Content uploaded by Fridtjof Lund-Johansen
Author content
All content in this area was uploaded by Fridtjof Lund-Johansen
Content may be subject to copyright.
Proc. Natl. Acad. Sci. USA
Vol. 94, pp. 12551–12556, November 1997
Immunology
Dendritic cell ontogeny: A human dendritic cell lineage of
myeloid origin
JOHANNA OLWEUS*†,ANDREW BITMANSOUR*, ROGER WARNKE‡,PETER A. THOMPSON*, JOSE CARBALLIDO§,
LOUIS J. PICKER¶,AND FRIDTJOF LUND-JOHANSEN§
*Becton Dickinson Immunocytometry Systems, 2350 Qume Drive, San Jose, CA 95131; ‡Department of Pathology, Stanford University, Stanford, CA 94305-5324;
§DNAX Research Institute, 901 California Avenue, Palo Alto, CA 94304; and ¶Southwestern Medical Center, University of Texas, 5323 Harry Hines Boulevard,
Dallas, TX 75235-9072
Edited by Irving L. Weissman, Stanford University School of Medicine, Stanford, CA, and approved September 10, 1997 (received for review April
16, 1997)
ABSTRACT Dendritic cells (DC) have been thought to
represent a family of closely related cells with similar func-
tions and developmental pathways. The best-characterized
precursors are the epidermal Langerhans cells, which migrate
to lymphoid organs and become activated DC in response to
inflammatory stimuli. Here, we demonstrate that a large
subset of DC in the T cell-dependent areas of human lymphoid
organs are nonactivated cells and belong to a separate lineage
that can be identified by high levels of the interleukin 3
receptor
a
chain (IL-3R
a
hi
). The CD34
1
IL-3R
a
hi
DC progen-
itors are of myeloid origin and are distinct from those that give
rise to Langerhans cells in vitro. The IL-3R
a
hi
DC furthermore
appear to migrate to lymphoid organs independently of in-
flammatory stimuli or foreign antigens. Thus, DC are heter-
ogeneous with regard to function and ontogeny.
Dendritic cells (DC) in lymphoid organs are potent antigen-
presenting cells, which play an important role in the initiation
of immune responses (1). Studies showing that epidermal
Langerhans cells are precursors of DC have suggested that the
unique role of DC as ‘‘natures adjuvant’’ is linked to their
developmental pathway. Langerhans cells reside in the epi-
dermis where the cells are capable of antigen uptake but have
low ability for antigen presentation (2). In response to inflam-
matory signals, the cells migrate rapidly to lymphoid tissues
and differentiate into mature, activated DC with potent ability
for stimulation of T cells (3–6). Cells with characteristics of DC
precursors have also been found in other tissues (7), and such
cells also migrate in response to inflammatory mediators (6, 8).
Thus, DC in lymphoid organs have been widely considered to
represent the end stage of a stepwise differentiation and
migration process, which is completed during inflammation
and serves to initiate immune responses (9–12).
Presently, most of the knowledge about the developmental
pathway of DC is based on results obtained by cell culture. Cells
with characteristics of Langerhans cells and DC can be generated
in vitro by culture of CD34
1
cells in the presence of granulocytey
macrophage colony-stimulating factor (GM-CSF) and tumor
necrosis factor
a
(TNF-
a
) (13–19). Results from studies of in vitro
colony formation have further indicated that the GM-CSFyTNF-
a
-responsive progenitors represent a separate DC colony forming
cell (15). These and other observations have supported the view
that DC are a family of closely related cells that constitute a
distinct ‘‘DC lineage.’’ However, progenitors committed to be-
come DC have not yet been identified directly in bone marrow.
The interpretation of results obtained by colony assays and cell
culture is furthermore complicated by the fact that populations of
lymphoid progenitors, granulomonoc ytic progenitors and periph-
eral blood monocytes also assume characteristics of DC in vitro
(14, 18–22).
Primitive hematopoietic progenitors and cells committed to
become lymphocytes, monocytes, granulocytes, and erythroid
cells can be identified as discrete populations of freshly isolated
CD34
1
bone marrow cells using specific cell surface markers
(23–29). Similar characterization of DC progenitors has been
difficult due to the lack of selective markers that identify the cells
at an early stage of differentiation. In the present study, however,
we demonstrate that antibodies to the interleukin 3 receptor
a
chain (IL-3R
a
) selectively react with a large subset of DC in
lymphoid organs and identify their precursors in blood and bone
marrow. The CD34
1
IL-3R
a
hi
progenitors are of myeloid origin
but committed to become DC and distinct from those that give
rise to Langerhans cells. Unlike Langerhans cells, IL-3R
a
hi
DC
home to lymphoid tissue independently of inflammation or
stimulation with foreign antigens.
MATERIALS AND METHODS
Tissue. Tissue from aborted fetuses of gestational age 19–21
weeks was obtained from Advanced Bioscience Resources
(Alameda, CA), a nonprofit organization which provides tissue
in compliance with state and federal laws. Blood donor buffy
coats were obtained from the Stanford Blood Bank (Stanford,
CA). Tonsils and adult lymph nodes were obtained from the
tissue acquisition service and the clinical flow cytometry
laboratory, Department of Pathology, University of Texas
Southwestern Medical Center (Dallas).
Cell Preparation. Mononuclear cell suspensions were ob-
tained by Lymphoprep gradient centrifugation (Nycomed,
Oslo). Antibody-free CD3
1
CD4
1
T cells were isolated to 99%
purity from peripheral blood mononuclear cells (PBMC) using
CD4 Dynabeads in combination with Detachabead reagent
(Dynal, Oslo) after depletion of myeloid cells (CD14
1
and
CD36
1
) by Dynabeads (Dynal). Fetal bone marrow CD34
1
cells were isolated by positive immunomagnetic selection
(Miltenyi Biotech, Auburn, CA), as described (29). Where
noted, subsets of immunostained cells were sorted using a
FACSVantage flow cytometer (Becton Dickinson).
Immunophenotypic Analysis. Multicolor immunofluorescence
staining and analysis was performed by standard methods (see
ref. 29). Primary or secondary antibodies were conjugated to
biotin, fluorescein isothiocyanate (FITC), phycoerythrin (PE),
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked ‘‘advertisement’’ in
accordance with 18 U.S.C. §1734 solely to indicate this fact.
© 1997 by The National Academy of Sciences 0027-8424y97y9412551-6$2.00y0
PNAS is available online at http:yywww.pnas.org.
This paper was submitted directly (Track II) to the Proceedings office.
Abbreviations: DC, dendritic cell; IL-3R
a
, interleukin 3 receptor
a
chain; GM-CSF, granulocyteymacrophage colony-stimulating factor;
G-CSF, granulocyte CSF; M-CSF, macrophage CSF; M-CSFR, M-
CSF receptor; TNF-
a
, tumor necrosis factor
a
; PBMC, peripheral
blood mononuclear cells; FITC, fluorescein isothiocyanate; PE, phy-
coerythrin; FACS, fluorescence-activated cell sorter.
†To whom reprint requests should be sent at the present address: J. O.,
Buskerud Central Hospital, Dronninggt. 28, N-3004 Drammen, Nor-
way. e-mail: f-johans@online.no.
12551
peridinin-chlorophyll protein, or allo-phycocyanin. Antibodies
and streptavidin conjugates were from Becton Dickinson, except
CD40, CD86 PE, anti-IL-3R
a
biotin, and anti-IL-3R
a
PE
(PharMingen); CD64 (Meda Rex, West Lebanon, NH); anti-M-
CSFR (Santa Cruz Biotechnology); donkey anti-rat IgG PE,
goat-anti-mouse IgG PE, and goat-anti-human IgM FITC (Jack-
son ImmunoResearch). Cells were analyzed using a FACSCali-
bur flow cytometer (Becton Dickinson).
Media and Cytokines. Except where specifically noted, cells
were cultured in Yssel’s medium (30) supplemented with
heat-inactivated 10% pooled human AB
1
serum and 10% fetal
FIG. 1. Antibodies to IL-3R
a
selectively stain DC in extrafollicular
regions of human tonsils. (A–C) Tonsillar mononuclear cells were
stained with anti-HLA-DR peridinin-chlorophyll protein (PerCP),
CD4 PE, anti-IL-3R
a
biotin 1Streptavidin allo-phycocyanin
(ALPC), and a mixture of FITC-conjugated lineage markers (‘‘lin’’)
for lymphocytes and monocytes (CD3, CD14, CD16, CD19, CD20,
Table 1. Expression of surface molecules on
HLA-DR
1
lin
2
IL-3R
a
hi
DC from mononuclear tonsillar cells,
PBMC, and fetal lymph node (LN) cells, and on CD34
1
IL-3R
a
hi
cells from fetal bone marrow (BM)
Tonsil Fetal LN PBMC Fetal BM
CD1a 22 2 2
CD3 22 2 2
CD4 11y111 11y111 11 11
CD5 22 2 2
CD11b 22 2 2
CD11c 22 2 2
CD13 2y11 2y11
CD14 22 2 2
CD15 2ND 22
CD16 22 2 2
CD19 22 2 2
CD20 22 2 2
CD32 11 1 1
CD33 11(1)1(1)1y11
CD34 2y11 2y1 11y111
CD36 11y111*11y111 11y111 11y1111
CD40 11 1 11 1
CD45RA 11(1)11(1)11(1)11
CD45RO 22 2 2
CD54 11y111 ND 11y111 1y11
CD56 22 2 2
CD58 1ND 1y11 1y11
CD62L 211y11 111
CD64 22 2 2
CD80 22 2 2
CD86 1ND 11(1)
HLA-DR 11 11 11 11
HLA-DQ 2ND 2y12
Cells were stained as described in legends to Figs. 1, 3, and 4. Mean
fluorescence intensity (MFI) levels for the IL-3R
a
hi
populations are
expressed as 2, indicating MFI in the first decade on a four log scale,
which corresponds to isotype control levels. The 1,11, and 111
indicate MFI in the second, third, and fourth decades, respectively.
Aysign means that MFI is on the border between two decades.
Parenthesis means that the MFI is in the upper end of a decade. ND,
not determined. The data are representative of at least three exper-
iments.
*In one out of three experiments tonsillar HLA-DR
1
lin
2
IL-3R
a
hi
DC
were negative for CD36.
CD56, and goat-anti-human IgM). The cells were analyzed by four-
color flow cytometry. Dendritic cells were identified as HLA-
DR
1
CD4
1
lin
2
—i.e., cells that simultaneously satisfy the criteria of
the box regions in Aand B, and are represented by large black dots.
Dashed lines represent isotype control levels. (D) Wright–Giemsa
staining of a cytocentrifuge slide of freshly FACS-sorted tonsillar
HLA-DR
1
lin
2
IL-3R
a
hi
cells. (3600.) (E) FACS-sorted tonsillar
HLA-DR
1
lin
2
IL-3R
a
hi
cells were cultured for 24 h with GM-CSF and
IL-3 and photographed in situ.(3400.) (F) A frozen section of tonsillar
tissue was stained with anti-IL-3R
a
, biotinylated anti-mouse IgG and
streptavidin peroxidase. (3100). Staining was visualized by diamino-
benzidine and hydrogen-peroxide, and the section was counterstained
with methylene blue.
12552 Immunology: Olweus et al. Proc. Natl. Acad. Sci. USA 94 (1997)
bovine serum. Recombinant human cytokines were used as
noted at the following concentrations: GM-CSF (10 ngyml),
IL-3 (10 ngyml), IL-6 (500 unitsyml), IL-7 (50 ngyml) (all from
Collaborative Biomedical Products, Bedford, MA), erythro-
poietin (2.5 unitsyml; CILAG, Schaffhausen, Switzerland),
stem cell factor (40 ngyml; Peprotech, Rocky Hill, NJ), gran-
ulocyte-CSF (G-CSF, 50 ngyml, Amgen Biologicals), and
macrophage-CSF (M-CSF, 10 ngyml;R&DSystems).
Cell Cultures. IL-3R
a
hi
lin
2
HLA-DR
1
cells from tonsil and
PBMC and CD14
hi
monocytes from PBMC were sorted by
fluorescence-activated cell sorter (FACS) and cultured for 36 h
before mixing with T cells to allow maturation of DC precursors.
GM-CSF and IL-3 were added to IL-3R
a
hi
lin
2
HLA-DR
1
cells to
enhance survival. Stimulator cells were washed twice before
coculture with CD4
1
T cells (10
5
per well) in flat-bottomed
96-well plates for 6 days. Bromodeoxyuridine (BrdU; Sigma) (50
m
M) was added 12 h before harvest latex particles (10
5
per well)
were added immediately prior to harvest as a reference for cell
counts (29). The cells were stained with CD3 PE, followed by
fixation, permeabilization, staining with anti-BrdU FITC, and
flow cytometric analysis, as described (29). Cocultures with
stimulator cells derived by culture of fetal bone marrow cells were
pulsed with 1
m
Ci (1 Ci 537 GBq) [
3
H]thymidine for 8 h before
collecting and counting.
RESULTS AND DISCUSSION
IL-3R
a
Is a Selective Marker for a Large Subset of DC in
T Cell-Rich Zones of Human Peripheral Lymphoid Organs.
Human tonsils contain DC that can be identified as cells that
lack lineage markers for monocytes and lymphocytes (lin
2
)
and are positive for HLA-DR and CD4 (HLA-DR
1
CD4
1
lin
2
)
(Fig. 1 Aand B) (31, 32). In an attempt to identify DC-selective
markers, we screened antibodies to leukocyte differentiation
antigens for selective reactivity with this population. Antibod-
ies to IL-3R
a
reacted strongly with more than 85% of HLA-
DR
1
CD4
1
lin
2
cells (0.32–0.37% of tonsillar mononuclear
cells, n53), but weakly with most other cells (n53) (Fig. 1C).
The staining was sufficiently specific to allow a 200-fold
enrichment of HLA-DR
1
CD4
1
lin
2
cells with a single positive
immuno-magnetic selection (n52). HLA-DR
1
lin
2
IL-3R
a
hi
cells were also found in adult cervical, axillar, intramammary,
mesenteric, and femoral lymph nodes. The frequency ranged
from 0.1%–1.8% (average 0.5%, n59). In all cases, cells
staining brightly with the anti-IL-3R
a
were found within the
HLA-DR
1
lin
2
population and constituted the majority of
these cells (on average 60.2 611.2%, n59).
Freshly sorted HLA-DR
1
lin
2
IL-3R
a
hi
cells showed an imma-
ture morphology without cytoplasmic protrusions (Fig. 1D). The
majority of the cells died rapidly in culture, but could be partially
rescued by the addition of cytokines (GM-CSF and IL-3). Under
these conditions, the cells rapidly formed large aggregates of cells
as previously reported for tonsillar DC (32) (Fig. 1E). After 3–5
days the cells were more dispersed and showed multiple long
processes characteristic of DC (data not shown).
IL-3R
a
hi
cells were found almost exclusively in the T
cell-rich extra-follicular regions of the tonsil (Fig. 1F). The
localization as well as the the cytokine requirements and
phenotype of the HLA-DR
1
lin
2
IL-3R
a
hi
cells (Table 1) sug-
gest that they are identical to the ‘‘plasmacytoid T cell’’ DC
that were recently characterized by Grouard et al. (33).
IL-3R
a
hi
DC Are Immature and Appear in Lymphoid Organs
Independently of Stimuli That Cause Up-Regulation of Major
Histocompatibility Complex (MHC) Class II and Costimulatory
Molecules. Originally, it was concluded that DC in lymphoid
tissues such as tonsils are activated, mature antigen-presenting
cells, because isolated cells expressed high levels of MHC class II
and costimulatory molecules (31, 32). However, in those studies
the cells were isolated after 1–2 days of cell culture. As shown in
Fig. 2Aand Table 1, HLA-DR
1
lin
2
IL-3R
a
hi
cells in fresh
preparations of tonsillar mononuclear cells expressed low levels
of CD80 (B7.1), CD86 (B7.2), and HLA-DQ. Overnight culture
of unseparated cells in the absence of cytokines was sufficient to
induce the mature phenotype (Fig. 2A). Sorted HLA-
DR
1
lin
2
IL-3R
a
hi
cells that were allowed to mature in culture for
36 h and kept viable with GM-CSF and IL-3 were potent
stimulators of allogeneic CD4
1
T cells (Fig. 2B). These results are
surprising in view of the hypothesis that DC migrate to lymphoid
organs in response to signals that lead to cell activation. However,
these data are in agreement with those of Grouard et al. (33), who
also studied freshly isolated cells, and with previous reports
FIG. 2. IL-3R
a
hi
DC are immature and appear in lymphoid organs
independently of stimuli that cause up-regulation of major histocompat-
ibility complex class II and costimulatory molecules. (A) Mononuclear
cells from tonsil were stained with anti-HLA-DR, anti-IL-3R
a
and
lineage markers, and either CD86, CD80, HLA-DQ, or isotype control
mAbs before (open bars) and after (filled bars) a 16-h incubation at 37°C
in Yssel’s medium (30). The HLA-DR
1
lin
2
IL-3R
a
hi
population was
analyzed for mean fluorescence intensit y (MFI) staining with the markers
indicated on the figure, and the bars represent MFI after isotype control
levels were subtracted. (B) T cells (10
5
) were cocultured with indicated
numbers of allogeneic IL-3R
a
hi
lin
2
HLA-DR
1
cells from tonsil (stimu-
lator cells). T cell proliferation was measured as the total number of CD3
1
BrdU
1
cells per well at day 6 of coculture. (C) Fetal ly mph node cells were
stained with anti-HLA-DR, anti-IL-3R
a
, and lineage markers (data not
shown) and analyzed by flow cytometry as described in Fig. 1 A–C. Data
are representative of three experiments.
FIG. 3. HLA-DR
1
lin
2
IL-3R
a
hi
DC are present in peripheral blood. (Aand B) PBMC were stained with anti-HLA-DR, anti-IL-3R
a
, and lineage
markers (not shown) and analyzed by flow cytometry as described in Fig. 1 A–C. R1 and R2 in Brepresent regions used to sort IL-3R
a
hi
blood cells that
were positive or negative for HLA-DR, respectively. (Cand D) HLA-DR
1
lin
2
IL-3R
a
hi
cells sorted from blood according to R1 in Bwere first cultured
separately with IL-3 and GM-CSF for 36 h and then incubated with allogeneic (C) or autologous (D) CD4
1
T cells. T cell proliferation was measured
as total number of CD3
1
BrdU
1
cells per well at day 6 of coculture with indicated numbers of IL-3R
a
hi
lin
2
HLA-DR
1
cells (
■
) or CD14
hi
monocytes
(
E
) from the same donor (stimulator cells). Individual displays show data that are representative of three experiments. OLS, ortogonal light scatter.
Immunology: Olweus et al. Proc. Natl. Acad. Sci. USA 94 (1997) 12553
FIG. 4. Proliferating progenitors for IL-3R
a
hi
DC are found as a discrete CD34
1
IL-3R
a
hi
population that is distinct from the cells that give
rise to Langerhans cells in response to GM-CSF and TNF-
a
.(Aand B) Isolated CD34
1
fetal bone marrow cells were stained with CD34 and
anti-IL-3R
a
. The cells were analyzed by flow cytometry as described in Fig. 1 A–C. The CD34
1
IL-3R
a
hi
population was defined according to the
region in A(blue dots). CD34
1
IL-3R
a
lo
cells were defined according to the region in B(red dots). OLS, ortogonal light scatter. (C) Freshly sorted
Wright–Giemsa-stained CD34
1
IL-3R
a
hi
cells display mitotic figures. (3600.) (D–G) CD34
1
IL-3R
a
hi
cells (blue) and CD34
1
IL-3R
a
lo
cells (red)
were sorted according to the regions in Aand B, respectively, and cultured in the presence of indicated cytokines, stained with CD1a and CD45RA
after 5 days of culture, and analyzed by FACS. L 5B lymphoid cells staining brightly with CD19 (data not shown). None of the cultured
CD34
1
IL-3R
a
hi
cells (Dand E) stained positively with CD19 (data not shown). Dashed lines indicate isotype control levels. (H) Transmission
electron microscograph (35,000) of a CD1a
1
cell sorted from CD34
1
IL-3R
a
lo
cells cultured with GM-CSF and TNF-
a
, as described in G. The
arrows point to areas containing Birbeck granules, (shown in Insets,345,000). (I) CD4
1
T cells (10
5
) were cocultured with indicated numbers of
12554 Immunology: Olweus et al. Proc. Natl. Acad. Sci. USA 94 (1997)
showing that DC in murine spleens were immature immediately
after isolation and differentiated rapidly in vitro (34, 35).
The nonactivated phenotype of DC in tonsils may seem like a
paradox since tonsils are typically removed after repeated in-
flammations. One can further not exclude the possibility that the
isolated cells had migrated to lymphoid tissue in response to
inflammation, but not yet assumed the activated phenotype. To
determine whether the presence of HLA-DR
1
lin
2
IL-3R
a
hi
cells
in lymphoid tissue depends on previous exposure to foreign
antigens or inflammatory stimuli, we examined whether the cells
could be found in fetal lymph nodes. Presumably, these lymph
nodes drain sterile, noninflamed tissues. As shown in Fig. 2C,
HLA-DR
1
lin
2
IL-3R
a
hi
cells were present at high frequencies in
the fetal lymph nodes (2.6 61.1%, n53). These cells were more
frequent than cells expressing high levels of myeloid markers,
such as CD13 and CD33 (1.5 60.2%, n53). The HLA-
DR
1
lin
2
IL-3R
a
hi
cells in fetal lymph nodes expressed the same
combination of markers as those in tonsils (Table 1). This cell type
is therefore most likely capable of migrating to lymphoid tissue
independently of inflammation and foreign antigens and without
initiating immune responses.
IL-3R
a
hi
DC Are Present in Peripheral Blood. HLA-
DR
1
lin
2
IL-3R
a
hi
cells were readily detectable as a population
with low orthogonal light scatter in blood from adult donors, and
constituted 0.47 60.14% (n58) of PBMC (Fig. 3 Aand B, R1).
A second population of IL-3R
a
hi
cells was present among PBMC,
but these cells were HLA-DR
2
(Fig. 3B, R2) and were found to
be basophilic granulocytes (data not shown). The HLA-
DR
1
lin
2
IL-3R
a
hi
blood cells expressed the same combination of
markers and had similar morphology as shown for the tonsillar
counterparts (Table 1 and data not shown). However, an inter-
esting difference was that whereas tonsillar DC did not express
the lymph node homing molecule L-selectin (CD62L), the cells
from PBMC were positive for this marker (Table 1).
Whereas multiple cell types can stimulate allogeneic T cells,
DC are characterized by their higher potency relative to other
antigen-presenting cells (13, 15, 21, 31, 33, 36–39). The cells
also induce proliferation of autologous T cells in vitro (37). The
data in Fig. 3 Cand Ddemonstrate that HLA-DR
1
lin
2
IL-
3R
a
hi
blood cells were up to 100-fold more potent than
monocytes in stimulating both allogeneic and autologous T
cells. Cells with similar characteristics have previously been
identified in blood and were referred to as CD11c
2
DC or
CD33
dim
CD14
2
CD16
2
DC (36, 37).
Precursors of IL-3R
a
hi
DC Are Found Among CD34
1
Bone
Marrow Cells and Are Distinct from the Cells That Give Rise to
Langerhans Cells in Response to GM-CSF and TNF-
a
.IL-3R
a
hi
cells were readily identified as a distinct subset of CD34
1
fetal
bone marrow cells with low orthogonal light scatter (3.1 60.9%,
n55) (Fig. 4 Aand B). The majority of the CD34
1
IL-3R
a
hi
cells
were within the CD34
lo
subset, which contains lineage-committed
progenitors (Fig. 4B) (23). The CD34
1
IL-3R
a
hi
cells had mor-
phology and immunophenotype similar to IL-3R
a
hi
lin
2
HLA-
DR
1
cells in blood and tonsil, but appeared more immature, and
mitotic figures were frequently observed (Fig. 4Cand Table 1).
Previous studies have demonstrated that DC and Langerhans
cells can be derived by culture of CD34
1
cells with GM-CSF and
TNF-
a
(13–19). The subset of CD34
1
cells that contains these
TNF-
a
-dependent progenitors has not yet been identified. We
therefore examined whether they were identical to the CD34
1
IL-
3R
a
hi
cells. The sorted IL-3R
a
hi
population (Fig. 4A) formed
aggregates of cells with DC morphology during culture with
either GM-CSF and IL-3 or GM-CSF and TNF-
a
(data not
shown), and after 5 days the cells were positive for CD1a (Fig. 4
Dand E). However, differentiation into DC occurred indepen-
dently of TNF-
a
, and the CD1a
1
cells co-expressed CD45RA,
which is absent from Langerhans cells (39) (Fig. 4 Dand E).
Tonsillar HLA-DR
1
lin
2
IL-3R
a
hi
cells also obtained a
CD1a
1
CD45RA
1
phenotype when cultured under the same
conditions (data not shown). In contrast, sorted CD34
1
cells with
low levels of the IL-3R
a
(Fig. 4B) gave rise to few CD1a
1
cells
when cultured with GM-CSF and IL-3, but large numbers of
CD45RA
2
CD1a
1
cells in the presence of GM-CSF and TNF-
a
(Fig. 4 Fand G). Consistent with a phenotype of Langerhans
cells, 30–40% of these CD45R A
2
CD1a
1
cells contained Birbeck
granules (Fig. 4H). The frequency of Birbeck granule-positive
cells among CD1a
1
cultured CD34
1
IL-3R
a
hi
cells was less than
4%, and may therefore reflect contamination of sorting gates.
Finally, CD34
1
IL-3R
a
hi
cells cultured with GM-CSF and IL-3
were potent stimulators of allogeneic CD4
1
T cells, and more
potent than CD14
1
macrophages generated by culture of
CD34
1
M-CSFR
1
cells from the same donor in M-CSF (Fig. 4I).
IL-3R
a
hi
DC Are of Myeloid Origin. The distribution of
M-CSF receptor (M-CSFR) and IL-3R
a
among CD34
1
cells
suggested that CD34
1
IL-3R
a
hi
cells may derive from cells in the
M-CSFR
hi
population, which down-regulate the M-CSFR as they
up-regulate the IL-3R
a
(arrow in Fig. 5A). Because M-CSFR
expression on CD34
1
cells is restricted to granulomonocytic
progenitors, this would indicate that IL-3R
a
hi
cells belong to the
granulomonocytic lineage (29). To test this possibility, immature
progenitors (i.e., CD34
hi
, see Fig. 4B) with high levels of M-CSFR
and low levels of IL-3R
a
(region in Fig. 5A) were sorted and
cultured. After 60 h of culture two populations of cells that had
downmodulated the M-CSFR were observed, with high and low
levels of IL-3R
a
, respectively (Fig. 5B). These cells were sorted
and cultured for 5 additional days in medium supplemented with
GM-CSF and IL-3. At this stage, the cultures from the IL-3R
a
lo
population contained CD15
1
granulocytic cells (Fig. 5C) whereas
cultures from the IL-3R
a
hi
population contained CD1a
1
cells
with DC morphology (Fig. 5Dand data not shown). The latter
cells induced strong proliferation of allogeneic T cells compared
with equal numbers of CD14
1
macrophages generated by culture
of CD34
1
M-CSFR
1
cells for 5 days with M-CSF (29) (n52, data
not shown). Addition of TNF-
a
to the medium did not increase
the number of CD34
1
IL-3R
a
hi
cells generated from M-CSFR
hi
myeloid progenitors, further suggesting that the cells are distinct
from the TNF-
a
-dependent Langerhans cell progenitors (data
not shown).
Thymus Contains Small Numbers of IL-3R
a
hi
DC That
Express Low Levels of CD34. Earlier studies have demon-
strated that some DC may share the differentiation pathway of
the T cell lineage and that DC are generated by progenitors in
the thymus (18, 19, 40). We therefore investigated whether
IL-3R
a
hi
DC were present in the thymus. The results showed
that less than 0.1% of the total thymic cell population were
IL-3R
a
hi
DC and that the IL-3R
a
hi
cells constituted only 15 6
2% (n53) of HLA-DR
1
lin
2
subset. This may indicate that
other DC populations than the IL-3R
a
hi
cells are predominant
in thymus. In addition, the IL-3R
a
hi
DC in thymus expressed
low levels of CD34 compared with bone marrow IL-3R
a
hi
cells
from the same donor (data not shown). Similar low levels of
CD34 were found on the cells in fetal lymph nodes (Table 1).
It therefore seems likely that the few IL-3R
a
hi
DC in the
thymus represent cells that have migrated from bone marrow
and are distinct from thymic DC progenitors.
IL-3R
a
hi
DC Constitute a Separate Lineage of DC That Follow
a Differentiation Pathway Distinct from Langerhans Cells. The
data presented here demonstrate that a subset of DC in human
allogeneic DC (
■
) or macrophages (
E
) from the same donor. The DC were generated by culturing sorted CD34
1
IL-3R
a
hi
cells for 5 days with
GM-CSF and IL-3 (
■
). The macrophages were generated by culture of CD34
1
M-CSFR
1
cells for 5 days with M-CSF and purified by FACS sorting
of CD14
1
cells. T cell proliferation was measured as incorporation of [
3
H]thymidine at day 6 of coculture. c.p.m; counts per minute. Data are
representative of three experiments.
Immunology: Olweus et al. Proc. Natl. Acad. Sci. USA 94 (1997) 12555
lymphoid organs and blood represent a separate lineage of cells.
The cells can be readily identified and isolated on the basis of their
high levels of the IL-3R
a
. Their precursors in the bone marrow
appear to follow the myeloid differentiation pathway to the
branching point of the granulocytic and monocytic lineages.
Progenitors that have committed to this DC lineage are distinct
from those that give rise to Langerhans cells when cultured in the
presence of GM-CSF and TNF-
a
. Their progeny furthermore
lack several characteristics of Langerhans cells that were previ-
ously considered common to DC. Unlike Langerhans cells, the
IL-3R
a
hi
DC appear to undergo little differentiation during
transit from the bone marrow to lymphoid organs. The IL-3R
a
hi
DC further seem capable of migrating to lymphoid tissue inde-
pendently of inflammatory stimuli. The results concur with
previous reports showing that some DC spend less than 24 h in
transit from the progenitor pool to lymphoid tissue and that DC
turnover occurs constantly during steady state conditions (1,
41–43). Thus, the DC system is constituted by multiple cell types
with distinct developmental pathways and functional properties.
A large proportion of the cells may enter lymphoid tissue without
inducing immune responses.
We thank Drs. Lewis Lan ier, Anne O’Garra, Rene de Waal Malefyt,
and Gosse Adema for critical reading of the manuscript; David Houck
for expert assistance with cell sorting; and Barny Abrams, Dr. Qudrat
Nasraty, and Dr. Kenneth A. Davis for conjugation of antibodies. J.O.
and F.L.-J. were supported by grants from the Norwegian Research
Counsel. DNAX Research Institute is supported by Schering-Plough.
1. Steinman, R. M. (1991) Annu. Rev. Immunol. 9, 271–296.
2. Schuler, G. & Steinman, R. (1985) J. Exp. Med. 161, 526–546.
3. Silberberg-Sinakin, I., Thorbecke, J., Baer, R., Rosenthal, S. &
Berezowsky, V. (1976) Cell. Immunol. 25, 137–151.
4. Macatonia, S., Knight, S., Edwards, A., Griffiths, S. & Fryer, P.
(1987) J. Exp. Med. 166, 1654–1667.
5. Kripke, M., Munn, C., Jeevan, A., Tang, J. & Bucana, C. (1990)
J. Immunol. 145, 2833–2838.
6. Roake, J. A., Rao, A. S., Morris, P. J., Larsen, C. P., Hankins,
D. F. & Austyn, J. M. (1995) J. Exp. Med. 181, 2237–2247.
7. Hart, D. & Fabre, J. (1981) J. Exp. Med. 153, 347–361.
8. MacPherson, G. G., Jenkins, C. D., Stein, M. J. & Edwards, C.
(1995) J. Immunol. 154, 1317–1322.
9. Fearon, D. & Locksley, R. (1996) Science 272, 50–52.
10. Ibrahim, M. A., Chain, B. M. & Katz, D. R. (1995) Immunol.
Today 16, 181–186.
11. Austyn, J. M. (1996) J. Exp. Med. 183, 1287–1292.
12. Lanzavecchia, A. (1996) Curr. Opin. Immunol. 8, 348–354.
13. Caux, C., Dezutter-Dambuyant, D. C., Schmitt, D. & Banche-
reau, J. (1992) Nature (London) 360, 258–261.
14. Reid, C. D., Stackpoole, A., Meager, A. & Tikerpae, J. (1992)
J. Immunol. 149, 2681–2688.
15. Young, J. W., Szabolcs, P. & Moore, M. A. (1995) J. Exp. Med.
182, 1111–1119.
16. Caux, C., Vanbervliet, B., Massacrier, C., Durand, I. & Banche-
reau, J. (1996) Blood 87, 2376–2385.
17. Caux, C., Vanbervliet, B., Massacrier, C., Dezutter-Dambuyant,
D. C., de Saint-Vis, B., Jacquet, C., Yoneda, K., Imamura, S.,
Schmitt, D. & Banchereau, J. (1996) J. Exp. Med. 184, 695–706.
18. Galy, A., Travis, M., Cen, D. & Chen, B. (1995) Immunity 3,
459– 473.
19. Res, P., Martinez, C. E., Jaleco, C. A., Staal, F., Noteboom, E.,
Weijer, K. & Spits, H. (1996) Blood 87, 5196–5206.
20. Inaba, K., Inaba, M., Deguchi, M., Hagi, K., Yasumizu, R.,
Ikehara, S., Muramatsu, S. & Steinman, R. M. (1993) Proc. Natl.
Acad. Sci. USA 90, 3038–3042.
21. Zhou, L. J. & Tedder, T. F. (1996) Proc. Natl. Acad. Sci. USA 93,
2588–2592.
22. Peters, J., Gieseler, R., Thiele, B. & Steinbach, F. (1996) Immu-
nol. Today 17, 273–278.
23. Terstappen, L. W., Huang, S., Safford, M., Lansdorp, P. M. &
Loken, M. R. (1991) Blood 77, 1218–1227.
24. Baum, C. M., Weissman, I. L., Tsukamoto, A. S., Buckle, A. M.
& Peault, B. (1992) Proc. Natl. Acad. Sci. USA 89, 2804–2808.
25. Lansdorp, P. M. & Dragowska, W. (1992) J. Exp. Med. 175,
1501–1509.
26. Terstappen, L. W. M. M., Huang, S. & Picker, L. J. (1992) Blood
79, 666– 677.
27. Olweus, J., Lund-Johansen, F. & Terstappen, L. W. M. M. (1994)
Blood 85, 2402–2413.
28. Olweus, J., Terstappen, L., Thompson, P. & Lund-Johansen, F.
(1996) Blood 88, 1594–1607.
29. Olweus, J., Thompson, P. & Lund-Johansen, F. (1996) Blood 88,
3741–3754.
30. Yssel, H., de Vries, J., Koken, M., Van Blitterswijk, W. & Spits,
H. (1984) J. Immunol. Methods 72, 219–224.
31. Hart, D. N. J. & McKenzie, J. L. (1988) J. Exp. Med. 168, 157–170.
32. Cameron, P. U., Lowe, M. G., Sotzik, F., Coughlan, A. F., Crowe,
S. M. & Shortman, K. (1996) J. Exp. Med. 183, 1851–1856.
33. Grouard, G., Rissoan, M., Filgueira, L., Durand, I., Banchereau,
J. & Liu, J. (1997) J. Exp. Med. 185, 1101–1112.
34. Crowley, M., Inaba, K., Witmer-Pack, M., Gezelter, S. & Stein-
man, R. (1990) J. Immunol. Methods 133, 55–66.
35. Inaba, K., Witmer-Pack, M., Inaba, M., Hathcock, K. S., Sakuta,
H., Azuma, M., Yagita, H., Okumura, K., Linsley, P. S., Ikehara,
S., Muramatsu, S., Hodes, R. J. & Steinman, R. M. (1994) J. Exp.
Med. 180, 1849–1860.
36. O’Doherty, U., Peng, M., Gezelter, S., Swiggard, W. J., Betjes, M.,
Bhardwaj, N. & Steinman, R. M. (1994) Immunology 82, 487– 493.
37. Thomas, R. & Lipsky, P. E. (1994) J. Immunol. 153, 4016–4028.
38. Grouard, G., Durand, I., Filgueira, L., Banchereau, J. & Liu, Y.
(1996) Nature (London) 384, 364–367.
39. Pope, M., Betjes, M. G., Hirmand, H., Hoffman, L. & Steinman,
R. M. (1995) J. Invest. Dermatol. 104, 11–17.
40. Ardavin, C., Wu, L., Li, C. L. & Shortman, K. (1993) Nature
(London) 362, 761–763.
41. Pugh, C., MacPherson, G. & Steer, H. (1983) J. Exp. Med. 157,
1758–1779.
42. Fossum, S. (1989) Curr. Top. Pathol. 79, 101–124.
43. Matsuno, K., Ezaki, T., Kudo, S. & Uehara, Y. (1996) J. Exp. Med.
183, 1865–1878.
FIG. 5. IL-3R
a
hi
DC follow the myeloid differentiation pathway to the
branching point of the granulocytic and monocytic lineages. (A) CD34
1
cells were incubated for 12 h in serum-free medium (25) to allow
up-regulation of the M-CSFR, and stained with CD34, anti-M-CSFR, and
anti-IL-3R
a
. The arrow shows the suggested differentiation pathway of
CD34
1
IL-3R
a
hi
cells. The region shows sorting criteria for M-CSFR
hi
y
IL-3R
a
lo
cells. An additional gate was set to include only CD34
hi
cells,
shown in Fig. 4B, to restrict the sort to immature progenitors (29). (B)
After a 60-h culture of CD34
hi
M-CSFR
1
cells in serum-free medium (25)
containing stem cell factor, G-CSF, GM-CSF, IL-3, and IL-6, the cells
were stained with anti-M-CSFR and anti-IL-3R
a
. The regions indicate
criteria for sorting of the two populations that had downmodulated the
M-CSFR during the culture period. (Cand D) After a 5-day secondary
culture of IL-3R
a
lo
M-CSFR
lo
cells (green) and IL-3R
a
hi
M-CSFR
lo
cells
(blue) with IL-3 and GM-CSF, the cells were stained with CD1a and
CD15 and analyzed by FACS. The small subset of CD15
lo
cells in C
represent basophilic granulocytes (27, 29).
12556 Immunology: Olweus et al. Proc. Natl. Acad. Sci. USA 94 (1997)