Biofabrication 9 (2017) 034105 https://doi.org/10.1088/1758-5090/aa7fdd

PAPER

Bioprinting of a functional vascularizedmouse thyroid glandconstruct

ElenaABulanova1,6 , ElizavetaVKoudan1,6, JonathanDegosserie2, CharlotteHeymans2,FredericoDASPereira1, VladislavAParfenov1, Yi Sun3, QiWang3, SurayaAAkhmedova4,IrinaK Sviridova4, Natalia S Sergeeva4, GeorgyAFrank5, YusefDKhesuani1, Christophe EPierreux2 andVladimirAMironov1

1 Laboratory for Biotechnological Research ‘3DBioprinting Solutions’,Moscow 115409, Russia2 deDuve Institute, Université Catholique de Louvain (UCL), B-1200 Brussels, Belgium3 Department ofMathematics and InterdisciplinaryMathematics Institute, University of SouthCarolina, Columbia, SC 29208,United

States of America4 NationalMedical ResearchRadiological Centre of theMinistry ofHealth of the Russian Federation,Moscow 125284, Russia5 RussianMedical Academy of Postgraduate Education Studies,Moscow 125993, Russia6 Both authors contributed equally to this paper.

E-mail: vladimir.mironov54@gmail.com

Keywords: 3Dbioprinting,mouse thyroid gland, tissue spheroids, vascularization

Supplementarymaterial for this article is available online

AbstractBioprinting can be defined as additive biofabrication of three-dimensional (3D) tissues and organconstructs using tissue spheroids, capable of self-assembly, as building blocks. The thyroid gland, arelatively simple endocrine organ, is suitable for testing the proposed bioprinting technology. Herewereport the bioprinting of a functional vascularizedmouse thyroid gland construct from embryonictissue spheroids as a proof of concept. Based on the self-assembly principle, we generated thyroidtissue starting from thyroid spheroids (TS) and allantoic spheroids (AS) as a source of thyrocytes andendothelial cells (EC), respectively. Inspired bymathematicalmodeling of spheroid fusion, we used anoriginal 3Dbioprinter to print TS in close associationwithASwithin a collagen hydrogel. During theculture, closely placed embryonic tissue spheroids fused into a single integral construct, EC fromASinvaded and vascularized TS, and epithelial cells from theTS progressively formed follicles. In thisexperimental setting, we observed formation of a capillary network around follicular cells, as observedduring in utero thyroid development when thyroid epithelium controls the recruitment, invasion andexpansion of EC around follicles. To prove that EC fromAS are responsible for vascularization of thethyroid gland construct, we depleted endogenous EC fromTS before bioprinting. EC fromAScompletely revascularized depleted thyroid tissue. The cultured bioprinted construct was functionalas it could normalize blood thyroxine levels and body temperature after grafting under the kidneycapsule of hypothyroidmice. Bioprinting of functional vascularizedmouse thyroid gland constructrepresents a further advance in bioprinting technology, exploring the self-assembling properties oftissue spheroids.

Introduction

The thyroid gland is an endocrine organ required forthe production of hormones such as thyroxine (T4)and triiodothyronine (T3). These are essential fornormal growth, neurological development and home-ostasis. Hypofunctionality of the thyroid gland or itscomplete removal due to thyroid cancer requires

compensation for the lost function [1]. Hypothyroid-ism is usually treated with synthetic hormone replace-ment therapy, which is a lifelong treatment [2].Although hormone replacement therapy does providea certain level of thyroid hormones it almost comple-tely excludes fine regulation of thyroid status accord-ing to current physiological conditions. The use ofautologous thyrocytes and thyroid follicles derived

RECEIVED

16April 2017

REVISED

22 June 2017

ACCEPTED FOR PUBLICATION

14 July 2017

PUBLISHED

18August 2017

© 2017 IOPPublishing Ltd

https://doi.org/10.1088/1758-5090/aa7fddhttps://orcid.org/0000-0001-8460-3113https://orcid.org/0000-0001-8460-3113mailto:vladimir.mironov54@gmail.comhttps://doi.org/10.1088/1758-5090/aa7fddhttp://crossmark.crossref.org/dialog/?doi=10.1088/1758-5090/aa7fdd&domain=pdf&date_stamp=2017-08-18http://crossmark.crossref.org/dialog/?doi=10.1088/1758-5090/aa7fdd&domain=pdf&date_stamp=2017-08-18

from induced pluripotent stem cells opens up theunique possibility of developing a new therapeuticmodality to compensate for loss of thyroid functionand eliminate the subsequent lifelong dependence ofpatients on treatment with synthetic hormones [3].

The recent advances in rapidly emerging three-dimensional (3D) bioprinting technology will even-tually enable the biofabrication of functional and vas-cularized tissue from autologous human cells suitablefor clinical transplantation [4]. The original concept ofrobotic layer-by-layer biofabrication of 3D human tis-sues and organs according to a digital model, using tis-sue spheroids as building blocks, was proposed morethan a decade ago and has been gradually improved[5–8]. Tissue fusion is a recurrent process duringembryonic development [9] and a fundamental bio-mimetic principle of the proposed 3D bioprintingtechnology. The fusion of closely placed tissue spher-oids enables rapid post-printing self-assembly of the3D tissue construct.

The anatomical and histological structure of thethyroid gland is relatively simple compared with morecomplex organs such as the lung, liver or kidney due tothe absence of a ductal system. The thyroid gland isthus an attractive target for testing the feasibility ofbioprinting technology. The structural and functionalunit of the thyroid gland consists of independent thyr-oid follicles composed of a polarized epithelium sur-rounded by fenestrated blood capillaries, therebyenabling import of iodine and delivery of thyroid hor-mones directly into the blood circulation. Branchedsegments of thyroid arteries and veins are connectedwith microvessels of multiple angio-follicular unitsand form the vascularized thyroid gland [10]. More-over, it has been demonstrated recently that thyroidfolliculogenesis is closely associated with and guidedby angiogenesis [11], thus indicating that during thyr-oid organogenesis the level of vascularization regulatesfunctional differentiation of thyroid follicles.

Vascularization, i.e. the development of a micro-vascular network, remains a big challenge and a majortechnological impediment in bioprinting. The micro-vascular network within an engineered tissue shouldmaintain high cell viability and function, providingthe requisite nutrients and oxygen. Without vascu-lature the size of a bioprinted construct is limited to200 μm, the diffusion limit of oxygen [12]. Differentapproaches exist for engineering vascularized tissues.The classical seeding of sacrificial scaffolds with endo-thelial cells (EC) supported by angiogenic factors hasshowed promising results but is limited to construc-tion of simple tissues only [13, 14]. Theoreticalresearch based on mathematical modeling and comp-uter simulation [15], along with a growing body ofexperimental evidence [16], strongly indicates that it ispossible to bioprint 3D tissue and organ constructswith a built-in intra-organ branched vascular treeusing different types of self-assembling vascular tissuespheroids as building blocks. It has been shown that:

(i) solid vascular tissue spheroids could be effectivelyused for biofabrication of large-diameter extraorganand intra-organ arterial and venous blood vessels [6];(ii) lumenized vascular tissue spheroids could be usedfor biofabrication of arteriolar and venular vascularsegments with intermediate diameter [9, 17]; and,finally, (iii) prevascularized organospecifc tissuespheroids or adjacently placed endothelial tissuespheroids could be used for bioprinting of tissue con-structs with a well-developed microvascular network[18]. We propose a solid scaffold-free approach inwhich the ‘self-assembling spheroid’ strategy enablesgeneration of a complex bioprinted tissue constructwith a built-in vascular system consisting of hier-archically branched blood vessels.

In our previous paper [19] we demonstrated thattissue spheroids when closely placed in fusion-permis-sive collagen hydrogel can coalesce and form a toroid-like construct, exactly as predicted by mathematicalmodeling and computer simulation. Here we reportthe bioprinting of a functional and vascularizedmousethyroid gland construct using two types of tissuespheroids, thyroid spheroids (TS) and allantoic spher-oids (AS), obtained from embryonic thyroid and allan-toic tissues, respectively. We demonstrate that ECfrom AS are endowed with the capacity to completelyvascularize thyroid explants, even when the endothe-lial pool of TS is ablated by selective inhibition of vas-cular endothelial growth factor receptor 2 (VEGFR2).More importantly, we confirmed the functionality ofbioprinted tissue construct in radioiodine-ablatedthyroid (hypothyroidmice).

Materials andmethods

Reagents andmaterialsThe following reagents and materials were used:Hank’s Balanced Salt Solution (Paneco, cat. no. R020),tungsten wire (Goodfellow, cat. no. W 005160), anti-E-cadherin (BD Biosciences, cat. no. 610182, 1:1000),anti-Ezrin (Thermo Scientific, cat. no. MS-661-P1,1:400), anti-PECAM (BD Biosciences, cat. no. 550274,1:100), anti-caspase-3a (Cell Signaling, cat. no. 9661),anti-pHH3 (Cell Signaling, cat. no. 9701), anti-TTF1(Dako, cat. no. M3575), Hoechst 33258 (Sigma, cat.no. B2883), Alexa Fluor 488 goat anti-mouse IgG1(γ1) antibody (Thermo Scientific, cat. no. A-21121),Alexa Fluor 647 goat anti-mouse IgG2a (γ2a) (ThermoScientific, cat. no. A-21241), Alexa Fluor 647 chickenanti-rat IgG (H+L) (Thermo Scientific, cat. no.A-21472), sodium hydroxide (Merck, cat. no.106498), sodium bicarbonate (Paneco, cat. no. F022),phosphate-buffered saline (PBS; Gibco, cat. no.18912-014), Corning Transwell®-COL collagen-coated polytetrafluoroethylene (PTFE) membraneinserts (Corning, cat. no. 3491, 0.4 μm), gelatin (Fluka,cat. no. 48723-500G-F), sucrose (Sigma, cat. no.

2

Biofabrication 9 (2017) 034105 EABulanova et al

S0389), SU5416 (Cayman, cat. no. 13342), 100 mmplastic cell culture dish (Corning, cat. no. 430167).

Complete culturemediumThe complete culture medium comprised Dulbecco’smodified Eagle’s medium (DMEM; Gibco, cat. no.12491-015) containing 10% fetal bovine serum (FBS;Gibco, cat. no. 16000-044) supplemented with anantibiotic/antimycotic mix (Gibco, cat. no. 15240-062) and 1 mML-glutamine (Paneco, cat. no. F032).

Collagen isolation from rat tails and collagen gelpreparationRat tail type I collagen was prepared according to themethod described by Rajan et al [20]. The establishedconcentration of collagen in solution was3.12 mgml−1. For preparation of collagen gel 890 μlof collagen solution was mixed with 60 μl of 1 Msodiumhydroxide, 250 μl of 7.5% sodiumbicarbonateand 300 μl of PBS.

Generation of spheroidsThyroid explants were microdissected from e14.5mouse embryos as described by Delmarcelle et al [21].Allantoic tissue was microdissected from e8.5 mouseembryos as described by Drake and Fleming [22].Dissection was not performed under a sterile atmos-phere, so explants and allantoides were washed severaltimes in sterile culture medium. Individual thyroidexplants and allantoides were then suspended in 20 μlof complete culture medium and spotted onto theunderside of a lid of a 100 mm plastic cell Petri dish.The lids were then inverted and placed onto culturedishes to create hanging drops. SU5416 (VEGFR2kinase inhibitor) was dissolved in dimethyl sulfoxide(DMSO; 20 mM stock solution) and added at 5 μM tothe culture medium for thyroid explants. Controlthyroid explants were exposed to the same concentra-tion of vehicle as the test samples. The hanging dropcultures were placed in a humidified atmosphere andcultured at 37 °C, 5% CO2 for 18–24 h. After 18–24 hof culture, the resulting spheroidal thyroid explantsand allantoides were used for bioprinting of mousethyroid gland constructs.

Kinetics of spheroid fusionFusion angles between adjacent spheroids were mea-sured as described by Susienka et al [23]. Briefly, fusionangles were measured manually and plotted as afunction of time using GraphPad Prism software(GraphPad Software, Inc., La Jolla, CA,USA).

Bioprinting processTo print the thyroid mouse gland construct within thecollagen hydrogel spread on a transparent PTFEmembrane we used a multifunctional Fabion 3Dbioprinter with the turnstile system (see figures 3(A)–(C)). This device consists of the following functional

systems: syringe pump; fluidic chip; loading piston(d); trapping piston (e); depositing piston (f) withgroove for liquid removal (g) (figure 3(B)). The syringeloaded with spheroid suspension was installed into thesyringe pump and thus connected to printing head. Assoon as the spheroid suspension entered the printinghead, the automation system switched the syringepump on and then switched it off at themoment whenspheroids came into contact with trapping piston. Thedepositing piston (f)moved upward, closing the liquidremoval outlet (g) and freeing the channel for spheroidpassage; the trapping piston (e) slides downward,opening the passage for spheroids. Then loadingpiston (d) immediatelymoved the spheroids rightwardand under the depositing piston (f). The depositingpiston transported spheroids downward until contactwith the printing surface and then returned to itsinitial position, thus completing one bioprinting cycle.The depositing piston features a groove slightly abovethe base (g) providing a channel for the draining ofliquid. This resource is applied to minimize theamount of liquid being deposited with the tissuespheroids. This section with reduced diameter ispositioned between the outlet and the horizontalchannel. A vacuum pump removes the liquid, whilethe tissue spheroids are held in place by the pistonitself.

The bioprinter has three air-free syringe dis-pensers, with one of them able to keep the temperaturedown to 4 °C, one spray nozzle and one pneumaticsyringe dispenser. TheXY tablemoves with the resolu-tion of 5 μm and has a heating base with a maximumtemperature of 100 °C. An experimental single spher-oid printer head was developed in house and took theplace of one syringe for this experiment. The config-uration was designed for convenient incorporationinto a bioprinter and subsequent maintenance. Theprinting head module can be sterilized with either UVirradiation or ethyl alcohol aqueous solution.

To enable collagen hydrogel printing, a specialcooling/heating system was developed in house. Thesystem consists of a controllable heating aluminumbase (with temperature is ranging from room temper-ature up to 200 °C in steps of 0.1 °C) and a cooling sys-tem (with temperature ranging from roomtemperature down to−4 °C in steps of 0.1 °C) to cooldown the syringe contents. Supplementary figure2(A), available online at stacks.iop.org/BF/9/034105/mmedia, shows the elements and their connections.The temperature of the syringe was controlled by acooling system based on a Peltier module. This mod-ule allowed cooling by a recirculating liquid that trans-fers the heat via a coiled bronze tube around thesyringe. The cooled syringe kept the collagen mixliquid at 4 °C. The base plate uses a cartridge heater tochange its temperature up to 100 °C. Upon depositionon the heated plate (28 °C), the collagen mix poly-merizes, allowing a controlled 3D print shape (supple-mentaryfigures 2(B), (C)).

3

Biofabrication 9 (2017) 034105 EABulanova et al

http://stacks.iop.org/BF/9/034105/mmediahttp://stacks.iop.org/BF/9/034105/mmedia

Culture of printed thyroid constructsAfter the printing, thyroid gland constructs werefurther cultured within collagen hydrogel spread onthe surface of a transparent PTFE membrane for 4days. The culture medium did not cover the printedconstructs but reached the level of the PTFE mem-brane to allow better oxygenation of the tissue. After 4days in culture the printed samples were subjected toimmunolabeling or grafting. Bright-field imaging wasperformed on a Nikon Eclipse Ti-E inverted micro-scope with a Nikon DS-Qi2 camera. Images wereprocessed with NIS-Elements D Research softwareversion 4.0. PhotoshopCS5 (Adobe)was used to adjustbrightness, contrast and picture size.

ImmunolabelingAfter the culture, printed constructs were fixed at 4 °Cin 4% paraformaldehyde in PBS for 2 h, transferred insucrose 20%/PBS and then embedded in 15%sucrose/7.5% gelatin/PBS. Immunofluorescence onthin tissue sections was performed as described [24].Primary antibodies and dilutions were as follows:monoclonal mouse anti-E-cadherin at 1:250, rat anti-PECAM at 1:100, monoclonal mouse ant-Ezrin at1:300, rabbit anti-caspase-3a at 1:200, rabbit anti-pHH3 at 1:100, monoclonal mouse anti-TTF1 at1:200. Nuclei were counterstained with Hoechst inPBS during incubation with secondary antibodies.Fluorescence on sections was observed with a ZeissCell Axiovert 200 inverted fluorescence microscope orwith a Zeiss Cell Observer Spinning Disk (COSD)microscope.

Generation of 131I-induced hypothyroidismmousemodel and transplantation of bioprintedmousethyroid gland constructsA hypothyroidism mouse model was generated asdescribed previously [25, 26]. Briefly, experimentalhypothyroidism was induced by administering 150μCi of [131I] by intraperitoneal injection tomice. Eightweeks after the administration of [131I], plasma levelsof T4were analyzed to confirm the hypothyroid status.For transplantation of bioprintedmouse thyroid glandconstructs the hypothyroid mice were anesthetizedwith 5 ml kg−1 of an anesthetic solution composed of a5% solution of ketamine (Moscow Endocrine Plant,Russia) and 2.5 mgml−1 droperidol (Moscow Endo-crine Plant), then constructs were placed under therenal capsule using a blunt-pointed needle. Aftersurgery, the skin wound was treated with gluten BF-6(Vertex, Russia). Three and 5 weeks later, the graftedmice were subjected to body temperature measure-ment, blood sampling for plasma T4 measurementsand 5 weeks after transplantation sacrifice for histolo-gical examination of the kidneys.

All mice were of the CD1 strain. Mice were raisedand treated according to the principles of laboratory

animal care of the institute’s animal ethical commit-tee. All animal experiments and care were in com-pliance with institutional guidelines and local ethicalcommittees. Animals weremaintained on a 12 h light/12 h dark schedule (light on at 6 am) and fed withlaboratory chow andwater ad libitum.

Body temperaturemeasurementBody temperature was measured in conscious miceusing a ‘Precision’ 841 thermometer (RST, cat. no.RST07841). Animals were restrained and keptmotionless to obtain a stable body temperature.

PlasmaT4 levelmeasurementsThe total T4 level was assayed using a T4 mouse/ratenzyme-linked immunosorbent assay (ELISA) kit(BioVision, cat. no. K7421-100), according to themanufacturer’s instructions.

Results

Mathematicalmodeling and computer simulationof bioprinted thyroid glandIn order to demonstrate the feasibility of using tissuespheroids as building blocks in 3D bioprinting we firststarted with mathematical modeling and computersimulation (figure 1). We capitalized on previouslypublished data and experimental results, whichreported 3D bioprinting of sequential segments of abranched vascular tree. Here, for the first time, wereport mathematical modeling and computer simula-tion for biofabrication of a complete intra-organbranched vascular tree including proximal and distalvascular segments. The fusion of spheroids composedof tightly packed epithelial cells or ECwas theoreticallymodeled and simulated. To investigate the fusionprocess, a multicellular lattice-based model, whichdescribes the interactions between cells based on thedifferential adhesion hypothesis (DAH), was used[27]. This hypothesis states that: (i) cell adhesion inmulticellular systems depends on energy differencesbetween different cell types and (ii) cells in aggregatesare motile enough to reach the configuration whichminimizes the interfacial energy of the system. In themodel, each site of a 3D cubic lattice is occupied eitherby a cell or a similar-sized volume unit of hydrogelmedium. For simulation of the evolution of a multi-cellular system, the kinetic Monte Carlo (KMC)method was applied [28]. In the KMC method, self-assembly of cells in the cellular aggregate system isdescribed in terms of the transition rates corresp-onding to possible conformational changes of thesystem, and then the corresponding time evolution ofthe system is expressed in terms of these rates. Thedynamics of the cells depend on the transition rates ofcell swapping with adjacent cells of different type and/or with hydrogel particles [29, 30], which are given by

4

Biofabrication 9 (2017) 034105 EABulanova et al

the Arrhenius relation. In each step, the transitionrates for all possible changes from the current config-uration are calculated and then a new configuration ischosen with a probability proportional to the rate ofthe corresponding transition. Figures 1(A)–(D) depictthe computer simulations of the fusion processesbetween two or three tissue spheroids from differentinitial settings for bioprinting of the thyroid glandconstruct. Next, in order to examine the bioprintingsettings wemodeled the fusion of three TS surroundedby six vascular spheroids which form the core struc-ture of the engineered thyroid construct (figure 1(E)).This particular setting was then tested experimentally(see below). The model of thyroid gland construct (seeinside the circle) is placed in the context of branchedvascular tree bioprinting using self-assembly of vascu-lar tissue spheroids (figure 1(F)). Such an approach isnecessary in the case of orthotopic implantation usingsurgical vascular anastomoses. In the case of hetero-topic implantation under a highly vascularized kidneycapsule, which has been implemented in our study, theproposed bioprinting of a complete branched vasculartree is not required to achieve a desirable level ofvascularization in the implanted thyroid glandconstruct.

A detailed description of the mathematical modeland computer simulation is given as supplementaryinformation (SI).

Generation of tissue spheroids frommouse thyroidand allantoidesBased on the self-assembly principle, we aimed toprint mouse thyroid gland constructs using roundedmouse embryonic thyroid explants for the follicularcompartment and rounded mouse allantois explantsas a vascular, i.e. EC, source. Thyroid explants weremicrodissected from e14.5 mouse embryos asdescribed in [21] and cultured overnight in a hangingdrop to obtain rounded pieces of tissue suitable forbioprinting (figure 2(A)). It has been demonstratedpreviously that mouse embryonic thyroid folliculo-genesis starts at e14.5 and is concomitant withvascularization of embryonic thyroid tissue [11, 21].

Allantoides were microdissected from e8.5 mouseembryos as described in [22] and cultured overnight ina hanging drop to obtain rounding of the tissue pieces(figure 2(B)). Dissected e8.5mouse allantoides have anelongated rod-like shape and consist of cells withangioblastic phenotype. It has been previously repor-ted that isolation of allantoides at this stage withsequential incubation in hanging drop in the presenceof VEGF induces endothelial differentiation and for-mation of a cluster of EC, with subsequent generationof blood microvessels [22]. Allantoides were thus usedas a source of EC; thyroid explants as a source of folli-cular cells and of VEGF [21].

When thyroid explants and allantoides were culturedinhangingdrops, they tookona spherical shapeovernight

Figure 1.Mathematicalmodeling and computer stimulation of fusion processes starting fromdifferent initial settings for bioprintingofmouse thyroid gland construct. (A)Time evolution of the fusion between two thyroid spheroids (TS) in KMC simulations. Initially,each spheroid contains 2176 cells. The three snapshots are taken at t=0, 4×104 and 2×106 steps, respectively. (B)Time evolutionof three fusing TS. The three snapshots are taken at t=0, 4×104 and 8×105 steps, respectively. (C)Time evolution of two fusingvascular spheroids. Initially, each spheroid contains 1336 cells, which include 848 smoothmuscle cells (SMCs in red) in the outer layerand 488 endothelial cells (ECs in green) in the inner layer that engulfs 840 internal hydrogelmediumparticles (in yellow). The threesnapshots of cross-sectional views are taken at t=0, 2×105 and 6×106 steps, respectively. (D)Time evolution of three fusingvascular spheroids. The three snapshots of cross-sectional views are taken at t=0, 4×105, and 2×106 steps, respectively. (E)Timeevolution of a systemof six vascular spheroids with three TS in themiddle, which form the central part of the bioprinted thyroid gland.The three snapshots of cross-sectional views (top) and external views (bottom) are taken at t=0, 2×106 and 1×107 steps,respectively. (F)Time evolution of a systemof vascularized thyroid gland construct with two large-diameter vessels fabricated fromvascular tissue spheroids. The two snapshots of cross-sectional views (top) and external views (bottom) are taken at t=0 and 5×107

steps, respectively.

5

Biofabrication 9 (2017) 034105 EABulanova et al

with mean outer diameters of 388.2μm±45.3(n=168) and 493.6μm±114.3 (n=28), respectively.On assessing the circularities of AS and TS, they were esti-mated close to a value of 1.0, 0.92±0.06 (n=28) and0.90±0.05 (n=168), respectively, indicating an almostperfect circle (figure 2(C)). This study was performedusing Image J software (NIH,Bethesda,MD,USA).

Fusion of thyroid explants and allantoides in ahanging drop andwithin collagen hydrogelIn order to investigate the capacity of TS to fuse, a pairof them of equal size were first placed in contact witheach other in a hanging drop and incubated overnight(supplementary figure 1(A)). Eighteen hours later theywere found to merge together as a single spheroid(supplementary figure 1(B)). This fusion was observedin the case of a pair of allantoides as well, incubatedunder similar conditions. Fusion is an inherentproperty of tissue spheroids, and hence this acts as a

proof of principle that spheroids can be used asbuilding blocks in our bioprinting technology.

We then evaluated the consequence of fusion ofembryonic TS under conditions in which the spher-oids are capable of engaging in adhesive interactionswith their culture environment. For that matter, threespheroids were placed within collagen hydrogel(figure 2(D)). In contrast to the spherical structuregenerated earlier, an oviform, elongated structure wasformed within the collagen. A few cells, endowed withmigratory properties, left the elongated structure andinvaded the 3D collagen hydrogel. The cohesiveness ofthe TS was measured using the fusion angle, whichincreased to 162 °C during culture within the collagengel indicating complete spheroid fusion (figure 2(E)).

We concluded that TS and AS are endowed withthe property of endogenous fusion, thereby validatingtheir use as building blocks for organ bioprintingtechnology.

Figure 2.Generation of rounded building blocks formouse thyroid gland construct printing. Representative images of dissected tissueand spheroids of thyroid lobes (A) and allantoides (B) after hanging drop culture. The circularity before and after hanging drop culturewas calculated (C). The circularity of thyroid lobes and allantoides increased after hanging drop culture. An unpaired t-test was usedfor statistical analysis (*P

Bioprinting usingmouse thyroid and allantoicspheroidsFor this study we developed a novel original multi-functional 3D bioprinter, Fabion (figure 3(A)), withtwo unique functional characteristics: first, it isequipped with a technology, based on the turnstileprinciple [31, 32] that facilitates dispensing of a singletissue spheroid at a time (figures 3(B), (C)); second, itincludes a cooling/heating system for better controlover the polymerization of the collagen hydrogel(supplementary figure 2). Our system allows the use ofa collagenmixwith a physiological pH and prevents itspremature gelation while being stored at 4 °C in anextrusion needle and during the printing. The poly-merization starts as soon as the hydrogel is depositedon the hot base plate, resulting in a collagen gel with astable and predictable structure. The first step was toprint a 2 mm×2 mm collagen bed on cell strainers inthe Petri dish to serve as a base. After polymerization,three e14.5 TS were loaded into the print head andplaced in a line and adjacent to each other on top of thecollagen bed according to the mathematical model(figure 3(D)). We also developed a more advanced

version of the bioprinted construct (TS+AS). Sixe8.5 mouse AS were loaded into the printing head andprinted adjacent to and on both sides of the three TS,again to mimic the mathematical model (figures 3(C),(F); compare with figure 1(E)). To complete theprinting process, a thin layer of collagen hydrogel wasdeposited on top of three TS, or three TS with six AS,depending on the construct version.

Endothelial cells fromallantoic spheroids areresponsible for the vascularization of thyroid glandconstructThe vascularization of bioprinted thyroid gland con-struct is themost important and critical issue since it isrequired for a surrogate endocrine organ to befunctional and both capture iodine and deliver hor-mones into the blood flow. We first analyzed theprinted thyroid constructs containing TS only. After 4days in culture, constructs were fixed and immunola-beled with antibodies against TTF1, E-cadherin andezrin to visualize thyrocyte nuclei, basolateral andapical domains, respectively. TTF1+ nuclei displayed amonolayered and organized circular pattern.

Figure 3.Bioprinting ofmouse thyroid gland construct using the turnstile device. (A)The Fabion 3Dbioprinter developed by 3DBioprinting Solutions and used for thyroid gland construct printingwith single tissue spheroids. (B)Turnstile device for bioprintingof single spheroids: spheroids loading point (a), channel with spheroids inline (b); output formedia culture (c); loading piston (d);trapping piston (e); deposition piston (f); recess in the piston formedia extraction (g). (C) Step-by-step progression of a single tissuespheroid using the turnstile device for bioprinting. (D)Bioprintedmouse thyroid gland construct composed of three spheroids fromembryonic thyroid explants (3TS) or (E) composed of three spheroids from embryonic thyroid explants in between six allantoidesspheroids (3TS+6AS) after 1 day in culture. (F)Bioprintedmouse thyroid gland construct labeled, after 1 day in culture, withantibodies against platelet and endothelial cell adhesionmolecule (PECAM; green) andE-cadherin (red) to visualize endothelial andepithelial cells, respectively. The insert in F shows thewidespread localization of PECAM-positive cells in bioprintedmouse thyroidconstruct (TS+AS) after 4 days in culture.

7

Biofabrication 9 (2017) 034105 EABulanova et al

E-cadherin confirmed the formation of epithelialmonolayers, each delineating a central space. Ezrin+

epithelial cells demonstrated apical polarization ofcells facing a lumen (figures 4(A), (A′), (A″)). Printedspheroid tissue was alive and growing as cells positivefor the proliferation marker pHH3 were found in theconstruct (figure 4(A″)). However, compared with thewidespread localization of EC in the TS at the end ofthe hanging drop culture (supplementary figure 3.),EC were only found at the periphery of thyroid tissueafter 4 days in collagen (figure 4(A)). This suggests thatthyroid EC localized deep inside the tissue constructdo not support long-term culture in collagen, prob-ably due to low pO2. Interestingly, the constructsprinted from TS and AS (3TS+6AS) displayedabundant and widespread localization of EC aroundepithelial cells of the thyroid tissue after 4 days incollagen. Ezrin and TTF1 visualization revealed devel-oping follicular structure in thyroid tissue(figures 4(B′), (B″)). These observations suggest thatendothelial progenitors present in AS responded andwere attracted by the angiogenic factor, VEGF-A,highly secreted by thyrocyte progenitors of the TS [11].However, one cannot exclude that thyroid ECresponded to survival and proliferating signalssecreted by the allantoides. To distinguish betweenthese two possibilities, we cultured e14.5 thyroidexplants in hanging drops in the presence of thespecific VEGFR2/Flk-1 kinase inhibitor SU5416(TSSU5416). As VEGFR2 signaling is essential for ECsurvival, inhibition of this signaling pathway leads to

progressive EC death [11]. Addition of 5 μM SU5416to the culture medium of thyroid explants for 24 hindeed removed all PECAM-positive EC from TS(figure 4(C)), indicating that SU5416 treatment waseffective.We then bioprinted three SU5416-treated TS(TSSU5416) with six AS, as above, and cultured theconstruct in collagen hydrogel for 4 days. Bioprintedconstructs showed invasion of EC deep into thyroidtissue (figure 4(D)), resulting in the generation of adense capillary network surrounding epithelial cells.This indicates that EC detected around follicles onthese thin sections were clearly derived from allan-toides. SU5416 treatment did not affect the prolifera-tion of thyrocytes (figures 4(C″), (D″)) or theformation of prefollicular structures (figures 4(C′),(D′), (C″), (D″)). Altogether, these experimentsdemonstrate that addition and fusion of allantoides tothyroid tissue within collagen hydrogels allowimproved vascularization of bioprinted TS.

Bioprintedmouse thyroid gland constructs arefunctional in vivoIn order to validate the functionality of bioprintedcultured mouse thyroid constructs (3TS+6AS), wegrafted these latter under the kidney capsule ofhypothyroid mice, following 131I injection(figure 5(A)). Implantation of tissue pieces under thevascularized kidney capsule is a standard and well-established approach for in vivo testing [33–36].Experiments on hypothyroid animals were performed8 weeks after the administration of 131I, when all mice

Figure 4.Vascularization of bioprintedmouse thyroid constructs. Immunolabeling of thin sections of bioprinted constructs after 4days of culture with antibodies as indicated. (A)Bioprinted thyroid spheroids (TS) display endothelial cells (EC) only at the peripheryof the tissue. (B)TS bioprinted together with allantoic spheroids (AS) reveal intense PECAM labeling around and inside the TS. (C)TSpretreated with SU5416 (TSSU5416) during hanging drop culture do not show endothelial labeling. (D) In the presence of AS, TSdepleted of EC (TSSU5416) are invaded by EC fromAS. Thyroid follicle formation (A′)–(D′) aswell as cell proliferation (A″)–(D″)wascomparable in all the cultured bioprinted constructs.

8

Biofabrication 9 (2017) 034105 EABulanova et al

had undetectable blood concentrations of T4(figure 5(B)). Histological evaluation of the nativethyroid region 8 weeks after administration of 131Irevealed loss of thyroid tissue (supplementary figures4(A), (B)). The blood level of T4 was measured as aspecific and most informative marker. Three and 5weeks after grafting, recipient mice showed a substan-tial elevation of T4 blood levels (figure 5(C) andsupplementary figure 5(A)), indicating functional

rescue with the bioprinted thyroid construct(3TS+6AS).

Additionally, we analyzed the body temperature inmice from the different groups. As expected,decreased body temperature was found in hypothyr-oid mice (figure 5(D)). Mice grafted with the printedand cultured thyroid gland construct showed a gra-dual normalization of body temperature at 3 and 5weeks after transplantation, providing a reliable

Figure 5. Functional rescue of in vivo thyroid function using bioprintedmouse thyroid construct. (A) Schematic representation andtime line ofmouse 131I-induced thyroid ablation and construct grafting. (B)T4 blood level and (D) body temperature in untreated andhypothyroidmice 8weeks after 131I-induced thyroid ablation. 131I-treatedmice present clear hypothyroidism. An unpaired t-test wasused for statistical analysis (*P

example for symptomatic recovery (figure 5(E), sup-plementary figure 5(B)).

Histological evaluation of the kidney region 3 and5 weeks after transplantation demonstrated successfulintegration andmaturation of grafted organoids in thehost niche. At the grafting site, numerous follicles con-taining a monolayered epithelium were present at thecortical area of the host organ (figure 5(F)). Regions ofnew capillary formation filled with erythrocytes weredetectable (supplementary figure 5(C)).

Our in vivo data clearly demonstrate that printedthyroid gland constructs have potent functional capa-city to compensate for the lack of native thyroid tissueallowing functional rescue of experimentally inducedhypothyroidism.

Discussion

In the present study we validated the earlier proposedprinciple of bioprinting technology based on the use oftissue spheroids as building blocks [5, 6]. Our data onbioprinting of functional vascularized mouse thyroidgland construct from TS and AS represents furtheradvance in the emerging solid scaffold-free approachfor 3D bioprinting, exploring self-assembling proper-ties of tissue spheroids [6, 15, 37]. Tissue spheroidscapable of fusion are already used for scaffold-freebiofabrication of cartilage [38], bone [39], liver [40],ovary [41] and blood vessels [9, 42]. Using tissuespheroids as opposed to single cells has substantialadvantages [43, 44]. First, tissue spheroids representdensely packed spherical aggregates of living cells.Spheroids thus ensure the maximum achievable celldensity during the printing, while displaying a higherresistance to shear stress, radiation and other unfavor-able factors, which results in decreasing risks ofkaryotype alteration [45]. Secondly, precise embed-ding of tissue spheroids into sequentially applied layersof hydrogel ensures the desired allocation of closelyplaced tissue spheroids contacting each other in bothhorizontal and vertical directions [19]. This createsproper conditions for formation of 3D tissue con-structs as a result of tissue fusion. Finally, the sphericalform of tissue spheroids is ideal for their roboticdispensing through the cylindrical nozzle of 3Dbioprinter.

However, tissue spheroid printing has also somedrawbacks, and problems can emerge during bio-printing: (i) spontaneous fusion of spheroids resultingin the formation of clusters or aggregates seriouslycomplicates bioprinting with individual tissue spher-oids; (ii) adhesion of tissue spheroids to the walls ofdispensing tubes obstructs the passage of further tissuespheroids; (iii) both the above phenomena causeundesired damage, defects, deformation and evendestruction of tissue spheroids. To overcome theaforementioned problems, we created a device allow-ing 3D bioprinting of individual tissue spheroids,

using the turnstile principle. Bioprinting achievedwith our device is stable and reliable.

Collagen is amajor component of the extracellularmatrix, it supports cell adhesion and culture but itsapplication in 3D bioprinting still has limitationsbecause it is sensitive to both pH and temperature andforms a gel at physiological pH and 37 °C [46]. It isthus difficult to keep a collagenmix liquid when print-ing and produce a solid scaffold immediately afterprinting [47]. The first approach to printing with col-lagen was changing its pH by addition of sodiumhydrogen carbonate (NaHCO3) before deposition toinitiate the transition of collagen solution to a gel whenthe temperature gradually increases during the print-ing. Although this system improved deposition of thehydrogel by the bioprinter, the process was time con-suming, resulting in clogging inside the extrusion nee-dle due to anticipatory polymerization caused bygradual transition to room temperature.We improvedthe printing of collagen hydrogel by implementing acooling/heating system in our extrusion bioprinter,which had been previously used [48], but its bio-compatibility was not evaluated.

Bioprinting of thyroid gland construct containinga dense network of EC is another important achieve-ment of this study. 3D bioprinted tissue constructsquickly develop necrotic regions without sufficientvascularization [49]. Printing the channels from sacri-ficial hydrogels followed by the seeding of EC andsmooth muscle cells for vessel wall formation is anapproach developed recently [13, 50]. Carbohydrateglass [13] or pluronic acid F-127 [50]was used to printsacrificial tunnels within a supportingmatrix and thenextruded by dissolution, leaving room for EC popula-tion. This approach is highly effective and repro-ducible for simple homogeneous constructs.However, natural tissues and organs are effectivelyperfused by a hierarchically branched complex intra-organ vascular tree. These will be difficult to mimicusing sacrificial matrices. In this study we employedthe self-assembly approach which represents an alter-native to the scaffold-free technique for engineeredtissue vascularization. It has been demonstrated thatlarge-diameter blood vessels could be bioprinted fromvascular tissue spheroids biofabricated from smoothmuscle cells and EC [43]. The capacity of lumenizedvascular AS to fuse into linear and branched inter-mediate-diameter vascular segments has been repor-ted [6]. Finally, the microvascular networks can beself-assembled either from single cells placed intohydrogel by vacuole accumulation [51] or from endo-thelial tissue spheroids [52]. Thus, all three main partsof the intra-organ branched vascular tree—large-dia-meter vessel segments, intermediate-diameter vas-cular segments and microvascular segments—couldbe bioprinted using vascular tissue spheroids as build-ing blocks.

In the present study we showed that sufficientdevelopment of a microvascular network could be

10

Biofabrication 9 (2017) 034105 EABulanova et al

achieved in the engineered tissue construct by placingAS adjacent to TS, known to produce high levels ofVEGF-A [11]. Indeed, we found that the endothelialdensity was more prominent in AS+TS than in TSalone. More importantly, TS depleted of EC wereeffectively invaded by EC when AS were added in theengineered construct. The presence of a dense micro-vascular network is not only significant for survival ofthe cells upon in vivo grafting. In the developing thyr-oid, EC invasion, from e12.5 onwards also positivelycontrols the induction of thyroid folliculogenesis bypromoting basal lamina deposition [11, 24]. Thus,allantoides-derived ‘vasculature’ could contribute toadvancing thyroid folliculogenesis and sequential dif-ferentiation and maturation of follicles in the thyroidgland construct. Most importantly, better survival andexpansion of EC from AS could allow formation oflarge, intermediate and microvessels in the construct,thereby allowing functional connection with the vas-cularization from the host mice. Taken together theseresults strongly support the practical feasibility ofdeveloping vascularized-competent tissue using bio-printing of vascular tissue spheroids. It should notescape to our attention that the advance in bioprintingof functional tissue depends on our understanding ofthe principles of cellular self-assembly and the abilityto take advantage of this knowledge.

The thyroid gland construct must include theartery and the vein to establish anastomosis with thevascular system of the receiving organism. This isimportant for orthotopic implantation, when theprinted construct is to be transplanted to the usualthyroid gland site. However, this is not critical if theimplantation is heterotopic. It is well known that thyr-oid gland transplants can survive under the skin, inmuscles or under the kidney capsule and still producehormones [53, 54]. In case of heterotopic implant-ation under highly vascularized kidney capsula there isno need for large-diameter blood vessels connected tothe bioprinted thyroid gland construct. Our resultssuggest that the connection of the microvascular net-work from the bioprinted construct, with that of thekidney capsula allows vascularization of the thyroidfollicles.

Validation of the functionality of bioprintedmouse thyroid gland construct is another significantresult of our study. Grafting of the bioprinted mousethyroid construct under the kidney capsula of hypo-thyroid mice restored blood T4 level and body temp-erature, demonstrating functionality of bioprintedmouse thyroid construct. The amplification of TS,improvement of thyroid follicle vascularization andthe level of their histological differentiation andmaturation should eventually achieve 100% rescue oflost function in athyroidmice.

A consistent question arises concerning the possi-ble clinical translation of our proposed variant of bio-printing technology. In order to bioprint human

thyroid tissue in the future it is necessary to have clini-cally relevant source of therapeutic-grade thyroid epi-thelial cells. However, the effective propagation andscalable generation of large number of autologous dif-ferentiated human thyroid epithelial cells from biopsyis still a challenge. As of today, our efforts to isolate andpropagate human thyroid epithelium or even viablethyroid follicles directly from human thyroid glandhave not been successful (data not shown). So thisdirect approach is looking increasingly elusive andpractically unfeasible. In this context, scalable genera-tion of functional thyroid epithelium from humanstem cells looks much more promising. The reportedgeneration of thyroid follicles at first from mouseembryonic stem cells [25], then from human embryo-nic stem cells [55], and most recently, from humaninduced pluripotent stem cells [56] opens an opportu-nity for bioprinting of human thyroid gland. Tissuespheroids biofabricated from human smooth muscleand EC could be successfully generated from humanfat tissue stem cells [57] or from human induced plur-ipotent stem cells using directed tissue differentiation[58, 59]. Alternatively, it would be possible to employhuman endothelial tissue spheroids, which are capableof forming a vascular network in vitro and in vivo[52, 60]. Autologous EC could be isolated in sufficientnumber from human fat tissues [61]. Thus, 3D bio-printing of thyroid human tissue from organo-specifictissue spheroids and vascular tissue spheroids capableof post-printing tissue fusion is technologicallyfeasible.

Conclusion

A thyroid gland construct has been bioprinted usingtwo types of rounded embryonic tissue spheroids:thyroid spheroidswere printed in close associationwithallantoic spheroids within collagen hydrogel using anovel bioprinter capable of precise placing of onespheroid at a time.Within 4 days, tissue spheroids fusedinto a single and integral thyroid gland construct, inwhich endothelial cells from allantoic spheroidsinvaded and vascularized thyroid spheroids, generatinga dense capillary network around developing follicles.Thebioprinted construct is functional as it could restorethyroid homeostasis after grafting under the kidneycapsule of hypothyroid mice. Thus, our data demon-strated that bioprinting of tissue spheroids (i) derivedfrom dissected embryonic mouse tissues, (ii) used asbuilding blocks and (iii) capable of fusion can developin vascularized-competent thyroid gland construct,functional in vivo. Our data represent a significantadvance in the development of organ printing technol-ogy and important step toward eventual bioprinting ofvascularized functional human tissue.

11

Biofabrication 9 (2017) 034105 EABulanova et al

Acknowledgments

We thank Dr Sabina Costagliola for expertise inestablishing a radioiodine-ablated animal model ofthyroid gland hypofunction and helpful discussions,Anne-Sophie Delmarcelle for thyroid microdissectionprocedure and Emy Tassenoey for immunohistologi-cal analyses of bioprinted constructs. This workwas supported in part by grants from: Fonds dela Recherche Scientifique (FNRS) (FRS-FNRS;J.0126.16); Action de Recherche Concertées fromUCL (ARC 15/20-065); Fondation Roi Baudouin toCEP; JDwas supported by FRIA, andCEP is a ResearchAssociate of the FRS-FNRS. Y Sun is partiallysupported by the NSF Grants DMS-1318866, DMS-1620212 and a SC EPSCoR GEAR Award. Qi Wang’sresearch is partially supported by NIH and NSFthrough awards DMS-1200487, DMS-1517347 andR01GM078994-05A1 and a SCEPSCoRGEAR award.

Author contributions

EAB,CEP andVAMdesigned the research; EVK,VAP,FDP, JD, CH and SAA performed the research; YS andQW created the mathematical model and computersimulation, EVK, NSS, GAF, EAB and CEP analyzedthe data; EAB, YDK,CEP andVAMwrote the paper.

ORCID iDs

ElenaABulanova https://orcid.org/0000-0001-8460-3113

References

[1] Eligar V, Taylor PN,OkosiemeOE, LeeseGP andDayanCM2016Thyroxine replacement: a clinical endocrinologist’sviewpointAnn. Clin. Biochem. 53 421–33

[2] Biondi B andWartofsky L 2014Treatmentwith thyroidhormoneEndocr. Rev. 35 433–512

[3] MaR,Morshed SA, Latif R andDavies T F 2015Thyroid celldifferentiation frommurine induced pluripotent stem cellsFront Endocrinol. 6 56

[4] Bajaj P, Schweller RM,Khademhosseini A,West J L andBashir R 2014 3Dbiofabrication strategies for tissueengineering and regenerativemedicineAnnu. Rev. Biomed.Eng. 16 247–76

[5] MironovV, BolandT, Trusk T, Forgacs G andMarkwald RR2003Organ printing: computer-aided jet-based 3D tissueengineeringTrends Biotechnol. 21 157–61

[6] MironovV, Visconti R P, KasyanovV, ForgacsG,DrakeC J andMarkwald RR 2009Organ printing: tissuespheroids as building blocksBiomaterials 30 2164–74

[7] Morimoto Y,HsiaoAY andTakeuchi S 2015 Point-, line-, andplane-shaped cellular constructs for 3D tissue assemblyAdv.Drug. Deliv. Rev. 95 29–39

[8] SchonB S,HooperG J andWoodfield TBF 2017Modulartissue assembly strategies for biofabrication of engineeredcartilageAnn. Biomed. Eng. 45 100–14

[9] Fleming PA, ArgravesWS,Gentile C,NeaguA, ForgacsG andDrakeC J 2010 Fusion of uniluminal vascular spheroids: amodel for assembly of blood vesselsDev. Dyn. 239 398–406

[10] Mohebati A and ShahaAR 2012Anatomy of thyroid andparathyroid glands and neurovascular relationsClin. Anat. 2519–31

[11] HickAC et al 2013Reciprocal epithelial: endothelial paracrineinteractions during thyroid development govern follicularorganization andC-cells differentiationDev. Biol. 381 227–40

[12] RadisicM,Malda J, Epping E, GengW, Langer R andVunjak-Novakovic G 2006Oxygen gradients correlate withcell density and cell viability in engineered cardiac tissueBiotechnol. Bioeng. 93 332–43

[13] Miller J S et al 2012Rapid casting of patterned vascularnetworks for perfusable engineered three-dimensional tissuesNat.Mater. 11 1–7

[14] Rouwkema J,RivronNCandvanBlitterswijkCA2008Vascularization in tissue engineeringTrendsBiotechnol.26434–41

[15] Visconti R P, KasyanovV,Gentile C, Zhang J,MarkwaldRR andMironovV 2010Towards organ printing:engineering an intra-organ branched vascular treeExpert Opin.Biol. Ther. 10 409–20

[16] RichardsD, Jia J, YostM,MarkwaldR andMei Y 2017 3DBioprinting for vascularized tissue fabricationAnn. Biomed.Eng. 45 132–47

[17] Gentile C, Fleming PA,MironovV, Argraves KM,ArgravesWS andDrakeC J 2008VEGF-mediated fusion in thegeneration of uniluminal vascular spheroidsDev.Dyn. 2372918–25

[18] Kelm JM et al 2005VEGF profiling and angiogenesis in humanmicrotissues J. Biotechnol. 118 213–29

[19] JakabK,NeaguA,MironovV,Markwald RR and ForgacsG2004 Engineering biological structures of prescribed shapeusing self-assemblingmulticellular systems Proc. Natl Acad.Sci. USA 101 2864–9

[20] RajanN,Habermehl J, CotéM-F, DoillonC J andMantovaniD 2006 Preparation of ready-to-use, storable andreconstituted type I collagen from rat tail tendon for tissueengineering applicationsNat. Protoc. 1 2753–8

[21] Delmarcelle A-S, VillacorteM,Hick A-C and PierreuxCE2014An ex vivo culture system to study thyroid developmentJ. Vis. Exp. 8–15

[22] DrakeC J and Fleming PA 2000Vasculogenesis in the day 6.5to 9.5mouse embryoBlood. 95 1671–9

[23] SusienkaM J,Wilks B T andMorgan J R 2016Quantifying thekinetics andmorphological changes of the fusion of spheroidbuilding blocksBiofabrication 8 45003

[24] VillacorteM et al 2016Thyroid follicle development requiresSmad1/Smad5- and endothelial-dependent basementmembrane assemblyDevelopment 143 1958–70

[25] Antonica F et al 2013Generation of functional thyroid fromembryonic stem cellsNature 491 66–71

[26] Abel ED et al 1999Divergent roles for thyroid hormonereceptor beta isoform inthe endocrine axis and auditory systemJ. Clin. Invest. 104 291–300

[27] SteinbergMS 1963Reconstruction of tissues by dissociatedcells Science 141 401–8

[28] Bortz A, KalosM and Lebowitz J 1975Anew algorithm forMonte Carlo simulation of Ising spin systems J. Comput. Phys.17 10–8

[29] Flenner E, Janosi L, Barz B,NeaguA, ForgacsG andKosztin I2012KineticMonte Carlo and cellular particle dynamicssimulations ofmulticellular systems Phys. Rev.E 85 1–10

[30] SunY et al 2013Modeling and simulations ofmulticellularaggregate self-assembly in biofabrication using kineticMonteCarlomethods SoftMatter 9 2172

[31] LangM,WangW,ChenXQ andWoodfield T 2010 Integratedsystem for 3D assembly of bio-scaffolds and cells 2010 IEEEInt. Conf. Autom. Sci. Eng. CASE 2010 pp 786–91

[32] LangM,ChenX,WangWandWoodfield TBF 2012 Injectionsystem for cellular assembly of 3Dbio-tissue engineeredconstructs IEEE Int. Conf. Autom. Sci. Eng. pp 291–6

[33] MaSN, YuanYH,GuoXR and LiD S 2016 Subcapsularimplantation of pancreatic islets in syngeneic, allogeneic, andxenogeneicmiceTransplant Proc. 48 2821–5

12

Biofabrication 9 (2017) 034105 EABulanova et al

https://orcid.org/0000-0001-8460-3113https://orcid.org/0000-0001-8460-3113https://orcid.org/0000-0001-8460-3113https://orcid.org/0000-0001-8460-3113https://orcid.org/0000-0001-8460-3113https://doi.org/10.1177/0004563216642255https://doi.org/10.1177/0004563216642255https://doi.org/10.1177/0004563216642255https://doi.org/10.1210/er.2013-1083https://doi.org/10.1210/er.2013-1083https://doi.org/10.1210/er.2013-1083https://doi.org/10.3389/fendo.2015.00056https://doi.org/10.1146/annurev-bioeng-071813-105155https://doi.org/10.1146/annurev-bioeng-071813-105155https://doi.org/10.1146/annurev-bioeng-071813-105155https://doi.org/10.1016/S0167-7799(03)00033-7https://doi.org/10.1016/S0167-7799(03)00033-7https://doi.org/10.1016/S0167-7799(03)00033-7https://doi.org/10.1016/j.biomaterials.2008.12.084https://doi.org/10.1016/j.biomaterials.2008.12.084https://doi.org/10.1016/j.biomaterials.2008.12.084https://doi.org/10.1016/j.addr.2015.09.003https://doi.org/10.1016/j.addr.2015.09.003https://doi.org/10.1016/j.addr.2015.09.003https://doi.org/10.1007/s10439-016-1609-3https://doi.org/10.1007/s10439-016-1609-3https://doi.org/10.1007/s10439-016-1609-3https://doi.org/10.1002/dvdy.22161https://doi.org/10.1002/dvdy.22161https://doi.org/10.1002/dvdy.22161https://doi.org/10.1002/ca.21220https://doi.org/10.1002/ca.21220https://doi.org/10.1002/ca.21220https://doi.org/10.1002/ca.21220https://doi.org/10.1016/j.ydbio.2013.04.022https://doi.org/10.1016/j.ydbio.2013.04.022https://doi.org/10.1016/j.ydbio.2013.04.022https://doi.org/10.1002/bit.20722https://doi.org/10.1002/bit.20722https://doi.org/10.1002/bit.20722https://doi.org/10.1038/nmat3357https://doi.org/10.1038/nmat3357https://doi.org/10.1038/nmat3357https://doi.org/10.1016/j.tibtech.2008.04.009https://doi.org/10.1016/j.tibtech.2008.04.009https://doi.org/10.1016/j.tibtech.2008.04.009https://doi.org/10.1517/14712590903563352https://doi.org/10.1517/14712590903563352https://doi.org/10.1517/14712590903563352https://doi.org/10.1007/s10439-016-1653-zhttps://doi.org/10.1007/s10439-016-1653-zhttps://doi.org/10.1007/s10439-016-1653-zhttps://doi.org/10.1002/dvdy.21720https://doi.org/10.1002/dvdy.21720https://doi.org/10.1002/dvdy.21720https://doi.org/10.1002/dvdy.21720https://doi.org/10.1016/j.jbiotec.2005.03.016https://doi.org/10.1016/j.jbiotec.2005.03.016https://doi.org/10.1016/j.jbiotec.2005.03.016https://doi.org/10.1073/pnas.0400164101https://doi.org/10.1073/pnas.0400164101https://doi.org/10.1073/pnas.0400164101https://doi.org/10.1038/nprot.2006.430https://doi.org/10.1038/nprot.2006.430https://doi.org/10.1038/nprot.2006.430https://doi.org/10.3791/51641https://doi.org/10.3791/51641https://doi.org/10.3791/51641https://doi.org/10.1088/1758-5090/8/4/045003https://doi.org/10.1242/dev.134171https://doi.org/10.1242/dev.134171https://doi.org/10.1242/dev.134171https://doi.org/10.1038/nature11525https://doi.org/10.1038/nature11525https://doi.org/10.1038/nature11525https://doi.org/10.1172/JCI6397https://doi.org/10.1172/JCI6397https://doi.org/10.1172/JCI6397https://doi.org/10.1126/science.141.3579.401https://doi.org/10.1126/science.141.3579.401https://doi.org/10.1126/science.141.3579.401https://doi.org/10.1016/0021-9991(75)90060-1https://doi.org/10.1016/0021-9991(75)90060-1https://doi.org/10.1016/0021-9991(75)90060-1https://doi.org/10.1103/PhysRevE.85.031907https://doi.org/10.1103/PhysRevE.85.031907https://doi.org/10.1103/PhysRevE.85.031907https://doi.org/10.1039/c2sm27090khttps://doi.org/10.1109/COASE.2010.5584476https://doi.org/10.1109/COASE.2010.5584476https://doi.org/10.1109/COASE.2010.5584476https://doi.org/10.1109/CoASE.2012.6386394https://doi.org/10.1109/CoASE.2012.6386394https://doi.org/10.1109/CoASE.2012.6386394https://doi.org/10.1016/j.transproceed.2016.06.045https://doi.org/10.1016/j.transproceed.2016.06.045https://doi.org/10.1016/j.transproceed.2016.06.045

[34] vanGurp L et al 2016 Sequential intravital imaging revealsin vivo dynamics of pancreatic tissue transplanted under thekidney capsule inmiceDiabetologia 59 2387–92

[35] Raikwar S P, KimEM, SivitzW I, Allamargot C,ThedensDR andZavazavaN2015Human iPS cell-derivedinsulin producing cells form vascularized organoids under thekidney capsules of diabeticmice PLoSOne 10 1–15

[36] Xinaris C et al 2012 In vivomaturation of functional renalorganoids formed from embryonic cell suspensions J. Am. Soc.Nephrol. 23 1857–68

[37] Rezende RA et al 2013 Scalable biofabrication of tissuespheroids for organ printing Procedia CIRP 5 276–81

[38] Lee J I, SatoM,KimHWandMochida J 2011Transplantatation of scaffold-free spheroids composed ofsynovium-derived cells and chondrocytes for the treatment ofcartilage defects of the knee Eur. CellMater. 22 275–90

[39] TangD, Tare R S, Yang LY,WilliamsDF,OuKL andOreffo ROC2016 Biofabrication of bone tissue: approaches,challenges and translation for bone regenerationBiomaterials83 363–82

[40] Lee J, CuddihyM J, CaterGMandKotovNA2009Engineering liver tissue spheroids with inverted colloidalcrystal scaffoldsBiomaterials 30 4687–94

[41] Liao J et al 2014Ovarian cancer spheroid cells with stem cell-like properties contribute to tumor generation,metastasis andchemotherapy resistance through hypoxia-resistantmetabolism PLoSOne 9 1–13

[42] InamoriM,MizumotoH andKajiwara T 2009An approachfor formation of vascularized liver tissue by endothelial cell-covered hepatocyte spheroid integrationTissue Eng.A 152029–37

[43] ItohM et al 2015 Scaffold-free tubular tissues created by a bio-3Dprinter undergo remodeling and endothelializationwhenimplanted in rat aortaePLoSOne 10 e0136681

[44] RagoAP,DeanDMandMorgan J R 2009Controlling cellposition in complex heterotypic 3Dmicrotissues by tissuefusionBiotechnol. Bioeng. 102 1231–41

[45] YamadaKMandCukierman E 2007Modeling tissuemorphogenesis and cancer in 3DCell 130 601–10

[46] Roth EA,XuT,DasM,Gregory C,Hickman J J and BolandT2004 Inkjet printing for high-throughput cell patterningBiomaterials 25 3707–15

[47] Rutz A L,HylandKE, Jakus AE, BurghardtWR and ShahRN2015Amultimaterial bioinkmethod for 3Dprinting tunable,cell-compatible hydrogelsAdv.Mater. 27 1607–14

[48] NoceraAD, SalvatierraNA andCidMP2014 PrintingCollagen 3D Structures IFMBE Proc. 49 136–9

[49] Griffith CK et al 2005Diffusion Limits of an in vitro thickprevascularized tissueTissue Eng. 11 257–66

[50] Bertassoni L E et al 2014Hydrogel bioprintedmicrochannelnetworks for vascularization of tissue engineering constructsLabChip 14 2202–11

[51] KameiM, SaundersWB, Bayless K J, Dye L,Davis G E andWeinstein BM2006 Endothelial tubes assemble fromintracellular vacuoles in vivoNature 442 453–6

[52] Alajati A et al 2008 Spheroid-based engineering of a humanvasculature inmiceNat.Methods 5 439–45

[53] IwaiH et al 1993Thyroid follicle transplantation bypercutaneous injectionActaOto-Laryngol. 113 135–7

[54] LaRosa FG, SmilekD, TalmageDW,LaffertyK J, BaulingP andAmmonsT J 1992Evidence that tolerance to cultured thyroidallografts is an active immunological process. Protectionofthird-party grafts bearing new antigenswhen associatedwithtolerogenic antigensTransplantation 53903–13

[55] MaR, Latif R andDavies T F 2015Human embryonic stemcells form functional thyroid folliclesThyroid. 25 455–61

[56] KurmannAA et al 2015Regeneration of thyroid function bytransplantation of differentiated pluripotent stem cellsCellStemCell 17 527–42

[57] MinedaK et al 2015Therapeutic potential of human adipose-derived stem/stromal cellmicrospheroids prepared by three-dimensional culture in non-cross-linked hyaluronic acid gelStemCells Transl.Med. 4 1511–22

[58] VodyanikMA et al 2010Amesoderm-derived precursor formesenchymal stem and endothelial cellsCell StemCell 7718–29

[59] PatschC et al 2015Generation of vascular endothelial andsmoothmuscle cells fromhuman pluripotent stem cellsNat.Cell Biol. 17 994–1003

[60] Pfisterer L andKorff T 2016 Spheroid-based in vitroangiogenesismodelMethodsMol. Biol. 1430 167–77

[61] Tiwari A, Salacinski H J,HamiltonG and Seifalian AM2001Tissue engineering of vascular bypass grafts: role of endothelialcell extraction Eur. J. Vasc. Endovasc. Surg. 21 193–201

[62] Graner F andGlazier J A 1992 Simulation of biological cellsorting using a 2-dimensional extended potts-model Phys. Rev.Lett. 69 2013–6

[63] Glazier J A andGraner F 1993 Simulation of the differentialadhesion driven rearrangement of biological cells Phys. Rev.Lett. 47 2128–54

13

Biofabrication 9 (2017) 034105 EABulanova et al

https://doi.org/10.1007/s00125-016-4049-6https://doi.org/10.1007/s00125-016-4049-6https://doi.org/10.1007/s00125-016-4049-6https://doi.org/10.1371/journal.pone.0116582https://doi.org/10.1371/journal.pone.0116582https://doi.org/10.1371/journal.pone.0116582https://doi.org/10.1681/ASN.2012050505https://doi.org/10.1681/ASN.2012050505https://doi.org/10.1681/ASN.2012050505https://doi.org/10.1016/j.procir.2013.01.054https://doi.org/10.1016/j.procir.2013.01.054https://doi.org/10.1016/j.procir.2013.01.054https://doi.org/10.22203/eCM.v022a21https://doi.org/10.22203/eCM.v022a21https://doi.org/10.22203/eCM.v022a21https://doi.org/10.1016/j.biomaterials.2016.01.024https://doi.org/10.1016/j.biomaterials.2016.01.024https://doi.org/10.1016/j.biomaterials.2016.01.024https://doi.org/10.1016/j.biomaterials.2009.05.024https://doi.org/10.1016/j.biomaterials.2009.05.024https://doi.org/10.1016/j.biomaterials.2009.05.024https://doi.org/10.1371/journal.pone.0084941https://doi.org/10.1371/journal.pone.0084941https://doi.org/10.1371/journal.pone.0084941https://doi.org/10.1089/ten.tea.2008.0403https://doi.org/10.1089/ten.tea.2008.0403https://doi.org/10.1089/ten.tea.2008.0403https://doi.org/10.1089/ten.tea.2008.0403https://doi.org/10.1371/journal.pone.0136681https://doi.org/10.1002/bit.22162https://doi.org/10.1002/bit.22162https://doi.org/10.1002/bit.22162https://doi.org/10.1016/j.cell.2007.08.006https://doi.org/10.1016/j.cell.2007.08.006https://doi.org/10.1016/j.cell.2007.08.006https://doi.org/10.1016/j.biomaterials.2003.10.052https://doi.org/10.1016/j.biomaterials.2003.10.052https://doi.org/10.1016/j.biomaterials.2003.10.052https://doi.org/10.1002/adma.201405076https://doi.org/10.1002/adma.201405076https://doi.org/10.1002/adma.201405076https://doi.org/10.1007/978-3-319-13117-7_36https://doi.org/10.1007/978-3-319-13117-7_36https://doi.org/10.1007/978-3-319-13117-7_36https://doi.org/10.1089/ten.2005.11.257https://doi.org/10.1089/ten.2005.11.257https://doi.org/10.1089/ten.2005.11.257https://doi.org/10.1039/C4LC00030Ghttps://doi.org/10.1039/C4LC00030Ghttps://doi.org/10.1039/C4LC00030Ghttps://doi.org/10.1038/nature04923https://doi.org/10.1038/nature04923https://doi.org/10.1038/nature04923https://doi.org/10.1038/nmeth.1198https://doi.org/10.1038/nmeth.1198https://doi.org/10.1038/nmeth.1198https://doi.org/10.3109/00016489309126197https://doi.org/10.3109/00016489309126197https://doi.org/10.3109/00016489309126197https://doi.org/10.1097/00007890-199204000-00036https://doi.org/10.1097/00007890-199204000-00036https://doi.org/10.1097/00007890-199204000-00036https://doi.org/10.1089/thy.2014.0537https://doi.org/10.1089/thy.2014.0537https://doi.org/10.1089/thy.2014.0537https://doi.org/10.1016/j.stem.2015.09.004https://doi.org/10.1016/j.stem.2015.09.004https://doi.org/10.1016/j.stem.2015.09.004https://doi.org/10.5966/sctm.2015-0037https://doi.org/10.5966/sctm.2015-0037https://doi.org/10.5966/sctm.2015-0037https://doi.org/10.1016/j.stem.2010.11.011https://doi.org/10.1016/j.stem.2010.11.011https://doi.org/10.1016/j.stem.2010.11.011https://doi.org/10.1016/j.stem.2010.11.011https://doi.org/10.1038/ncb3205https://doi.org/10.1038/ncb3205https://doi.org/10.1038/ncb3205https://doi.org/10.1007/978-1-4939-3628-1_11https://doi.org/10.1007/978-1-4939-3628-1_11https://doi.org/10.1007/978-1-4939-3628-1_11https://doi.org/10.1053/ejvs.2001.1316https://doi.org/10.1053/ejvs.2001.1316https://doi.org/10.1053/ejvs.2001.1316https://doi.org/10.1103/PhysRevLett.69.2013https://doi.org/10.1103/PhysRevLett.69.2013https://doi.org/10.1103/PhysRevLett.69.2013https://doi.org/10.1103/PhysRevE.47.2128https://doi.org/10.1103/PhysRevE.47.2128https://doi.org/10.1103/PhysRevE.47.2128

IntroductionMaterials and methodsReagents and materialsComplete culture mediumCollagen isolation from rat tails and collagen gel preparationGeneration of spheroidsKinetics of spheroid fusionBioprinting processCulture of printed thyroid constructsImmunolabelingGeneration of 131I-induced hypothyroidism mouse model and transplantation of bioprinted mouse thyroid gland constructsBody temperature measurementPlasma T4 level measurements

ResultsMathematical modeling and computer simulation of bioprinted thyroid glandGeneration of tissue spheroids from mouse thyroid and allantoidesFusion of thyroid explants and allantoides in a hanging drop and within collagen hydrogelBioprinting using mouse thyroid and allantoic spheroidsEndothelial cells from allantoic spheroids are responsible for the vascularization of thyroid gland constructBioprinted mouse thyroid gland constructs are functional in vivo

DiscussionConclusionAcknowledgmentsAuthor contributionsReferences