bionic human 7 questions

ISSN:1369 7021 © Elsevier Ltd 2006DECEMBER 2006 | VOLUME 9 | NUMBER 1226

Scaffolds for stem cells Today, most people know at least something about stem cells. Embryonic stem cells enjoy regular mentions in news programs and magazines, probably because of the controversial way in which they are generated but also because of their huge potential in medicine. Here we distinguish between and define types of stem cells, discuss techniques used so far to create various cells and tissues from stem cells, and discuss how three-dimensional supports and stem cells have been and should be used to encourage the development of functional replacement tissue.

Nicholas D. Evans1*, Eileen Gentleman1, and Julia M. Polak2

1Department of Materials, Royal School of Mines, Imperial College London, South Kensington, London SW7 2AZ, UK

2Tissue Engineering and Regenerative Medicine, Department of Chemical Engineering, Room 144, Roderic Hill Building, Imperial College, South

Kensington Campus, London SW7 2AZ, UK

*E-mail: [email protected]

Stem cells can be thought of as versatile, unspecialized cells that

have the potential either to divide to make more stem cells or to

differentiate into one or more cell type, usually in response to

some kind of signal. Ultimately, these cells are used in the hope

of addressing the shortfall in the quantity of tissue available for

transplantation, either alone, as may be the case for replacing lost

pancreatic beta cells in type 1 diabetes mellitus1, or in combination

with a scaffold, as may be necessary in the engineering of bone

tissue2. The term ‘stem cell’ refers to a rather confusing assortment

of different and distinct cell types all sharing this property, but for

simplicity stem cells are usually divided into adult stem cells and

embryonic stem cells.

Stem cells Adult stem cells Adult stem cells have been used for many years now, with particular

success in the treatment of cancers of the blood system. Hematologists

often point out that they pioneered stem cell biology long before the

more recent explosion in scientific and public interest3. Throughout the

1950s and 1960s, these scientists demonstrated that transplantations

of ‘hematopoietic stem cells’ (HSCs), isolated from the bone marrow,

could generate a new immune system composed of many distinct

specialist cell types in organisms in which the host immune system had

been destroyed4. This culminated in 1963 when Mathé demonstrated

the long term survival of a leukemia patient treated with HSCs5. Bone

marrow transplantation is now a routine medical procedure.

Following these successes, Friedenstein et al.6 noticed another cell

type in bone marrow explants, initially called the fibroblast colony-

forming cell because it stuck down on cell culture plastic, that was

later shown to be a stem cell7,8. They are now referred to as marrow

stromal cells or mesenchymal stem cells (MSCs) (Fig. 1). These cells

resemble cells of the connective tissue (fibroblasts) and, in contrast

to HSCs, can be grown easily in cell culture dishes. By changing the

composition of the medium in which they are grown, MSCs can

Scaffolds for stem cells REVIEW

DECEMBER 2006 | VOLUME 9 | NUMBER 12 27

be selectively differentiated into bone cells (osteocytes), fat cells

(adipocytes), and cartilage cells (chondrocytes)8. This property has made

them an attractive choice for bone and cartilage tissue engineering,

especially since they may be used to treat the person from whom they

were isolated – as an ‘autologous’ transplant. There are also numerous

examples of evidence in the literature that these cells can differentiate

into other lineages, including heart cells9 and neurons10. But they

do have limitations – they can only divide a finite number of times

(depending on the age of the donor)11, which limits their supply, and

they may accumulate genetic changes over time12.

Stem cells are also known to be distributed around the body

in various other ‘niches’. For example, neural stem cells can be

isolated from brain tissue, grown in vitro, and induced to differentiate

into the three cell types of the brain – neurons, astrocytes, and

oligodendrocytes13,14. They also appear to be capable of turning into

other cell types – after injection into a developing mouse blastocyst,

they can be found later in the adult organism in several tissues,

including heart, kidney, and liver15. Similar stem cells are also thought

to reside in other tissues as a repair mechanism against injury, for

example in the skin16. Again, however, these stem cells cannot be grown

easily in vitro and are thought to have a limited replicative capacity.

Embryonic stem cells Embryonic stem cells (ESCs), on the other hand, are renowned for their

ability to divide indefinitely and their capacity to differentiate into most,

if not all, of the tissues of the body. This makes them a potentially far

more versatile cell type than adult stem cells. In mammals, they were

first isolated in 1981 from the blastocyst of the mouse, a ball of cells

formed several days following fertilization. Because of technical and

ethical hurdles, it was not until 1998 that Thomson and colleagues17

were able to do the same in humans. Human ESCs can now be routinely

cultured as preserved ‘pluripotent’ cells, which retain their ability to

divide indefinitely in an undifferentiated state and, when stimulated

with the right signals, to differentiate into all of the tissues of the adult

(Fig. 2). This raises the exciting possibility that they may be able to

provide an unlimited source of cells for tissue replacement.

Following this advance, many groups rushed to make clinically-

important cell types using human ESCs as a starting point. It was

known that human ESCs would differentiate spontaneously into many

cell types, both in vitro in free-floating structures analagous to the

early embryo called ‘embryoid bodies’ and following implantation

into experimental animals17, but methods were needed to turn them

selectively into a cell type of interest. The simplest and most common

strategies involved simply growing ESCs in a medium designed for the

required cell type. For example, Bielby et al.18 grew human ESCs in a

medium containing β-glycerophosphate, vitamin C, and dexamethasone, which is used routinely for the growth of osteoblasts in cell culture

experiments, and demonstrated the formation of bone nodules and

cells that expressed bone-specific genes. Other researchers have used

similar methods, usually by including a variety of growth factors in the

medium, to make pancreatic β cells19, neurons20, cardiomyocytes21, lung cells22,23, and even eggs24 and sperm25. Another simple strategy

involves culturing human ESCs in the presence of the target cell

or a cell type postulated to have a role in differentiation. In this

way, Vats et al.26, Mummery et al.27 and Van Vranken et al.28 have

shown that human ESCs can be differentiated into chondrocytes,

cardiomyocytes, and pneumocytes respectively. Genetic manipulation

is also a useful technique for directing the differentiation of ES cells.

Kim and colleagues29 introduced the gene for nurr1 into ESCs, and

demonstrated differentiation into dopamine-producing neurons and

an improvement in the condition of Parkinson’s disease-afflicted rats

following their implantation. Tai et al.30 have shown that increased

numbers of osteoblasts could be produced by ESCs over-expressing

osterix. Extracellular matrix – the proteinaceous amalgam in which cells

grow – is also thought to be an important factor in the differentiation of

ESCs. For example, Coraux et al.31 showed that the culture of ESCs on a

matrix derived from skin fibroblasts could generate a tissue that looked

remarkably like real skin.

The literature on the directed differentiation of ESCs into various

cell types is now vast, with over 1600 papers published since 1998,

most of which use variations or combinations of the methods given

Fig. 1 Diagram demonstrating how adult stem cells can be used in tissue

engineering. Marrow is removed from the adult bone and placed in a culture

dish. Adherent mesenchymal stem cells can then be expanded or directed to

differentiate into bone, cartilage, or fat cells.

REVIEW Scaffolds for stem cells

DECEMBER 2006 | VOLUME 9 | NUMBER 1228

above. Despite this, however, the induction of differentiation to a

specific cell type remains a largely hit-and-miss affair, and ESCs appear

to have a malicious tendency to differentiate into a host of other

cell types in addition to the cell type of interest. So most reviewers

suggest that it is probably necessary to select the cell type of interest

for most applications, either by sorting the cells using fluorescent

antibodies or genetic markers, or by engineering a lethal marker into

the cells that can be switched on should they not turn into the right

kind of cell32-34. Perhaps for this reason, ESC research to date has lain

firmly in the domain of cell biologists, who perform differentiation

assays on cells grown on two-dimensional surfaces in cell culture

dishes and flasks. Cellular differentiation and tissue development is,

however, an inherently three-dimensional process and so to investigate

differentiation and tissue formation in vitro fully, it may be necessary

to turn to the field of tissue engineering, where three-dimensional cell

culture systems have been used for many years.

Tissue engineering Tissue engineers often focus their efforts on providing a three-

dimensional environment, or scaffold, for cell attachment and growth,

and hope that, by mimicking the in vivo environment, cells can be

coaxed into creating a desired tissue type. The ultimate aim of tissue

engineering is to make a three-dimensional cell-containing scaffold

that can be implanted in the body to cure a disease or repair a defect

(Fig. 3). The standard in vivo culture system – where cells are grown in

a monolayer on a charged, flat, plastic surface – cannot replicate

the complexity of the cells’ natural environment and rarely supports

the assembly of cells into a functioning tissue. Providing an

appropriate scaffold that will lead to the development of a functional

tissue is certainly not a simple matter, however, and tissue engineers

have approached the problem in many ways, using a variety of

materials.

Conventional scaffolds Biomedical implants have been used since ancient times – for example,

a Brazilian group recently reported that the ancient Incas successfully

used Au plates to repair cranial defects35. Until the last few decades of

the 20th century, the criteria used in choosing materials for implants

has fundamentally changed very little and usually implant materials

were chosen that were functional because of their inertness. Since the

discovery in the 1960s that some glass ceramics actively bond to living

bone, however, the focus has shifted away from inert materials and

toward materials that are bioactive – those that deliberately elicit a

specified response from the body. Currently, most scaffolds provide a

three-dimensional environment in which tissue can grow and develop,

so that it is able to reproduce the functions of the tissue it is intended

to replace. Some scaffolds may be designed to be implanted without

any cellular component36 – instead they are designed to encourage

tissue ingrowth and de novo tissue synthesis in vivo – while most are

intended to have some kind of cellular component engineered in vitro

before implantation (an example can be found elsewhere37). The latter

strategies require that cells have access to nutrients and space to grow.

Fig. 2 (a) Photomicrograph of undifferentiated human ESCs (H1) grown on a mouse fibroblast feeder layer. (b) Diagram explaining the derivation and

differentiation of ESCs. ESCs can be derived from the inner cell mass of a preimplantation blastocyst and expanded in culture on fibroblast feeder layers. Colonies of

ESCs can be directed to differentiate into cell types from the three germ layers.

(b)(a)

Scaffolds for stem cells REVIEW

DECEMBER 2006 | VOLUME 9 | NUMBER 12 29

Fig. 3 Diagram outlining the potential for tissue engineering. Stem cells can be derived from embryos or adult tissues and expanded in culture. They can then be

either seeded directly on a scaffold or differentiated in culture and sorted to obtain a purified population of a target cell type before seeding on the scaffold. Cell-

seeded scaffolds can then be grown in culture to develop a desired tissue prior to implanting in the body.

For this reason most scaffolds, regardless from which material they are

made, are constructed with some kind of porous network and cultured

with cells in a manner that encourages nutrient transport (Fig. 4). For

instance, inorganic materials such as bioactive glasses and calcium

phosphates have been used extensively for bone tissue engineering

because of the similarities to and their ability to bond with bone’s

natural mineral backbone. Bioactive glasses can be sintered in powder

form to create porous networks38 or, when in solution, can simply

be ‘foamed’ using soap and then gelled to make sol-gels39. Similarly,

porosity can be engineered into polymers, such as polyesters (which

Fig. 4 Schematic illustration demonstrating four methods for achieving nutrient transport in an engineered tissue construct. (a) Construct is placed in static culture

where it is reliant on simple diffusion for delivery of nutrients. (b) Construct is placed in either an environment where the media is mixed, such as a spinner flask,

or in a rotating bioreactor where it achieves a zero-gravity state as fluid is moved around it. (c) Construct is placed in a system that forces fluid and nutrients to

be continuously perfused directly through it. (d) Small channels are engineered into the construct allowing nutrients to be delivered in a similar manner to blood

vessels in the body.

(a) (b) (c) (d)

REVIEW Scaffolds for stem cells

DECEMBER 2006 | VOLUME 9 | NUMBER 1230

have the advantage of being biodegradable), either by foaming the

polymer solution40 or by molding the dissolved polymer around lumps

of another material such as salt, allowing the polymer to harden and

then leaching out the salt with water41. Porous networks can also be

engineered into natural molecules – for example, collagen gels can be

freeze-dried before cell seeding42.

Alternatively, hydrogels can be used as scaffolds for cell growth and

cell delivery. Since the gelling process is often nontoxic, cells can be

introduced into the solution prior to gelation. In the case of alginate,

a natural polymer made up of chains of guluronic and manuronic acid,

calcium is usually added to cell/gel solutions, which crosslinks these

chains and hardens the gel43. Likewise, collagen gels can be hardened

by altering the pH of the solution44,45 (Fig. 5) and poly(ethene) glycol

can be solidified using light46. Hydrogels have different mechanical

properties from other scaffolds, so the material must be selected on

the basis of its properties, keeping in mind the intended application.

But all of these scaffolds have their disadvantages. Inorganic

scaffolds such as ceramics and glasses tend to be too brittle and weak

to be used in load-bearing applications, and even bioactive glasses,

discovered more than 30 years ago, are limited to non-load-bearing

applications such as the replacement of small bones in the middle

ear47. Artificial polymers, on the other hand, may be viewed by the

body as foreign material because they lack sticky surface molecules

for cell adhesion. Their degradation products are, in the case of

polyesters, acidic, and though not directly toxic, may create a possibly

unphysiological acidic microenvironment. This is particularly important

and often overlooked in bone tissue engineering – the natural

mechanism by which bone is degraded in vivo by osteoclasts involves

the formation of an acidic microenvironment! Collagen may be a

better bet in this case, as the natural mechanism of bone formation

involves the mineralization of a collagen scaffold created by osteoblasts

– unsurprisingly collagen scaffolds are readily mineralized in tissue-

engineering experiments48. Another problem with porous scaffolds is

that because cells are seeded onto the internal porous matrix of the

scaffold it becomes arguable whether the cells experience a truly three-

dimensional environment – they merely ‘see’ a slightly curved two-

dimensional surface. This can be solved to some extent by decreasing

the pore size and adding surface texture, or embedding cells in a soft

extracellular matrix, but then problems arise as to how to keep deeply

embedded cells supplied with nutrients. Most importantly in the case of

stem cells, there are challenges involved in providing the correct signals

to encourage differentiation and to pattern cells as they differentiate

into an organized tissue.

Micro- and nanopatterned scaffolds To solve these problems, various groups have begun to investigate

scaffold patterning – both at microscale resolution, possibly including

channels so that cells can be supplied with nutrients and areas where

different cell-types can be deposited, and at nanoscale resolution,

where cells are supplied with the correct ligands to enhance and direct

their function and to induce differentiation.

To address the first challenge, a number of new technologies

are being investigated. Photopolymerizable poly(ethene) glycol

hydrogels have been used as a material for encapsulating cells in

three dimensions – these gels can be mixed with cells as a liquid and

light can be used to solidify them46. Recently, several groups have

used this property to add an element of patterning into hydrogels49-51.

In these cases, a layer of cell-containing gel is selectively exposed

to light in a pattern dictated by a light-blocking mask. Following

gelation of this layer, another layer of cells, possibly containing

another cell type, can be poured over this solid mold and gelled

using another pattern. To illustrate this technique, Liu and Bhatia50

encapsulated cells in recognizable shapes and layered different cell

types in geometric patterns (Fig. 6). Another interesting technique

involves the use of photolithography, a technique used extensively

to make electronic components. This simply involves exposing the

(b)(a)

Fig. 6 Illustration of patterned cell-containing hydrogels. (a) HepG2 cells,

a liver cancer cell line, encapsulated in various three-dimensional shapes

in photopolymer ized PEG hydrogels. (b) Cells labeled with red or green

dyes, partitioned in distinct areas within a three-dimensional hydrogel layer.

(Reprinted with permission from50. © 2002 Springer.)

Fig. 5 Epifluorescence micrograph of a short collagen fiber/collagen hydrogel

composite seeded with rat skin fibroblasts prior to hardening of the gel. Live

cells appear green and have long projections into the gel matrix, while dead

cells are small and round and stain red. The collagen fibers autofluoresce

orange.

Scaffolds for stem cells REVIEW

DECEMBER 2006 | VOLUME 9 | NUMBER 12 31

surface of a ‘photoresist’ mounted on a Si wafer to ultraviolet light.

The exposed areas can be hardened – again in a pattern determined

by a mask – and the patterned surface can be used to mold a soft

substrate. This technique has been used extensively for microfluidic

applications, where poly(dimethyl siloxane) (PDMS) is indented and

adhered to a glass slide to make a series of channels (the technique has

been reviewed elsewhere52). But more recently, tissue engineers have

started to realize that this could be an excellent way of introducing

something akin to a vascular system in a biomaterial. For example,

Shin et al.53 simulated a vascular network in a PDMS microfluidic

device and showed the growth of endothelial cells – blood vessel cells

– on the walls of the ‘vascular tree’ (Fig. 7), while Tan and Desai54

made a fibroblast-containing scaffold based on collagen. The ability to

develop three-dimensional, perfusable scaffolds has great potential in

both tissue engineering and bioreactor technology as it could provide

a way to keep deeply-embedded cells supplied with nutrients within a

scaffold.

Another possible way of patterning scaffolds more intricately might

involve rapid prototyping (RP). Put simply, RP encompasses a range

of different techniques, all of which have the property of producing a

physical object based on a computer design55. Most of these devices

are analogous to printers and can print scaffolds using a variety of

materials. Arguably, the most interesting, however, are devices that

appear to be able to print cells and matrices in combination56,57, which

could allow precise control over tissue microstructure in the future.

Developing technologies such as these allow us to control the

immediate surroundings of cells in a three-dimensional environment

more precisely, and may provide a more authentic environment in

which to direct the differentiation of cells to form a coherent tissue.

But these approaches do not directly consider the chemical interactions

that go on between a cell and its substrate (Fig. 8). To address this

‘nanoscale’ problem, tissue engineers are beginning to fabricate

bioactive scaffolds, where the surface of the scaffold is engineered to

stimulate cell function.

One such method involves using self-assembling amphiphilic

peptides. These molecules are engineered with hydrophilic heads and

hydrophobic tails that, under the correct conditions, can self-assemble

into a network of ‘nanofibers’ with the heads sticking out into solution

and the tails hidden in the core of the thread (Fig. 9)58. Networks

F ig. 7 Cells growing in microfabricated networks in PDMS. Channels are

constructed by selectively exposing a photoresist to light to create an intricate,

raised relief pattern. This relief pattern is then used to indent PDMS, which

is then adhered to a glass slide. Cells and medium can be perfused through

the channels and cells adhere to the walls. The cells here are HMEC-1, an

immortalized endothelial cell line. (Reprinted with permission from53. © 2004

Springer.)

Fig. 8 Diagram demonstrating the relationship between cells and scaffolds. The microenvironment created around a cell adherent to a tissue-engineering scaffold

is complicated. Nutrient transport brings growth factors, ligands, and other signals that can bind to cell receptors. The degrading scaffold can likewise release

chemical messengers that bind to membrane receptors and influence intracellular communication and cellular processes such as gene transcription. Cells also

attach to the scaffold via integrin receptors. Integrin receptors are closely connected to the cell’s cytoskeleton and relay further information to the cell thereby

affecting cell function.

REVIEW Scaffolds for stem cells

DECEMBER 2006 | VOLUME 9 | NUMBER 1232

such as these have the advantage of porosity and living cells can be

combined with such materials before the scaffold is fabricated. More

importantly, peptides common to the extracellular matrix of the tissue

under investigation can be engineered into the hydrophilic heads of

such molecules, which can improve cell attachment and promote

differentiation and function. Since the meshwork is so fine, the internal

surface area of the scaffold is very large and cells are exposed to a

high density of these ligands from all directions, which may encourage

three-dimensional growth. Another fabrication method currently under

investigation for making very fine scaffold networks is electrospinning,

in which tiny threads of material can be fabricated to provide a porous

mesh. This is accomplished by extruding a material from a fine nozzle

using electrostatic force to form fibers between 3 nm and 5 µm in

diameter. Again, extracellular matrix ligands can be attached to such

fibers, or the fibers can be made out of biologically derived materials

such as collagen59.

Such micro- and nanopatterned scaffolds, therefore, introduce

a greater level of control over the fine structure of a scaffold than

conventional scaffolds and may allow control over cell patterning and

cell differentiation. This level of control may be invaluable if a scaffold

is to be used in combination with stem cells.

Scaffolds and stem cells Stem cells, of course, are reliant on the extracellular environment

not only to survive but also to develop into a functional tissue. So

increasingly, tissue engineers are beginning to use the composition of

scaffolds to persuade stem cells to differentiate. Arinzeh et al.60 have

shown that adjusting the ratio of hydroxyapatite to tri-calcium

phosphate could influence the degree to which osteogenic

differentiation of MSCs occurs, while others have begun to engineer

bioactive factors into porous scaffolds. For instance, Kim et al.61 have

created a polyester scaffold that slowly exudes vitamin C and

β-glycerophosphate and demonstrated an increase in osteogenesis from MSCs, while Yang et al.62 have demonstrated increased

osteogenic differentiation in a polylactic acid (PLA) scaffold spiked

with a bone-specific growth factor. Alternatively, mechanical force can

be used to stimulate differentiation – Altman et al.63 have recently

shown that applying a mechanical force to a collagen-gel scaffold can

encourage MSCs to differentiate into ligament tissue.

Micro- and nanopatterned scaffolds have been investigated less

well in regard to stem cells, although two recent studies highlight

their attractiveness. Silva and colleagues64 included a five amino

acid, laminin-specific cell-binding domain (which binds to specific

integrins on cell surfaces) at the hydrophilic head of their amphiphiles,

and showed that neural stem cells could be induced to differentiate

into neurons when cultured within the network. In contrast, cells

grown in control scaffolds without the laminin-specific domain or on

two-dimensional tissue culture plastic coated with laminin solution

differentiated much less. This was hypothesized to be largely as a

result of the density of the cell-binding ligands to which the cells

were exposed, indicating clearly the importance of extracellular matrix

in influencing cell function. In a similar study, Hosseinkhani et al.65

replaced the laminin-specific domain in the amphiphilic molecule

with the amino acid sequence, arginine-glycine-aspartate (RGD), a

common cell-binding domain in many extracellular matrix proteins,

especially collagen. They then showed that the differentiation of MSCs

to osteoblasts is significantly enhanced compared with amphililic

nanofibers without this sequence on to two-dimensional controls.

So far, remarkably few studies have been published on the effect of

three-dimensional environments and scaffolds on ESC differentiation.

In two rare examples, Levenberg and colleagues66,67 have shown that

human ESCs embedded in an extracellular matrix gel called Matrigel

can be differentiated in three dimensions on conventional polyester

scaffolds. In these cases, several structures that resemble primitive

tissues were generated, depending on the content of the growth

medium. The authors also show that tissues grown in three dimensions

express higher levels of differentiation-associated proteins than those

on coated two-dimensional surfaces. Interestingly, another group has

Fig. 9 Illustration of how amphiphilic peptides self-assemble to produce

nanofibers. (a) The chemical structure of each peptide, including a long

hydrophobic tail (1) and the three amino acid RGD motif at the head (5). (b)

and (c) show how these peptides assemble to form a fiber. (Reprinted with

permission from58. © 2001 AAAS.)

(b)

(a)

(c)

Scaffolds for stem cells REVIEW

DECEMBER 2006 | VOLUME 9 | NUMBER 12 33

recently reported that the chondrogenic differentiation of human ESCs

in a PEG hydrogel is dependent upon whether or not the hydrogel

contains adhesive RGD sites, illustrating the importance of the cell

matrix and microenvironment in ESC differentiation68.

The use of more novel, patterned scaffolds should provide ESC

biologists with an important new tool to stimulate and model

differentiation in vitro. For instance, nanopatterned scaffolds, such as

those using amphiphilic peptides, could be used to partition cells within

a mixed population of ESC-derived cells, based on the specificity of the

ligands to which different cells bind. In this way, scaffolds could play a

role in directing tissue organization, not only with the aim of producing

tissue for transplant but also for studying differentiation in vitro.

Similarly, ESCs could be compartmentalized within scaffolds to study

cell-cell interactions and their effect on cell differentiation and tissue

formation. Such scaffolds will undoubtedly find exciting applications in

the study of ESC differentiation in the future.

Conclusions With stem cells, we are presented with a versatile material with which

we may be able to rebuild many structures found in the body. But the

challenge of how to construct three-dimensional tissues from them

still remains. Recently, biologists and materials scientists have realized

that scaffolds can and should be designed that direct and enhance cell

function and differentiation. We have only scratched the surface of

how these scaffolds can be used in concert with stem cells, however,

and this provides us with great encouragement that they may provide

us with the power to construct three-dimensional functional and

intricate replicas of human tissue in the near future.

Acknowledgments NDE acknowledges the support of Medical Research Council, UK for funding.

REFERENCES

1. Hussain, M. A., and Theise, N. D., Lancet (2004) 364, 203

2. Heng, B. C., et al., J. Bone Miner. Res. (2004) 19, 1379

3. Jansen, J., et al., J. Cell Mol. Med. (2005) 9, 37

4. Ford, C. E., et al., Nature (1956) 177, 452

5. Mathé, G., et al., Br. Med. J. (1963) 5373, 1633

6. Friedenstein, A. J., et al., Cell Tissue Kinet. (1970) 3, 393

7. Haynesworth, S. E., et al., Bone (1992) 13, 81

8. Pittenger, M. F., et al., Science (1999) 284, 143

9. Liechty, K. W., et al., Nat. Med. (2000) 6, 1282

10. Hofstetter, C. P., et al., Proc. Natl. Acad. Sci. USA (2002) 99, 2199

11. Baxter, M. A., et al., Stem Cells (2004) 22, 675

12. Tolar, J., et al., Stem Cells (2006), in press

13. Reynolds, B. A., et al., Science (1992) 255, 1707

14. Levison, S. W., et al., Neuron (1993) 10, 201

15. Clarke, D. L., et al., Science (2000) 288, 1660

16. Toma, J. G., et al., Nat. Cell Biol. (2001) 3, 778

17. Thomson, J. A., et al., Science (1998) 282, 1145

18. Bielby, R. C., et al., Tissue Eng. (2004), 10, 1518

19. Lumelsky, N., et al., Science (2001) 292, 1389

20. Zhang, S. C., et al., Nat. Biotechnol. (2001) 19, 1129

21. Yuasa, S., et al., Nat. Biotechnol. (2005) 23, 607

22. Rippon, H. J., et al., Stem Cells (2006) 24, 1389

23. Samadikuchaksaraei, A., et al., Tissue Eng. (2006) 12, 867

24. Hubner, K., et al., Science (2003) 300, 1251

25. Geijsen, N., et al., Nature (2004) 427, 148

26. Vats, A., et al., Tissue Eng. (2006) 12, 1687

27. Mummery, C., et al., Circulation (2003) 107, 2733

28. Van Vranken, B. E., et al., Tissue Eng. (2005) 11, 1177

29. Kim, J. H., et al., Nature (2002) 418, 50

30. Tai, G., et al., Tissue Eng. (2004) 10, 1456

31. Coraux, C., et al., Curr. Biol. (2003) 13, 849

32. Keller, G., Genes Dev. (2005) 19, 1129

33. Passier, R., et al., Cardiovasc. Res. (2003) 58, 324

34. Pera, M. F., et al., Development (2004) 131, 5515

35. Marino, R., Jr., et al., Neurosurgery (2000) 47, 940

36. Stevens, M. M., et al., Proc. Natl. Acad. Sci. USA (2005) 102, 11450

37. Weinand, C., et al., Bone (2006) 38, 555

38. Jones, J. R., et al., Biomaterials (2006) 27, 964

39. Sepulveda, P., et al., J. Biomed. Mater. Res. (2002) 59, 340

40. Ginty, P. J., et al., Proc. Natl. Acad. Sci. USA (2006) 103, 7426

41. Murphy, W. L., et al., Tissue Eng. (2002) 8, 43

42. O’Brien, F. J., et al., Biomaterials (2004) 25, 1077

43. Simpson, N. E., et al., Biomaterials (2004) 25, 2603

44. Gentleman, E., et al., Tissue Eng. (2006) 12, 1639

45. Gentleman, E., et al., Tissue Eng. (2004) 10, 421

46. Mann, B. K., et al., Biomaterials (2001) 22, 3045

47. Rust, K. R., et al., Am. J. Otol. (1996) 17, 371

48. Phillips, J. E., et al., Biomaterials (2006) 27, 5535

49. Albrecht, D. R., et al., Lab Chip. (2005) 5, 111

50. Liu, V. A., and Bhatia, S. N., Biomed. Microdevices (2002) 4, 257

51. Luo, Y., and Shoichet, M. S., Nat. Mater. (2004) 3, 249

52. Falconnet, D., et al., Biomaterials (2006) 27, 3044

53. Shin, M., et al., Biomed. Microdevices (2004) 6, 269

54. Tan, W., and Desai, T. A., Tissue Eng. (2003) 9, 255

55. Mironov, V., et al., Trends Biotechnol. (2003) 21, 157

56. Xu, T., et al., Biomaterials (2005) 26, 93

57. Boland, T., et al., Anat. Rec. A Discov. Mol. Cell Evol. Biol. (2003) 272, 497

58. Hartgerink, J. D., et al., Science (2001) 294, 1684

59. Pham, Q. P, et al., Tissue Eng. (2006) 12, 1197

60. Arinzeh, T. L., et al., Biomaterials (2005) 26, 3631

61. Kim, H., et al., Biochem. Biophys. Res. Commun. (2005) 332, 1053

62. Yang, X. B., et al., Tissue Eng. (2004) 10, 1037

63. Altman, G. H., et al., FASEB J. (2002) 16, 270

64. Silva, G. A., et al., Science (2004) 303, 1352

65. Hosseinkhani, H., et al., Biomaterials (2006) 27, 4079

66. Levenberg, S., et al., Proc. Natl. Acad. Sci. USA (2003) 100, 12741

67. Levenberg, S., et al., Tissue Eng. (2005) 11, 506

68. Elisseeff, J., et al., Stem Cells Dev. (2006) 15, 295