Efficient enucleation of
in vitro from hematopoietic
stem and progenitor cells
Kenichi Miharada1,2, Takashi Hiroyama1,
Kazuhiro Sudo1, Toshiro Nagasawa2&
Erythroblast enucleation is thought to be largely dependent on
signals mediated by other cells, such as macrophages. In an
attempt to improve the in vitro production of red blood cells
(RBCs) from immature hematopoietic progenitor cells, we have
developed a method to produce enucleated RBCs efficiently in
the absence of feeder cells. Our method may represent an
efficient way to produce transfusable RBCs on a large scale
from hematopoietic progenitors.
The mechanism of erythroblast enucleation, one of the more critical
steps of RBC production, remains to be fully elucidated1,2. The role of
the interaction of erythroblasts with other cells such as macrophages is
a controversial topic in the study of erythroblast enucleation3–7. In
retinoblastoma-deficient (Rb?/?) embryos, macrophages are unable
to physically interact with erythroblasts and RBC production is
impaired6. In addition, in vitro production of enucleated RBCs
from immature hematopoietic progenitor cells proceeds efficiently
in the presence8but not in the absence9of feeder cells. However,
enucleation appears to initiate in vitro in the absence of feeder cells in
erythroblasts induced to differentiate in vivo to a developmental stage
that is competent for nuclear self-extrusion7,10.
Because the supply of transfusable RBCs for medical use is
insufficient in many countries, we have attempted to improve upon
existing methods for producing enucleated RBCs from human hema-
topoietic progenitor cells8,9. We used human CD34+cells comprising
hematopoietic stem and progenitor cells as the starting material8,9. We
developed a four-passage culture protocol. In the first step (passage I),
we cultured CD34+cells collected by magnetic cell sorting (MACS)
from human umbilical cord blood (approximate purity 95%) (Sup-
plementary Fig. 1 online) in erythroid differentiation medium (EDM;
Supplementary Methods online) in the presence either of stem cell
factor (SCF), erythropoietin (EPO) and interleukin-3 (IL-3) (protocol A)
or of SCF, EPO, IL-3, vascular endothelial growth factor (VEGF)
and insulin-like growth factor-II (IGF-II) (protocol B; Table 1). In
passages II and III, we cultured the cells in EDM in the presence of
SCF and EPO alone (Table 1).
We observed that a greater number of cells were generated by
protocol B than by protocol A after passage I (Table 1; Supplementary
Fig. 2 online). This result is consistent with the fact that the majority
of the human CD34+hematopoietic stem and progenitor cells used as
the starting material expressed VEGF receptor-2 (a functional receptor
for VEGF), at a low level, and also the IGF-I receptor (a functional
receptor for both IGF-I and IGF-II) (Fig. 1a). The relative increases in
cell numbers after passage III of protocol B were similar to earlier
results obtained by a method that avoided the use of feeder cells9
(Table 1; Supplementary Fig. 3 online).
Two lines of evidence suggested that protocol B led to efficient
induction of erythroid differentiation. First, although the majority of
cells produced by protocols A and B expressed both CD71 (the
transferrin receptor) and glycophorin-A (Gly-A, a specific marker of
mature erythroid cells) after passage II, CD71 expression was lower in
the cells cultured by protocol B (Fig. 1b). This implied that the cells
cultured by protocol B had differentiated to more mature stages.
As immature stem and progenitor cells mature into erythroid lineages,
their phenotype progresses from CD71–Gly-A–to CD71+Gly-A–
© 2006 Nature Publishing Group http://www.nature.com/naturebiotechnology
Table 1 Characteristics of cultured cells
Day 6Day 10 Day 16Day 20
% Ortho. cells
% Enucleated cells
133.8 ± 5.1
55.2 ± 3.3
29.3 ± 2.2
167.5 ± 14.9
64.5 ± 4.2
34.1 ± 3.6
61.8 ± 9.6
84.5 ± 3.2
42.2 ± 1.2
5.4 ± 2.3
1.6 ± 1.3
86.4 ± 10.1
86.2 ± 4.6
47.3 ± 4.6
11.5 ± 5.2
1.9 ± 1.7
40.4 ± 5.3
90.3 ± 5.3
73.9 ± 3.4
10.8 ± 4.4
6.2 ± 2.9
49.6 ± 7.4
87.1 ± 3.6
78.1 ± 2.0
20.6 ± 5.5
14.0 ± 2.3
1.1 ± 0.1
92.1 ± 4.8
94.6 ± 4.7
45.4 ± 4.2
50.0 ± 2.8
1.0 ± 0.1
93.6 ± 2.5
95.8 ± 3.2
21.2 ± 3.2
77.5 ± 2.8
Days 6, 10, 16 and 20 represent cells after passages I, II, III and IV, respectively. Stem cell factor (SCF), erythropoietin (EPO), interleukin-3 (IL-3), vascular endothelial growth factor (VEGF) and
insulin-like growth factor-II (IGF-II) were used as follows: in protocol A, SCF+EPO+IL-3 in passage I and SCF+EPO in passages II and III; in protocol B, SCF+EPO+IL-3+VEGF+IGF-II in passage I
and SCF+EPO in passages II and III. The culture method for passage IV was common to both protocols (see Supplementary Methods). Proliferation is defined as the relative increase in cell number
after each passage. Gly-A, glycophorin-A; Ortho. cells, orthochromatic erythroblasts. Values are mean ± s.d. (n ¼ 3).
Received 24 January; accepted 1 August; published online 17 September 2006; doi:10.1038/nbt1245
1Cell Engineering Division, RIKEN BioResource Center, Koyadai 3-1-1, Tsukuba, Ibaraki 305-0074, Japan.2Division of Hematology, Institute of Clinical Medicine,
University of Tsukuba, Ten-nodai 1-1-1, Tsukuba, Ibaraki 305-8577, Japan. Correspondence should be addressed to Y.N. (firstname.lastname@example.org).
ADVANCE ONLINE PUBLICATION1
B R I E F C O M M U N I C AT I O N S
to CD71+Gly-A+and finally to CD71–Gly-A+(Supplementary Fig. 4
online). Second, protocol B gave rise to a greater number of ortho-
chromatic erythroblasts (mature erythroblasts) after the completion of
passage III, and we observed a greater number of enucleated cells in the
cells produced by protocol B (Table 1; Supplementary Fig. 5 online).
In passage IV (the enucleation step), the cells expanded in passage
III were cultured in enucleation medium (Supplementary Methods)
for 4 d. To increase the stability of the RBCs, we added to the medium
mannitol, adenine and phosphate11(MAP; Supplementary Methods),
a mixture that is commonly used to preserve RBCs in Japan.
We observed many aggregated cells in cultures grown in the absence
but not in the presence of MAP (Supplementary Fig. 6 online),
suggesting that MAP functioned to prevent the loss of viability of
erythroblasts, cells undergoing enucleation and/or enucleated RBCs.
Glucocorticoids promote proliferation and inhibit differentiation of
erythroid progenitors12. Glucocorticoids are usually present to some
degree in serum, suggesting that the use of serum-free medium may be
preferable for inducing terminal differentiation of cultured cells. How-
ever, removal of serum from the medium may also impair cell viability.
In fact, serum was necessary in passage IV to maintain cell viability.
(Because we found that Plasmanate, a commercially available human
serum, could be successfully substituted for fetal bovine serum, we used
Plasmanate for this purpose.) Hence, we added mifepristone, an
antagonist of glucocorticoid function, to the medium in passage IV,
which accelerated the process of enucleation (Fig. 1c). Mifepristone was
more effective at inducing enucleation in the presence of a lower (0.5%)
rather than a higher (10%) concentration of Plasmanate (Fig. 1c),
strongly suggesting that the Plasmanate contained glucocorticoids.
With both protocols, the vast majority of cells produced after pass-
age IV expressed Rh-D antigen, a specific marker of erythroid pro-
genitors and terminally differentiated RBCs (Table 1; Supplementary
Fig. 7 online). Notably, protocol B produced a much larger proportion
of enucleated cells (nearly 80%) than did protocol A (Fig. 1d; Table 1).
Given that reticulocytes and even erythroblasts produced in vitro can
differentiate into fully mature RBCs in vivo8,9, we believe that mature
erythroid cells produced by protocol B are also likely to do so.
In contrast to previously published methods8,9, the method des-
cribed here included VEGF and IGF-II in the culture medium. These
two factors have been reported to promote the survival, proliferation
and/or differentiation of hematopoietic progenitors13–15. Consistent
with these findings, these factors promoted the expansion of erythroid
progenitors (Table 1). However, a much more important feature of our
culture system is that it allows erythroid cells to differentiate to a
developmental stage competent for nuclear self-extrusion. In fact, we
observed many enucleated cells after the completion of our culture
process (Fig. 1d; Table 1).
It has generally been thought that efficient enucleation of erythro-
blasts depends on cells in their local environment3–6. However, our
findings demonstrate that the interaction of erythroblasts with other
cells is not necessary and that signals mediated by humoral factors
seem to be sufficient for efficient autonomous completion of erythro-
Blood transfusion is indispensable for many clinical procedures,
yet the supply of transfusable materials is insufficient in many
countries. Because culture without feeder cells is technically easier
and less expensive than culture with feeder cells, our method may
represent an efficient way to produce transfusable RBCs on a large scale
from immature hematopoietic progenitors in umbilical cord blood.
Note: Supplementary information is available on the Nature Biotechnology website.
We thank the Stem Cell Resource Network for providing cord blood.
K.M. designed and performed research and wrote the manuscript; T.H. and K.S.
provided expertise; T.N. provided useful discussion; Y.N. coordinated research,
analyzed data and wrote the manuscript.
COMPETING INTERESTS STATEMENT
The authors declare that they have no competing financial interests.
Published online at http://www.nature.com/naturebiotechnology/
Reprints and permissions information is available online at http://npg.nature.com/
1. Lee, J.C. et al. Blood 103, 1912–1919 (2004).
2. Kingsley, P.D., Malik, J., Fantauzzo, K.A. & Palis, J. Blood 104, 19–25 (2004).
3. Ohneda, O. & Bautch, V.L. Br. J. Haematol. 98, 798–808 (1997).
4. Yanai, N., Sato, Y. & Obinata, M. Leukemia 11 (Suppl. 3), 484–485 (1997).
5. Hanspal, M., Smockova, Y. & Uong, Q. Blood 92, 2940–2950 (1998).
6. Iavarone, A. et al. Nature 432, 1040–1045 (2004).
7. Spike, B.T. et al. EMBO J. 23, 4319–4329 (2004).
8. Giarratana, M.-C. et al. Nat. Biotechnol. 23, 69–74 (2004).
9. Neildez-Nguyen, T.M. et al. Nat. Biotechnol. 20, 467–472 (2002).
10.Yoshida, H. et al. Nature 437, 754–758 (2005).
11.Hess, J.R. & Greenwalt, T.G. Transfus. Med. Rev. 16, 283–295 (2002).
12.Kolbus, A. et al. Blood 102, 3136–3146 (2003).
13.Gerber, H.P. & Ferrara, N. J. Mol. Med. 81, 20–31 (2003).
14.Zhang, C.C. & Lodish, H.F. Blood 103, 2513–2521 (2004).
15.Hiroyama, T. et al. Exp. Hematol. 34, 760–769 (2006).
© 2006 Nature Publishing Group http://www.nature.com/naturebiotechnology
Day 0Day 6
50 µm 50 µm
Figure 1 Production of erythroid cells from CD34+hematopoietic stem
and progenitor cells isolated from human umbilical cord blood. (a) Flow
cytometric analysis of IGF-I and VEGF receptor-2 expression. Cells were
stained with anti-IGF-I receptor and anti-VEGF receptor-2 monoclonal
antibodies and analyzed by flow cytometry. (a) Day 0, cells before culture.
Day 6, cells cultured by protocol B for 6 d (cells following passage I).
Results with isotype controls are also shown. (b) Flow cytometric analysis of
CD71 expression. Cells following passage II (day 10) were stained with anti-
CD71 monoclonal antibody. Solid line with gray shadow, staining with an
isotype control antibody. Solid lines with blue and red shadow, staining of
the cells cultured by protocol A and protocol B, respectively. (c) Effect
of mifepristone and Plasmanate on enucleation. Cells were cultured
by protocol B. The percentage of enucleated cells was calculated by the
formula 100 ? (number of enucleated cells/number of enucleated cells
plus nucleated cells). Values are mean ± s.d. (n ¼ 3). (d) Morphology
of the cells produced after passage IV. Enucleated RBCs were more
abundant in the cells cultured by protocol B: that is, nearly 80% of
RBCs were enucleated. Results shown in a,b,d are representative of
three independent experiments.
2ADVANCE ONLINE PUBLICATIONNATURE BIOTECHNOLOGY
B R I E F C O M M U N I C AT I O N S