LETTERS
Pluripotent stem cells induced from adult neural stem
cells by reprogramming with two factors
Jeong Beom Kim
1
*, Holm Zaehres
1
*, Guangming Wu
1
, Luca Gentile
1
, Kinarm Ko
1
, Vittorio Sebastiano
1
,
Marcos J. Arau´zo-Bravo
1
, David Ruau
2
, Dong Wook Han
1
, Martin Zenke
2
& Hans R. Scho¨ler
1
Reprogramming of somatic cells is a valuable tool to understand
the mechanisms of regaining pluripotency and further opens up
the possibility of generating patient-specific pluripotent stem
cells. Reprogramming of mouse and human somatic cells into
pluripotent stem cells, designated as induced pluripotent stem
(iPS) cells, has been possible with the expression of the transcrip-
tion factor quartet Oct4 (also known as Pou5f1), Sox2, c-Myc and
Klf4 (refs 1–11). Considering that ectopic expression of c-Myc
causes tumorigenicity in offspring
2
and that retroviruses them-
selves can cause insertional mutagenesis, the generation of iPS
cells with a minimal number of factors may hasten the clinical
application of this approach. Here we show that adult mouse
neural stem cells express higher endogenous levels of Sox2 and
c-Myc than embryonic stem cells, and that exogenous Oct4
together with either Klf4 or c-Myc is sufficient to generate iPS cells
from neural stem cells. These two-factor iPS cells are similar to
embryonic stem cells at the molecular level, contribute to develop-
ment of the germ line, and form chimaeras. We propose that, in
inducing pluripotency, the number of reprogramming factors can
be reduced when using somatic cells that endogenously express
appropriate levels of complementing factors.
Mouse and human somatic cells can be reprogrammed into iPS
cells by the expression of a defined set of factors (Oct4, Sox2, c-Myc
and Klf4, as well as Nanog and human LIN28)
1–11
. It is possible to
generate iPS cells from mouse and human fibroblasts in the absence
of c-Myc retrovirus
6,7
, and therefore it was suggested that endogenous
expression of c-Myc could have a role in the reprogramming process.
Neural stem cells (NSCs) endogenously express Sox2, which may
function in maintaining the stemness and multipotency of
NSCs
12,13
, and Sox2 was suggested in maintaining cellular pluripo-
tency by regulating the expression of Oct4 (ref. 14). NSCs were estab-
lished from adult Oct4–GFP (OG2)/ROSA26 heterozygous
transgenic mouse brains
15–17
, expressing green fluorescent protein
(GFP) under the control of the Oct4 promoter (Oct4–GFP) and
the Escherichia coli lacZ transgene from the constitutive ROSA26
locus (ROSA26 lacZ). The established NSCs had the capacity for
self-renewal and multipotency (Supplementary Fig. 1).
Compared to embryonic stem cells (ESCs), expression of Sox2 was
approximately twofold higher in NSCs. c-Myc and Klf4 were also
expressed at levels about tenfold higher and eightfold lower in
NSCs than in ESCs, respectively (Fig. 1a). Western blot analyses
showed that the relationship between protein and RNA levels in
NSCs corresponded to that in ESCs for Oct4, Sox2 and Klf4; the
c-Myc protein level was comparable in NSCs and ESCs (Fig. 1b).
In this study, we attempted to reprogram NSCs into iPS cells by
introducing different combinations of the four factors Oct4, Sox2,
c-Myc and Klf4 (Supplementary Table 1) using the retroviral MX
vector system. We assessed the ability of 15 different transcription
factor combinations to induce Oct4–GFP-positive colony formation.
Six combinations were able to induce the generation of iPS cells from
NSCs, as judged by the formation of GFP
1
colonies and the estab-
lishment of iPS cell lines. We observed GFP
1
cells 4 days after trans-
duction with the combination containing all four factors—that is, the
control combination—and noted a gradual increase in the number of
GFP
1
colonies during the first 2 weeks post-infection
(Supplementary Fig. 2a). We established four-factor iPS cells from
GFP
1
ESC-like colonies on day 14. These four-factor iPS cells stained
positive for stage-specific embryonic antigen-1 (SSEA-1) and alkal-
ine phosphatase (Supplementary Fig. 2b), showed messenger RNA
expression patterns similar to those in ESCs (Fig. 2a, Supplementary
Fig. 2c), and led to teratoma formation on injection into nude mice
(Supplementary Fig. 2d). Our results demonstrate that NSCs can be
reprogrammed into iPS cells by the four transcription factors: Oct4,
Sox2, c-Myc and Klf4.
Three different combinations of three factors were also capable of
generating iPS cells from NSCs: Oct4, Klf4 and c-Myc (OKM); Oct4,
Klf4 and Sox2 (OKS); and Oct4, c-Myc and Sox2 (OMS;
Supplementary Table 1). We did not observe GFP
1
colonies for
the three-factor combinations that did not include Oct4. GFP
1
col-
onies were observed 1 week after transduction with the OKM com-
bination (without Sox2). However, GFP
1
colony formation was
observed only after 14–21 days with the OKS combination (without
c-Myc), and was even more delayed with the OMS combination
(without Klf4; Supplementary Table 1). Nonetheless, these OKM,
OKS and OMS three-factor iPS cells had similar gene expression
profiles to ESCs (Supplementary Fig. 3a), and all types of three-factor
iPS cells differentiated into all three germ layers (Supplementary Fig.
3b). Taken together, these results indicate that three-factor iPS cells
could be generated in the absence of Sox2, Klf4 or c-Myc retroviruses
in NSCs, which endogenously express these three factors.
We next assessed the ability of two-factor combinations to induce
the generation of iPS cells from NSCs. Only two combinations were
successful in reprogramming NSCs. We first observed GFP
1
colonies
in NSC cultures infected with Oct4 and Klf4 (OK) and 1–2 weeks later
in those infected with Oct4 and c-Myc (OM; Supplementary Table 1).
The two-factor OM iPS cells showed an ESC-like gene expression
pattern and contributed to the three germ layers in teratomas
(Supplementary Fig. 3a, b). Klf4 and c-Myc may exert similar func-
tions. It is known that Klf4 has a role in the inactivation of the p53 (also
known as Trp53) tumour-suppressor gene, which leads to cell immor-
talization; Klf4 also works in conjunction with the RAS
V12
oncogenic
signal transduction protein to stimulate cellular proliferation
18,19
.
1
Department of Cell and Developmental Biology, Max Planck Institute for Molecular Biomedicine, Ro¨ntgenstrasse 20, 48149 Mu¨nster, NRW, Germany.
2
Institute for Biomedical
Engineering, Department of Cell Biology, RWTH Aachen University Medical School, Pauwelsstrasse 30, 52074 Aachen, NRW, Germany.
*These authors contributed equally to this work.
Vol 454 | 31 July 2008 | doi:10.1038/nature07061
646
©2008 Macmillan Publishers Limited. All rights reserved

Similarly, the immortalizing gene product c-Myc, in conjunction with
mutant RAS, exhibits an oncogenic effect
20
. It has been reported that
c-Myc increases telomerase activity in NSCs, a mechanism possibly
responsible for the immortalization of NSCs
21
. Because c-Myc
increases tumorigenicity in chimaera pups
2
, the recent studies dem-
onstrating iPS cell generation without the c-Myc retrovirus
6,7
repre-
sent a significant finding. Importantly, our results of inducing iPS cells
from NSCs with Oct4 and Klf4 and without c-Myc bring us even closer
to the generation of iPS cells suitable for transplantation.
We compared two-factor OK iPS cells with four-factor iPS cells
and ESCs. On day 14 post-infection, five GFP
1
colonies were dis-
sociated and propagated under ESC culture conditions (Fig. 1c, f),
yielding three (that is, 60%) two-factor OK iPS cell clones (B-2, D-7
and F-4) that were morphologically indistinguishable from ESCs
(Fig. 1d, g). No colonies formed from NSCs infected with control
virus (MX) (Fig. 1e, h). We estimated the reprogramming efficiencies
from the number of Oct4–GFP
1
colonies and transduction rates with
MX–GFP control virus on NSCs for the two-factor OK iPS and the
four-factor iPS by time course (Fig. 1i, j). Thereby, we calculated a
reprogramming efficiency of 3.6% for four-factor reprogramming of
NSCs and 0.11% for the two-factor approach, which is comparable to
reprogramming of fibroblasts with selection (,0.08%)
1–3
and
without selection (0.5%)
5
(Fig. 1j and Supplementary Table 2).
Transduction with all four factors had a positive impact on the tim-
ing and number of GFP
1
colonies. Integration of the viral transgenes
was confirmed by genotyping by polymerase chain reaction (PCR;
Supplementary Fig. 4). The viral transgenes of all four factors were
detected in four-factor iPS cells, whereas two-factor OK iPS cells only
contained the Oct4 and Klf4 transgenes.
Two-factor OK iPS cells stained positive for SSEA-1 and alkaline
phosphatase (Supplementary Fig. 5), and exhibited expression pat-
terns of ESC marker genes similar to those in four-factor iPS cells and
ESCs (Fig. 2a). Quantitative real-time PCR (qRT-PCR) results
demonstrated that expression of endogenous Oct4, Sox2, c-Myc and
Klf4 in two-factor OK iPS cells was comparable to that in ESCs
(Supplementary Fig. 6a), and showed the silencing of the viral tran-
scripts in two-factor OK iPS cells with a 1,000-fold reduction after 30
days (Supplementary Fig. 6b). In addition, in four-factor iPS cells, the
endogenous expression levels were similar to those in ESCs, and the
expression of the transgenes was completely silenced (Supplementary
Fig. 6c, d). Global gene expression of two-factor iPS also clustered
close to ESCs and four-factor iPS (Fig. 2b and Supplementary Fig. 7).
Scatter plots of DNA microarray analyses demonstrated a higher simi-
larity between two-factor iPS cells and ESCs than between two-factor
a
0
20
40
60
80
G
F
P
+

c
o
l
o
n
y

n
u
m
b
e
r
Two factors
Four factors
Day 7 Day 14 Day 21
i
Oct4
Nanog
Klf4
Sox2
c-Myc
b-actin
E
S
C
s
N
S
C
s
Oct4
Klf4
Sox2
c-Myc
β-actin
E
S
C
s
N
S
C
s
b
Four factors Two factors (Oct4 + Klf4)
GFP
+
colony GFP
+
colony
Day 7 7 ± 3 0 ± 0 0 ± 0
Day 14 29 ± 7 1.45 ± 0.3
0.35 ± 0.1
5 ± 2
Day 21 73 ± 11 3.6 ± 0.5 11 ± 2 0.11 ± 0.02
j
Efficiency (%) Efficiency (%)
0.05 ± 0.02
Day 14 Day 30
Oct4–GFP Oct4–GFP
Control
Mock
NSCsd
g
e
h
c
f
Oct4 Klf4Sox2 c-Myc
G
e
n
e

e
x
p
r
e
s
s
i
o
n

r
e
l
a
t
i
v
e

t
o

E
S
C
s
10
2
10
0
10
–2
10
–4
10
–6
Figure 1 | Generation of two-factor Oct4/Klf4 (OK) iPS cells from adult
NSCs of OG2/ROSA26 transgenic mice. a, RT–PCR and qRT–PCR
analyses of Oct4, Nanog, Klf4, Sox2 and c-Myc in ESCs and NSCs (n53; error
bars indicate s.d.). b-actin was used as a loading control. b, Western blot
analyses of the four factors in ESCs and NSCs. Anti-actin antibody was used
as a loading control. c, Morphology of two-factor OK iPS cell colony on day
14 post-infection. Shown is an ESC-like colony expressing Oct4–GFP
(f). d, g, Morphology of established two-factor OK iPS cells (clone F-4) on
day 30 post-infection, grown on irradiated MEFs. Phase contrast (d) and
Oct4–GFP (g) are shown. e, h, Morphology of NSCs and mock infection on
day 30 post-infection. i, Generation of GFP-positive colonies at day 7, 14 and
21 after two-factor OK and four-factor infection (n 5 3; error bars indicate
s.d.). j, Reprogramming efficiency of generating two-factor and four-factor
iPS cells (n 5 3). Indicated are the total numbers of GFP
1
colonies per
50,000 plated NSCs on day 7, 14 and 21 after infection.
NATURE | Vol 454 | 31 July 2008 LETTERS
647
©2008 Macmillan Publishers Limited. All rights reserved