Tissue Engineering by Self-Assembly of Cells Printed into Topologically Defined Structures

Tissue Engineering by Self-Assembly of Cells Printed

into Topologically Defined Structures

KAROLY JAKAB, Ph.D.,1 CYRILLE NOROTTE,1 BROOK DAMON, Ph.D.,1

FRANCOISE MARGA, Ph.D.,1 ADRIAN NEAGU, Ph.D.,1,2 CYNTHIA L. BESCH-WILLIFORD, Ph.D.,3

ANATOLY KACHURIN, Ph.D.,4 KENNETH H. CHURCH, Ph.D.,5 HYOUNGSHIN PARK, Ph.D.,6

VLADIMIR MIRONOV, M.D., Ph.D.,7 ROGER MARKWALD, Ph.D.,7

GORDANA VUNJAK-NOVAKOVIC, Ph.D.,8 and GABOR FORGACS, Ph.D.1,9,10

ABSTRACT

Understanding the principles of biological self-assembly is indispensable for developing efficient strategiesto build living tissues and organs. We exploit the self-organizing capacity of cells and tissues to constructfunctional living structures of prescribed shape. In our technology, multicellular spheroids (bio-ink par-ticles) are placed into biocompatible environment (bio-paper) by the use of a three-dimensional deliverydevice (bio-printer). Our approach mimics early morphogenesis and is based on the realization that thegenetic control of developmental patterning through self-assembly involves physical mechanisms. Three-dimensional tissue structures are formed through the postprinting fusion of the bio-ink particles, inanalogy with early structure-forming processes in the embryo that utilize the apparent liquid-like behaviorof tissues composed of motile and adhesive cells. We modeled the process of self-assembly by fusion of bio-ink particles, and employed this novel technology to print extended cellular structures of various shapes.Functionality was tested on cardiac constructs built from embryonic cardiac and endothelial cells. Thepostprinting self-assembly of bio-ink particles resulted in synchronously beating solid tissue blocks,showing signs of early vascularization, with the endothelial cells organized into vessel-like conduits.

INTRODUCTION

GENES, THROUGH DIFFERENTIATION, SET UP the inherent

physical and chemical properties of cells, extracellular

matrix, and tissues. These in turn generate forces that drive

structure formation and cause subsequent alterations in gene

activity. It is this delicate interplay of genetic, molecular,

and physical factors that underlies the evolving modern

understanding of morphogenesis as a self-assembly pro-

cess.1–3 To paraphrase Whitesides and Grzybowski,4 self-

assembly is the autonomous organization of components,

from an initial state into final pattern or structure without

1Department of Physics, University of Missouri, Columbia, Missouri.2Department of Biophysics, Victor Babes University of Medicine and Pharmacy, Timisoara, Romania.3Department of Veterinary Pathobiology, University of Missouri, Columbia, Missouri.4vaxDesign, Inc., Orlando, Florida.5nScrypt, Inc., Orlando, Florida.6Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts.7Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina.8Department of Biomedical Engineering, Columbia University, New York, New York.9Department of Biology, University of Missouri, Columbia, Missouri.10Department of Biomedical Engineering, University of Missouri, Columbia, Missouri.

TISSUE ENGINEERING: Part AVolume 14, Number 3, 2008# Mary Ann Liebert, Inc.DOI: 10.1089/tea.2007.0173

413

external intervention. This definition clearly applies to

processes taking place in living organisms, in particular

during early morphogenesis. We have demonstrated that

biological self-assembly can be employed to construct liv-

ing structures of prescribed geometry.5 Our approach re-

lies on morphogenetic driving forces known to operate in

early embryonic development, specifically on the liquid-like

properties of tissues composed of motile and adhesive

cells.6–9 In particular, aggregates of such cells spontaneously

round up,10 and upon contact, fuse, similarly to coalescing

liquid drops.5,11 We used our earlier findings to develop a

novel tissue engineering technology with sound scientific

underpinning rooted in developmental biology that has

a number of advantages when compared with classical

approaches.

The basic idea underlying classical tissue engineering12 is

to seed living cells into biocompatible and eventually bio-

degradable environment (the scaffold), and then culture the

system in a bioreactor so that the initial cell population can

expand into a tissue.13 With an appropriate scaffold that

mimics the biological extracellular matrix, it is expected that

the developing tissue will adopt both the form and function

of the desired organ. The organ would then be implanted into

the recipient.14 This approach has lead to some spectacular

results.15,16 Despite these successes, the classical tissue en-

gineering approach faces serious hurdles. Selection of the

ideal biomaterial scaffold for a given cell type is problem-

atic17 and has been accomplished to date mostly by trial and

error. Even if the right biomaterial is available, achiev-

ing high enough cell density to construct a viable tissue is

extremely time consuming. Preshaping the scaffold may

present further difficulties. Recent efforts have concentrated

on scaffoldless tissue engineering18–21 and produced prom-

ising results even if they are limited to specific applications

and are not scalable. A major unresolved problem in tissue

engineering that hampers its commercial viability is that no

technology exists at present to produce off-the-shelf prod-

ucts. To address some of these issues, it has been suggested

that the process of building three-dimensional (3D) biolog-

ical structures could significantly be accelerated by the

technology of bioprinting: the automated, computer-aided

deposition of cells and cell aggregates.22 Commercially

available inkjet printers have been successfully redesigned

or new ones built to specifically deliver biological material

into scaffold fabricated according to a computer-aided de-

sign template.23,24 Pressure-operated mechanical extruders

have been developed to handle biomolecular assemblies or

live cells.25–28 The novel technology described in this

work is compatible with methods of rapid prototyping, is

scalable, exploits intrinsic properties of cells and tissues,

and makes it possible to reliably and reproducibly build

organoids of defined topology and functionality in vitro.

The 3D structures are arrived at via a three-phase process:

(i) preprocessing, or bio-ink preparation; (ii) processing, that

is, the actual automated delivery/printing of the bio-ink

particles into the bio-paper by the bio-printer; and (iii)

postprocessing, that is, the maturation/incubation of the

printed construct in the bioreactor. Final structure formation

takes place during postprocessing via the fusion of the bio-

ink particles (hence ‘‘ink’’).

MATERIALS AND METHODS

Cushion tissue fusion

Leghorn chicken eggs (Ozark Hatcheries, Neosho, MO)

were incubated at 408C in 80% humidity for 4–5 days.

Excised cushion tissue explants were washed in phosphate-

buffered saline (PBS) and cut into similar-size fragments.

Fragments were incubated in gyratory shaker (at 120 rpm,

5% CO2, 378C). Within 24–36 h (depending on size) this

procedure reproducibly yielded round aggregates. For

in vivo fusion studies, the eggs were cut open, embryos

were placed in Earle’s balanced salt solution (EBSS; Invi-

trogen, Carlsbad, CA), and the heart was dissected. Fusing

atrio-ventricular (AV) cushion tissue buds of approximately

400 mm were pinched from the opened myocardial tube at

regular time intervals between Hamburger-Hamilton (HH)

stages 26 and 28 (5–6 days), washed in PBS, placed in a

Petri dish with fresh EBSS, and photographed with a charge-

coupled device camera (Diagnostic Instruments, Sterling

Heights, MI) attached to a dissecting microscope (Olympus,

Center Valley, PA).

Preparation of bio-ink particles

and cardiac constructs

AV tissue fragments from 9-day-old Leghorn chicken

embryos were dissociated into single cells, then plated in

tissue culture dishes in Dulbecco’s modified Eagle’s me-

dium (DMEM) (Invitrogen) supplemented with 10% fetal

bovine serum (FBS) (US Biotechnologies, Parkerford, PA),

and allowed to grow to confluence. Human vascular endo-

thelial cells (CRL1730) were purchased from ATCC

(Manassas, VA) and grown in F12K medium (Invitrogen)

supplemented with 10% FBS, 0.1mg/mL heparin (Sigma,

St. Louis, MO), and 0.03mg/mL endothelial cell growth

supplement (Upstate, Lake Placid, NY). To prepare spheri-

cal cell aggregates, cell cultures were washed twice with

Hank’s balanced salt solution containing 2mM calcium

chloride, and treated with 0.1% Trypsin for 10min. Cardiac

and endothelial cells were counted using a hemacytometer,

mixed in 4:1 ratio, and then centrifuged at 2500 rpm. Cul-

tures of endothelial cells between seven and nine passage

doublings were used. The resulting pellet was aspirated into

capillary micropipettes of 500 mm diameter and incubated at

378C with 5% CO2 for 10min. The micropipette was

transferred into an in-house-built special apparatus, which

extruded the cellular ‘‘sausage’’ and cut it into equal-size

cylinders (height-to-diameter ratio¼ 1). Cellular cylinders

414 JAKAB ET AL.

rapidly rounded into spheres. Spheroids were aspirated into

the capillary micropipettes to form the bio-ink cartridge.

Similar procedure was followed with the Chinese hamster

ovary (CHO) cells.

Preparation of bio-paper

Rat tail collagen type 1 (Sigma) was dissolved in 1M

acetic acid, the pH of the solution was adjusted to 7.2

with 2M sodium hydroxide and HEPES buffer, and then

the resulting solution was diluted to 1mg/mL with

DMEM–F12K mixture. Vascular endothelial growth factor

(VEGF165, Sigma) was added to the gel before polymeri-

zation (20 ng/mL).

Visualization

To visualize details of fusion of bio-ink particles made of

CHO cells or mixture of cardiac and endothelial cells, these

were stained with red and green membrane-intercalating

dyes (PKH26 and PKH2, Sigma) according to the proto-

col supplied by the manufacturer. Cells were replated af-

ter labeling to recover from the staining procedure and

then used to make aggregates of different fluorescent col-

ors. Transparency of the collagen gels permitted the use of

bright-field imaging to monitor the assembling structures.

Histology

Cell–gel constructs were fixed in paraformaldehyde

(PFA). PFA was subsequently washed out. Samples were

dehydrated and prepared for immunohistochemistry (sec-

tioning into 5-mm-thick layers, paraffin embedding, etc.).

Immunohistology was performed to assess cellular distri-

bution in the constructs using hematoxylin and eosin (H&E);

evaluate the expression of cardiac proteins, using troponin I

antibody (primary: rabbit anti-troponin antibodies; second-

ary: anti-rabbit immunoglobulin G conjugated with FITC

[fluorescein isothiocyanate; Chemicon International, Inc.,

Temecula, CA]); and visualize the pattern of endothelial

cells, using CD31 antibody (primary: mouse anti-human

CD31 monoclonal antibody; secondary: EnVisionþ, a horse-

radish peroxidase–labeled polymer conjugated with anti-

mouse antibodies [DAKO, Glostrup, Denmark]).

Contractile behavior

Functionality of the engineered cardiac constructs was

evaluated by monitoring contractile activity at 10�magni-

fication using a Nikon Diaphot microscope (Nikon Instru-

ments, Inc., Melville, NY) as in our previous studies.29

To measure the amplitude of contractions, video-recorded

beating sequences were digitized at the rate of 30 frames/

second. The en-face area of each construct was measured as

function of time using image analysis (Scion Image soft-

ware). Amplitude of contraction was expressed as frac-

tional area change. Decrease in the total en-face area was

used as a measure of the contraction amplitude to avoid

determining the axis of maximal shortening in a construct

of an irregular rectangular shape.

RESULTS

Early morphogenesis based on tissue liquidity

To demonstrate that the outlined tissue engineering

program indeed is executed similarly to morphogenetic

processes, we specifically considered the in vitro, midline

fusion of embryonic chicken cardiac cushions, the process

leading to septation of the heart into the four chambers and

valve formation.30 We dissected early chicken embryonic

heart to extract AV cushions at the stage when their in vivo

fusion would commence (HH 26, day 5). When irregularly

shaped tissue explants were isolated and incubated in cul-

ture medium, they naturally and progressively rounded into

almost perfect spherical shape (Fig. 1A). This observation

suggests that these tissue spheroids have properties in

common with liquid droplets in that they try to minimize

their interfacial area with the surroundings. Indeed, tissues

composed of adhesive and motile cells have measurable

surface tensions and viscosities.1,10,31–34

When two similar cushion tissue spheroids were placed

contiguously, they fused into a single spheroid (Fig. 1B).

The in vitro fusion process resembled the in vivo process

(Fig. 1C) of AV septum formation during heart develop-

ment.30 Monte Carlo (MC) simulations35 based on a model

of binary liquids5,26 were in excellent agreement with ex-

perimental results (Fig. 1D). The finding that the function

r2¼A(1� e�t=t) fitted well the time variation of the cir-

cular interfacial area (of instantaneous radius r) between

the two spherical fragments (Fig. 1E) gave further support

to the notion of tissue liquidity. This expression is consis-

tent with the analytical result r2¼sRt=(Z) for the fusion of

droplets of highly viscous materials (Z: viscosity; s: inter-facial tension between the liquid and the embedding me-

dium; R: initial radius of the fusing drops), valid in the limit

t/t�136 and relates the relaxation time t, characterizing therate of fusion, to the material properties of the tissue:

t¼22=3RZ=s. (Here we used the relation A¼ 22=3R2, which

follows from the conservation of volume during fusion.)

The exponential growth of the interfacial area is similar to

experimental results obtained in real liquids.11 Measure-

ment of cushion tissue–culture medium interfacial tension

with earlier described methods34 resulted in s � 16:0 dyn/

cm (0.016N/m). This value, when combined with the ex-

pression and experimentally obtained information for t,leads to Z � 107 Poise for chicken embryonic cushion tis-

sue, a value consistent with earlier estimates obtained with

an independent method for various embryonic chicken

tissues.33 The exponential function r2 ¼ A(1� e�MCS=t)

described accurately the evolution of fusion also within the

TISSUE ENGINEERING BY PRINTING CELL AGGREGATES 415

× 104

0 250 500 750

Time (min)

1000 1250 1500 000

500

MCS/103

1000 1500 2000

50

100

150

200

250

300

350

400

(Cro

ss s

ecti

on

rad

ius

/ Cel

l dia

met

er)2

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5 f

r2 (µm

2 )

e

FIG. 1. Liquid-like properties of embryonic cushion tissue. (A) An irregular tissue fragment, excised from a day 5 (HH 26) chick

embryo, rounds up into a spheroid in about 24 h. (B) Two apposed spheroids in culture medium, in hanging drop configuration (as well

as in collagen; not shown), fuse in time, similarly to liquid drops. (C) In vivo cushion tissue fusion during chicken heart development.

Panels (from the left: 5 day, HH 26 embryo) show the gradual blending of the two superior AV cushion tissues (middle: 5.5 day, HH 27).

Fusion is complete by day 6.0 (HH 28; right panel). Scale bars: 100mm. (D) MC simulation of the fusion of two model cellular

aggregates. Each initial aggregate is of 30 cell diameters. The simulation is based on a lattice gas model of a binary liquid, the two

components being the cellular material and the surrounding tissue culture medium (for details see the works by Jakab et al.5 and Neagu

et al.26). Pattern evolution is driven by the minimization of the interfacial energy between the two components, which can conveniently

be expressed in terms of a single model parameter, the interfacial tension g between the two liquids. The figure shows the fusion pattern

for g ¼ 0:5 (in units of the biological fluctuation energy,42 the analogue of the thermal energy, kBT , with kB and T being Boltzmann’s

constant and the absolute temperature, respectively). (E) The interfacial circular area between the two fusing cellular spheroids [shown

in (B)] increases with time, and eventually saturates. The fit to the data points is exponential with A¼ 46,414mm2, t ¼ 387:2min (see

discussion in section ‘‘Early morphogenesis based on tissue liquidity’’). The inset shows schematically the quantity plotted along the

vertical axis. (F) The analogue of (E) in the MC simulation. Markers depict the mean value of 30 simulations, which only differ by the

seed of the pseudo-random number generator used in MC. The length of the symmetric error bars is twice the standard deviation. The

solid line is the corresponding theoretical exponential curve (see discussion in section ‘‘Early morphogenesis based on tissue liquidity’’),

with A ¼ 379:6 and t ¼ 286:65· 103 MCS, and time t replaced with MCS.

416

MC simulation as the process approached its completion

but not at its initiation (Fig. 1F). To simulate the fusion pro-

cess, time has been replaced by Monte Carlo Steps (MCSs),

one step being defined as a sequence of computations, which

give each interfacial cell a chance to experience a move

(dictated by the random process implemented in MC sim-

ulations5,35). This is to be expected since in MC pattern

evolution is driven by energy minimization, thus being

more accurate near equilibrium than far from it.

These results imply that the time evolution of the fusion

pattern is indeed similar to that of coalescing liquid drops.

Our findings suggest that fusion of early (primitive) cardiac

cushions during heart development, once under way, may

be driven by the inherent physical properties of the tissue.37

The novel tissue engineering technology

Tissue liquidity, as described above, provides the sci-

entific basis for our tissue engineering technology to

build complex biological structures. The components of this

a

f b

c

d

e

FIG. 2. Components of the novel technology. (A) A patent-

pending device (schematic) cuts the earlier prepared tissue/

cellular slurry into cylindrical aggregates with aspect ratio 1.

(B) The spheroidal bio-ink particles result from the rapid rounding

of the tissue/cellular cylinders upon incubation (the one shown is

of 500 mm and is made of CHO cells fluorescently labeled with

a membrane dye). (C) The printer cartridge is a micropipette

housing the spherical multicellular aggregates. (D) The printer.

The one shown has been designed and built by nScrypt (Orlando,

FL). It is a fully computer-controlled delivery device. Three-

dimensional deposition of bio-ink particles (with accuracy of mm)

is achieved by the movable x–y stage and three z-directional

printing heads. Two of these are used to print the spheroids, which

are extruded from the cartridge/micropipette by the positive dis-

placement of a piston within the micropipette. The third unit is

pressure operated and is used to print the embedding hydrogel

(e.g., collagen gel). Each extruder is equipped with a camera,

providing full visual control of printing. (E) Spheroids are de-

livered one by one into the hydrogel according to a computer

script encoding the printable shape. Note that the hydrogel itself

is also printed, here a 1�1�0.2 cm layer of collagen type 1 gel.

(F) The outcome of the printing process is a set of discrete

spheroids, here 500mm aggregates of CHO cells, fluorescently

labeled with red or green membrane dyes.

a

d

e6

Am

plit

ude (

µm)

4 < A > = 1.78 µm < T > = 1.84 s

2

0

0 5 10 15

Time (s)

20 25 30

b c

FIG. 3. Printing cellular structures of defined topology.

(A) Upper panel: a circular arrangement of bio-ink particles,

composed of spherical aggregates of CHO cells (neighboring

aggregates were labeled alternatively with red and green dyes).

Lower panel: a fused circle, after about 60 h. (B) The time evo-

lution of a vertically printed three-layered tubular structure of

CHO cells, built from circles shown in panel (A). (C) Upper

panel: a horizontally printed tube of 12 layers composed of hu-

man umbilical smooth muscle cells. Lower panel: a horizontally

printed branching CHO tube. (D) Building a pulsatile cardiac

construct. A 6�6 array of cardiac bio-ink particles (without en-

dothelial cells; left) printed into 1mg/mL collagen 1 bio-paper. If

endothelial cells are present (middle panel; labeled with red dye;

cardiac cells labeled with green dye), they first accumulate at the

boundaries between neighboring aggregates and later in the

originally acellular region between four aggregates (although

many remain also inside the aggregates). In a 6�6 array, fusion

is complete in about 70 h (right panel). (E) The synchronous

beating of the fused construct in (D). <A> and <T> are the

amplitude and period of beating, respectively. All bio-ink particles

in this figure are of 500 mm diameter. The colored panels [in

(A) and (D)] were prepared by fluorescently labeling the cells with

red or green membrane dyes.

TISSUE ENGINEERING BY PRINTING CELL AGGREGATES 417

technology are shown in Figure 2. We first prepared

monodisperse spherical bio-ink particles (Fig. 2A, B) with

composition dictated by the properties of the desired struc-

ture. The bio-ink particles were packaged into cartridges

(Fig. 2C; micropipettes of 300–500 mm inner diameter).

Cartridges were inserted into the bio-printer (Fig. 2D) and

delivered into the bio-paper/hydrogel (Fig. 2E) according to

a computer script that encodes the shape of the structure to

be printed (Fig. 2F).

Building organoids of definite topology

Printing results in a set of discrete biological structures,

which subsequently undergo fusion into continuous struc-

tures. As a proof of concept, we have applied this technology

to print bio-ink particles in the form of circles (Fig. 3A),

cylinders (Fig. 3B, C), or sheets (Fig. 3D), which upon fu-

sion yielded constructs of specific shape, in particular hol-

low tubes and solid blocks (Fig. 3B–D). To illustrate the

flexibility of the method, we prepared tubular structures in

two different ways: vertical and horizontal. In the first case

(Fig. 3B), bio-ink particles are delivered in circular pattern,

and circles are printed layer by layer in the vertical direction.

In the second case (Fig. 3C), chains of bio-ink particles are

printed again layer by layer, with the number of chains

varying in each layer. Horizontally printed cylinders can be

arranged to arrive at branching tubular structures upon fu-

sion of the bio-ink particles (Fig. 3C). (Once the right con-

ditions for structure formation had been established [see

‘‘Discussion’’ section], reproducibility was verified by per-

forming the relevant experiments at least four times.)

Functionality of the printed tissue constructs

To verify that postprinting structure formation leads

to functional tissue constructs, we prepared multicellular

spheroids employing fragments of 9-day chicken embryonic

heart tissue, containing various cell types. In particular, the

fragments contained the fused AV cushions, whose mesen-

chyme at this stage begins to differentiate into endothelial

and fibrous (fibroblastic) tissue and, uniquely in the chick,

also into myocardial cells (e.g., the large muscular leaflet of

the triscupid valve38,39). We used the cells of the dissociated

fragments to prepare 500 mm spherical aggregates with or

without endothelial cells (human umbilical endothelial cells,

randomly dispersed in the spheroids in the presence of

VEGF). We then printed a 6�6 array of the spheroids in a

collagen type 1 gel bio-paper (Fig. 3D). After deposition the

aggregate–gel construct was incubated in tissue culture

medium for 5 days. The bio-ink particles completely fused

over a period of 70 h to form a thick graft. The engineered

tissue manifested characteristic properties of authentic

chicken cardiomyocytes as revealed by its synchronous

beating after about 90 h (Fig. 3E) and immunohistochemis-

try (Fig. 4).

Vascularization of the printed tissue constructs

Long-term survival and function of engineered tissues can

be maintained only with appropriate vascularization.40 We

hypothesized that if printing of the cardiac construct, as de-

scribed above, is repeated with aggregates composed of a

mixture of cardiac and endothelial cells (human umbilical

vein endothelial cell), self-organizing properties will result

in the formation of primitive vasculature. Thus, a 6�6 array

(similar to that shown in Fig. 3D) was printed onto the col-

lagen gel bio-paper. VEGF, an endothelial mitogen and

vasculogenic stimulator, was added to the collagen solution

before gelation. For the postprinting incubation, the culture

medium was also supplemented with VEGF. Synchronous

beating was again observed, after approximately 90 h. Con-

structs that formed without endothelial cells (Fig. 4A–C)

were always compact and had cells present throughout

their entire volume. However, tissue constructs that formed

with endothelial cells (Fig. 4D) consistently showed regions

void of cells with internal boundaries notably decorated

with characteristically elongated cells (Fig. 4E), as well as

conduit-like structures (Fig. 4F). The formation of such

structures seems to be facilitated by the intrinsic features of

the method, the acellular regions between four neighboring

spherical aggregates, and the boundaries between pairs of

aggregates, where endothelial cells preferentially accumu-

late (middle panel in Fig. 3D). Further, cells with either nu-

merous intracellular vacuoles or a single large internal

vacuole, characteristic for lumen-forming endothelial cells,41

could be observed (Fig. 4F). Such structures were never

observed in the constructs without endothelial cells. Immu-

nohistochemistry provided evidence that these cells and those

lining the periphery of the hollow acellular regions indeed

were endothelial cells (Fig. 4G–J). Structures suggestive of

forming capillaries could also be discerned (Fig. 4I, J).

DISCUSSION

This work had four objectives. First, to further demon-

strate that embryonic tissues (or, more generally, cell ag-

gregates composed of adhesive and motile cells) have

liquid-like properties. In particular, we showed both exper-

imentally and by modeling that this concept applies not only

to equilibrium cellular arrangements (such as those observed

in sorting10,42) but also to time-dependent phenomena, such

as the fusion of two tissue fragments (Fig. 1). Second, we

wished to provide an example of a true early morphogenetic

phenomenon that can be interpreted in terms of tissue li-

quidity. For this, we considered the in vivo fusion of em-

bryonic chicken cardiac cushions, the hallmark of cardiac

development. We provided evidence that this morphoge-

netic process indeed is similar to its in vitro counterpart,

which in turn obeys the laws of fluid mechanics (Fig. 1).

Third, we described a novel tissue engineering technology,

418 JAKAB ET AL.

bioprinting that allows the reproducible building of 3D liv-

ing structures of desired geometry (Fig. 2). The technology

has firm theoretical and experimental foundation. It employs

spherical bio-ink particles, with the composition consistent

with that of the tissue construct to be built. These are de-

livered by automated methods into a biocompatible envi-

ronment, the bio-paper. Thus, the approach represents an

example of rapid prototyping. Structures form by self-

assembly, in the course of postprinting fusion of the liquid-

like bio-ink particles in a process akin to structure formation

in early development. The method is robust: errors in de-

position (up to a limit) are self-corrected during structure

formation (similarly to early development). (Such an error

occurs in the top ring of the initial construct in Figure 3B,

which by 24 h is barely noticeable.) The success of the

technology depends strongly on the bio-paper. It should be a

soft enough gel to allow the smooth deposition of cellular

aggregates. It should provide the appropriate environment

for fusion and thus for cell movement. Cell movement

should occur at the ‘‘right’’ rate at which the desired struc-

ture forms before the cellular system collapses into a single

large spherical aggregate (the lowest energy state). The

desired structure should thus be sufficiently long-lived to

allow the elimination of the bio-paper once fusion is com-

plete. These requirements are not too restrictive. (Indeed, we

were able to produce the same structure with several bio-

papers.5) Importantly, they clearly distinguish the bio-paper

from the scaffold used in classical tissue engineering. MC

simulations provided strong evidence that tissue liquidity

indeed is the mechanism driving structure formation via the

fusion of the bio-ink particles (Fig. 1). In particular, some of

the above-listed conditions imposed on the bio-paper can be

fulfilled in the simulations by the tuning of the bio-paper–

bio-ink interfacial tension, g. It is, however, important

to realize that tissue liquidity and g provide merely a use-

ful analogy that can conveniently be employed for the

FIG. 4. Histology of the 3D printed cardiac constructs. (A) Troponin 1 staining (green; blue is nuclear staining) of a construct without

endothelial cells (inset: higher magnification). The fused sheet was sectioned perpendicular to its plane (sagittally). Upon fusion of the

originally 500mm aggregates, the thickness of the final construct is slightly less than 500 mm. (It is interesting that the troponin positive

cardiomyocytes sort to the upper part of the sheet, in direct contact with collagen.) (B) H&E staining of a sagittally sectioned construct

without endothelial cells. (C) and (D) H&E staining of a construct sectioned parallel to its plane, respectively without and with

endothelial cells. Constructs with endothelial cells show regions void of cells. (E) Higher magnification images of H&E-stained samples

reveal that the periphery of regions void of cells in (D) is decorated with cells with morphology typical for endothelial cells lining the

lumen of blood vessels. (F) Vacuolization. The arrowhead and arrows point at regions suggestive of cells with one or several vacuoles,

respectively. (G–J) CD31 staining for endothelial cells (brown) confirmed the suggestive results of the H&E staining. Vacuolization

indeed takes place in endothelial cells (G), and they indeed line the regions void of cells (H, I). Arrows in (I) and (J) point at structures

reminiscent of sprouting capillary vessels. Scale bars: 100 mm in (A, B, I, J); 200 mm in (C, D); 50mm in (E, F, G); 400 mm in (H).

TISSUE ENGINEERING BY PRINTING CELL AGGREGATES 419

characterization of the outlined tissue engineering method.

Ultimately, postprinting structure formation is governed by

molecular mechanisms underlying cell–cell and cell–matrix

adhesion, as well as cell motility. (Indeed, g can be related tothe density and nature of cell–cell31,32 and cell–matrix ad-

hesion molecules.43) Finally, our forth objective was to

present specific examples of structures built using the novel

technology (Fig. 3) and provide evidence that these tissue

structures manifest the functionality of the real tissue they

are supposed to replace. We showed that building small-

diameter tubes (major structures in complex organisms)

or extended 3D grafts can easily be accomplished. Sig-

nificantly, the self-organizing capacity of cells assures the

onset of vascular network formation within the 3D con-

structs (Fig. 4), even without their implantation into a living

organism.44–47 Our work implies that functional tissue

constructs containing multiple cell types could be built by

rapid prototyping (i.e., bio-printing), suggesting that the off-

the-shelf availability of organoid-like structures could be

achieved.

ACKNOWLEDGMENTS

Support for this research was provided by the National

Science Foundation FIBR program (EF-0526854 to VM,

RM, KC, GF), the National Institutes of Health (P41

EB002520 and RO1 HL076485 to GV-N) and the Medical

University of South Carolina Bioprinting Research Center

(to VM).

REFERENCES

1. Forgacs, G., and Newman, S.A. Biological Physics of the

Developing Embryo. Cambridge: Cambridge University Press,

2005.

2. Newman, S.A. Generic physical mechanisms of tissue mor-

phogenesis: a common basis for development and evolution.

J. Evol. Biol. 7, 467, 1994.

3. Wolpert, L. Principles of Development. Oxford: Oxford

University Press, 1998.

4. Whitesides, G.M., and Grzybowski, B. Self-assembly at all

scales. Science 295, 2418, 2002.

5. Jakab, K., Neagu, A., Mironov, V., Markwald, R.R., and

Forgacs, G. Engineering biological structures of prescribed

shape using self-assembling multicellular systems. Proc. Natl.

Acad. Sci. USA 101, 2864, 2004.

6. Godt, D., and Tepass, U. Drosophila oocyte localization is

mediated by differential cadherin-based adhesion. Nature 395,

387, 1998.

7. Gonzalez-Reyes, A., and St. Johnston, D. Patterning of the

follicle cell epithelium along the anterior-posterior axis dur-

ing Drosophila oogenesis. Development 125, 2837, 1998.

8. Hayashi, T., and Carthew, R.W. Surfacemechanics mediate pat-

tern formation in the developing retina. Nature 431, 647, 2004.

9. Steinberg, M.S., and Poole, T.J. Liquid behavior of embry-

onic tissues. In: Bellairs, R., Curtis, A.S.G., and Dunn, G.,

eds. Cell Behaviour. Cambridge: Cambridge University Press,

1982, p. 583.

10. Foty, R.A., Pfleger, C.M., Forgacs, G., and Steinberg, M.S.

Surface tensions of embryonic tissues predict their mutual

envelopment behavior. Development 122, 1611, 1996.

11. Yao, W., Maris, H.J., Pennington, P., and Seidel, G.M.

Coalescence of viscous liquid drops. Phys. Rev. E Stat.

Nonlin. Soft Matter Phys. 71, 016309, 2005.

12. Langer, R., and Vacanti, J.P. Tissue engineering. Science 260,

920, 1993.

13. Khademhosseini, A., Langer, R., Borenstein, J., and Vacanti,

J.P. Microscale technologies for tissue engineering and

biology. Proc. Natl. Acad. Sci. USA 103, 2480, 2006.

14. Griffith, L., and Naughton, G. Tissue engineering—current

challenges and expanding opportunities. Science 295, 1009,

2002.

15. Atala, A., Bauer, S.B., Soker, S., Yoo, J.J., and Retik, A.B.

Tissue-engineered autologous bladders for patients needing

cystoplasty. Lancet 367, 1241, 2006.

16. Oberpenning, F., Meng, J., Yoo, J.J., and Atala, A. De novo

reconstitution of a functional mammalian urinary bladder by

tissue engineering. Nat. Biotechnol. 17, 149, 1999.

17. Dahl, S.L., Rhim, C., Song, Y.C., and Niklason, L.E. Me-

chanical properties and compositions of tissue engineered and

native arteries. Ann. Biomed. Eng. 35, 348, 2007.

18. L’Heureux, N., Dusserre, N., Konig, G., Victor, B., Keire, P.,

Wight, T.N., Chronos, N.A., Kyles, A.E., Gregory, C.R.,

Hoyt, G., Robbins, R.C., and McAllister, T.N. Human tissue-

engineered blood vessels for adult arterial revascularization.

Nat. Med. 12, 361, 2006.

19. L’Heureux, N., Paquet, S., Labbe, R., Germain, L., and Au-

ger, F.A. A completely biological tissue-engineered human

blood vessel. FASEB J. 12, 47, 1998.

20. McGuigan, A.P., and Sefton, M.V. Vascularized organoid

engineered by modular assembly enables blood perfusion.

Proc. Natl. Acad. Sci. USA 103, 11461, 2006.

21. Shimizu, T., Yamato, M., Isoi, Y., Akutsu, T., Setomaru, T.,

Abe, K., Kikuchi, A., Umezu, M., and Okano, T. Fabrication

of pulsatile cardiac tissue grafts using a novel 3-dimensional

cell sheet manipulation technique and temperature-responsive

cell culture surfaces. Circ. Res. 90, e40, 2002.

22. Mironov, V., Boland, T., Trusk, T., Forgacs, G., and Mark-

wald, R.R. Organ printing: computer-aided jet-based 3D tis-

sue engineering. Trends Biotechnol. 21, 157, 2003.

23. Boland, T., Xu, T., Damon, B., andCui, X. Application of inkjet

printing to tissue engineering. Biotechnol. J. 1, 910, 2006.

24. Sun, W., Darling, A., Starly, B., and Nam, J. Computer-aided

tissue engineering: overview, scope and challenges. Bio-

technol. Appl. Biochem. 39, 29, 2004.

25. Landers, R., Hubner, U., Schmelzeisen, R., and Mulhaupt, R.

Rapid prototyping of scaffolds derived from thermoreversible

hydrogels and tailored for applications in tissue engineering.

Biomaterials 23, 4437, 2002.

26. Neagu, A., Jakab, K., Jamison, R., and Forgacs, G. Role of

physical mechanisms in biological self-organization. Phys.

Rev. Lett. 95, 178104, 2005.

27. Smith, C.M., Christian, J.J., Warren, W.L., and Williams,

S.K. Characterizing environmental factors that impact the

viability of tissue-engineered constructs fabricated by a

direct-write bioassembly tool. Tissue Eng. 13, 373, 2007.

420 JAKAB ET AL.

28. Smith, C.M., Stone, A.L., Parkhill, R.L., Stewart, R.L., Simp-

kins, M.W., Kachurin, A.M., Warren, W.L., and Williams, S.K.

Three-dimensional bioassembly tool for generating viable tis-

sue-engineered constructs. Tissue Eng. 10, 1566, 2004.

29. Radisic, M., Park, H., Shing, H., Consi, T., Schoen, F.J.,

Langer, R., Freed, L.E., and Vunjak-Novakovic, G. Func-

tional assembly of engineered myocardium by electrical

stimulation of cardiac myocytes cultured on scaffolds. Proc.

Natl. Acad. Sci. USA 101, 18129, 2004.

30. Wessels, A., and Sedmera, D. Developmental anatomy of the

heart: a tale of mice and man. Physiol. Genomics 15, 165,

2003.

31. Forgacs, G., Foty, R.A., Shafrir, Y., and Steinberg, M.S.

Viscoelastic properties of living embryonic tissues: a quan-

titative study. Biophys. J. 74, 2227, 1998.

32. Foty, R.A., and Steinberg, M.S. The differential adhesion

hypothesis: a direct evaluation. Dev. Biol. 278, 255, 2005.

33. Gordon, R., Goel, N.S., Steinberg, M.S., and Wiseman, L.L.

A rheological mechanism sufficient to explain the kinetics of

cell sorting. J. Theor. Biol. 37, 43, 1972.

34. Hegedus, B., Marga, F., Jakab, K., Sharpe-Timms, K.L., and

Forgacs, G. The interplay of cell-cell and cell-matrix inter-

actions in the invasive properties of brain tumors. Biophys.

J. 91, 2708, 2006.

35. Metropolis, N., Rosenbluth, A.W., Rosenbluth, M.N., Teller,

A.H., and Teller, E. Equation of state calculations by fast

computing machines. J. Chem. Phys. 2121, 1087, 1953.

36. Frenkel, J. Viscous flow of crystalline bodies under the action

of surface tension. J. Phys. 9, 351, 1945.

37. Sugi, Y., Yamamura, H., Okagawa, H., and Markwald, R.R.

Bone morphogenetic protein-2 can mediate myocardial reg-

ulation of atrioventricular cushion mesenchymal cell forma-

tion in mice. Dev. Biol. 269, 505, 2004.

38. van den Hoff, M.J., Kruithof, B.P., Moorman, A.F., Mark-

wald, R.R., and Wessels, A. Formation of myocardium after

the initial development of the linear heart tube. Dev. Biol.

240, 61, 2001.

39. van den Hoff, M.J., Moorman, A.F., Ruijter, J.M., Lamers,

W.H., Bennington, R.W., Markwald, R.R., and Wessels, A.

Myocardialization of the cardiac outflow tract. Dev. Biol.

212, 477, 1999.

40. Laschke, M.W., Harder, Y., Amon, M., Martin, I., Farhadi, J.,

Ring, A., Torio-Padron, N., Schramm, R., Rucker, M., Junker,

D., Haufel, J.M., Carvalho, C., Heberer, M., Germann, G.,

Vollmar, B., and Menger, M.D. Angiogenesis in tissue engi-

neering: breathing life into constructed tissue substitutes.

Tissue Eng. 12, 2093, 2006.

41. Kamei, M., Saunders, W.B., Bayless, K.J., Dye, L., Davis,

G.E., and Weinstein, B.M. Endothelial tubes assemble from

intracellular vacuoles in vivo. Nature 442, 453, 2006.

42. Beysens, D.A., Forgacs, G., and Glazier, J.A. Cell sorting is

analogous to phase ordering in fluids. Proc. Natl. Acad. Sci.

USA 97, 9467, 2000.

43. Robinson, E.E., Zazzali, K.M., Corbett, S.A., and Foty, R.A.

Alpha5beta1 integrin mediates strong tissue cohesion. J. Cell

Sci. 116, 377, 2003.

44. Kelm, J.M., Djonov, V., Hoerstrup, S.P., Guenter, C.I., Ittner,

L.M., Greve, F., Hierlemann, A., Sanchez-Bustamante, C.D.,

Perriard, J.C., Ehler, E., and Fussenegger, M. Tissue-

transplant fusion and vascularization of myocardial micro-

tissues and macrotissues implanted into chicken embryos and

rats. Tissue Eng. 12, 2541, 2006.

45. Kelm, J.M., Djonov, V., Ittner, L.M., Fluri, D., Born, W.,

Hoerstrup, S.P., and Fussenegger, M. Design of custom-

shaped vascularized tissues using microtissue spheroids as

minimal building units. Tissue Eng. 12, 2151, 2006.

46. Levenberg, S., Rouwkema, J., Macdonald, M., Garfein, E.S.,

Kohane, D.S., Darland, D.C., Marini, R., van Blitterswijk,

C.A., Mulligan, R.C., D’Amore, P.A., and Langer, R. En-

gineering vascularized skeletal muscle tissue. Nat. Bio-

technol. 23, 879, 2005.

47. Shimizu, T., Sekine, H., Isoi, Y., Yamato, M., Kikuchi, A.,

and Okano, T. Long-term survival and growth of pulsatile

myocardial tissue grafts engineered by the layering of cardi-

omyocyte sheets. Tissue Eng. 12, 499, 2006.

Address reprint requests to:

G. Forgacs, Ph.D.

Departments of Physics, Biology, and Biomedical

Engineering

University of Missouri

Columbia, MO 65211

E-mail: [email protected]

TISSUE ENGINEERING BY PRINTING CELL AGGREGATES 421