ARTICLES
Disease-corrected haematopoietic
progenitors from Fanconi anaemia induced
pluripotent stem cells
A´ngel Raya1,2,3, Ignasi Rodrı´guez-Piza`1, Guillermo Guenechea4,5, Rita Vassena1, Susana Navarro4,5,
Marı´a Jose´ Barrero1, Antonella Consiglio1,6, Maria Castella`5,7, Paula Rı´o4,5, Eduard Sleep1,3, Federico Gonza´lez1,
Gustavo Tiscornia1, Elena Garreta1,3, Trond Aasen1,3, Anna Veiga1, Inder M. Verma8, Jordi Surralle´s5,7,
Juan Bueren4,5 & Juan Carlos Izpisu´a Belmonte1,9
The generation of induced pluripotent stem (iPS) cells has enabled the derivation of patient-specific pluripotent cells and
provided valuable experimental platforms to model human disease. Patient-specific iPS cells are also thought to hold great
therapeutic potential, although direct evidence for this is still lacking. Here we show that, on correction of the genetic defect,
somatic cells from Fanconi anaemia patients can be reprogrammed to pluripotency to generate patient-specific iPS cells.
These cell lines appear indistinguishable from human embryonic stem cells and iPS cells from healthy individuals. Most
importantly, we show that corrected Fanconi-anaemia-specific iPS cells can give rise to haematopoietic progenitors of the
myeloid and erythroid lineages that are phenotypically normal, that is, disease-free. These data offer proof-of-concept that
iPS cell technology can be used for the generation of disease-corrected, patient-specific cells with potential value for cell
therapy applications.
Thepossibility of reprogrammingmature somatic cells to generate iPS
cells1–5 has enabled the derivation of disease-specific pluripotent cells,
thus providing unprecedented experimental platforms to model
human disease6–9. In addition, the generation of patient-specific iPS
cells may have a wide range of applications in cell and gene therapy,
and could be particularly relevant for the treatment of inherited bone
marrow failure (BMF) syndromes, where the progressive decline in
haematopoietic stem cell (HSC) numbers limits the production of
peripheral blood cells.
Among the different inherited BMF syndromes, Fanconi anaemia
(FA) is the most common10. FA is a rare recessive, autosomal or
X-linked, chromosomal instability disorder caused by mutations in
any of the 13 genes so far identified in the FA pathway11. Cells from
these patients display typical chromosomal instability and hypersensi-
tivity toDNAcrosslinking agents, characteristics that are used tomake
the diagnosis of FA12. Most FA patients develop BMF, which typically
appears during the first decade of life, and some patients show
increased predisposition to develop malignancies (cumulative inci-
dence of,30% by 40 years of age)13. Currently, the therapy of choice
for BMF in FA patients is the transplantation of haematopoietic grafts
from HLA-identical siblings, whereas the output of transplants from
non-related donors is more limited14,15. Somatic mosaicism, acting as
a natural gene therapy in FA patients, showed that genetic correction
confers a selective growth advantage to HSCs from FA patients, a
process that can ultimately restore the haematopoietic system of the
patient with phenotypically normal cells16–18. A selective proliferation
advantage has also been observed in FA mouse models after ex vivo
genetic correction of their HSCs with lentiviral vectors19. In spite of
these observations, gene therapy trials conducted so far in FA patients
have not been clinically successful20,21, owing to the paucity and poor
quality of HSCs in the bone marrow of FA patients20–23.
As a consequence of the genetic instability of FA cells, genetic
defects eventually produced before gene therapy correction would
not be repaired. Nevertheless, the generation of genetically corrected
FA-specific iPS cells by the reprogramming of non-haematopoietic
somatic cells would result in the production of large numbers of
autologous, genetically stable HSCs that may be used to treat BMF
in FA patients.
Generation of patient-specific iPS cells
We obtained samples from six FA patients, four from the FA-A com-
plementation group (patients FA5, FA90, FA153 and FA404) and two
from the FA-D2 complementation group (FA430 and FA431).
Samples from patients FA5, FA90, FA153, FA430 and FA431 were
cryopreserved primary dermal fibroblasts that had undergone an
undetermined number of passages. From patient FA404 we obtained
a skin biopsy, from which we established primary cultures of dermal
fibroblasts and epidermal keratinocytes. We first attempted to opti-
mize the reprogramming protocol using primary dermal fibroblasts
from a foreskin biopsy of a healthy donor (see Supplementary
Information and Supplementary Fig. 1). Our improved reprogram-
ming protocol consisted of two rounds of infection with mouse-
stem-cell-virus-based retroviruses encoding amino-terminal
Flag-tagged versions of OCT4 (also known as POU5F1), SOX2,
KLF4 and c-MYC (also known as MYC), performed 6 days apart.
Transduced fibroblasts were passaged after 5 days onto a feeder layer
1Center for Regenerative Medicine in Barcelona, Dr. Aiguader 88, 08003 Barcelona, Spain. 2Institucio´ Catalana de Recerca i Estudis Avanc¸ats (ICREA), 3Networking Center of
Biomedical Research in Bioengineering, Biomaterials and Nanomedicine (CIBER-BBN), 4Hematopoiesis and Gene Therapy Division, Centro de Investigaciones Energe´ticas,
Medioambientales y Tecnolo´gicas (CIEMAT), Av. Complutense 22, 28040 Madrid, Spain. 5Networking Center of Biomedical Research in Rare Diseases (CIBERER), 6Department of
Biomedical Science and Biotechnology, University of Brescia, Viale Europa 11, 25123 Brescia, Italy. 7Department of Genetics and Microbiology, Universitat Autonoma de Barcelona,
08193 Bellaterra, Spain. 8Laboratory of Genetics, 9Gene Expression Laboratory, Salk Institute for Biological Studies, 10010 North Torrey Pines Road, La Jolla, California 92037, USA.
Vol 460 | 2 July 2009 |doi:10.1038/nature08129
53
Macmillan Publishers Limited. All rights reserved©2009
of primary human fibroblasts and switched to human embryonic stem
(ES) cellmedium the next day.We also included a selection step based
on the combined inhibition of MAP2K1 and GSK3B with inhibitors
PD0325901 and CT99021 (a combination termed 2i that enhances
derivation and growth of mouse ES cells24) for 1week, starting 1week
after plating onto feeders.
Because of the genetic instability and apoptotic predisposition of
FA cells25, somatic cells were reprogrammed either directly or after
genetic correction with lentiviral vectors encoding FANCA or
FANCD2, respectively. We have previously shown that genetic com-
plementation of human and mouse FA cells with these vectors effi-
ciently corrects the FA phenotype19,23,26. We were not successful at
obtaining iPS-like colonies from fibroblasts of patients FA5, FA153
or FA430, either unmodified or corrected, after at least five repro-
gramming attempts, probably owing to the cells having accumulated
too many passages and/or karyotypic abnormalities (Supplementary
Table 1). However, from patient FA90 we readily obtained iPS-like
colonies when using genetically corrected fibroblasts (Fig. 1a).
Overall, we obtained 10–15 iPS-like colonies in each of 3 independent
experiments. Of these, we randomly picked ten colonies, all of which
could successfully be expanded and grew as coloniesmorphologically
indistinguishable from human ES cells (Fig. 1b) that stained strongly
positive for alkaline phosphatase activity (Fig. 1c). Five of these lines
(cFA90-44-1, -11, -14, -20 and -21) were selected for further char-
acterization. All of them displayed a normal karyotype (46 XX) at
passages 12–16 and could be maintained in culture for, at least, 20
passages. At the time of the writing, cFA90-44-14 had undergone 43
passages without signs of replicative crisis, while maintaining a nor-
mal karyotype (Supplementary Fig. 2). Immunofluorescence ana-
lyses of the five lines revealed expression of transcription factors
(OCT4, SOX2, NANOG) and surface markers (SSEA3, SSEA4,
TRA1-60, TRA1-81) characteristic of pluripotent cells (Fig. 1d–f
and Supplementary Fig. 3).
With somatic cells from another FA-A patient, patient FA404, we
obtained similar results. Fibroblasts that had been transduced with
lentiviruses encoding FANCA (Fig. 1g) were readily reprogrammed
to generate iPS-like cells (Fig. 1h). We established two cell lines
(cFA404-FiPS4F1 and cFA404-FiPS4F2), which displayed typical
human ES-like morphology and growth characteristics, stained posi-
tive for alkaline phosphatase activity and expressed all the pluripo-
tency-associated markers tested (Fig. 1i–l and Supplementary Fig. 4).
From patient FA404 we also derived primary epidermal keratino-
cytes, which we reprogrammed using a protocol recently set up in
our laboratory27. We generated three iPS cell lines (cFA404-KiPS4F1,
-KiPS4F3 and -KiPS4F6) from genetically corrected keratinocytes,
which displayed all the main characteristics of bona fide iPS cells
and human ES cells (Supplementary Fig. 4) and a normal 46 XY
karyotype (Supplementary Fig. 2).
Wewere also successful at reprogramming fibroblasts from patient
FA431 (Supplementary Fig. 5a), a FA-D2 patient. In this case, iPS-
like colonies appeared in roughly equal numbers from either
unmodified or genetically corrected fibroblasts (Supplementary
Table 1). We picked two iPS-like colonies from either condition,
which grew after passaging and stained positive for alkaline phos-
phatase activity (Supplementary Fig. 5c, g). However, whereas those
derived from corrected fibroblasts (cFA431-44-1 and cFA431-44-2)
could be maintained in culture for extended periods of time (18
passages at the time of writing) and showed expression of pluripo-
tency-associated transcription factors and surface markers (Sup-
plementary Fig. 5d–f and data not shown), those derived from
unmodified fibroblasts experienced a progressive growth delay and
could not be maintained over the third passage (Supplementary
Fig. 5g). The observation that uncorrected FA-D2 fibroblasts from
patient FA431 could be reprogrammed, while we only obtained iPS
cells from FANCA-complemented fibroblasts from patients FA90 or
FA404, could be explained by the fact that FA-D2 patients, in par-
ticular FA431, carry hypomorphic mutations compatible with the
expression of residual FANCD2 protein28. Therefore, it appears that
restoration of the FA pathway is a pre-requisite for iPS cell generation
from somatic cells of FA patients (in total, 12 out of 28 independent
reprogramming attempts were successful when using genetically
corrected cells—also including the patients for which reprogram-
ming was never successful—versus 0 out of 28 successful attempts
when uncorrected cells were used; x2 [1]5 15.27, P5 9.33 1025).
OCT4
SSEA3
SOX2
SSEA3
NANOG
SSEA4
OCT4
SSEA3
SOX2
SSEA4
NANOG
TRA1-60
OCT4
SSEA3
SOX2
SSEA3
NANOG
SSEA4
OCT4
SSEA3
SOX2
SSEA4
NANOG
TRA1-60
a
b
c
d
e
f
g
h
i
j
k
l
Figure 1 | Derivation of patient-specific induced pluripotent stem cells
from Fanconi anaemia patients. a–f, Successful reprogramming of
genetically corrected primary dermal fibroblasts (a) derived from patient
FA90. b, Colony of iPS cells from the cFA90-44-14 line grown on Matrigel-
coated plates showing human-ES-cell-like morphology. c–f, The same iPS
cell line shows strong alkaline phosphatase staining (c) and expression of the
transcription factors OCT4 (d), SOX2 (e) and NANOG (f) and the surface
markers SSEA3 (d, e) and SSEA4 (f). g, Genetically corrected fibroblasts
from patient FA404. h, Colony of iPS cells from the cFA404-FiPS4F1 line
grown on feeder cells displaying typical human ES cell morphology. i–l, The
same iPS cell line shows strong alkaline phosphatase staining (i) and
expression of the pluripotency-associated transcription factors OCT4
(j), SOX2 (k) and NANOG (l) and surface markers SSEA3 (j), SSEA4 (k) and
TRA1-80 (l). Cell nuclei were counterstained with 4,6-diamidino-2-
phenylindole (DAPI) in d–f and j–l. Scale bars, 100mm (a, c–g, i–l) and
250mm (b, h).
ARTICLES NATURE | Vol 460 | 2 July 2009
54
Macmillan Publishers Limited. All rights reserved©2009
Characterization of iPS cells
Out of the 19 FA-iPS cell lines generated in these studies, we selected
10 for more thorough characterization (Supplementary Table 1). We
confirmed the presence of the reprogramming transgenes integrated
in their genome by polymerase chain reaction (PCR) of genomic
DNA (Fig. 2a and Supplementary Fig. 6), as well as the origin of
the iPS cell lines by comparing their HLA type and DNA fingerprint
with those of patients’ somatic cells (Supplementary Table 2). In all
lines tested, transgenic expression of the four reprogramming factors
was reduced to low or undetectable levels, compared to an iPS cell
line (KiPS4F3) previously shown not to have silenced the retroviral
expression ofOCT4 and c-MYC27 (Fig. 2b). Furthermore, all the FA-
iPS cell lines tested showed re-activation of endogenous OCT4 and
SOX2 expression, as well as of other pluripotency-associated tran-
scription factors such asNANOG, REX-1 (also known as ZFP42) and
CRIPTO (also known as TDGF1; Fig. 2c). Taking advantage of the
fact that our retroviral transgenes were Flag-tagged, we confirmed by
immunofluorescence that iPS cells displayed negligible anti-Flag
immunoreactivity (Fig. 2d–g). Finally, the promoters of
the pluripotency-associated transcription factors OCT4 and
NANOG, heavily methylated in patients’ fibroblasts, were demethy-
lated in FA-iPS cells (Fig. 2h), indicating epigenetic reprogramming
to pluripotency.
We next analysed the differentiation ability of FA-iPS cells. In vitro,
iPS-derived embryoid bodies readily differentiated into endoderm,
ectoderm and mesoderm derivatives as judged by cell morphology
and specific immunostaining with antibodies against a-fetoprotein/
FOXA2, TuJ1/GFAP and a-actinin, respectively (Fig. 3a–c, and
Supplementary Fig. 7). Following specific in vitro differentiation pro-
tocols, iPS cells gave rise to specialized mesoderm-derived cell types
such as rhythmically beating cardiomyocytes (Supplementary Movie
K
LF
4
c-
M
Y
C
O
C
T4
S
O
X
2
K
LF
4
c-
M
Y
C
O
C
T4
S
O
X
2
K
LF
4
c-
M
Y
C
O
C
T4
S
O
X
2
cFA90
iPS
cFA404
iPS Fibr.
Retroviral-derived expression
G
en
e
ex
p
re
ss
io
n
re
la
tiv
e
to
G
A
P
D
H
(%
)
OCT4 SOX2 KLF4 c-MYC
cFA90 fibr.
cFA90-44-1
cFA90-44-11
cFA90-44-14
cFA90-44-21
Total expression
OCT4 SOX2 KLF4 c-MYC NANOG REX-1 CRIPTO
b
d
f
h
0.0%
20.0%
cFA404
iPS
2.5%
12.5%
cFA90
iPS
cFA90
fibr.
48.0%
57.1%
–2261
–2231
–2182
–1401
–1368
–1337
NANOG
–1449
–1295
OCT4
–2305
–2137
e
g
0
5
10
15
20a
0
10
20
30
40
50
60
cFA404 fibr.
cFA404-KiPS4F1
cFA404-KiPS4F3
cFA404-KiPS4F6
cFA404-FiPS4F1
cFA404-FiPS4F2
ES[4]
KiPS4F3
G
en
e
ex
p
re
ss
io
n
re
la
tiv
e
to
G
A
P
D
H
(%
)
c
Figure 2 | Molecular characterization of FA-iPS cell lines. a, PCR of
genomic DNA to detect integration of the indicated retroviral transgenes in
FA-iPS cell lines cFA90-44-14 (cFA90) and cFA404-FiPS4F1 (cFA404).
Genetically corrected fibroblasts (Fibr.) from patient FA404 before
reprogramming were used as negative control. b, c, Quantitative PCR with
reverse transcription (RT–PCR) analyses of the expression levels of
retroviral-derived reprogramming factors (b) and of total expression levels
of reprogramming factors and pluripotency-associated transcription factors
(c) in the indicated patients’ fibroblasts (fibr.) and FA-iPS cell lines. Human
ES cells (ES[4]) and partially silenced iPS cells (KiPS4F3) are included as
controls. Transcript expression levels are plotted relative to GAPDH
expression. d–g, Colony of cFA90-44-14 iPS cells showing high levels of
endogenous NANOG expression (e, green channel in d) and absence of Flag
immunoreactivity (f, red channel in d). Cell nuclei were counterstained with
DAPI (g, blue channel in d). h, Bisulphite genomic sequencing of the OCT4
and NANOG promoters showing demethylation in FA-iPS cell lines cFA90-
44-14 and cFA404-KiPS4F3, compared to patient’s fibroblasts. Open and
closed circles represent unmethylated andmethylated CpGs, respectively, at
the indicated promoter positions. Scale bar, 100 mm.
α-fetoprotein
FOXA2
TuJ1
GFAP α-actinin
α-fetoprotein
α-fetoprotein
TuJ1 α-actinin
a b c
d e f
Figure 3 | Pluripotency of FA-iPS cells. a–c, In vitro differentiation
experiments of cFA404-FiPS4F2 iPS cells reveal their potential to generate
cell derivatives of all three primary germ cell layers. Immunofluorescence
analyses show expression of markers of a, endoderm (a-fetoprotein, green;
FOXA2, red), b, neuroectoderm (TuJ1, green; GFAP, red), and, c, mesoderm
(a-actinin, red). d–f, Injection of cFA90-44-14 iPS cells into the testes of
immunocompromised mice results in the formation of teratomas
containing structures that represent the three main embryonic germ layers.
Endoderm derivatives (d, e) include glandular structures that stain positive
for endoderm markers (a-fetoprotein, green); ectoderm derivatives
(e) include structures that stain positive for neuroectoderm markers (TuJ1,
red); mesoderm derivatives (f) include structures that stain positive for
muscle markers (a-actinin, red). All images are from the same tumour. Scale
bars, 100mm (a, b, d, e) and 25mm (c, f).
NATURE | Vol 460 | 2 July 2009 ARTICLES
55
Macmillan Publishers Limited. All rights reserved©2009
1) and haematopoietic progenitor cells (see below). We also sub-
jected our FA-iPS cells to the most stringent test available to assess
pluripotency of human cells, the formation of bona fide teratomas29.
For this purpose, we injected cells from eight different lines into the
testes of immunocompromised mice. In all cases, teratomas could be
recovered after 8–10weeks that were composed of complex struc-
tures representing the three main embryonic germ layers, including
glandular formations that stained positive for definitive endoderm
markers, neural structures that expressed neuroectodermal markers,
and mesoderm derivatives such as muscle and cartilage (Fig. 3d–f,
Supplementary Fig. 8 and data not shown). Using comparable assays,
we have recently characterized the in vitro differentiation and tera-
toma induction abilities of a variety of normal human pluripotent
stem cell lines, including human ES cells30 and iPS cells generated
from healthy donors27. Overall, we did not detect conspicuous differ-
ences in the differentiation ability of FA-iPS cell lines compared to
that of either human ES cells or normal iPS cells.
FA-specific iPS cells are disease-free
Consistent with the previous genetic correction of somatic cells used
for reprogramming, we could detect the presence of integrated copies
of the gene therapy vectors by quantitative PCRof genomicDNA in all
FA-iPS cell lines tested (Supplementary Fig. 9a). A concern with gene
therapy strategies is the silencing of the correcting transgene. For this
reason, we chose lentiviruses as gene therapy vectors, because lenti-
viral transgenes are particularly resistant to silencing in human ES
cells31. However, this resistance appears to be promoter-dependent32
andnearly complete silencingof lentiviral transgeneshas been recently
observed in the context of induced reprogramming3,8. In our experi-
ments, the FANCA lentivirus was partially silenced in FA-iPS cells, as
evidenced by the loss of IRES-GFP (internal ribosome entry site-green
fluorescent protein) fluorescence (data not shown), whichwas detect-
able in transduced fibroblasts (Supplementary Fig. 9b). However,
transgene silencing was not complete, as we could detect FANCA
expression in all the FA-iPS cell lines analysed, but not in the patients’
fibroblasts (Fig. 4a). To test the functionality of the FA pathway in FA-
iPS cells, we induced subnuclear accumulation of stalled replication
forks by high-energy local ultraviolet irradiation across a filter with
5mm pore