Review
Mimicking
the
Articular
Joint
with
In
Vitro
Models
Susanna
Piluso
,
1,2,3Yang
Li
,
1,3Florencia
Abinzano,
1,3Riccardo
Levato
,
1,3Liliana
Moreira
Teixeira
,
2,3,4Marcel
Karperien
,
2Jeroen
Leijten
,
2René
van
Weeren
,
3,4and
Jos
Malda
,
1,3,4,*
Treatingjointdiseasesremainsasignificantclinicalchallenge.Conventionalin
vitro cultures and animal models have been helpful, but suffer from limited
predictive power for the human response. Advanced models are therefore
requiredtomimicthecomplexbiologicalinteractionswithinthehumanjoint.
However,theintricatestructureofthejointmicroenvironmentandthecomplex
natureofjointdiseaseshavechallengedthedevelopmentofinvitromodelsthat
canfaithfullymimictheinvivophysiologicalandpathologicalenvironments.In
thisreview,wediscussthecurrentinvitromodelsofthejointandtheprogress
achievedinthedevelopmentofnovelandpotentiallymorepredictivemodels,
and highlight theapplication ofnew technologiesto accurately emulate the
articularjoint.
TheSynovial Joint:AComplexOrgan
The proper functioning of the joint depends on the maintenance of joint homeostasis, a dynamicequilibrium between anabolic andcatabolic processeswithin all the components ofthejoint[1,2].Thejointisacomplexmultitissueorganencompassingthearticularcartilage, thesubchondralbone,thesynovialmembrane,and,insomejoints,additionalintra-articular structures,suchasligamentsandmenisci(Box1).Thesynoviumisessentialforjoint homeo-stasis;infact,synovialmacrophagesareresponsibleforthemaintenanceofafinebalance betweenproinflammatoryandanti-inflammatorycytokines(seeGlossary)inthesynovialfluid [3].Synovialinflammation is nowrecognized to play a key role inthe progression of joint diseases,withthereleaseofinflammatorycytokinesbeingmediatedbythecrosstalkbetween synoviumandcartilage[4].Further,alterationsinthecompositionandstructureofthe sub-chondralbonecanaffectthebehavioroftheoverlyingcartilage,suggestingtheexistenceofa physical and molecular crosstalk between the two tissues [5,6]. The intricate interaction betweenthesedifferenttissuestructuresandcelltypesmakesitquitechallengingto recapitu-latebothhealthyandpathological[e.g.,inthecase ofosteoarthritis(OA)orrheumatoid arthritis(RA)] jointphysiology inamodel. Clearly,conventional2Dinvitro static cultures cannotaccuratelyrecreatethislevelofcomplexity.
Therefore,tounravel theintricatemechanismsinvolvedinjointhomeostasisanddisease,a widerangeofinvivomodelshavebeenused[7,8].Examplesincludemodelsinsmallanimals(i. e.,inthemouse,rat,rabbit,orguineapig)thatareoftenusedforinitialdrugscreening,asthese aregenerallycheaperandeasiertohandlethan thelargeanimalmodels.Incontrast,large animals(i.e.,inthedog,goat,sheep,pig,orhorse)showmoresimilaritiestohumansintermsof joint anatomy and cartilage morphology, but are more expensive and require specialized facilities[9,10].Nevertheless,animalmodels allowforthe studyofdiseasesinthe naturally
Highlights
Whiledifferentinvitroandinvivo
mod-elsofjointdisordershavebeen
devel-oped,therearenoeffectivetoolsfor
theevaluationofnewtherapiesforjoint
diseases,suchasosteoarthritis(OA).
Recent advances in (bio)fabrication
technologies enable the generation
ofinvitromodelsthancanfurther
reca-pitulatearticular physiology with the
potentialofreplacinganimalmodels.
Thevalidation oftheseadvancedin
vitromodelsiscrucialtoexploittheir
translationalpotential.
1
DepartmentofOrthopaedics,
UniversityMedicalCenterUtrecht,
UtrechtUniversity,Utrecht,The
Netherlands 2
DepartmentofDevelopmental
BioEngineering,TechnicalMedical
Centre,UniversityofTwente,
Enschede,TheNetherlands
3
RegenerativeMedicineUtrecht,
UtrechtUniversity,Utrecht,The
Netherlands 4
DepartmentofEquineSciences,
FacultyofVeterinaryMedicine,
UtrechtUniversity,Utrecht,The
Netherlands
*Correspondence:
j.malda@umcutrecht.nl(J.Malda).
occurringenvironmentofthewholejoint.However,duetospecies-specificdifferences[10,11], manytherapeutictreatmentsfailwhentranslatinganimalstudiestohumanclinicaltrials[12]. Furthermore,ethicalconcernsandthesocietalambitiontoreduceanimalexperimentationhave driventhedevelopmentofadvancedinvitromodelsthatcanmoreaccuratelyrepresentstages ofhumandisease.
Recentadvancesinengineeringandbiologyhaveresultedinthedevelopmentoffunctional microscaleunits ofhuman organsthatare ableto recapitulate humandiseases[13,14].A uniquefeatureofthesesystemsistheirabilitytorecreatethecomplextissue microenviron-mentsandfacilitatecommunicationbetweendifferenttissues,accuratelymimickingtheinvivo situation[15].Invivo,cellsresideinahighlysophisticated3Dmicroenvironmentthatprovides biochemicalandbiomechanicalcuesguidingtheirbehavior,includingmigration,proliferation, anddifferentiation[16].Alterationsofaspecificextracellularmatrix(ECM)componentcan alreadygreatlyimpactthebiochemical–biomechanicalbalance,disruptingtissuehomeostasis andfunction[17,18].
Glossary
3Dbioprinting:theautomated
processofpatterningand
assemblinglivingandnon-living
materialswithaspatiallycontrolled
organization,toproduce
bioengineeredstructuresfor
applicationinregenerativemedicine.
Chondrocytes:cartilagemature
residentcells.Theyproduceand
maintaintheextracellularmatrixthat
keepscartilagehealthyand
functional.
Cytokinesandgrowthfactors:
signalingmoleculesreleasedbycells,
withtheobjectiveofaffectingthe
behaviorofothercells,including
growth,proliferation,migration,
inflammatorystate,and
differentiation.
Explant:extractedpiecesofnative
tissuesororgansthatcanbe
culturedinthelaboratory.
Extracellularmatrix(ECM):dense
networkofmacromolecules,suchas
proteinsandpolysaccharides,that
providesstructuralandbiochemical
supporttothesurroundingcells.
Hydrogel:hydrophilicpolymer
networks,whichabsorbandretaina
largeamountofwater.
Mesenchymalstromalcells
(MSCs):multipotentadultcellsthat
candifferentiateintobone,cartilage,
adipose,andmuscletissue.They
canbeobtainedfrombonemarrow,
adiposetissue,andothersources.
Microfluidics:amultidisciplinary
fieldbetweenchemistry,physics,
engineering,andmicro/
nanotechnologywhichdealswiththe
precisemanipulationoffluidsat
submillimeterscale.
Organ-on-a-chip:amultichannel
microfluidicdevicewithintegrated
microscalecellularco-culturesthat
offersphysiologicalbiochemicaland/
orbiophysicalstimulationtomore
realisticallyrecapitulatethenative
microenvironment.
Osteoarthritis(OA):degenerative
jointdiseaseassociatedwith
cartilagedestruction,inflammationof
thesynovialmembrane,and
subchondralboneremodeling.
Rheumatoidarthritis(RA):a
systemicinflammatorydiseasethat
involvessynovialcellproliferationand
structuraldamagetocartilage,bone,
andligaments.
Box1.TissuesoftheJoint
Articularcartilageisahighlyspecializedconnectivetissuethatcoverstheendsofourlongbones.Itisanavascular,
alymphatic,andaneuraltissuethatconsistsofadenseECMwithalowdensityofcells(about1–2%)[5].Subchondral
boneisthelayerofbonebeneaththecartilage.Togetherwiththecartilageitformsabiocomposite,knownasthe
osteochondralunit,which isspecializedtotransfer loadduringweight-bearingandjoint motion.Thesynovium
constitutestheenvelopeofthearticularjointsand,byactingasaninterfacewiththesystemicbloodcirculation,
ensuresnutrientsupplytothearticularcartilageviathesynovialfluid.Moreover,itsupplementsthesynovialfluidwiththe
keymoleculesnecessaryforjointfunction(i.e.,lubricin,hyaluronan,andimmunomodulatorycytokines)[5].Thesynovial
membraneconsistsoftwocelltypes,synovialfibroblasts,whichconstituteupto75%ofallcellsinahealthysynovium,
andmacrophages.Themacrophagescanbeclassifiedintoclassicallyactivated(M1)andalternativelyactivated(M2).
Thelattersubtypeisinvolvedintheproductionofimmunoregulatoryfactors[e.g.,IL-10andchemokineligand
(CCL)-18],whereasM1macrophagesproduceproinflammatorymediatorssuchasTNF-a,IL-1b,andIL-6,whichplayakey
roleinsynovialinflammation[88].Thenumberofmacrophagesincreasesdrasticallyduringinflammationand,together
withsynovialfibroblasts,secreteproinflammatorycytokinesandECM-degradingenzymes.Theproductionofcytokines
attractinflammatorycells(Tcells,macrophages,monocytes)intothesynovium,resultingintheformationofamassof
inflamedtissueorpannus[3].
Additionalcomponentsofahealthykneejointarethemenisci(i.e.,wedge-shapedtissuesthatperformcomplex
functionsinload-bearing,loadtransmission,stabilizationofthejoint,shockabsorptionduringmovements,nutritionof
articularcartilage,andlubrication[89]).Lubricin,asuperficialzoneprotein,whichhasakeyroleinthemaintenanceof
jointintegrity,ishighlyexpressedinhealthykneemenisci.Theexpressionoflubricinisdownregulatedintheknee
menisciandsynovialfluidofOApatients,leadingtoincreasedfrictioninthejointandcartilagedegeneration[80].Further,
meniscuscellsincreasedtheirproductionofmatrix-degradingenzymes,cytokines,andchemokinesinresponseto
stimulationwithproinflammatoryfactors(IL-1b,IL-6,orfibronectinfragments).Thissuggeststhattheroleofthe
meniscusinOAgoesbeyondthemechanicalaspectandmightbeduetobiologicalinteraction[90].
Tendonsandligamentsaresoftconnectivetissuescomposedofcloselypacked,highlyalignedcollagenfiberbundles
thatjoinbonetomuscleandbonetobone,respectively.Thesestructurestransfertensileloadstoguidemotionand
stabilizethediarthrodialjoint[91].Tendonorligamentfailuremayresultinjointdestabilizationandhenceleadtodamage
byalteringthebiomechanicalbalancebetweenneighboringtissues(e.g.,meniscusandarticularcartilage)inthejoint
[92].Further,afteranteriorcruciateligamentinjuries,thegeneexpressionofkeydegradativeenzymes(MMPs)was
upregulatedintheligamentsandthesynovium,suggestingacloseinteractionbetweenthesetissuesinresponseto
injuriesandakeyroleincartilagedegradation[93].
ThekneejointalsocontainsHoffa’sfatpadorinfrapatellarfatpad(IFP),asofttissueinterposedbetweenthejoint
capsuleandthesynovium.TheIFPisahighlyinnervatedtissueand,therefore,acommonsourceofkneepain.Although
itsphysiologicalfunctioninthekneeremainsstillelusive,earlystudiessuggestthatinflammationofthisadiposetissue
Ideally,invitro models ofthe joint mustrecapitulate the intricate microenvironment ofthe synovialcavityandcapturetheinteractionsbetweenthevariousjointelements(Figure1,Key Figure).Thetissuesthatthejointiscomposedofareextremelysensitivetotheirmechanical environment,andloadingplaysakeyroleinthemaintenanceofjointhomeostasis[19].While moderate mechanicalloading contributes to the maintenance of tissueintegrity, reduced loadingoroverloadingcantriggerpathologicalchangesinjointtissues[20].Therefore,invitro modelsofthejointshouldideallyallowfortheinclusionofmechanicalstimulationand repro-ducetheinteractionsbetweenthedifferenttissues.
Here, we reviewthe existingin vitro models and we propose newavenues for the future developmentofmoresophisticatedmodelsofthearticulatingjoint.
KeyFigure
Schematic
of
the
Synovial
Joint
Femur Femur
Tendon Tendon
Synovial membrane Synovial membrane
Hoffa’s fat pad Hoffa’s fat pad
Tibia Tibia
Ar cular car lage Ar cular car lage
Meniscus Meniscus
Healthy
joint
Damaged
+ diseased
joint
Joint-on-chip Ligament-on-chip Meniscus-on-chip Tendon-on-chip Synovial membrane-on-chip Osteochondralunit-on-chip Hoffa’s fat pad-on-chip
Figure1.Thesynovialjointconsistsofseveraltissuesthatworkcloselytogethertorealizeandmaintainjointfunctionandhomeostasis.Thesetissuescomprise
articularcartilage,subchondralbone,synovium,menisci,tendon,ligamentandfatpad,whichallarecharacterizedbyaspecificstructureandwhichareexposedto
differentbiochemicalandbiomechanicalstimulations.Impairmentoffunctionofonetissueleadstoalteredbehavioroftheothertissues,whichcanleadtojoint
degeneration.Theidealinvitromodelmustrecapitulatethiscomplexenvironment;themostsuitableapproachtorealizesuchamodelisthedevelopmentofseveraljoint
TheState oftheArt:InVitroModelsofTissuesintheJoint
2DMonolayerCultureModels
Todate,limitedinvitromodelsexisttostudytheinteractionsbetweenthedifferenttissuesofthe joint.Withinthemajorityofcurrentinvitromodels,thesimulationofinteractionsisoftenlimited totheintroductionofbiochemicalcuesortoco-cultureofcellsortissues[21].Forexample, interactionswiththesynoviumareoftensimulatedbysupplementingtheculturemediawith cytokines[22],orbyadditionofsynovialfluid[23].Alternatively,fibroblast-likesynoviocytescan beincludedinthemodel[24].Recently,aninvitromodelusingsynovialcellswasdevelopedto studytheresponseofthesynoviumtocartilagewearparticles[25].Themodelconsistedofa densecellsheetoffibroblast-likesynoviocytesinculturewithcartilageparticles.Theco-culture resulted inan increased production ofcytokines and thickeningof the cell sheet through increasedcollagencontent.Interestingly,theseresultscorrelatewithdatafrompreviousanimal studiesshowingthatcartilagedebriscaninducesynovitis[25].
Choosing a cell source that is representative can, however,be a challenge, withprimary chondrocytesand chondrogenicallydifferentiated mesenchymalstromal cells(MSCs) beingthemostcommonlychosencellsforthissimplifiedviewofthejoint[21].Additionally,cells culturedon2Dsurfacesarepronetochangetheirphenotypecomparedwiththeirnativeinvivo milieu.Tounderstandthepathwaysleadingtode-differentiation,changesingeneexpression wereinvestigatedinmonolayerculturesofhumanchondrocytes[26].
Notwithstandingtheirrelative simplicity,thesemodels doprovidevaluableinsightsthatcan enhanceourunderstandingoftheeventsinvolvedinjointdiseases. However,thesesimple models provide information only on isolatedevents within a specific tissue and allow for changingonlyonefactoratatime(cytokines,growthfactors,osmoticpressure,etc.).
Biomaterial-Based3DCultureModels
Themost simplistic 3D culture model is basedon aggregation of cells into spheres [27]. Previousstudieshave,indeed,shownthatchondrospheresoutperformsinglecellsintermsof cartilagematrixproduction[28],allowingthestudyofcell-to-cellandcell-to-matrixinteractions. Thisprovidesapowerfulyetsimpletooltostudycartilageformationandtopreliminarilyscreen drugsfor OA[29].Combiningmicroaggregateswithhydrogels furtherexploits thenatural environmentalcuesto promote growth factor-freechondrogenic differentiation [30].These constructscanalsobesubjectedtomechanicalloadstoinvestigatethecellresponsetothese typesofstimuli[31,32].Mechanicalloadscaninducechangesinthepericellularandterritorial matrixofchondrocytes[33].Further,cellularresponsetomechanicalstimulidependsonthe materialwithinwhichcellsareencapsulated[34].Forexample,hydrogelsprovide3Dmatrices withpropertiessimilartonaturalECMs,suchashighwatercontent,porosity,and biocompati-bility[35].TocapturethebiologicalfeaturesofECM,hydrogelscanbemodifiedto provide bioactivecues,suchasthearginine-glycine-asparticacid(RGD)adhesivemotiforthepeptide bindingmotifofN-cadherinthatmimicscell–cellinteractions,whichmightenhance chondro-genesisofMSCs[36,37].Further,thedynamicremodelingofECMthroughcell-responsive enzymaticdegradation canbereplicatedbycrosslinkinghydrogels withdifunctionalmatrix metalloproteinase (MMP)-degradable peptides [38]. The inclusion of protease-degradable crosslinksinhyaluronicacidhydrogelssignificantlyaffectedthemorphologyofencapsulated MSCs,whichdisplayedanelongatedshape.Meanwhile,cellsremainedroundedinhydrogels thatinhibitedcellularremodeling[38].
Additionally, hydrogel mechanical properties can be reinforced and tailored to match the stiffnessofthetargettissue, forexample,bycombining thehydrogelwithpolymericfibers
[39].Thefiberscouldalsomimictissuetopography,whichisknowntoinfluencecellbehavior. Forexample,compositesofelectrospunfiberswithbioglass-derivedfoamsorbiphasic scaf-foldshavebeenusedtorepairosteochondraldefects[40].Recently,anengineeredscaffold consistingofhydrogelsandelectrospunfiberswasdevelopedtosimulatethemechanicaland topographical characteristics of native tendon tissue [41]. Additionally, hydrogels can be fabricatedto mimic structuralfeatures oftissueinterfaces,suchas thetendon/ligament-to boneinterface,byusingbiphasicscaffoldsthatmimicthealignmentofcollagenmolecules[42]. Further,thecombinationofmicrofluidicmixingtechnologieswithhydrogelfunctionalization approachesenablethepreparationofhydrogelswithacontrolledspatialgradientofcellsand biomoleculesignals[43].
Table1.RepresentativeExamplesofInVitroModelsofJointTissues
Tissue Model Findings Refs
Fibroblast-likesynoviocytes
2Dmonolayer
culture
Treatmentoffibroblast-likesynoviocytesresultedincelldeathandproduction
ofproinflammatorymediators(IL–1b,IL-6,andTNF-a)
[24]
2Dmonolayer
culture
Culturingfibroblast-likesynoviocyteswithcartilagewearparticlesresultedinan
increaseinproliferationandECMcontent,similartothethickeningofsynovial
liningobservedinOApatients
[25]
Periostealstemcells Biomaterial-based3Dculture Encapsulationofmacroaggregatedperiostealstemcellsintobiomaterials
resultedinimprovedinvivocartilagetissueformation,comparedwithsingle
cell-ladenhydrogels
[30]
Cartilage Tissueexplants
CollagenasewasusedoncartilageexplantstoimitateearlyOA;mechanical
loadingaffectedthedegradedcartilagemorethanhealthyexplants
[44]
AgeingaffectedTGFandbonemorphogeneticprotein(BMP)signaling
pathways,whichmightcontributetothedevelopmentofOA
[82]
Combininginsulin-likegrowthfactor-1(IGF-1)anddexamethasone(Dex)
preventedmatrixlossinaninflammatoryenvironment
[45]
Bone Tissueexplants Trabecularboneexplantsweresubmittedtoinducedshearstresstoanalyze
theeffectsonciliaexpressionofbonemarrowcells
[83]
Meniscus Tissueexplants
Cytokinesinhibitedmeniscalrepairofexplantsinvitro [84]
InducingoverexpressionofTGF-bviarecombinantadeno-associatedvirus
(rAAV)-mediatedgenetransferstimulatedinvitrohealingofmeniscusexplants
[85]
Tendon Tissueexplants Cyclicloadinginducedexpressionofinflammatorymarkersontendon
fascicles
[86]
Synovium Tissueexplants Synovialexplantsfromrheumatoidarthritispatientswereusedtoevaluatethe
effectsofbiologicdiseasemodifyingantirheumaticdrugtreatmentinvitro;
theseresultswerecorrelatedtoclinicalperformanceofthedrugs
[87]
Cartilageandsynovium
Co-culture Addingosteoarthritissynoviumtocartilageexplantsinhibited
glycosaminoglycanproduction,andthiseffectwascounteractedbythe
additionoftriamcinolone
[55]
Osteochondralplug Anosteochondralplugmodelwithindependentcompartmentsforcartilage
andbonewasusedtoevaluatetheeffectofcelldistributionwithinahydrogel
forcartilagerepairapplications
[51–53]
Osteochondralmicrotissue Multichamber
bioreactor
TheeffectsofIL-1bwereevaluatedonanosteochondralmicrotissuemodel,
whereMSCsseededwithinahydrogelweremaintainedusingseparate
chondralandosseouscompartmentsinwhichthecellsdifferentiatedinto
cartilageandbone-liketissues
[50]
Osteochondralunitandsynoviallining Multichamberbioreactor ThesynovialliningwasreproducedbyincorporatingMSCsseededwithina
polyethyleneglycolhydrogelontheosteochondralinterface,consistingofa
collagenhydrogel
TissueCultureModels
Tissueexplants,however,keepthecellsintheirnaturalmicroenvironment(Table1).Cartilage explantshaveprovidedvaluableinformationonthewholetissueresponsetoseveralstimuli, recreatingbothphysiologicalandpathologicalenvironmentsindiversejointtissues.Forexample, toemulatethedamageobservedinearlystagesofOA,cartilageexplantswerefirsttreatedwith collagenaseandthen subjectedto repetitive mechanicalstress. This modelprovidednovel informationonthemechanismsassociatedwithcartilagedegeneration(e.g.,mechanicalloading) [44].In anotherstudy to elucidatethe mechanismof actionofdexamethasone, a potential therapeuticdrug,cartilageexplantsweretreatedwith inflammatorycytokinesandsubjected tomechanicalinjurytosimulateearlystagesofOA[45].Amajordrawbackisthatonlyalimited numberofexplantscanbeobtainedfromeachdonorandthereisahighintradonorvariability betweensamples,dependingonthelocationofextractionfromthejoint[46].Thisvariationiseven greaterbetweendonors,presentingachallengeforreproducibility.Explantscommonlyfeature celldeathontheedgeswherethetissuewascut[21].Long-termstudiesarealsocomplicated,as thepropertiesofcartilageandboneexplantschangeovertimeintermsofmechanicalproperties andECMcomposition[47].Moreover,althoughitisrelativelyeasytoobtainexplantsfromanimal sources,availabilityofhealthyhumandonortissueislimited.
Bioreactors
Bioreactorsaredevicesabletoculturecellsunderacontrolledenvironment(e.g.,temperature, pH,nutrientsupply,mechanicalstimuli).Differenttypesofbioreactorshavebeendeveloped basedontissueandapplication,suchasspinnerflasks,rotatingwallvessels,andperfusion bioreactors[48,49].Inarecentwork,abioreactorconsistingoftwoseparatecompartments hasbeenused to recreate anengineered osteochondralunit.Thetwo microenvironments couldbeindividuallycontrolled,thusallowingcontrolofconditionsintheboneandcartilage part,followingexposuretoproinflammatorycytokines[50].Further,thedesignofthis bioreac-torwasfurtheradvancedtoenablecontinuousopticalmonitoringduringculture[48]. Preservingtheintegrity ofanosteochondralunitduring invitro culturehasbeenshown to extendthelifeandqualityofexplants[51].Previousworkonosteochondralexplants demon-stratedthatbioreactorplatforms(six-wellplateformat)providingdistinctmediatothecartilage andboneregionsoftheimplantallowforlong-termcultureoftheosteochondralplugs,while maintainingcartilagetissuecontent,structure,andmechanicalproperties[52].Thisplatform wasalsousedtoinvestigatetheeffectofthespatialchondrocytedistributionincartilagerepair mechanisms[53].Despitethefactthatthisosteochondralplug-basedmodelprovidednew insightintotheinteractionbetweentissues,itisstilllimitedtotheinterplaybetweenboneand cartilageandhasnotbeenyetadaptedtorepresentinflammatoryjointconditions.
Co-CultureofTissueExplants
Theabove-describedinvitromodelsfocusonindividualtissuesofthejointortheosteochondral unit,withoutincludinginteractionswithotherelementsofthejoint.Thisaspectwashighlighted inarecentstudydescribingco-culturesofbovinecartilageexplantswithexplantsoffibrousjoint capsuleandsynovium[54].Specifically,theco-incubationofmechanicallyinjuredcartilagewith fibrousjointcapsuleandsynoviumresultedinincreasedproteolyticdegradationofaggrecanby both MMPs and aggrecanase.This enhanced proteolyticactivity was notobserved when cartilagewasculturedalone[54].Similarly,theco-incubationofexplantsofhumanarticular cartilagewithhumansynoviumresultedintheproductionofMMPsandinflammatorycytokines thatwerenotdetectedinthecartilageexplantmonoculture[55].Thecytokinesanddegradative enzymesdetectedinthe synovium–cartilage co-culturesweresimilarto thosefoundinthe synovialfluidofpatientswithOA,suggestingthatco-cultureshaveahigherpredictivepower
comparedwithmonocultures.Thisfurthersuggeststhatinvitromodelsofthejointmustbe multitissuesystemsinwhichthesynoviumplaysacrucialrole.
TheChallenge:MoreAdvanced InVitroTechnologies forMimickingArticular Function
Although static culture of tissue explants and co-cultures can recapitulate some in vivo functionalitiesofthejointtissues,thesemodelsstilllackmechanicalstimuli,suchastension, compression,andshearstressthatcellsexperienceinvivo.
Advancesinbiofabricationandmicrofluidicstechnologiesprovidetheopportunitytorecreate dynamicflowconditionsandmechanicalstimulation(Figure2)thatmayaidthedevelopmentof more predictive models of the human synovial joint. To this end,a variety of 3D culture techniques,including3Dhumanorganoids,humanorgan-on-a-chip[56],andbiofabricated tissue-likestructures[57],havebeenexploredtomodelphysiologicalandpathologicalhuman conditions.
Anoverviewofthecurrentinvitromodelsusedtoculturearticularjointtissuesisdepictedin Figure 3, highlighting advantages and limitations of each model type, thereby enablinga
Sy nov ia l me mbr ane Os te ochondr al unit
Joint ssue
Main cells
Type of mechanical load
Fibroblast-like synovial cells
Macrophage-like synovial
cells
T and B cells
Chondrocytes
Osteoblasts, osteocytes, and
osteoclasts
Microvascular and nerve cells
Tension
Fluid induced shear strain
Li ga m en t te nd on
Shear stress
Tension
Compression
Ligamentoblast/
ligamentocytes
Tenoblasts/tenocytes
Fibrochondrocytes
Vascular cells
Ho ffa’ s f at pad M eni sc usAdipocytes
Vascular and immune cells
Pericytes and mesenchymal
progenitor cells
Fibroblast-like cells
Chondrocyte-like cells
Nerve and vascular cells
Shear, tension, and
compression
Tension
Fluid induced shear
Tension
Fluid induced shear stress
Figure2.MechanicalStimulationoftheSynovialJoint.Illustrativetablewithrepresentativeexamplesofthejoint
Reproducibility and compaƟbility with high-throughput screening Simple and easy to work Models isolated events
ReducƟonisƟc ReducƟonisƟc No Ɵssue–Ɵssue interface No gradients of chemicals and oxygen StaƟc StaƟc
Includes interacƟons with other cells/Ɵssues
Limited 3D environment No direct Ɵssue–Ɵssue interface
Higher predicƟve power compared with monocultures
2D monolayer
Current in vitro models Advantages LimitaƟons
Tissue explants Bioreactors 2D cell culture 3D cell culture Tissue complexity Dynamic culture MulƟƟssue Organs-on-chips Ex vivo mulƟƟssue 3D biomaterial based Limited number of explants from each donor High intra- and
interdonor variaƟon; reproducibility challenge No gradients of chemicals and oxygen No Ɵssue–Ɵssue interface Lacks mechanical actuaƟon
Cells prone to change their phenotype Challenge to select a representaƟve cell source
Semidynamic 3D culturing
Extended culture Ɵme and quality of explants/Ɵssues Enables whole Ɵssue response to several sƟmuli The cells are cultured in their 3D naƟve microenvironment Can provide natural environmental cues Allow the study of cell-to-cell and cell-to-cell-to-matrix interacƟons
(A)
(B)
Figure3.CurrentandEnvisionedModelsoftheArticularJointTissues.(A)Illustrativetablewithmostrelevantin
vitrocell/tissueculturemodelsofthejoint,highlightingtheadvantagesanddrawbacksofeachmodel.(B)Schematic
comparisonbetweenmodels(A).Theevolutionthroughoutthemodelsystemsisillustratedin Figure3B,featuringincreasingcomplexitystepstowardsorgan-on-chipplatforms. Organs-on-chiparemicrofluidicdevicesthatcansimulatephysiologicalfunctionsoftissuesandorgans [14].Numerousorgans-on-chipmodelshavebeendevelopedsofar,however,therearestill issuestobeaddressed.First,weshouldchooseamatrix(e.g.,ahydrogel)withasuitable compositionandtopographytomimictheECMofthedifferenttissues.Forexample,cartilage exhibitsa highly organized structure which is differentfrom the bone or synoviumtissue. Further,the stiffnessof the matrixis also an importantfactor, since themechanicalloads experiencedbythecellsmightbeaffectedbythestiffnessofthematrix(i.e.,shielding)[58].The needtofindsuitableflowratesforeachofthedifferenttissuemodelswillalsobechallenging, especiallywhen connecting the differenttissues.Anotherimportant issue isrelated to the sourceofcells.Themajorityofthemicrofluidicdeviceshavebeenoperatedusingcelllines, eventhoughhuman-derivedprimarycellswouldbeabettermimicofthehumanphysiology. Themainlimitationsforusinghumanprimarycellsarethelimiteddonoravailability,potential variability,andtheircosts[59].
Basedonthefundamentalunderstandingoffunctionalandpathologicalorgans,acombination oftechnologicalapproaches(e.g.,softlithography,microfluidics,3Dbioprinting)isadoptedto simulateinvitrothenative3Dorganizationbyincorporatingbiomaterialscaffolds,continuous vascular-likeperfusion,mechanicalstimulation,andchemicalcueswithinthesameplatform. 3Dbioprintingrepresentsapromisingtechnologyforreplicatingtissue–tissueinterfacesdueto itsinherentcapabilitytodepositheterogeneousbioinks(livingcellsand/orbiomaterials)ina definedspatiotemporalmannerrelevanttobiologicalarchitectures.Biofabricated3Dinvitro modelsofferagreatopportunitytoinvestigatethephysiologicalandpathologicalprocessesof multitissues, as well as to perform drug screening and toxicological studies. Up to now, numerousinvitromodelsoftissuesandorgans,rangingfrombonetomicrovasculature,have beendevelopedwithavarietyofshapes,lengthscales,resolutions,andmechanicalproperties ofbiomaterialsandmostlyusinghuman-derivedcells[56,60].Recently,theperformanceofan osteochondralbioreactorwassignificantlyimprovedbyusing3Dprinting.Specifically,thenew modelenabledthefluidtransportthroughthecentralchambertobemaximizedandallowedfor opticalaccesswithinthe3Dconstruct,whilemaintainingdimensionscompatiblewitha96-well plate[48].Additionally,arecentlydescribed3Dprinterbioreactorallowsfortheprintingof3D constructs directly inside the bioreactor, reducing both contamination risk and the riskof damagetotheconstruct[61].3D-printingtechnologyhasgainedincreasedattentioninthe fabricationofmicrofluidicsystems.Thesedevicesarecommonlyfabricatedusing polydime-thylsiloxane (PDMS), which is easyto mold,biocompatible, transparent,and inexpensive. Additionally,PDMSmoldinghasveryhighresolution.However,thefabricationofthedevices involvesstackingandbondingdifferentlayerstogether,whichincreasesthefinalcostandlimits the 3Dcomplexity that can be achieved[62].Although, 3D-printingis an automated and assembly-freetechnology,withlow-costset-ups,limitedonlybytheresolutionoftheprinting process.Printingtechnologiessuchasstereolithography(SLA)ormeltelectrospinningwriting (MEW) could significantly improve the resolution of perfusable channels. For example, a custom-built projection SLA has been recently used to prepare complex 3D structures withpatterningfeaturesof<5mmresolution[63],whileMEWenablesthefabricationoffibers of4–7mm[64].
Organ-on-ChipTechnology
Microfluidicsystemscanbeconnectedwitheachothertoformmulticompartmentmicrofluidic devices.Animportantchallengewhenconnectingdifferentmicrofluidicdevicesinvolvesthe requirementforacommonmedia,orbloodsubstitute,suitableforeachtissueoftheinteracting
system[65].Thismulticompartmentorganizationisidealforthedesignofinvitromodelsofthe joint,amultitissueorganwherecellsareexposedtodifferenttypesofmechanicalstimuli.For example,withinthesynovialjoint,cellsexperienceahighmechanicalstress,boththrough load-bearingandbyshearforcescreatedbythemotionofthesynovialfluidduringexercise[66].The magnitude,direction,andtypeofstressappliedwasfoundtoaffectthetypeandseverityof inflammationofthejoints[67].However,thebiochemicaleffectsofshearforceonthesynovial membranehavenotbeenstudiedsystematicallyinaninvitroplatform,andamulticompartment microfluidicchamberwouldbeapromisingsolution.
Dependingontheorientationofthecompartmentalizedchannels,mainlytwotypesof con-figurationshave beendeveloped(Figure4C,D)(i.e.,upper–lowerchambersseparatedbya semiporous,stretchablemembraneandlateralchambersobtainedbyanarrayofmicroposts [68–70]oraremovabletemplate[69]).Inaddition,dynamiccompressivebioreactorshavebeen
Figure4.RecreatingMicrophysiologicallyArticularJoint-RelevantEnvironmentswithOrgan-on-Chips.(A)Generationoftheboneperivascularnichefor
studiesofbreastcancercolonization.Humanbonemarrow-derivedmesenchymalstemcells(MSCs)andendothelialcells(ECs)wereculturedinmonolayersandin3D
decellularizedbonematrix.Imagereproducedwithpermission,from[78].(B)Multiple‘organs-on-a-chip’platformtomodelmetastasisfromlungtobrain,bone,and
liverdownstreamorgans.Thismodularstrategycanbeusedtomodelthetissuesfromthearticularjoint.Imagereproducedwithpermission,from[79].(C)Schemeofa
microfluidicvascularizedbonetissuemodelformimickingrealboneangiogenesis.Thebone-mimickingchannelconsistsofamixtureoffibrinwithhydroxyapatite.
Imagereproducedwithpermission,from[69].(D)Schematicdrawingandphotographofthemicrofluidickidneyglomeruluschip,withtheurinaryandcapillary
compartmentsoftheglomerulusseparatedbyapolydimethylsiloxanemembrane.Cyclicstrainisappliedonthesidechanneltomimicthetissuestretch.Itisenvisioned
thatasimilardesigncanbeusedforjointarticulartissues,whichinvolvesimilarmechanicaltension,suchasthesynovialmembraneorthetendons.Imagereproduced
withpermission,from[80].(E)Pictureofacompleteactuatorchipimplementedinmechano-stimuliresponsivestudiesofneuronalcellnetworksonchip.Theimages
representaschematicdrawingofitsmechanismofactuationusinggasflow.Itisenvisionedthatasimilardesigncanbeusedforjointarticulartissuesinvolvingsimilar
developed to screen the effect of biomaterials and of mechanical stimulation on cellular behaviors(Figure4E)[71].
RecapitulatingtheMicrophysiologicalEnvironmentoftheSynovialJoint Thecurrentdevelopmentsdescribedaboveopenupnewwaystorecapitulatewholeorgansor combinationsof multiple organs ina body-on-a-chip fashion. However, the integrationof different organs still faces some challenges, including maintaining the functionality of the differentorgansandthestudyofintertissueresponsestodrugadministration.The develop-mentofthesemulticompartmentsystems,thatcaninteractwitheachotherinaphysiologically relevantmanner,wouldbecrucialforthedevelopmentofmodelsofthejoint,forexamplea joint-on-a-chip.
Organ-on-chiptechnologyprovides a platformto recreate physiologicallyrelevant environ-ments.Althoughthisnoveltechnologyhasbeenusedtodevelopbiomimeticsystemsofseveral organs(lung,kidney,liver),boneistheonlytissueofthejointthathasbeendevelopedonchip thusfar.Tomimictheinvivomicroenvironmentofbone,amicrofluidicapproachwasusedto developa vascularized bone tissuemodel. Fibrin was used as a model of the ECM and combined with various concentrations of hydroxyapatite nanocrystals to mimic the bone structure.Thepresenceofhydroxyapatiteresultedinenhancedangiogenicproperties, induc-ingimprovedsproutlength,sproutspeed,andlumendiameter[72].Otherstudiesfocused mainlyonbonemarrow-on-chip.Forexample,abonemarrow-on-a-chipsystemwas devel-opedbyfirstengineeringnewboneinvivoandthenperfusingitinamicrofluidicdeviceableto reconstitutethehematopoieticnichephysiology[73].Totestthefunctionalityandorgan-level response,theengineeredbonemarrowwasexposedtovaryingdosesofg-radiation,andthe resultscloselymimickedtheeffectsobservedinthebonemarrowoflivemice[73]. Toreconstitutethemicroenvironmentofthearticularjoint,aninvitro3Dmicrosystemmodelof theosteochondralunitwasdevelopedbyfittingamultichamberbioreactorintoamicrofluidic base[50].Tomimicthechondrogenicandosteogenictissues,humanbonemarrowstemcells (hBMSCs)wereseededwithinahyaluronicacid-basedhydrogelandagelatin/hydroxyapatite construct, respectively. The two compartments were supplied by two different medium streamsto reconstitutethechondral andosseousmicroenvironments andpromote tissue-specific differentiationof hBMSCs.In anextstep, theosteochondralunit was exposedto interleukin-1b(IL-1b)toevaluatetheresponsetoproinflammatorycytokinesandthe commu-nicationbetweenthetwocompartments.TheIL-1btreatmentoftheosseouscompartment resultedinastrongcatabolicresponseinthechondrallayer,whichwassignificantlyhigherthan the response observed after local exposure to IL-1b, suggesting an active biochemical communicationbetweenthetwolayers.Tobettermimicthejointenvironment,themicrotissue modelwasimprovedbyincorporatinganosteochondralinterface,consistingofanMSC-laden collagenhydrogel,andasynovialliningproducedwithMSCsseededonapolyethyleneglycol hydrogel[74].Althoughthismicrotissuerepresentsavaluablemodeloftheosteochondralunit, somekeyelementssuchasotherjointcomponents(menisci/tendons)andmechanical stimu-lation(e.g.,compression,shearstress)arestillmissing.
Besidesarticularcartilage,subchondralbone,menisci,tendons,andfatpad,akeycomponent ofthejointorganisthesynovium.Indeed,synovialinflammationisintimatelyassociatedwith jointdegenerationindiseasessuchasOAandRA[3].RecentstudiesindicatethatOAtissue andsynovialfluidcontainhighlevelsofcytokines,andthatchondrocytesandsynovialcellsin OA overproduce several inflammatory mediators, including IL-1b, tumor necrosis factor-a (TNF-a),andnitricoxide,whicharecharacteristicofinflammatoryarthritis.Further,the lipid
metabolismmightalsocontributetothepathogenesisofOAthroughinflammatorymediators, forinstance,adipokinesproducedbyadiposetissuesuchaspresentinHoffa’sfatpad[75].It becomesclearthatthesynoviumtogetherwiththeosteochondralunit,menisci,tendons,and fatpadareessentialcomponentsforthedevelopmentofanadvancedinvitromodelofthejoint. Further,synovialinflammationalsorepresentsaneffectivetargetforthedevelopmentofnovel therapeuticstrategies.However,amajorlimitationsofaristheinabilitytospecificallytarget synovialcells,without affectingthe whole-organismphysiology[3].Thislimitation couldbe overcomebyfabricatingindividualjointcomponentsusingmicrofluidictechnologyonchipto reconstitutethephysiologicalcontextofeachtissue,includingmechanicalandbiochemical stimulation,andcombinetheseindividualtissuesonchipinamultimodulardevice:the joint-on-chip.Thisapproachprovidestheopportunitytounderstandthemolecularpathwaysinvolvedin thephysiology andetiologyofeach tissueindividuallyandmay, asan example,aidinthe discoveryofsynovium-targetingtherapies.Furthermore,bydesigninginterconnectable organ-on-chipmodelsofcartilage,bone,synovium,tendons,andfatpadandarrangingthemasin vivo,itisalsopossibletostudycrosstalkbetweenthevarioustissues.Aninterconnectedmodel of the joint could be valuable for understanding joint homeostasis and to develop novel therapiestopreventandslowtheprogressionofjointdiseases.
ValidationofInVitroModelsoftheJoint
Theseadvancedmodels,duetotheirgreatlevelofphysiologicalmimicry,havethepotentialto replace or be integrated with animal studies for preclinical testing. To achieve this, one importantchallengestillneedstobeaddressedtotranslatethe‘on-chip’resultstotheclinic; thatis,thevalidationofthemodel.
Here,akeystepwillbetodemonstratethepredictivepowerofthemultitissuemodelofthejoint forhumandisease.Therefore,thereisastrongneedtodevelopvalidationmethodsthatcan assessthereproducibility, reliability,and thetranslationalpotential ofthose models[60].A possiblestrategycouldbethetestingofarangeofdrugswithknowneffectagainstRA(inthe caseofthejointmodels)andcomparetheresultswithdataobtainedfrominvitrotissue co-culturemodels, animal studies,and efficacy studiesin patients.Forinstance, celecoxib,a nonsteroidalanti-inflammatorydrugusedforthetreatmentofOAandRA,hasbeenusedto evaluatetheaccuracyofamodelbasedontheco-cultureofatissue-engineeredcartilage constructwithsynovialfibroblastsandmacrophages.Thesystemcouldreplicateonlysomeof theinvivoresponsesofthedrug[76].Animportantlimitation,however,isthatthismodelonly mimicscartilage–synoviuminteractions,whilenumerousstudiesdemonstratedaclose inter-actionbetweencartilage–subchondralboneandvasculature[49].Thisfurtherhighlightsthe necessityto developmultitissue modelsthat can bettermimic the invivo behavior.These multitissuedevicescouldbeadvantageousfordrugscreening,enablingthetestingofvarying concentrationsof adrugor combinationsof drugs,simultaneously andina cost-effective manner.Most importantly,these devicescan beoperated withhuman cells and thusare capableoffurthermimickingthehumanphysiologyandmetabolism[77].
ConcludingRemarksandFuturePerspectives
Understandingthemolecularmechanismsinvolvedinjointhomeostasisanddiseaseis funda-mentalforthedevelopmentofnoveltherapeutictreatments.Amajorchallengesofarhasbeen thelackofmodelsthatcanfaithfullymimicjointphysiologyinhealthanddisease.Thelimited predictivepowerofexistinganimalmodelshasdriventheneedtodevelopadvancedinvitro modelsthatcanmoreaccuratelyrepresenttheinvivo-likeenvironment.Recentadvancesin engineeringandbiologyhave enabledthe developmentofinvitromodels thatrecapitulate
OutstandingQuestions
Canthecomplexmicroenvironmentof
thesynovialjointbefaithfully
recapitu-lated using the organ-on-chip
technology?
Isitpossible tostudyonsetofjoint
diseasesandthesequenceof
patho-logicaleventsrecapitulatingthe
com-plexity of human disease using a
multicompartmentjoint-on-a-chip?
Willsuchahumanizedjointmodelbe
abletopredicttheefficacyandtoxicity
ofnoveltreatmentsforarthritis?
Can a multicompartment
joint-on-a-chipenablethediscoveryofearly
complex 3D organ-level structures with integrated mechanical, biochemical, and physical stimulation(see OutstandingQuestions). This technologyhasthe potential to emulatethe variouscomponentsofthejointorganandtheinterplaybetweenthedifferentelementsofthe joint.Inadditiontoa‘humanized’jointmodel,thedevelopmentofjointmodelsforanimalsis alsoimportantfortheapplicationinveterinarymedicine.Specifically,theuseofhumanand animalmodelsunderasimilarexperimentalsettingcouldbringtolightspecies-specific differ-ences.Ultimately,the developmentofan advancedinvitro modelofthe jointwill assistin reducingorreplacingtheuseofanimalmodelsinbiomedicalresearch,personalizedmedicine, andpharmaceuticalstudies.
Acknowledgments
TheauthorsacknowledgethefinancialsupportprovidedbytheDutchArthritisFoundation(LLP12,LLP22,andLLP25)and
thefinancialsupportofthestrategicallianceprogramentitled:Advancedbiomanufacturing,fundedbytheUniversityof
Twente,UtrechtUniversityandUniversityMedicalCenterUtrecht,projecttitle:Bioprintingfunctionaltissuesfromstem
cellsandenablingbiomaterials.
References
1. teMoller,N.C.andvanWeeren,P.R.(2017)Howexercisein flu-encesequinejointhomeostasis.Vet.J.222,60–67
2. McIlwraith,C.W.andvanWeeren,R.(2015)JointDiseaseinthe Horse,ElsevierHealthSciences
3. Bhattaram,P.andChandrasekharan,U.(2017)Thejoint syno-vium:acriticaldeterminantofarticularcartilagefatein inflamma-toryjointdiseases.Semin.CellDev.Biol.62,86–93
4. Pearson,M.J.etal.(2017)IL-6secretioninosteoarthritispatients ismediatedbychondrocyte-synovialfibroblastcross-talkandis enhancedbyobesity.Sci.Rep.7,3451
5. Goldring,S.R.andGoldring,M.B.(2016)Changesinthe osteo-chondral unit during osteoarthritis: structure, function and cartilage–bonecrosstalk.Nat.Rev.Rheumatol.12,632
6. Findlay,D.M.andKuliwaba,J.S.(2016)Bone–cartilagecrosstalk: aconversationforunderstandingosteoarthritis.BoneRes.4, 16028
7. Hu,Y.etal.(2013)Advancesinresearchonanimalmodelsof rheumatoidarthritis.Clin.Rheumatol.32,161–165
8. McCoy,A.(2015)Animalmodelsofosteoarthritis:comparisons andkeyconsiderations.Vet.Pathol.52,803–818
9. Thysen,S.etal.(2015)Targets,modelsandchallengesin oste-oarthritisresearch.Dis.Model.Mech.8,17–30
10.McNamee,K.etal.(2015)Animalmodelsofrheumatoidarthritis: howinformativearethey?Eur.J.Pharmacol.759,278–286
11.Denayer,T.etal.(2014)Animalmodelsintranslationalmedicine: validationandprediction.NewHoriz.Transl.Med.2,5–11
12.Wang,K.etal.(2015)Investigationaldrugsforthetreatmentof osteoarthritis.ExpertOpin.Investig.Drugs,24,1539–1556
13.Esch,E.W.etal.(2015)Organs-on-chipsatthefrontiersofdrug discovery.Nat.Rev.DrugDiscov.14,248
14.Bhatia,S.N.andIngber,D.E.(2014)Microfluidic organs-on-chips.Nat.Biotechnol.32,760
15.Skardal,A.etal.(2016)Organoid-on-a-chipandbody-on-a-chip systemsfordrugscreeninganddiseasemodeling.DrugDiscov. Today,21,1399–1411
16.Das,R.K.etal.(2016)Stress-stiffening-mediatedstem-cell com-mitmentswitchinsoftresponsivehydrogels.Nat.Mater.15,318
17.Gattazzo,F.etal.(2014)Extracellularmatrix:adynamic micro-environmentforstemcellniche.Biochim.Biophys.Acta,1840, 2506–2519
18.Gilbert,P.M.andWeaver,V.M.(2017)Cellularadaptationto biomechanicalstressacrosslengthscalesintissuehomeostasis anddisease.Semin.CellDev.Biol.67,141–152
19.Sun,H.B.(2010)Mechanicalloading,cartilagedegradation,and arthritis.Ann.N.Y.Acad.Sci.1211,37–50
20.O’Conor,C.J.etal.(2014)TRPV4-mediated mechanotransduc-tion regulates the metabolic response of chondrocytes to dynamicloading.Proc.Natl.Acad.Sci.U.S.A.111,1316–1321
21.Johnson,C.I.etal.(2016)Invitromodelsforthestudyof osteo-arthritis.Vet.J.209,40–49
22.Li,Y.etal.(2015)Effectsofinsulin-likegrowthfactor-1and dexamethasoneoncytokine-challengedcartilage:relevanceto post-traumaticosteoarthritis.Osteoarthr.Cartil.23,266–274
23.Beekhuizen,M.etal.(2013)InhibitionofoncostatinMin osteoar-thriticsynvialfluidenhancesGAGproductioninosteoarthritic cartilagerepair.Eur.CellMater.26,80–90
24.Braun,H.J.etal.(2014)Theeffectofplatelet-richplasma for-mulationsandbloodproductsonhumansynoviocytes: implica-tionsforintra-articularinjuryandtherapy.Am.J.SportsMed.42, 1204–1210
25.Silverstein,A.M.etal.(2017)Towardunderstandingtheroleof cartilageparticulatesinsynovialinflammation.Osteoarthr.Cartil. 25,1353–1361
26.Ma,B.etal.(2013)Geneexpressionprofilingofdedifferentiated humanarticularchondrocytesinmonolayerculture.Osteoarthr. Cartil.21,599–603
27.Johnstone,B.etal.(1998)Invitrochondrogenesisofbone marrow-derivedmesenchymalprogenitorcells.Exp.CellRes. 238,265–272
28.Teixeira,L.M.etal.(2012)Highthroughputgenerated micro-aggregatesofchondrocytesstimulatecartilageformationinvitro andinvivo.Eur.CellMater.23,387–399
29.Pothacharoen,P.etal.(2014)Effectsofsesaminonthe biosyn-thesisofchondroitinsulfateproteoglycansinhumanarticular chondrocytesinprimaryculture.Glycoconj.J.31,221–230
30.Leijten,J.etal.(2016)Bioinspiredseedingofbiomaterialsusing threedimensionalmicrotissuesinduceschondrogenicstemcell differentiationandcartilageformationundergrowthfactorfree conditions.Sci.Rep.6,36011
31.Zellner,J.etal.(2015)Dynamichydrostaticpressureenhances differentiallythechondrogenesisofmeniscalcellsfromtheinner andouterzone.J.Biomech.48,1479–1484
32.Fahy,N.etal.(2018)Mechanicalstimulationofmesenchymal stemcells:implicationsforcartilagetissueengineering.J.Orthop. Res.36,52–63
33.Gilbert,S.J.andBlain,E.J.(2018)Cartilagemechanobiology: howchondrocytesrespondtomechanicalload.In Mechanobi-ologyinHealthandDisease(Verbruggen,S.,ed.),pp.99–126, Elsevier
34.Steward,A.J.andKelly,D.J.(2015)Mechanicalregulationof mesenchymalstemcelldifferentiation.J.Anat.227,717–731
35.Yang,J.etal.(2017)Cell-ladenhydrogelsforosteochondraland cartilagetissueengineering.ActaBiomater.57,1–25
36.Cosgrove,B.D.etal.(2016)N-cadherinadhesiveinteractions modulatematrixmechanosensingandfatecommitmentof mes-enchymalstemcells.Nat.Mater.15,1297
37.Bian,L.etal.(2010)Dynamicmechanicalloadingenhances functionalpropertiesoftissue-engineeredcartilageusingmature caninechondrocytes.TissueEng.A,16,1781–1790
38.Holloway,J.L.etal.(2014)Modulatinghydrogelcrosslinkdensity anddegradationtocontrolbonemorphogeneticproteindelivery andinvivoboneformation.J.Control.Release,191,63–70
39.Visser,J.etal.(2015)Reinforcementofhydrogelsusing three-dimensionallyprintedmicrofibres.Nat.Commun.6,6933
40.Khorshidi,S.andKarkhaneh,A.(2018)Areviewongradient hydrogel/fiberscaffoldsforosteochondralregeneration.J.Tissue Eng.Regen.Med.12,e1974–e1990
41.Yang,G.etal.(2016)Multilayeredpolycaprolactone/gelatin fiber-hydrogelcompositefortendontissueengineering.Acta Bio-mater.35,68–76
42.FontTellado,S.etal.(2017)Fabricationandcharacterizationof biphasicsilkfibroinscaffoldsfortendon/ligament-to-bonetissue engineering.TissueEng.A,23,859–872
43.Mahadik,B.P.etal.(2014)Microfluidicgenerationofgradient hydrogelstomodulatehematopoieticstemcellculture environ-ment.Adv.Healthc.Mater.3,449–458
44.Grenier,S.etal.(2014)Aninvitromodelforthepathological degradationofarticularcartilageinosteoarthritis.J.Biomech.47, 645–652
45.Wang,Y.et al.(2017) Dexamethasonetreatmentaltersthe responseofhumancartilageexplantstoinflammatorycytokines andmechanicalinjuryasrevealed bydiscoveryproteomics. Osteoarthr.Cartil.25,S381–S382
46.Brama,P.etal.(2000)Topographicalmappingofbiochemical propertiesofarticularcartilageintheequinefetlockjoint.Equine Vet.J.32,19–26
47.vanCaam,A.P.etal.(2015)TGFb1blockschondrocyte hyper-trophyandmaintainscellviabilityinculturedcartilageexplantsbut doesnotprotectagainstproteoglycanloss.Osteoarthr.Cartil.23, A137–A138
48.Nichols,D.A.etal.(2018)Designandvalidationofan osteochon-dralbioreactorforthescreeningoftreatmentsforosteoarthritis. Biomed.Microdevices,20,18
49.Pirosa,A.etal.(2018)Engineeringin-vitrostemcell-based vas-cularizedbonemodelsfordrugscreeningandpredictive toxicol-ogy.StemCellRes.Ther.9,112
50.Lin,H.etal.(2014)Stemcell-basedmicrophysiological osteo-chondralsystemtomodeltissueresponsetointerleukin-1b.Mol. Pharm.11,2203–2212
51.DeVries-vanMelle,M.L.etal.(2011)Anosteochondralculture modeltostudymechanismsinvolvedinarticularcartilagerepair. TissueEng.CMethods,18,45–53
52.Schwab,A.etal.(2017)Exvivocultureplatformforassessment ofcartilagerepairtreatmentstrategies.ALTEX,34,267–277
53.Mouser,V.H.etal.(2018)Exvivomodelunravellingcell distribu-tioneffectinhydrogelsforcartilagerepair.ALTEX,35,65–76
54.Swärd,P.etal.(2017)Cocultureofbovinecartilagewith syno-viumandfibrousjointcapsuleincreasesaggrecanaseandmatrix metalloproteinaseactivity.ArthritisRes.Ther.19,157
55.Beekhuizen,M.etal.(2011)Osteoarthriticsynovialtissue inhibi-tionofproteoglycanproductioninhumanosteoarthriticknee cartilage:establishment and characterizationof along-term cartilage–synoviumcoculture.ArthritisRheum.63,1918–1927
56.Zhang,B.etal.(2018)Advancesinorgan-on-a-chipengineering. Nat.Rev.Mater.3,257–278
57.Moroni,L.etal.(2018)Biofabricationstrategiesfor3Dinvitro modelsandregenerativemedicine.Nat.Rev.Mater.3,21–37
58.Lee,S.H.etal.(2017)Hydrogel-basedthree-dimensionalcell culturefororgan-on-a-chipapplications.Biotechnol.Prog.33, 580–589
59.Kimura,H.etal.(2018)Organ/body-on-a-chipbasedon micro-fluidictechnologyfordrugdiscovery.DrugMetab. Pharmacoki-net.33,43–48
60.Ahadian,S.etal.(2017)Organ-on-a-chipplatforms:a conver-genceofadvancedmaterials,cells,andmicroscaletechnologies. Adv.Healthc.Mater.PublishedonlineOctober16,2017.http:// dx.doi.org/10.1002/adhm.201700506PublishedonlineOctober 16,2017
61.Smith,L.J.etal.(2018)FABRICA:abioreactorplatformfor printing,perfusing,observing,& stimulating3D tissues.Sci. Rep.8,7561
62.Bhattacharjee,N.etal.(2016)Theupcoming3D-printing revolu-tioninmicrofluidics.LabChip,16,1720–1742
63.Raman,R.etal.(2016)High-resolutionprojection microstereo-lithographyforpatterningofneovasculature.Adv.Healthc.Mater. 5,610–619
64.Castilho,M.etal.(2017)Meltelectrospinningwritingof poly-hydroxymethylglycolide-co-e-caprolactone-based scaffolds for cardiactissueengineering.Adv.Healthc.Mater.6,1700311
65.Coppeta,J.etal.(2017)Aportableandreconfigurable multi-organplatformfordrugdevelopmentwithonboardmicrofluidic flowcontrol.LabChip,17,134–144
66.Schett,G.etal.(2001)Thestressedsynovium.ArthritisRes.Ther. 3,80
67.Manunta,A.F.etal.(2014)Sports-relatedchangesofthesynovial membrane.Joints,2,181
68.Adriani,G.etal.(2017)A3Dneurovascularmicrofluidicmodel consistingofneurons,astrocytesandcerebralendothelialcellsas ablood–brainbarrier.LabChip,17,448–459
69.Yamada,A.etal.(2016)Transientmicrofluidic compartmentali-zationusingactionablemicrofilamentsforbiochemicalassays, cellcultureandorgans-on-chip.LabChip,16,4691–4701
70.Mi,S.etal.(2018)Constructionofaliversinusoidbasedonthe laminarflowonchipandself-assemblyofendothelialcells. Bio-fabrication,10,025010
71.Seo,J.etal.(2018)Interconnectabledynamiccompression bio-reactorsforcombinatorialscreeningofcellmechanobiologyinthree dimensions.ACSAppl.Mater.Interfaces,10,13293–13303
72.LiáJeon,N.(2015)Microfluidicvascularizedbonetissuemodel withhydroxyapatite-incorporatedextracellularmatrix.LabChip, 15,3984–3988
73.Torisawa,Y.S.etal.(2014)Bonemarrow–on–a–chipreplicates hematopoieticnichephysiologyinvitro.Nat.Methods,11,663
74.Lozito, T.P.et al. (2013) Three-dimensional osteochondralmicrotissue tomodelpathogenesisofosteoarthritis.StemCellRes.Ther.4,S6
75.Robinson,W.H.etal.(2016)Low-gradeinflammationasakey mediatorofthepathogenesisofosteoarthritis.Nat.Rev. Rheu-matol.12,580–592
76.Peck,Y.etal.(2018)Establishmentofaninvitro three-dimen-sionalmodelforcartilagedamageinrheumatoidarthritis.J. TissueEng.Regen.Med.12,e237–e249
77.Esch,M.B.etal.(2014)Howmulti-organmicrodevicescanhelp fosterdrugdevelopment.Adv.DrugDeliv.Rev.69,158–169
78.Marturano-Kruik,A.etal.(2018)Humanboneperivascular niche-on-a-chipforstudyingmetastaticcolonization.Proc.Natl.Acad. Sci.115,1256–1261
79.Ma,Y.-H.V.etal.(2018)Areviewofmicrofluidicapproachesfor investigatingcancerextravasationduringmetastasis.Microsyst. Nanoeng.4,17104
80.Musumeci,G.etal.(2014)Lubricinexpressioninhuman oste-oarthritickneemeniscusandsynovialfluid:amorphological, immunohistochemicalandbiochemicalstudy.ActaHistochem. 116,965–972
81.Xie,S.etal.(2018)Nanoscalemembraneactuatorforinvitro mechano-stimuliresponsivestudiesofneuronalcellnetworkson chip.J.Micromech.Microeng.28,085011
82.vanCaam,A.etal.(2016)ExpressionofTGFb-familysignalling componentsinageingcartilage:age-relatedlossofTGFband BMPreceptors.Osteoarthr.Cartil.24,1235–1245
83.Coughlin,T.etal.(2016)Primaryciliaexpressioninbonemarrow inresponsetomechanicalstimulationinexplantbioreactor cul-ture.Eur.CellMater.32,111–122
84.Hennerbichler,A.etal.(2007)Interleukin-1andtumornecrosis factoralphainhibitrepairoftheporcinemeniscusinvitro. Osteo-arthr.Cartil.15,1053–1060
85.Cucchiarini,M.etal.(2015)OverexpressionofTGF-bvia rAAV-mediatedgenetransferpromotesthehealingofhumanmeniscal lesionsexvivoonexplantedmenisci.Am.J.SportsMed.43, 1197–1205
86.Spiesz,E.M.etal.(2015)Tendonextracellularmatrixdamage, degradationandinflammationinresponsetoinvitrooverload exercise.J.Orthop.Res.33,889–897
87.Andersen,M.etal.(2018)AssociationbetweenIL-6productionin synovialexplantsfromrheumatoidarthritispatientsandclinical andimagingresponsetobiologictreatment:apilotstudy.PLoS One,13,e0197001
88.Fahy,N.etal.(2014)Humanosteoarthriticsynoviumimpacts chondrogenicdifferentiationofmesenchymalstemcellsvia mac-rophagepolarisationstate.Osteoarthr.Cartil.22,1167–1175
89.Makris,E.A.etal.(2011)Thekneemeniscus:structure–function, pathophysiology,currentrepairtechniques,andprospectsfor regeneration.Biomaterials,32,7411–7431
90.Stone,A.V.etal.(2014)Pro-inflammatorystimulationofmeniscus cellsincreasesproductionofmatrixmetalloproteinasesand addi-tionalcatabolicfactorsinvolvedinosteoarthritispathogenesis. Osteoarthr.Cartil.22,264–274
91.Woo,S.L.etal.(2019)Functionaltissueengineeringofligament andtendoninjuries.InPrinciplesofRegenerativeMedicine(Atala, A.,ed.),pp.1179–1198,Elsevier
92.Jung,H.-J.etal.(2009)Roleofbiomechanicsinthe understand-ingofnormal,injured,andhealingligamentsandtendons.BMC SportsSci.Med.Rehabil.1,9
93.Haslauer,C.M.etal.(2013)Lossofextracellularmatrixfrom articularcartilageismediatedbythesynoviumandligamentafter anteriorcruciateligamentinjury.Osteoarthr.Cartil.21,1950– 1957
94.Ioan-Facsinay,A.andKloppenburg, M.(2013)Anemerging playerinkneeosteoarthritis:theinfrapatellarfatpad.Arthritis Res.Ther.15,225