines derived from three differentpatients.

9

Biofabrication 10 (2018) 045006 COnofrillo et al

3.2.3. Generation of hyaline-like cartilageTo analyse the process of biofabrication of chondraltissue in our Biopen extruded samples, we cryosec-tioned the bioscaffolds at the indicated time pointsand performed histological analysis by immunostain-ing with antibodies against Collagen type II (Col II),Collagen type I (Col I) (figure 6) and Proteoglycan(supp. figure 3). The data obtained matched the geneexpression analysis: Col I was already expressed at thestart of the chondrogenic differentiation, while Col IIwas detected after 4weeks (figure 6(A)).

Despite the fact that the intensity of the Col I signalincreased during chondrogenesis, the intercellular ColII deposition and accumulation in the hydrogel matrix

dramatically differed fromCol I, reaching a strong anduniform pattern at week 8 of chondrogenesis. Thethickness of the Col II-rich stripe increased from anaverage length of 400 μm at week 4 up to 800 μm atweek 8 (figure 6(A)). Those data are consistent withRT-PCR expression analyses where we demonstratedthat the ratio between Collagen II and I is in favor ofCollagen type II higher expression. Those results cor-relates with the generation of hyaline-likeneocartilage.

To confirm these data we analysed the fluores-cence intensity profile and the relative percentage ofthe area that was positive for Col II on three fields ofview for each of the three patient derived cell lines

Figure 5.Cellular response to chondrogenic stimuli. (A)The graphs represent the fold changes calculatedwith the 2^ΔΔCΤmethodof Collagen type II (COL2A1) and type I (COL1A2), Aggrecan (ACAN) and Sox-9 (SOX9)markers. Fold changeswere normalized forRPL13a andGAPDH. Error bars represents standard deviation between three biological replicates and significativity was calculatedwith unpaired t-test. These analyses have been performed on lines derived from three different patients. The lower panels show theelectrophoretic analysis of the amplified cDNA to underline the switch inCOL2A1 andACANexpression during chondrogenesis. (B)Western blot analysis of SOX9 protein expression in hADASCs undergoing chondrogenesis inGelMa/HAMabioscaffolds.

10

Biofabrication 10 (2018) 045006 COnofrillo et al

(figure 6(B)). The results showed a significant differ-ence in the fluorescence intensity between the samplescollected at the start of the chondrogenesis and the dif-ferentiated samples at weeks 4 and 8 on an area of3 mm2 for each field analysed. Moreover, the fluor-escent area was 26%of the 3 mm2 area at week 4 with a

significant increase up to 40% at week 8 (figure 6(C)).The changes in cell morphology during differentiationrepresents an important feature to estimate the degreeof chondrogenesis [47]. As detected by actin stainingof cytoskeleton, hADSCs modified their morphologyover time from round shape to a more elongated

Figure 6.Neocartilage building and accumulation of Collagen type II. (A) Immunostaining analysis on 10 μmcryosections fromBiopen biofabricated samples at weeks 0, 4 and 8 of chondrogenesis, performed using antibodies against Collagen type II (Col II, ingreen), CollagenType I (Col I, in cyan) and counterstainedwith cellmarker nuclei DAPI (in grey) andTexas Red-X phalloidin Actin(in red) to detect cellmorphology. The superimposed signals are shown in the last columns (merge+Nuclei Actin). The images areorientatedwith the surface of the samples toward the top. (B)The panels showhighermagnification of sections shown inA.Note: theimages are themost representative selected from the immunohistology performed on the three hADSCs patient derived cell lines. (C)The graph shows the quantification analysis of Col IIfluorescence intensity (in panel A)normalized for the positive area (bars) and thepercentage of the positive area forCol II (area fill). The calculationswere performed on three different fields of view.Oneway AnovawithDunnett’s correctionwas applied to statistically analyse the differences in thefluorescence intensity, while unpaired t test wasused for thefluorescence area.

11

Biofabrication 10 (2018) 045006 COnofrillo et al

appearance (figure 6(B)). Collagen type II was stronglyconcentrated along the cell’s membranes and in theECM, while collagen type I remained confined in theintracellular area (figure 6(B)). Moreover, Collagentype X, a marker of hypertrophic differentiation,resulted to be expressed after 4 weeks of chondrogen-esis, also with an intracellular accumulation (supp.figure 4).

Conversely, in the control bioscaffolds, in absenceof chondrogenic stimuli, the expression of Collagentype I and II was not detectable (supp. figure 5).

In order to visualize the spatial distribution of cellsand ECMdeposition along the differentiation, we usedLight Sheet microscopy, which allows the reconstruc-tion of a 3D image, while maintaining the physiologi-cal integrity of the bioscaffold through opticalsectioning of the sample. We labelled the bioscaffoldswith Actin to detect cells and with Collagen type II todetect ECM generation. From our imaging analysis,we observed how hADSCs exerted the ability to con-dense in the GelMa/HAMa hydrogel scaffold duringthe chondrogenesis (figure 7). Cells are evenly dis-tributed along the entire depth of the printed scaffoldright after the printing (figures 7(A), (C)), but con-densed toward the superficial areas of the bioscaffoldsduring the chondrogenic differentiation (figures 7(B),(D)). In contrast, the condensation phenomenon wasnot visible without chondrogenic stimuli (supp. figure4). Collagen type II deposition is clearly visible at week4 in the area where cells are condensed. The stem cellcondensation is in fact required for the inception ofthe chondrogenic program during limb develop-ment [48, 49].

3.2.4. Collagen II fibrillary organization in GelMa/HAMabioscaffoldsVisualization of the 3D arrangement of the ECM andliving cells in native hydrated conditions is crucial toensure the formation of a hyaline-like tissue [49]. Inaddition, it is important to understand if the accumu-lation of Collagen type II is combined with theorganization of mature collagen fibrils that providesthe strength and the function to the neotissue.

Hence, the immunohistochemistry analysis wascomplemented with nonlinear microscopy, a non-invasive methodology that allows to simultaneouslyacquire SHG images of collagen and TPE auto-fluorescence images of living cells in label-free biofab-ricated scaffolds (figure 8(A)).

In the acellular scaffolds that were maintained inchondrogenic media for 8 weeks, TPE and SHG werenot detectable, indicating that the composition ofGelMa/HAMa did not interfere with the two signals’generation (figure 8(B)). In the chondrogenic inducedbioscaffolds, we were able to detect TPE signals fromcells embedded in the GelMa/HAMa and SHGderived image of collagen fibrils surrounding the cells(figure 8(D)). In contrast, in absence of chondrogenicstimuli, only TPE signal was detectable (figure 8(C).

The collagen-like fibrils detected by SHG are onlyattributable to the building of neocartilage from stemcells undergoing chondrogenesis. Importantly, giventhat the corresponding immunostaining performedon the same sections was able to detect only Collagentype II protein accumulation in the ECM, rather thanCollagen type I or Collagen type X (figure 6 and supp.figure 3), mature Collagen type II can be considered asamajor contributor of the SHG signal.

3.2.5. Mechanical evaluation of GelMa/HAMabioscaffolds during chondrogenesisIn order to characterize the functional development ofthe in vitro generated tissue, we used unconfinedcompression testing (figure 9) on the three hADSCscell lines in triplicate for each indicated time point.The unconfined compressive modulus measured innon chondrogenic bioscaffolds was 57.28 kPa(+/−26.8) and 29.9 kPa (+/−23.4) after 8 weeks ofchondrogenesis without any significative differencebetween the groups (figure 9(A)).

Moreover, the bioscaffolds exerted a compressivemodulus of 61.2 (+/−36.5) at week 0 (data notshown). Overall, we measured that the bioscaffoldsmaintain consistent mechanical properties along thechondrogenic differentiation. However, by analysingthe surface of our bioscaffolds at a microscale level byatomic force microscopy indentation, we found thatthe Young’s modulus of five different localized area ofthree different samples was 660 kPa (+/−1119) after 8weeks of chondrogenesis and only 191.7 kPa(+/−140.7) in their relative non chondrogenic con-trols. The high variability of the chondrogenic samplesreflects the accumulation of biological derived ECM,for which organization is based on collagen fibrils andcannot result in a flat homogeneous surface. The stiff-ness is on average 3.45 times higher than their controlcounterparts, which in contrast showed a much lowerstandard deviation (figure 9(B)).

Thus, the secretedmatrix causes a gain inmechan-ical strength that cannot be detected by unconfinedmechanical tests due to the non-homogeneous orga-nization of the neocartilage through the entire thick-ness of the sample.

4.Discussion

The results of this study demonstrate that the Biopencan be used to produce human hyaline-like cartilagein vitro by co-axial extrusion of hADSCs laden inGelMa/HAMahydrogel.

The necessity for the development of a handhelddevice is dictated by surgical needs: cartilage injurieshave to be cleaned and debrided before any regen-erative solution can be attempted surgically, thereforeimpeding the perfect matching of laboratory-pro-duced bioscaffolds.

12

Biofabrication 10 (2018) 045006 COnofrillo et al

In this study, we have used the handheld extrusiondevice (Biopen) in the same fashion as it would be usedsurgically, to simulate its clinical applicability.

The ideal cell type to be used for cartilage regenera-tion is still a matter of debate. A number of cell sourceshave been investigated, the most common beingmature articular chondrocytes, chondrocyte progeni-tors, and stem cells [50]. It is accepted that overallMSCs must possess a proven chondrogenic capacity,must not cause donor-site morbidity, and must bereadily expandable in culture without losing their phe-notype [51]. The number of stem cells-based clinicaltrials currently running worldwide and focused oncartilage repair is in fact increasing [52]. Recently, agroup of stem cells isolated from joint tissue, such aschondrogenic stem/progenitors from cartilage itself,synovial fluid, synovial membrane, and IPFP have

gained great attention due to their increased chondro-genic capacity over the bone marrow and sub-cutaneous adipose-derived stem cells [53]. Indeed, theIPFP, has gained much attention as a source ofhADSCswith high chondrogenic capacity both in vitroand in tissue regenerative medicine approaches [53].Human ADSCs also express CD44, which is the hya-luronic acid receptor involved in cell–cell and cell–matrix interactions, therefore playing an importantrole in the early stages of cartilage regeneration [38].By expressing CD44, these cells are more prone tobind the hyaluronic acid residues present in theGelMa/HAMa bioink favouring the chondrogenicprocess. Therefore, for our study we have selectedhADSCs harvested from the IPFP, as the source forhuman cartilage regeneration. Furthermore, we haveshown that stem cells isolated from the IPFP of

Figure 7.Neocartilage building and cells condensation during chondrogenesis. (A), (B) Immunostaining analysis on 10 μmcryosections fromBiopen biofabricated samples at weeks 0 and 4 of chondrogenesis, performed using Texas Red-X Phalloidin (in red)to detect cell condensation. (C), (D)Light sheetmicroscopy performed on the entire bioscaffolds (shown in grey, 1 mmheight) atweeks 0 and 4, after labellingwith Texas Red-X Phalloidin andCollagen type II antibody. The stainedActin is shown in red in thesuperimposed panel with the bioscaffold (left column), and in rainbow colour code to show the depth and the localization of cellsthroughout the entire depth of the bioscaffold (middle column). The raibow color code represents a range of depth that goes from0 mm (violet colour) to 1 mm (red colour). Collagen type II is shown in green (right column) to detect neocartilage formation. Theimages are orientatedwith the surface of the samples toward the top.

13

Biofabrication 10 (2018) 045006 COnofrillo et al

Figure 8.Neocartilage building and organization of collagen type IIfibrils. (A) Schematic cartoon showing the nonlinearmicroscopywhereby dual-photon excitation gives rise to two-photon excitation (TPE) and second harmonic generation (SHG). Under dual-photon excitation conditions, non-centrosymmetric structures (such as collagenmolecules) absorb and release energy in a differentway to subcellular proteins. The light emitted from the excited non-centrosymmetric structures return at double energy, but half thewavelength of the original excitation photons generating images of thefibrillary structures of collagen proteins. (B)Representativeimage of the acellular control scaffold. (C)Representative images of hADCSs laden inGelMa/HAMa after 8weeks in controlmedia.(D)Representative images of hADCSs laden inGelMa/HAMa after 8weeks of chondrogenic differentiation.Note: Samples have beenfixed and cryosectioned and directly imaged for nonlinearmicroscopy (seeMaterial andmethods section formore detailedinformation).

Figure 9.Mechanical properties evaluation. (A)The graph shows the unconfined compressivemodulus of control and chondrogenicsamples after 8weeks. Error bars represents standard deviation between three biological replicates and significativity was calculatedwith unpaired t-test. (B)The graph indicates Young’smodulus values obtained from analysing force curves using theHertzindentationmodel of control and chondrogenic samples after 8weeks. Error bars represents standard deviation between threebiological replicates and significativity was calculatedwith unpaired t-test. These analyses have been performed on three hADSCs,undergoing chondrogenic differentiation, derived from three different patients.

14

Biofabrication 10 (2018) 045006 COnofrillo et al

patients affected by mild/severe OA, retain theirchondrogenic potential, thus establishing the mile-stone for the treatment of cartilage injuries via using apersonalized medicine approach using autologousstem cells [54]. Despite the heterogeneity of the clinicalstatus of the patients, the data obtained showed con-sistent results among the different stem cells lines tes-ted, defining our stem cells source, the human fat pad,as a good candidate for the clinical application.

In our previous study we have demonstrated that acore/shell distribution of our scaffold reliably protectsthe survival and proliferation capacity of hADSCs [31].

The co-axial extrusion approach have been pro-posed by other groups, they aimed to use a viscoussacrificial shell to allow the deposition of a liquid cel-lular compartment that cannot otherwise hold a pre-determined shape, thus performing the hardeningprocedure before the removal of the sacrificial com-partment [55–57].

The advantage of our technology relies instead ona different application of core/shell geometry. Thesegregation of the photocrosslinkable shell away fromthe cellular compartment, which is not photo-crosslinked, is designed to protect the cells during thebiofabrication process without compromising struc-tural or cellular integrity. Our co-axial strategy geome-trically compartmentalizes a solid phase that facilitatesappropriate stiffness requirements, and a viscousliquid phase that preserves cell viability by separatingcells from PIs and the cytotoxic chemical by-productscoming from the crosslinking reaction going onwithin the solid phase. As shown in our previous study[31], the segregation of the core and shell compart-ments is well defined right after the biofabricationprocess.

Thus, we produced GelMa/HAMa bioscaffolds bymeans of Biopen mediated co-axial extrusion to assesswhether the fabricated environment is permissive forneocartilage generation in vitro.

Expression of hyaline-like cartilage componentswas assessed first by RT-qPCR. SOX9, the master reg-ulator of chondrogenesis, is expressed at transcrip-tional level but not at translational level before thechondrogenic induction, reflecting the chondrogenicpotential of the stem cells source. However, SOX9transcript accumulates over time under chondrogenicstimuli, reaching an expression plateau at week 4, witha consequent accumulation of the translated proteinthat continue to accumulate until week 8. This appar-ent discrepancy can be ascribed to the several posttranslational modification that regulates the stabilityof SOX9 in the cells. In fact, SOX9 is subjected to ubi-quitin mediated proteasomal degradation effected bythe E3 ubiquitin ligase FBW7 [58, 59] and by the E6ubiquitin ligase UBE3A [60]. Moreover, the activationof TGFβ pathway leads to the stabilization of SOX9 bythe phosphorylation of Serine 211 [61]. Thus, despitethe transcriptional expression of SOX9 reaches a pla-teau, post translational modifications, triggered by

chondrogenic stimuli, can positively affect the stabilityof the protein, leading to the activation of its transcrip-tional targets. In fact, COL2A1 and ACAN are notexpressed at the start of the chondrogenesis, but aredetectable at week 4 showing a 300 and 20 fold increaserespectively at week 8. COL1A2, an acceptedmarker offibrocartilage formation, is express in hADSCs popu-lations and shows an increase of only 10 times over theentire length of chondrogenic stimulation. This is con-sistent with other studies that analysed the chondro-genic potential of 3D bioprinted hADSCs [42]. Theproduction and accumulation of ECM containingCollagen type II, Proteoglycan and GAGs was demon-strated in a time-related manner both histologicallyand immunophenotypically. hADSCs secreted a sig-nificant amount of Collagen type II and Proteoglycanaround the cell bodies and in the scaffold matrix. Incontrast, no significant extracellular accumulation ofCollagen type I was observed over time in the analysedbioscaffold. It has been shown that the ratio betweenCollagen type II and Collagen type I is the dis-criminant to define hyaline-like ECM [45]. Afterchondrogenic stimulation, thementioned ratio is con-siderably higher in the GelMa/HAMa bioscaffoldsboth at transcriptional and translational level. Our set-tings are consistent with other studies showing Col IIaccumulation after 28 days of differentiation and per-formed usingGelMa based bioinks and animal derivedstem cells [62]. Moreover, the production of hyaline-like matrix we observed is highly efficient with respectto that generated from adult chondrocytes loaded inGelMa ink [63], underlining once again the superiorityof hADSCs as a source of cells for the production ofECM in scaffold-based cartilage regenerationapplications.

To further characterise our bioscaffolds, we haveshown collagen fibrils formation and distributionusing nonlinear two photon microscopy and SHG[64, 65]. This technique can detect only mature fibril-lary collagen. Our results showed that Collagen type Iand X are expressed at the intracellular level. In con-trast, Collagen II are secreted and accumulated in anorganized ECM after 8 weeks of chondrogenesis.Thus, is very likely thatmature Collagen II is themajorcontributor of the SHG signal.

The pattern of matrix deposition in bioscaffoldsresembles the early stages formation of the growthplate with an accumulation of collagen type II in thesurface of the developing tissue. It is in fact well estab-lished that cartilage formation begins with mesenchy-mal condensation leading to chondrogenicdifferentiation of mesenchymal cells. Then, a densematrix is produced, serving as the cartilage anlage, atemplate for the subsequent generation of both thearticular cartilage and the subchondral bone [48, 49].The formation of physiologic tissue by progenitor cellsultimately required lessons taken from native tissuemorphogenesis. Self-assembly has been proposed asan in vitro method for recapitulating mesenchymal

15

Biofabrication 10 (2018) 045006 COnofrillo et al

condensation that precedes chondrogenesis [66].However current tissue-engineered scaffolds forchondrogenesis pay little attention to this phenom-enon or focused more on the modification of thebioink by adding growth factors or decellularizedECM components to help the cellular condensation[67], while others use cells pre-clustered into pelletsprior to 3Dmanufacturing tomimic themesenchymalcondensation [68]. With our setting instead, wedemonstrated that hADSCs are able to condenseunder in vitro chondrogenic stimulation and success-fully produce hyaline-like bioscaffolds. Moreover, theperipheral distribution of matrix formation is promis-ing in the view of the entire osteochondral tissueregeneration. In the future, the possibility to deliver agradient of osteogenic and chondrogenic growth fac-tors within the hydrogel could promote the selectivetissue differentiation allowing the formation of boneand cartilage simulating the entire osteochondral unit[69]. Importantly, our culture technique was per-formed in a static fashion, therefore not stimulatingthe orientation of the collagen fibres in away to recapi-tulate hyaline-like cartilage structures. To better simu-late the physiological environment within the joint,the use of a dynamic culture systems will be necessaryto improve the orientation of the collagen fibres, asshown in previous studies [70]. Mechanical resistanceto stress is undoubtedly a key element in cartilage tis-sue engineering: just like native articular cartilage, theregenerated tissue needs to be able to sustain the loadsapplied on the joint and distribute them evenly in theunderlying subchondral bone. There is currently noconsensus on the expected results from mechanicaltesting following cartilage regeneration in hydrogels.While several studies have shown that after 4 or 8weeks in chondrogenic culture media, 3D printedsamples show increase in the compression modulus[71], other studies show decrease or stagnation ofmechanical properties with time in compression [72],and in tension [73]. In our study, by unconfined com-pression tests, we found no statistically significant dif-ference of the bulk mechanical properties betweenchondrogenic bioscaffolds and their control counter-parts. Importantly, several studies aiming at cartilagematrix growth in hydrogels demonstrated a compres-sive modulus after 8 weeks reaching about 50 kPa[71, 72, 74], which is the constant modulus of ourhydrogel. While this compressive modulus is achievedduring the chondrogenic process in vitro, with ourapproach it is possible to obtain a comparable stiffnessright after the biofabrication process. The mechanicalbehaviour of the bioscaffolds is determined by theinterplay between intrinsic degradation of the bioma-terial components, cell mediated degradation andECMdeposition [75]. However, by performing atomicforce microscopy indentation, we have demonstratedthat in correspondence of accumulated ECM there is amarked increase in Young’s modulus at 8 weeks ofchondrogenesis, compared to undifferentiated

samples. This analysis highlights the importance ofconsidering spatial variations in constructs. Theseresults reflect the fact that ECMaccumulation lead to again in mechanical resistance in localized regions thatcorrespond to the area of where cell condensationhappens. In our next studies, the phenotype andmigration of the cells, matrix deposition, proteolyticactivity and construct degradation rate will be eval-uated and correlated with de novo cartilage formationto better understand the interaction between hADSCsand crosslinkedGelMa/HAMahydrogel [76].

5. Conclusion

The clinical application of hydrogel based bioscaffoldshas the potential to effect the regeneration of articularcartilage. In this study, we have demonstrated thathuman IPFP adipose-derived stem cells can producehyaline-like cartilage in GelMa/HAMa bioscaffolds.The in vitro biofabrication of human neocartilage via ahandheld extrusion device is a key step in the pathtowards the development of tissue regeneration strate-gies in the clinical practice.

Acknowledgments

This work was supported by (1) Arthritis Australia—Zimmer Australia Grant, (2) Victorian OrthopaedicResearch Trust, (3) The John Loewenthal Foundationfor Surgery (Royal Australasian College of Surgeons),(4) The Australian Research Council Centre of Excel-lence Scheme (Project Number CE 140100012), and(5) StVincent’sHospital (Melbourne).

Research Endowment Found. Equipment GrantformMTPConnect is gratefully acknowledged. Accessto atomic force microscope facilities was providedthrough the Materials Characterisation and Fabrica-tion Platform at theUniversity ofMelbourne.

The authors would like to thank the AustralianNational Nanofabrication Facility—Materials nodefor equipment use; Dr Andre Tan for technical assis-tance in performing SHG microscopy. ConfocalMicroscopy and IMARIS software usage were per-formed at the Biological OpticalMicroscopy Platform,TheUniversity ofMelbourne

(www.microscopy.unimelb.edu.au).

Author contributions statement

CO, SD designed and performed the experiments,executed the data analyses, wrote the manuscript andprepared the figures.

CO’C design the analyses, performed the AFMtests, gave intellectual input throughout the project,and reviewed themanuscript draft.

16

Biofabrication 10 (2018) 045006 COnofrillo et al

RB performed and analysed the compression test-ing analysis as well as gave intellectual input through-out the project, and reviewing of themanuscript draft.

AJO provided intellectual input and edited themanuscript.

MS performed the image acquisition with lightsheetmicroscope.

GGWand PFMC conceived the idea of the surgicalhandheld device, have been involved in developmentof concept, edited and approvedmanuscript.

CDB designed the overall project, supervised theexperiments, analysed and interpreted the data, wrotethemanuscript and approved the final version.

All authors approved the finalmanuscript.

Author disclosure statement

We declare that this manuscript is original, has notbeen published before and is not currently beingconsidered for publication elsewhere. We wish toconfirm that there are no conflicts of interest asso-ciated with this publication and there have been nocompeting financial interests for this work that couldhave influenced its outcome.

We confirm that themanuscript has been read andapproved by all named authors and that there are noother persons who satisfied the criteria for authorshipbut are not listed. We further confirm that all of ushave approved the order of authors listed in themanuscript.

We confirm that we have given due considerationto the protection of intellectual property associatedwith this work and that are no impediments to pub-lication, including the timing of publication, withrespect to intellectual property. In so doing we con-firm that we have followed the regulations of our insti-tutions concerning intellectual property.

Ethical statement

St. Vincent Hospital Ethics Committee [HREC/16/SVHM/186] approved use of all human samples andprocedures (isolation of hADSCs from human infra-patellar fat pad) in this study and all the experimentswere performed in accordance with relevant guide-lines and regulations.

ORCID iDs

CarmineOnofrillo https://orcid.org/0000-0003-2537-6672

References

[1] Hollander A P,Dickinson SC andKafienahW2010 Stem cellsand cartilage development: complexities of a simple tissueStemCells 28 1992–6

[2] Muiznieks LD andKeeley FW2013Molecular assembly andmechanical properties of the extracellularmatrix: a fibrousprotein perspectiveBiochim. Biophys. Acta—Mol. Basis Dis.1832 866–75

[3] Mansour JM2009 Biomechanics of cartilageKinesiology: TheMechanics and Pathomechanics of HumanMovement edCAOatis 2nd edn (Philadelphia, PA:Wolters KluwerHealth)pp 66–79

[4] KonE, FilardoG,DiMartinoA andMarcacciM2012ACI andMACI J. Knee Surg. 25 017–22

[5] Makris EA, Gomoll AH,Malizos KN,Hu JC andAthanasiouKA2016Repair and tissue engineering techniquesfor articular cartilageNat. Rev. Rheumatol. 11 21–34

[6] Wood J J,MalekMA, Frassica F J, Polder J A,MohanAK,BloomEDAT, BraunMMandCoté TR2006Autologouscultured chondrocytes: adverse events reported to the unitedstates food and drug administration J. Bone Joint Surg. Am. 88503-7

[7] LinZ, Fitzgerald J B, Xu J,Willers C,WoodD,Grodzinsky A J andZhengMH2008Gene expression profilesof human chondrocytes during passagedmonolayercultivation J. Orthop. Res. 26 1230–7

[8] DewanAK,GibsonMA, Elisseeff JH andTriceME2014Evolution of autologous chondrocyte repair and comparisonto other cartilage repair techniquesBiomedRes. Int. 2014272481

[9] DengZ, Jin J, Zhao J andXuH2016Cartilage defecttreatments: with orwithout cells?Mesenchymal stem cells orchondrocytes? Traditional ormatrix-assisted? A systematicreview andmeta-analyses StemCells Int. 2016 9201492

[10] GoldringMB andGoldring SR 2010Articular cartilage andsubchondral bone in the pathogenesis of osteoarthritisAnn.NYAcad. Sci. 1192 230–7

[11] Omori G 2005 Epidemiology of knee osteoarthritisActaMed.Biol. 53 1–11

[12] Vos T et al 2012 Years livedwith disability (YLDs) for 1160sequelae of 289 diseases and injuries 1990–2010: a systematicanalysis for theGlobal Burden ofDisease Study 2010 Lancet380 2163–96

[13] OndrésikM et al 2017Management of knee osteoarthritis.Current status and future trendsBiotechnol. Bioeng. 114717–39

[14] BaugéC andBoumédiene K 2015Use of adult stem cells forcartilage tissue engineering: current status and futuredevelopments StemCells Int. 2015 438026

[15] Martin I, IrelandH, BaldomeroH and Passweg J 2015Thesurvey on cellular and engineered tissue therapies in Europe in2012Tissue Eng.A 21 1–13

[16] CornelissenD-J, Faulkner-Jones A and ShuW2017Currentdevelopments in 3Dbioprinting for tissue engineeringCurr.Opin. Biomed. Eng. 2 76–82

[17] Liu Y, ZhouG andCaoY2017Recent progress in cartilagetissue engineering—our experience and future directionsEngineering 3 28–35

[18] Daly AC, Freeman FE,Gonzalez-Fernandez T, Critchley S E,Nulty J andKellyD J 2017 3Dbioprinting for cartilage andosteochondral tissue engineeringAdv.Healthc.Mater. 6 1–20

[19] DiBella C, FosangA,Donati DM,WallaceGG andChoong P FM2015 3Dbioprinting of cartilage for orthopedicsurgeons: reading between the lines Front. Surg. 2 1–7

[20] Iwasa J, Engebretsen L, Shima Y andOchiM2009Clinicalapplication of scaffolds for cartilage tissue engineeringKneeSurg., Sports Traumatol. Arthrosc. 17 561–77

[21] KangH-W, Yoo J J andAtala A 2015Bioprinted Scaffolds forCartilage Tissue Engineering BT—Cartilage Tissue Engineering:Methods and Protocols ed PMDoran (NewYork: Springer)pp 161–9

[22] RaiV,DilisioMF,DietzNE andAgrawalDK 2017Recentstrategies in cartilage repair: a systemic review of the scaffolddevelopment and tissue engineering J. Biomed.Mater. Res.A105 2343–54

[23] Koh J L,Wirsing K, Lautenschlager E andZhang L-O 2004Theeffect of graft heightmismatch on contact pressure following

17

Biofabrication 10 (2018) 045006 COnofrillo et al

osteochondral grafting: a biomechanical studyAm. J. SportsMed. 32 317–20

[24] Mollon B, Kandel R, Chahal J andTheodoropoulos J 2013Theclinical status of cartilage tissue regeneration in humansOsteoarthr. Cartil. 21 1824–33

[25] Keriquel V, Guillemot F, Arnault I, Guillotin B,Miraux S,Amédée J, Fricain J-C andCatros S 2010 In vivo bioprinting forcomputer-and robotic-assistedmedical intervention:preliminary study inmiceBiofabrication 2 014101

[26] Ozbolat I T 2018 Bioprinting scale-up tissue and organconstructs for transplantationTrends Biotechnol. 33 395–400

[27] SchmutzerM andAszodi A 2017Cell compaction influencesthe regenerative potential of passaged bovine articularchondrocytes in an ex vivo cartilage defectmodel J. Biosci.Bioeng. 123 512–22

[28] Bartz C,MeixnerM,Giesemann P, Roël G, BulwinG-C andSmink J J 2016An ex vivo human cartilage repairmodel toevaluate the potency of a cartilage cell transplant J. Transl.Med.14 317

[29] O’Connell CD et al 2016Development of the Biopen: ahandheld device for surgical printing of adipose stem cells at achondral wound siteBiofabrication 8 015019

[30] DiBella C et al In situ handheld three-dimensional bioprintingfor cartilage regeneration J. Tissue Eng. Regen.Med. 12 611–21

[31] Duchi S et al 2017Handheld co-axial bioprinting: applicationto in situ surgical cartilage repair Sci. Rep. 7 1–12

[32] MoyadT F 2011Cartilage injuries in the adult knee: evaluationandmanagementCartilage 2 226–36

[33] YeK, FelimbanR, Traianedes K,Moulton S E,WallaceGG,Chung J, Quigley A, Choong P FMandMyersDE 2014Chondrogenesis of infrapatellar fat pad derived adipose stemcells in 3Dprinted chitosan scaffold PLoSOne 9 e102638

[34] Estes B T,Diekman BO,Gimble JM andGuilak F 2010Isolation of adipose-derived stem cells and their induction to achondrogenic phenotypeNat. Protocols 5 1294–311

[35] FelimbanR, YeK, Traianedes K,Di Bella C, Crook J,WallaceGG,Quigley A, Choong P FMandMyersDE 2014Differentiation of stem cells fromhuman infrapatellar fat pad:characterization of cells undergoing chondrogenesisTissueEng.A 20 1–11

[36] Lucarelli E et al 2014 In vitro biosafety profile evaluation ofmultipotentmesenchymal stem cells derived from the bonemarrowof sarcoma patients J. Transl.Med. 12 95

[37] LoessnerD,Meinert C, Kaemmerer E,Martine LC, YueK,Levett PA, Klein T J,Melchels F PW,Khademhosseini A andHutmacherDW2016 Functionalization, preparation and useof cell–laden gelatinmethacryloyl-based hydrogels asmodulartissue culture platformsNat. Protocols 11 727–46

[38] KnudsonCB 2003Hyaluronan andCD44: strategic players forcell–matrix interactions during chondrogenesis andmatrixassemblyBirthDefects Res.C 69 174–96

[39] Barry F andMurphyM2013Mesenchymal stem cells in jointdisease and repairNat. Rev. Rheumatol. 9 584–94

[40] DuD, SugitaN, Liu Z,Moriguchi Y,Nakata K,Myoui A andYoshikawaH2015Repairing osteochondral defects of criticalsize usingmultiple costal grafts: an experimental studyCartilage 6 241–51

[41] Katagiri H,Mendes L F and Luyten F P 2017Definition of acritical size osteochondral knee defect and its negative effect onthe surrounding articular cartilage in the ratOsteoarthr. Cartil.25 1531–40

[42] Squitieri L 2005Method for quantitative analysis ofglycosaminoglycans and type II collagen in chondrocyte-seededarticular cartilage scaffolds with varied cross-linking densityUndergraduate ThesisMassachusetts Institute of Technologyhttp://hdl.handle.net/1721.1/32924

[43] BiW,Deng JM, ZhangZ, Behringer RR anddeCrombrugghe B 1999 SOX9 is required for cartilageformationNat. Genet. 22 85

[44] BiW,HuangW,WhitworthD J, Deng JM, Zhang Z,Behringer RR and deCrombrugghe B 2001Haploinsufficiencyof SOX9 results in defective cartilage primordia and prematureskeletalmineralization Proc. Natl Acad. Sci. USA 98 6698–703

[45] Marlovits S,HombauerM, TruppeM,Vècsei V andSchlegelW2004Changes in the ratio of type-I and type-IIcollagen expression duringmonolayer culture of humanchondrocytesBone Joint J. 86-B 286–95

[46] Daly AC ,Chritcley S E, Rencsok EMandKelly D J 2016Acomparison of different bioinks for 3Dbioprinting offibrocartilage and hyaline cartilageBiofabrication 8 045002

[47] YourekG,HussaiMA andMao J J 2014Cytoskeletal changesofmesenchymal stem cells during differentiationASAIO J. 53219–28

[48] ShimizuH, Yokoyama S andAsaharaH 2007Growth anddifferentiation of the developing limb bud from theperspective of chondrogenesisDev. GrowthDiffer. 49 449–54

[49] Olsen BR, ReginatoAMandWangW2000BonedevelopmentAnnu. Rev. Cell Dev. Biol. 16 191–220

[50] WangM et al 2017Advances and prospects in stem cells forcartilage regeneration StemCells Int. 2017 4130607

[51] Savkovic V, LiH, Seon J K,HackerM, Franz S and Simon JC2014Mesenchymal stem cells in cartilage regenerationCurr.StemCell Res. Ther. 9 469–88

[52] LeeWY andWangB 2017Cartilage repair bymesenchymalstem cells: clinical trial update and perspectives J. Orthop.Transl. 9 76–88

[53] doAmaral R J FC, AlmeidaHV,KellyD J,O’Brien F J andKearneyC J 2017 Infrapatellar fat pad stem cells: fromdevelopmental biology to cell therapy StemCells Int. 20176843727

[54] Sawitzke AD2013 Personalizedmedicine for osteoarthritis:where arewe now?Ther. Adv.Musculoskelet. Dis. 5 67–75

[55] GaoQ,HeY, Fu J, Liu A andMaL 2015Coaxial nozzle-assisted3Dbioprintingwith built-inmicrochannels for nutrientsdeliveryBiomaterials 61 203–15

[56] Colosi C, Shin SR,ManoharanV,Massa S, CostantiniM,Barbetta A,DokmeciMR,DentiniM andKhademhosseini A2016Microfluidic bioprinting of heterogeneous 3D tissueconstructs using low-viscosity bioinkAdv.Mater. 28 677–84

[57] CostantiniM, Idaszek J, SzökeK, Jaroszewicz J, DentiniM,Barbetta A, Brinchmann J E andŚwięszkowskiW2016 3Dbioprinting of BM-MSCs-loaded ECMbiomimetic hydrogelsfor in vitroneocartilage formationBiofabrication 8 035002

[58] SuryoRahmantoA et al 2016 FBW7 suppression leads to SOX9stabilization and increasedmalignancy inmedulloblastomaEMBO J. 35 2192–212

[59] HongX et al 2016 SOX9 is targeted for proteasomaldegradation by the E3 ligase FBW7 in response toDNAdamageNucleic Acids Res. 44 8855–69

[60] Hattori T, KishinoT, Stephen S, EberspaecherH,Maki S,TakigawaM, deCrombrugghe B andYasudaH 2013 E6-AP/UBE3Aprotein acts as a ubiquitin ligase toward SOX9 proteinJ. Biol. Chem. 288 35138–48

[61] CoricorG and Serra R 2016TGF-β regulates phosphorylationand stabilization of SOX9protein in chondrocytes throughp38 and Smad dependentmechanisms Sci. Rep. 6 38616

[62] Levato R,WebbWR,Otto I A,Mensinga A, Zhang Y,vanRijenM, vanWeerenR, Khan IMandMalda J 2017Thebio in the ink: cartilage regenerationwith bioprintablehydrogels and articular cartilage-derived progenitor cellsActaBiomater. 61 41–53

[63] SchuurmanW, Levett PA, PotMW, vanWeeren PR,DhertW JA,HutmacherDW,Melchels F PW,Klein T J andMalda J 2013Gelatin-methacrylamide hydrogels as potentialbiomaterials for fabrication of tissue-engineered cartilageconstructsMacromol. Biosci. 13 551–61

[64] ZipfelWR,WilliamsRMandWebbWW2003Nonlinearmagic:multiphotonmicroscopy in the biosciencesNat.Biotechnol. 21 1369

[65] ZoumiA, LuX, KassabG S andTromberg B J 2004 Imagingcoronary arterymicrostructure using second-harmonic andtwo-photon fluorescencemicroscopyBiophys. J. 87 2778–86

[66] PacificiM, Koyama E, ShibukawaY,WuC, Tamamura Y,Enomoto-IwamotoMand IwamotoM2006Cellular andmolecularmechanisms of synovial joint and articular cartilageformationAnn.NYAcad. Sci. 1068 74–86

18

Biofabrication 10 (2018) 045006 COnofrillo et al

[67] Kim IG,Ko J, LeeHR,Do SHand ParkK2016Mesenchymalcells condensation-induciblemesh scaffolds for cartilage tissueengineeringBiomaterials 85 18–29

[68] Suzuki S,Muneta T, Tsuji K, Ichinose S,MakinoH,UmezawaA and Sekiya I 2012 Properties and usefulness ofaggregates of synovialmesenchymal stem cells as a source forcartilage regenerationArthritis Res. Ther. 14R136

[69] Censi R,Matricardi AD andPietro 2015 Bioactive hydrogelscaffolds—advances in cartilage regeneration throughcontrolled drug deliveryCurr. Pharm.Des. 21 1545–55

[70] El-Ayoubi R,DeGrandpré C,DiRaddoR, YousefiA-MandLavigne P 2009Design and dynamic culture of 3D-scaffolds forcartilage tissue engineering J. Biomater. Appl. 25 429–44

[71] ChungC, BeechamM,MauckR L andBurdick J A 2009Theinfluence of degradation characteristics of hyaluronic acidhydrogels on in vitroneocartilage formation bymesenchymalstem cellsBiomaterials 30 4287–96

[72] Roberts J J, NicodemusGD,Greenwald EC andBryant S J2011Degradation improves tissue formation in (un)loadedchondrocyte-laden hydrogelsClin. Orthop. Relat. Res. 4692725–34

[73] DuanB,Hockaday LA, KangKHandButcher J T 2013 3DBioprinting of heterogeneous aortic valve conduits withalginate/gelatin hydrogels J. Biomed.Mater. Res.A 101A1255–64

[74] Marrella A, LagazzoA, Barberis F, Catelani T, QuartoR andScaglione S 2017 Enhancedmechanical performances andbioactivity of cell laden-graphene oxide/alginate hydrogelsopen new scenario for articular tissue engineering applicationsCarbon 115 608–16

[75] KimK, YoonDM,Mikos AG andKasper FK 2012Harnessingcell–biomaterial interactions for osteochondral tissueregeneration BTTissue Engineering III: Cell–SurfaceInteractions for Tissue Culture edCKasper et al (Berlin:Springer) pp 67–104

[76] SaremM,AryaN,HeizmannM,Neffe AT, BarberoA,Gebauer T P,Martin I, Lendlein A and Shastri V P 2018Interplay between stiffness and degradation of architecturedgelatin hydrogels leads to differentialmodulation ofchondrogenesis in vitro and in vivo Acta Biomater. 69 83–94

19

Biofabrication 10 (2018) 045006 COnofrillo et al