EMBRYONIC STEM CELLS IN DRUG
The completed sequencing of the human genome has identified numerous potential drug
targets, which are expected to deliver the next generation of new medicines. However, for drug
companies to realize this opportunity, they must rely on improved prognostic applications of high-
throughput technologies, from target identification to preclinical compound evaluation. Reducing
the timelines and attrition rate of new therapeutics for clinical evaluation requires cell-based
methods for testing the efficacy and safety of new compounds. Drug discoverers are beginning
to use stem cells as a new resource for increasing confidence in the mechanism of action of
new targets and the safety of modulating their activity.
TARGET The application of mammalian stem-cell technology to Drug discovery investigators worldwide have used
Gene product that therapeutic drug discovery has afforded new opportunities for genetic engineering in mES cells to derive highly sophis-
approaches are directed against evaluating TARGETS and novel therapeutics in relevant ticated GeMM to evaluate potential targets. The ability
in the search for new medicines.
cell-based assays. The recent identification and isolation to inactivate genes in vivo has become a central tool in
EMBRYONIC STEM CELLS
of human stem cells might offer superior methods to determining target function, selectivity and toxicity1,2;
(ES). Undifferentiated cells format predictive, high-throughput assays that shorten as a result, researchers in drug discovery have gained
typically derived from the inner the timelines for the identification of new therapeutics access to, and characterized, GeMM through both
cell mass of a blastocyst-stage and reduce the amount of in vivo testing. The potential internal and external resources.
embryo. In culture, these cells can
self-renew in the undifferentiated
impact of these cellular systems is broad, and processes Before the demonstration of the ability to engineer
state or reveal their pluripotentcy that could be affected include target identification and genetic modifications in mice, the species was rarely
on differentiation into cell types validation, chemical screening, secondary assays for drug used as either an in vivo clinical or physiological
of the three embryonic germ efficacy, metabolism and safety studies, compound eval- model. Therefore the challenge of applying technol-
lineages. Mouse ES cells readily
uation for human genetic variants and the identification ogies in GeMM required the miniaturization of estab-
differentiate into all somatic and
germ lineages when incorporated of clinically relevant biomarkers (FIG. 1). This review will lished in vivo assays, including behavioural tests for
into chimeric offspring through describe the current application of stem cells in drug clinical conditions of the central nervous system3,4,
blastocyst microinjection or discovery and the advantages that these technologies blood pressure monitors for cardiovascular diseases5,
morula aggregation. have compared with other cell systems. The focus will and sophisticated micro computed tomography imaging
be the use of murine EMBRYONIC STEM (ES) CELLS, including equipment for the in vivo evaluation of tumours or
the development of genetically modified mice, but the metabolic changes in muscle, fat and bone6,7. These
emphasis will be on the application of ES cells for in vitro shifts in technology allowed GeMM to become a crucial
methodologies. This review will also provide a forward- factor in making decisions on the utility of established
looking description of the potential applications and and novel targets. The strength of this approach was
Genetic Technologies, contributions that advancements with human ES (hES) demonstrated in an excellent review of KNOCKOUT (KO)
Pfizer Global Research and cells could deliver to drug discovery. MOUSE phenotypes for targets for the top 100 best-sell-
Development, Groton ing drugs, which reported that the phenotypes of
06340, Connecticut, USA.
Murine ES cells and genetically modified mice mouse KOs correlate well with the therapeutic effects in
pfizer.com The most expansive use for murine ES (mES) cells is the humans of antagonists of the pathways modulated
doi:10.1038/nrd1281 development of genetically modified mice (GeMM). in the KO mice8.
70 | JANUARY 2004 | VOLUME 3 www.nature.com/reviews/drugdisc
a specific catalytic domains or the alteration of drug-
binding sites in a target protein. In addition, single-base
changes allow the expression of conserved human muta-
tions, notably the presenilin 1 allelic variants associated
with early onset Alzheimer’s disease12,13. In 1995, Jaenisch
and colleagues14 demonstrated that replacement of the
entire murine Myf5 gene into the murine myostatin
locus resulted in functional redundancy in the develop-
ing mouse. Drug companies have used this approach to
replace entire murine genes with the orthologous human
target gene. Specific examples of human target KIs have
been reported for chemokine receptor-2 (CCR2) (REF. 15)
and glucagon receptor16 by GlaxoSmithKline and Merck,
respectively. These humanized mice could serve as
important in vivo models for screening lead compounds
Drug efficacy and are especially valuable when the lead chemical
material is highly species selective.
Random-mutagenesis strategies are also readily
applicable to ES cells using gene trapping (reviewed in
REF. 2) and chemical mutagens (for example, N-ethyl-N-
nitrosourea (ENU)). These strategies have generated
b Idea Lead Candidate Clinic important observations in GeMM. For example, gene
trapping has recently corroborated human genetic
studies on WNK1 kinase in the regulation of blood
pressure17. The ENU-induced chemical mutagenesis
Target Secondary Human
HTS ADME Toxicology approach has resulted in the rapid identification of a
validation assays variant
specific allelic series of genes in the transforming
Figure 1 | Applications of stem-cell technology in drug discovery. a | Numerous growth factor-β (TGF-β) signalling pathway in ES cells,
genomics-based technologies are now routinely applied to drug discovery. A central role for which allows functional evaluation in GeMM18.
these technologies is validating the next generation of therapeutics from novel targets identified
through genomics. The aim of these technologies is to accurately identify the next generation
of targets that demonstrate therapeutic efficacy and safety. b | Embryonic stem cells have
In vitro differentiation: induction
unique attributes that can be applied in drug discovery, from initial target ideas to clinical trials. Since the isolation of mES cells in the early 1980s, inves-
The specific applications listed here are not necessarily listed in order of application. ADME, tigators have come to recognize the intrinsic tendency of
absorption, distribution, metabolism and excretion; HTS, high-throughput screening. these cells to spontaneously DIFFERENTIATE into numerous
cell types in a pattern reminiscent of the developing
mouse embryo. Doetschmann and colleagues19 reported
that altering the culture conditions resulted in develop-
The pharmaceutical industry has wide reliance on ment of EMBRYOID BODIES, structures similar to the embry-
KO mice in applications associated with drug discov- onic yolk sac, with differentiated cells for the three
ery, varying from target validation to toxicity deter- principal germ layers of the developing embryo: ecto-
mination1. KO mice are valuable for understanding the derm, endoderm and mesoderm (FIG. 2). Early studies of
validity and safety of novel targets and are crucial in the the in vitro differentiation (IVD) of ES cells lacked the
KNOCKOUT MICE advancement or withdrawal of new initiatives. The KO ability to consistently derive purified populations of
Mice derived from gene- mouse results in 100% antagonism in vivo and can dis- specific cells types, and research on in vitro differentia-
targeting experiments in ES cells
tinguish differences in selectivity between genetic and tion technology was limited. In fact, with the rapid
with mutations in selected genes
that result in the complete pharmacological inactivation or the presence of sec- interest in gene targeting for the development of
ablation of gene expression. ondary pharmacology. Important models of clinical GeMM, most of the effort directed towards ES cells was
These induced mutations are conditions have been developed using KO mice, in the consistent maintenance of ES cells in an undiffer-
passed through the germline, including the robust atherosclerosis model generated by entiated state. This includes the identification of
allowing the derivation of novel
lines of gene KO mice for
the inactivation of apolipoprotein E9,10. In addition, KO leukaemia inhibitory factor (LIF) for the sustained
functional evaluation in vivo. mice illuminate the mechanisms of drug metabolism culture of ES cells in an undifferentiated state without
and toxicity. The use of 5-lipoxygenase KO mice feeder fibroblast co-cultures20. Although research on
HOMOLOGOUS demonstrated that the induction of hepatic enzymes in vitro differentiation of ES cells was limited, investiga-
was related to a specific chemical series of leukotriene tors demonstrated a greater understanding of the differ-
Genetic recombination between
identical (or nearly identical) B4 antagonists, and was not due to the pharmacology entiated cells and the methods that are required to
double-stranded DNA associated with leukotriene depletion11. derive specific cell populations. It is noteworthy that
sequences. This process occurs The precision of genetic modifications that can Wobus et al.21 demonstrated that introducing nerve
at a relatively high frequency in achieved by HOMOLOGOUS RECOMBINATION in ES cells allows growth factor to embryoid bodies accelerated the devel-
ES cells, allowing the
engineering of gene-targeting
the pre-planned replacement of mutations in endo- opment of neuronal cell populations. Further studies
vectors to direct pre-planned genous murine alleles, which have been coined knock-ins demonstrated that the spontaneously derived contrac-
mutagenesis in ES cells. (KIs). Single-base changes allow the inactivation of tile myogenic cells from embryoid bodies express genes
NATURE REVIEWS | DRUG DISCOVERY VOLUME 3 | JANUARY 2004 | 7 1
Zygote Cell type References: Murine Human
Neurons 28,38,48 104, 105
Glia 29,35,36 105
Mouse 3.5 days Ectoderm Keratinocytes 51
Human 7 days Dendritic cells 52
Cardiomyocytes 19,22,23,34,43 103
Chondrocytes 32, 33
ES cells Mesoderm Osteocytes 39,56
Vasculature 54 107
Skeletal muscle 55
Pancreatic cells 37,50
GeMM Endoderm Hepatocytes 57 106
Figure 2 | ES cell differentiation potential. This figure demonstrates the pluripotential capacity of ES cells to deliver differentiated
cells types from the ectodermal, mesodermal and endodermal lineages. The figure is presented with an axis to demonstrate the
advantage in time to delivery of in vitro differentiated ES cells have compared with genetically modified mice (GeMM). Selected cell
types of therapeutic relevance and the alignment with the germ lineage derivation for both murine and human ES cells are listed
with the corresponding reference. ES cell, embryonic stem cell.
of the α- and β-cardiac myosin isoforms characteristic of cells in vitamin D and dexamethasone produces
cardiomyocyte development in vivo22,23. The two-step osteoclasts cells39; similarly, co-culture of murine stromal
method of generating cells committed to haematopoi- cells with haematopoietic growth factors produces
etic lineages was developed using embryoid bodies that megakaryocytes40.
were disaggregated and plated in methylcellulose24,25, A well-established inductive method to evaluate
which resulted in haematopoietic progenitor cells of the PLURIPOTENCY, which dates back to 1970 (REF. 41), is ectopic
various lineages. These haematopoietic cells26,27 tempo- implantation into syngenic or immune-compromised
rally expressed cell-specific transcripts that mimicked mice. This method has been applied to test the pluri-
the embryonic expression patterns and functional char- potency of ES cells42. Although a powerful approach,
acteristics associated with terminally differentiated cell this is not a practical method for deriving cells to be
types. Of particular excitement was the ability to repro- evaluated in drug discovery, because this method
ducibly develop populations of neurogenic lineages requires manipulations of the cells in animals, results in
and glial cells from embryoid bodies in response to the multiple cell types and the populations of cells generated
morphogen retinoic acid28,29. cannot be developed at scale.
Process in which a cell The differentiation of ES cells in suspension culture
progresses in a linear manner to results in the emergence of various cell types and gene- In vitro differentiation: selection
a specialized state. Stem cells can expression patterns resembling the normal developing Although inductive procedures have resulted in many cell
develop into any specialized cell mammalian embryo25,30. An understanding of cellular types, to further enrich in vitro differentiation popula-
type – for example, neuronal,
muscle, hepatic and
differentiation during embryogenesis has led to methods tions, investigators have developed selective methods that
haematopoietic cells. for enriching populations of specific cell types using are based on the expression of marker proteins. These
culture conditions supplemented with established selective markers can include the expression of an anti-
EMBRYOID BODIES growth factors. Several examples of pharmacologically biotic-resistance gene, colorimetric markers or cell-
Spherical cell clusters observed
relevant, specialized cell types derived by inductive surface markers. For example, highly purified cell
after spontaneous or induced
differentiation of ES cells in differentiation protocols include adipocytes31, chondro- populations (>99%) have been generated through the
culture: Embryoid bodies show cytes32,33, atrial and ventricular cardiomyocytes34, micro use of antibiotic-resistance genes, transgenes and gene
differentiation that recapitulates and macroglial cells35,36, pancreatic islet cells37 and targeting to insert a selectable marker directly into an
the early stages of mammalian motor neurons38. Co-culture conditions have been endogenous cell-specific promoter. A transgenic
including cell types derived from
described that improve inductive differentiation methods approach resulted in the selective differentiation and
endoderm, mesoderm and for specific cell populations. For example, the co-culture purification of cardiomyocytes using a neomycin-
ectodermal lineages. of undifferentiated mouse ES cells with ST2 stromal resistance marker under the control of the α-cardiac
72 | JANUARY 2004 | VOLUME 3 www.nature.com/reviews/drugdisc
myosin heavy-chain promoter43. In addition , a gene- OCT4) and NANOG61,62,63, will most likely aid the identi-
targeting approach in ES cells — which directs cell- fication of PLURIPOTENT cell lines. In addition, determining
type-specific expression of the selectable marker either the biochemical differences between pluripotent embry-
through introduction into the coding sequence or onic and precursor stem cells — or ‘stemness’64,65,66 —
without disrupting the coding sequence using a will allow clonal isolates of precise populations of clini-
dicistronic vector — has been reported44. This cally relevant cellular reagents to be derived. Molecular
approach led to the purification of neuronal cells profiling using array-based genome-expression profiling
through the precise introduction of a neomycin- with reproducible methods of ES cell in vitro differentia-
resistance marker into the murine Sox1 locus45. The tion will result in an improved understanding of the
incorporation of colorimetric markers, such as β-galac- genetic hierarchy operating in the development of
tosidase and enhanced green fluorescent protein normal mammalian cell types67. Molecular profiling, in
(eGFP), into gene-targeting vectors has resulted in selec- concert with genetic and pharmacological manipula-
tive protocols that have been successfully applied to the tion, will also prove enlightening in the functional
isolation of purified cardiac precursors46. Fluorescence- evaluation of therapeutically relevant pathways in vitro.
activated cell sorting (FACS) has been applied to cell-
specific surface markers of ES cell derivatives, such as the Stem cell advantages
isolation and purification of macrophage by sorting for Because all therapies are directed against targets of
the interleukin-1 (IL-1) receptor47. normal or pathological cells, the availability of reliable
Investigators have also expressed specific genes to cell types is essential for their successful application in
influence the in vitro differentiation of ES cells. For the drug discovery process. Cells for drug discovery
example, dopaminergic neurons are a target cell for research have typically been obtained from primary
psychiatric diseases; however, in vitro methods for this tissue, immortalized tumour cells or genetically trans-
specialized cell type are limited. The overexpression of formed cells. Primary mammalian cells are genetically
rat Nurr1 complementary DNA in ES cells, combined normal (that is, diploid), yet have very limited survival
with a specific inductive protocol for mid- and hindbrain times in culture, which affects the applicability of primary
neurons, resulted in a greater than tenfold enrichment of explants in screening technology. Also, the inconsistent
dopaminergic neurons48,49. These cells demonstrated availability, and the inherent donor variation, of human
molecular and functional properties associated with primary cells types restricts opportunities for the use
dopaminergic neurons, including selective secretion of of primary cells in drug discovery. Immortalized cells
dopamine in vitro and electrophysiological characteris- derived from tumours or oncogenic transformation
tics and behavioural improvements when engrafted into offer more consistent sources of cellular reagents, which
a rat model of Parkinson’s disease. Constitutive overex- makes them suitable for use in high-throughput screen-
pression of murine Pax4 in ES cells, combined with an ing (HTS) and secondary assays. Immortalized cells can
inductive protocol for pancreatic cells, resulted in an be maintained indefinitely and transfected with DNA
enrichment of nestin-positive progenitor and insulin- constructs that express target proteins or reporters.
producing β-cells versus other cells producing hormones However, these cells are typically genetically abnormal
(for example, glucagon) found in pancreatic islets50. The (ANEUPLOID), and conclusions based on gene function
Pax4 protocol resulted in functional cells that secrete could be limiting. Stem cells offer considerable advan-
insulin in response to glucose or sulphonylurea and tages, compared with primary or immortalized cells, in
which have improved glucose homeostasis after trans- that these cells are genetically normal, demonstrate uni-
plantation into streptozotocin-induced diabetic rats. form physiological responses, are maintained in culture
for long periods of time and are grown at scale, all of
In vitro differentiation: summary which enhances their usefulness in screening processes.
Various methods for directing the in vitro differentiation Furthermore, the unique strength of ES cells is their
of mES cells can efficiently deliver cells of all three germ ability to undergo homologous recombination at a rela-
lineages (that is, endoderm, mesoderm and ectoderm). tively high frequency, which enables the selection of
In fact, the differentiation of mES cells has recently been reproducible and precise genetic modifications of the
shown to form germ cells, including mature oocytes59 endogenous genome. However, it should be noted that
and spermatocytes60. A list of various murine cell types homologous recombination has been demonstrated at
reported by inductive and selective methods from mES very low frequency in other cell types, including trans-
cells is provided in FIG. 2. For completeness, in vitro differ- formed cells68 and haematopoietic stem cells69.
entiated cell types from hES cells are included. In the In summary, mES cells offer several important
near future, it is likely that consistent and reproducible advantages over primary or immortalized cells in drug
in vitro differentiation protocols for purified cells of all discovery applications (TABLE 1). This results in more
The ability to differentiate into
all cell types derived from the germ lineages will be available to drug discoverers. reproducible and accurate evaluations of targets in func-
three embryonic germ lineages: FIGURE 3 summarizes the inductive and selective differ- tional and screening paradigms. It is likely that as tech-
ectoderm, mesoderm and entiation methods investigators have used to derive nology advances with regard to stem cells derived from
endoderm. populations of pharmacologically relevant cell types. other sources and species, the same level of precise
Furthermore, the recently improved understanding of genetic modifications could be available to investigators
Cells with abnormal number of the transcriptional machinery associated with stem-cell and result in improved applications in basic biology and
chromosome sets. potency, including the roles of POUSF1 (also known as drug discovery.
NATURE REVIEWS | DRUG DISCOVERY VOLUME 3 | JANUARY 2004 | 7 3
Embryonic stem cells Target evaluation: undifferentiated ES cells
Spontaneous: Growth factors The understanding of gene function in drug discovery is
LIF negative Morphogens
fundamental to the successful management of target
Serum-free media portfolios. The use of undifferentiated ES cells to gain
functional information rapidly and without the costs
required to generate and evaluate KO mice has proven
to be effective. The first reported example of this
Embryoid bodies, stem, precursor and approach by a pharmaceutical company was the devel-
Differentiation of ES cells
differentiated cells opment and evaluation of nuclear factor-κB (NF-κB)
Genetic modification: Growth factors precursor p105 in ES cells at Bristol-Myers Squibb73.
Antibiotic resistance Cell-surface proteins
Gene overexpression The genetically modified ES cells had specific deletions
Fluorescence marker in the C-terminal region of the p105 gene, which
encodes the ankyrin-containing domains, but yet main-
tained the N-terminal coding region for a functional
Selection Cell sorting
p50 protein that associates with RelA and complexes
Fully differentiated cells — enriched populations
with inhibitor of NF-κB-α (IκB-α). These cells
demonstrated that p105 was important in the control
of p50 dimerization and the regulation of NF-κB-
binding activity of p50-RelA complex, detailing the role
of NF-κB in inflammatory responses.
An important application of mES cells has been to
the evaluation of target genes that result in embryonic
lethality in KO mice74, because if gene-deficient
Figure 3 | Strategies to derive purified populations of in vitro differentiated ES cells. embryos die in early development then evaluation is
This figure summarizes the methods presently used to first differentiate ES cell populations, limited. Investigators have used ES cells deficient in
and second to purify these populations for use in applied or basic research programs. The
target genes in the undifferentiated state to evaluate the
approaches are considered general methods, and specific examples of these methods are
included in the text and references. The specific differentiated cell populations shown here were
gene function in vitro. ES cells deficient for the gene
selected based on clear differences in cellular morphology. The examples are hepatocytes, encoding the tumour suppressor BRCA1 have increased
neurons and glial cells from left to right. ES cell, embryonic stem cell. sensitivity to DNA-damaging agents with DNA-repair
insufficiency75 and reduced rates of homologous recom-
bination compared with wild-type ES cells76. These
observations, and the association with RAD51, indicate
Derivation of gene null ES cells a key role for BRCA1 in double-strand DNA repair.
To facilitate the functional evaluation by mutagenesis
in ES cells, methods have been described for the inacti- Target evaluation: differentiated ES cells
vation of both target alleles in undifferentiated ES cells. Investigators have demonstrated that cells derived from
This requires developing two unique targeting vectors ES cells differentiate in vitro reflecting the cellular physiol-
that take advantage of selection protocols using antibi- ogy of clinically relevant primary cells (BOX 1). Gene
otic-resistance genes (for example, neomycin, hygro- inactivation can be applied to IVD systems directly to
mycin and puromycin) or purine metabolism in evaluate complex molecular events in signalling path-
hypoxanthine phosphoribosyl transferase-negative ES ways that could be relevant to decisions on target
cell lines (HAT selection). Alternatively, after initial advancement. For example, Minamino et al.77 evaluated
gene targeting in ES cells, site-specific recombination the role of mitogen-activated protein kinase kinase
using the Cre/lox or Flp/frt system can be used to excise kinase-1 (MEKK1) in a cellular model of post-
the selectable marker, followed by repeated targeting ischaemic reperfusion injury in ES cell-derived car-
using the original vector to inactivate the second diomyocytes. The MEKK1-deficient cardiomyocytes
allele70. These methods although effective have limita- were significantly more sensitive to apoptosis induced
tions, including the need to generate multiple targeting by reactive oxygen species, which resulted from
vectors and reduced efficiency in the second targeting increased expression of tumour-necrosis factor-α
reaction. An alternative approach demonstrated that (TNF-α) by the cardiomyocytes. This was the result of
increased selection pressure with G418, an aminglyco- the specific inhibition of stress-induced signalling via the
side antibiotic that is structurally related to gentamycin, c-Jun kinase (JNK) pathway. This direct approach
in +/– ES cells targeted with a neomycin-resistance clearly allows time-saving evaluation of targets relative
gene could gene convert the + allele resulting in null to KO mice and can guide decisions on resources allo-
–/– ES cells71. This method does not result in –/– ES cated to the generation of the KO mice.
cells lines for all targeted cells and might be a conse- Alternatively, ES cells deficient in target genes have
quence of gene targeting with a neomycin-resistance been evaluated following in vitro differentiation into
gene that is less resistant to G418 selection. Finally, specific cell types. The mitogen-activated protein kinase
investigators have derived –/– ES cells from blastocysts p38-α is a key component of the novel class of com-
derived from targeted mouse lines when wishing to pounds known as cytokine suppressor anti-inflam-
perform in vitro analysis72. matory drugs (CSAIDS). Allen et al.45 produced p38α
74 | JANUARY 2004 | VOLUME 3 www.nature.com/reviews/drugdisc
Table 1 | Characteristics of commonly used cell types in drug discovery
Cell types Genetic stability Growth Quantity Cellular variability Gene targeting
Tumour or Aneuploid Abnormal Unlimited Change over time Very limited
transformed cell lines in culture
Primary cells Diploid Do not proliferate Restricted Highly donor-dependent No
Embryonic stem cells Diploid Normal Unlimited Uniform Yes
–/– ES cells that were differentiated and FACS sorted to cells deficient in the murine orthologue. An important
isolate IL-1-receptor-positive macrophages. The p38α- feature of cell-based reporter screens is that they require
deficient macrophage had reduced response to the fewer cells than cell-based functional screens, which
cytokine IL-1, when assayed for IL-6 production, estab- allows formatting in ultra-HTS or miniaturized HTS.
lishing a mechanism of action of p38-α in the IL-1
signalling pathway. In addition, results from KO cells Teratology/toxicity screening
derived from ES cells can provide clues as to the devel- A leading cause of drug candidate attrition is reproduc-
opmental cause of embryonic lethality. KO mice for the tive toxicity. As stated previously, the cellular and tissue
ryanodine receptor (RyR2) are embryonic lethal78. differentiation that generates the three germ lineages
Cardiomyocytes derived from RyR2 –/– ES cells demon- (ectoderm, endoderm and mesoderm), and the tempo-
strated that the release of Ca2+ from the sarcomplasmic ral expression of genes, in ES cell IVD parallels the
reticulum of normal cardiomyocytes occurs exclusively pattern seen in the normal developing mammalian
through RyR2 and is crucial for modulation of the beat- embryo. This led to the hypothesis that ES cells could be
ing rate of the myocardium79. These results indicated to applied to in vitro teratology or toxicity assays83. The use
the authors that the RyR2 effect on Ca2+ homeostasis of ES cells to evaluate teratogens has been championed
and beating rate probably accounts for the lethality at the European Center for the Validation of Alternative
observed in RyR2 KO mice. Methods (ECVAM). In 2001, the ECVAM Scientific
The implantation of undifferentiated ES cells into Advisory Board approved an embryonic stem cell test
ectopic sites in mice results in TERATOMAS , which comprise (EST) as a scientifically validated test for embryotoxicity
cells from all three embryonic germ lineages and there- in vitro84,85. This group recommended that although the
fore demonstrate the potential for differentiation of new EST could not replace current evaluation in vivo, this
ES cells. Investigators have applied this technique to eval- method could reduce or refine the use of animal pro-
uate the role of target gene expression on the cell prolifer- cedures. The EST is based on the effects that known
ation and differentiation of the resulting teratoma. This compounds have on mES cell differentiation of car-
type of analysis has demonstrated a reduced proliferative diomyocytes in hanging drop cultures86. The EST was
capacity of ES cells null for cyclooxygenase 2 (COX2) used to test 20 coded compounds87 that were blindly
compared with COX1-deficient or wild-type ES cells80. distributed to four European testing laboratories. The
Altered differentiation potential in ES cell that were participation by Novartis and Schering as test sites
null for N- and P-cadherin has been reported81. This reflects the positive attributes this assay could have on
approach could result in an effective strategy to evaluate preclinical teratology evaluation. These groups’ results
the role of potential cancer targets on cell proliferation. were 100% predictive for strongly embryotoxic com-
pounds, and 70% and 72% predictive for compounds
Screening applications for stem cells of weak embryotoxicity or non-embryotoxic, respec-
Innovative approaches in HTS technology have increased tively88; furthermore, there was a 78% correlation
the scope of applications for cell-based methods in drug between in vivo and in vitro data. Accordingly, the
discovery. These technologies, and developments in Scientific Advisory Board for ECVAM unanimously
automation, have resulted in the emergence of cell-based endorsed the EST as a scientifically validated test,
assays, which were previously evaluated in single which is ready to be considered for regulatory purposes.
cuvettes, that can be used in 96, 384 and 1,586 well The next hurdle for alternative screening technology
formats82. The ability to precisely modify the murine will be the reception this approach receives from regu-
genome using gene targeting in ES cells with or without latory agencies in the United States or Europe during
differentiation of the genetically modified cells into var- drug approval.
TERATOMA ious cell types allows the application of this technology To further explore the evaluation of alternative cell-
A rare tumour type that to the identification of novel drugs through HTS. mES based assays for developmental toxicity, investigators
typically arises in the gonads
cells can be modified to incorporate reporter genes developed ES cells lines that express reporters when they
and demonstrates mixed cellular
populations of all three directly into a gene promoter or as a dicistronic fusion differentiate into specific cell lineages. The aim is to over-
embryonic germ layers. gene such that agonists or antagonist compounds that come the limitation in quantitative assessment of drug
Investigators can assess the interact with the gene product can be readily identified in actions by focusing the evaluation of toxicity in selective
differentiation capacity of stem HTS methods. The human target gene or human gene- cell and tissue types, rather than the heterogeneous cell
cells by injection of pluripotent
cells into laboratory animals
reporter fusion products can be expressed through the populations observed during spontaneous in vitro differ-
and inducing the formation of replacement of the murine endogenous orthologue entiation of ES cells. Cell lines that express GFP selec-
teratomas in situ. directly as a knock-in or through complementation in ES tively in cardiomyocytes89 and endodermal lineages90
NATURE REVIEWS | DRUG DISCOVERY VOLUME 3 | JANUARY 2004 | 7 5
mice produced from mES cells up to 140 passages92. We
Box 1 | ES cells in defining gene function: the erythropoiesis example
have recently developed protocols to significantly
A robust readout in a cell-based assay is required to define gene function, whether using increase the scale of mES cell-derived embryoid bodies
genetic, biological or chemical tools. Colorimetric reporters are commonly introduced through bioreactor culture methods. These embryoid
to monitor changes in gene expression in cell-based assays. A naturally occurring bodies can undergo cryopreservation and reproducible
thawing to derive differentiated cell types with expan-
colorimetric reporter is the expression of haemoglobin in erythrocytes. In order to
sion and inductive protocols (J. McNeish, unpublished
evaluate genes that regulate erythropoiesis, potential targets can be inactivated by
data). This embryoid bodies protocol allows for the
homologous recombination in undifferentiated ES cells which are then differentiated rapid delivery of differentiated cells, with or without
into red blood cells. The dual-specificity tyrosine-regulated kinase (DYRK) family are genetic modifications, for analysis. Protocols are also
conserved protein kinases implicated as regulators of cellular growth and development. being developed for the expansion and cryopreserva-
A member of the DYRK family that is principally expressed in erythroid progenitor cells tion of specific ES cell-derived precursor or terminally
is the regulatory erythroid kinase (REDK). To understand the functional role of REDK differentiated lines (FIG. 4). Some cells cannot be cryo-
in the development of mature red blood cells, REDK –/– ES cells were derived and preserved and recovered in the differentiated state (for
differentiated according to established erythropoiesis protocols51. As demonstrated in the example, neurons). However, protocols have been estab-
figure, with in vitro differentiation a significant increase in the number of erythrocytes is lished that enable the production of functional neurons
consistently observed in the REDK –/– ES cells compared with wild-type control ES cells. from thawed progenitors93. Compared with undifferen-
tiated ES cells, the delivery of embryoid bodies, pre-
This indicates that the expression of REDK is important in the early stages of
cursor or progenitor cells is preferred by cell-culture
erythropoiesis and functions as a lineage-restricted, stage-specific suppressor of the
experts in various disease areas, which therefore
development of red blood cells. This example demonstrates the advantages stem cells can increases the opportunities to use these reagents in drug
bring to functional genomic approaches, including a reliable cell-based assay that allows discovery programmes.
the modification of gene expression or protein function using various methods in a rapid
and potentially high-throughput manner118. Human gene variant screening
The tremendous progress in human genomics has
Wild type REDK knockout resulted in the identification of variants of target genes
from ethnic populations around the world; however, in
most instances little is known about the function of
these gene variants. Knowledge of the effects of single
nucleotide polymorphisms (SNP) is required to deter-
mine whether genetic variants result in altered expres-
sion of the target gene, or alters the response of the gene
product to therapeutic compunds. An improved under-
standing of functional human variants could facilitate
several aspects of the drug discovery process, from
target validation, understanding pharmacokinetic dif-
ferences and improving drug safety, to identifying those
subjects to include/exclude in clinical trials. Functional
have been reported. To facilitate this assessment, specific pharmacogenomics approaches to determine the rele-
approaches were developed, including FACS for car- vance of DNA variation could become a specific area of
diomyocytes and a fluorescence microscope image drug discovery in which ES cells become widely used.
analysis for endodermal cells within embryoid bodies. The ability to precisely modify target sequences in
Compared with immunological or molecular readouts, mES cells has enabled the development of GeMM with
fluorescence-based approaches allow the quantifiable engineered variations in conserved DNA codons in the
measurement of dose-dependency in toxicity assays and endogenous murine genome (knock-in), as well as
development of high-throughput methods. A limitation the generation of transgenic mice in which murine
of in vitro teratology assays is the absence of maternal genes have been replaced with human paralogues con-
metabolism enzymes in the system, as evidenced by the taining specific SNPs. An in vivo approach to a target
inability of the EST to identify cyclophosphamide, a gene of interest with multiple GeMM lines for specific
strong pro-teratogen that forms the active metabolite on human variants has limitations including the time to
first-pass metabolism89,91. deliver the variant GeMM, the cost of maintaining
several GeMM lines and the chemical requirements for
Scaling up growth of ES cells the in vivo validation of potential drugs. Therefore, an ES
For stem cells to be applied to drug discovery, methods cell-based in vitro approach to evaluate multiple human
will need to be developed that facilitate the operational variants is a preferred strategy and hastens the under-
use of these reagents. Specifically, this will require the standing of variant function in drug discovery processes.
ability to stably expand and preserve undifferentiated ES Of course, the use of murine cells to understand
cells, or ES cell-derived cells, at a scale required for cell- human gene function by complementation must be rig-
based screening. Undifferentiated mES cells can be orously scrutinized. Although the completion of the
grown almost indefinitely and maintain pluripotentcy, sequencing of the human and mouse genomes has
as demonstrated by germline competency in chimeric demonstrated robust synteny between these two
76 | JANUARY 2004 | VOLUME 3 www.nature.com/reviews/drugdisc
numerous passages and population doublings without
affecting their genetic stability97. An evaluation of hES
cells, in comparison with mES cells, demonstrated
Undifferentiated ES cells similar expression of POUSF1 and telomerase94,97.
However, hES cells differ from mES cell lines in their
expression of cell-surface markers98, and the rate of
growth of hES cells is slower, with reported doubling
times approximately threefold longer than for mES
cells99. Unlike mES cells, hES cells do not maintain an
Embryoid bodies undifferentiated state and proliferate in the presence of
LIF. This limitation was overcome when investigators at
Geron97 reported the development of feeder-independent
methods for growing hES cells that use mES cell-con-
ditioned media. These methods will facilitate the
scaled-up growth of hES cells that is required for their
Precursors use in drug discovery, and regenerative or transplanta-
hES cells have demonstrated in vitro differentiation
into functionally relevant derivatives of the three
embryonic germ layers100,101. Inductive in vitro differenti-
ation of hES cells using eight different growth factors
Differentiated cells followed by expression profiling for 24 different cell-
specific transcripts has been evaluated102. This compre-
hensive analysis indicated that the human cells have a
Figure 4 | Growth and delivery of ES cells and derivatives for use in drug discovery broad capacity for enrichment into numerous cell types.
applications. The figure illustrates a typical ES cell differentiation from undifferentiated ES cells Cardiomyocytes derived from spontaneous differentia-
(a) to embryoid bodies (b), followed by precursor cells (c), and fully differentiated and purified cell tion have functional characteristics, including subcellular
populations (d). The ES cell differentiation used in this figure as an example is a macrophage. structures, cell-specific protein expression, Ca2+ flux
Methods have been developed to scale the growth at each level of the differentiation protocol.
with electrical activation and appropriate pharmaco-
Additionally, cryopreservation and thawing techniques at each specific stage of the
differentiation process enable rapid utilization of these reagents by drug discovery investigators.
logical responses; furthermore, these cells show positive
Many cell types — for example neuronal cell-types — cannot be cryopreserved and thawed at and negative chronotropic responses with β-agonists and
the fully differentiated state. muscarinic agonists, respectively103. Inductive proto-
cols to identify enriched populations of neuronal cells
and precursors have been reported104,105. The neuronal
species, there are hundreds of genes that are not found precursors, when transplanted to the ventricles of
in both species and significant differences in the expres- neonatal mice, resulted in widespread migration and
sion pattern of many genes. The use of hES cells or adult differentiation into mature neuronal and glial cells in vivo,
stem cells that are engineered to express specific variants, without the formation of teratomas105. The generation
or which are isolated from individuals or ethnic popula- of human hepatocyte-like cells from hES cells has been
tions harbouring specific variants, will be invaluable in reported using inductive procedures106. This result
overcoming this limitation. might lead to reproducible populations of human hepa-
tocytes for crucial metabolic studies in drug discovery.
hES cells hES-cell-derived endothelial cells were isolated from
This review has focused primarily on the opportunities embryoid bodies using antibodies to PECAM1 (REF. 107).
for mES cells to contribute to human drug discovery. These cells were transplanted to severe combined
Indeed, an interpretation of the relevance of findings in immunodeficiency mice in biodegradable polymer scaf-
murine physiology and the human situation is required folds and removed after one-week, which resulted in the
in the application of mES cells; however, the scientific in vivo formation of functional human microvessels
power of mES cells make them a welcome resource in containing murine blood after anastamosis with the
drug discovery. In the future, investigators might gain murine vasculature.
the ultimate in predictive cellular systems, hES cells. In The discoveries and technical breakthroughs that are
1998, Thompson and colleagues94 reported the estab- being made with hES cells in culture conditions, and the
lishment of an hES cell line derived from explanted establishment of protocols to derive essentially all cell
human blastocysts that were donated from in vitro fertil- types through in vitro differentiation, hold promise for
ization clinics with informed consent and approval by the eventual use of hES cells in regenerative medicine.
the associated university’s institutional review board. However, the application of hES cells in drug discovery
Additional hES cell lines have been reported from blasto- will be greatly facilitated by the ability to overexpress or
cysts95 and from primordial germ cells96. Undiffer- modify endogenous target genes of interest in these cells.
entiated hES cells demonstrate many characteristics Eiges et al.108 evaluated the stable transfection efficiency
similar to well-established murine cells. The hES cells of various conditions of a murine Rex1 promoter–eGFP
have been maintained in an undifferentiated state for fusion transgene. The authors determined that hES cells
NATURE REVIEWS | DRUG DISCOVERY VOLUME 3 | JANUARY 2004 | 7 7
differed from mES cells with respect to the efficiency of Pharmaceutical and biotechnology companies rely on
the methods for the delivery of DNA to the two cell information gained from GeMM to contribute to var-
types. Specifically, the cationic liposome ExGen 500 ious aspects in the drug discovery pipeline; however,
(Fermentas) demonstrated an approximately tenfold increased throughput and reduced time periods for
increase in transfection efficiency compared with target analysis and evaluation are preferred. In-
other chemical reagents or electroporation. The pre- creasingly, investigators are demanding precise tempo-
ferred method for transfection in mES cells is electro- ral and spatial control of the modification of the
poration; however, the authors reported that hES cells murine target gene using inducible and conditional
did not survive these conditions well. Rex1 is expressed technologies to mimic drug-like effects in vivo. This tech-
in murine pre-implantation embryos and ES cells, nology is very powerful; however, the time and cost of
and is downregulated with ES cell differentiation. delivering the specific GeMM required make this
Therefore Rex1–eGFP expression can serve as a marker approach limiting. Consequently, in vitro differentia-
for purified populations or for identifying isolates of tion of ES cells could enhance the throughput of tar-
pluripotent ES cells. gets, and reduce the time required to deliver the
In addition, hES cells demonstrated the ability to assessment of a gene product’s mechanism of action
stably express gain-of-function transgenes using self- and safety as a novel therapeutic target.
inactivating human lentiviral vectors109. These lentiviral ES-cell-based approaches are aligned with advances
vectors, which carry an eGFP reporter driven by the in molecular biology, including bacterial artificial
ubiquitous human elongation factor-1α promoter chromosome expression, RNA interference, inducible
(EF1α–eGFP) resulted in high-level and sustained (>60 systems and so on. In addition, the use of cell-based
days) transgene expression in hES cells. The reporter methods have overcome issues related to developmental
expression was not compromised by IVD into CD34+ consequences associated with genetic modifications in
haematopoietic cells, which demonstrates that the some GeMM, and can now be applied directly to HTS
lentiviral transfection did not affect hES differentiation paradigms. Therefore, the application of ES cells and
capacity. Notably, the high transfection efficiency will IVD could eventually replace many current applications
allow expression cloning with complex cDNA libraries of the more time-consuming and cost-prohibitive uses
to identify specific genes that result in cell-based pheno- of GeMM. The crucial issues that are presently faced in
typic selection methods, which is a promising approach the industrial application of ES cell IVD are consistent
in functional genomics using human cells. and reproducible delivery of relatively pure populations
Recently, Zwaka and Thomson110 demonstrated that of specific differentiated cell types, phenotypic and
gene targeting by homologous recombination can be pharmacological evaluation of cell types, and the delivery
applied to hES cells. These investigators established new of cells at the scale needed for drug discovery applica-
electroporation protocols for hES cells on the basis of tions in screening. With the increased focus on stem-cell
the increased size of human cells compared with mouse applications, optimized protocols for differentiation
cells, and the observation that human cells cannot be and scaling should overcome these limitations and
efficiently cloned from single cells107. Therefore, the hES deliver benefits to drug discovery in terms of cost, time
cells were electroporated in clumps and plated at high and capacity.
density, in contrast to single cells when using mES cells. Drug discovery has relied on numerous methods
Using this electroporation procedure, HPRT and to identify novel therapeutics; undoubtedly the
POU5F1 were successfully targeted in the hES cells. The opportunity to utilize homogeneous human reagents
POU5F1-targeting vector used a construction that will affect the drug discovery process. The ability to
inserted an eGFP reporter. POU5F1 is expressed specifi- precisely modify endogenous loci in hES cells will
cally in undifferentiated ES cells and the GFP KI hES lead to an improved understanding of gene function
cells express the reporter only in the undifferentiated and, when applied to druggable genes, could demon-
state and are turned off after differentiation. As with the strate the clinical significance of targets for novel
rex1–eGFP transgenic hES cells108, the POU5F1–eGFP therapeutics. Genetically stable human cellular reagents
KI cell line will be important in optimizing the hES cell will improve the attrition rate of drug candidate
culture conditions. As stated previously, mES cells are compounds, and result in new and improved oppor-
readily manipulated using virally mediated gene trap- tunities to identify safe and efficacious new medi-
ping and chemical mutagenesis. In the future, these cines. At present, established hES cell lines are used
random approaches could play an important role in for all federally sponsored research with these cells;
the application of human stem cells to drug discovery however, these cells are scarce and therefore availabil-
and development. ity to them is low113. Although human, pluripotent
adult stem-cell lines are being identified from bone
A forward-looking view marrow 114,115, cord blood 116, adipose tissue 117 and
Although, the sequencing of the human genome has other sources, it is questionable whether these cells
been essentially completed, the promise of the informa- will have the broad differentiation capacity and
tion contained in this mass of data for delivering new growth potential of ES cells. Although stem cells offer
therapeutics is years away. The race now is to understand great promise for regenerative medicine, the near-
the chemically druggable target space of the human term applications of stem cells will probably be in
genome, which is apparently shrinking rapidly111,112. drug discovery. Drug companies have begun to
78 | JANUARY 2004 | VOLUME 3 www.nature.com/reviews/drugdisc
develop policies to work with human pluripotent store of stem cells lines derived from human popula-
stem cells and will begin applying these cells to specific tions of numerous ethnic populations and human
programmes that can affect the discovery of the next gene variations. These cells will provide a qualitative
generation of medicines. The recent demonstration advantage to investigators that previously used aneu-
that gene targeting has proven successful for two ploid human cells or animal models in discovery
genes in hES cells is encouraging and investigators research. These advantages will probably result in
could soon be creating KO cell lines for functional stem cells becoming invaluable in high-throughput
evaluation and KI cells lines for screening applica- drug discovery and the development of effective and
tions. In the future, drug discoverers might have a safe new therapeutics.
1. Harris, S. Transgenic knockouts as part of high-throughput, 19. Doetschman, T. C., Eistetter, H., Katz, M., Schmidt, W. & 38. Wichterle, H., Lieberam, I., Porter, J. A. & Jessell, T. M.
evidence based target selection and validation strategies. Kemler, R. The in vitro development of blastocyst-derived Directed differentiation of embryonic stem cells into motor
Drug Discov. Today 6, 628–636 (2001) embryonic stem cell lines: formation of a visceral yolk sac, neurons. Cell 119, 385–397 (2002).
2. Abuin, A., Holt, K. H., Platt, K. A., Sands, A. T. & blood islands and myocardium. J. Embryol. Exp. Morphol. 39. Yamane, T. et al. Development of osteoclasts from
Zambrowicz, B. P. Full-speed mammalian genetics; in vivo 87, 27–45 (1985). embryonic stem cells through a pathway that is c-fms but
target validation in the drug discovery process. Trends The first report of in vitro differentiation in murine not c-kit dependent. Blood 90, 3516–3523 (1997).
Biotechnol. 20, 261–268 (2002). embryonic stem cells. The authors describe the 40. Berthier, R. et al. The MS-5 murine stromal cell line and
References 1 and 2 are excellent reviews on the appli- spontaneous differentiation potential of the ES cells hematopoietic growth factor synergize to support the
cation of genetically modified mice in drug discovery. and offer a forward-looking perspective of this megakaryocyte differentiation of embryonic stem cells.
3. Crawley, J. N. Behavioral phenotyping of transgenic and observation. Exp. Hematol. 25, 481–490 (1997).
knockout mice: experimental design and evaluation of 20. Williams, R. L. et al. Myeloid leukemia inhibitory factor 41. Stevens, L. C. The development of transplantable
general health, sensory functions, motor abilities and maintains the developmental potential of embryonic stem teratocarcinomas from intratesticular grafts of pre- and
specific behavioral tests. Brain Res. 835, 18–26 (1999). cells. Nature 336, 684–687 (1988). postimplantation embryos. Dev. Biol. 21, 364–382 (1970).
4. van der Staay, F. J. & Steckler, T. Behavioral phenotyping 21. Wobus, A. M., Grosse, R. & Schoneich, J. Specific effects of 42. Damjanov, I., Damjanov, J. & Solter, D. in Teratocarcinomas
of mouse mutants. Behav. Brain Res. 125, 3–12 (2001). nerve growth factor on the differentiation pattern of mouse and Embryonic Stem Cells: A Practical Approach (ed
5. Doevendans, P. A., Daemen, M. J., de Muinck, E. D. & embryonic stem cells in vitro. Biomed. Biocem. Acta. 47, Robertson, E. J.) 1–18 (IRL Oxford, Washington, 1987).
Smits, J. F. Cardiovascular phenotyping in mice. 965–973 (1988). 43. Klug, M. G., Soonpaa, M. H., Koh, G. Y. & Fields, L. J.
Cardiovasc. Res. 39, 34–49 (1998). 22. Robbins, J., Gulick, J., Sanchez, A., Howles, P. & Genetically selected cardiomyocytes from differentiating
6. Weissleder, R. Scaling down imaging: molecular mapping of Doetschman, T. Mouse embryonic stem cells express the embryonic stem cells form stable intracardiac grafts.
cancer in mice. Nature Rev. Cancer 2, 11–18 (2002). cardiac myosin heavy chain genes during development J. Clin. Invest. 98, 216–224 (1996).
7. Massoud, T. F. & Gambhir, S. S. Molecular imaging in living in vitro. J. Biol. Chem. 265, 11905–11909 (1990). References 41–43 clearly establish lineage selection
subjects: seeing fundamental biological processes in a new 23. Maltsev, V. A., Rohwedel, J., Herscheler, J. & Wobus A. M. as an important method in the derivation and
light. Genes Dev. 17, 545–580 (2003). Embryonic stem cells differentiate in vitro into purification of some in vitro differentiated ES cells.
8. Zambrowicz, B. P. & Sands, A. T. Knockouts of the 100 cardiomyocytes representing sinusoidal, atrial and 44. Mountford, P. et al. Dicistronic targeting constructs:
best-selling drugs — will they model the next 100? Nature ventricular cell types. Mech. Dev. 44, 41–50 (1993). reporters and modifiers of mammalian gene expression.
Rev. Drug Discov. 2, 38–51 (2003). 24. Schmitt, R. M., Bruyns, E. & Snodgrass, H. R. Proc. Natl Acad. Sci. USA 91, 4303–4307 (1994).
An important review aligning genetically modified Hematopoietic development of embryonic stem cells in vitro: 45. Li, M., Pevny, L., Lovell-Badge, R., & Smith, A. Generation
mice with therapeutic target evaluation. The authors cytokine and receptor gene expression. Genes Dev. 5, of purified neural precursors from embryonic stem cells by
describe the phenotypes associated with KO mice for 728–740 (1991). lineage selection. Curr. Biol. 8, 971–974 (1998).
targets of many of the most successful medicines and 25. Wiles, M. V. & Keller, G. Multiple hematopoietic lineages 46. Kolossov, E. et al. Functional characteristics of ES cell-derived
conclude that this is a cost-effective strategy for develop from embryonic stem (ES) cells in culture. cardiac precursor cells identified by tissue-specific expression of
defining novel target mechanisms for the next Development 111, 259–267 (1991). green fluorescent protein. J. Cell Biol. 143, 2045–2056 (1998).
generation of medicines. 26. Chen, U. Differentiation of mouse embryonic stem cells into 47. Allen, M. et al. Deficiency of the stress kinase p38α results in
9. Zhang, S. H., Reddick, R. L., Piedrahita, J. A. & Maeda, N. lympho-hematopoietic lineages in vitro. Dev. Immunol. 2, embryonic lethality: characterization of the kinase
Spontaneous hypercholesterolemia and arterial lesions in 29–50 (1992). dependence of stress responses of enzyme-deficient
mice lacking apolipoprotein E. Science 258, 468–471 (1992). 27. Keller, G., Kennedy, M., Papayannopoulou, T. & Wiles, M. V. embryonic stem cells. J. Exp. Med. 191, 859–870 (2000).
10. Plump, A. S. et al. Severe hypercholesterolemia and athero- Hematopoeitic commitment during embryonic stem cell This paper describes how murine ES cell in vitro
sclerosis in apolipoprotein E-deficient mice created by homo- differentiation in culture. Mol. Cell. Biol. 13, 473–486 (1993). differentiation was used to overcome the embryonic
logous recombination in ES cells. Cell 71, 343–353 (1992). 28. Strubing, C. et al. Differentiation of pluripotent embryonic lethality of the target gene KO mouse and detail the
11. Beierschmitt, W. P. et al. Induction of hepatic microsomal stem cells into neuronal lineages in vitro gives rise to mature role of the p38 kinase in inflammatory signalling. The
drug-metabolizing enzymes by inhibitors of 5-lipoxygenase inhibitory and excitatory neurons. Mech. Dev. 53, 275–287 results were important to drug discovery investigators
(5-LO) studies in rats and 5-LO knockout mice. Toxicol. Sci. (1995). and the potential impact of in vitro-based stem-cell
63, 15–21 (2001). 29. Fraichard, A. et al. In vitro differentiation of embryonic stem technology was highlighted.
12. Davis, J. A. et al. An Alzheimer’s disease-linked PS1 variant cells into glial and functional neurons. J. Cell Sci. 108, 48. Kim, J.-H. et al. Dopaminergic neurons derived embryonic
rescues the developmental abnormalities of PS1-deficient 3181–3188 (1995). stem cells function in an animal model of Parkinson’s
embryos. Neuron 20, 603–609 (1998). 30. Desbaillets, I., Ziegler, U., Groscurth, P. & Gasserman, M. disease. Nature 418, 50–56 (2002).
13. Grilli, M. et al. Presenilin-1 regulates the neuronal threshold Embryoid body: an in vitro model of mouse embryogenesis. 49. Lee, M. A. et al. Overexpression of midbrain-specific trans-
to excitotoxicity both physiologically and pathologically. Exp. Physiol. 85, 645–651 (2000). cription factor Nurr1 modifies susceptibility of mouse neural
Proc. Natl Acad. Sci. USA 97, 12822–12827 (2000). 31. Dani, C. et al. Differentiation of embryonic stem cells into stem cells to neurotoxins. Neurosci. Lett. 333, 74–78 (2002).
14. Wang, Y., Schnegelsberg, P. N., Dausman, J. & adipocytes in vitro. J. Cell Sci. 110, 1279–1285 (1997). 50. Blyszczuk, P. et al. Expression of Pax4 in embryonic stem
Jaenisch, R. Functional redundancy of the muscle-specific 32. Kramer, J., Hegeret, C., Guan, K., Wobus, A. M., Muller, P. K. cells promotes differentiation of nestin positive progenitor
transcription factor Myf5 and myogenin. Nature 379, & Rohedel, J. Embryonic stem cell-derived chondrogenic and insulin-producing cells. Proc. Natl Acad. Sci. USA 100,
823–825 (1996). differentiation in vitro: activation by BMP-2 and BMP-4. 998–1003 (2003).
15. Prosser, H. M. et al. Targeted replacement of rodent CCR2 Mech. Dev. 92, 193–205 (2000). 51. Bagutti, C., Wobus, A. M., Fassler, R. & Watt, F. M.
with the human orthologue CCR2B: a mouse model for in 33. Hegert, C. et al. Differentiation plasticity of chondrocytes Differentiation of embryonic stem cells into keratinocytes:
vivo analysis of human target-selective small molecule MCP-1 derived from mouse embryonic stem cells. J. Cell Sci. comparisons of wild-type and β1-integrin deficient cells.
receptor antagonists. Drug Discov. Res. 55, 197–209 (2002). 115, 4617–4628 (2002). Dev. Biol, 179, 184–196 (1996).
16. Shiao, L. L., Cascieri, M. A., Trumbauer, M., Chen, H. & 34. Boheler, K. R., Czyz, J., Tweedie, D., Yang, H. T., 52. Fairchild, P. J. et al. Directed differentiation of dendritic cells
Sullivan, K. A. Generation of mice expressing the human Anisimov, S. V. & Wobus, A. M. Differentiation of from mouse embryonic stem cells. Curr. Biol. 10,
glucagons receptor with a direct replacement vector. pluripotent embryonic stem cells into cardiomyocytes. 1515–1518 (2000).
Transgenic Res. 8, 295–302 (1999). Circ. Res. 91, 189–201. (2002). 53. Nakano, T., Kodama, H. & Honjo, T. In vitro development
17. Zambrowicz, B. P. et al. Wnk1 kinase deficiency lowers 35. Angelov, D. N. et al. Temporospatial relationships of primitive and definitive erythryocytes from different
blood pressure in mice: A gene-trap screen to identify between macroglia and microglia during in vitro precursors. Science 272, 722–724 (1996).
potential targets for therapeutic intervention. Proc. Natl differentiation of murine stem cells. Dev. Neurosci. 54. Risau, W. et al. Vasculogenesis and angiogenesis in
Acad. Sci. USA 10 November 2003 20, 42–51 (1998). embryonic-stem-cell-derived embryoid bodies.
(doi:10.1073pnas.2336103100). 36. Brustle, O. et al. Embryonic stem cell-derived glial Development 102, 471–478 (1988).
18. Vivian, J. L., Chen, Y., Yee, D., Schneider, E. & Magnuson, T. precursors: a source of myelinating transplants. Science 55. Rohwedel, J. et al. Muscle cell differentiation of embryonic
An allelic series of mutations in Smad2 and Smad4 identified 285, 754–756 (1999). stem cells reflects myogenesis in vivo: developmentally
in genotype-based screen of N-ethyl-N-nitrosourea- 37. Lumelsky, N. et al. Differentiation of embryonic stem cells regulated expression of myogenic determination genes and
mutagenized mouse embryonic stem cells. Proc. Natl Acad. to insulin-secreting structures similar to pancreatic islets. functional expression of ionic currents. Dev. Biol 164,
Sci. USA 99, 15542–15547 (2002). Science 292, 1389–1394 (2001). 87–101 (1994).
NATURE REVIEWS | DRUG DISCOVERY VOLUME 3 | JANUARY 2004 | 7 9
56. Buttery, L. D. et al. Differentiation of osteoblasts and in vitro 82. Hertzberg, R. P. & Pope, A. J. High-throughput screening: 103. Kehat, I. et al. Human embryonic stem cells can
bone formation from murine embryonic stem cells. J. Cell new technology for the 21st century. Curr. Opinion Chem. differentiate into myocytes with structural and functional
Sci. 110, 1279–1285 (1997). Biol. 4, 445–451 (2000). properties of cardiomyocytes. J. Clin. Invest. 108,
57. Jones, E. A., Tosh, D., Wilson, D. I., Lindsay, S., & 83. Laschinski, G., Vogel, R. & Spielman, H. Cytotoxicity test 407–414 (2001).
Forrester, L. M. Hepatic differentiation of murine embryonic using blastocyst-derived euploid embryonal stem cells: a 104. Schuldiner, M. et al. Induced neuronal differentiation of
stem cells. Exp. Cell Res. 272, 15–22 (2002). new approach to in vitro teratogenesis screening. Reprod. human embryonic stem cells. Brain Res. 913, 201–205
58. Ali, N. N. et al. Derivation of type II alveolar epithelial cells from Toxicol. 5, 57–64 (1991). (2001).
murine embryonic stem cells. Tissue Eng. 8, 541–550 (2002). 84. Spielman, H., Pohl, I., Doring, B., Liebsch, M. & 105. Zhang, S.-C., Werning, M., Duncan, I. D., Brustle, O. &
59. Hubner, K. et al. Derivation of oocytes from mouse Moldenhauer, F. The embryonic stem cell test, an in vitro Thomson, J. A. In vitro differentiation of transplantable
embryonic stem cells. Science 300, 1251–1256 (2003). embryotoxicity test using two permanent mouse cell lines neural precursors from human embryonic stem cells.
60. Toyooka, Y., Tsunekawa, N., Akasu, R. & Noce, T. 3T3 fibroblasts and embryonic stem cells. Toxicol. In Vitro Nature Biotechnol. 19, 1129–1133 (2001).
Embryonic stem cells can form germ cells in vitro. Proc. Natl 10, 119–127 (1997). 106. Rambhatla, L., Chiu, C. P., Kundu, P., Peng, Y. &
Acad. Sci. USA 100, 11457–11462 (2003). 85. Balls, M. & Hellstein, E. Statement on the scientific validity of Carpenter, M. K. Generation of hepatocyte-like cells
61. Nichols, J. et al. Formation of pluripotent stem cells in the the embryonic stem cell test (EST) — an in vitro test for from human embryonic stem cells. Cell Transplant.
mammalian embryo depends on POU transcription factor, embryotoxicity. Altern. Lab. Anim. 30, 265–268 (2002). 12, 1–11 (2003).
Oct 4. Cell 95, 379–391 (1998). 86. Rudnicki, M. A. & McBurney, M. W. in Teratocarcinomas and 107. Levenberg, S., Golub, J. S., Amit, M., Itskovitz-Elder, J. &
62. Chambers, I. et al. Functional expression cloning of Nanog, Embryonic Stem Cells: A Practical Approach (ed Robertson, Langer, R. Endothelial cells derived from human embryonic
a pluripotency sustaining factor in embryonic stem cells. E. J.) 19–49 (IRL Oxford, Washington DC, 1987). stem cells. Proc. Natl Acad. Sci. USA 99, 4391–4396
Cell 113, 643–655 (2003). 87. Brown, N. Selection of test chemicals for the ECVAM (2002).
63. Mitsui, K. et al. The homeoprotein Nanog is required for international validation study on in vitro embryotoxicity test. 108. Eiges, R. et al. Establishment of human embryonic stem
maintenance of pluripotency in mouse epiblast and ES cells. Altern. Lab. Anim. 30, 177, 198 (2002). cell-transfected clones carrying a marker for undifferentiated
Cell 113, 631–642 (2003). 88. Spielman, H. et al. Preliminary results of the ECVAM cells. Curr. Biol. 11, 514–518 (2001).
64. Ramalho-Santos, M., Yoon, S., Matsuzaki, Y., Mulligan, R. C. validation study on three in vitro embryotoxicity tests. 109. Ma, Y., Ramezani, A., Lewis, R., Hawley, R. G. &
& Melton, D. A. “Stemness”: transcriptional profiling of Altern. Lab. Anim. 29, 301–330 (2001). Thomson, J. A. High-level sustained transgene expression
embryonic and adult stem cells. Science 298, 597–600 References 83–85, and 87 and 88, describe the in human embryonic stem cells using lentiviral vectors.
(2002). validated use of murine ES cells as an alternative Stem Cells 21, 111–117 (2003).
65. Ivanova, N. B. et al. A stem cell molecular signature. Science to animal testing in teratology and embryotoxicity 110. Zwaka, T. P. & Thomson, J. A. Homologous recombination
298, 601–604 (2002) testing. This embryonic stem cell test demonstrated in human embryonic stem cells. Nature Biotechnol. 21,
66. D’Amour, K. A. & Gage, F. H. Genetic and functional excellent alignment in predicting embryotoxicity with 319–321 (2003).
differences between multipotent neural and pluripotent known teratogens in vivo. The first demonstration that human ES cells maintain
embryonic stem cells. Proc. Natl Acad. Sci. USA 100, 89. Bremer, S. et al. Establishment of an in vitro reporter assay the genetic machinery capable of directing
11866–11872 (2003). for the development of cardiac toxicity. Toxicol. In Vitro 15, homologous recombination at frequencies previously
67. Potter, S. S., Valerius, M. T. & Brunskill, E. W. Using 215–223 (2001). observed only in murine ES cells. The investigators
progenitor cells and gene chips to define genetic pathways. 90. Paparella, M., Kolossov, E., Fleischmann, B. K., Hescheler, J. established novel protocols for achieving high
Methods Mol. Biol. 185, 269–284 (2002). & Bremer, S. The use of quantitative image analysis in the transfection efficiencies based on cellular differences
68. Hemmer, R., Wei, W., Dutriaux, A. & Sedivy, J. Somatic cell assessment of in vitro embryotoxicity endpoints based on a between murine and human ES cells.
knockouts of tumor suppressor genes. Methods Mol. Biol. novel embryonic stem cell clone with endoderm-related 111. Walke, D. W. et al. In vivo drug target discovery: identifying
223, 187–206 (2003). GFP expression. Toxicol. In Vitro 16, 589–597 (2002). the best targets from the genome. Curr. Opin. Biotechnol.
69. Hatada, S., Nikkuni, K., Bentley, S. A., Kirby, S. & Smithies, O. 91. Bremer, S., Pellizzer, C., Coecke, S., Paparella, M. & 12, 626–631 (2001).
Gene correction in hematopoietic progenitor cells by Catalani, P. Detection of the embryotoxic potential of 112. Hopkins, A. L. & Groom, C. R. The druggable genome.
homologous recombination. Proc. Natl Acad. Sci. USA 97, cyclophosphamide by using a combined system of Nature Rev. Drug Discov. 1, 727–730 (2002).
13807–13811 (2000). metabolic competent cells and embryonic stem cells. 113. Holden, C. & Vogel, G. “Show us the cells” U. S. researchers
70. Abuin, A & Bradley, A. Recycling selectable markers in mouse Altern. Lab. Anim. 30, 77–85 (2002). say. Science 297, 923–925 (2002).
embryonic stem cells. Mol. Cell. Biol. 16, 1851–1856 (1996). 92. Suda, Y., Suzuki, M., Ikawa, Y. & Aizawa, S. Mouse 114. Pittenger, M. F. et al. Multilineage potential of adult human
71. Mortenson, R. M., Conner, D. A., Chao, S., Geisterfer- embryonic stem cells exhibit indefinite proliferation potential. mesenchymal stem cells. Science 284, 143–147 (1999).
Lowrance, A. A. & Seidman, J. G. Production of J. Cell Physiol. 133, 197–201 (1987). 115. Jiang,Y. et al. Pluripotency of mesenchymal stem cells
homozygous mutant ES cells with single targeting 93. Hancock, C. R., Wetherington, J. P., Lambert, N. A. & derived from adult bone marrow. Nature 418, 41–49
construct. Mol. Cell. Biol. 12, 2391–2395 (1992). Condie, B. G. Neuronal differentiation of cryopreserved (2002).
72. Langa, F. et al. Teratocarcinomas induced by embryonic neural progenitor cells derived from mouse embryonic 116. Broxmeyer, H. E. et al. High efficiency recovery of functional
stem (ES) cells lacking vimentin: an approach to study the stem cells. Biochem. Biophys. Res. Commun. 271, hematopoietic progenitor and stem cells from human cord
role of vimentin in tumorigenesis. J. Cell Sci. 113, 418–421 (2000). blood cryopreserved for 15 years. Proc. Natl Acad. Sci. USA
3463–3472 (2000). 94. Thomson, J. A. et al. Embryonic stem cells lines derived 100, 645–650 (2003).
73. Ishikawa, H., Ryseck, R. P. & Bravo, R. Characterization of from human blastocysts. Science 282, 1145–1147 (1998). 117. Zuk, P. A. et al. Human adipose tissue is a source of
ES cells deficient for the p105 precursor (NF-κB1): role of The first report of a human embryonic stem cell line multipotent stem cells. Mol Biol. Cell 13, 4279–4295 (2002).
p50 NLS. Oncogene 13, 255–263 (1996). established from a blastocyst-stage embryo. 118. Roach, M. et al. Attenuation of erythropoiesis by REDK
74. Fassler, R. et al. Differentiation and integrity of cardiac 95. Reubinoff, B. E., Peras, M. F., Fong, C. Y., Trounson, A. & (DYRK3) in an embryonic stem cell-based in vitro
muscle cells are impaired in the absence of β1 integrin. Bongso, A. Embryonic stem cell lines from human differentiation model. First Annu. Mtg Int. Soc. Stem Cell
J. Cell Sci. 109, 2989–2999 (1996). blastocysts: somatic differntiation in vitro. Nature Biotechnol. Res. A86 (2003).
75. Gowen, L. C., Johnson, B. L., Latour, A. M., Sulik, K. K. & 18, 399–404 (2000).
Koller, B. H. BRCA1 deficiency results in early embryonic 96. Shamblott, M. J. et al. Derivation of pluripotent stem cells Acknowledgements
lethality characterized by neuroepithelial abnormalities. from cultured primordial germ cells. Proc. Natl Acad. Sci. The author acknowledges the contributions of G. Cezar, K. Haskell,
Nature Genet. 12, 191–194 (1996). USA 95, 13726–13731 (1998). M. Roach with figures, J. Hambor and K. Neote in generating Box 1,
76. Snouwaert, J. N. et al. BRCA1 deficient embryonic stem The first report of a human embryonic stem cell line P. Vickers for critically reading of the manuscript and T. Kelleher for
cells display a decreased homologous recombination established from primary germ cells of a developing encouragement. Finally without the love and support of my family
frequency and an increased non-homologous embryo. this manuscript would not have been possible.
recombination that is corrected by expression of BRCA1 97. Lebkowski, J. S. et al. Human embryonic stem cells: culture,
transgene. Oncogene 18, 7900–7907 (1999). differentiation and genetic modification for regenerative Competing interest statement
77. Minamino, T. et al. MEKK1 suppresses oxidative stress- medicine applications. Cancer J. 7 (Suppl. 2), S83–S93 The author declares that he has no competing financial interests.
induced apoptosis of embryonic stem cell-derived cardiac (2001).
myocytes. Proc. Natl Acad. Sci. USA 96, 15127–15132 98. Henderson, J. K. et al. Preimplantation human embryos and
(1999). embryonic stem cells show comparable expression of stage Online links
78. Takeshima, H. et al. Embryonic lethality and abnormal specific antigens. Stem Cells 20, 329–337 (2002).
cardiac myocytes in mice lacking ryanodine receptor type 2. 99. Amit, M. et al. Clonally derived human embryonic stem cell DATABASES
EMBO J. 17, 3309–3316 (1998). lines maintain pluripotency and proliferative potential for The following terms in this article are linked online to:
79. Yang, H. T. et al. The ryanodine receptor modulates the prolonged periods in culture. Dev. Biol. 227, 271–278 (2000). LocusLink: http://www.ncbi.nlm.nih.gov/LocusLink/
spontaneous beating rate of cardiomyocytes during 100. Itskovitz-Elder, J. et al. Differentiation of human embryonic CCR2 | glucagon receptor | LIF | MEKK1 | Myf5 | NANOG | nestin |
development. Proc. Natl Acad. Sci. USA 99, 9225–9230 stem cells into embryoid bodies comprising the three p50 | p105 | PAX4 | POUSF1 | RelA | WNK1
(2002). embryonic germ layers. Mol Med. 6, 88–95 (2000).
80. Zhang, X., Morham, S. G., Langenbach, R., Baggs, R. B. & 101. Amit, M. & Itskovitz-Elder, J. Derivation and spontaneous FURTHER INFORMATION
Young, D. A. Lack of cyclooxygenase-2 inhibits growth in differentiation of human embryonic stem cells. J. Anat. 200, Online Mendelian Inheritance in Man:
teratocarcinomas in mice. Exp Cell Res. 254, 232–240 (2000). 225–232 (2002). http://www.ncbi.nlm.nih.gov/Omim/
81. Moore, R., Radice, G. L., Dominis, M. & Kemler, R. The 102. Schuldiner, M., Yanuka, O., Itskovitz-Elder, J., Melton, D. A. Alzheimer’s disease | Parkinson’s disease
generation and in vivo differentiation of murine embryonal & Benvenisty, N. Effects of eight growth factors on the National Institutes of Health Stem Cell Information:
stem cells genetically null for either N-cadherin or differentiation of cells derived from human embryonic stem http://stemcells.nih.gov/index.asp
P-cadherin. Int. J. Dev. Biol. 43, 831–834 (1999). cells. Proc. Natl Acad. Sci. USA 97, 11307–11312 (2000). Access to this interactive links box is free online.
80 | JANUARY 2004 | VOLUME 3 www.nature.com/reviews/drugdisc