| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Medicine, University of Pennsylvania, Philadelphia, PA 19104
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
40%
of these activated cells fail entirely to participate in clonal
expansion. To address how prior cell division influences the subsequent
function of primary T cells at the single cell level, primary
CD4+ T cells were subjected to polyclonal stimulation,
sorted based on the number of cell divisions they had undergone, and
restimulated by ligation of TCR/CD28. We find that individual
CD4+ T cells exhibit distinct secondary response patterns
that depend upon their prior division history, such that cells that
undergo more rounds of division show incrementally greater IL-2
production and proliferation in response to restimulation.
CD4+ T cells that fail to divide after activation exist in
a profoundly hyporesponsive state that is refractory to both
TCR/CD28-mediated and IL-2R-mediated proliferative signals. We find
that this anergic state is associated with defects in both TCR-coupled
activation of the p42/44 mitogen-activated protein kinase
(extracellular signal-related kinase 1/2) and IL-2-mediated
down-regulation of the cell cycle inhibitor
p27kip1. However, these defects are selective,
as TCR-mediated intracellular calcium flux and IL-2R-coupled STAT5
activation remain intact in these cells. Therefore, the process of cell
division or cell cycle progression plays an integral role in anergy
avoidance in primary T cells, and may represent a driving force in the
formation of the effector/memory T cell pool. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
locus
(9). This diversity observed in primary T cell responses has suggested that an individual T cell’s proliferative behavior during primary culture might influence its eventual effector function (10, 11, 12). To further address this question, we have labeled T cells with the fluorescent dye 5- and 6-carboxyfluorescein diacetate succinimidyl ester (CFSE),3 stimulated these cells in vitro, and sorted them by flow cytometry based upon their proliferative history. Our data demonstrate that the responsiveness of individual T cells to secondary TCR engagement is quantitatively tuned to the number of mitotic events achieved during primary stimulation. We also define a T cell fate that is refractory to both TCR- and IL-2R-mediated signals, and is associated with the failure to proliferate following primary activation in the presence of CD28 costimulation. These data support a model in which T cells must divide after Ag encounter to avoid anergy.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Pooled spleen and lymph node cells from female BALB/c mice, aged
8–12 wk, were used for all experiments. mAbs against CD3 (145-2C11)
and CD28 (37.51) were purified from hybridomas obtained from J.
Bluestone (University of Chicago, Chicago, IL) and J. Allison
(University of California, Berkeley, CA), respectively. Purified,
fluorochrome-conjugated mAb against CD16/CD32 (Fc-block), Thy-1.2, CD4,
CD25, CD122, common 
-chain (c), IL-2, and
IFN-, and biotinylated anti-CD3 and anti-CD28 were purchased
from PharMingen (San Diego, CA). Abs reactive with both the
phosphorylated and unphosphorylated forms of extracellular
signal-related kinase (ERK)1, ERK2, and STAT5 were purchased from Zymed
Laboratories (San Francisco, CA). mAb specific for
p27kip1 was purchased from Transduction
Laboratories (Lexington, KY). Abs specifically reactive with only the
phosphorylated forms of ERK1 and ERK2 were purchased from New England
Biolabs (Beverly, MA), and Ab reactive with the phosphorylated form of
STAT5 was purchased from Zymed Laboratories (San Francisco, CA). Rabbit
antiserum against actin was purchased from Sigma (St. Louis, MO). CFSE,
Indo-1 AM, FuraRed AM, and TOPRO-3 were purchased from Molecular Probes
(Eugene, OR). PMA and ionomycin were purchased from Sigma and were used
at 5 ng/ml and 0.25 µM, respectively. Murine rIL-2 was obtained from
Genzyme (Cambridge, MA), and was used at 15–30 U/ml.
Cell labeling and culture conditions
Cell isolation and fluorescent labeling of cells with CFSE were performed as previously described (8). Briefly, pooled spleen and lymph node cells were incubated with CFSE in PBS at a final concentration of 2 µM for 3 min. CFSE-labeled cells were stimulated with soluble anti-CD3 mAb (1 µg/ml, unless otherwise stated) at 2–5 x 106/ml in either 24-well plates or 75-cm2 flasks. Where indicated, cultures contained anti-CD28 mAb (1 µg/ml) and/or IL-2 (15 U/ml). For primary cultures, the cells were stimulated for 4 days, washed, and replated in fresh medium for 48 h, then sorted based on CFSE fluorescence. Sorted T cells were restimulated in 96-well round-bottom plates with soluble Abs, as indicated, for 4 days in the presence of a 4-fold excess of irradiated syngeneic splenocytes, and secondary proliferation (CFSE fluorescence) was analyzed by flow cytometry.
Flow cytometry and cell sorting
Cell surface marker staining was performed as described (8), and flow cytometric analysis was performed on a FACScalibur flow cytometer using CellQuest software (Becton Dickinson, San Jose, CA). Methods using CFSE labeling to calculate the responder frequency (defined as the proportion of input T cells that undergo one or more cell divisions during the culture period) and the absolute number of mitotic events occurring in the culture have been described (8). The vital dye TOPRO-3 was used to discriminate live and dead cells (8). Live, Thy-1.2+, or CD4+ T cells were sorted based on CFSE fluorescence using a FACSvantage flow cytometer/sorter (Becton Dickinson).
Intracellular cytokine staining
Cytokine expression was assessed at the single cell level, as described previously (5), with some modifications. Cells that had been activated in primary culture for 4 days and rested for 48 h were cultured for 5 h with plate-bound anti-CD3 plus anti-CD28 mAbs (5 µg/ml each) in the presence of 2 µM monensin (Sigma). As primed T cells do not divide during this 5-h restimulation period (data not shown), cytokine production can be assessed as a function of primary proliferative history without the use of cell sorting.
Measurement of intracellular Ca2+ concentration ([Ca2+]i)
Primed T cells were loaded with either Indo-1 AM (2 µg/ml) or FuraRed AM (10 µg/ml) (Molecular Probes) for 30 min at 30°C in RPMI 1640 with 1% serum (13). Agonistic, biotinylated anti-CD3 and anti-CD28 Abs were also added at this time. Cells were washed, resuspended in RPMI 1640 lacking serum and sodium bicarbonate, and warmed to 37°C for 5 min, and Indo-1 or FuraRed fluorescence before and after receptor cross-linking was assessed on a FACStarPlus or FACScalibur flow cytometer, respectively (13). Receptor cross-linking was achieved by the addition of streptavidin (0.2 µg/ml, final concentration). Maximal [Ca2+]i flux was achieved by the addition of ionomycin (1 µM). Kinetic analysis of [Ca2+]i was achieved using FloJo flow cytometry software (Tree Star Software, San Carlos, CA).
Assessment of ERK activation, STAT5 activation, and p27kip1 degradation
CD4+ T cells sorted based on division cycle were rested at 37°C for 2–4 h in medium, and a portion of each pool was set aside for analysis of total ERK content. The remaining cells were restimulated for 10 min with polystyrene beads coated with anti-CD3 mAb ± anti-CD28 mAbs (14), or with PMA. Whole cell lysates (6–8 x 105 cell equivalents per lane) were subjected to SDS-PAGE, transferred to nitrocellulose membranes, and probed with anti-ERK or anti-phospho-ERK antisera (1/1000 dilution). Bands were quantified by densitometric analysis using NIH Image software, and the phospho-ERK signals were normalized to the corresponding total ERK signals to determine relative ERK activation. For the assessment of p27kip1 degradation, cells primed and rested as above were cultured for 48 h in the presence of 50 U/ml IL-2. The cells were then sorted into divided and undivided fractions, and the whole cell lysates were subjected to immunoblot analysis as above using a mAb against p27kip1. For the assessment of STAT5 activation, splenic T cells were primed for 4 days with anti-CD3 (1 µg/ml), rested for 24 h, and restimulated by the addition of IL-2 (100 U/ml) for 10 min. The cells were fixed for 5 min in cold 1% formaldehyde, permeabilized for 10 min in cold methanol, and washed with PBS containing 3% nonfat dry milk. The fixed/permeabilized cells were then stained for either total STAT5 or phosphorylated STAT5 using 1 µg of specific primary Ab for 1 h at room temperature. Cells were washed four times with PBS containing 3% milk, and Cy5-conjugated donkey anti-mouse IgG F(ab')2 (0.3 µg) was added for 30 min at room temperature. Cells were washed twice in PBS and subjected to flow cytometric analysis.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Previously, we have shown that while optimal stimulation of
freshly isolated T cells by anti-CD3 ± anti-CD28 mAb
induces the activation of 95–98% of the T cell population, as
assessed by CD25 and CD69 expression, a large proportion of the
activated T cells (up to 50%) fails to undergo even a single round of
cell division (8). This raised the question as to whether
there might be an association between the heterogeneity in
proliferative behavior observed within the CD4+ T
cell pool and heterogeneity in subsequent effector function (i.e.,
cytokine production and proliferation). To test this hypothesis, murine
spleen and lymph node cells were labeled with CFSE and cultured in
vitro with an optimal concentration of T cell mitogenic anti-CD3 Ab
for 4 days. The cultures were then washed and rested for 48 h, and
T cells that had divided once, twice, or had remained undivided
following primary stimulation were purified by FACS and restimulated
(Fig. 1
A).
|
demonstrates this proliferative
heterogeneity. After the peak of the clonal expansion phase (day 4),
25.7% of the input T cells responded to primary stimulation by
dividing once, 18.7% divided twice, 11% divided three times, and less
than 4% divided four or more times, while 41% remained undivided.
Following cell sorting, T cells were cultured with fresh syngeneic
accessory cells in medium alone, or in the presence of anti-CD3
mAb. As the cells remained labeled with CFSE, we were able to monitor
their proliferative response to restimulation. Sorted
CD4+ T cells cultured with feeders in medium
alone did not divide during the 4-day restimulation period (Fig. 1B), confirming that all of the T cells had exited the cell
cycle following the initial phase of activation and rest. When
subjected to TCR reengagement by the addition of anti-CD3 to the
restimulation cultures, those CD4+ T cells that
had undergone at least a single round of cell division during primary
activation initiated a second burst of proliferation (Fig. 1C, middle and lower panels). In
contrast, those cells that failed to divide following primary
activation were unable to proliferate in response to TCR reengagement
(Fig. 1C, upper panel).
Quantitative, single cell analysis of the dynamics of secondary clonal
expansion shows that the proportion of input CD4+
T cells that responded to restimulation by proliferating (i.e., the
responder frequency) was less than 5% in the undivided pool, whereas
65% of the input CD4+ T cells that divided once
during primary activation, and 75% of the CD4+ T
cells that divided twice were able to reenter the cell cycle when
subjected to TCR ligation again (Fig. 1D). Measuring the
accumulation of cell divisions within the CD4+
subset during the restimulation period shows that 10,000 undivided
CD4+ T cells cultured in the presence of
anti-CD3 gave rise to <1,500 total mitotic events. In contrast,
the same number of cells from the pool of CD4+ T
cells that divided once gave rise to 50,000 mitotic events, while the
CD4+ T cells that divided twice during primary
activation gave rise to >130,000 mitotic events (Fig. 1E).
These absolute mitotic event values correspond to the generation of an
average of between three and four daughter cells per responder from the
pool of precursors with a prior division history of one, while the
average CD4+ T cell that divided twice during the
primary response generated 10 daughters during secondary clonal
expansion.
Therefore, we find that both the probability that an individual CD4+ T cell will participate in the secondary phase of clonal expansion, and the number of rounds of cell division it subsequently achieves are linked to its proliferative behavior during the primary response.
Relationship between primary proliferation and cytokine production
To explore the potential basis for the observed relationship
between primary and secondary proliferative behavior, we analyzed the
production of IL-2 by individual CD4+ T cells as
a function of their primary division status. CFSE-labeled splenocytes
were primed as described above, and then briefly (5 h) restimulated by
coligation of TCR and CD28, after which cytokine production was
assessed by flow cytometry. As was the case for proliferation, the
frequency of CD4+ T cells that produced IL-2 upon
secondary TCR ligation increased with each successive division cycle
(Fig. 2, A and C),
such that while only 3% of undivided CD4+ T
cells could secrete IL-2, 12% of those cells that had divided three
times were able to produce IL-2. We have observed a similar
relationship between IL-2 production and cell division in an MHC class
II-restricted, TCR-transgenic T cell model, in which the frequency of
IL-2 producers is <5% within the undivided Ag-specific
CD4+ T cell pool, but increases with each cell
division such that up to 80% of the cells that have divided five or
six times are able to produce IL-2 (9). This trend in cell
division-associated IL-2 production does not generalize to all
cytokines in this in vitro model, as all primed
CD4+ T cells exhibited a similar capacity to
secrete IFN-, regardless of their previous proliferative behavior
(Fig. 2, B and D). These results suggest that T
cells that fail to divide after activation are subsequently unable to
produce IL-2. This defect is a hallmark of T cell clonal anergy
(15) and could possibly explain the inability of the
undivided cells to proliferate in response to restimulation.
|
The failure of the undivided CD4+ T cells to
produce IL-2 and proliferate in response to mitogenic anti-CD3 Abs,
despite normal expression of the ß TCR (data not shown), suggested
a biochemical defect in TCR-coupled signal transduction. As a first
step to localize this defect, we examined TCR-proximal signals involved
in the elevation of
[Ca2+]i following
engagement of TCR/CD3. All T cells exhibited an increase in
[Ca2+]i in response to
CD3 ligation (Fig. 3A), and
the relative increase in
[Ca2+]i was comparable in
T cells with varied proliferative history (Fig. 3B). The
ability of the nonproliferative cells to flux calcium and secrete
IFN- (Fig. 2) in response to TCR ligation emphasizes that these
cells are not globally unresponsive.
|
4-fold (ERK1)
and 2-fold (ERK2) in undivided CD4+ T cells
compared with the levels in divided CD4+ T cells
(Fig. 4, A and B,
upper panels). In addition, we observed a ligand
density-dependent defect in the capacity to activate these ERK species
in the undivided compared with the divided CD4+ T
cell pool. In response to restimulation with a relatively high density
of TCR ligand (1 µg/ml anti-CD3), the undivided
CD4+ T cells exhibited a 3.5-fold reduction in
the relative activation of ERK1 compared with the activation of this
isoform in divided cells, but ERK2 was activated to a comparable degree
(Fig. 4A, middle panel). At half this ligand
density (0.5 µg/ml anti-CD3), the defect was more profound:
activation of ERK1 was not detected in the undivided
CD4+ T cells, and ERK2 activation was reduced
4-fold compared with the activation of this isoform in the divided
cells (Fig. 4A, lower panel). Therefore, we find
that CD4+ T cells that fail to proliferate in
response to primary stimulation suffer from a quantitative defect in
TCR-coupled activation of the Ras-Raf-MAPK pathway. This signaling
defect has been described previously in anergic T cell clones
(22, 23, 24, 25), and could explain the inability of the undivided
cells to produce IL-2.
|
B, lower
panel). This result suggests that the diminished activation of ERK
in the undivided cells is not due to an intrinsic defect in the
Ras-Raf-MAPK cascade, but rather that this pathway is inefficiently
coupled to the TCR. Because maximal signals through TCR and CD28
cooperated to overcome the defect in ERK activation, and because CD28
costimulation is also important for the activation of other signal
transduction pathways required for optimal IL-2 production and
proliferation (28, 29), we tested whether coligation of
TCR and CD28 by agonistic Abs could overcome the proliferative defect
exhibited by previously undivided CD4+ T cells.
While the combination of agonistic anti-CD3 and anti-CD28 Abs
induced a nearly 4-fold expansion in the pool of previously divided
cells, these signals failed to restore efficient clonal expansion by
the undivided CD4+ T cells (in this experiment,
20,000 T cells gave rise to 76,100 mitoses within the previously
divided pool vs 10,660 within the undivided pool; a 7-fold difference)
(Fig. 4C). This is consistent with the fact that coligation
of CD28 and TCR did not result in IL-2 production by the undivided
cells (Fig. 2). These results suggest that an additional defect(s)
exists either downstream of ERK, or more likely, within a signal
transduction pathway distinct from the Ras-Raf-MAPK pathway, and that
this pathway is required for T cell proliferation. Nonproliferative CD4+ T cells are unable to use IL-2 as a growth factor
The inability of the previously undivided
CD4+ T cells to activate the Ras-Raf-MAPK
cascade, produce IL-2, and proliferate in response to restimulation
constitutes a phenotype that is remarkably similar to that of anergic T
cell clones. However, another hallmark of T cell clonal anergy is that
it can be readily reversed by the addition of IL-2 (30).
Therefore, we examined whether the hypoproliferative phenotype of the
undivided CD4+ T cells could be reversed by the
addition of exogenous IL-2. Previously divided or undivided
CD4+ T cells were restimulated by TCR ligation in
the presence or absence of exogenous IL-2 (Fig. 5A). In response to TCR
ligation alone, the previously divided and undivided
CD4+ T cell pools exhibited differences in
responder frequency and proliferative capacity similar to those shown
in Fig. 1, which together translated to a greater than 20-fold
difference in the number of total mitotic events. The previously
undivided CD4+ T cells were likewise refractory
to IL-2 alone. Surprisingly, even the combination of TCR or TCR/CD28
ligation and exogenous IL-2, which induced a 30-fold expansion of the
previously divided T cell pool, failed to induce comparable expansion
of the previously undivided population (Fig. 5A). Therefore,
the apparent link between primary T cell proliferative behavior at the
single cell level and subsequent proliferative responsiveness upon
reactivation could be explained by the fact that the capacity of an
individual CD4+ T cell to not only produce IL-2
(Fig. 2), but also to respond to IL-2 (Fig. 5), is quantitatively
associated with cell division. The inability of the undivided cells to
respond to IL-2 represents a distinction between this
division-associated hyporesponsive state and anergy in T cell
clones.
|
To understand the basis of the association between cell division
and IL-2 responsiveness, we first assessed the expression of IL-2R
chains on CD4+ T cells of varying proliferative
histories. The majority of primed CD4+ T cells,
regardless of prior division history, were induced to express IL-2R
-chain (CD25) upon TCR reengagement (Fig. 5B, •),
although the relative level of CD25 per cell was successively higher on
cells that had undergone more rounds of cell division (Fig. 5B, 
). Flow cytometric analysis of IL-2R ß-chain
(CD122) and c-chain expression on primed
CD4+ T cells showed no significant differences as
a function of division cycle (data not shown). Furthermore, both
divided and undivided CD4+ T cells were able to
up-regulate CD25 in response to IL-2 (Fig. 5C). Therefore,
the differential responsiveness of the divided vs the undivided
CD4+ T cells to IL-2 cannot be explained simply
by the presence vs absence of high-affinity IL-2R, suggesting that the
undivided CD4+ T cell pool suffers from a
biochemical defect(s) in signal transduction downstream of IL-2R.
Two major IL-2R-coupled signaling pathways have been reported to be
absolutely necessary for the transduction of IL-2-mediated
proliferative signals in T cells. One pathway involves the Janus kinase
(JAK)3-mediated activation of the transcription factor STAT5
(31), while the other pathway involves the
phosphatidylinositol-3 kinase (PI3K)-mediated
activation of protein kinase B (PKB), also known as Akt
(32). Interestingly, up-regulation of CD25 in response to
IL-2, which occurs normally in both the divided and undivided T cell
populations (Fig. 5, B and C), requires the
functional activity of STAT5 (33). This suggests that the
JAK3/STAT5 pathway is not compromised in the undivided
CD4+ T cell pool. To confirm this biochemically,
we assessed the activation of STAT5 in IL-2-stimulated
CD4+ T cells by measuring STAT5 phosphorylation.
A relatively large amount of total, intracellular STAT5 could be
detected by flow cytometry in all CD4+ T cells
(Fig. 6A). Specific
phosphorylation of STAT5 could likewise be detected, in an
IL-2-dependent manner, in all CD4+ T cells (Fig. 6A). Quantitative analysis of phospho-STAT5 content as a
function of cell division (Fig. 6, B and C)
showed that all CD4+ T cells could activate STAT5
to a similar degree in response to IL-2, regardless of their
proliferative behavior during the primary stimulus. These data suggest
that the inability of the previously undivided
CD4+ T cells to proliferate in response to IL-2
results from a biochemical defect in a pathway distinct from the
JAK3/STAT5 pathway. To test the integrity of the
PI3K/Akt pathway as a function of cell division,
we compared the ability of divided vs nondivided
CD4+ T cells to down-regulate the
cyclin-dependent kinase inhibitor p27kip1 in
response to IL-2. IL-2-mediated down-regulation of
p27kip1 is crucial for the progression of T
cells from the G1 phase to the S phase of the
cell cycle (34), and activation of the
PI3K/Akt pathway is both necessary and sufficient
to achieve p27kip1 down-regulation in T cell
lines (35). Interestingly, we found that while those
CD4+ T cells that participated in clonal
expansion during the primary stimulus contained very little
p27kip1 after restimulation with IL-2 (Fig. 6D, right lane), CD4+ T
cells that failed to divide were unable to down-regulate this cell
cycle inhibitor (Fig. 6D, left lane). These
results suggest that the inability of the undivided
CD4+ T cells in this model may be due to a defect
in IL-2R-coupled signal transduction involving the
PI3K, PKB/Akt pathway.
|
The data above suggest that the integrity of both TCR-coupled and
IL-2R-coupled signal transduction pathways is linked to cell division,
and that activated T cells that fail to divide suffer from defects in
both these pathways that render them unable to proliferate upon further
mitogenic stimulus. We next tested whether these functional defects
could be overcome by bypassing T cell surface receptors. The
combination of the phorbol ester PMA and the calcium ionophore
ionomycin can uncouple the Ras-Raf-MAPK cascade from the Ag receptor by
directly activating PKC (36, 37, 38). Receptor-independent
activation of PKC using PMA resulted in comparable ERK1/2
phosphorylation in both divided and undivided
CD4+ T cells (Fig. 7A). PKC also lies downstream
of PI3K in the IL-2R signal transduction pathway
(39, 40), and is normally activated in response to IL-2
(41). To test whether receptor-independent activation of
PKC could overcome the defect in IL-2-mediated
p27kip1 down-regulation exhibited by the
undivided CD4+ T cell subset, we restimulated
primed cells with PMA and ionomycin instead of IL-2. This combination
of phorbol ester and calcium ionophore was able to induce comparable
p27kip1 down-regulation in both the divided and
the undivided CD4+ T cells (Fig. 7B).
We next compared the ability of the divided vs the undivided
CD4+ T cells to proliferate upon restimulation
with anti-CD3 vs PMA/ionomycin. As seen before, the previously
undivided CD4+ T cells exhibited little
proliferation in response to TCR engagement compared with the divided
population (Fig. 7C). However, the combination of PMA and
ionomycin was able to stimulate a large and equal degree of
proliferation by both the divided and the undivided subsets (Fig. 7C). These results suggest that receptor-independent
activation of PKC is able to bypass the defect in
PI3K-mediated signal transduction normally
exhibited by the undivided CD4+ subset.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
An association between cell division and cytokine production has been observed previously during the primary phase of clonal expansion in CD4+ T cells in vitro (10, 11) and in vivo (9), leading to the hypothesis that multiple rounds of cell division may be required to render chromosomal loci associated with T cell effector function accessible to specific transcription factors induced during earlier activation (10, 12, 42). Our data show that this division-associated trend in T cell effector differentiation (i.e., cytokine production and proliferation) is stably maintained within the pool of relatively long-lived cells present after a period of primary stimulation and rest in vitro, and is again operative when these cells reencounter Ag.
We show that those CD4+ T cells that fail
entirely to divide in response to primary activation exist in a
hyporesponsive state that is refractory to TCR ligation, and is
associated with a quantitative defect in TCR-coupled Ras-Raf-MAPK
signal transduction. A rest period of at least 2 days following primary
stimulation is required for the development of this MAPK defect (our
unpublished observations), suggesting either that this defect arises
only after cessation of primary signal transduction, or possibly that
cell death due to growth factor withdrawal may select for cells that
exhibit this phenotype. However, this defect is selective, as although
TCR-coupled MAPK activation is attenuated in these cells, the
TCR-coupled calcium response appears intact. In normal T cells, the
most TCR-proximal events following Ag recognition are initiated by the
nonreceptor tyrosine kinases Lck and ZAP-70 (43). The
Ras-Raf-MAPK cascade is coupled to the TCR via adaptor molecules such
as linker of activated T cells (LAT), SH2 domain-containing leukocyte
protein of 76 kDa (SLP-T6), and Grb2/Sos (44), and
involves initial events mediated by the SH3 domain of Lck
(45). TCR-induced increase in
[Ca2+]i, which is
mediated by phosphatidylinositol, is dependent on the activation of
PLC-, and is coupled to the TCR through ZAP-70 and Rho/Vav
(45, 46). TCR-coupled PLC- activation, generation of
phosphatidylinositol, and
[Ca2+]i elevation are not
dependent on the SH3 domain of Lck (45). The selective
defect in MAPK activation, but not PLC- activation, in the undivided
CD4+ T cells suggests a defect in the coupling of
the SH3 domain of Lck to the Ras-Raf-MAPK pathway. Also, we observe in
this study that signal transduction through the costimulatory receptor
CD28, which has been shown to augment TCR-coupled ERK activation in
normal T cells (27), and phorbol ester, which bypasses the
most proximal TCR-coupled events through the direct activation of PKC,
both restored maximal activation of the MAPK cascade in the undivided T
cells. This also suggests that the defect is not in the MAPK pathway
itself, but rather is in a pathway that couples Ras-Raf-MAPK to the
TCR. PI3K, Raf, and mitogen-activated
protein/extracellular signal-related kinase kinase are capable of
binding to the SH3 domain of Lck (47, 48, 49), suggesting that
uncoupling of these factors from Lck in undivided cells could result in
this selective defect in MAPK activation. The ability of PMA/ionomycin
to induce proliferation in these cells (see below) may also suggest
that PKC-mediated activation of the NF-
B pathway, which is required
for the production of IL-2, could be defective in these cells.
Alternatively, such negative regulatory factors as Ras-GAP and c-Cbl
also bind to SH3 domain of Lck (50, 51), suggesting that
selective coupling of these factors to the TCR in undivided cells, but
not divided cells, could likewise explain this selective MAPK defect.
Differential coupling of negative regulatory molecules to the TCR has
been demonstrated previously in anergic human T cell clones, in which
the TCR is coupled not to Ras, but to Rap1 (25), a Ras
homologue that binds Raf, but is unable to activate the MAPK cascade.
Interestingly, the Ras-Raf-MAPK defect exhibited by the undivided cells
in this model appears to preferentially affect the ERK1 isoform, as
TCR-coupled activation of ERK2, but not ERK1, remains relatively
intact. The presence of a significant amount of activated ERK2 in the
undivided pool following TCR ligation might suggest that the
Ras-Raf-MAPK pathway is not significantly compromised in these cells;
however, recent studies in ERK1-/- mice have
defined an obligatory role for ERK1 in T cell activation that cannot be
fulfilled by ERK2 (52). Together, these results support a
model in which a selective defect in ERK1 might result in a highly
attenuated response to TCR ligation.
The inability of maximal TCR-coupled calcium and Ras-Raf-MAPK signaling
to overcome the proliferative defect in the undivided
CD4+ T cells suggests the presence of further
defect(s) in growth factor-mediated signal transduction. Unlike the
defect in TCR-coupled MAPK activation, a rest period following primary
stimulation is not required for the development of this IL-2 refractory
phenotype (our unpublished observations). CD4+ T
cells from both the divided and undivided pools were able to express
all components of the high-affinity IL-2R in response to both TCR
engagement and IL-2 itself, demonstrating that the inability of the
undivided subset to proliferate is not due to a lack of growth factor
receptor expression. This suggests that the integrity of signal
transduction from the IL-2R must be coupled to cell division. Two major
IL-2R-coupled signaling pathways have been reported to be absolutely
necessary for the transduction of IL-2-mediated proliferative signals
in T cells. One pathway, which is specifically coupled to the
c-chain of the IL-2R, involves the
JAK3-mediated activation of the transcription factor STAT5
(31), while the other pathway is coupled to the IL-2R
ß-chain and involves the PI3K-mediated
activation of PKB/Akt (32). All T cells were able to
phosphorylate STAT5 in response to IL-2 stimulation, regardless of
their prior division history, suggesting that the defect in
IL-2R-coupled signal transduction does not reside in this JAK/STAT
pathway. PI3K is an integral component of many
growth factor receptor signal transduction pathways (53),
and is absolutely required for normal T cell responses
(54). D3 phosphoinositides generated by
PI3K recruit PKB/Akt and
phosphoinositide-dependent kinase-1 (PDK1) to the cell membrane, in
which PDK1 phosphorylates and activates PKB/Akt (55). The
PKB/Akt kinase is also a crucial signal transduction component for many
growth factor receptors (56, 57), and has been shown to be
both necessary and sufficient for the down-regulation of
p27kip1, hyperphosphorylation of the
retinoblastoma tumor suppressor protein, and release of active E2F in T
cell lines (35). However, the mechanism by which PKB/Akt
activity leads to these downstream effects is unclear. Our studies show
that CD4+ T cells that fail to divide following
activation are neither able to down-regulate
p27kip1 nor able to proliferate in response to
restimulation with IL-2, suggesting that PI3K
and/or PKB/Akt are not coupled properly to the IL-2R, or that another
biochemical step(s) leading to p27kip1 is not
coupled properly to PKB/Akt in these cells. In this sense, the defect
in IL-2 responsiveness exhibited by the nonproliferative
CD4+ T cell subset resembles the phenotype of T
cells treated with the pharmacologic agent rapamycin (34),
which blocks IL-2R signal transduction by binding to mammalian target
of rapamycin (mTOR)/FK506-binding protein-rapamycin-associated protein
(FRAP) (58), a downstream target of PKB/Akt.
Interestingly, receptor-independent activation of PKC in the undivided
CD4+ T cells using phorbol ester and calcium
ionophore resulted in efficient down-regulation of
p27kip1 and normal proliferative capacity. These
results suggest that the IL-2 refractory phenotype of the previously
undivided CD4+ T cells is due to a biochemical
defect that lies somewhere upstream of the action of PKC within the
PI3K-PKB/Akt pathway. PKC is normally coupled to
the IL-2R through PI3K, as PKC is activated by
PDK1-dependent phosphorylation (40), and a role for PKC in
IL-2-mediated signal transduction has been defined using the
IL-2-dependent cell line CTLL (59, 60). Interestingly,
full activation of the PKC-
isoform requires the activity of
mTOR/FRAP (61), which is in turn dependent on PKB/Akt
(62). These two distinct
PI3K-dependent pathways leading to PKC
activation, which can be bypassed using phorbol esters, represent a set
of biochemical events that could potentially be defective in the
undivided pool. Additionally, PKC can phosphorylate other downstream
targets of PKB/Akt, including GSK-3 and cAMP response element binding
protein (56). Therefore, the defect in IL-2-mediated
signal transduction exhibited by the undivided
CD4+ T cells in this model might result
specifically from the failure to activate downstream targets of PKB/Akt
or PKC.
The phenotype of the CD4+ T cells that fail to divide following primary activation, i.e., an inability to secrete IL-2 and proliferate upon TCR reengagement, closely resembles that of clonal anergy, a phenomenon normally associated with TCR occupancy in the absence of CD28 costimulation (15, 63, 64). The cell division-associated hyporesponsive state that we observe also shares similarities with clonal anergy at the molecular level. Similar to the undivided CD4+ T cells described in this work, both mouse and human T cell clones rendered anergic by TCR engagement in the absence of CD28 costimulation exhibit a normal increase in intracellular calcium following TCR engagement (15), but suffer from quantitative defects in Ras-coupled MAPK activation (22, 23, 24, 25). Also, the cyclin-dependent kinase inhibitor p27kip1, which we show is highly elevated in the CD4+ T cells that fail to divide after activation, has recently been shown to function as a molecular anergy factor in human T cell clones stimulated in the absence of CD28 costimulation (65). In addition, the authors discovered that p27kip1 could actively inhibit IL-2 gene transcription by sequestering the AP-1 coactivator JAB1, an activity not previously attributed to this cyclin-dependent kinase inhibitor. Therefore, the inability of the undivided CD4+ T cells in our model to down-regulate p27kip1 in response to IL-2 could explain not only why these cells are unable to progress from the G1 phase to the S phase of the cell cycle, but when paired with the coincident defect in TCR-coupled ERK1 activation, could also explain why these cells are unable to produce IL-2 as well. Finally, the combination of phorbol ester and calcium ionophore, which was able to overcome the molecular and functional defects exhibited by the undivided T cells in this study, has been shown to reverse anergy in T cell clones (15).
While these two refractory states share many functional and biochemical
similarities, they differ in at least two respects. As mentioned above,
classical T cell clonal anergy results from TCR occupancy in the
absence of CD28 costimulation, whereas the division-associated
unresponsiveness described in this work occurs despite the presence of
sufficient CD28 costimulatory signals. Second, the failure of anergic T
cell clones to proliferate in the models described above is due to a
defect in the production, but not in the utilization, of IL-2. The
provision of IL-2 during restimulation in both models restored clonal
responsiveness (30, 65). This is in contrast to the
phenotype of the undivided CD4+ T cells described
in this work, which not only fail to produce IL-2 during restimulation,
but are also refractory to IL-2 provided exogenously. This difference
may point to an important distinction between the division-associated,
hypoproliferative state and clonal anergy induced by TCR occupancy in
the absence of costimulation. Alternatively, these two phenotypes could
arise from analogous biochemical circumstances, and the differential
IL-2 responsiveness may reflect differences in the capacity of
IL-2R-coupled signal transduction pathways (e.g.,
PI3K or PKB/Akt) to remain poised in primary T
cells vs long-term clones. Notably, IL-2 unresponsiveness has also been
observed in T cells from mice chronically treated with staphylococcal
superantigen (66). However, in this case, IL-2
unresponsiveness apparently resulted from a defect in
c-coupled activation of the JAK3-STAT5
pathway.
These data show that cell division or a process associated with cell
cycle progression controls the ability of a T cell to respond to both
Ag and growth factor by modulating the integrity of the signals
transduced through both the Ras-Raf-MAPK cascade and the
PI3K, PKB/Akt pathway. This further suggests that
activated T cells must proliferate to avoid the induction of anergy.
The hypothesis that cell cycle progression may be required for anergy
avoidance by T cells was originally proposed by Schwartz and Jenkins
(15, 67). These investigators have suggested that CD28
costimulation promotes anergy avoidance indirectly by mediating
efficient IL-2 production. In this scenario, it is the subsequent
IL-2-driven cell division that allows T cells to escape anergy. Several
lines of evidence, including our results in this study, support this
model. Partial agonist ligands, which signal through the TCR, but do
not cause proliferation, induce anergy in T cells despite the presence
of CD28 costimulatory signals (68). Furthermore, whether a
given peptide ligand induces productive activation or anergy correlates
with its capacity to induce IL-2 (and proliferation), not earlier
signaling patterns such as TCR-
phosphorylation (69).
Second, overt blockade of IL-2-mediated signal transduction, using
either anti-IL-2/IL-2R Abs (67) or the pharmacologic
agents butyrate (70) or rapamycin (71),
induces anergy. Interestingly, primary T cells arrested with butyrate
during in vitro stimulation with agonistic anti-CD3 and
anti-CD28 Abs are also rendered anergic, and like the T cells that
naturally fail to divide during primary stimulation, these
butyrate-treated cells are also refractory to IL-2 (our unpublished
results). These results are consistent with previous studies using T
cells clones; however, not all pharmacological inhibitors of cell cycle
induce anergy in T cells (70, 71) (our unpublished
observations). More specifically, anergy avoidance may be associated
with the G1 to S phase transition, as opposed to
mitosis per se, as drugs that block mitosis, but allow
G1 to S phase transition (e.g., hydroxyurea), do
not induce anergy (71).
The use of primary T cells isolated from normal mice has allowed us to
assess the size of the nonresponsive pool within a normal T cell
repertoire. Surprisingly, this pool represents between 30 and 40% of
the mature, peripheral T cells in the mouse. Although we believe that
this population is generated during in vitro stimulation (see below),
because our analysis involves a heterogeneous starting T cell
population, we cannot eliminate the possibility that the
nonproliferative cells in our study represent a subset of the
peripheral CD4+ T cell repertoire with previous
antigenic experience, or that has otherwise been rendered anergic in
vivo before our analysis. However, several lines of evidence argue
against this deterministic model. First, the use of mitogenic Abs
against the monotypic CD3 component of the TCR effectively bypasses the
Ag specificity, and therefore the affinity, of the individual T cells.
We have also detected this nonproliferative subset in populations of
TCR-transgenic T cells stimulated in vitro and in vivo with specific
peptide (9). These data suggest that the nonproliferative
population is not merely a subset of the peripheral repertoire with
reduced TCR affinity. We have addressed the possibility that the
nonproliferative pool represents a subset of T cells with previous
antigenic exposure by fractionating the peripheral T cell repertoire of
normal mice and D011.10 TCR transgenic mice into naive and memory
subsets based on surface phenotype. In these experiments, both pools
exhibited a similar degree of proliferative heterogeneity and gave rise
to both divided and undivided cells after stimulation through TCR/CD28
(our unpublished observations). This suggests to us that the
nonproliferative pool is not comprised preferentially of cells with any
given antigenic history. Our previous studies using Ag-specific T cells
from TCR-transgenic Rag2-/- mice also argue
against this possibility. These cells represent a population of purely
clonal, naive T cells, and despite this, these cells still exhibit
marked heterogeneity in proliferative behavior at the single cell
level, and although they exhibit higher responder frequencies than T
cells from recombination-competent mice, the few
Rag2-/- D011 T cells that fail to divide in
response to antigenic peptide also fail to produce IL-2 upon
restimulation (9). Finally, the studies described above
using cell cycle inhibitors (butyrate and rapamycin) show that T cells
that would normally proliferate in response to TCR/CD28 stimulation can
be rendered unresponsive by overtly blocking their cell cycle
progression. For these reasons, we favor a stochastic model in which
any given naive T cell has the potential to respond to an antigenic
stimulus by dividing, but that the probability that any given cell will
achieve this goal is <1; from our data, we estimate this number to be
between 0.5 and 0.7. The failure to divide in this scenario fixes the
cell in a state that is refractory to further stimulation. Support for
this stochastic model can be observed in our data in this study, which
show that even among previously divided cells, a sizeable proportion
(25%) fails to divide upon restimulation; i.e., a pure population
of responders can give rise to both responders and nonresponders (see
Fig. 1). This indicates that the responsive phenotype can be transient,
and leaves open the possibility that random events that control gene
expression in complex systems may regulate proliferation
(72). However, whether the origin of this nonresponsive
population is stochastic or deterministic, its presence is remarkable,
as it represents approximately one-third of the peripheral
CD4+ T cell pool in a normal mouse.
The results reported in this work, together with several recent studies (9, 10, 11), suggest that secondary T cell responses are influenced by a mitotic clock (73, 74). Specifically, T cells that fail to divide following activation are refractory to secondary stimulation, while T cells that do proliferate remain responsive. Additionally, those cells that progress through many division cycles during primary stimulation tend to exhibit better secondary responses than those T cells that divide fewer times. Thus, the role of T cell clonal expansion in generating effective Ag-specific immunity may be 2-fold: The progression of naive T cells through a phase of exponential growth not only increases the frequency of Ag-specific T cells, but may also ensure that the resultant T cell pool consists preferentially of those T cells that have undergone multiple rounds of cell division and therefore carry the greatest potential to respond to future antigenic challenge.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. Laurence A. Turka, University of Pennsylvania, 700 Clinical Research Building, 415 Curie Boulevard, Philadelphia, PA 19104-6144.
3 Abbreviations used in this paper: CFSE, 5- and 6-carboxyfluorescein diacetate succinimidyl ester; [Ca2+]i, intracellular Ca2+ concentration; ERK, extracellular signal-related kinase; JAK, Janus kinase; MAPK, mitogen-activated protein kinase; MFI, mean fluorescence intensity; PDK, phosphoinositide-dependent kinase; PI3K, phosphatidylinositol 3-kinase; PKB, protein kinase B; PKC, protein kinase C; c, common -chain; LAT, linker of activated T cells; SLP-76, SH2 domain-containing leukocyte protein of 76 kDa; mTOR, mammalian target of rapamycin; FRAP, FK506-binding protein-repamycin-associated protein.
Received for publication September 7, 1999. Accepted for publication June 15, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and ZAP70. J. Exp. Med. 186:1645.
) or anti-CD3 (•). Values are
shown graphed as a function of primary division cycle. The results
shown are representative of six independent experiments.
) or had remained undivided throughout primary stimulation (
