1Stem Cell Transplantation Program, Division of Pediatric Hematology/Oncology, Manton Center for Orphan Disease Research, Howard Hughes Medical Institute, Children’s Hospital Boston and
Dana Farber Cancer Institute, 300 Longwood Avenue, Boston, Massachusetts 02115, USA. 2Division of Hematology, Brigham and Women’s Hospital, 75 Francis Street, Boston, Massachusetts
02115, USA. 3Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, Massachusetts 02115, USA. 4Broad Institute, Cambridge, Massachusetts 02142,
USA. 5Harvard Stem Cell Institute, Cambridge, Massachusetts 02138, USA.
The field of stem-cell biology has been catapulted forward by the startling development of reprogramming technology. The
ability to restore pluripotency to somatic cells through the ectopic co-expression of reprogramming factors has created
powerful new opportunities for modelling human diseases and offers hope for personalized regenerative cell therapies.
While the field is racing ahead, some researchers are pausing to evaluate whether induced pluripotent stem cells are
indeed the true equivalents of embryonic stem cells and whether subtle differences between these types of cell might
affect their research applications and therapeutic potential.
stem cells from a patient’s cells. In a momentous contribution, in 2006
Takahashi and Yamanaka illustrated how cell fates can be altered by the
ectopic co-expression of transcription factors1. The manipulation of cell
fates through reprogramming has altered fundamental ideas about the
stability of cellular identity, stimulating major new directions in research
into human disease modelling, tissue differentiation in vitro and cellular
transdifferentiation. Despite heady progress, a major question remains:
are the new induced pluripotent stem (iPS) cells equivalent to the classic
embryonic stem (ES) cells and thus a suitable alternative for research
and therapy? Whereas the initial wave of papers argued convincingly
that the two cell types were functionally equivalent, a more refined
analysis of how iPS cells behave in vitro, coupled with genome-wide
genetic and epigenetic analysis, has revealed numerous subtle but sub-
stantial molecular differences, probably owing to technical limitations
inherent in reprogramming. In this Review, we describe the derivation
of iPS cells, outline the functional assessments of pluripotency, and then
recount how global assessments of gene expression, gene copy num-
ber variation, DNA methylation and chromatin modification provide
a more nuanced comparison of iPS cells and ES cells. We detail how
these features influence the utility of each of these cell types for disease
modelling and therapeutics, and offer predictions for the evolution of
the art of reprogramming somatic cells.
The promise of induced pluripotent stem
cells in research and therapy
Daisy A. Robinton1–5 & George Q. Daley1–5
fter a decade of constraints, pluripotent stem-cell biology is now
a flourishing research area, following the achievement of a long-
standing ambition — the successful derivation of pluripotent
Pluripotent stem cells
The years since Takahashi and Yamanaka’s breakthrough have seen a
flood of papers touting advances in reprogramming technology, includ-
ing alternative methods for reprogramming and the successful deriva-
tion of iPS cells from various cell types. Although the field has advanced
at a breathtaking pace, investigators have recently taken a step back to
more critically evaluate iPS cells relative to ES cells and have endeav-
oured to fully understand how these cell populations differ from one
another in the hope of closing the gap between the two populations.
Taking clues from the data, it seems that researchers should attempt
to define each cell type more accurately and to understand its inherent
properties rather than ask whether these two classes of pluripotent cell
are identical. Although ES cells and iPS cells are arguably equivalent in
all their functions, these cells are bound to harbour subtle differences
and to have distinct but complementary roles in research because of
their distinct origins and modes of derivation. To appreciate the differ-
ences between ES cells and iPS cells, we must first define what it means
to be pluripotent.
The term pluripotency has been assigned to a variety of cell types
with a wide range of functional capacities. In its loosest sense, pluripo-
tent describes a cell that can generate cell types from each of the three
embryonic germ layers: the endoderm, mesoderm and ectoderm. At
the strict end of the range of definitions, however, pluripotent describes
a cell that can give rise to an entire organism, generating every cell
type within that organism. The property of cell pluripotency was first
exposed by Driesch in 1891, when he separated the two cells of an
early sea urchin blastocyst and observed the development of two com-
plete sea urchins2. Many decades later, studies of embryo aggregation
and blastocyst chimaerism by Mintz and colleagues3, Gardner4 and
Brinster5, in the 1960s and 1970s, solidified the idea that the cells of
the inner cell mass of the mouse blastocyst were pluripotent, and the
isolation of mouse teratocarcinoma stem cells and native ES cells by
Evans and Kaufman6 and Martin7, in 1981, ushered in the era of cul-
turing pluripotent stem cells in a dish. The first successful isolation of
human ES cells, by Thomson and colleagues in 1998, brought forth
a surge of excitement in the scientific community and beyond8. The
potential to understand early human development, tissue formation
and differentiation in vitro through studying ES cells seemed to offer
limitless possibilities. The opportunity to model diseases, discover
disease mechanisms and, ultimately, use cell therapy for previously
untreatable conditions was particularly alluring.
The derivation of ES cells from the human embryo, however,
sparked controversy in the United States and led to a presidential
executive order that restricted government funding9. The limited
numbers of stem cell lines that were approved for research lacked
the diversity necessary to address some of the most compelling ques-
tions, particularly those related to modelling and treating disease10.
Most ES cells represented generic cells isolated from presumably
normal embryos — except for those from embryos that had been
tested by pre-implantation diagnostics and found to carry genetic
1 9 J A N U A R Y 2 0 1 2 | V O L 4 8 1 | N A T U R E | 2 9 5
© 2012 Macmillan Publishers Limited. All rights reserved
diseases. The generic lines were not matched to a particular patient,
so products derived from them for transplantation purposes would
face rejection by the transplant recipient’s immune system or neces-
sitate that the recipient receive lifelong therapy with toxic immu-
nosuppressive medication. To compound these limitations, when
human ES cells are cultivated on mouse feeder cells, the human cells
can incorporate mouse components that render the ES cells subject
to immune rejection.
To realize the full potential of ES cells, researchers foresaw that
customized, personalized pluripotent stem cells specific to each
patient would be generated by using somatic-cell nuclear transfer
(SCNT) — the procedure that had been used successfully to clone
Dolly the sheep from adult mammary cells. Nuclear-transfer-gener-
ated ES (ntES) cell lines would capture a patient’s complete genome
in a cell that could be induced to become any tissue, thus allow-
ing differentiation into disease-relevant cells for analysis or cell-
replacement therapy. Despite successful proof of principle in mouse
studies11, and the clear distinctions between generating ntES cells
for medical research and creating cloned blastocysts for reproduc-
tion, the ethical controversy driven by widespread opposition to
human cloning has severely curtailed research into human SCNT.
Only this year, when investigators gained access to a large number of
human oocytes, was the derivation of pluripotent stem cells through
human SCNT accomplished12. However, the investigators in this
study needed to leave the oocyte nucleus intact to derive pluripo-
tent stem cells, so the resultant cells were triploid, thus affording
research applications for these cells but limiting their suitability for
Despite the many hindrances to the study and derivation of
human ES and ntES cells over the past decade, great strides were
being made in understanding the pathways that regulate the main-
tenance and pluripotency of ES cells. This progress was not lost on
those seeking an alternative source of personalized patient-specific
stem cells, and in 2006 Takahashi and Yamanaka announced the suc-
cessful derivation of iPS cells from adult mouse fibroblasts through
the ectopic co-expression of only four genes1. In an elegant screen
of 24 gene candidates selected for their links to ES-cell pluripo-
tency, these researchers found four factors that were sufficient to
reprogram adult fibroblasts into iPS cells: OCT4 (also known as
POU5F1), SOX2, Krüppel-like factor 4 (KLF4) and c-MYC. This
historic contribution inspired an astonishing flurry of follow-up
studies, with successful reprogramming quickly translated to
human fibroblasts13–15 and then to a wide variety of other cell types,
including pancreatic β cells16, neural stem cells17,18, mature B cells19,
stomach and liver cells20, melanocytes21, adipose stem cells22 and
keratinocytes23, demonstrating the seemingly universal capacity to
alter cellular identity.
Mouse and human iPS cells differ in appearance. Mouse iPS-cell
colonies appear more dome-like and refractile than human iPS-cell
colonies. Human iPS-cell colonies are flatter than those of mice and
are akin to a distinct type of pluripotent stem cell that is derived from
the epiblast of the early mouse embryo24, a feature that indicates that
mouse and human iPS cells, like mouse and human ES cells, probably
reflect distinct developmental states (Fig. 1). The pluripotent state
of mouse stem cells is called a ‘naive’ state because it closely resem-
bles the most primitive state, or ground state, of the mouse inner cell
mass; this is different from the more ‘primed’ state of human stem
cells, which proliferate in response to different cytokines, reflecting
the distinct developmental states of these populations25. Regardless
of the method of derivation, iPS cells maintain the key features of ES
cells, including the ability to propagate in culture indefinitely and the
capacity to generate cells from each of the three embryonic germ layers
(see ref. 26 for a review). Such broad similarities are not proof that iPS
cells are molecularly or functionally equivalent to ES cells. Yamanaka’s
intention was to derive an alternative source of pluripotent stem cells
with the same range of functions as ES cells but offering even greater
potential for clinical use. To determine the degree of success garnered
by reprogramming, we must explore the set of assays that were devel-
oped to assess the key characteristic of ES cells: pluripotency.
Assessment of pluripotency
In the past few years, consistent standards for the identification and
evaluation of iPS cells and for the assessment of their functional equiv-
alence to ES cells have become widely accepted27. A variety of repro-
gramming methods have been developed to derive iPS cells, and each
has advantages and disadvantages (Table 1). Assessing reprogramming
begins with identifying compact colonies that have distinct borders
and well-defined edges, and are comprised of cells with a large nucleus,
large nucleoli and scant cytoplasm. A wide range of colony morpholo-
gies result from reprogramming, and although many colonies appear
morphologically similar to ES cells, only a subset of these have com-
parable molecular and functional features. To accurately distinguish
reprogrammed, bona fide iPS cells from those that are only partially
reprogrammed, investigators look for a series of molecular hallmarks.
Markers of pluripotency
Fully reprogrammed cells express a network of pluripotency genes,
including OCT4, SOX2 and NANOG, in levels comparable to ES cells,
and they reactivate telomerase gene expression, downregulate THY1
and upregulate SSEA1 (ref. 28). Positive staining for alkaline phos-
phatase activity has been widely used as a marker of pluripotency; how-
ever, recently published data have shown this to be insufficient as a test
for true iPS cells, because intermediately reprogrammed cells also stain
positively29. The same report shows that iPS cells that are generated by
virus-mediated reprogramming silence proviral genes when the endog-
enous pluripotency genes are activated, and that this event is paired with
the expression of the embryonic antigens SSEA3, TRA-1-60, TRA-1-81,
DNA methyltransferase 3β (DNMT3β) and REX1 (ref. 29). Genome-wide
epigenetic reprogramming is crucial for deriving fully reprogrammed
cells, and the degree of success is measured, in part, by evaluating the
methylation status at the promoters of the genes responsible for maintain-
ing pluripotency, as well as at the genes important for driving differentia-
tion30. A crucial event during epigenetic reprogramming is the reactiva-
tion of the silent X chromosome, which occurs late in reprogramming
and represents a hallmark of ground-state pluripotency28,30,31. If iPS cells
acquire all of these molecular features, they are expected to behave like ES
cells and to demonstrate reprogramming-factor independence, which is
marked by silencing of the proviral transgenes. Variations in epigenetic
reprogramming, the extent of methylation, the persistence of expression
of integrated proviruses and other known and unknown factors can alter
the differentiation potential of iPS cells. Because of the potential for het-
erogeneity, it is essential to know as much as possible about the nature of
a cell line before labelling it pluripotent.
Figure 1 | Morphology of pluripotent stem cell types. Mouse ES (a) and
iPS (b) cells form dome-shaped, refractile colonies. These colonies are in
contrast to the flat morphology of mouse epiblast-derived stem cells (f), which
resemble human ES (d) and iPS (e) cells. Human iPS cells induced into a naive
pluripotent state by treatment with chemical inhibitors97–100 (c) parallel the
morphology of mouse ES and iPS cells. Scale bars, 50 μm.
2 9 6 | N A T U R E | V O L 4 8 1 | 1 9 J A N U A R Y 2 0 1 2
© 2012 Macmillan Publishers Limited. All rights reserved
Functional assays of pluripotency
When iPS cell lines are isolated and documented to carry the molecular
features of fully reprogrammed cells, they are typically also assessed
in functional assays. Characterization of the functional abilities of iPS
cells begins with in vitro differentiation. The cells can be differenti-
ated as embryoid bodies — compact balls of loosely organized tissues
that resemble the gastrulating embryo — or through two-dimensional
directed differentiation in a culture dish. Such cultures can then be
assessed for markers of each of the three germ layers. Analysis of the
pluripotency of mouse cells typically entails the development of a chi-
maera, which evaluates the potential of iPS cells to contribute to the
normal development of adult tissues after injection into the blastocyst.
Whether germline transmission occurs after blastocyst chimaerism
is measured by the ability of chimaeras to produce all-iPS-cell mice
in their offspring. These offspring have the genomic integrity of the
injected iPS cell line, as well as the ability to form functional germ cells.
The highest stringency test for mouse iPS cells — tetraploid comple-
mentation — entails the injection of iPS cells into tetraploid blastocysts
to measure the ability of the iPS cells to direct the normal development
of an entire organism. This test has been accomplished for only a limited
subset of iPS cells32–34, although with an efficiency that parallels tetra-
ploid complementation carried out with ES or ntES cells33,35,36.
The current functional gold standard for human iPS cells involves the
evaluation of teratoma formation. In this assay, the in vivo differentiation
potential of human iPS cells is measured after their injection subcu-
taneously or intramuscularly into immunodeficient mice37,38. If the
cells are truly pluripotent, they will form well-differentiated tumours
comprised of elements from each of the three germ layers. This assay
provides information about the spontaneous differentiation potential
of the injected iPS cells. Although it is the most stringent assay avail-
able for human iPS cells, it is not powerful enough to assess whether
iPS cells can produce all the cell types of the human body, and it cannot
assess the contribution of iPS cells to the germ line. The caveat to all
these functional assays lies in the fact that the standards for iPS cells
are still hotly debated, especially when anticipating the use of iPS cells
for therapy39. Adopting a consistent set of standards that can be applied
uniformly worldwide is essential as stem-cell research and applications
Functional differences between iPS cells and ES cells
Despite the multitude of assays used to evaluate pluripotency, and
although many parallels have been found between iPS cells and ES
cells, there is a wide range of evidence showing that there are subtle
yet substantial differences between these cell types. Disparities were
first observed in the differentiation abilities of iPS cells and ES cells
in both teratoma-forming and in vitro differentiation assays. Some
mouse iPS cells showed lower efficiencies of teratoma formation than
mouse ES cells, whereas some human iPS cells showed less propensity
Table 1 | Methods for reprogramming somatic cells to iPS cells
Vector type Cell types Factors*Efficiency (%) AdvantagesDisadvantages
Fibroblasts, neural stem
cells, stomach cells,
liver cells, keratinocytes,
amniotic cells, blood cells
and adipose cells
OSK + VPA, or
OS + VPA
~0.001–1 Reasonably efficient Genomic integration,
incomplete proviral silencing
and slow kinetics
cluster + VPA
~0.1–1.1 Reasonably efficient and
transduces dividing and non-
Genomic integration and
incomplete proviral silencing
Fibroblasts, β cells,
keratinocytes, blood cells
~0.1–2Reasonably efficient and allows
controlled expression of factors
Genomic integration and
requirement for transactivator
FibroblastsOSKM ~0.1Reasonably efficient and no
Labour-intensive screening of
Fibroblasts OSK ~0.1–1Reasonably efficient and no
Labour-intensive screening of
excised lines, and loxP sites
retained in the genome
Fibroblasts and liver cellsOSKM~0.001 No genomic integrationLow efficiency
Fibroblasts OSNL ~0.001Only occasional genomic
Low efficiency and occasional
vector genomic integration
FibroblastsOSKM~1No genomic integrationSequence-sensitive RNA
replicase, and difficulty in
purging cells of replicating
FibroblastsOS ~0.001No genomic integration,
direct delivery of transcription
factors and no DNA-related
Low efficiency, short half-life,
and requirement for large
quantities of pure proteins and
multiple applications of protein
OSKML + VPA
~1–4.4No genomic integration,
antiviral response, faster
controllable and high efficiency
Requirement for multiple
rounds of transfection
Adipose stromal cells and
~0.1 Efficient, faster reprogramming
kinetics than commonly used
lentiviral or retroviral vectors,
no exogenous transcription
factors and no risk of
Lower efficiency than other
commonly used methods
*OSKM and similar factor names represent combinations of reprogramming factors: K, KLF4; L, LIN28; M, c-MYC; N, NANOG; O, OCT4; S, SOX2; and VPA, valproic acid.
1 9 J A N U A R Y 2 0 1 2 | V O L 4 8 1 | N A T U R E | 2 9 7
© 2012 Macmillan Publishers Limited. All rights reserved
to differentiate along haematopoietic, neuroepithelial and neuronal
lineages than human ES cells40–42. Some researchers interpreted these
findings to mean that iPS cells have an intrinsically lower differentia-
tion capacity than ES cells41, whereas other research groups have offered
different explanations, including that the cell of origin might have a
specific effect on the differentiation capacities of the derived iPS cells.
The results from cell-of-origin studies indicate that the parental cell
can influence the differentiation capacity of the resultant iPS cells. In one
study, mouse bone-marrow-derived and B-cell-derived iPS cells showed
more efficient differentiation along haematopoietic lineages than did
fibroblast-derived iPS cells or neural-progenitor-derived iPS cell lines.
Interestingly, treatment of the neural-progenitor-derived iPS cells with
trichostatin A, a potent histone-deacetylase inhibitor, plus 5-azacyti-
dine, a methylation-resistant cytosine analogue, increased the blood-
forming capacity of these cells, suggesting that their limitations were
due to epigenetic modifications. Whereas the bone-marrow-derived
and neural-progenitor-derived iPS cells contributed well to all tissues
in the chimaera assay, including to the germ line, the fibroblast-derived
iPS cells contributed only poorly43. This study laid some of the early
groundwork for later lines of investigation that probed the molecular
differences between iPS cells and ES cells, and provided an explanation
for the functional differences between these cells.
One investigation found that some iPS cells derived from human
retinal-pigment epithelial cells show an increased propensity to differ-
entiate back into this cell type than do ES cells or iPS cells derived from
other tissues44. More recently, Bar-Nur and colleagues showed that iPS
cells generated from human pancreatic islet β cells retain open chroma-
tin at the loci of key β-cell genes and that this correlates with a greater
capacity to differentiate into insulin-producing cells both in vivo and
in vitro than that of ES cells or isogenic non-β-cell-derived iPS cells45.
These functional differences extend beyond differentiation and potency
to disease modelling. For example, fragile X syndrome is caused by aber-
rant silencing of the FMR1 gene during human development; iPS cells
that were reprogrammed from adult skin fibroblasts from an individual
with fragile X syndrome failed to reactivate the FMR1 gene, whereas
ES cells derived from embryos with this syndrome, as diagnosed by
pre-implantation testing, expressed FMR1 (ref. 46). Consequently, the
potential for epigenetic memory in the fragile-X-syndrome-derived iPS
cells, and substantial differences between fragile-X-syndrome-derived
iPS and ES cells, must be considered when studying this condition and
potentially many other conditions. To determine whether the pluripo-
tent cells being used are appropriate to address a particular question or
to use in a given application, it is crucial to compare not only the in vivo
and in vitro differentiation potentials but also the genetic and epigenetic
disparities that underpin these functional differences.
iPS cells versus ES cells
Refined analyses, described in this section, have addressed whether
iPS cells are suitable alternatives to classic ES cells for use in research
Genetics and epigenetics
Global gene-expression analysis and bisulphite genomic sequencing
accompanying early derivations of iPS cells provided the initial evidence
for differences between iPS cells and ES cells at the epigenetic level14.
Further exploration of these differences led to the identification of only
a few, seemingly consistent, differences in global gene expression that
were more pronounced in earlier passages of iPS cells47. Many of the
differentially expressed genes were imprinted in ES cells48.
Looking beyond the expression patterns to the DNA sequence itself
has revealed genetic variation between iPS cells and ES cells. A recent
publication suggested that chromosomal aberrations are a common
feature of stem-cell populations that are propagated in vitro, with each
type of stem cell — whether ES cell, iPS cell or multipotent stem cell —
being prone to distinct abnormalities49. Both human iPS and ES cells
showed a tendency for gains at chromosomes 12 and 17. Whereas iPS
cells had additional gains at chromosomes 1 and 9, ES cells had additional
gains at chromosomes 3 and 20. Other work identified an accumula-
tion of point mutations in reprogrammed cells, particularly occurring
in oncogenic pathways45, whereas another study noted an increase in
copy number variants (CNVs) in early-passage human iPS cells relative
to intermediate-passage iPS cells or ES cells50. The number and size of
these CNVs were negatively correlated with the passage number in iPS
cells, suggesting that a selective disadvantage is conferred by these aber-
rations. A comparison of iPS cells and their parental cell of origin showed
that the majority of CNVs were created de novo in fragile regions of the
genome50. A comprehensive study by Laurent and colleagues found a
higher frequency of subchromosomal CNVs in pluripotent cell samples
than in non-pluripotent cell samples51. This work uncovered variation
between genomic regions enriched for CNVs in human ES cells and iPS
cells. A small subset of samples of ES cells harboured a large number
of duplications, whereas several iPS-cell samples contained moderate
numbers of deletions. Reprogramming was associated with deletions
in tumour-suppressor genes, whereas extended time in culture led to
duplications of oncogenes in human iPS cells.
The finding that human iPS cells derived from a variety of tissues have
residual, persistent donor-cell-specific gene-expression patterns sparks
the question of whether the current measure of a fully reprogrammed
cell is sufficient52 or whether iPS cells retain some type of ‘somatic mem-
ory’ from their past identity. To understand this observation better, a
more detailed analysis at the epigenetic level is required.
Reprogramming cells to a pluripotent state entails global epigenetic
remodelling and introduces epigenetic changes, some of which are
necessary for reprogramming to occur and others of which are inad-
vertently introduced during the process. A failure to demethylate pluri-
potency genes is associated with partial reprogramming in iPS cells.
Whole-genome profiling of the DNA methylomes of multiple human
iPS and ES cell lines, as well as somatic and progenitor cell lines, from
different laboratories using different reprogramming techniques and
with a variety of cells derived from distinct germ layers has shown that
although overall iPS-cell DNA methylomes closely resemble human
ES-cell DNA methylomes, iPS cells have significant variability in their
somatic memory, as well as aberrant iPS-cell-specific differential meth-
ylation. Some studies have suggested that this occurs in a passage-
dependent manner, but others have shown that differentially methylated
regions (DMRs) in iPS cells are transmitted to differentiated progeny at
a high frequency and cannot be erased through passaging29,53–55. Overall,
there are remarkable global similarities between the DNA methylomes
of generic iPS and ES cells; however, a core set of DMRs that seems to
represent hot spots of failed epigenetic reprogramming has been identi-
fied55. These DMRs are enriched for genes that are important for devel-
opmental processes51,55. The high incidence of unique DMRs in iPS cells
compared with progenitor somatic cells or ES cells suggests that these
patterns are stochastic and arise during reprogramming. In the most
exhaustive comparison so far, Kim and colleagues reported that more
DMRs were present in mouse iPS cells than in ntES or embryo-derived
ES cells43. However, these DMRs did not pertain to specific loci and thus
do not represent consistent differences between iPS cells and ES cells.
This lack of consistency suggests that aberrant DMRs in mouse iPS cells
reflect the technical limitations inherent in reprogramming, rather than
indicating loci that can reliably distinguish ES cells from iPS cells43.
In addition, the residual iPS-cell-specific methylation in many iPS-cell
isolates links these cells to their tissue of origin and, ultimately, affects
their differentiation propensity43,55. Residual signatures can be distinct
enough to enable the myeloid and lymphoid origins of blood-derived
iPS cells to be discerned43. In iPS cells derived from non-haematopoietic
cells, such as fibroblasts and neural progenitors, there can be residual
repressive methylation at loci that are required for haematopoietic fates,
reducing the blood-forming potential in vitro43. Exogenous supplementa-
tion of neural-progenitor-derived iPS cells with the cytokine WNT3A
can increase the blood-forming potential of these cells, supporting the
idea that incomplete reprogramming owing to epigenetic marks can be
2 9 8 | N A T U R E | V O L 4 8 1 | 1 9 J A N U A R Y 2 0 1 2
© 2012 Macmillan Publishers Limited. All rights reserved
overcome by manipulating the culture conditions. Treatment of cultures
with demethylating agents or knockdown of DNMT1 expression has
been shown to convert intermediately reprogrammed cells into fully
pluripotent cells, further supporting this idea48. When iPS cells are
forced to differentiate along a particular lineage, they become more
amenable to generating cells of that lineage after another round of repro-
gramming43. This finding shows that the differentiation propensity and
DNA methylation profile can be reset, and it suggests that the ‘epigenetic
memory’ of the donor cell can be exploited, especially in cases in which
directed differentiation is particularly challenging43.
Another important feature of epigenetic reprogramming is the
reactivation of the inactivated X chromosome. During normal devel-
opment in eutherian mammals (those with a placenta), one X chro-
mosome is randomly inactivated in each cell in females. Whether this
epigenetic silencing event is reset in iPS cells remains an area of contro-
versy, in part because of the poor fidelity of X-inactivation markers in
pluripotent cells56. Some studies have shown that the majority of female
human iPS clones retain an inactivated X chromosome (which is tran-
scriptionally silent)57, whereas others have indicated that some human
iPS clones lose immunostaining for trimethylated H3K27 on the X
chromosome (a marker of epigenetic silencing), indicating X reacti-
vation58. In addition, some of the earliest studies of iPS cells showed
X reactivation in reprogrammed female mouse fibroblasts31,58. How-
ever, recently published data support the finding that X reactivation
does not occur in human iPS cells and, interestingly, reprogramming
was found to favour expression of a particular X chromosome when
induced from a mixed X-inactivated population of fibroblasts59. Finally,
epigenetic reprogramming sometimes fails to properly restore bivalent
domains, which mark developmental loci with active and repressive
Although many of the studies cited here have generated data
suggesting that there are epigenetic differences between iPS cells
and ES cells, there are several limitations on extending these data to
all iPS cells in a more general (and more useful) sense. The published
comparisons were often made using iPS cells derived from a multi-
tude of laboratories by a variety of methodologies, and reanalysis of
the gene-expression microarray data using an unsupervised clustering
algorithm shows a strong correlation between transcriptional signatures
and specific laboratories for both iPS cells and ES cells. This finding
indicates that specific culture protocols and laboratory environments
can affect the transcriptional profile of iPS and ES cells. Therefore, the
data produced in a particular laboratory might be specific to the cells
In addition, most iPS colonies are clones derived from a single
reprogrammed cell, whereas ES cells used for analysis are typically non-
clonal. The subcloning of ES cells has revealed genetic and epigenetic
anomalies that would probably have otherwise gone undetected in the
heterogeneous ES-cell population62. With regard to somatic memory,
there is poor overlap between the gene sets that have been reported to
be characteristic of a particular cell type of origin, suggesting that the
retention of somatic memory is stochastic and is a reflection of the tech-
nical failure of reprogramming to fully erase the somatic epigenome.
To exacerbate the issue, the iPS cells used for comparison often have
different genetic backgrounds and have frequently been derived from
fibroblasts that were already heterogeneous in their make-up, affecting
both the gene-expression patterns and the functionality of the iPS cells.
Throughout the literature, many publications lack correlation
between the gene-expression patterns and the epigenetic patterns
observed. An additional consideration is the presence of different
viral insertions in individual iPS cell lines, which can also affect the
functionality of the derived cells1. Evidence to support this idea is
provided by the reduced number of differences observed among iPS
cells and between iPS and ES cells when transgenes are removed63.
Many of the aforementioned studies have focused on differences in
either transcriptional profiles or changes in epigenetic marks; how-
ever, the most recent studies have evaluated iPS cells and ES cells from
both of these angles in parallel, together with their in vitro differentia-
tion potential, generating the most comprehensive and compelling
data that have been published so far.
Stadtfeld and colleagues explored the epigenetic and functional
discrepancies between iPS cells and ES cells using a new reprogram-
ming strategy that allowed direct comparison of genetically matched
cells derived from the same source32. These authors derived iPS and ES
cells from mice carrying an integrated doxycycline-inducible reprogram-
ming cassette in every cell, a strategy that sidesteps the confounding
effects of variable genetic backgrounds and viral integration that have
been observed in other studies. The overall messenger RNA and micro-
RNA expression patterns of iPS cells and ES cells were indistinguishable
except for the aberrant silencing of a few transcripts localized to the
imprinted Dlk1–Dio3 gene cluster on chromosome 12qF1, a region that
is important for development. A failure to reactivate this locus meant that
iPS cells contributed poorly to chimaeras and were unable to generate
all-iPS-cell mice. By contrast, iPS cells with normal Dlk1–Dio3 expres-
sion contributed to high-grade chimaeras and supported the develop-
ment of viable all-iPS-cell mice. The treatment of iPS cells that failed to
reactivate Dlk1–Dio3 with a histone-deacetylase inhibitor rescued the
ability of these clones to support the development of all-iPS-cell mice
by relieving this region of aberrant hypermethylation. However, recent
data from iPS and ES cells derived from a mouse strain carrying a dis-
tinct reprogramming cassette suggest that different levels of expression
of reprogramming factors, rather than aberrant silencing of Dlk1–Dio3,
account for the different behaviour of the cell lines64. The disparate results
from these studies highlight that iPS cells can behave differently based
on subtle variations in the expression of only a few loci.
To systematically compare human iPS cells derived from different
somatic cell types and ES cells, Ohi and colleagues compared ES cells
with iPS cells reprogrammed from somatic cells representative of the
three embryonic germ layers65. Transcriptional and epigenetic profil-
ing of these cells showed transcriptional differences, owing, in part, to
incomplete promoter methylation, which enabled iPS cells to be dis-
cerned on the basis of their cell of origin. The differential methylation
between iPS cells and ES cells did not correlate with varying levels in
DNA methyltransferases; however, the authors found a nonrandom pat-
tern of incompletely silenced genes in genetic regions that are isolated
from other genes that undergo silencing during reprogramming. This
finding could be explained by inefficient or delayed recruitment of the
silencing machinery and DNA methyltransferases to particular somatic
genes because of the isolated location of these genes65.
In another comprehensive study, Polo and colleagues evaluated the
effect of cellular origin on the gene-expression pattern, epigenetic prop-
erties and functional abilities of genetically matched mouse iPS cells66.
Using the same ‘secondary’ reprogramming strategy used by Wernig
and colleagues53, whereby reprogramming is assessed using tissues
from a mouse generated from iPS cells carrying integrated, doxycy-
cline-inducible reprogramming factors, Polo and colleagues generated
iPS cells from tail-tip fibroblasts, splenic B cells, bone-marrow-derived
granulocytes and skeletal muscle precursors, and showed that each iPS
cell line retained a transcriptional memory of its cell of origin. This
memory was evident in that markers for each respective cell of origin
remained actively expressed, and was supported by the finding that
iPS cell lines derived from the same cell of origin clustered together
on the basis of global transcriptional data. A similar correlation was
found on evaluation of methylation patterns, which showed subtle but
substantial differences, reflecting the consequences of different histone
marks. The effects of somatic memory extended beyond the genetic
and epigenetic levels to functional significance, affecting the autono-
mous differentiation potential of the different iPS cell lines after embry-
oid-body formation. A clear bias that reflected the cells of origin was
observed in the iPS cell lines. Notably, the transcriptional, epigenetic
and functional effects evaluated in early-passage iPS cell lines became
1 9 J A N U A R Y 2 0 1 2 | V O L 4 8 1 | N A T U R E | 2 9 9
© 2012 Macmillan Publishers Limited. All rights reserved
less significant with continued passaging. This finding indicated that
complete reprogramming is a gradual process that extends beyond the
time frame necessary to observe the activation of endogenous pluripo-
tency genes, transgene-free growth and differentiation into cell types
from each of the three germ layers.
Having considered data from a multitude of published studies, generated
by the painstaking efforts of many research groups, we return to our
earlier question: are iPS cells equivalent to ES cells? The answer is not
straightforward. Rather, there is an emerging consensus that iPS cells
and ES cells are neither identical nor distinct populations. Instead, they
are overlapping, with greater variability inherent within each population
than between the populations. The heterogeneity and behaviour of each
class of cells is more complex than has previously been thought. The
two pluripotent stem-cell types are, in theory, functionally equivalent;
however, in practice, they harbour genetic and epigenetic differences
that reflect their different histories. It remains to be seen whether there
are any consistent molecular distinctions between iPS cells and ES cells.
It is also important to consider that, in contrast to long-standing
belief, ES cells themselves have considerable epigenetic heterogeneity
and have differing propensities for differentiation — much like those
found in iPS cells67,68. These observations, paired with those discussed
in this Review, are a call for researchers to take a step back from the
direct comparison of iPS cells and ES cells, and they highlight the need
to redefine what it means to belong to either of these cell classes. Some
researchers have already taken heed of this message and generated a
bioinformatics assay for pluripotency63, whereas others have pro-
duced a ‘scorecard’ to evaluate the character of both iPS and ES cell
lines and predict the quality and utility of any pluripotent cell line in a
high-throughput manner47,69. Using DNA methylation mapping, gene-
expression profiling and a quantitative differentiation assay, Bock and
colleagues made a systematic comparison of 20 established ES cell lines
and 12 iPS cell lines47. They confirmed that, despite overall similarities,
transcriptional and epigenetic variation is common between iPS cell
lines, between ES cell lines, and between iPS cell and ES cell populations.
These data provide a reference for the variation among human pluripo-
tent cell lines, which assists in predicting the functional consequences
of these differences. We can conclude from these studies that any given
iPS cell line generated by today’s technology might not be completely
equivalent to the ideal ES cell.
The differences between iPS cells and ES cells, as well as those among
iPS cells, clearly affect the utility of these cells in research, disease model-
ling and therapeutics, providing an impetus for investigators to evaluate
their cell populations carefully and precisely. The differences do not
diminish the potential of iPS cells, given that iPS cells have considerable
advantages over ES cells. Rather than replacing ES cells with iPS cells,
it is becoming clear that these two cell types complement one another.
Researchers are still in the process of developing the necessary protocols
to harness the potential of iPS cells; however, as it becomes clear how
to evaluate the genetic, epigenetic and functional status of different iPS
cell lines, further applications of these cells will be uncovered, and pro-
gress will be made in creating iPS cell lines and designing protocols to
accomplish the ambitious goals of the field.
Medical applications of iPS cells
Generating patient-specific stem cells has been a long-standing goal
in the field of regenerative medicine. Despite considerable challenges,
generating disease-specific and patient-specific iPS cells through
reprogramming has become almost routine. These cells provide a
unique platform from which to gain mechanistic insight into a variety
of diseases, to carry out in vitro drug screening, to evaluate potential
therapeutics and to explore gene repair coupled with cell-replace-
ment therapy (Fig. 2). In the past few years, the number of reports
on applications of iPS cells has steadily increased, testifying to the
broad influence of this breakthrough technology (Table 2). Despite
the continued presence of substantial hurdles, the pace of this work
is such that no review can capture the current state of the field; thus,
we point to a few publications that highlight the promising medical
applications of iPS cells but also indicate their key limitations.
In 2009, Lee and colleagues harnessed iPS cells to demonstrate disease
modelling and drug screening for familial dysautonomia, a rare genetic
disorder of the peripheral nervous system70. In almost all cases, familial
dysautonomia is caused by a single point mutation in the gene encoding
the inhibitor of nuclear factor-κB (IκB)-kinase-complex-associated pro-
tein (IKBKAP) that manifests as an extensive autonomic nervous system
deficit and dysfunction in small-fibre sensory neurons. Although many
traditional cell-based models have been used to study the pathogenesis
of familial dysautonomia and to screen for candidate drugs, none has
used symptom-relevant human cell types. With the successful derivation
of iPS cells from patients with familial dysautonomia, investigators pro-
duced central and peripheral nervous system precursors and subsequently
found three disease-related phenotypes, thus providing validation that
disease-relevant cell types could accurately reflect disease pathogenesis in
vitro70. After screening with multiple compounds, the authors showed that
the disease phenotype could be partially normalized by kinetin, a plant
hormone. This initial report demonstrated how iPS cells can facilitate
the discovery of therapeutic compounds and described how these cells
provided a platform for modelling different severities of familial dysau-
tonomia and for generating predictive tests to determine differences in
the clinical manifestation of the disorder.
Such applications of iPS cells in drug screening and discovery are
destined to expand to encompass numerous disease conditions. Sev-
eral research groups have generated models of long QT syndrome,
a congenital disease with 12 types, each of which is associated with
abnormal ion-channel function, a prolonged QT interval on an
electrocardiogram and a high risk of sudden cardiac death due to
ventricular tachyarrhythmia. Much work has been carried out in ani-
mal models to probe the underlying mechanisms of this syndrome,
but cardiomyocytes have distinct and complex electrophysiological
properties that differ between species. In addition, the lack of in vitro
sources of human cardiomyocytes and the inability to model patient-
specific variations of this disease has impeded studies.
In a proof-of-principle study for using iPS cells to capture the
physiological mechanisms of genetic variation, Moretti and colleagues
differentiated iPS cells from individuals with type 1 long QT syndrome
into cardiomyocytes and, as predicted, observed prolonged action poten-
tials in the ventricular and atrial cells71. Using this model system, these
investigators uncovered a dominant-negative trafficking defect associ-
ated with the particular mutation that causes this variant of long QT
syndrome. Further investigation of long QT syndrome iPS-cell-derived
cardiomyocytes showed that these cells had an increased susceptibil-
ity to catecholamine-induced tachyarrhythmia, and compounds that
exacerbated the condition (including isoprenaline) were identified71.
Treatment of these cardiomyocytes with β-adrenergic receptor blockers
attenuated the long QT phenotype.
Type 2 long QT syndrome has also been modelled in cardiomyocytes,
by Itzhaki and colleagues72. The authors derived type 2 long QT syndrome
iPS cells to evaluate the potency of existing and new pharmacological
agents that might exacerbate or ameliorate the condition. Their studies
show that the long QT syndrome phenotype was aggravated by blockers
of ERG-type potassium channels, whereas nifedipine, a calcium-channel
blocker, and pinacidil, an agonist of ATP-sensitive potassium channels,
both ameliorated the long QT syndrome phenotype, as shown by the
decreased duration of action potentials in long QT syndrome cardio-
myocytes, as well as the elimination of early after-depolarizations and the
abolishment of all triggered arrhythmias. A possible limitation of these
beneficial drugs is excessive shortening of the action-potential duration,
leading to short QT syndrome.
Importantly, these studies established that the iPS-cell model can be
used to identify complex cardiotoxic effects of drugs, as well as to define
protective pharmacological agents, including optimal drug dosages.
3 0 0 | N A T U R E | V O L 4 8 1 | 1 9 J A N U A R Y 2 0 1 2
© 2012 Macmillan Publishers Limited. All rights reserved
Given the number of drugs that have notoriously been withdrawn from
the market because of their tendency to induce arrhythmias, it is highly
likely that the current inadequate approaches for assessing cardiotoxic-
ity will be complemented by iPS-cell-based assessments of drug effects.
A study from our laboratory explored dyskeratosis congenita, a dis-
order of telomere maintenance, and provided an unanticipated insight
into the basic biology of telomerase that has therapeutic implications73.
In its most severe form, dyskeratosis congenita is caused by a mutation
in the dyskerin gene (DKC1), which is X linked, leading to shortened tel-
omeres and premature senescence in cells and ultimately manifesting as
the degeneration of multiple tissues. Because the reprogramming of cells
to an induced pluripotent state is accompanied by the induction of the
gene encoding telomerase reverse transcriptase (TERT), we investigated
whether the telomerase defect would limit the derivation and mainte-
nance of iPS cells from individuals with dyskeratosis congenita. Although
the efficiency of iPS-cell derivation was poor, we were able to successfully
reprogram patient fibroblasts. Surprisingly, whereas the mean telomere
length immediately after reprogramming was shorter than that of the
parental fibroblast population, continued passage of some iPS cell lines
led to telomere elongation over time. This process was accompanied by
upregulation of the expression of TERC, which encodes the RNA subunit
Further analysis established that TERT and TERC, as well as DKC1,
were expressed at higher levels in dyskeratosis-congenita-derived iPS cells
than in the parental fibroblasts73. We determined that the genes encoding
these components of the telomerase pathway — including a cis element
in the 3ʹ region of the TERC locus that is essential for a transcriptionally
active chromatin structure — were direct binding targets of the pluri-
potency-associated transcription factors. Further analysis indicated that
transcriptional silencing owing to a 3ʹ deletion in the TERC locus leads to
the autosomal dominant form of dyskeratosis congenita by diminishing
TERC transcription. Although telomere length is restored in dyskeratosis-
congenita-derived iPS cells, differentiation into somatic cells is accompa-
nied by a return to pathogenesis with low TERC expression and a decay in
telomere length. This finding showed that TERC RNA levels are dynami-
cally regulated and that the pluripotent state of the cells is reversible, sug-
gesting that drugs that elevate or stabilize TERC expression might rescue
defective telomerase activity and provide a therapeutic benefit. Although
we set out to understand the pathogenesis of dyskeratosis congenita with
this study, we showed that a high expression level of multiple telomerase
components was characteristic of the pluripotent state more generally,
illustrating how iPS cells can reveal fundamental aspects of cell biology.
An independent study of the reprogramming of cells from patients with
dyskeratosis congenita confirmed the general transcriptional upregula-
tion of multiple telomerase components and the maintenance of telomere
lengths in clones74; however, in this study, no clones with elongated telom-
eres were identified. The different outcomes of these studies highlight the
limitations of iPS-cell-based disease models that are imposed by clonal
variation as a result of the inherent technical infidelity of reprogram-
ming75. This point also introduces an additional important consideration.
Before a given iPS-cell disease model can be claimed to be truly represent-
ative of the disease, how many patients must be involved, and how many
iPS cell lines must be derived from each patient? Although the answers to
these questions are unclear, it is crucial to keep these issues in mind when
generating disease models and making claims based on these models.
Although iPS cells are an invaluable tool for modelling diseases in vitro,
the goal of developing patient-specific stem cells has also been motivated
by the prospect of generating a ready supply of immune-compatible cells
and tissues for autologous transplantation. At present, the clinical trans-
lation of iPS-cell-based cell therapies seems more futuristic than the in
vitro use of iPS cells for research and drug development, but two ground-
breaking studies have provided the proof of principle in mouse models
that the dream might one day be realized. Hanna, Jaenisch and colleagues
used homologous recombination to repair the genetic defect in iPS cells
derived from a humanized mouse model of sickle-cell anaemia76. Directed
differentiation of the repaired iPS cells into haematopoietic progenitors
followed by transplantation of these cells into the affected mice led to
the rescue of the disease phenotype. The gene-corrected iPS-cell-derived
haematopoietic progenitors showed stable engraftment and correction of
the disease phenotype.
In another landmark study from Jaenisch’s research group, Wernig
and colleagues derived dopaminergic neurons from iPS cells that, when
implanted into the brain, became functionally integrated and improved
the condition of a rat model of Parkinson’s disease77. The successful
implantation and functional recovery in this model is evidence of the
therapeutic value of pluripotent stem cells for cell-replacement therapy
in the brain — one of the most promising areas for the future of iPS-
Figure 2 | Medical applications of iPS cells.
Reprogramming technology and iPS cells have
the potential to be used to model and treat
human disease. In this example, the patient has
a neurodegenerative disorder. Patient-specific
iPS cells — in this case derived by ectopic
co-expression of transcription factors in cells
isolated from a skin biopsy — can be used in one
of two pathways. In cases in which the disease-
causing mutation is known (for example, familial
Parkinson’s disease), gene targeting could be
used to repair the DNA sequence (right). The
gene-corrected patient-specific iPS cells would
then undergo directed differentiation into the
affected neuronal subtype (for example, midbrain
dopaminergic neurons) and be transplanted into
the patient’s brain (to engraft the nigrostriatal
axis). Alternatively, directed differentiation of
the patient-specific iPS cells into the affected
neuronal subtype (left) will allow the patient’s
disease to be modelled in vitro, and potential drugs
can be screened, aiding in the discovery of novel
Patient-specifc iPS cells
Afected cell type
Transplantation of genetically
matched healthy cells
Repaired iPS cells
Use gene targeting to repair
1 9 J A N U A R Y 2 0 1 2 | V O L 4 8 1 | N A T U R E | 3 0 1
© 2012 Macmillan Publishers Limited. All rights reserved
Together, these findings provide proof of principle for using
reprogramming with gene repair and cell-replacement therapy for
treating diseases. Using iPS cells in cell-replacement therapy offers
the promise of therapeutic intervention that is not compounded
by the use of immunosuppressive drugs to prevent tissue rejection,
while harnessing targeted gene-repair strategies, such as homolo-
gous recombination and zinc-finger nucleases, to repair genetic
defects. These strategies provide the opportunity for generating an
unlimited population of stem cells that can be differentiated into
the desired cell type for studying disease mechanisms, for screening
and developing drugs or for developing a suitable cell-replacement
therapy. There have been considerable advances and successes to this
end; however, selecting an appropriate disease target, directing the
differentiation of iPS cells into phenotype-relevant cell populations
and identifying disease-relevant phenotypes remain major hurdles.
It is unclear whether iPS cells used for cell-replacement therapy
would completely evade an immune response when returned to the
patient, because a recent study has shown the immune rejection of
teratomas formed from iPS cells, even in syngeneic mice78. Neverthe-
less, iPS cells provide a promising model with which to study disease
mechanisms, discover new therapies and develop truly personalized
Predictions for the evolution of the art
Few fields have enjoyed the remarkable upsurge in activity and
excitement that followed the initial report of the reprogramming of
somatic cells into iPS cells in 2006. Despite heady progress, crucial
challenges must be met for the field to realize its full potential. There
is as yet no consensus on the most consistent or optimal protocol
for deriving the most reliable and, ultimately, the safest iPS cells.
Increasing the reprogramming efficiency and effecting reprogram-
ming without genetically modifying the cells are goals that have been
achieved. Using more-uniform protocols and more-rigorous controls
would facilitate experimental and potentially therapeutic consistency
Table 2 | Diseases modelled with iPS cells
DiseaseMolecular defect of donor cell Cell type differentiated from iPS cellsDisease phenocopied
in differentiated cells
Amyotrophic lateral sclerosis (ALS)Heterozygous Leu144Phe mutation
Mutations in SMN1
Motor neurons and glial cells NDNo
Spinal muscular atrophy (SMA)Neurons and astrocytes, and mature motor
Parkinson’s disease Multifactorial; mutations in LRRK2
72 CAG repeats in the huntingtin gene
Huntington’s diseaseNone NANo
Down’s syndromeTrisomy 21 Teratoma with tissue from each of the three
Fragile X syndrome CGG triplet repeat expansion resulting
in the silencing of FMR1
Mutation in IKBKAP
Familial dysautonomia Central nervous-system lineage, peripheral
neurons, haematopoietic cells, endothelial cells
and endodermal cells
Rett’s syndrome Heterozygous mutation in MECP2 Neural progenitor cells YesYes
Mucopolysaccharidosis type IIIB
Homozygous mutation in NAGLUNeural stem cells and differentiated neuronsPartially Yes
cerebral ALD (CCALD) and
Mutation in ABCD1Oligodendrocytes and neuronsPartially Yes
ADA SCIDMutation or deletion in ADANone ND No
Fanconi’s anaemia FAA and FAD2 correctedHaematopoietic cells No (corrected)No
Homozygous HbS mutationNoneNANo
β-ThalassaemiaHomozygous deletion in the β-globin
Heterozygous Val617Phe mutation
Heterozygous mutation in JAK2
Haematopoietic cells ND No
Polycythaemia veraHaematopoietic progenitors (CD34+CD35+)Partially No
Lesch–Nyhan syndrome (carrier) Heterozygous mutation in HPRT1NoneNA No
Type 1 diabetes Multifactorial; unknown
β-Cell-like cells (express somatostatin,
glucagon and insulin; glucose-responsive)
Gaucher’s disease, type IIIMutation in GBANone NA No
α1-Antitrypsin deficiency (A1ATD) Homozygous mutation in the
Hepatocyte-like cells (fetal) YesNo
3 0 2 | N A T U R E | V O L 4 8 1 | 1 9 J A N U A R Y 2 0 1 2
© 2012 Macmillan Publishers Limited. All rights reserved
between laboratories and would yield standardized cell lines that
could be used with confidence in both basic and applied studies.
Barring that, researchers must agree on standards of molecular
analysis to ensure that the reprogrammed cells that most closely
approximate the generic state of the naive genome can be identified.
Because iPS cells are subject to the same type of culture adaptations
that affect karyotypic integrity as human ES cells79, it is important
to define protocols that minimize the time in culture. In addition,
cell lines used in clinical applications will need to be evaluated
frequently for aberrant culture-induced changes at all stages: from
the somatic cells to the reprogrammed and differentiated cells80.
Understanding the genomic alterations that take place during the
reprogramming, culture and differentiation of iPS cells will be cru-
cial for designing experiments and ensuring that the derived cells
are functional, pure and appropriate for use in research and therapy.
Minimizing any aberrations is important, but as long as research-
ers understand that aberrations will arise — and can describe and
control their effects — even imperfect cells can be used, and prefer-
ential differentiation can be taken advantage of whenever possible.
Characteristics of iPS cells that were initially perceived as flaws,
including varying differentiation propensities, might prove useful
in clinical settings to generate cell types that have been difficult to
obtain thus far.
Generating more stringent markers of pluripotency and assays
to distinguish the abilities of a given iPS cell line are key priorities.
Building on the progress that has already been made using ES cells81,
researchers must continue to improve the understanding of directed
differentiation and to develop new protocols. With refined differen-
tiation protocols, researchers will be able to investigate the patho-
physiological basis of genetic diseases and carry out drug screening
on affected cell types. These protocols will bring the field a step
closer to patient-matched cells and tissues for clinical transplanta-
tion, a long-standing ambition of the stem-cell field that might be
its ultimate measure of success. ■
DiseaseMolecular defect of donor cellCell type differentiated from iPS cells Disease phenocopied
in differentiated cells
Glycogen storage disease Ia
Defect in glucose-6-phosphate gene Hepatocyte-like cells (fetal) YesNo
Autosomal dominant mutation in LDLR Hepatocyte-like cells (fetal) YesNo
Crigler–Najjar syndromeDeletion in UGT1A1Hepatocyte-like cells (fetal) ND No
Hereditary tyrosinaemia, type 1Mutation in FAHD1 Hepatocyte-like cells (fetal)NDNo
Pompe diseaseKnockout of GAASkeletal muscle cells Yes No
Progressive familial cholestasisMultifactorial Hepatocyte-like cells (fetal)ND No
Hurler syndrome (MPS IH) Genetic defect in IDUA Haematopoietic cellsNoNo
LEOPARD syndrome Heterozygous mutation in PTPN11 CardiomyocytesYes No
Type 1 long QT syndromeDominant mutation in KCNQ1 Cardiomyocytes YesNo
Type 2 long QT syndrome Missense mutation in KCNH2 Cardiomyocytes Yes Yes
SCID or leaky SCID Mutation in RAG1None NANo
Omenn syndrome (OS) Mutation in RAG1 None NA No
Cartilage-hair hypoplasia (CHH) Mutation in RMRP None NA No
Herpes simplex encephalitis (HSE)
Mutation in STAT1 or TLR3 Mature cell types of the central nervous system No No
Duchenne muscular dystrophyDeletion in the dystrophin gene NoneNA No
Becker muscular dystrophyUnidentified mutation in dystrophinNone NA No
Dyskeratosis congenita (DC) Deletion in DKC1 None NANo
Cystic fibrosis Homozygous deletion in CFTRNone NA No
Friedreich’s ataxia (FRDA) Trinucleotide GAA repeat expansion
Heterogeneity in causative genes and
mutations: mutations in RP9, RP1,
PRPH2 or RHO
Mutation in COL7A1
Sensory and peripheral neurons, and
Retinal progenitors, photoreceptor precursors,
retinal-pigment epithelial cells and rod
Haematopoietic cells, and epidermis-like
keratinocytes that differentiate into cells of all
three germ layers in vivo
Retinitis pigmentosa Yes Yes
epidermolysis bullosa (RDEB)
Scleroderma Unknown NANo
Osteogenesis imperfecta Mutation in COL1A2None NANo
An extended version of this table includes references and more information about drug and functional tests (Supplementary Table 1). ABCD1, ATP-binding cassette, sub-family D, member 1; ADA,
adenine deaminase; CFTR, cystic fibrosis transmembrane conductance regulator; COL1A2, α2-chain of type I collagen; COL7A1, α1-chain of type VII collagen; DKC1, dyskerin; FAA, Fanconi’s anaemia,
complementation group A; FAD2, Fanconi’s anaemia, complementation group D2; FAHD1, fumarylacetoacetate hydrolase; FMR1, fragile X mental retardation 1; FXN, frataxin; GAA, acid α-glucosidase;
GBA, acid β-glucosidase; HbS, sickle haemoglobin; HPRT1, hypoxanthine phosphoribosyltransferase 1; IDUA, α-L-iduronidase; JAK2, Janus kinase 2; KCNH2, potassium voltage-gated channel, subfamily H
(eag-related), member 2; KCNQ1, potassium voltage-gated channel, KQT-like subfamily, member 1; LDLR, low-density lipoprotein receptor; LRRK2, leucine-rich repeat kinase 2; MECP2, methyl CpG binding
protein 2; NA, not applicable; NAGLU, α-N-acetylglucosaminidase; ND not determined; PRPH2, peripherin 2; PTPN11, protein tyrosine phosphatase, non-receptor type 11; RAG1, recombination activating
gene 1; RHO, rhodopsin; RMRP, RNA component of mitochondrial-RNA-processing endoribonuclease; RP, retinitis pigmentosa; SCID, severe combined immunodeficiency; SMN1, survival of motor neuron 1;
SNCA, α-synuclein; SOD1, superoxide dismutase 1; STAT1, signal transducer and activator of transcription 1; TLR3, Toll-like receptor 3; UGT1A1, UDP glucuronosyltransferase 1 family, polypeptide A1.
1 9 J A N U A R Y 2 0 1 2 | V O L 4 8 1 | N A T U R E | 3 0 3
© 2012 Macmillan Publishers Limited. All rights reserved
1. Takahashi, K. & Yamanaka, S. Induction of pluripotent stem cells from mouse
embryonic and adult fibroblast cultures by defined factors. Cell 126, 663–676
This breakthrough paper describes the derivation of iPS cells directly from
mouse somatic cells through the ectopic co-expression of reprogramming
transcription factors, providing an alternative source of pluripotent cells for
Dreisch, H. Entwicklungsmechanische Studien I. Der Wert der ersten beiden
Furchungszellen in der Echinodermenentwickelung. Experimentelle Erzeugung
von Teil und Doppelbildungen. Z. Wiss. Zool. 53, 160–183 (1891).
Dewey, M. J., Martin, D. W. Jr, Martin, G. R. & Mintz, B. Mosaic mice
with teratocarcinoma-derived mutant cells deficient in hypoxanthine
phosphoribosyltransferase. Proc. Natl Acad. Sci. USA 74, 5564–5568 (1977).
Gardner, R. L. Mouse chimeras obtained by the injection of cells into the
blastocyst. Nature 220, 596–597 (1968).
Brinster, R. L. The effect of cells transferred into the mouse blastocyst on
subsequent development. J. Exp. Med. 140, 1049–1056 (1974).
Evans, M. J. & Kaufman, M. H. Establishment in culture of pluripotential cells
from mouse embryos. Nature 292, 154–156 (1981).
Martin, G. R. Isolation of a pluripotent cell line from early mouse embryos
cultured in medium conditioned by teratocarcinoma stem cells. Proc. Natl Acad.
Sci. USA 78, 7634–7638 (1981).
Thomson, J. A. et al. Embryonic stem cell lines derived from human blastocysts.
Science 282, 1145–1147 (1998).
Vogel, G. & Holden, C. Ethics questions add to concerns about NIH lines.
Science 321, 756–757 (2008).
10. Mosher, J. T. et al. Lack of population diversity in commonly used human
embryonic stem-cell lines. N. Engl. J. Med. 362, 183–185 (2010).
11. Tabar, V. et al. Therapeutic cloning in individual parkinsonian mice. Nature Med.
14, 379–381 (2008).
12. Noggle, S. et al. Human oocytes reprogram somatic cells to a pluripotent state.
Nature 478, 70–75 (2011).
13. Park, I. H. et al. Reprogramming of human somatic cells to pluripotency with
defined factors. Nature 451, 141–146 (2008).
14. Takahashi, K. et al. Induction of pluripotent stem cells from adult human
fibroblasts by defined factors. Cell 131, 861–872 (2007).
15. Yu, J. et al. Induced pluripotent stem cell lines derived from human somatic
cells. Science 318, 1917–1920 (2007).
16. Stadtfeld, M., Brennand, K. & Hochedlinger, K. Reprogramming of pancreatic
β cells into induced pluripotent stem cells. Curr. Biol. 18, 890–894 (2008).
17. Eminli, S., Utikal, J., Arnold, K., Jaenisch, R. & Hochedlinger, K. Reprogramming
of neural progenitor cells into induced pluripotent stem cells in the absence of
exogenous Sox2 expression. Stem Cells 26, 2467–2474 (2008).
18. Kim, J. B. et al. Pluripotent stem cells induced from adult neural stem cells by
reprogramming with two factors. Nature 454, 646–650 (2008).
19. Hanna, J. et al. Direct reprogramming of terminally differentiated mature
B lymphocytes to pluripotency. Cell 133, 250–264 (2008).
20. Aoi, T. et al. Generation of pluripotent stem cells from adult mouse liver and
stomach cells. Science 321, 699–702 (2008).
21. Utikal, J., Maherali, N., Kulalert, W. & Hochedlinger, K. Sox2 is dispensable
for the reprogramming of melanocytes and melanoma cells into induced
pluripotent stem cells. J. Cell Sci. 122, 3502–3510 (2009).
22. Sun, N. et al. Feeder-free derivation of induced pluripotent stem cells from
adult human adipose stem cells. Proc. Natl Acad. Sci. USA 106, 15720–15725
23. Maherali, N. et al. A high-efficiency system for the generation and study of
human induced pluripotent stem cells. Cell Stem Cell 3, 340–345 (2008).
24. Tesar, P. J. et al. New cell lines from mouse epiblast share defining features with
human embryonic stem cells. Nature 448, 196–199 (2007).
25. Wray, J., Kalkan, T. & Smith, A. G. The ground state of pluripotency. Biochem.
Soc. Trans. 38, 1027–1032 (2010).
26. Hanna, J. H., Saha, K. & Jaenisch, R. Pluripotency and cellular reprogramming:
facts, hypotheses, unresolved issues. Cell 143, 508–525 (2010).
27. Maherali, N. & Hochedlinger, K. Guidelines and techniques for the generation of
induced pluripotent stem cells. Cell Stem Cell 3, 595–605 (2008).
28. Stadtfeld, M., Maherali, N., Breault, D. T. & Hochedlinger, K. Defining molecular
cornerstones during fibroblast to iPS cell reprogramming in mouse. Cell Stem
Cell 2, 230–240 (2008).
This study enumerated the molecular markers and functional criteria for
defining reprogrammed cell populations.
29. Chan, E. M. et al. Live cell imaging distinguishes bona fide human iPS cells from
partially reprogrammed cells. Nature Biotechnol. 27, 1033–1037 (2009).
30. Payer, B., Lee, J. T. & Namekawa, S. H. X-inactivation and X-reactivation:
epigenetic hallmarks of mammalian reproduction and pluripotent stem cells.
Hum. Genet. 130, 265–280 (2011).
31. Maherali, N. et al. Directly reprogrammed fibroblasts show global epigenetic
remodeling and widespread tissue contribution. Cell Stem Cell 1, 55–70 (2007).
32. Stadtfeld, M. et al. Aberrant silencing of imprinted genes on chromosome
12qF1 in mouse induced pluripotent stem cells. Nature 465, 175–181 (2010).
33. Boland, M. J. et al. Adult mice generated from induced pluripotent stem cells.
Nature 461, 91–94 (2009).
34. Zhao, X. Y. et al. iPS cells produce viable mice through tetraploid
complementation. Nature 461, 86–90 (2009).
35. Eakin, G. S., Hadjantonakis, A. K., Papaioannou, V. E. & Behringer, R. R.
Developmental potential and behavior of tetraploid cells in the mouse embryo.
Dev. Biol. 288, 150–159 (2005).
36. Eggan, K. & Jaenisch, R. Differentiation of F1 embryonic stem cells into viable
male and female mice by tetraploid embryo complementation. Methods
Enzymol. 365, 25–39 (2003).
37. Lensch, M. W., Schlaeger, T. M., Zon, L. I. & Daley, G. Q. Teratoma formation
assays with human embryonic stem cells: a rationale for one type of human-
animal chimera. Cell Stem Cell 1, 253–258 (2007).
38. Park, I. H. et al. Disease-specific induced pluripotent stem cells. Cell 134,
This study derived iPS cells from patients with a range of diseases,
demonstrating the applications of iPS cells for disease modelling,
pathogenesis studies and drug development.
39. Daley, G. Q. et al. Broader implications of defining standards for the
pluripotency of iPSCs. Cell Stem Cell 4, 200–202 (2009).
40. Miura, K. et al. Variation in the safety of induced pluripotent stem cell lines.
Nature Biotechnol. 27, 743–745 (2009).
41. Feng, Q. et al. Hemangioblastic derivatives from human induced pluripotent
stem cells exhibit limited expansion and early senescence. Stem Cells 28,
42. Hu, B. Y. et al. Neural differentiation of human induced pluripotent stem cells
follows developmental principles but with variable potency. Proc. Natl Acad.
Sci. USA 107, 4335–4340 (2010).
43. Kim, K. et al. Epigenetic memory in induced pluripotent stem cells. Nature 467,
This report describes an exhaustive comparison of pluripotent stem cells
derived from mouse embryos or reprogrammed through nuclear transfer or
transcription factor co-expression, by using several in vitro differentiation
assays and methylation analysis; it revealed that iPS cells manifest molecular
and behavioural features of the donor tissue of origin, indicating a ‘memory’
of the somatic tissue.
44. Hu, Q., Friedrich, A. M., Johnson, L. V. & Clegg, D. O. Memory in induced
pluripotent stem cells: reprogrammed human retinal-pigmented epithelial cells
show tendency for spontaneous redifferentiation. Stem Cells 28, 1981–1991
45. Bar-Nur, O., Russ, H. A., Efrat, S. & Benvenisty, N. Epigenetic memory and
preferential lineage-specific differentiation in induced pluripotent stem cells
derived from human pancreatic islet Beta cells. Cell Stem Cell 9, 17–23 (2011).
46. Urbach, A., Bar-Nur, O., Daley, G. Q. & Benvenisty, N. Differential modeling of
fragile X syndrome by human embryonic stem cells and induced pluripotent
stem cells. Cell Stem Cell 6, 407–411 (2010).
47. Bock, C. et al. Reference maps of human ES and iPS cell variation enable high-
throughput characterization of pluripotent cell lines. Cell 144, 439–452 (2011).
This study generated genome-wide reference maps of DNA methylation and
gene expression, together with the differentiation potential of each cell line,
providing a resource for assessing the similarity of ES cells and iPS cells,
as well as for predicting the differentiation efficiency of a particular cell
line and creating a scorecard for the comprehensive characterization of any
pluripotent cell line.
48. Pick, M. et al. Clone- and gene-specific aberrations of parental imprinting in
human induced pluripotent stem cells. Stem Cells 27, 2686–2690 (2009).
49. Ben-David, U., Mayshar, Y. & Benvenisty, N. Large-scale analysis reveals
acquisition of lineage-specific chromosomal aberrations in human adult stem
cells. Cell Stem Cell 9, 97–102 (2011).
50. Hussein, S. M. et al. Copy number variation and selection during
reprogramming to pluripotency. Nature 471, 58–62 (2011).
This study showed that significantly more CNVs were present in early-
passage human iPS cells than in intermediate-passage human iPS cells,
fibroblasts or human ES cells; it also provided evidence that CNVs conferred
a selective disadvantage.
51. Laurent, L. C. et al. Dynamic changes in the copy number of pluripotency and
cell proliferation genes in human ESCs and iPSCs during reprogramming and
time in culture. Cell Stem Cell 8, 106–118 (2011).
52. Ghosh, Z. et al. Persistent donor cell gene expression among human induced
pluripotent stem cells contributes to differences with human embryonic stem
cells. PLoS ONE 5, e8975 (2010).
53. Wernig, M. et al. A drug-inducible transgenic system for direct reprogramming
of multiple somatic cell types. Nature Biotechnol. 26, 916–924 (2008).
54. Mikkelsen, T. S. et al. Dissecting direct reprogramming through integrative
genomic analysis. Nature 454, 49–55 (2008).
55. Lister, R. et al. Hotspots of aberrant epigenomic reprogramming in human
induced pluripotent stem cells. Nature 471, 68–73 (2011).
This study analysed the methylomes of human iPS cells, ES cells, somatic
cells, and differentiated iPS and ES cells, and revealed megabase-scale
DMRs in iPS cells, indicating incomplete reprogramming of these cells.
56. Shen, Y. et al. X-inactivation in female human embryonic stem cells is in a
nonrandom pattern and prone to epigenetic alterations. Proc. Natl Acad.
Sci. USA 105, 4709–4714 (2008).
57. Tchieu, J. et al. Female human iPSCs retain an inactive X chromosome. Cell
Stem Cell 7, 329–342 (2010).
58. Marchetto, M. C. et al. A model for neural development and treatment of Rett
syndrome using human induced pluripotent stem cells. Cell 143, 527–539
59. Pomp, O. et al. Unexpected X chromosome skewing during culture and
reprogramming of human somatic cells can be alleviated by exogenous
telomerase. Cell Stem Cell 9, 156–165 (2011).
60. Bernstein, B. E. et al. A bivalent chromatin structure marks key developmental
genes in embryonic stem cells. Cell 125, 315–326 (2006).
3 0 4 | N A T U R E | V O L 4 8 1 | 1 9 J A N U A R Y 2 0 1 2
© 2012 Macmillan Publishers Limited. All rights reserved
61. Newman, A. M. & Cooper, J. B. Lab-specific gene expression signatures in
pluripotent stem cells. Cell Stem Cell 7, 258–262 (2010).
62. Humpherys, D. et al. Epigenetic instability in ES cells and cloned mice. Science
293, 95–97 (2001).
63. Soldner, F. et al. Parkinson’s disease patient-derived induced pluripotent stem
cells free of viral reprogramming factors. Cell 136, 964–977 (2009).
64. Carey, B. W. et al. Reprogramming factor stoichiometry influences the
epigenetic state and biological properties of induced pluripotent stem cells.
Cell Stem Cell 9, 588–598 (2011).
65. Ohi, Y. et al. Incomplete DNA methylation underlies a transcriptional memory of
somatic cells in human iPS cells. Nature Cell Biol. 13, 541–549 (2011).
66. Polo, J. M. et al. Cell type of origin influences the molecular and functional
properties of mouse induced pluripotent stem cells. Nature Biotechnol. 28,
67. Martinez, Y. et al. Cellular diversity within embryonic stem cells: pluripotent
clonal sublines show distinct differentiation potential. J. Cell. Mol. Med. http://
dx.doi.org/10.1111/j.1582-4934.2011.01334.x (in the press).
68. Osafune, K. et al. Marked differences in differentiation propensity among
human embryonic stem cell lines. Nature Biotechnol. 26, 313–315 (2008).
69. Muller, F. J. et al. A bioinformatic assay for pluripotency in human cells. Nature
Methods 8, 315–317 (2011).
70. Lee, G. et al. Modelling pathogenesis and treatment of familial dysautonomia
using patient-specific iPSCs. Nature 461, 402–406 (2009).
71. Moretti, A. et al. Patient-specific induced pluripotent stem-cell models for
long-QT syndrome. N. Engl. J. Med. 363, 1397–1409 (2010).
72. Itzhaki, I. et al. Modelling the long QT syndrome with induced pluripotent stem
cells. Nature 471, 225–229 (2011).
This study generated iPS cells from patients with long QT syndrome and
stimulated them to differentiate into cardiomyocytes that paralleled the
disease phenotype in vitro, and these cells were then used to evaluate the
potency of existing and new therapeutic agents.
73. Agarwal, S. et al. Telomere elongation in induced pluripotent stem cells from
dyskeratosis congenita patients. Nature 464, 292–296 (2010).
74. Batista, L. F. et al. Telomere shortening and loss of self-renewal in dyskeratosis
congenita induced pluripotent stem cells. Nature 474, 399–402 (2011).
75. Agarwal, S. & Daley, G. Q. Telomere dynamics in dyskeratosis congenita: the
long and the short of iPS. Cell Res. 21, 1157–1160 (2011).
76. Hanna, J. et al. Treatment of sickle cell anemia mouse model with iPS cells
generated from autologous skin. Science 318, 1920–1923 (2007).
77. Wernig, M. et al. Neurons derived from reprogrammed fibroblasts functionally
integrate into the fetal brain and improve symptoms of rats with Parkinson’s
disease. Proc. Natl Acad. Sci. USA 105, 5856–5861 (2008).
78. Zhao, T., Zhang, Z. N., Rong, Z. & Xu, Y. Immunogenicity of induced pluripotent
stem cells. Nature 474, 212–215 (2011).
79. Harrison, N. J., Baker, D. & Andrews, P. W. Culture adaptation of embryonic stem
cells echoes germ cell malignancy. Int. J. Androl. 30, 275–281 (2007).
80. Mayshar, Y. et al. Identification and classification of chromosomal aberrations in
human induced pluripotent stem cells. Cell Stem Cell 7, 521–531 (2010).
81. Murry, C. E. & Keller, G. Differentiation of embryonic stem cells to clinically
relevant populations: lessons from embryonic development. Cell 132, 661–680
82. Lowry, W. E. et al. Generation of human induced pluripotent stem cells from
dermal fibroblasts. Proc. Natl Acad. Sci. USA 105, 2883–2888 (2008).
83. Huangfu, D. et al. Induction of pluripotent stem cells from primary human
fibroblasts with only Oct4 and Sox2. Nature Biotechnol. 26, 1269–1275 (2008).
84. Sommer, C. A. et al. Induced pluripotent stem cell generation using a single
lentiviral stem cell cassette. Stem Cells 27, 543–549 (2009).
85. Anokye-Danso, F. et al. Highly efficient miRNA-mediated reprogramming of
mouse and human somatic cells to pluripotency. Cell Stem Cell 8, 376–388
86. Woltjen, K. et al. piggyBac transposition reprograms fibroblasts to induced
pluripotent stem cells. Nature 458, 766–770 (2009).
87. Somers, A. et al. Generation of transgene-free lung disease-specific human
induced pluripotent stem cells using a single excisable lentiviral stem cell
cassette. Stem Cells 28, 1728–1740 (2010).
88. Zhou, W. & Freed, C. R. Adenoviral gene delivery can reprogram human
fibroblasts to induced pluripotent stem cells. Stem Cells 27, 2667–2674
89. Stadtfeld, M., Nagaya, M., Utikal, J., Weir, G. & Hochedlinger, K. Induced
pluripotent stem cells generated without viral integration. Science 322,
90. Okita, K., Nakagawa, M., Hyenjong, H., Ichisaka, T. & Yamanaka, S. Generation
of mouse induced pluripotent stem cells without viral vectors. Science 322,
91. Si-Tayeb, K. et al. Generation of human induced pluripotent stem cells by
simple transient transfection of plasmid DNA encoding reprogramming factors.
BMC Dev. Biol. 10, 81 (2010).
92. Fusaki, N., Ban, H., Nishiyama, A., Saeki, K. & Hasegawa, M. Efficient induction
of transgene-free human pluripotent stem cells using a vector based on Sendai
virus, an RNA virus that does not integrate into the host genome. Proc. Jpn Acad.
85, 348–362 (2009).
93. Kim, D. et al. Generation of human induced pluripotent stem cells by direct
delivery of reprogramming proteins. Cell Stem Cell 4, 472–476 (2009).
94. Zhou, H. et al. Generation of induced pluripotent stem cells using recombinant
proteins. Cell Stem Cell 4, 381–384 (2009).
95. Warren, L. et al. Highly efficient reprogramming to pluripotency and directed
differentiation of human cells with synthetic modified mRNA. Cell Stem Cell 7,
96. Miyoshi, N. et al. Reprogramming of mouse and human cells to pluripotency
using mature microRNAs. Cell Stem Cell 8, 633–638 (2011).
97. Hanna, J. et al. Human embryonic stem cells with biological and epigenetic
characteristics similar to those of mouse ESCs. Proc. Natl Acad. Sci. USA 107,
98. Buecker, C. et al. A murine ESC-like state facilitates transgenesis and
homologous recombination in human pluripotent stem cells. Cell Stem Cell 6,
99. Li, W. et al. Generation of rat and human induced pluripotent stem cells by
combining genetic reprogramming and chemical inhibitors. Cell Stem Cell 4,
100. Nichols, J. & Smith, A. Naive and primed pluripotent states. Cell Stem Cell 4,
Supplementary Information is linked to the online version of the paper at
Acknowledgements G.Q.D. is supported by grants from the National Institutes of
Health (RO1-DK70055, RO1-DK59279 and UO1-HL100001, as well as special funds
from the American Recovery and Reinvestment Act of 2009 (RC2-HL102815), the
Roche Foundation for Anemia Research, Alex’s Lemonade Stand Foundation and
the Harvard Stem Cell Institute. G.Q.D. is an affiliate member of the Broad Institute,
a recipient of Clinical Scientist Awards in Translational Research from the Burroughs
Wellcome Fund and the Leukemia and Lymphoma Society, and an investigator of
the Howard Hughes Medical Institute and the Manton Center for Orphan Disease
Research. We gratefully acknowledge A. de Los Angeles for providing the images for
Fig. 1a, d, f; T. Onder, for Fig. 1b; and A. Cherry, for Fig. 1c, e.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The author declares competing financial interests:
details accompany the full-text HTML version of the paper at www.nature.
com/nature. Readers are welcome to comment on the online version of this
article at www.nature.com/nature. Correspondence should be addressed to
1 9 J A N U A R Y 2 0 1 2 | V O L 4 8 1 | N A T U R E | 3 0 5
© 2012 Macmillan Publishers Limited. All rights reserved