Noncanonical NF-kappaB regulates inducible costimulator (ICOS) ligand expression and T follicular helper cell development.

Abstract

Follicular helper T (Tfh) cells have a central role in mediating humoral immune responses. Generation of Tfh cells depends on both T-cell intrinsic factors and the supporting function of B cells, but the underlying molecular mechanisms are incompletely understood. Here we show that NF-κB-inducing kinase (NIK), a central component of the noncanonical NF-κB signaling pathway, is required for Tfh cell development. Unlike other known Tfh regulators, NIK acts by controlling the supporting function of B cells. NIK and its upstream BAFF receptor regulate B-cell expression of inducible costimulator ligand (ICOSL), a molecule required for Tfh cell generation. Consistently, injection of a recombinant ICOSL protein into NIK-deficient mice largely rescues their defect in Tfh cell development. We provide biochemical and genetic evidence indicating that the ICOSL gene is a specific target of the noncanonical NF-κB. Our findings suggest that the noncanonical NF-κB pathway regulates the development of Tfh cells by mediating ICOSL gene expression in B cells.

Full text

Page 1

Noncanonical NF-κB regulates inducible costimulator
(ICOS) ligand expression and T follicular helper
cell development
Hongbo Hua,1, Xuefeng Wua,1,2, Wei Jina,3, Mikyoung Changa, Xuhong Chenga, and Shao-Cong Suna,b,4
aDepartment of Immunology, University of Texas MD Anderson Cancer Center, Houston, TX 77030 andbUniversity of Texas Graduate School of Biomedical
Sciences, Houston, TX 77030
Edited by Michael Karin, University of California San Diego School of Medicine, La Jolla, CA, and approved June 27, 2011 (received for review April 12, 2011)
Follicular helper T (Tfh) cells have a central role in mediating
humoral immune responses. Generation of Tfh cells depends on
both T-cell intrinsic factors and the supporting function of B cells,
but the underlying molecular mechanisms are incompletely un-
derstood. Here we show that NF-κB–inducing kinase (NIK), a cen-
tral component of the noncanonical NF-κB signaling pathway, is
required for Tfh cell development. Unlike other known Tfh regu-
lators, NIK acts by controlling the supporting function of B cells.
NIK and its upstream BAFF receptor regulate B-cell expression of
inducible costimulator ligand (ICOSL), a molecule required for Tfh
cell generation. Consistently, injection of a recombinant ICOSL pro-
tein into NIK-deficient mice largely rescues their defect in Tfh cell
development. We provide biochemical and genetic evidence indi-
cating that the ICOSL gene is a specific target of the noncanonical
NF-κB. Our findings suggest that the noncanonical NF-κB pathway
regulates the development of Tfh cells by mediating ICOSL gene
expression in B cells.
H
lymphoid organs and subsequent B-cell differentiation events,
such as antibody isotype switching and selection of high-affinity
B-cell clones (1). The successful progress of antibody responses
requires cognate help of the antigen-stimulated B cells by a
special CD4 T-cell subset, termed T follicular helper (Tfh) cells
(1, 2). These T cells express the chemokine receptor CXCR5 and
thus are capable of migrating to the B-cell follicles for efficient
T cell–B cell interaction. Tfh cells then direct the differentiation
of B cells by secreting cytokines, such as IL-21, and expressing
surface molecules such as CD40 ligand (CD40L) and pro-
grammed death 1 (PD1) (2). In addition, Tfh cells characteristi-
cally express high levels of inducible costimulator (ICOS), which
delivers a major T-cell costimulatory signal in response to ligation
by ICOS ligand (ICOSL; also termed B7h and B7RP-1) (3).
The development of Tfh cells is a multistep process, including
the initial CD4 T-cell activation by dendritic cells in the T-cell
zone and the subsequent interaction of Tfh precursor cells with
B cells at the T–B border of peripheral lymphoid organs (4). In
addition to the T-cell receptor (TCR) and CD28 signals, re-
quired for T-cell activation, the costimulatory signal mediated by
ICOS is critical for Tfh cell production (5–8). Because B cells
constitutively express high levels of ICOSL (9), the T cell–B cell
interaction may provide an important mechanism of ICOS cos-
timulation on T cells. Indeed, B cells are known as essential
supporting cells in the development of Tfh cells (2, 4). Several
recent studies suggest that the T cell–Bcell interaction is critical
for Tfh cell development (10–12), and that this supporting
function of B cells requires their surface expression of ICOSL
(7). However, the signaling pathways mediating the homeostatic
expression of ICOSL and the Tfh-supporting function of B cells
are poorly defined.
The NF-κB signaling pathway has an important role in regu-
lating lymphocyte development and activation (13–15). NF-κB
comprises a family of transcription factors, including RelA,
RelB, c-Rel, NF-κB1 p50, and NF-κB2 p52, which form different
dimeric complexes and transactivate target genes by binding to
umoral immune responses to protein antigens involve ger-
minal center (GC) formation in B-cell follicles of peripheral
a κB enhancer (16, 17). NF-κB is normally sequestered in the
cytoplasm by inhibitory proteins (IκBs), and NF-κB activation
typically involves inducible degradation of IκBα and nuclear
translocation of p50/RelA and p50/c-Rel NF-κB dimers. The
IκBα degradation is in turn triggered through its phosphorylation
by an IκB kinase (IKK) complex, composed of IKKα and IKKβ
as well as a regulatory subunit, NEMO (NF-κB essential mod-
ulator) (16, 17). In addition to this canonical pathway of NF-κB
activation, a noncanonical NF-κB pathway mediates specific
functions of NF-κB in certain cell types, including B cells (18,
19). This pathway depends on inducible processing of the NF-
κB2 precursor protein p100 (20, 21). Because p100 contains an
IκB-homologous C-terminal portion, it functions as not only the
precursor of p52, but also an IκB-like molecule that specifically
inhibits a noncanonical NF-κB member, RelB. Thus, the in-
ducible processing of p100 serves to both generate p52 and in-
duce the nuclear translocation of the noncanonical NF-κB dimer
p52/RelB. The activated RelB also can function as heterodimers
with other NF-κB members, particularly p50.
A central signaling component of the noncanonical NF-
κB pathway is NF-κB–inducing kinase (NIK), which integrates
noncanonical NF-κB–stimulating signals from a subset of TNF
receptor (TNFR) family members, including CD40, B-cell acti-
vating factor belonging to TNFR family receptor (BAFFR), and
lymphotoxin-β receptor (LTβR) (18, 19, 22). NIK and its down-
stream kinase IKKα stimulate p100 processing by mediating p100
phosphorylation and ubiquitination (20, 21). A major function of
the noncanonical NF-κB pathway is regulating humoral immune
responses. Deficiency in NIK or other components of this path-
way attenuates GC formation and production of antibodies (18,
23). However, how precisely NIK and the noncanonical NF-κB
signaling pathway regulate antibody responses is incompletely
understood. In this study, we have demonstrated that NIK has
a critical role in antigen-stimulated generation of Tfh cells. This
function of NIK is not T-cell intrinsic but is mediated through
regulating the supporting role of B cells. Interestingly, NIK is
required for maintaining the high-level expression of ICOSL in B
cells. We provide genetic and biochemical evidence that non-
canonical NF-κB members are directly involved in ICOSL gene
regulation. Thus, our data suggest that the noncanonical NF-κB
signaling pathway regulates Tfh cell development by controlling
ICOSL gene expression in B cells.
Author contributions: H.H. and X.W. designed research; H.H., X.W., W.J., M.C., and X.C.
performed research; H.H. and X.W. analyzed data; and S.-C.S. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1H.H. and X.W. contributed equally to this work.
2Present address: Laboratory of Gene Regulation and Signal Transduction, Department of
Pharmacology, School of Medicine, University of California San Diego, La Jolla, CA 92093.
3Present address: Howard Hughes Medical Institute and Immunology Program, Memorial
Sloan-Kettering Cancer Center, New York, NY 10065.
4To whom correspondence should be addressed. E-mail: ssun@mdanderson.org.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1105774108/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1105774108PNAS
| August 2, 2011
| vol. 108
| no. 31
| 12827–12832
IMMUNOLOGY

Page 2

Results
NIK Regulates Tfh Cell Development in a B-Cell–Dependent Manner.
To understand how the noncanonical NF-κB signaling pathway
regulates humoral immune responses, we investigated the role of
NIK in Tfh cell development. We immunized WT and NIK KO
mice with a strong protein antigen, sheep red blood cells
(SRBC), which induces robust GC formation and Tfh cell de-
velopment in the absence of adjuvants (5, 24). We detected Tfh
cells in the spleen of the immunized mice by flow cytometry,
based on their typical surface markers CXCR5 and PD1. As
expected, the spleen of immunized WT mice produced a clear
population of Tfh cells characterized by a CXCR5+PD1+phe-
notype (Fig. 1A). Importantly, the generation of Tfh cells was
attenuated in the spleen of NIK KO mice (Fig. 1A). Analyses of
multiple animals showed thatthe NIK KOmice had a significantly
lower percentage of Tfh cells out of total CD4 T cells (Fig. 1B).
The development of Tfh cells requires a complex signaling
program in activated CD4 T cells. B cells also play a critical role
in supporting the differentiation of CD4 T cells to Tfh cells (4).
To understand how NIK regulates Tfh cell differentiation, we
performed lymphocyte adoptive transfer studies to examine
whether this function of NIK is in T cells or in B cells. Purified T
and B cells were transferred to Rag2 KO recipient mice, which
lack endogenous lymphocytes. After lymphocyte transfer, the
recipient mice were immunized with SRBC and then subjected
to Tfh cell analysis. As expected, the mice transferred with WT
T cells plus WT B cells efficiently developed Tfh cells after im-
munization (Fig. 1C). The transfer of NIK KO T cells plus WT B
cells also was associated with the effective development of Tfh
cells in recipient mice, suggesting that the function of NIK in Tfh
cell regulation is not T-cell intrinsic. On the other hand, the mice
transferred with WT T cells plus NIK KO B cells exhibited se-
riously defective Tfh cell production (Fig. 1C). Taken together,
these results indicate that NIK regulates Tfh cell development by
modulating the supporting function of B cells.
NIK and Its Upstream Receptor BAFFR Regulate ICOSL Expression in B
Cells. ICOS/ICOSL interaction is crucial for the development of
Tfh cells (5–7, 25). In particular, ICOSL is highly expressed on B
cells and is involved in ICOS stimulation during the interaction
of B cells and CD4 T cells (7). Given that B cells are constantly
exposed to noncanonical NF-κB stimuli, particularly BAFF, we
reasoned that the noncanonical NF-κB pathway might contribute
to the high levels of ICOSL expression in B cells. To test this
hypothesis, we examined the expression of ICOSL on splenic B
cells derived from WT and NIK KO mice. As expected, freshly
isolated WT B cells displayed constitutive ICOSL expression
(Fig. 2A). Moreover, the expression level of ICOSL was sub-
stantially reduced, but not completely blocked, in the NIK-de-
ficient B cells (Fig. 2A).
Because homeostatic activation of NIK and noncanonical NF-
κB in splenic B cells is mediated primarily by the BAFF/BAFFR
system (26), we tested whether ICOSL expression is also subject
to regulation by BAFFR. For these studies, we used the control
A/J mouse and its mutant variant, A/WySnJ, which carries a ge-
netic defect in the BAFFR gene (27). As seen with the NIK KO
B cells, the B cells isolated from A/WySnJ mice had a signifi-
cantly lower level of ICOSL expression (Fig. 2B). These results
indicate that the noncanonical NF-κB signaling pathway, which
is chronically activated in vivo, mediates induction of the high-
level expression of ICOSL in B cells, although other mechanisms
contribute to this gene expression event as well.
NIK Mediates in Vitro ICOSL Induction by Noncanonical NF-κB
Inducers. To further assess the role of NIK in mediating the in-
duction of ICOSL expression, we examined whether non-
canonical NF-κB inducers stimulate ICOSL expression in vitro,
and whether NIK is required for this gene induction event. For
these studies, we incubated the B cells in vitro for 48 h to reduce
the level of constitutive ICOSL expression. Consistent with the
need for BAFFR in maintaining constitutive ICOSL expression
in vivo, incubation of WT B cells with BAFF led to potent in-
duction of ICOSL expression in vitro (Fig. 2C). Importantly, the
BAFF-stimulated ICOSL expression was dependent on NIK, as
demonstrated by its absence in NIK KO B cells. Furthermore,
stimulation of CD40, a noncanonical NF-κB inducer mediating
T-cell–dependent B-cell activation during an immune response,
also led to induction of ICOSL in a NIK-dependent manner (Fig.
2C). In contrast, the canonical NF-κB inducer LPS failed to in-
duce ICOSL expression in both the WT and KO B cells (Fig.
2C). These results indicate a specific role for the noncanonical
NF-κB pathway in mediating the induction of ICOSL expression.
We also performed studies using the murine M12 B-cell line
as a model system, because it had been well characterized
for noncanonical NF-κB activation by the B-cell–specific non-
canonical NF-κB inducers BAFF and anti-CD40 (28, 29). The
surface expression of ICOSL was strongly induced by stimulation
with either BAFF or anti-CD40 (Fig. 2D). Consistent with the
persistent nature of noncanonical NF-κB signaling, the induction
of ICOSL was prolonged, with 48 h as the optimal induction time
(Fig. 2D). As seen with primary B cells, the expression of ICOSL
in M12 B cells was induced only slightly by the canonical NF-κB
inducer LPS (Fig. 2D). These results indicate the involvement of
the noncanonical NF-κB pathway in the induction of ICOSL
gene expression. This idea was further suggested by parallel real-
time quantitative PCR (qPCR) analyses showing the induction of
ICOSL mRNA by BAFF and anti-CD40, but not by LPS (Fig. 2E).
Noncanonical NF-κB Binds Directly to ICOSL Promoter and Mediates
the Induction of ICOSL Gene Expression. To examine whether
ICOSL serves as a direct target of noncanonical NF-κB, we
performed ChIP assays, a technique that detects in vivo binding
Fig. 1.
(A and B) Age-matched NIK+/+(WT) and NIK−/−(KO) mice were immunized
with SRBC and killed on day 7 after immunization. The frequency of Tfh cells
among CD4 T cells was quantified by flow cytometry and presented as
a representative flow cytometry graph (A) and the mean value of multiple
mice (with each circle or square representing an individual mouse) (B). (C)
Rag2 KO mice were adoptively transferred with a combination of T and
B cells derived from either WT or NIK KO mice. The recipient mice were
immunized with SRBC antigen and subjected to Tfh cell analyses as described
in A and B. Data are presented as mean value of multiple recipient mice.
NIK regulates Tfh cell development in a B-cell–dependent manner.
12828
| www.pnas.org/cgi/doi/10.1073/pnas.1105774108 Hu et al.

Page 3

of transcription factors to the regulatory regions of target genes.
Analysis of the murine and human ICOSL gene locus revealed
two major conserved noncoding sequence (CNS) elements, one
located between −800 and −30 nucleotides and the other located
between −2,300 and −2,100 nucleotides relative to the tran-
scription start site (Fig. S1). We first performed sequential ChIP
assays to examine which regions were bound by RelB, the core
component of the noncanonical NF-κB complex. As expected, in
nonstimulated cells (NT), RelB did not bind to any of the CNS
regions (Fig. 3A). Interestingly, on BAFF stimulation, RelB was
bound to the promoter-proximal region, as demonstrated by its
pull down of a DNA fragment spanning −434 to −190 (Fig. 3A).
In contrast, the upstream regions (−2,371 to −2,150 and −882
to −675) did not appreciably bind RelB. Additional ChIP anal-
yses using both primary spleen B cells and the M12 B-cell line
demonstrated that in addition to RelB, p50 and p52 also bound
to the promoter-proximal region of ICOSL (Fig. 3B). On the
other hand, we detected very weak binding of the ICOSL pro-
moter by RelA, the core component of the canonical NF-κB
complex. These results thus demonstrate that the inducible ex-
pression of ICOSL in B cells is associated with binding of non-
canonical NF-κB members to the ICOSL promoter.
RelB is the core subunit of the noncanonical NF-κB that
functions as a heterodimer with either p52 or p50 (22). To
functionally examine the requirement of the noncanonical NF-
κB in ICOSL gene induction, we performed RNAi-mediated
knockdown of RelB. We infected M12 B cells with a control
lentiviral vector, pLKO.1, or the same vector encoding a RelB
shRNA. Compared with the control cells, the RelB shRNA-
infected cells showed markedly lower RelB expression (Fig. 3C).
The RelB knockdown moderately reduced the basal level of
ICOSL expression and strongly attenuated the BAFF-induced
ICOSL expression (Fig. 3D). As an additional approach to
confirm the important role of the noncanonical NF-κB pathway
in ICOSL gene regulation, we reconstituted the NIK KO B cells
with an NIK expression vector via retroviral infection. Expres-
sion of exogenous NIK rescued both the basal (Fig. S2) and
BAFF-induced (Fig. 3E) ICOSL expression. Taken together with
the ChIP assay results, these findings identify ICOSL as a target
gene of the noncanonical NF-κB signaling pathway.
ICOSL Promoter Has a κB Site That Binds Noncanonical NF-κB
Members and Mediates ICOSL Promoter Activation. Through DNA
sequence analysis, we identified a κB-like element in the pro-
moter region of ICOSL (−347 to −338). This sequence differs
slightlyfromthetypicalκBconsensussequence, GGGRNTTTCC
(30) (Fig. 4A). EMSA revealed binding of this κB-like element by
nuclear proteins stimulated by BAFF and anti-CD40 (Fig. 4A).
The ICOSL κB-binding complexes also were detected in spleen
B cells isolated from WT mice. This κB-binding activity was de-
pendent on NIK,asdemonstrated by its absence in B cells derived
from the NIK KO mice (Fig. 4A). Parallel antibody supershift
assays revealed that the major NF-κB complexes formed with
the ICOSL κB site contained the noncanonical NF-κB members
p52 and RelB, as well as p50 (Fig. 4B); in contrast, little binding
activity of RelA and c-Rel was detected.
To further examine the specificity of the ICOSL in binding to
different NF-κB members, we performed EMSA with overex-
pressed NF-κB components. When expressed alone or together,
p50 and p52 bound to the ICOSL κB and common κB with
similar efficiency (Fig. 4C). In contrast, RelA failed to bind to
the ICOSL κB (Fig. 4C, Upper), despite its efficient binding to
the common κB probe (Fig. 4C, Lower). The RelA/p50 hetero-
dimer also barely bound to the ICOSL κB (Fig. 4C, Upper).
Furthermore, although RelB alone did not bind ICOSL κB or
common κB, in line with its inability to form stable homodimers
(31, 32), it bound to the ICOSL probe when expressed together
with p52 or p52 plus p50 (Fig. 4C, Upper). These results, together
with those presented in Fig. 4 A and B, suggest that ICOSL
contains a κB site that preferentially binds noncanonical NF-κB.
To examine whether this intriguing κB site is important for
BAFF-stimulated ICOSL promoter activity, we generated lucif-
erase reporters driven by either WT ICOSL promoter or the
same promoter harboring a mutation in the ICOSL κB site.
Consistent with the induction of ICOSL expression (Fig. 2), the
WT ICOSL promoter was stimulated by anti-CD40 and BAFF,
but not by LPS (Fig. 4D). Importantly, mutation of the κB site
abolished the ICOSL promoter activation (Fig. 4D). These
results identify ICOSL κB as a functional DNA element that
mediates the response to noncanonical NF-κB signals.
Injection of Recombinant ICOSL into NIK KO Mice Largely Rescues
Their Tfh Defect. The foregoing results establish ICOSL as a tar-
get gene of the noncanonical NF-κB signaling pathway and pro-
vide a possible molecular mechanism through which this NF-κB
signaling axis regulates Tfh cell development. To further validate
the functional significance of the ICOSL gene expression, we
tested whether recombinant ICOSL is able to partially or com-
Fig. 2.
isolated spleen B cells of WT and NIK KO mice were analyzed by flow
cytometry to determine the level of ICOSL expression. Data are represen-
tative of three independent experiments with multiple mice. (B) Spleen B
cells were isolated from control A/J mice or the BAFFR-deficient A/WySnJ
mice and subjected to flow cytometry analysis of ICOSL surface expression
level. (C) WT and NIK KO B cells were cultured in vitro for 48 h either in the
absence (NT) or the presence of BAFF, an agonistic anti-CD40 antibody, or
LPS. The intensity of ICOSL surface expression was measured by flow
cytometry. (D) M12 B cells were either not treated (NT) or stimulated for the
indicated times with BAFF, anti-CD40, or LPS. The ICOSL surface expression
was analyzed by flow cytometry. (E) WT and NIK KO spleen B cells were
stimulated in vitro as indicated. For the untreated control (NT), cells were
cultured for 12 h without inducers. Total RNA was isolated and subjected to
real-time qPCR to analyze the level of ICOSL mRNA. Data are presented as
fold relative to the NT sample.
NIK mediates the inducible expression of ICOSL on B cells. (A) Freshly
Hu et al.PNAS
| August 2, 2011
| vol. 108
| no. 31
| 12829
IMMUNOLOGY

Page 4

pletely rescue the defect of NIK KO mice in Tfh cell development.
A recombinant ICOSL-Fc fusion protein or a control Fc protein
was injected into NIK KO mice on the day of SRBC immunization
and at different times after the immunization, followed by flow
cytometry analysis of the generation of Tfh cells (Fig. 5A). As
expected, the control Fc-injected NIK KO mice displayed a sig-
nificantly reduced level of Tfh cell generation compared with the
WT mice (Fig. 5 B and C). Importantly, injection of recombinant
ICOSL-Fc largely (although not completely) rescued the defect of
the NIK KO mice in Tfh cell development (Fig. 5 B and C).
Because ICOSL-Fc might have an effect on non-B cells, we
performed additional ICOSL-Fc rescue experiments using the
lymphocyte adoptive transfer model. In brief, Rag2 KO mice
were adoptively transferred with WT T cells plus NIK KO B cells
and then subjected to ICOSL-Fc rescue studies (Fig. 5 D and E).
Under these conditions, ICOSL-Fc again efficiently rescued the
defect of the NIK KO B cells in supporting Tfh cell develop-
ment. Collectively, these results further emphasize the critical
role for NIK-mediated ICOSL expression in regulating Tfh cell
development.
Discussion
The results presented in this paper identify NIK and its down-
stream noncanonical NF-κB as critical factors in the regulation
of Tfh cell development. Unlike the currently known signaling
factors, which function in T cells, these factors regulate the
supporting role of B cells. We obtained biochemical and genetic
evidence indicating that the noncanonical NF-κB pathway reg-
ulates the expression of ICOSL, a costimulatory molecule re-
quired for stimulation of Tfh cell development. These findings
identify ICOSL as a target gene of the noncanonical NF-κB
signaling pathway and provide insight into the mechanism by
which this pathway regulates humoral immune responses.
Some previous studies have demonstrated the requirement for
ICOSL in the induction of Tfh cells. Genetic deficiency in ICOS
and ICOSL or blockade of ICOS/ICOSL interactions impairs
Tfh cell development in mice (5–8). Notably, the high-level ex-
pression of ICOSL on B cells is particularly important for B-cell–
mediated supporting function in the generation of Tfh cells (7).
In agreement with these previous studies, we found an associa-
tion between the attenuated expression of ICOSL in NIK-deficient
B cells and reduced production of Tfh cells in immunized NIK
KO mice. This defect is due to the impaired supporting function
of B cells, as demonstrated by adoptive transfer experiments.
Furthermore, we found that injection of a recombinant ICOSL
protein into the NIK KO mice largely rescued the defective an-
tigen-stimulated Tfh cell generation. Thus, our data further
emphasize the critical role of ICOSL/ICOS interaction in the
induction of Tfh cell differentiation.
B cells are characteristic for their constitutive expression of
high levels of ICOSL (9). Although the mechanism mediating
constitutive ICOSL expression in B cells has remained obscure,
this gene expression pattern is correlated with constitutive acti-
vation of NF-κB (33). Unlike T cells and many other cell types,
B cells are constantly exposed to homeostatic NF-κB stimuli in
peripheral lymphoid organs and display chronic NF-κB activity.
One important homeostatic NF-κB–inducing signal is triggered
through the binding of BAFF to BAFFR, and this signal pre-
dominantly stimulates the noncanonical NF-κB signaling path-
way (26, 29). We have shown that the BAFFR signal is critical
for the constitutive expression of ICOSL in B cells in vivo.
Consistently, BAFF stimulated ICOSL expression in vitro in
both primary B cells and the M12 B-cell line. Our in vitro studies
also suggest the involvement of the CD40 signal in ICOSL gene
induction. But because CD40 stimulation requires antigen-
stimulated T cells, the CD40 signal likely contributes to ICOSL
induction only during an immune response.
We found strong evidence suggesting an essential role for the
noncanonical NF-κB in ICOSL gene expression. First, inducers
that trigger the activation of noncanonical NF-κB, such as BAFF
and anti-CD40, effectively induce the expression of ICOSL in B
cells. In contrast, canonical NF-κB inducer LPS was insufficient
to trigger ICOSL expression in B cells. Moreover, RNAi-mediated
knockdown of the core noncanonical NF-κB member RelB at-
tenuated ICOSL gene induction, providing genetic evidence of
the need for noncanonical NF-κB in ICOSL gene induction.
Finally, our ChIP assays revealed the binding of noncanonical
NF-κB members to the promoter region of ICOSL, suggesting
their direct involvement in ICOSL gene regulation. Of course,
our data do not exclude the possibility that canonical NF-κB is
also involved in the regulation of ICOSL gene expression. In
particular, BAFF is known to stimulate p50 in addition to the
noncanonical NF-κB members p52 and RelB (34). We found
that p50 also binds to the ICOSL promoter in BAFF-stimulated
B cells. However, because RelA is not a major component of
the NF-κB complex bound to the ICOSL promoter, it is likely
that p50 may function as a homodimer or a partner of the
Fig. 3.
region of ICOSL gene and is critical for ICOSL in-
duction. (A) Spleen B cells from WT mice were either
not treated (NT) or stimulated with BAFF for 24 h.
Chromatin IP was performed using either a control
Ig (Ig) or anti-RelB antibody, and the precipitated
DNA was subjected to PCR using primers that am-
plify the indicated regions of the ICOSL promoter.
Input DNAs also were subjected to PCR to show the
efficiency of the primers. (B) WT spleen B cells (Left)
or M12 B cells (Right) were stimulated with BAFF for
the indicated times. Chromatin IP was performed
using either a control Ig or the indicated antibodies,
and the precipitated DNA was subjected to PCR us-
ing primers that amplify a 300-bp DNA fragment
(−490 to −190) of the ICOSL promoter. Data are
representative of three independent experiments. (C
and D) M12 cells were infected with either the
pLKO.1 lentiviral vector or the same vector encoding
RelB shRNA. After puromycin selection, the bulk of
infected cells were subjected to IB to determine the
efficiency of RelB knockdown (C). The control and
RelB knockdown cells were either not treated (NT) or
stimulated with BAFF for 48 h, and the ICOSL expression level was analyzed by flow cytometry (D). (E) WT or NIK KO splenocytes were infected with ret-
roviruses carrying the pCLXSN(GFP) vector (vector), or pCLXSN(GFP)-NIK (NIK). Infected cells were stimulated with BAFF for 48 h, and ICOSL expression on
infected B cells was analyzed by flow cytometry (gated on B220+GFP+cells).
Noncanonical NF-κB binds to the promoter
12830
| www.pnas.org/cgi/doi/10.1073/pnas.1105774108 Hu et al.

Page 5

noncanonical NF-κB member RelB, because RelB is known to
form both p52/RelB and p50/RelB heterodimers.
The specific involvement of noncanonical NF-κB in ICOSL
gene induction appears to be due to two different regulatory
mechanisms. First, noncanonical NF-κB members are the pre-
dominant components of chronically activated NF-κB complexes
in B cells exposed to the homeostatic inducer BAFF. Second, the
ICOSL promoter contains a κB element that favors binding by
the noncanonical NF-κB members. Mutation of this κB element
abolished the activation of ICOSL promoter by anti-CD40 and
BAFF. It is important to note, however, that the role of NF-κB in
ICOSL gene regulation appears to vary among different cell
types. Previous studies have suggested that canonical NF-κB
stimuli, such as IL-1 and TNF-α, induce the expression of ICOSL
in endothelial cells and fibroblasts (35–37). Our findings suggest
that LPS is inefficient in the induction of ICOSL expression in
B cells. Whether other canonical NF-κB stimuli induce ICOSL
expression in B cells remains to be investigated. Notwithstand-
ing, our findings suggest that the NIK-regulated noncanonical
NF-κB signaling pathway plays a predominant role in mediating
the high level of ICOSL expression in B cells, a signaling function
required for the supporting role of B cells in antigen-stimulated
production of Tfh cells.
Materials and Methods
Mice. NIK KO mice on a 129Sv/Ev background (38) were provided by Amgen
and were maintained in the specific pathogen-free facility of the University
of Texas MD Anderson Cancer Center. NIK+/−heterozygous mice were bred
to generate the age-matched NIK+/+(WT) and NIK−/−(KO) mice used in the
experiments. Rag2−/−(Rag2 KO) mice, on a 129Sv/Ev background, were
obtained from Taconic. A/J and A/WySnJ mice were obtained from Jackson
Laboratory. All animal experiments were performed in accordance with
protocols approved by the University of Texas MD Anderson Cancer Center’s
Institutional Animal Care and Use Committee.
Antibodies, Reagents, and Plasmids. Antibodies for p50 (D17), p52 (c-5), RelB
(C-19), c-Rel (sc-71×), and HSP60 (H-1), as well as control rabbit Ig, were
obtained from Santa Cruz Biotechnology. Fluorescence-labeled antibodies
for CD4 (L3T4), CD3 (145-2C11), PD-1 (J43), and ICOSL (HK5.3) were pur-
chased from eBioscience. Anti-mouse CD40 (553721) and fluorescence-
labeled antibodies for CD19 (1D3) and CXCR5 (2G8) were purchased from BD
Biosciences. Other antibodies were as reported previously (28).
Recombinant ICOSL-Fc (also called B7RP-1-Fc) fusion protein and control Fc
protein were provided by Amgen. Recombinant BAFF protein (PHC1674) was
purchased from Biosource. LPS (derived from E. coli 0127:B8) was obtained
from Sigma-Aldrich, and SRBC was purchased from Cocalico Biologicals.
The pLKO.1-puromycin lentiviral vector and the same vector encoding
mouse RelB shRNAs were purchased from Sigma-Aldrich. Three shRNAs
targeting different regions of the RelB mRNA were used. To generate the
luciferase reporter driven by the mouse ICOSL promoter (pGF-ICOSL), a 705-
bp ICOSL promoter DNA fragment (−570 to +135) was inserted upstream of
the luciferase gene in a lentiviral reporter plasmid, pGreenFire (pGF; System
Biosciences). pGF-ICOSLΔκB, a mutant form of pGF-ICOSL that contains point
mutations in a κB-like element of the ICOSL promoter (−347 to −338), was
created using the QuikChange Site-Directed Mutagenesis Kit (Stratagene).
The following primers were used: sense, 5′-CAGGGACCAGGCCGTTAAC-
GTTCTGGGCAGCGTTG-3′; antisense, 5′-CAACGCTGCCCAGAACGTTAACGGC-
CTGGTCCCTG-3′.
The pcDNA expression vectors encoding Flag-tagged p50, p52, and RelB
werepurchasedfromAddgene.ThepCMV4-p65wasdescribedpreviously(39).
Fig. 4.
canonical NF-κB members and mediates ICOSL promoter activation. (A) The
sequence of an ICOSL κB was aligned with the consensus κB sequence (Up-
per). EMSA was performed using the ICOSL κB probe and nuclear extracts
isolated from nontreated or BAFF- and anti-CD40-stimulated (24 h) M12 cells
or from freshly purified WT and NIK KO spleen B cells. (B) A supershift assay
was performed using nuclear extracts of BAFF-stimulated M12 B cells and
the ICOSL κB probe, in either the absence (none) or the presence of the
indicated antibodies or an Ig control. The supershifted bands are indicated
by arrows. (C) HEK293 cells were transfected with the indicated NF-κB
members, either alone or in combination. Nuclear extracts were subjected to
EMSA using the ICOSL κB and a general κB probe. Immunoblot analysis was
performed to monitor the expression of the different NF-κB proteins. (D)
M12 cells were infected with pGreenFire lentiviral vectors carrying a lucifer-
ease gene driven by either WT ICOSL promoter (ICOSL-luc) or mutant ICOSL
promoter with mutated κB site (ICOSLΔκB-luc). The cells were either not
treated (NT) or stimulated for 14 h with LPS, anti-CD40, or BAFF. Luciferase
activity is presented as fold induction compared with the NT pGF cells.
A κB sequence of the ICOSL promoter preferentially binds non-
Fig. 5.
WT and NIK KO mice were immunized with SRBC on day 0 along with in-
jection of a recombinant ICOSL-Fc fusion protein or a control Fc protein. The
mice received three additional injections of ICOSL-Fc or control Fc at the
indicated days postimmunization and then subjected to Tfh cell analyses.
(B and C) The frequency of Tfh cells among CD4 T cells was quantified by
flow cytometry and presented as a representative flow cytometry graph (B)
and mean value of multiple mice (C). (D and E) Rag2 KO mice were adop-
tively transferred with a combination of WT T cells and NIK KO B cells. The
recipient mice were immunized with SRBC along with injection with either
control Fc or ICOSL-Fc, as described in A. The frequency of Tfh cells among
CD4 T cells was quantified by flow cytometry and presented as a represen-
tative flow cytometry graph (D) and mean values (E).
Recombinant ICOSL rescues the Tfh cell defect in NIK KO mice. (A)
Hu et al. PNAS
| August 2, 2011
| vol. 108
| no. 31
| 12831
IMMUNOLOGY

Page 6

Cell Culture and shRNA Knockdown. MurineB-celllineM12.4.1(designatedM12
in this paper) was described previously (28). The cells were infected with len-
tiviruses carrying either the empty pLKO-1 vector or RelB shRNA clones. The
infected cells were then enriched by selection using puromycin (2.0 μg/mL) for
5 d, and the bulk of the infected cells were used in experiments. To produce
the lentiviral particles, the pLKO.1 vectors were transfected into HEK293 cells
(using the calcium method) along with packing vectors psPAX2 and pMD2
(provided by Dr. Xiaofeng Qin, MD Anderson Cancer Center, Houston, TX).
B cells were purified from splenocytes using anti-B220 conjugated mag-
netic beads (Miltenyl Biotec) and were either directly subjected to flow
cytometry or stimulated in vitro by anti-mouse CD40 (500 ng/mL), BAFF (200
ng/mL), or LPS (100 ng/mL).
Mouse Immunization and ICOSL-Fc Injection. For induction of Tfh cells, age-
matched WT and NIK KO mice were immunized i.p. with 2 × 109SRBC (24).
In some experiments, the NIK KO mice were injected i.p. with 50 μg of ICOSL-
Fc or 25 μg control Fc (40) on the day of immunization and on days 2, 4,
and 6 after immunization. On day 7, spleen cells were isolated for flow
cytometry analyses.
Lymphocyte Adoptive Transfer. B220+B cells and CD90.2+T cells were isolated
from the splenocytes of WT or NIK KO mice using magnetic beads (Miltenyi
Biotec). The isolated cells were >95% pure, as determined by flow cytom-
etry. WT or NIK KO T cells (5 × 106) were mixed with either WT or NIK KO B
cells (5 × 106) and then injected via a tail vein into Rag2 KO mice. After 16 h,
the recipient mice were subjected to immunization and ICOSL-Fc injection
studies as described above.
Flow Cytometry. Cell suspensions were subjected to flow cytometry analyses
as described previously (41) using a BD Biosciences LSRII flow cytometer. Data
were analyzed using FlowJo software.
ChIP Assays. ChIP assays were performed using the Millipore EZ-ChIP Kit
following the manufacturer’s instructions. In brief, nontreated and treated
M12 cells were crosslinked with 1% of formaldehyde (final concentration,
vol/vol) for 10 min, lysed in SDS lysis buffer, and sonicated to shear the DNA.
The chromatin DNA was subjected to IP using the indicated antibodies or
a control IgG. After purification, the precipitated DNA was analyzed by PCR
using primers that amplify different regions of the ICOSL promoter.
The primer sequences were as follows: −2371 to −2150, 5′-ACAGGTT-
GAGAACCATTCTTCC-3′ and 5′- GAATCCCAGAAAGCCAAATGC-3′; −882 to
−675, 5′-TAGCCTCAGACTCAAGAGATC-3′ and 5′-CCAGACTTGGCAATCCT-
GTTC-3′; −434 to −190, 5′-CCAGGTCCGGGCTTTGAACC-3′ and 5′-CATGAGT-
TACAGGTGCCAGGGTG -3′.
Statistical Analysis. Two-tailed unpaired t tests were performed using Prism
software. A P value < 0.05 was considered significant.
ACKNOWLEDGMENTS. We thank Amgen for the NIK KO mice and ICOSL-Fc
recombinant protein and Xiaofeng Qin for the lentiviral packaging vectors.
We also thank the personnel from the flow cytometry core facility (Karen
Martinez, David He, and Amy Cortez) and the animal facility at MD Ander-
son Cancer Center for technical assistance. This study was supported by
National Institutes of Health Grants AI057555, AI064639, GM84459-S1,
and GM84459.
1. Vinuesa CG, Tangye SG, Moser B, Mackay CR (2005) Follicular B helper T cells in an-
tibody responses and autoimmunity. Nat Rev Immunol 5:853–865.
2. King C, Tangye SG, Mackay CR (2008) T follicular helper (TFH) cells in normal and
dysregulated immune responses. Annu Rev Immunol 26:741–766.
3. Simpson TR, Quezada SA, Allison JP (2010) Regulation of CD4 T cell activation and
effector function by inducible costimulator (ICOS). Curr Opin Immunol 22:326–332.
4. Linterman MA, Vinuesa CG (2010) Signals that influence T follicular helper cell dif-
ferentiation and function. Semin Immunopathol 32:183–196.
5. Akiba H, et al. (2005) The role of ICOS in the CXCR5+follicular B helper T cell main-
tenance in vivo. J Immunol 175:2340–2348.
6. Bossaller L, et al. (2006) ICOS deficiency is associated with a severe reduction of
CXCR5+CD4 germinal center Th cells. J Immunol 177:4927–4932.
7. Nurieva RI, et al. (2008) Generation of T follicular helper cells is mediated by in-
terleukin-21 but independent of T helper 1, 2, or 17 cell lineages. Immunity 29:
138–149.
8. Hu YL, Metz DP, Chung J, Siu G, Zhang M (2009) B7RP-1 blockade ameliorates au-
toimmunity through regulation of follicular helper T cells. J Immunol 182:1421–1428.
9. Yoshinaga SK, et al. (1999) T-cell co-stimulation through B7RP-1 and ICOS. Nature 402:
827–832.
10. Haynes NM, et al. (2007) Role of CXCR5 and CCR7 in follicular Th cell positioning and
appearance of a programmed cell death gene-1 high germinal center–associated
subpopulation. J Immunol 179:5099–5108.
11. Johnston RJ, et al. (2009) Bcl6 and Blimp-1 are reciprocal and antagonistic regulators
of T follicular helper cell differentiation. Science 325:1006–1010.
12. Zaretsky AG, et al. (2009) T follicular helper cells differentiate from Th2 cells in re-
sponse to helminth antigens. J Exp Med 206:991–999.
13. Gerondakis S, Siebenlist U (2010) Roles of the NF-κB pathway in lymphocyte de-
velopment and function. Cold Spring Harb Perspect Biol 2:a000182.
14. Zhu M, Fu Y (2010) The complicated role of NF-κB in T-cell selection. Cell Mol Immunol
7:89–93.
15. Hayden MS, Ghosh S (2011) NF-κB in immunobiology. Cell Res 21:223–244.
16. Hayden MS, Ghosh S (2008) Shared principles in NF-κB signaling. Cell 132:344–362.
17. Vallabhapurapu S, Karin M (2009) Regulation and function of NF-κB transcription
factors in the immune system. Annu Rev Immunol 27:693–733.
18. Dejardin E (2006) The alternative NF-κB pathway from biochemistry to biology: Pit-
falls and promises for future drug development. Biochem Pharmacol 72:1161–1179.
19. Sun SC (2011) Non-canonical NF-κB signaling pathway. Cell Res 21:71–85.
20. Xiao G, Harhaj EW, Sun SC (2001) NF-κB–inducing kinase regulates the processing of
NF-κB2 p100. Mol Cell 7:401–409.
21. Senftleben U, et al. (2001) Activation by IKKα of a second, evolutionary conserved,
NF-κB signaling pathway. Science 293:1495–1499.
22. Beinke S, Ley SC (2004) Functions of NF-κB1 and NF-κB2 in immune cell biology. Bi-
ochem J 382:393–409.
23. Yamada T, et al. (2000) Abnormal immune function of hemopoietic cells from
alymphoplasia (aly) mice, a natural strain with mutant NF-κB–inducing kinase. J Im-
munol 165:804–812.
24. Dong C, Temann UA, Flavell RA (2001) Cutting edge: Critical role of inducible cos-
timulator in germinal center reactions. J Immunol 166:3659–3662.
25. Gigoux M, et al. (2009) Inducible costimulator promotes helper T-cell differentiation
through phosphoinositide 3-kinase. Proc Natl Acad Sci USA 106:20371–20376.
26. Claudio E, Brown K, Park S, Wang H, Siebenlist U (2002) BAFF-induced NEMO-
independent processing of NF-κB2 in maturing B cells. Nat Immunol 3:958–965.
27. Thompson JS, et al. (2001) BAFF-R, a newly identified TNF receptor that specifically
interacts with BAFF. Science 293:2108–2111.
28. Liao G, Zhang M, Harhaj EW, Sun SC (2004) Regulation of the NF-κB–inducing kinase
by tumor necrosis factor receptor–associated factor 3–induced degradation. J Biol
Chem 279:26243–26250.
29. Morrison MD, Reiley W, Zhang M, Sun SC (2005) An atypical tumor necrosis factor
(TNF) receptor–associated factor–binding motif of B cell–activating factor belonging
to the TNF family (BAFF) receptor mediates induction of the noncanonical NF-κB
signaling pathway. J Biol Chem 280:10018–10024.
30. Chen FE, Ghosh G (1999) Regulation of DNA binding by Rel/NF-κB transcription fac-
tors: Structural views. Oncogene 18:6845–6852.
31. Ruben SM, et al. (1992) I-Rel: A novel rel-related protein that inhibits NF-κB tran-
scriptional activity. Genes Dev 6:745–760.
32. Huang DB, Vu D, Ghosh G (2005) NF-κB RelB forms an intertwined homodimer.
Structure 13:1365–1373.
33. Sen R, Baltimore D (1986) Multiple nuclear factors interact with the immunoglobulin
enhancer sequences. Cell 46:705–716.
34. Hatada EN, et al. (2003) NF-κB1 p50 is required for BLyS attenuation of apoptosis but
dispensable for processing of NF-κB2 p100 to p52 in quiescent mature B cells. J Im-
munol 171:761–768.
35. Swallow MM, Wallin JJ, Sha WC (1999) B7h, a novel costimulatory homolog of B7.1
and B7.2, is induced by TNFα. Immunity 11:423–432.
36. Khayyamian S, et al. (2002) ICOS-ligand, expressed on human endothelial cells, cos-
timulates Th1 and Th2 cytokine secretion by memory CD4+T cells. Proc Natl Acad Sci
USA 99:6198–6203.
37. Klingenberg R, et al. (2005) Endothelial inducible costimulator ligand expression is
increased during human cardiac allograft rejection and regulates endothelial cell–
dependent allo-activation of CD8+T cells in vitro. Eur J Immunol 35:1712–1721.
38. Yin L, et al. (2001) Defective lymphotoxin-β receptor–induced NF-κB transcriptional
activity in NIK-deficient mice. Science 291:2162–2165.
39. Ganchi PA, Sun S-C, Greene WC, Ballard DW (1992) IκB/MAD-3 masks the nuclear
localization signal of NF-κB p65 and requires the transactivation domain to inhibit
NF-κB p65 DNA binding. Mol Biol Cell 3:1339–1352.
40. Tesciuba AG, et al. (2008) ICOS costimulation expands Th2 immunity by augmenting
migration of lymphocytes to draining lymph nodes. J Immunol 181:1019–1024.
41. Reiley WW, et al. (2006) Regulation of T cell development by the deubiquitinating
enzyme CYLD. Nat Immunol 7:411–417.
12832
| www.pnas.org/cgi/doi/10.1073/pnas.1105774108 Hu et al.