CHONDROCYTES

Collagen Type II enhances chondrogenic differentiation inagarose-based modular microtissues

RAMKUMAR TIRUVANNAMALAI ANNAMALAI, DAVID R. MERTZ, ETHAN L.H. DALEY& JAN P. STEGEMANN

Department of Biomedical Engineering, University of Michigan,Ann Arbor, Michigan, USA

AbstractBackground aims. Cell-based therapies have made an impact on the treatment of osteoarthritis; however, the repair and re-generation of thick cartilage defects is an important and growing clinical problem. Next-generation therapies that combinecells with biomaterials may provide improved outcomes. We have developed modular microenvironments that mimic thecomposition of articular cartilage as a delivery system for consistently differentiated cells. Methods. Human bone marrow–derived mesenchymal stem cells (MSC) were embedded in modular microbeads consisting of agarose (AG) supplementedwith 0%, 10% and 20% collagen Type II (COL-II) using a water-in-oil emulsion technique. AG and AG/COL-II microbeadswere characterized in terms of their structural integrity, size distribution and protein content. The viability of embeddedMSC and their ability to differentiate into osteogenic, adipogenic and chondrogenic lineages over 3 weeks in culture werealso assessed. Results. Microbeads made with <20% COL-II were robust, generally spheroidal in shape and 80 ± 10 μm indiameter. MSC viability in microbeads was consistently high over a week in culture, whereas viability in corresponding bulkhydrogels decreased with increasing COL-II content. Osteogenic differentiation of MSC was modestly supported in bothAG and AG/COL-II microbeads, whereas adipogenic differentiation was strongly inhibited in COL-II containing microbeads.Chondrogenic differentiation of MSC was clearly promoted in microbeads containing COL-II, compared with pure AGmatrices. Conclusions. Inclusion of collagen Type II in agarose matrices in microbead format can potentiate chondrogenicdifferentiation of human MSC. Such compositionally tailored microtissues may find utility for cell delivery in next-generation cartilage repair therapies.

KeyWords: Agarose,Biomaterials,Cartilage tissue engineering,Chondrogenesis,Collagen,Mesenchymal stromal cells,Microencapsulation,Modular tissue engineering

Introduction

Repair and regeneration of cartilage is a difficult or-thopedic problem due to the low inherent healingcapacity of the native tissue [1–3]. Articular carti-lage is a well-organized tissue with remarkabledurability; however, damage may result in debilitat-ing joint pain and functional impairment. Clinicalapproaches such as osteotomy and osteochondral grafttransplantation [4] have shown benefits in terms ofrelieving pain, delaying further deterioration and re-storing partial function, but these therapies do notresult in the regeneration of fully competent tissues,and their sustainability in a load-bearing environ-ment remains uncertain.The lack of long-term clinical

solutions for cartilage repair has motivated the searchfor improved regenerative therapies that achieve fullrecovery of intractable and large defects.

Cell-based approaches have emerged as a clinicaltherapy to regenerate damaged cartilage [5–10].However, challenges associated with naked chondro-cyte delivery [11,12] have led to the development ofmatrix-assisted strategies for cell implantation [13–16].Such biomaterial-based approaches use a preformedscaffold or a hydrogel matrix, often supplemented withbiochemical cues to provide mechanical stability whilesustaining chondrogenic differentiation. It has also beenshown that native extracellular matrix components incombination with autologous cells can to some degreerecapitulate the native microenvironment and

Correspondence: Jan P. Stegemann, PhD, Department of Biomedical Engineering, University of Michigan, 1101 Beal Avenue, Ann Arbor, MI 48109, USA.E-mail: jpsteg@umich.edu

(Received 14 May 2015; accepted 13 October 2015)

ISSN 1465-3249 Copyright © 2015 International Society for Cellular Therapy. Published by Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.jcyt.2015.10.015

Cytotherapy, 2016; 18: 263–277

architecture, which can improve clinical outcomes [17].Pre-clinical studies have also shown that cell-seededscaffolds can be maintained in perfusion cultures withexogenous stimuli before implantation to furtherimprove graft maturation and host–implant integra-tion [18–20]. Despite these efforts, a repair tissue withfunctional properties and stability comparable to ar-ticular cartilage has yet to be engineered.

Bone marrow-derived mesenchymal stem cells(MSC) have been investigated widely for cartilagerepair applications [21–24]. These progenitor cellsare readily available, have demonstrated multi-lineage potential [25] and also exhibit valuableimmunomodulatory and tissue homing properties [26].MSC have been shown to differentiate into both car-tilage [27–30] and bone [31–33], supporting their usein orthopaedic applications. In addition, MSC canexhibit both trophic [34] and chemotactic effects [35]to create a regenerative tissue environment. Notably,even MSC isolated from the marrow of patients withadvanced osteoarthritis retain their chondrogenic po-tential and synthesize cartilage-specific matrix [27].Taken together, the tissue-specific and pro-regenerativecapabilities of MSC make them an excellent cell sourcefor engineering of cartilage tissue. Combination ofMSC with an appropriate biomaterial scaffold thatmimics key aspects of the native extracellular matrixarchitecture and biochemistry may further enhancetheir regenerative potential.

The major material components of articular car-tilage are large proteoglycans with interspersed fibrillarcollagen, which constitute approximately 15–25% and50–60% of dry weight, respectively [36–39].Proteoglycans contribute to the compressive stiff-ness of the cartilage, while the tensile strength andresilience are dependent on collagen fibers [40]. In ar-ticular cartilage, chondrocytes are embedded in thematrix at a relatively low cell density (~104 cells/mm3) and account for only approximately 1% of thetissue volume [41,42]. In their native differentiatedstate, chondrocytes have a generally spheroidal mor-phology and synthesize collagenType II (COL-II) andlarge proteoglycans. However, when isolated and placedin conventional two-dimensional culture, chondrocyteswill de-differentiate, spread on the culture surface andproduce predominantly collagen Type I and smallproteoglycans [43–45]. Although collagen Type I iscommonly used as a scaffold in tissue engineeringbecause of its wide availability [46,47], it has beenshown that COL-II can preferentially promote cell pro-liferation, extracellular matrix deposition and woundhealing by chondrocytes [48–50]. The polysaccha-ride agarose (AG) has been used as a mimic of theproteoglycan component of cartilage and has beenshown to maintain the spherical morphology ofchondrocytes [51,52], as well as support the deposi-

tion of an appropriate pericellular matrix [51–54]. Invivo studies have demonstrated that AG provides a mi-croenvironment that supports the non-hypertrophicand non-proliferative chondrogenic phenotype [55].The physiological response to AG resembles a wound-healing response similar to other biomaterialscommonly used in the context of tissue repair [56].Studies in both animals and humans have shown thatAG can be completely biodegraded and cleared afterimplantation without adverse effects [55,57].

The goal of the present study was to fabricate andcharacterize modular microenvironments that mimicthe proteoglycan-protein composition of cartilage tissue,with an emphasis on the ability to support lineage-specific differentiation of human bone marrow–derived MSC. The proteoglycan component wasrepresented by AG, a polysaccharide that can easilybe formed into a hydrogel and that has found utilityas a matrix in cartilage tissue engineering [51,58].Theprotein component was represented by reconsti-tuted COL-II, which can form fibrillar structures asin the native tissue [37]. Cells were embedded direct-ly into the AG/COL-II matrix using a water-in-oilemulsion technique, producing discrete “microbeads”(~80 μm in diameter) consisting of MSC embeddedin a spheroidal hydrogel matrix.The modular formathas several potential advantages over bulk gel methodsin terms of reducing diffusion path lengths, allowingpre-culture of matrix-adhered cells, and delivering adifferentiated cell population. Microbeads made withspecific COL-II contents were characterized for theirstructural integrity, size distribution and proteincontent. MSC viability and their potential to differ-entiate into osteogenic, adipogenic and chondrogeniclineages were also assessed.The long-term goal of thiswork is to develop injectable, cell-based modular mi-croenvironments that promote specific tissueregeneration, as shown schematically in Figure 1 inthe case of articular cartilage. Such a therapy wouldprovide a three-dimensional matrix environment andcells of specified function that could be delivered min-imally invasively to sites of tissue damage.

Methods

Biopolymers and MSC culture

AG (Fisher Scientific) was low melting point gradewith an average molecular weight of 120 kDa, gellingtemperature between 34.5 and 37.5°C and gel-strength of ~500 g/cm2 min. AG stock solution wasmade at 2.0 wt% by dissolving the appropriate amountof powdered AG in DI water heated to 60°C and ster-ilized using a 0.2-μm filter. COL-II (Elastin ProductsCompany) was from mouse sternum with a molecu-lar weight of 1000 kDa, <0.4% proteoglycan and wastested for the absence for collagenType I. COL-II stock

264 R.T. Annamalai et al.

solution was made at a concentration of 0.4 wt% bydissolving sterile lyophilized COL-II in 0.02 N aceticacid solution.

Human bone marrow–derived MSC were ob-tained from a commercial vendor (RoosterBio Inc.).The cells were from an individual male donor 31–45years of age. These cells tested positive for CD166,CD105, CD90 and CD73 and negative for CD14,CD34 and CD45. Cells were expanded in standardMSC growth medium consisting of α-MEM (GibcoLife Technologies) supplemented with 10% MSC-FBS (Gibco) and penicillin (50 U/mL)/streptomycinsulfate (50 μg/100 mL, Gibco). MSC from passage 4–7were maintained at 37.0°C in standard cell culture in-cubators, and the medium was replenished every 2days. For encapsulated cultures, the microbeads weresuspended in a suitable volume of cell culture mediumand maintained in 15-mL vented polypropylene bio-reaction tubes (Celltreat).

Microbead and bulk gel fabrication and characterization

Cells were encapsulated in AG and AG/COL-IImicrobeads at a concentration of 0.5 million cells/mL using an adaptation of a water-in-oil emulsification

procedure described previously [46], and the final con-centrations of each component in the microbeadformulations are listed in Table I. Briefly, MSC weresuspended in matrix solution consisting of AG andCOL-II solutions, 0.1 N sodium hydroxide (Sigma)and 10% fetal bovine serum (FBS; Gibco).The cell-matrix mixture was injected into 80 mL of stirredpolydimethylsiloxane (PMX-200, 100 cS; XiameterDow Corning). Emulsification was carried outusing an impeller speed of 800 rpm. Immediatelyafter injection of the cell-matrix suspension, thepolydimethylsiloxane was kept at 37°C for 5 min usinga water bath. Once the matrix suspension was emul-sified, the water bath was replaced with a crushed icebath and stirred for a further 25 min. The resultingmicrobeads were then separated from the oil phaseby centrifugation at 150 g for 5 min. Microbeads werewashed twice in MSC culture medium and main-tained under standard culture conditions. Themicrobead format allowed aliquoting of precise volumesof suspended microbeads for use in biochemical andother assays. For microbead maintenance, the culturemedium was replenished every second day. Bulk hy-drogel constructs were also made for comparison tomicrobeads at the same cell concentration. For bulk

Figure 1. Schematic of the preparation of injectable microbeads for cartilage repair. A water-in-oil emulsification technique is used to gen-erate spheroidal microbeads containing embedded cells. The microbead matrix consists of agarose and COL-II to mimic the compositionof cartilage. Microbeads can be cultured in suspension and delivered via injection, and their small size facilitates mass transport to theembedded cells. The long-term goal is to develop an injectable, living articular cartilage tissue analog. Parts of this figure were obtainedfrom Servier Medical Art (http://www.servier.com).

Table I. Volumes (μL) of AG, COL-II, NaOH and FBS used to create collagen-agarose bead formulations.

Formulations AGa COL-IIb NaOH FBS α-MEM

AG/0% COL-II 1500 0 0 300 1200AG/10% COL-II 964 536 107 300 1093AG/20% COL-II 667 833 167 300 1033AG/100% COL-II 0 1500 300 300 900

% COL-II is given as wt/wt (collagen/agarose).aAG stock concentration is 20 mg/mL of deionized water (2 wt%).bCOL-II stock concentration is 4.0 mg/mL of 0.02 N acetic acid (0.4 wt%).

Agarose/collagen II–based microtissues for cartilage repair 265

gels, MSC were suspended in the corresponding hy-drogel solution and after through mixing the hydrogelwas poured into glass-bottom petri dishes and allowedto gel at 37°C for 30 min. Gelled bulk constructs wereapproximately 1.5 mm in thickness. Culture mediumwas added on top of each gel and cultures were main-tained at 37°C.

Freshly made microbeads were imaged using aninverted phase-contrast microscope (Nikon). Fourrandom fields were imaged for each formulation, to-taling approximately 150 microbeads in each count.Microbead diameter was measured manually usingImageJ software (National Institutes of Health) to de-termine the size distribution of the formulations tested.Microbeads were also visualized microscopically to de-termine their structural integrity.The formulations werequalitatively assessed based on the ratio of intactmicrobeads to broken or fragmented microbeads ina specific volume of the sample.

Collagen content in the microbeads was ana-lyzed via protein staining using Coomasie Blue G-250(Bio-Rad). Staining solution was made by adding10 mL glacial acetic acid to 45 mL double-distilledwater. Three hundred milligrams of powderedCoomassie blue was then dissolved in 45 mL of meth-anol, and the two solutions were combined and sterilefiltered.The microbeads were suspended in the stain-ing solution for 5 min and washed multiple times inphosphate-buffered saline (PBS) to remove non-specific binding.The microbeads were then observedand imaged with a bright-field microscope (Nikon).

Analysis of MSC morphology and viability

Cells were encapsulated in microbeads and culturedfor 1 week to study short-term viability and cell mor-phology. For Live/Dead Cell Assay, Calcein-AM(4.0 mmol/L), ethidium homodimer (2.0 mmol/L) and4′,6-diamidino-2-phenylindole (DAPI; 2.0 μg/mL) so-lutions were used (Molecular Probes). After aspiratingthe culture medium, microbeads were washed in PBSand incubated in dye solution at 37°C for 45 min. Aftertwo washes in 10 mmol/L PBS, microbeads were re-suspended in fresh PBS and imaged under an invertedfluorescence microscope (Nikon) using filter sets forgreen and red (Excitation/Emission 488/530 for calcein-AM and 488/630 for ethidium homodimer). Multiplenon-overlapping images were taken for every sampleand the percentage of viable cells was quantified usingImageJ. Briefly, 16-bitTIF images were separated into8-bit red and green channels and converted to binaryimages for particle analysis counts. For cell morphol-ogy analysis,Texas Red phalloidin (Molecular Probes,LifeTechnologies) was used to stain cytoskeletal actinwith DAPI as a nuclei counterstain. Samples werewashed twice with PBS and fixed with Z-fix fixative

for 10 min at 4°C. After permeabilization with 0.5%Triton X-100 in PBS for 20 min at room tempera-ture, the samples were washed twice in PBS and stainedwith Texas Red phalloidin (1:40 dilution) and DAPI(2 ng/mL) for 45 min at room temperature. Finally,after two more PBS washes, the samples were imagedusing a fluorescence microscope (Nikon).

Quantitation of protein and DNA content

A commercially available double-stranded DNA assaykit was used (Quanti-iT PicoGreen, Invitrogen, LifeTechnologies) to quantify DNA content, which is re-flective of cell content. Microbead samples were washedin PBS and digested in 50 mmol/L Tris-HCl/4 molGuanidine-HCl solution (pH 7.5) for 2–3 hours at 4°C.Digested samples were then centrifuged at 150 g, andthe samples were diluted five times in deionized waterto reduce the guanidine concentration to <150 mmol/Lto avoid interference. The DNA content in the su-pernatant was then quantified using 1X PicoGreen inTris-HCl buffer added to the diluted samples by mea-suring fluorescence at EX:485, EM:518 nm after 5 minincubation. Purified calf thymus DNA served as thestandard for the assay.

The COL-II content of the fabricated microbeadswas verified using a commercially available total proteinassay kit (BCA Protein Assay Reagent Kit, Pierce).Microbead samples were washed three times in PBSand suspended in ice-cold PBS, then sonicated (BransonUltrasonics) for 20 seconds, with a 10-second inter-val to avoid heat denaturation. Homogenized sampleswere then diluted three times in deionized water andthe protein content was quantified according to themanufacturer’s protocol.Purified bovine serum albuminserved as the standard for the assay.

Histology

Microbeads were fixed overnight in buffered zinc for-malin (Z-Fix, Anatech Ltd.), embedded in 12-mm-diameter collagen gel (3.0 mg/mL) disks and fixed againfor 2 h. Disks were then infiltrated with paraffin usingan automated tissue processor (TP1020, LeicaBiosystems), embedded in paraffin blocks and cut into6-μm sections using a rotary microtome. To differen-tiate the acidic mucins indicative of chondrogenesis fromneutral mucins, alcian blue–periodic acid–Schiff (PAS)staining was carried out using a commercially avail-able kit (Polyscientific R&D Corp.) according to themanufacturer’s protocol.The sections were then imagedwith a bright-field microscope (Nikon).

Osteogenic, adipogenic and chondrogenic differentiation

Osteogenic differentiation medium consisted of growthmedium (α-MEM, 10% FBS) supplemented with

266 R.T. Annamalai et al.

0.2 mmol/L L-ascorbic acid 2-phosphate (Sigma),10 mmol/L β-glycerophosphate (Sigma) and100 nmol/L dexamethasone (Sigma). Osteogenic dif-ferentiation of MSC was assessed by quantifyingosteocalcin secretion and calcium deposition of func-tional osteoblasts. Osteocalcin is a protein secreted bymature osteoblasts and is a marker for osteogenesis[59]. Microbeads were washed in PBS and digestedin 0.2 N HCl overnight at 37°C and neutralized beforethe assay using 10 N NaOH. A human intact-osteocalcin enzyme-linked immunosorbent assay kit(Alpha Aesar) was used to quantify intact osteocalcinpresent in the samples. This kit is specific for intactosteocalcin (1–49 amino acid residues) and not for frag-ments of the protein and was used according to themanufacturer’s protocol. In brief, samples were addedto a pre-coated (captured monoclonal antibody) 96-well plate. Horseradish peroxidase–conjugatedstreptavidin was then added and developed (Strep-Av-HRP), and absorbance was measured at 450 nm.Purified human osteocalcin served as standards. Foranalysis of calcium deposition, microbead samples werefixed in 10% buffered zinc-formalin (Anatech) andwashed three times in PBS and twice in deionized waterbefore the assay.The samples were then stained in 1.0%alizarin red solution for 10 min prepared by dissolv-ing 1.0 g of alizarin red powder (Sigma) in 100 mLof ultrapure water (pH 4.2).The microbeads were thenwashed in deionized water twice and detained in1.0 mL of 0.5 N HCl/5% sodium dodecyl sulfate so-lution for 20 min.The destained extracted solution wasthen sampled and quantified calorimetrically by mea-suring absorbance at 415 nm.

Adipogenic medium consisted of α-MEMsupplemented with 10% FBS, 10 μg/mL humanrecombinant insulin (Sigma), 0.5 mmol/L 3-isobutyl-1-methylxanthine (IBMX) and 1 μmol/L dexa-methasone. Adipogenic differentiation of MSC wasassessed by quantifying the total lipid content of thecells. Microbead samples were fixed in 10% buff-ered zinc-formalin (Anatech) and washed three timesin PBS and once in 60% propylene glycol before theassay. The samples were then stained with warm oilred O staining solution (0.5% in propylene glycol,Sigma) for 10 min.The samples were then washed in85% propylene glycol solution for 2 min, followed bydeionized water for 2 min.The stained samples weredestained using 100% isopropanol for 15 min, and theextracted solution was sampled and quantified calo-rimetrically by measuring absorbance at 500 nm.

The chondrogenic differentiation medium con-sisted of Dulbecco’s Modified Eagle’s Medium highglucose supplemented with 1% FBS, 1% ITS PremixUniversal Culture Supplement (Corning),0.35 mmol/LL-proline (Sigma),1 mmol/L sodium pyruvate (Gibco),l glutamine (4 mmol/L, Gibco), 0.2 mmol/L L-ascorbic

acid 2-phosphate (Sigma), 100 nmol/L dexametha-sone (Sigma) and 10 ng/mL rhTGF-b1 (Peprotech).Chondrogenic differentiation of MSC was assessed byquantifying sulfated glycosaminoglycans (sGAG) de-posited by functional chondrocytes. Microbeads werewashed in PBS and digested using papain extractionsolution consisting of 20 μg/mL crystallized papain sus-pension (Corning), 0.2 mol/L sodium phosphate(Sigma),0.1 mol/L sodium acetate (Sigma),0.01 mol/Lethylenediaminetetraacetic acid (Sigma) and 5 mmol/LL-cysteine (Sigma). sGAG in the papain-digestedmicrobeads was quantified using 1,9-dimethylmethyleneblue (DMMB) as previously reported [60,61]. Briefly,25 μL of the sample was added to 200 μL of DMMB,and the absorbance of the solution was measured at525 nm.Purified chondroitin sulfate (Sigma) from sharkcartilage was used for standards. All conditions weresampled and analyzed in triplicate.

Statistical analysis

All measurements were performed at least in tripli-cate. Data are plotted as means with error barsrepresenting standard deviation. Statistical compari-sons were done using Student’s t-test with a 95%confidence limit. Differences with P < 0.05 were con-sidered statistically significant.

Results

Microbead fabrication and characterization

Microbeads made at AG/COL-II mass ratios shownin Table I were characterized 1 day after fabricationin terms of general morphology, population size dis-tribution, structural integrity and collagen content.Phase-contrast and bright-field images of represen-tative microbeads are shown in Figure 2A. Pure AGmicrobeads were essentially transparent and smooth,compared with the visibly more textured appearancethat was evident with increasing collagen content. Asimilar trend was seen when microbeads were stainedfor protein using Coomasie blue (Figure 2B), with thedegree of texture and staining increasing with COL-II content. The median microbead diameter(Figure 2C) was approximately 80 ± 10 μm, regard-less of composition; however, the variation in microbeadsize increased marginally with increasing COL-IIcontent. Microbeads made with 20 wt% COL-II werefragile but could be processed, whereas at COL-II con-centrations >20 wt% the microbeads exhibitedinsufficient integrity for further study (data not shown).Quantitative of the protein content of microbeads(Figure 2D) showed a linear increase in collagen con-centration with increasing mass of collagen used formicrobead fabrication (R2 = 0.97).

Agarose/collagen II–based microtissues for cartilage repair 267

MSC morphology and viability in AG/COL-II matrices

The morphology of MSC embedded in matrix for-mulations containing 0%, 10% and 20% COL-IIwas assessed by staining the actin cytoskeleton of

MSC entrapped in bulk gels (Figure 3A) andmicrobeads (Figure 3B). In pure AG bulk gels,MSC were highly rounded with little spreading,whereas the degree of spreading and pseudopodextension increased in the 10% and 20% COL-II

Figure 2. Morphology, collagen content, and size distribution of AG/COL-II microbeads. (A) Phase contrast and (B) bright-field imagesof acellular microbeads with varying COL-II concentration, with collagen stained with Coomassie blue in the bright-field images. (C) Di-ameter of microbeads as a function of composition. The solid center line in the box plot represents the median, the dotted center line inthe box represents the mean and the lower and upper boundaries of the box represent the 25th and 75th percentiles, respectively. Whis-kers (error bars) above and below the box indicate the 90th and 10th percentiles. (D) Quantitation of protein content of AG/COL-II microbeadsusing a total protein assay.

268 R.T. Annamalai et al.

bulk gels. A similar trend was observed in theAG/COL-II microbead format in which cells inCOL-II-containing matrices showing increased spread-ing and extension of pseudopods, which wasparticularly evident in the microbeads containing20% COL-II.

Quantitation of the DNA content of microbeads(Figure 3C) showed that those fabricated with pureAG and 10% COL-II contained the same numberof cells at fabrication. The viability of MSC embed-ded bulk gels and microbeads assayed 1 day afterfabrication is shown in Figure 3D. Cells showedgenerally high viability across all matrix formula-tions. In bulk gels, MSC viability was >85% in the0% and 10% COL-II formulations, although bulkgels containing 20% COL-II exhibited a lower initialviability around 70%. In microbeads, MSC viabilitywas consistently >75% in all formulations. Microbeadswere pre-cultured in expansion medium for a periodof 1 week before being exposed to differentiationmedia, and viability remained high during this pre-culture period.

Osteogenic differentiation of MSC in microbeads

Lineage-specific differentiation of MSC encapsu-lated in AG and AG/COL-II microbeads was assessedover time in culture. Because of the relative fragilityof microbeads with 20% COL-II content, themultilineage differentiation experiments were per-formed using only pure AG (control) microbeads andthe 10% COL-II microbead formulation, at an MSCconcentration of 0.5 × 106 cells/mL hydrogel.Microbeads were initially maintained in MSC expan-sion medium for a 1-week pre-culture period to allowacclimation post-encapsulation and were subse-quently switched to lineage-specific medium andcultured for another 21 days to induce differentia-tion. In the description of results that follow, the dayof analysis refers to the number of days of differen-tiation culture, such that “day 7” refers to the timepoint at which the microbeads have been in lineage-specific medium for 7 days.

Results from osteogenic differentiation experi-ments are shown in Figure 4.The images in panel 4A

Figure 3. Cell morphology and viability in bulk gels and microbeads. (A) Fluorescence staining of actin cytoskeleton (red) and nucleus(blue) of MSC embedded in bulk gels. (B) Fluorescence staining of actin cytoskeleton (red) and nucleus (blue) of MSC embedded inAG/COL-II microbeads with overlay of bright-field image to visualize microbead boundaries. (C) DNA content of microbeads made withpure AG or AG/10% COL-II 1 day after microbead fabrication. (D) Percentage of viable cells in bulk gels and microbeads 1 day after gelfabrication.

Agarose/collagen II–based microtissues for cartilage repair 269

show differential interference contrast (DIC) imagesof microbeads and confocal fluorescence images of cellviability using a vital stain over time in culture.Microbeads maintained their integrity throughout theculture period. MSC morphology was generally spher-ical, although evidence of moderate cell spreading canbe seen in the AG/COL-II microbeads, particularlyat the later time point. MSC viability, shown inTable II, was higher in AG/COL-II microbeads (75–85%), compared with pure AG microbeads (60–

70%) at both early and later time points. The DNAcontent in differentiating microbead populations(Figure 4B) was used to estimate cell content. In pureAG microbeads, DNA content was essentially con-stant over the differentiation culture period, althoughtended toward a decrease at later time points. In AG/COL-II microbeads, the cell content started at a lowerlevel than pure AG samples but exhibited a steady in-crease over time in culture, such that at day 21, thetotal DNA content in 10% COL-II microbeads was

Figure 4. Osteogenic differentiation of MSC in microbeads. (A) DIC images of microbeads with corresponding Live/Dead confocal flu-orescence images at day 7 (top two rows) and day 21 (bottom two rows). (B) DNA content of AG and AG/COL-II microbeads over timein culture. (C) Osteocalcin production by human MSC in AG and AG/COL-II microbeads over time in culture. (D) Calcium phosphatemineral deposition by human MSC in AG and AG/COL-II microbeads over time in culture. All images at same magnification. *P < 0.05.

Table II. Quantification of cell viability in AG and AG/COL-II microbead formulations.

Agarose Agarose/Collagen-II

Time Culture media (n = 3) Total no. of cells % Live cells % Dead cells Total no. of cells % Live cells % Dead cells

Day 7 Expansion 357 92 8 348 61 39Osteogenic 210 60 40 204 72 28Adipogenic 171 77 23 111 62 38Chondrogenic 159 60 40 174 67 33

Day 21 Expansion 245 50 50 48 69 31Osteogenic 174 67 33 270 88 12Adipogenic 306 63 37 165 80 20Chondrogenic 453 32 68 147 59 41

270 R.T. Annamalai et al.

significantly higher than that of pure AG microbeads(P < 0.05).

Osteocalcin production (Figure 4C) increased inboth pure AG and AG/COL-II microbeads by 21 daysin osteogenic culture and was significantly higher inthe populations cultured in osteogenic medium(P < 0.05) compared with the control medium.However, there was no statistically significant differ-ence between AG and AG/COL-II microbeadsmaintained under osteogenic conditions at any timepoint. In addition, when cultured in control medium,there was a significant decrease in osteocalcin pro-duction in AG/COL-II beads compared with AG beads.Calcium deposition was assessed using an alizarin redassay (Figure 4D), and followed a similar trend,showing a significant increase from day 1 to day 21in both pure AG and AG/COL-II microbeads, but nosignificant between the two matrices.The calcium de-position data were reinforced by the DIC images ofmicrobeads, which showed an accumulation of ma-terial and a more opaque appearance in microbeadscultured in osteogenic medium over time, presum-

ably due to the deposition of calcium phosphatemineral by differentiating MSC.

Adipogenic differentiation of MSC in microbeads

MSC encapsulated in pure AG and AG/COL-II werealso subjected to an adipogenic differentiation pro-tocol consisting of acclimation culture in control MSCexpansion medium for 7 days followed by 21 days ofculture in adipogenic medium. DIC images ofmicrobeads and confocal fluorescence images of vital-stained embedded cells (Figure 5A) show that themicrobeads retained their integrity over time inadipogenic culture. MSC viability in AG/COL-IImicrobeads was somewhat lower (~60%, Table II) atday 7, compared with pure AG microbeads (~80%),but by day 21, both formulations exhibited similarlyhigh cell viability (80–85%).

Measurement of DNA (Figure 5B) showed thatin pure AG microbeads, there was a significant dropin cell content from day 7 to day 21 in pure AGmicrobeads (P < 0.05), whereas AG/COL-II

Figure 5. Adipogenic differentiation of MSC in microbeads. (A) DIC images of microbeads with corresponding Live/Dead confocal flu-orescence images at day 7 (top two rows) and day 21 (bottom two rows). (B) DNA content of AG and AG/COL-II microbeads over timein culture. (C) Intracellular lipid droplet accumulation in AG and AG/COL-II microbeads over time in culture. All images at same mag-nification. *P < 0.05; **P < 0.001.

Agarose/collagen II–based microtissues for cartilage repair 271

microbeads exhibited a steady and statistically signif-icant increase in cell content from day 7 to day 21(P < 0.05). A quantitative colorimetric assay using oilred O (Figure 5C) showed a sixfold increase(P < 0.001) in intracellular lipid droplet accumula-tion in pure AG microbeads maintained in adipogenicmedium, compared with that of pure AG microbeadscultured in control medium, and a threefold in-crease (P < 0.001) compared with AG/COL-IImicrobeads maintained in adipogenic medium.However, AG/COL-II microbeads cultured inadipogenic differentiation medium did not show a sig-nificant increase in lipid content, relative to controlmedium cultures. The DIC and fluorescence imagesshow an increased cell volume in adipogenic cul-tures, which can be attributed to lipid accumulationand was more pronounced in pure AG microbeadscompared with AG/COL-II microbeads.

Chondrogenic differentiation of MSC in microbeads

Pure AG and AG/COL-II microbeads were also cul-tured in chondrogenic differentiation medium for 21

days, after an initial 7-day acclimation period in controlMSC medium. DIC images (Figure 6A) showed thatmicrobeads retained their generally spherical shape andintegrity throughout the culture period. Correspond-ing confocal fluorescence images showed cell shape,which remained generally rounded and exhibited clus-tering by day 21. MSC viability, shown by vital stainingin Figure 6A and presented quantitatively in Table II,was initially high in AG microbeads in expansionmedium and somewhat lower in lineage-specific media.In chondrogenic medium, viability in pure AGmicrobeads dropped from ~60% to ~30% over the 21-day differentiation culture period. In contrast, viabilityin AG/COL-II microbeads remained essentially con-stant at ~60–70% over time in culture.

Pure AG microbeads showed a steady decline inDNA content (Figure 6B) over time in chondro-genic culture, and cell content at day 21 wassignificantly lower than at day 0 (P < 0.05). In AG/COL-II microbeads, the initial cell content was lowerbut stayed more constant over time in culture and wasstatistically unchanged over 21 days. Deposition ofsGAG, a marker of chondrogenic phenotype

Figure 6. Chondrogenic differentiation of MSC in microbeads. (A) DIC images of microbeads with corresponding Live/Dead confocalfluorescence images at day 7 (top two rows) and day 21 (bottom two rows). (B) DNA content of AG and AG/COL-II microbeads overtime in culture. (C) sGAG production by MSC embedded in AG and AG/COL-II microbeads over time in culture. All images at samemagnification. *P < 0.05.

272 R.T. Annamalai et al.

(Figure 6C), generally increased over time in culturein both AG and AG/COL-II microbeads. Chondro-genic differentiation was particularly evident in AG/COL-II microbeads, which had statistically significantlyhigher sGAG deposition by day 14 (P < 0.05) and byday 21 showed a 60% increase over their pure AGcounterparts. Staining of histological sections withalcian blue/periodic acid–Schiff to detect polysaccha-rides (Figure 7) confirmed a rounded morphology andrelatively little matrix deposition in pure AG microbeadscultured in chondrogenic medium (Figure 7A,B). Incontrast, AG/COL-II microbeads cultured in chon-drogenic conditions (Figure 7C,D) exhibited clearlymore robust polysaccharide matrix deposition at bothdays 7 and 21.The overlaid DIC and calcein-AM flu-orescence images in Figure 7 also show that thedifferentiating MSC maintained a rounded morphol-ogy, with visible matrix deposition around the cells.

Discussion

Creation of cellular microenvironments that mimic car-tilage tissue has been approached in a variety of ways[4,5,8,13,24]. In the present study, our goal was to

fabricate and characterize modular microtissues thatmimic both the proteoglycan content and the proteincontent of native cartilage. To this end, we createdmicrobeads consisting of various mass ratios of thepolysaccharide AG and the protein COL-II. AG hasbeen used previously as a matrix in cartilage tissue en-gineering. Although this polysaccharide does notrecapitulate all of the properties of cartilageproteoglycans, it has been shown to have relevance indirecting cell phenotype [53–55,58].The protein COL-II is highly associated with hyaline cartilage but hasnot been used widely as a scaffold material becauseof its relative scarcity. By creating microscalepolysaccharide-protein tissue constructs, we aimed toharness the mechanical robustness of agarose gels, whilealso augmenting the matrix with a tissue-specificprotein isoform.

MSC were chosen as a cell source because of theirdemonstrated ability to differentiate into specific tissuelineages, including orthopaedic tissues, under the ap-propriate stimulation [27–33]. A major advantage ofusing MSC as a cell source in cartilage tissueengineering is that they can be expanded to clinical-ly significant numbers without loss of their

Figure 7. Histological evaluation of microbeads in chondrogenic medium at day 7 and day 21. First three columns show representativemicrobeads stained with alcian blue/periodic acid–Schiff (PAS), and fourth column shows cells in microbeads stained with calcein-AMwith overlay of DIC image of microbeads. (A and B) pure AG beads. (C and D) AG/10% COL-II beads.

Agarose/collagen II–based microtissues for cartilage repair 273

undifferentiated phenotype [62]. In the present study,we used a commercially available source of MSC, andfuture work will need to address variability betweencells from different donors and their capacity for dif-ferentiation. Well-characterized and expanded cellpopulations can then be differentiated toward thechondrocytic phenotype for use in cell-based thera-pies for cartilage repair. In contrast, removal ofchondrocytes from their native environment and ex-pansion in monolayers can cause dedifferentiation andloss of function [43–45]. Partial function may be re-covered by transferring cultured chondrocytes to athree-dimensional environment, but the recovery is in-complete [51,52]. Donor site morbidity also limits theuse of autologous chondrocytes because any defect tohealthy cartilage may further degenerate the surround-ing tissue [63]. A consistent progenitor cell source andmethods for reliably differentiating and delivering suchcells is therefore an important need for cell-basedtherapies.

The system we used to fabricate polysaccharide-protein microbeads is an adaptation of a methodthat we have used for a variety of modular tissueengineering applications [64–68]. This water-in-oilemulsion process is easy to implement and createslarge batches of essentially identical microbeads in ashort period of time. The average size and sizedistribution of the microbeads can be tailored bycontrolling the impeller geometry, the impeller-to-vessel diameter ratio, and the viscosity of the dispersedphases [69,70], and the process can be scaled tocreate smaller (<1 mL) or larger (>10 mL) volumesof microbeads. Emulsification produces microbeadpopulations with a distribution of sizes, which canbe an advantage when creating packed beds ofmicrobeads, but it is difficult to tightly controlmicrobead size. The composition of the microbeadsis determined by controlling the type and amount ofconstituents that make up the aqueous phase, includ-ing the cell type. A variety of gelable materials andrelevant cell types have been used in this system,and they are being investigated for advancedbiomaterial-based cell delivery. More broadly, modularapproaches to creating engineered tissue are emerg-ing as a promising approach to creating complexstructures [71–76]. Such methods have advantagesover conventional scaffold-based techniques [64,72],and new methods are being developed at a rapidrate.

In the present study, we created hydrogelmicrobeads by mixing AG and COL-II in specific massratios. MSC were embedded directly in the microbeadmatrix at the time of gelation by suspending them inthe matrix formulation used to create the microbeads.All the AG/COL-II hydrogel formulations we testedemulsified consistently; however, retrieval of the formed

microbeads from the oil phase became challenging asthe COL-II mass concentration was increased.Microbeads made with ≥20% COL-II were fragile andbroke apart during collection and therefore were notsuitable for long-term cultures. In contrast, microbeadsfabricated with <20% COL-II were more robust andstayed intact over time in culture.The relatively stronggel formed by the AG component aided in protect-ing microbeads from compressive forces during theemulsification and collection processes because in apolysaccharide-protein co-gel such as AG/COL-II, thenon-fibrillar component significantly enhances the me-chanical properties in compression [77,78].The medianmicrobead diameter in the present work was80 ± 10 μm in all formulations, but the consistencyof microbead size also decreased with an increase inCOL-II content, presumably due to the increase inviscosity caused by the addition of a macromolecu-lar protein component. The decrease in microbeadintegrity and increase in size heterogeneity in formu-lations ≥20% COL-II led us to focus on microbeadswith 10% COL-II in this study, which were com-pared with pure AG controls.

MSC were shown to remain viable upon encap-sulation in AG/COL-II microbeads, with only minorloss of cell viability 24 h after encapsulation, and main-tenance of high cell viability over 7 days when culturedin standard MSC expansion medium. By day 7, cellviability in all hydrogel formulations was also signifi-cantly higher in 3D microbeads compared with bulkgels made at the same matrix composition.This effectmay be attributed to the increased access to oxygenand other nutrients that microbeads can provide dueto their small size and associated short diffusion dis-tances [79]. Increasing the COL-II concentration inmicrobeads resulted in a noticeable increase in cellspreading within the matrix, presumably as a conse-quence of integrin-mediated cell attachment to theprotein component. While COL-II has been associ-ated with increased chondrogenic differentiation ofMSC [80,81], it has also been suggested that a non-spherical morphology can result in dedifferentiationof chondrocytes [62]. Therefore, it is important tobalance the effects of added extracellular matrix com-ponents in microbead formulations.

Our main interest in this study was the ability ofmicrobeads to support desired tissue-specific differ-entiation. We therefore compared the multilineagepotential of MSC embedded in AG/COL-II (10%)microbeads and pure AG microbeads over 3 weeks ofculture. Osteogenic differentiation was modestly sup-ported by both pure AG and AG/COL-II formulations,as evidenced by an increase in osteocalcin expres-sion and calcium deposition after 21 days in osteogenicmedium. However, there was no difference betweenthe matrix formulations in their ability to promote

274 R.T. Annamalai et al.

osteogenesis in vitro. Pure AG microbeads culturedin adipogenic medium showed robust lipidaccumulation after 21 days, whereas AG/COL-IImicrobeads produced only low levels of lipid even whenexposed to adipogenic stimuli. In contrast, chondro-genesis was significantly more robust in AG/COL-IImicrobeads, compared with pure AG, and strong ex-pression of sGAG was measured by days 14 and 21in chondrogenic culture. Deposition of sGAG was con-firmed using histology, which showed rounded cellsand more matrix deposition in the COL-II-containingmicrobeads. These data suggest that the inclusion ofCOL-II in agarose microbeads can potentiate differ-entiation toward the chondrogenic lineage, whileinhibiting adipogenic differentiation. The mecha-nism of this effect is not clear from our experiments;however, binding of MSC to specific proteins has beenshown to modulate differentiation [82,83]. In addi-tion, it has been suggested that cryptic peptidesequences on the COL-II molecule can affect chon-drocyte phenotype [1].

In summary, we have demonstrated a method forcreating modular microtissues designed to mimic spe-cific proteoglycan and protein components of thecartilage extracellular matrix. The resulting MSC-laden microbeads are robust, support cell viability andcan be maintained in culture. Furthermore, our datashow that the composition of the three-dimensionmatrix in the microbeads affects the lineage-specificdifferentiation of embedded MSC. In particular, COL-II was shown to preferentially promote thechondrogenic phenotype, while suppressing adipogenicdifferentiation.The microbead format has the advan-tage that diffusion path lengths are short, andmicrobead preparations can be delivered in a mini-mally invasive manner.The composition of microbeadscan also be varied to include other cell and matrix typesand to incorporate biochemical factors that promotethe desired regenerative effects. Cartilage defects area serious and growing problem, and cell-based thera-pies have started to have an impact on their treatment.The development of biomaterial-based approaches thatimprove cell differentiation, delivery and function willenable the next generation of cell-based cartilagetherapies.

Acknowledgments

Research reported in this publication was supportedin part by the National Institute of Arthritis andMusculoskeletal and Skin Diseases of the NationalInstitutes of Health under award numbersR21AR062709 and R01AR062636. The content issolely the responsibility of the authors and does notnecessarily represent the official views of the Nation-al Institutes of Health.

Disclosure of interest: The authors have no com-mercial, proprietary, or financial interest in the productsor companies described in this article.

References

[1] Hunziker EB. Articular cartilage repair: basic science andclinical progress. A review of the current status and prospects.Osteoarthritis Cartilage 2002;10(6):432–63.

[2] Hunziker EB, et al. An educational review of cartilage repair:precepts & practice—myths & misconceptions—progress &prospects. Osteoarthritis Cartilage 2015;23(3):334–50.

[3] Huey DJ, Hu JC, Athanasiou KA. Unlike bone, cartilageregeneration remains elusive. Science 2012;338(6109):917–21.

[4] Moran CJ, et al. Restoration of articular cartilage. J Bone JointSurg Am 2014;96(4):336–44.

[5] Mobasheri A, et al. Chondrocyte and mesenchymal stemcell-based therapies for cartilage repair in osteoarthritis andrelated orthopaedic conditions. Maturitas 2014;78(3):188–98.

[6] Diekman BO, Guilak F. Stem cell-based therapies forosteoarthritis: challenges and opportunities. Curr OpinRheumatol 2013;25(1):119–26.

[7] Westreich R, et al. Validating the subcutaneous model ofinjectable autologous cartilage using a fibrin glue scaffold.Laryngoscope 2004;114(12):2154–60.

[8] Crawford DC, et al. An autologous cartilage tissue implantNeoCart for treatment of grade III chondral injury to the distalfemur: prospective clinical safety trial at 2 years. Am J SportsMed 2009;37(7):1334–43.

[9] Medved F, et al. Severe posttraumatic radiocarpal cartilagedamage: first report of autologous chondrocyte implantation.Arch Orthop Trauma Surg 2013;133(10):1469–75.

[10] Crawford DC, DeBerardino TM, Williams RJ 3rd. NeoCart,an autologous cartilage tissue implant, compared withmicrofracture for treatment of distal femoral cartilage lesions:an FDA phase-II prospective, randomized clinical trial aftertwo years. J Bone Joint Surg Am 2012;94(11):979–89.

[11] Peterson L, et al. Two- to 9-year outcome after autologouschondrocyte transplantation of the knee. Clin Orthop RelatRes 2000;374:212–34.

[12] Niemeyer P, et al. Characteristic complications after autologouschondrocyte implantation for cartilage defects of the knee joint.Am J Sports Med 2008;36(11):2091–9.

[13] Makris EA, et al. Repair and tissue engineering techniquesfor articular cartilage. Nat Rev Rheumatol 2015;11(1):21–34.

[14] Musumeci G, et al. New perspectives for articular cartilagerepair treatment through tissue engineering: a contemporaryreview. World J Orthop 2014;5(2):80–8.

[15] Tuan RS. A second-generation autologous chondrocyteimplantation approach to the treatment of focal articularcartilage defects. Arthritis Res Ther 2007;9(5):109.

[16] Harris JD, et al. Failures, re-operations, and complicationsafter autologous chondrocyte implantation—a systematicreview. Osteoarthritis Cartilage 2011;19(7):779–91.

[17] Bornes TD, Adesida AB, Jomha NM. Mesenchymal stem cellsin the treatment of traumatic articular cartilage defects: acomprehensive review. Arthritis Res Ther 2014;16(5):432.

[18] Radisic M, et al. Cardiac tissue engineering using perfusionbioreactor systems. Nat Protoc 2008;3(4):719–38.

[19] Chen M, et al. Ectopic osteogenesis of macroscopic tissueconstructs assembled from human mesenchymal stem cell-laden microcarriers through in vitro perfusion culture. PLoSONE 2014;9(10):e109214.

[20] Pu F, et al. The use of flow perfusion culture andsubcutaneous implantation with fibroblast-seeded PLLA-

Agarose/collagen II–based microtissues for cartilage repair 275

collagen 3D scaffolds for abdominal wall repair. Biomaterials2010;31(15):4330–40.

[21] Wang Y, et al. Mesenchymal stem cells for treating articularcartilage defects and osteoarthritis. Cell Transplant2015;24:1661–78.

[22] Bian L, et al. Coculture of human mesenchymal stem cellsand articular chondrocytes reduces hypertrophy and enhancesfunctional properties of engineered cartilage. Tissue Eng PartA 2011;17(7–8):1137–45.

[23] Toh WS, et al. Advances in mesenchymal stem cell-basedstrategies for cartilage repair and regeneration. Stem Cell Rev2014;10(5):686–96.

[24] Gopal K, Amirhamed HA, Kamarul T. Advances of humanbone marrow-derived mesenchymal stem cells in the treatmentof cartilage defects: a systematic review. Exp Biol Med(Maywood) 2014;239(6):663–9.

[25] Pittenger MF, et al. Multilineage potential of adult humanmesenchymal stem cells. Science 1999;284(5411):143–7.

[26] Chamberlain G, et al. Concise review: mesenchymal stem cells:their phenotype, differentiation capacity, immunologicalfeatures, and potential for homing. Stem Cells2007;25(11):2739–49.

[27] Kafienah W, et al. Three-dimensional cartilage tissueengineering using adult stem cells from osteoarthritis patients.Arthritis Rheum 2007;56(1):177–87.

[28] Ko J-Y, et al. In vitro chondrogenesis and in vivo repair ofosteochondral defect with human induced pluripotent stemcells. Biomaterials 2014;35(11):3571–81.

[29] Wakitani S, et al. Human autologous culture expanded bonemarrow mesenchymal cell transplantation for repair of cartilagedefects in osteoarthritic knees. Osteoarthritis Cartilage2002;10(3):199–206.

[30] Kuroda R, et al. Treatment of a full-thickness articular cartilagedefect in the femoral condyle of an athlete with autologousbone-marrow stromal cells. Osteoarthritis Cartilage2007;15(2):226–31.

[31] Grayson WL, et al. Stromal cells and stem cells in clinical boneregeneration. Nat Rev Endocrinol 2015;11(3):140–50.

[32] Dupont KM, et al. Human stem cell delivery for treatmentof large segmental bone defects. PNAS 2010;107(8):3305–10.

[33] Kon E, et al. Autologous bone marrow stromal cells loadedonto porous hydroxyapatite ceramic accelerate bone repair incritical-size defects of sheep long bones. J Biomed Mater Res2000;49(3):328–37.

[34] Wu L, et al. Trophic effects of mesenchymal stem cells inchondrocyte co-cultures are independent of culture conditionsand cell sources. Tissue Eng Part A 2012;18(15–16):1542–51.

[35] Caplan AI. Adult mesenchymal stem cells for tissueengineering versus regenerative medicine. J Cell Physiol2007;213(2):341–7.

[36] Poole AR, et al. Composition and structure of articularcartilage: a template for tissue repair. Clin Orthop Relat Res2001;391(Suppl.):S26–33.

[37] Buckwalter JA, Mankin HJ. Articular cartilage: tissue designand chondrocyte-matrix interactions. Instr Course Lect1998;47:477–86.

[38] Buckwalter JA, Mankin HJ. Instructional course lectures, theAmerican Academy of Orthopaedic Surgeons—articularcartilage. Part I: tissue design and chondrocyte-matrixinteractions. J Bone Joint Surg Am 1997;79:600–11.

[39] Sophia Fox AJ, Bedi A, Rodeo SA. The basic science ofarticular cartilage: structure, composition, and function. SportsHealth 2009;1(6):461–8.

[40] Kempson GE, et al. Correlations between stiffness and thechemical constituents of cartilage on the human femoral head.Biochim Biophys Acta (BBA) Gen Subj 1970;215(1):70–7.

[41] Stockwell RA. The cell density of human articular and costalcartilage. J Anat 1967;101(Pt 4):753–63.

[42] Hunziker EB, Quinn TM, Häuselmann HJ. Quantitativestructural organization of normal adult human articularcartilage. Osteoarthritis Cartilage 2002;10(7):564–72.

[43] Benya PD, Padilla SR, Nimni ME. Independent regulationof collagen types by chondrocytes during the loss ofdifferentiated function in culture. Cell 1978;15(4):1313–21.

[44] Von Der Mark K, et al. Relationship between cell shape andtype of collagen synthesised as chondrocytes lose their cartilagephenotype in culture. Nature 1977;267(5611):531–2.

[45] Mayne R, et al. Changes in type of collagen synthesized asclones of chick chondrocytes grow and eventually lose divisioncapacity. Proc Natl Acad Sci U S A 1976;73(5):1674–8.

[46] Batorsky A, et al. Encapsulation of adult human mesenchymalstem cells within collagen-agarose microenvironments.Biotechnol Bioeng 2005;92(4):492–500.

[47] Hunziker EB. Articular cartilage repair: are the intrinsicbiological constraints undermining this process insuperable?Osteoarthritis Cartilage 1999;7(1):15–28.

[48] Nehrer S, et al. Canine chondrocytes seeded in type I andtype II collagen implants investigated in vitro. J Biomed MaterRes 1997;38(2):95–104.

[49] Buma P, et al. Cross-linked type I and type II collagenousmatrices for the repair of full-thickness articular cartilagedefects—a study in rabbits. Biomaterials 2003;24(19):3255–63.

[50] Veilleux NH, Yannas IV, Spector M. Effect of passage numberand collagen type on the proliferative, biosynthetic, andcontractile activity of adult canine articular chondrocytes intype I and II collagen-glycosaminoglycan matrices in vitro.Tissue Eng 2004;10(1–2):119–27.

[51] Benya PD, Shaffer JD. Dedifferentiated chondrocytes reexpressthe differentiated collagen phenotype when cultured in agarosegels. Cell 1982;30(1):215–24.

[52] Benya PD. Modulation and reexpression of the chondrocytephenotype; mediation by cell shape and microfilamentmodification. Pathol Immunopathol Res 1988;7(1–2):51–4.

[53] Chang J, Poole CA. Sequestration of type VI collagen in thepericellular microenvironment of adult chondrocytes culturedin agarose. Osteoarthritis Cartilage 1996;4(4):275–85.

[54] DiMicco MA, et al. Structure of pericellular matrix aroundagarose-embedded chondrocytes. Osteoarthritis Cartilage2007;15(10):1207–16.

[55] Emans PJ, et al. Autologous engineering of cartilage. PNAS2010;107(8):3418–23.

[56] Fernandez-Cossio S, et al. Biocompatibility of agarose gel asa dermal filler: Histologic evaluation of subcutaneous implants.Plast Reconstr Surg 2007;120(5):1161–9.

[57] Selmi TAS, et al. Autologous chondrocyte implantation in anovel alginate-agarose hydrogel—outcome at two years. J BoneJoint Surg-Br Vol 2008;90B(5):597–604.

[58] Mauck RL, et al. Functional tissue engineering of articularcartilage through dynamic loading of chondrocyte-seededagarose gels. J Biomech Eng 2000;122(3):252–60.

[59] Nakamura A, et al. Osteocalcin secretion as an early markerof in vitro osteogenic differentiation of rat mesenchymal stemcells. Tissue Engineering Part C-Methods 2009;15(2):169–80.

[60] Wise JK, et al. Comparison of uncultured marrowmononuclear cells and culture-expanded mesenchymal stemcells in 3D collagen-chitosan microbeads for orthopedic tissueengineering. Tissue Eng Part A 2014;20(1–2):210–24.

[61] Farndale RW, Buttle DJ, Barrett AJ. Improved quantitationand discrimination of sulphated glycosaminoglycans by useof dimethylmethylene blue. Biochim Biophys Acta1986;883(2):173–7.

276 R.T. Annamalai et al.

[62] Gomoll AH, et al. The subchondral bone in articularcartilage repair: current problems in the surgical management.Knee Surg Sports Traumatol Arthrosc 2010;18(4):434–47.

[63] Lee CR, et al. Effects of harvest and selected cartilage repairprocedures on the physical and biochemical properties ofarticular cartilage in the canine knee. J Orthop Res2000;18(5):790–9.

[64] Lund AW, et al. Osteogenic differentiation of mesenchymalstem cells in defined protein beads. J Biomed Mater Res BAppl Biomater 2008;87(1):213–21.

[65] Wang L, Rao RR, Stegemann JP. Delivery of mesenchymalstem cells in chitosan/collagen microbeads for orthopedic tissuerepair. Cells Tissues Organs 2013;197(5):333–43.

[66] Rao RR, et al. Dual-phase osteogenic and vasculogenicengineered tissue for bone formation. Tissue Eng Part A2015;21(3–4):530–40.

[67] Peterson AW, et al. Vasculogenesis and angiogenesis inmodular collagen-fibrin microtissues. Biomaterials Science2014;2(10):1497–508.

[68] Chen Z, Wang L, Stegemann JP. Phase-separated chitosan–fibrin microbeads for cell delivery. J Microencapsul2011;28(5):344–52.

[69] Chen HT, Middleman S. Drop size distribution in agitatedliquid-liquid systems. AIChE J 1967;13(5):989–95.

[70] Calabrese RV, et al. Measurement and analysis of drop sizein a batch rotor-stator mixer. In: van dan Akker HEA, DerksenJJ, editors. Proceedings of the 10th European Conference onMixing, Delft, the Netherlands. 2000.

[71] McGuigan AP, Sefton MV. Vascularized organoid engineeredby modular assembly enables blood perfusion. Proc Natl AcadSci U S A 2006;103(31):11461–6.

[72] Tiruvannamalai-Annamalai R, Armant DR, Matthew HW.A glycosaminoglycan based, modular tissue scaffold systemfor rapid assembly of perfusable, high cell density, engineeredtissues. PLoS ONE 2014;9(1):e84287.

[73] Nichol JW, Khademhosseini A. Modular tissue engineering:engineering biological tissues from the bottom up. Soft Matter2009;5(7):1312–19.

[74] Rao RR, Stegemann JP. Cell-based approaches to theengineering of vascularized bone tissue. Cytotherapy2013;15(11):doi:10.1016/j.jcyt.2013.06.005.

[75] Park IS, Rhie J-W, Kim S-H. A novel three-dimensional adipose-derived stem cell cluster for vascularregeneration in ischemic tissue. Cytotherapy 2014;16(4):508–22.

[76] Kittaka M, et al. Clumps of a mesenchymal stromal cell/extracellular matrix complex can be a novel tissue engineeringtherapy for bone regeneration. Cytotherapy 2015;17(7):860–73.

[77] Lake SP, Hald ES, Barocas VH. Collagen-agarose co-gels asa model for collagen–matrix interaction in soft tissues subjectedto indentation. J Biomed Mater Res A 2011;99A(4):507–15.

[78] Lake S, et al. Mechanics of a fiber network within a non-fibrillar matrix: model and comparison with collagen-agaroseco-gels. Ann Biomed Eng 2012;40(10):2111–21.

[79] Tanaka H, Matsumura M, Veliky IA. Diffusion characteristicsof substrates in Ca-alginate gel beads. Biotechnol Bioeng1984;26(1):53–8.

[80] Bosnakovski D, et al. Chondrogenic differentiation of bovinebone marrow mesenchymal stem cells (MSCs) in differenthydrogels: influence of collagen type II extracellular matrixon MSC chondrogenesis. Biotechnol Bioeng 2006;93(6):1152–63.

[81] Muhonen V, et al. Recombinant human type II collagenhydrogel provides a xeno-free 3D micro-environment forchondrogenesis of human bone marrow-derived mesenchymalstromal cells. J Tissue Eng Regen Med 2015. Epub ahead ofprint (http://onlinelibrary.wiley.com/doi/10.1002/term.1983/epdf).

[82] Wang Y, Sul HS. Pref-1 regulates mesenchymal cellcommitment and differentiation through Sox9. Cell Metab2009;9(3):287–302.

[83] Warzecha J, et al. Sonic hedgehog protein promotesproliferation and chondrogenic differentiation of bone marrow-derived mesenchymal stem cells in vitro. J Orthop Sci2006;11(5):491–6.

Agarose/collagen II–based microtissues for cartilage repair 277