Content uploaded by Maria Rescigno
Author content
All content in this area was uploaded by Maria Rescigno
Content may be subject to copyright.
nature immunology • volume 2 no 9 • september 2001 • http://immunol.nature.com
ARTICLES
882
Francesca Granucci,Caterina Vizzardelli,Norman Pavelka,Sonia Feau,Maria Persico, Ettore
Virzi,Maria Rescigno,Giorgio Moro and Paola Ricciardi-Castagnoli
Dendritic cells (DCs) are strong activators of primary T cell responses. Their priming ability is
acquired upon encounter with maturation stimuli. To identify the genes that are differentially
expressed upon maturation induced by exposure to Gram-negative bacteria, a kinetic study of DC
gene expression was done with microarrays representing 11,000 genes and ESTs (expressed
sequence tags). Approximately 3000 differentially expressed transcripts were identified. We found
that functional interleukin 2 (IL-2) mRNA, which gave rise to IL-2 production, was transiently up-
regulated at early time-points after bacterial encounter. In contrast, macrophages did not produce
IL-2 upon bacterial stimulation.Thus, IL-2 is an additional key cytokine that confers unique T cell
stimulatory capacity to DCs.
University of Milano-Bicocca, Department of Biotechnology and Bioscience,Piazza della Scienza 2,Milan,Italy.
Correspondence should be addressed to P.R.C.(paola.castagnoli@unimib.it)
Inducible IL-2 production by
dendritic cells revealed by global
gene expression analysis
Dendritic cells (DCs) are professional antigen-presenting cells that
are able to initiate adaptive immune responses to invading pathogens.
Key functions of DCs include uptake and processing of antigen and
priming of naïve T cells
1
, functions that are segregated in time.
Immature resting DCs located in nonlymphoid tissues, such as skin
and mucosae, take up antigen. Mature DCs loaded with antigen
1
and
capable of priming T cells migrate from nonlymphoid tissues to the T
cell area of lymph nodes or spleen. Thus, when immature DCs come
into contact with inflammatory stimuli, they undergo a maturation
process that transforms them from phagocytic and migratory cells to
nonphagocytic, highly efficient stimulators of naïve T cell responses
2
.
Mature DCs are programmed to undergo apoptotic death within 9–10
days
3
. Immature monocyte-derived human DCs (hMDCs) or imma-
ture bone marrow–derived mouse DCs (mBMDCs) can be induced to
mature in vitro with many different stimuli, including inflammatory
cytokines; bacteria cell products, such as lipopolysaccharide (LPS)
and lipoteichoic acid (LTA); bacterial DNA; double-stranded viral
RNA; and live bacteria. This last stimulus represents one of the most
potent catalysts of the DC terminal differentiation process in the
mouse: it induces a rapid and effective DC phenotypic and function-
al maturation
4
. A key part of the complexity of innate and adaptive
immunity to microorganisms relies on how they affect DC activation
5
.
Figure 1. Developmentally synchronized DCs. (a) DCs were
incubated with E. coli and activation tested by measuring surface
expression of CD40, B7-2 and MHC class II at the indicated time-
points by flow cytometry. RNA was extracted from synchronized
cells. (b) Example of gene expression comparison in duplicate
experiments (1 and 2 at time 0): 3340 genes present in the dupli-
cates were compared.The R
2
of duplicate linear regressions for each
time-point was never lower than 0.9 (data not shown). (c) Gene
expression comparison at two time-points (time 0 and time 6 of
experiment 1), excluding genes that were consistently absent in
both the analyses.R
2
of comparison analysis linear regressions usu-
ally ranging between 0.7–0.8 (data not shown).The degree of gene
expression is indicated by the average difference (Avg.Diff.) value.
a b
c
© 2001 Nature Publishing Group http://immunol.nature.com
© 2001 Nature Publishing Group http://immunol.nature.com
ARTICLES
http://immunol.nature.com • september 2001 • volume 2 no 9 • nature immunology
883
To study how DCs are conditioned by bacterial encounter, we car-
ried out a kinetic analysis of gene expression in immature mouse DCs
stimulated at different time-points with live Gram-negative bacteria.
Many genes, which were differentially expressed during maturation,
encoded transcription factors, signal transduction molecules and pro-
teins involved in cytoskeletal rearrangements or in the activation and
control of immune responses. We also observed induction of inter-
leukin 2 (IL-2) mRNA at early time-points after bacterial encounter
and induction of IL-2 protein in culture supernatants. To test whether
this cytokine could have a role in activating T cell responses, we com-
pared the ability of wild-type and IL-2
–/–
DCs to stimulate T cells in
primary mixed lymphocyte reactions (MLRs). The ability of IL-2
–/–
DCs to induce both CD4
+
and CD8
+
T cell proliferation was severely
impaired, which indicated that IL-2 is a key cytokine in conferring the
unique T cell stimulatory capacity to DCs. Bone marrow–derived
macrophages, activated with bacteria, did not produce any IL-2.
Results
Analysis of developmentally synchronized DCs
Transcription analysis requires homogeneous populations to avoid dilu-
tion and contamination of data. Due to their plasticity, mBMDCs are
extremely heterogeneous, and it is not feasible to obtain homogeneous
immature cells without contamination with mature and intermediate
DCs
6,7
. There is a DC culture system that allows the propagation of
homogeneous immature growth factor–dependent (granulocyte mono-
cyte–colony-stimulating factor, GM-CSF) mouse DCs that can fully
mature in response to bacteria, bacterial cell products or inflammatory
cytokines
3
, mimicking the in vivo DC maturation process. Studies of a
well characterized DC line, D1, done with this culture system show
similar maturation to that seen with fresh splenic or bone
marrow–derived DCs
8–11
.
We activated D1 cells with Gram-negative Escherichia coli and
transcriptionally analyzed immature cells as well as mature cells that
had been stimulated for 4, 6, 12, 18, 24 or 48 h with high-density
oligonucleotide arrays that displayed probes for 11,000 genes and
expressed sequence tags (ESTs). At each time-point after stimula-
tion, D1 cells were phenotypically characterized for their state of
developmental synchronization by analyzing surface expression of
major histocompatibility complex (MHC) class II, B7-2 and CD40
(Fig. 1a). Biotin-labeled cRNA was generated and hybridized onto
Table 1.Results of PCA on the entire dataset
Independent variables
a
PC1 PC2 PC3 PC4
(n=1) (n=2) (n=3) (n=4)
0 h (1) -0.26 0.53 -0.161 0.347
0 h (2) -0.252 0.643 -0.337 -0.228
4 h (1) -0.264 0.142 0.468 0.328
4 h (2) -0.267 0.164 0.327 -0.441
6 h (1) -0.268 -0.017 0.36 0.186
6 h (2) -0.269 0.015 0.339 -0.351
12 h (1) -0.272 -0.149 0.111 0.209
12 h (2) -0.273 -0.163 0.043 -0.121
18 h (1) -0.271 -0.185 -0.013 0.232
18 h (2) -0.27 -0.223 -0.14 -0.289
24 h (1) -0.268 -0.229 -0.246 0.093
24 h (2) -0.27 -0.196 -0.252 -0.243
48 h (1) -0.266 -0.139 -0.256 0.319
48 h (2) -0.271 -0.121 -0.259 -0.036
Eigenvalue
b
0.932 0.022 0.019 0.008
Cumulative (%)
c
93.2 95.4 97.3 98.1
The values in the columns are the coefficients of the PC indicated at the top.
a
The
independent variables are the seven different time-points analyzed in duplicate.
Numbers 1 and 2 in parentheses indicate the two independent experiments.
b
The
eigenvalues express the variance explained by a particular PC (overall variance=1).
c
The cumulative values are the percentage of information that is described by the
sum of the first n PCs.
Figure 2. PCA analysis.(a) Plot of PC2 coefficients against PC3
coefficients. Data from duplicate experiments (1 and 2) are shown.
(b) Hierarchical clusters showing the different types of gene expres-
sion profiles obtained (see Methods). For each group, the kinetic
points (0, 4, 6, 12, 18, 24 and 48) are represented from left to right.
The label above the upper left corner of each box identifies the clus-
ter and the number above the upper right corner indicates how
many genes were present in each cluster.
a
b
© 2001 Nature Publishing Group http://immunol.nature.com
© 2001 Nature Publishing Group http://immunol.nature.com
nature immunology • volume 2 no 9 • september 2001 • http://immunol.nature.com
ARTICLES
884
arrays. To assure the reproducibility of the assay, the experiment was
done in duplicate and gene expression profiles compared (Fig. 1b).
The variability observed never exceeded 5% of the genes and usual-
ly affected transcript expression to a degree that was close to the
lower limit of detection (1.5 pM). Approximately 30% of the genes
and ESTs displayed on the array were present at each time-point test-
ed. Excluding genes that always remained below the detection limit
during the entire kinetic assay, we obtained 9930 genes that were
expressed at least at one time-point.
Principal component analysis
To get an approximate visualization of our entire dataset, without los-
ing experimental information (variance), we first applied the principal
component analysis (PCA) method, which allows the dimensionality
of complex data to be reduced
12
. Thus, we were able to globally
describe features of the kinetic points that best explained the corre-
sponding transcriptome (Table 1). We analyzed only the first four
principal components (PCs) because they globally explained >98% of
the overall variance. Although PC1 was able to describe >93% of the
total variance, it did not contain any kinetic information, as it was sim-
ply a measure of the time-independent average expression.
Conversely, PC2 and PC3 were two time-dependent parameters. PC2
represented the trend of expression over time, giving a measure of
gene down-regulation; PC3 described the shape of this trend, indicat-
ing the concavity of the expression curve. PC2 versus PC3 coefficients
are shown (Fig. 2a).
Qualitatively this analysis indicates that, after activation, the gen-
eral organization of gene expression was immediately influenced.
Cells then progressively returned to gene expression, which was sim-
ilar to that of immature cells but clearly distinct from it. Thus, in addi-
tion to genes permanently expressed in either immature or terminally
differentiated cells and that characterize these two stages, there is a
group of genes that are transiently modulated during activation. The
process of DC maturation differentiation stabilized 24 h after activa-
tion. PC4 was able to discriminate between two independent repli-
cates (Table 1), which indicated the existence of a systematic exper-
imental error that contributed to the overall variance of <1%. Given
the systematic nature of the error and the correlation coefficient (R
2
)
Figure 3.Transcription response of DCs to bacteria. An example of 130 genes is shown.Genes were grouped according to their most likely function,deduced from
data available in public databases and published literature.
Figure 4.IL-2 expression by DCs. (a) IL-2 mRNA levels in D1 cells after activation with bacteria.Data represent average differences (Avg.Diff.) and standard deviations
calculated from four probes distributed on four different chips (duplicates of chip A and chip B, see Methods).(b) Semi-quantitative PCR of 40 ng samples of cDNA,which
showed IL-2 expression in mBMDCs after bacterial stimulation (done as in Fig.1) at the indicated time-points.mRNA was from mBMDCs.Double-stranded cDNA was tran-
scribed from mRNA,purified and quantified.β-actin acted as a loading control.(c) Quantification of the amount of IL-2 present in the supernatant of bacteria-activated DCs
and macrophages.IL-2 was measured in the supernatants, by ELISA, at the indicated time-points.This experiment was repeated four times with similar results.
a
b
c
© 2001 Nature Publishing Group http://immunol.nature.com
© 2001 Nature Publishing Group http://immunol.nature.com
ARTICLES
http://immunol.nature.com • september 2001 • volume 2 no 9 • nature immunology
885
of linear regression in correlation plots (Fig. 1), it was appropriate to
average the replicates and analyze the mean values further.
Gene clustering
To identify differentially expressed genes within the collection of
9930 genes and ESTs obtained, we first applied a clustering algo-
rithm, which groups genes according to the similarity of their expres-
sion patterns, based on self-organizing maps (SOMs)
13
. Average dif-
ferences (parameters that indicate the degree of expression
14
) between
the duplicates were used and data filtered to exclude all the genes or
ESTs that showed a fold change between maximum and minimum
kinetic values lower than 3. With this filter, sequences to be clustered
were reduced to 2951, about 50% of which were ESTs (Fig. 2b).
Then, a second clustering method (hierarchical clustering) was
applied to each SOM cluster to further investigate smaller profile dif-
ferences and to verify that all the sequences were grouped correctly
15
.
Using this approach, we obtained a good clustering of the genes, with
each hierarchical cluster having a correlation coefficient that was
higher than 0.79 (Fig. 2b). (For further details on Fig. 2, see
www.btbs.unimib.it/DCgenes.)
To classify genes and divide them into functional families, different
databases—GenBank, SwissProt, Kegg and Gene Ontology—were
searched (Fig. 3). As expected, most of the differentially expressed
genes encoded transcription factors and signal transduction molecules.
They were distributed in different clusters, some transiently induced at
different time-points and some stably induced or down-regulated. An
example of the genes affected by bacterial induction is shown (Fig. 3).
To control the validity of this analysis, we checked, within the differ-
ent clusters, whether well established markers for DC activation fol-
lowed the described regulation. For example, inflammatory products
such as tumor necrosis factor-α (TNF-α, cluster 26), macrophage
inflammatory protein 1α (MIP-1α, also known as CCL3, cluster 23),
MIP-1β (also known as CCL4, cluster 27) and MIP-2 (cluster 28) were
modulated during the kinetic assay, as observed in other DC systems
16,17
.
Figure 5.DC-derived IL-2 was a key molecule for T cell activation. (a) Activation profile of mBMDCs after stimulation with bacteria.Expression of the indicated mol-
ecules by unstimulated DCs (thin lines) or DCs that had been activated by bacteria for 15 h (thick lines) were analyzed by flow cytometry.(b) Survival curve of wild-type and
IL-2
–/–
DCs after bacterial encounter. Bacteria-activated DCs (10
6
) were added in six-well plates and the number of viable cells remaining at the indicated time-points evaluat-
ed by trypan blue exclusion. (c) Proliferative response of alloreactive T cells measured by tritiated thymidine incorporation. Graded numbers of bacteria-activated wild-type
and IL-2
–/–
DCs were incubated with 2×10
5
allogeneic T cells in 96-well plates.Background T cell proliferation was evaluated by coculturing syngeneic T lymphocytes with wild-
type DCs.Proliferation was assessed after 72 h by 16 h of exposure to [
3
H]thymidine,expressed as the mean cpm of duplicates.(d,e) Alloreactive T cells activation.DCs (5×10
5
)
were stimulated with bacteria in 24-well plates and 2×10
6
CSFE-labeled CD4
+
or CD8
+
allogeneic T cells were added to the culture.(d) Cycling cells were analyzed at the indi-
cated time-points by FACS analysis.(e) After 48 h of culture,CD69 expression was evaluated on small noncycling T lymphocytes (thin lines) and large T cell blasts (thick lines).
a
b
c
d
e
© 2001 Nature Publishing Group http://immunol.nature.com
© 2001 Nature Publishing Group http://immunol.nature.com
nature immunology • volume 2 no 9 • september 2001 • http://immunol.nature.com
ARTICLES
886
They showed a peak of induction at early time-points and were progres-
sively down-regulated towards the end of the kinetic assay. This sug-
gested that the inflammatory activity of DCs mainly occurs early after
activation, before DCs have left the inflammatory site (Fig. 3). In addi-
tion, IL-12p35 was transiently induced during activation (cluster 32)
(Figs. 2 and 3), with expression peaking at 4 h. Temporary expression of
IL-12p35 also occurs with LPS-activated hMDCs, but with delayed
kinetics
16
. IL-12p40 mRNA (cluster 17) was strongly up-regulated at
early time-points after bacterial encounter and down-regulated at later
time-points (Figs. 2 and 3).
Activation with bacteria induces DC maturation and survival
18
. The
anti-apoptotic genes that were up-regulated after bacteria encounter
were mouse inhibitor of apoptosis protein 1 (MIAP-1, cluster 17),
MIAP-2 (cluster 11), Bcl-x (cluster 22), TNF receptor–associated fac-
tor 1 (TRAF1, cluster 23) and TRAF2 (cluster 29). Conversely, Bcl-2
mRNA was already down-regulated 4 h after bacterial stimulation
(cluster 0), which showed that it was not involved in maintaining
mature DC survival (Fig. 3). Expression of members of the Vav family
and WASp, proteins involved in actin reorganization downstream of
specific signals
19
, was regulated during maturation (Fig. 3).
Induction of IL-2 production by DCs
We observed that cluster 32 contained the transcript encoding the T cell
growth factor IL-2. Bacterial encounter induced transient IL-2 mRNA
up-regulation in D1 cells at early time-points (4–6 h) after activation
(Fig. 4a). We therefore tested the hypothesis that this cytokine could be
an important costimulatory molecule produced by DCs. IL-2 expression
was confirmed by semi-quantitative polymerase chain reaction (PCR)
analysis of fresh mBMDCs, which showed expression kinetics that were
similar to those of D1 cells (Fig. 4b). Protein secretion was measured by
enzyme-linked immunosorbent assay (ELISA). Accumulation of IL-2 in
the supernatant was observed at 4–8 h and again 14–18 h after activation
(Fig. 4c). To test whether IL-2 production after bacterial encounter was
DC-specific, we did the same analysis on bone marrow–derived
macrophages. After activation with bacteria, supernatants were collect-
ed at similar time-points and IL-2 measured. No IL-2 production by
macrophages was observed (Fig. 4c). This indicated that, after bacterial
activation, IL-2 was specifically induced only in DCs.
DC-derived IL-2 mediates T cell activation
The role of IL-2 produced by early bacteria-primed DCs in T cell acti-
vation was investigated by analyzing the capacity of IL-2
–/–
and wild-
type DCs to stimulate alloreactive CD4
+
and CD8
+
T cells in a primary
mixed lymphocyte reaction (MLR) assay. mBMDCs from wild-type and
IL-2
–/–
mice
20
were activated with bacteria at a multiplicity of infection
(MOI) of 10 (Fig. 1). Wild-type and IL-2
–/–
DCs were equally activated
by bacterial stimulation (Fig. 5a) and did not show any difference in via-
bility (Fig. 5b). CD4
+
or CD8
+
allogeneic T cells were added to early
bacteria-primed DCs (4–7 h after activation) and DNA synthesis tested
by thymidine incorporation after three days of culture. The ability of IL-
2
–/–
DCs to induce T cell proliferation was severely impaired (Fig. 5c).
We also directly examined T cell division by flow cytometry analy-
sis. CD4
+
or CD8
+
allogeneic T cells labeled with the cell dye carboxy-
fluorescein diacetate succinimidyl ester (CFSE) were added to early
bacteria-activated wild-type and IL-2
–/–
DCs. The number of mBMDCs
able to induce maximal T cell proliferation activity, as identified in the
previous experiment, was used in this MLR assay. Very few cycling T
lymphocytes were observed in IL-2
–/–
DC cultures (Fig. 5d). This was
not due a nonspecific lack of DC function because IL-2
–/–
DCs were
able to activate T cells, as judged by the increased number of blast cells
expressing the early activation marker CD69 after 48 h of culture (Fig.
5e). No CD69 up-regulation was observed on T cells cultured in
absence of DCs (data not shown).
Discussion
To identify genes that confer DCs with a high T cell stimulatory
capacity, we examined, with kinetic global transcription analysis,
immature DCs and mature cells stimulated by Gram-negative bacte-
ria. The number of genes expressed at different stages of DC ontoge-
ny was similar and encompassed ∼30% of genes displayed on the
arrays. Thus, approximately the same number of genes were induced
and suppressed at different time-points after activation. The diversity
of transcripts expressed in immature, transitional and mature DCs
were similar in magnitude, as has already been suggested for resting
and activated T cells
21
.
The entire dataset was visualized with the PCA method. In these
types of studies, PCA allows one to describe and distinguish a process
of cellular activation from a process of cellular differentiation.
Activation is a reversible process: upon exposure to a given stimulus,
the cells undergo transitional functional and phenotypic modifications
and then return to their original state. In contrast, the differentiation
event is an irreversible process that induces progression toward a new
functional state. As the gene expression profile is a major determinant
of cellular phenotype and function, PCA applied to a cell gene expres-
sion pattern in a kinetic study allows the visualization of similarities
among different states within the same cell after stimulation. In the case
of DCs as visualized by PCA, the cells undergo differentiation because
their gene expression profile shows not only a profound reprogram-
ming at early time-points, which would be consistent with activation,
but it also progresses to a new, distinct steady-state. As indicated by
PCA, the process of DC differentiation is quite rapid. During the 24 h
after bacterial encounter, DCs experience all the transcription modifi-
cation necessary to progress from immature to mature cells.
As expected, activation with bacteria induced the modulation of
many genes involved in cytoskeleton rearrangements, antigen process-
ing, control of migration and apoptosis and regulation of inflammatory
responses. In particular, many factors that modulate the dynamic prop-
erties of actin filaments were differentially expressed during matura-
tion. These factors were proteins involved in coupling actin filaments to
the cell surface, for example, members of the VASP family or proteins
involved in cross-linking actin filaments, such as fascin, or in severing
them, such as gelsolin
22
. In addition, the Vav proteins that induce typi-
cal Rac-1 and RhoG-like cytoskeletal changes—including cell spread-
ing, membrane ruffling and the formation of lamellipodia as a conse-
quence of extensive reorganization of F-actin
23
—were also regulated
through transcription. Constitutive, but modulated, expression of the
transcript overlapping myelin, the transcript expressed outside the cen-
tral nervous system, was observed in D1 cells (cluster 23). Thymic DCs
express tissue-specific antigens to negatively select autoreactive T
cells
24
; analogously, peripheral DC expression of sequestered antigens
could be a mechanism for maintaining peripheral tolerance.
The most unanticipated finding of this study was that DCs produce
IL-2 in a tightly regulated time-frame. Thus, the adjuvant property of
bacteria is explained by induction, in DCs, not only of the up-regula-
tion of costimulatory surface proteins and the maximization of the effi-
ciency in presenting antigens
5
, but also by induction of the production
of costimulatory molecules such as IL-2. This seems to be a unique fea-
ture of DCs because macrophages are unable to produce IL-2 upon bac-
terial activation. Two waves of IL-2 production by DCs after bacterial
encounter were observed. The first was 4–8 h after bacterial uptake and
© 2001 Nature Publishing Group http://immunol.nature.com
© 2001 Nature Publishing Group http://immunol.nature.com
ARTICLES
http://immunol.nature.com • september 2001 • volume 2 no 9 • nature immunology
887
the second was 14–18 h after activation. This timing was compatible
with the appearance of MHC class II
+
–peptide and MHC class I
+
–pep-
tide complexes at the cell surface
4,8,25,26
. DCs are able to present exoge-
nous captured antigens to CD4
+
T cells in a few hours, whereas at least
8 h are required to process and present bacterial antigens in association
with MHC class I molecules
4
. Thus, early-activated DCs are perfectly
equipped to prime CD4
+
T cells, despite their relatively low expression
of MHC and membrane-associated costimulatory molecules and pro-
duction of T cell–inhibitory cytokines, such as IL-10
16
. At later time-
points, IL-2 could represent a key costimulatory protein in activating
CD8
+
T cells, even though DCs have not yet reached their terminal
maturation stage. This data could explain the ability of activated DCs
to prime CD8
+
T cell in a CD4-independent manner
27
. When the fre-
quency of responder antigen-specific T cells or their affinity for
MHC
+
–peptide complexes is low, exogenous sources of IL-2 for induc-
tion of T cell proliferation may be also required, as can frequently hap-
pen in vivo during immune responses to microorganisms.
Signals induced by T cell receptors and IL-2 are involved in activat-
ing and maintaining transcription of the IL-2 receptor (IL-2R) α chain,
which forms, together with the β and γ chains, the high affinity IL-2R.
The α subunit is up-regulated in vivo at the surface of specific T cells 8
h after superantigen injection
21
. The reduction in IL-2 concentration in
the supernatant, 8–10 h after activation, could be explained by re-
uptake of the cytokine by DCs expressing IL-2R
28
. It could be impor-
tant that the release of IL-2 by DCs is efficiently controlled to avoid
“bystander activation” of unrelated T cells. Secretion of IL-2 could also
be necessary to counteract the effect of IL-10, which is produced by
early-activated DCs
16
and which we also found at a high concentration
(data not shown). In absence of IL-2, IL-10 has a well documented
inhibitory function in alloantigen responses
29
.
IL-2 is also thought to activate natural killer (NK) cells in vitro.
However, this effect has never been considered relevant in vivo dur-
ing immunocompetent responses. It was commonly believed that IL-
2 is exclusively produced by T cells during the acquired immune
response, whereas activation of NK cells occurs prior to that, during
the innate response
30
. As DCs can produce IL-2 early after activation,
this assumption should be revised. DCs can activate NK cell respons-
es in direct NK cell–DC interactions
11
; thus IL-2 is likely to be the rel-
evant and obvious costimulatory factor. The finding that DCs can pro-
duce IL-2 at early time-points after bacterial uptake suggests a pri-
mary role of DCs in the activation of innate responses; it also helps
explain the unique ability of these cells to prime T lymphocytes. This
observation was made more than 20 years ago
31
, but, until now, the
molecular events responsible for T cell priming were not understood.
Methods
DC, macrophages and culture medium. D1 cells were cultured in Iscove’s modified
Dulbecco’s medium (IMDM, Sigma, St. Louis, MO) containing 10% heat-inactivated fetal
bovine serum (Gibco-BRL, Gaithersburg, MD), 100 IU of penicillin, 100 µg/ml of strepto-
mycin, 2mM
L-glutamine (all from Sigma) and 50 µM β-mercaptoethanol (in complete
IMDM) with 30% supernatant from R1 medium (supernatant from NIH3T3 fibroblasts
transfected with GM-CSF)
3
. Macrophages and DCs were derived from bone marrow cells
collected from four different mice and divided in two separate cultures. mBMDCs were col-
lected after 14 days of bone marrow culture in 30% R1 medium. Macrophages were col-
lected after 14 days of bone marrow culture in presence of M-CSF. Wild type and IL-2
–/–
DCs were from mice from the same litter. mBMDCs were used for bacterial activation only
if they were immature, as judged by low B7-2 expression and absence of CD40. Cells that
showed a partial spontaneous activation were discarded.
Infection with bacteria. E. coli (DH5α strain) were grown overnight in Luria-Bertani
medium (LB). Samples (100 µl) were inoculated in 10 ml of fresh LB and grown for an
additional hour at 37 °C. Bacteria were added to cell cultures at a MOI of 10. Cultures were
incubated for 1.5 h. DC and macrophage cultures were washed and supplemented with gen-
tamycin and tetracycline at a final concentration of 50 µg/ml and 30 µg/ml, respectively.
Sample preparation and array hybridization. Antisense cRNA was prepared following
Affymetrix (Santa Clara, CA) recommendations. Briefly, total RNA was extracted from
frozen pellets with the Trizol procedure. The Oligotex kit from Qiagen (Chatsworth, CA)
was used to purify mRNA. Double-stranded cDNA was retro-transcribed with a modified
oligo(dT) primer with a 5′ T7 RNA polymerase promoter sequence and the Superscript
Choice System for cDNA synthesis (Life Technologies, Gaithersburg, MD). Double-strand-
ed cDNA (1 µg) was transcribed to cRNA with the ENZO kit (Affymetrix). cRNA was puri-
fied on an affinity column (Rneasy, Qiagen) and then fragmented to an average size of
50–200 bp, by incubation for 35 min at 94 °C in 40 mM Tris-acetate at pH 8.1, 100 mM
potassium acetate and 30 mM magnesium acetate. Samples were diluted in the hybridiza-
tion solution (1 M NaCl, 10 mM Tris at pH 7.6, 0.005% Triton X-100, 0.1 mg/ml of herring
sperm DNA, BioB, BioC, BioD and cre control cRNAs at concentrations of 1.5, 5, 25 and
100 pM, respectively) at a final concentration of 0.05 µg/ml and heated to 94 °C for 5 min.
Analysis of the samples was done by hybridizing the fragmented cRNAs to the Affymetrix
Mu11k GeneChip
array, which consisted of two individual chips (called A and B) that col-
lectively represented ∼11,000 murine genes and ESTs. Probe array hybridizations were car-
ried out, as described
32
, by placing the samples in the hybridization cartridge at a final vol-
ume of 200 µl/chip. Hybridizations were done under rotation at 45 °C for 16 h. After
hybridization, the chips were rinsed with 6×SSPE-T (0.9 M NaCl, 60 mM NaH
2
PO
4
, 6 mM
EDTA, 0.005% Triton X-100 adjusted to pH 7.6), 0.5× SSPE-T and stained by incubation
with 2 µg/ml of phycoerythrin-streptavidin (Molecular Probes, Eugene, OR) and 1 mg/ml
of acetylated bovine serum albumin (Sigma). The arrays were read at a resolution of 7.5 µm
by a confocal scanner and analyzed with the MicroArray Suite 4.0 Gene Expression analy-
sis program (both from Affymetrix).
PCA analysis. Briefly, we first rewrote our dataset in terms of a matrix in which the 9930
rows represented the genes and ESTs (data observations) and the 14 columns represented
the seven different time-points in duplicate (independent variables). The a
ij
th element of the
matrix was the average difference, which represented gene expression
14
, of the ith gene at
the jth experimental condition. The method proceeded through a unitary transformation
(rotation) of the matrix, which returned two distinct matrices: the eigenvector matrix and the
eigenvalue matrix. The columns of the eigenvector matrix were the PCs of the dataset and
the rows were the time-points in duplicate. The eigenvalues represented the percentage of
the overall variance that each component described. PC1, PC2, PC3 and PC4 together were
able to describe >98% of the entire variance.
PCR primers and IL-2 ELISA. The sequences of the PCR primer pairs (5′→3′) used were
as follows. IL-2: sense, TCCTCACAGTGACCTCAAGTCC; antisense, TGACAGAAG-
GCTATCCATCTCC. β-actin: sense, CATCGTGGGCCGCTCTAGGCAC; antisense, CCG-
GCCAGCCAAGTCCAGACGC. IL-2 ELISA was done with the DuoSet kit (R&D
Systems, Minneapolis, MN), following the manufacturer recommendations.
MLR. Wild-type and IL-2
–/–
DCs (5×10
5
) were activated with bacteria and 4–7 h later incu-
bated with 2×10
6
CFSE-labeled T cells. T cell division was assessed by FACS analysis 48
or 72 h later. Alternatively graded numbers of bacteria-activated DCs were incubated with
2×10
5
T cells and proliferation tested by [
3
H]thymidine incorporation 72 h later. CD4
+
and
CD8
+
lymphocytes were purified (to 99% purity) from BALB/c or C57BL/6 mice lymph
nodes by negative selection of macrophages, DCs, B cells and CD4
+
or CD8
+
T cells. These
cell populations were eliminated with the MiniMACS columns (Miltenyi Biotec, Auburn,
CA, GmbH) after preincubation with Mac1, Cd11c, B220 and CD4 or CD8 antibodies (all
from PharMingen, San Diego, CA).
Hierarchical clusters. Clusters showing the different types of gene expression profiles
were obtained with the Xcluster program (http://genome-www.stanford.edu/∼sherlock/clus-
ter.html). A filter was used to examine genes that showed a difference in expression of at
least 3 between the maximum and minimum kinetic expression values. Genes were first
clustered with the SOM algorithm; then the hierarchical method was applied to each SOM
cluster. After various attempts at clustering, 36 clusters was chosen as the most appropriate
number: higher numbers resulted in the multiplication of clusters with very similar profiles
and lower numbers did not give good correlation coefficients in the hierarchical clusters.
Mice. Pathogen-free C57BL/6 and BALB/c mice were from Harlan Italy (Milan, Italy).
C57BL/6 IL-2
–/–
mice were kept in pathogen free conditions. All experiments were done in
compliance with the relevant laws and institutional guidelines.
Acknowledgments
We thank A. Schimpl for IL-2
–/–
mice, D. Grdic for helpful discussions,E. Bottani for manu-
script editing and Affymetrix for technical support.Supported by the EC Grant QLG1-
1999-00202-TAGAPO,the CNR Target Project on Biotechnology, the Italian Association
for Cancer Research (AIRC),Biopolo,Novuspharma and Lombardia Region.
Received 9 March 2001; accepted 25 June 2001.
1. Banchereau,R. & Steinman R. M. Dendritic cells and the control of immunity. Nature 392,
245–252 (1998).
2. Sallusto,F. & Lanzavecchia,A.Efficient presentation of soluble antigen by cultured human dendritic
cells is maintained by granulocyte/macrophage colony-stimulating factor plus interleukin-4 and
downregulated by tumor necrosis factor α. J.Exp. Med.179, 1109–1118 (1994).
3. Winzler, C. et al. Maturation stages of mouse dendritic cells in growth factor-dependent long-term
© 2001 Nature Publishing Group http://immunol.nature.com
© 2001 Nature Publishing Group http://immunol.nature.com
nature immunology • volume 2 no 9 • september 2001 • http://immunol.nature.com
ARTICLES
888
cultures.J.Exp. Med. 185, 317–328 (1997).
4. Rescigno,M. et al. Bacteria-induced neo-biosynthesis, stabilization, and surface expression of func-
tional class I molecules in mouse dendritic cells. Proc. Natl Acad. Sci. USA 95,5229–5234 (1998).
5. Banchereau, J. et al.Immunobiology of dendritic cells. Annu. Rev.Immunol.18,767–811(2000).
6. Lutz, M.B.et al.An advance culture method for generating large quantities of highly pure dendritic
cells from mouse bone marrow. J.Immunol. Meth. 223, 77–92 (1999).
7. Pierre, P.et al. Development regulation of MHC class II transport in mouse dendritic cells. Nature
388,787–792 (1997).
8. Rodriguez,A. et al. Selective transport of internalized antigens to the cytosol for MHC class I pre-
sentation in dendritic cells. Nature Cell Biol. 1,362–368 (1999).
9. Rescigno,M. et al. Fas engagement induces the maturation of dendritic cells (DCs),the release of inter-
leukin (IL)-1β,and the production of interferon γ in the absence of IL-12 during DC-T cell cognate
interaction.A new role for fas ligand in inflammatory responses. J.Exp.Med. 192, 1661–1668 (2000).
10. Singh-Jasuja, H.et al. Cross-presentation of glycoprotein 96-associated antigens on major histocom-
patibility complex class I molecules requires receptor-mediated endocytosis. J.Exp. Med. 191,
1965–1974 (2000).
11. Fernandez, N.C. et al. Dendritic cells directly trigger NK cell functions:cross-talk relevant in innate
anti-tumor immune responses in vivo. Nature Med.5, 405–411 (1999).
12. Raychaudhuri,S., Stuart,J. M. & Altman,R. B. Principal components analysis to summarize microarray
experiments:application to sporulation time series. Proceedings of the Pacific Symposium on
Biocomputing 455–466 (2000).
13. Tamayo, P.et al. Interpreting pattern of gene expression with self-organizing maps:methods and
application to hematopoietic differentiation. Proc. Natl Acad. Sci. USA 96, 2907–2912 (1999).
14. Lockhart,D. J.et al. Expression monitoring by hybridization to high-density oligonucleotide arrays.
Nature Biotechnol. 14, 1675–1680 (1996).
15. Staudt, L. M. & Brown,P. O.Genomic views of the immune system.Annu. Rev.Immunol. 18,
829–859 (2000).
16. Langenkamp,A.,Messi, M., Lanzavecchia,A. & Sallusto,F.Kinetics of dendritic cell activation: impact
on priming of Th1,Th2 and nonpolarized T cells. Nature Immunol. 1,311–316 (2000).
17. Sallusto, F.et al. Distinct patterns and kinetics of chemokine production regulate dendritic cell func-
tion.Eur.J.Immunol. 29, 1617–1625 (1999).
18. Rescigno, M. et al. Dendritic cell survival and maturation are regulated by different signaling path-
ways.J. Exp.Med. 188, 2175–2180 (1998).
19. Bustelo,X.R.Regulatory and signaling properties of the Vav family.Mol.Cell.Biol.20,1461–1477 (2000).
20. Kramer, S.et al.Thymic selection and peptide-induced activation of T cell receptor-transgenic CD8 T
cells in interleukin-2-deficient mice. Eur.J. Immunol. 24,2317–2322 (1994).
21. Teague,T.K.et al.Activation changes the spectrum but not the diversity of genes expressed by T
cells.Proc. Natl Acad.Sci. USA 96,12691–12696 (1999).
22. Borisy, G. G. & Svitkina,T. M.Actin machinery:pushing the envelope. Curr. Opin. Cell Biol. 12,
104–112 (2000).
23. Movilla, N.& Bustelo,X. N. Biological and regulatory properties of Vav-3, a new member of the Vav
family of oncoprotein.Mol.Cell.Biol. 19, 7870–7885 (1999).
24. Klein, L.et al. Shaping of the autoreactive T-cell repertoire by a splice variant of self protein
expressed in thymic epithelial cells. Nature Med. 6,56–61 (2000).
25. Cella, M.et al. Inflammatory stimuli induce accumulation of MHC class II complexes on dendritic
cells. Nature 388, 782–787 (1997).
26. Lutz, M.B.et al. Intracellular routes and selective retention of antigens in mildly acidic cathepsin
D/lysosome-associated membrane protein-1/MHC class II-positive vesicles in immature dendritic
cells. J. Immunol.159, 3707–3716 (1997).
27. Schuurhuis, D.H.et al. Immature dendritic cells acquire CD8
+
cytotoxic T lymphocyte priming
capacity upon activation by T helper cell-independent or -dependent stimuli.J. Exp.Med. 192,
145–150 (2000).
28. Kronin,V.,Vremec, D. & Shortman,K. Does the IL-2 receptor α chain induced on dendritic cells have
a biological function? Int.Immunol. 10, 237–240 (1998).
29. Zeller, J.C.et al. Induction of CD4
+
T cell alloantigen-specific hyporesponsiveness by IL-10 and TGF-
β. J. Immunol.163, 3684–3691 (1999).
30. Biron, C.A. et al.Natural killer cells in antiviral defense: function and regulation by innate cytokines.
Annu.Rev.Immunol.17, 189–220 (1999).
31. Steinman R. M. et al. Dendritic cells of the mouse:identification and characterization. J Invest.
Dermatol.75, 14–16 (1980).
32. Wodicka,L. et al. Genome-wide expression monitoring in Saccharomyces cerevisiae. Nature Biotechnol.
15,1359–1367 (1997).
© 2001 Nature Publishing Group http://immunol.nature.com
© 2001 Nature Publishing Group http://immunol.nature.com