-
Annu. Rev. Immunol. 12:881-922 Copyright 1994 by Annual Reviews
Inc. All rights reserved
THE CD40 ANTIGEN AND ITS
LIGAND
J. Banchereau, F. Bazan', D. Blanchard, F. Briere, J.P. Galizzi,
C. van Kooten, Y.l. Liu, F. Rousset, S. Saeland
Schering-Plough, Laboratory for Immunological Research,
Dardilly, France; ** DNAX Research Institute, Palo Alto,
California
KEY WORDS: CD40, CD40-ligand, lymphocyte activation, B cell(T
cell interactions, hyper IgM syndrome
Abstract CD40 is an integral membrane protein found on the
surface of B lymphocytes, dendritic cells, follicular dendritic
cells, hematopoietic progenitor cells, epithelial cells, and
carcinomas. It is a 45-50 kDa glycoprotein of 277 aa, which is a
member of the tumor necrosis factor receptor superfamily. The CD40
gene maps to human chromosome 20q 1 1 -2-q 1 3-2. CD40 binds to a
ligand (CD40-L) which is an 35 kDa glycoprotein of 261 aa, a member
of the tumor necrosis factor superfamily. The CD40-L gene maps to
human chromosome Xq24. This CD40-L is expressed on activated T
cells, mostly CD4 + but also some CD8 + as well as basophils/mast
cells. The CD40-L is defective in the X-linked hyper- IgM
syndrome.
Cross-linking of CD40 with immobilized anti-CD40 or cells
expressing CD40-L induces B cells to proliferate strongly, and
addition of I L-4 or IL-1 3 allows the generation of
factor-dependent long-term normal human B cell lines and the
secretion of IgE following isotype switching. Addition of IL- l O
results in very high immunoglobulin production with limited cell
proliferation. IL- IO induces naive B cells to produce IgG3, IgGI,
and IgAl , and further addition ofTGFfJ permits the secretion
ofIgA2. Several evidences suggest that CD40-dependent activation of
B cells is important for the generation of memory B cells within
the germinal centers: (i) CD40 activated germinal center B cells
cultured in the presence of IL-4 acquire a memory B cell phenotype,
(ii) CD40 activated B cells can undergo isotype switching, (iii)
the deficit of CD40-L results in the hyper-IgM syndrome
characterized by lack of germinal centers in secondary lymphoid
organ
88 1 0732-0582/94/0410-088 I $05 .00
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
882 BANCHEREAU ET AL
follicles and lack of IgG, IgA, and IgE, and (iv) CD40-L
positive T cells are present in secondary follicles. Thymic
epithelial cells, activated monocytes, and dendritic cells express
CD40 antigen which may be involved in an enhanced cytokine
production by these cells, allowing an amplification of T cell
proliferation. Finally, as other members of the tumor necrosis
factor receptor family have been shown to bind several l igands, it
is possible that CD40 may bind other l igands that may trigger CD40
on d ifferent cell types such as hematopoietic cells or epithelial
cells.
INTRODUCTION The CD40 antigen was independently identified in
1985 and 1 986 by monoclonal antibodies reacting with carcinomas
and B cells (antibody S2C6, antigen p50) ( 1 ) and showing
costimulatory effects for B lymphocyte (antibody G28-5, antigen
Bp50) (2). This antigen was designated as CDw40 at the Third
International Workshop on leukocyte antigens in Oxford in 1 986,
and as CD40 at the Fourth Workshop in Vienna in 1 989. A cDNA
encoding CD40 was isolated in 1 989 (3), and this sequence
demonstrated a relationship with the human low affinity nerve
growth factor receptor (LNGFR). These molecules are now considered
as part of the tumor necrosis factor receptor superfamily.
Cross-linking of CD40, in conjunction with IL-4, was then found to
induce B cells to undergo long-term B cell growth, as well as
isotype switching (4-6). In 1 992, expression cloning using CD40 Fc
fusion protein allowed the isolation of a CD40-ligand (CD40-L)
expressed on activated T cells (7), an observation which led to the
demonstration of the key role o f CD40-L/CD40 interactions in T
cell-dependent B cell activation by many groups (8). The CD40-L is
one of the members of the recently identified tumor necrosis factor
superfamily. In 1 993, a genetic alteration of the CD40-L was shown
to be responsible for the X-linked hyper-IgM syndrome, which is
characterized by the lack of circulating IgG and IgA and the
absence of germinal centers (9). While the function of CD40 has
principally been studied on mature B lymphocytes, more recent
studies show the presence of b iologically functional CD40 on other
cell types, such as epithelial cells ( 10), monocytesj macro phages
( 1 1 ), and hematopoietic progenitors ( 1 2, l 3) .
CD40 ANTIGEN Modular Structural Design and Evolutionary
Relationship of CD40 The CD40 antigen is a phosphorylated
glycoprotein which migrates in SDS polyacrylamide gel
electrophoresis as a 48 kDa polypeptide under
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 883
both reducing and nonreducing conditions. It is a hydrophobic
molecule with an acidic Pi of 3 .2. A significant proportion of
CD40 from the Burkitt lymphoma Raji cells and normal B cells is in
a dimeric form, whereas such dimers are virtually absent from
carcinoma lines or EBV -transformed cells ( 14, 1 5) .
A cDNA encoding CD40 was isolated by expression cloning from a
library of the Burkitt lymphoma Raji (3) . The mature molecule is
composed of 277 aa with a 1 93 aa extracellular domain, a 22 aa
transmembrane domain, and a 62 aa intracellular tail. While the
extracellular segment of CD40 displayed significant similarity to
the analogous protein domain of the p75 low-affinity, nerve growth
factor receptor (LNGFR) ( 1 6), the intracellular chain did not
betray a relationship to any other characterized molecule. Human
and murine CD 40 molecules share 62% amino acid identity in the
extracellular domains and 78% identity in the intracellu lar
extensions ( 1 7) .
The p75 LNGFR is the founding father of a superfamily of
receptorlike molecules that share a common binding domain composed
oftandemly repeated cysteine rich modules ( 1 8) (Figure I, Figure
2, Table 1) . It is remarkable that this nascent group of receptors
is more appropriately named after two later additions: the p55 and
p75 receptors for tumor necrosis factor (TNFR I and TNFR2,
respectively) ( 19-2 1 ) that bind the related cytokines tumor
necrosis factor-ex (TNF-ex; also known as cachectin) and
lymphotoxin (L T; also known as TNF-f3) (22, 23). LNGFR acts
principally to recruit a number of neurotrophin ligands, including
NGF, to the cell surface, aiding in the formation of the signaling
complex formed by dimeric trk receptors that possess intracellular
tyrosine kinase domains (24, 25). LNGFR does not apparently play a
role in signal transduction (26). In contrast, the two TNFRs are
both active signalling receptors and bind ligands with a "TNF fold"
that is unrelated to the neurotrophin structure (27). Subsequent
additions to the TNFR superfamily display conserved interactions to
TNF-like cytokine ligands (28), cementing the TNFR moniker for the
cognate receptors ; perhaps the LNGFR molecule represents an
"escaped" TNFR analogous to the tissue factor cell surface tether
for coagulation Factor VII that is surprisingly related to the
hematopoietic cytokine receptor superfamily (29).
In addition to the LNGFR, CD40, and the two TNF receptors, the
TNFR superfamily currently includes CD27, a molecule expressed on T
cells and activated B cells (30, 3 1) ; CD30, an activation
molecule ofT cells and B cells initially detected on Reed Sternberg
cells (32); OX40, a rat activated T cell antigen (33); 4-1BB,
another T cell molecule (34); Fas/Apol, a lymphocyte antigen whose
triggering induces apoptosis (35-37); and a receptor-like protein
from an expressed open reading frame
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
884 BANCHEREAU ET AL
rt Ii..NGFf( ntl.JlfQFR rltLNGm r1 hrNFA I rlmTN'FAl n 1'I1NFFtO
r1mTNmU
rll\TNfRh
t1 hC040 n mC(:)t1l rt tX.(l) r1 n4tH
r1 m.HSS n h'til n mFIU n hCot'1 n mCOOl r1 ncOlo
r3" hctn(t r1 mC:tO t1 my,v 111 "dV 12 n cplVermS nG4R (1
utI/AS!
rI ru
-
r tLJr4GR't
r3n.kGFR
t:) d .NGf'R rJ, hTNFff 1 r3mTNFRI
r31'1'tNffln
rl mTNFAIl
t$hTNffth rl h04o fl mC04Q-1'31'0.40 (3 h44&e rt m4-1Se ,3
hfu rlmf'a. t3 !'ICOn 11 t'rIC027 ,3 hCD:Jo r1 hCOao t3 meet!) t:J
myxV Tt r3 lilY 12 r3 eplV frn8 tllr.U'V G4R f3 ciiCJI1
r
-
uperfalllily ytoki"e
uperfamily Rceptors
.'J
".-1.ii! .. ..',. q,. ,.. ..,.,., '111_... .. -. . ... ... -..
... .. ........ .... ,-J
.... -I1ltf""I+I."'''''''''-1rr''[}-P-St .......
Figure 2_ Schematic organization of receptor and cytokine
superfamilies: The receptor supergroup comprises twelve molecules,
ten from mammalian cells: LNGRF, TNFR1 and TNFR2, the TNFR homolog
from human chromosome 12 (TNFRh), C040, OX-40, 4-1 SS, Fas, C02?
and C030 antigens. A single, representative poxvirus TNFR homolog
chain is noted (from myxoma virus: ref). The small, single-repeat
fungal TNFR homolog is ECP1 protein from Cladosporium fulvum.
Repeats are drawn as diamonds (red, purple, yellow and cyan denote
repeats 1-4, respectively) that are divided into two halves (see
box of the TNFR Cys-repeat fold) that likely form disulfide linked
loops. Ser-Thr-Pro rich chain segments that typically link the
repeat domains to the transmembrane segment are drawn as wavy
lines-note that the poxvirus and fungal receptor homologs are
soluble proteins. Two types of cytoplasmic domain homologies are
noted by blue or orange colored boxes. The TNFR2 chain does not
resemble either motif. The known cytokine ligands of the TNFR1,
TNFR2, C040, C02? and C030 are drawn as folded sheets with opposing
red and grey faces. A separate box shows the -strand construction
of the fold-the grey (inner) sheet is formed by strands 6"-8-I-O-G,
while the red (outer) sheet comprises strands 8'-C'C-H-E-F. Note
that all ligands, save LT, are membrane anchored by N-terminal
hydrophobic segments.
The bottom box shows a decomposition of structural features in
the receptor and cytokine chains. A composite receptor is shown
divided into four repeats each of which show a variation {)f the
Cys-repeat pattern: disulfide links are shown above. Repeats are
also colored (red, purple, yellow and cyan) as in the receptor
superfamily figure above. Intron positions are mapped to the
receptor chain and show some correlation to repeat and subrepeat
structure. The composite ligand shows the position of the eleven
-strands with the few identical amino acids in the alignment of
Fig. 5 noted above the chain-all of these contribute to the
hydrophobic core. Orange and green stars note the location of
residues in contact with two distinct receptor chains in the trimer
complex of L T and TNFR1.
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
00 00 0'\
Table 1 Members of the TNF receptor and cytokine superfamily
analyzed in the present work
t:C M olecule Chain length (aa) Gene structure Chromosome
location References :> Z
(") :r:
Receptors tr1 ;;c LNGFR 399; 396; 397 (h; r; c) 6 exons (h)
17q21-22 (h) (16) tr1
:> TNFR I 434; 433 (h; m) 10 exons 12pI 3; 6 (h; m) (19, 20,
61, 62) c:: TNFR II 493; 452 (h; m) 1; 4 (h; m) (21) tr1 ..., TNFR
h 408 (h) 12p (h) (38) :> r CD40 258; 286 (h; m) 9 exons
20qlI-q13; 2 (h; m) (3, 17, 63, 66, 67) OX-40 262 (r) (33) 4-1 BB
229; 229 (h; m) 1; 4 (h; m] (34, 39, 139a) Fas 319; 306 (h; m)
partial (m) IOq24.1; 19 (h; m) (35-37) CD27 240; 230 (h; m) 12q I 3
(h) (30, 31, 64) CD30 577; 480 (h; m) I p36 (h) (30, 40) Poxvirus
Rh 310; 309; 337; 330 single viral orfs (41, 43-46)
(myx; sf; cpx; var) ECP 1 65 3 exons (cf) (47)
Cytokines TNFo: 233; 235 (h; m) 4 exons (h; m) 6 (131) LT 205;
202 (h; m) 4 exons (h; m) 6 (132) LT-f3 244 (h) 4 exons (h) 6 (137)
CD40-L 260; 261 (h; m) 4 exons (h) Xq26-q27 (h) (7, 117, 119,
127-129) CD27-L 193 (h) 19p I 3.3 (h) (138) CD30-L 234; 239 (h; m)
9q33; 4 (h; m) (139) 4-1 BB-L 309 (m) 17 (m) (I39a) A
nnu.
Rev
. Im
mun
ol. 1
994.
12:8
81-9
26. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
Acc
ess
prov
ided
by
Uni
vers
ity o
f C
alif
orni
a -
Dav
is o
n 01
/25/
15. F
or p
erso
nal u
se o
nly.
-
CD40 AND ITS LIGAND 887
branch of the TNFR superfamily include the Variola virus G4R
(45) and Cowpox virus crmB (46) gene products.
The most intriguing member of the TNFR superfamily is the fungal
pathogenic protein ECP I isolated from Cladosporium julvum, a
tomato pathogen (47). ECPI is a secreted protein of 65 residues
that appears to contain a single, characteristic TNFR-like cysteine
repeat of 46 amino acid length; ECP l is suggested to play a role
in suppressing the tomato plant defense response by b inding plant
cytokine-like molecules secreted in the face of fungal attack (48).
This economical molecule may represent the minimal binding
structure of a TNFR homolog and is a functional analog of the
poxvirus T2 factors.
It is instructive to examine the modular protein architecture of
the TNFR superfamily from the vantage point of the recently
described xray crystal structure of the complex formed by soluble b
inding domains of p55 TNFR I and a LT trimer (49). Early
comparisons ofTNFR superfamily members highlighted the striking
division of the extracellular binding domains into three or four
imperfect repeats of ,.... 40 residues, anchored by a
superimposable pattern of six cysteines ( 1 6, 1 8-2 1 , 30, 32,
33). While the disulfide bridging pattern was unknown, each repeat
was assumed to form an independently folding entity with three main
intrachain disulfide bridges-covariant loss of Cys 3 and 5 in
several receptor modules led to the suggestion that these were
likely linked ( 1 8) . The protein fold of the TNFR repeat module
is shown to be a tandem arrangement of "tethered loops, " where the
N-terminal loop is fixed by a Cys l-Cys2 link, and the C-terminal
loop features an analogous Cys4-Cys6 bridge (49) (see Figure 2).
The non-essential Cys3-Cys5 link ties the stalk connecting the two
loop structures to the second loop (49) . A structural subdivision
nevertheless does seem to indicate that the TNFR modular fold arose
from the union of two smaller folding units (48). This finding
explains the occasional, puzzling truncations of exact half-repeats
in several superfamily receptors: TNFRI repeat 4, OX-40 repeat 3,
human CD30 repeat 4, and CD27 repeat 3-a half-repeat corresponds to
either the N- or C-terminal loop structure that is presumably
capable of correct folding. The comparative alignment of available
TNFR modules ( 1 8 ; see also Figure 1) also h ighlights the
conservation of additional residues, notably an aromatic amino acid
located 5 residues after Cys I that in the crystal structure is
shown to be important for the interaction of N- and C-terminal
loops (49). The stacked repeat modules form an elongated, s lightly
bent rod that intercalates in the groove between two L T subunits
in the packed trimer. In addition, the key receptor-ligand
interactions involve residues in repeats 2 and 3 (49). Together,
the three receptor subunits encage the LT trimer and scrupulously
avoid. contact between each other.
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
888 BANCHEREAU ET AL
A notable superstructural motif in the TNFRI rods is a
spiralling "staircase " formed by the successive stacking of
disulfide links perpendicular to the rod axis (49). The gross
spatial architecture of the CD40/CD40 ligand interaction is
schematically shown in Figure 9 and should help in the following
structural discussion on CD40 and its ligand. This energetically
favorable packing arrangement places the disulfide bridges in a
preferred hydrophobic core location (50); it has been observed in
other small, disulfide rich proteins like toxins and defensins (5
1-53), and the cystine-knot family of NGF/TGFP/pDGF folds (28).
These disulfide staircases may appear again in the binding domains
of other cytokine receptors in spite of distinct folding
topologies; notable candidates include cysteine-rich modules
present in the EGF and insulin receptor extracellular segments (54)
and the TGF-p family signalling receptors (55).
The cysteine repeat pattern of TN FRs resembles an
eight-cysteine motif in laminin-like modules (33, 56). These latter
laminin repeats are believed to fold in a manner similar to EGF,
with an "extra " disulfide-bridged loop at the C-terminus (57). An
X-ray crystallographic structural analysis of a representative
fragment of laminin containing a three-repeat segment of chain that
binds the protein nidogen appears to be in progress (58) and may
soon resolve the matter.
The modular organization of the TNFR binding structures invites
the question of an underlying genetic organization-do the repeats,
or halfrepeats, correspond to exon-encoded folding units as
observed for other protein modules (59) ? Several gene structures
have been elucidated for superfamily members: the mouse LNGFR gene
(60), the human and mouse TNFRI genes (61 , 62), the mouse CD40
(63), human CD27 (64), the fungal ECP l (47), and a fragment of the
mouse Fas gene (65). The CD40 murine gene is fairly typical of this
collection with nine exons that span 1 6.3 kb of genomic D NA (63).
A diagrammatic i llustration of exon boundaries is presented in
Figure 2. While there is no consistent subdivision of repeats into
exons for any given gene, the composite picture that emerges
suggests that type 1 introns (which intercut codons after the first
nucleotide base) typically map to the dividing boundaries of
repeats, or half-repeats. Type 1 modules are fairly common in the
diverse collection of small protein motifs found in mosaic cell
surface molecules (57).
The mouse CD40 gene is located on the distal region of
chromosome 2 which is syntenic to human chromosome 20q I l -q13.
Accordingly the human CD40 gene was mapped to chromosome 20 by
using human-rodent somatic cell hybrids (66) and to 20 q ll-20q1
3-2 by in situ hybridization (67). It is obvious that superfamily
receptor genes have been dispersed throughout the genome. Table 1
shows the chromosomal localization of other receptor genes.
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
Cells Expressing CD40
CD40 AND ITS LIGAND 889
B LYMPHOCYTE C D40 was early considered as a pan B cell antigen
(68). CD40 is detected on all B cells isolated from adult and cord
blood, tonsils and spleens (Figure 3A) . It is expressed on resting
virgin sIgD+ sIgM+ B cells in the primary follicles and the mantle
zone of secondary follicles within the peripheral lymphoid organs
(Figure 4). The density of CD40 is identical on naive B cells
centroblasts and centrocytes, composing the germinal centers of
secondary follicles, and on the CD38 - sIgD - memory B cells
(69).
Plasmablasts isolated from tonsils express CD40 (P Merville et
aI, manuscript in preparation), whereas roughly half of the
antibody secreting cells circulating in the blood eight days after
vaccination have lost CD40 (70). The fully differentiated plasma
cells of mucosal lamina propria and bone marrow do not express C
D40. Polyclonal activators such as anti-IgM, anti-CD20, or
anti-Bgp95 antibodies or phorbol esters slightly upregulate CD40
expression on B cells (3, 7 1) . IFNy (3) and IL-4 (72) are also
able to increase CD40 levels on B cells. The human C D40 gene is
expressed as a single 1 .4 kb species in B cells. The murine C D40
gene is expressed in B lymphocytes as two mRNA species, a
predominant one of l .7 kb and a minor one of 1 .4 kb, which are
generated by alternative usage of polyadenylation signals in the 3'
untranslated region. The activation of murine B lymphocytes
preferentially increases the level of the smaller transcripts (17).
Anti-human C D40 antibodies react with the CD40 antigen from
macaques and baboons (73, 74).
Virtually all chronic lymphocytic leukemia B cells and non
Hodgkin's lymphoma cells express C D40. Burkitt lymphoma cell lines
and EBVtransformed B cell lines all display CD40 ( 1 ) . The
plasmacytoma cell line RPMI 8226 displays low levels of C D40 while
U266 cells do not. Most IL-6-dependent myeloma cell lines are CD40
positive (B Klein, personal communication).
Numerous EBV-transformed B cell lines release soluble CD40
(sCD40) spontaneously, as detected with a specific ELISA (75).
Supernatants of Staphylococcus aureus strain Cowan I-activated
normal B cells cultured in the presence of IL-4 or IL-2 also
contain sCD40. This is in line with the described soluble forms of
the other members of the TNF -R family, which include sTNFR l and
sTNFR2 (76, 77), sLNGFR (78), and sCD27 (79). Although the
mechanisms of sCD40 release have not yet been studied, it is
possible that these molecules originate from proteolytic cleavage
as shown for sCD27 (80), and as indicated by the lack of
alternatively spliced C D40 mRNA. sCD40 of EBV cell line
supernatants is able to bind to the C D40-L expressed on activated
T cells.
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
890 BANCHEREAU ET AL
CD40
A
CD40
B
CD40
c
CD19 sIgM sIgD
CDlO CDl9 sIgM
CD34 CDIO CDl9
1\
Figure 3 Expression of CD40 on B lineage cells and hematopoietic
progenitor cells. Total B cells isolated from tonsils (panel A), B
cell precursors (CD I 9 + CDl O + sIg - ) obtained from mid-term
fetal bone marrow (232) (panel B), and CD34+ progenitors separated
from full-term umbilical cord blood (233) (panel C), were stained
with the anti-CD40 mAb 89 (72). In parallel, and as indicated,
cells were monitored for expression of CD19, CDlO, CD34, sIgM, and
sIgD. Histograms represent log of FITC-fiuorescence analyzed on a
fiowcytometer. Dotted lines correspond to negative control staining
with an unrelated murine mAb.
HEMATOPOIETIC PROGENITORS In human fetal and adult bone marrow,
CD40 is detectable on the majority of B cell precursors (BCP),
which express CDl9 and CDl O but lack surface Ig (Figure 3B; 1 3 ,8
1-83). CD40
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 89 1
is acquired early in B cell ontogeny, as it is present on BCP
expressing the CD34 progenitor cell antigen ( 1 3) and has been
reported on fetal liver CDI9+ cells prior to rearrangements at the
IgH locus (pro-B cells) (8 1 ).
CD40 is also expressed on malignant BCP in B lineage acute
lymphoblastic leukemias at various maturation stages, ranging from
pro-B cells to pre-B cells expressing cytoplasmic J1 chain (8 1 ,
84). About 28 to 44% of BCP-ALL cases have been shown to express
CD40 (81, 82). Among the positive BCP-ALL, CD40 is present only on
a proportion of the leukemic cells, a feature that has been
associated with clonogenic capacity (81).
Finally, CD40 is not restricted to early B lineage cells in
normal hematopoiesis; it is found on most cord blood CD34 +
progenitors, which lack a CDI9+ CDIO+ subset (Figure 3C), and on
the majority of bone marrow CD34 + cells ( 1 3). In this context,
CD40 expression is more heterogeneous on CD34 + progenitors than on
mature B cells (Figure 3A and 3C). CD40 expression is lost during
myeloid development in cultures of CD34+ cells ( 13). It is
important to further define the CD34+ CD40+ population, and to
investigate whether the most primitive nonlineage committed CD34+
cells, characterized by lack of CD38 antigen (85), express
CD40.
DENDRITIC CELLS AND MONOCYTES Immunohistological analysis (68)
on tonsil and spleen sections has shown high levels of CD40 on
interdigitating dendritic cells in the T cell-rich areas of
secondary lymphoid organs (Figure 4). These cells derive from
skin/mucosal Langerhans cells which only weakly express CD40 (86).
However, following culture, Langerhans cells express CD40 at high
levels. Dendritic cells can be generated in vitro by culturing CD34
+ hematopoietic progenitor cells in the presence of GMCSF /IL-3 and
TNFIX (87). These cells, which resemble Langerhans cells as they
express CDl a and display Birbeck granules, express CD40 at high
levels and may thus represent cells at a stage of differentiation
between Langerhans cells and interdigitating dendritic cells. In
fact, dendritic cells isolated from peripheral blood also express
CD40 and may represent Langerhans cells homing towards secondary
lymphoid organs (88).
Primary human monocytes freshly isolated or cultured for 48 hr
show low but detectable CD40 surface protein ( 1 1 ) . GM-CSF,
IL-3, and IFNy strongly upregulate their CD40 expression.
OTHER CELL TYPES CD40 is expressed in the CD45-negative stromal
cell population of human thymus ( 10). Immunohistology shows CD40
expression on cortical and medullary thymic epithelial cells, as
well as thymic interdigitating cells and B cells (Figure 4).
Expression of CD40 is specifically maintained on cultured thymic
epithelial cells but not on thymic
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
896 BANCHEREAU ET AL
CD40-L in NK cells and purified monocytes (1 27). Northern blot
analysis of the B cell line Daudi and the histiocytic lymphoma U937
have shown specific hybridization signals at 3.7 and 1 .7 kb,
suggesting expression of CD40-L related molecules in B cells and
monocytes (1 29).
CD40-Ligand Is a Member of an Emerging Cytokine Family
The initial characterization of the mouse CD40-L did not easily
yield i ts chain similarity to the TNF superfamily (7). Subsequent
analyses showed that an 200 aa domain which formed the major
extracellular domain of CD40-L could be aligned with available
TNF-O( and LT sequences in a structurally sound manner (27, 1 1 9,
1 29) (Figure 2, Figure 5). The region in question comprises the
bioactive, receptor-binding, globular portion of TNF-like molecules
that folds, by example of TNF-O( ( 1 30, 1 3 1) and LT (49, 132),
into a barrel-like structure reminiscent of viral capsid proteins.
This distinctive TNF fold consists of two packed sheets of eight
major antiparallel f3-strands linked in a "f3-jellyroll" topology
with an N-terminal loop insertion that contains two additional,
short f3-strands (1 3 1 , 1 32, 1 33). In this manner, and
following a s tandard nomenclature for viral capsid proteins ( 1
30, 1 33), the "inner " sheet which is primarily involved in trimer
contacts is formed by s trands B '-B-I-D-G in correct spatial
order, and the "outer" sheet analogously consists of s trands C'
-C-H-E-F (see Figure 2). While there is no detectable sequence
similarity with viral capsid proteins, detailed structural
comparisons u tilizing the TNF three-dimensional coordinates show
excellent backbone superposition, similar types of sidechain
contacts, conserved amino acids, and various geometric matches with
available fj-jellyroll folds ( 1 34).
The concept of structural conservation in spi te of amino acid
divergence has prompted the construction of detailed CD40-L models
( 1 35, 1 36) utilizing homology modeling techniques and the
available protein frameworks of TNF-O( and L T. These models show
that the sparse amino acid matches between CD40-L and TNF-O( or LT
are important for the construction of the fJ- jellyroll core fold,
and also predict that CD40-L is capable of trimerization. Gaps in
the protein alignment partition to variable loops in the model and
two cysteines (residues X and Y) are suggested to form a disulfide
link analogous to the Cys l 45-Cys 1 77 link in TNF-a ( 1 35).
Recent additions to the TNF cytokine superfamily have been a L
T-fj molecule capable of forming heterotrimers with L T (1 37), and
the cytokine ligands for CD27 ( 1 38), CD30 ( 1 39), and 4- 1 BB
(139a). It is interesting to note that all TNF superfamily
cytokines, with the exception of LT, are produced as type II
membrane-tethered molecules ( 1 39). Secreted LT has an alternative
membrane-bound form when complexed with a p33 factor
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
Spleen
Tonsil Thymus
Figure 4. Expression of CD40 antigen on frozen sections of human
spleen, tonsil and thymus: Expression of CD40 was analysed with
anti-CD40 antibody G28.5 (a kind gift of professor E. A. Clark)
using the APAAP method. Spleen: CD40 staining on B cells in
follicle (F) and marginal zone (MZ). Very strong CD40 staining is
observed on interdigitating cells in the periarteriolar lymphocytic
sheath (PALS) around the central arteriole (CA). Tonsil: CD40
staining on follicular mantle (FM) B cells, germinal center (GC) B
cells and follicular dendritic cells. Again, very strong CD40
staining is seen on interdigitating cells in the extrafollicular
area. Thymus: CD40 staining on thymic epithelial cells both in the
cortex (CT) and the medulla (M). Very strong CD40 staining is
observed on dendritic cells in the medulla.
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
892 BANCHEREAU ET AL
fibroblasts. IL-IO(, TNFO(, and IFNy signi ficantly upregulate
CD40 levels and 1 .4 kb CD40 transcripts on cultured epithelial
cells.
CD40 is also expressed on follicular dendritic cells (FDC) of
secondary lymphoid organs, as shown by immunohistology analysis on
tissue sections (Figure 4) and by flow cytometry on freshly i
solated FDC (89), and on FDC cultured in the presence of GM-CSF
(90).
Immunohistological analysis performed on many different tissues
(9 1 ) has also indicated staining of various cell types with the
anti-CD40 mAb G28.5 : endothelial cells (mixed pattern of
reactivity = + / - ), smooth muscle cells (+ / - ), cardiac
myocytes (+ ), epidermis (weakly + ), gastrointestinal mucosa (+ ),
gallbladder mucosa (+), bronchus mucosa (+ ), salivary gland ductal
and acinar cells ( + ), pancreas ductal cells, mammary ductal and
acinar cells (+ / - ), sweat gland (weakly +), prostate gland cells
(weakly + ), thyroid gland ( + ), parathyroid gland ( + ) .
Basophils isolated from the blood of chronic myeloid leukemia
patients have been reported to express CD40, shown by the binding
ofmAb B-El O (92).
The CD40 antigen was initially identified with the mAb S2C6,
which was generated from mice immunized with a urinary bladder
carcinoma (93). Subsequently, CD40 antigen has been identified on
carcinomas of other origins such as colon, prostate, breast, and
lung, as well as on melanomas (94). CD40 has also been detected on
T cell lines transformed with HTLV I and II (95). Accordingly
baboon T cell lymphomas, which display HTLV I, are CD40 positive
(74). Thus, CD40 is expressed on cells with high proliferation
potential such as hematopoietic progenitors, B lymphocytes, and
epithelial cells, and cells able to present antigen such as
dendritic cells, activated monocytes, B lymphocytes, and follicular
dendritic cells.
Signal Transduction Through CD40
The TNFR superfamily is primarily defined by common structural
motifs in the extracellular binding domains-there is no equivalent
organizing principle for the diverse cytoplasmic extensions ( 1 8)
. This is an important distinction: perhaps the TNFR superfamily is
a grouping of composite receptors with related binding folds
grafted onto a diverse collection of intracellular signalling
domains. After all, most growth factor/cytokine receptors operate
by a common signalling paradigm irrespective of their cytoplasmic
mass. Receptors rely on the ligand-induced association of
extracellular domains to drive the concommitant association of
intracellular structures, independently of whether these
cytoplasmic domains are enzymes (i.e. tyrosine or serine kinases)
or not (96, 97) . The TNFR intracellular domains clearly fall into
the latter category--by analogy to
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 893
the T cell antigen receptor complex (9S) or the hematopoietic
superfamily receptors (99, 100), there may be a reduced number of
protein motifs that act as binding epitopes for intracellular
signalling molecules like kinases.
Perhaps a cytoplasmic grouping is possible: it has been noticed
that the cytoplasmic domains of Fas, TNFR1, and LNGFR show a
distant but significant similarity (35, 37). The more
chain-economical domain ofCD40 has also been proposed to belong to
this group (35, 37), but further comparison suggests a fortuitous
alignment. The structural integrity of the intracellular domain of
Fas is required for apoptosis as assayed by both site-directed
mutagenesis and deletion analysis (101), or by naturally occurring
variants involved in murine lymphoproliferative disorders ( 1 02).
The Fas gene in mice is also the structural gene for
lymphoproliferation (ipr) mutation (102, 103 , 104). While the key
ipr mutation causes a rearrangement in gene intron 2 (and no viable
mRNA is detectable), the allelic iprcg defect involves a single
amino acid change in the Fas cytoplasmic domain, converting Ile225
Asn (102-104). The Fas-similar domain in TNFRI mediates
TNF-directed cell cytotoxic activity ( l05). Most recently, t he
homologous domain in the LNGFR was shown to be also involved in the
induction of apoptosis (103). Together, these observations suggest
that a subgroup of TNFR superfamily receptors share a common
signalling domain. That CD40 is not included in this group makes
some biological sense: while TNFRI and Fas require triggering by
antibodies or ligands to induce cell death, CD40 acts instead as a
survival factor (4) . Some functional analogy between CD40 and
LNGFR was observed in that LNGFR constitutively induces cell death
unless triggered by NGF or a monoclonal antibody (106) .
Of the remaining TNFR superfamily members, a cytoplasmic domain
homology has been noticed only between CD27 and 4-1 BB (31).
Interestingly, this region of 4-1BB displays a short sequence
bracketed by two cysteines that resembles the kinase binding motifs
of CD4 and CDS ( 107, l OS). Indeed, Kim et al (109) have recently
shown that the mouse T cell antigen 4-1 BB directly associates with
the tyrosine kinase p56 \Ckl through these motifs. CD40, OX-40,
CD30, and the TNFR homolog from human chromosome 12 in turn show a
short stretch of sequence similarity centered on a
Glu-(AspjGlu)-Gly-Lys motif at the C-terminal end of their
respective cytoplasmic domains. TNFR2 remains unclassified in this
organizational scheme. Further dissection of this multitude of
domains may uncover their role in TNFR superfamily signal
transduction (110).
The importance of the CD40 intracellular domain in signal
transduction has been directly inferred by the deleterious nature
of a Thr234 Ala mutation (1 11). Cross-linking of CD40 on
nonresting human B lymphocytes i nduces tyrosine phosphorylation of
four distinct substrates (S4).
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
894 BANCHEREAU ET AL
Activation of protein tyrosine kinases (PTK) appears important
for the transduction of CD40 signals because PTK inhibitors block B
cell aggregation (112). As the CD40 cytoplasmic segment does not
contain any enzymatic domain, it is likely that the CD40 mediated
protein tyrosine kinase activity (113) occurs through activation of
separate kinases. Accordingly, engagement of CD40 on the Daudi cell
line induces activation of the src type kinase, iyn, following its
increased phosphorylation (113). In contrast, anti-CD40 alters
neither the phosphorylation nor the activity of jyn, another kinase
of B lymphocytes. CD40 engagement results in phosphorylation of the
85 kDa subunit of phosphatidyl inositol 3 kinase (PI3K), while it
does not affect its 110 kDa subunit. CD40 cross-linking induces an
increased activity ofPI3K which catalyzes the phosphorylation of
phosphoinositols on the 3' moiety and which plays a key role in
mitogenesis. Finally, CD40 ligation induces within one minute
increased phosphorylation ofPLCy2 (phospholipase C). PLCyl , which
is present in lower amounts in B cells (114) does not appear to be
phosphorylated in response to CD40 ligation. Phosphorylation of
PLCy2 is consistent with anti-CD40 induced IP3 production (84).
CD40-LIGAND
Expression of CD40-Ligand on Activated T Cells
Fusion proteins made from the extracellular domain of human CD40
and the Fc region of human IgGI have been used to identify CD40-L.
CD40-L is expressed on activated EL-4 thymoma cells, activated
mature T cells but not on resting T cells (7, 115-123, 123a).
CD40-L can be detected on T cells very early (1-2 hr) after
activation. The expression of CD40-L is primarily on CD4+ T cells,
although a small population of CD8+ cells also acquires CD40-L. The
CD40-L can be induced on THO, THI and TH2 cells. The functional
expression of CD40-L, as detected with the CD40-chimera is
inhibited after co-culture with B cells (120). This effect can be
explained by a downregulation of CD40-L mRNA and by the release of
sCD40 which binds to CD40-L. As a consequence, T cells stain
positive with a polyclonal anti-CD40 antiserum (75). The capacity
of B cells to release CD40 in cultures was confirmed by a specific
ELISA. Studies performed with monoclonal antibodies specific for
the CD40-L (122, 124), isolated for their ability to block T
cell-dependent B cell activation, confirm the expression of CD40-L
observed with the CD40 fusion protein. Immunohistochemistry
demonstrates that human CD40-L is expressed on CD3 + CD4 + T
lymphocytes of the mantIe zone and germinal center light zone of
secondary follicles in all peripheral lymphoid tissues. CD40-L
positive cells can be identified within the interfolIicular T
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 895
cell rich areas of secon dary lymphoi d tissues and the medulla
and cortex of normal thymus.
Immunohistochemistry on murine spleen i solated three to four
days, following immunization with the thymus-dependent antigen KLH,
demonstrates an increase of CD4 + CD40-L + T cells in and around
the terminal arterioles, and on the periphery of the outer peri
arteriolar lymphoid sheath. Double immunohistochemical analysis
reveals that the B cells producing the specific antibodies are
juxtaposed to CD40-L + T cells ( 1 25). The latter study failed to
show the presence of CD40-L + T cells in germinal centers. This
difference with human data may be linked either to species or to
the reagents as the anti-murine CD40-L may not detect CD40-L
saturated with sCD40.
Lung mast cells and blood basophils also express CD40-L which i
s functional as shown by the ability of basophils to induce B cells
to secrete IgE ( l 25a).
Characterization of a CD40-Ligand cDNA
Murine EL-4 thymoma cells can induce the proliferation of
resting human B lymphocytes ( 1 26). A cDNA that encodes CD40-L (7)
was i solated from the EL-4 line following enrichment of cells
binding to a CD40-Fc fusion protein. The murine cDNA predicts a
polypeptide which has 260 aa consisting of a 22 aa cytoplasmic
domain, a 24 aa transmembrane domain, and a 214 aa extracellular
domain with four cysteines. Murine CD40-L i s a type I I membrane
protein which has an extracellular carboxy-terminus.
A human CD40-L cDNA has been i solated by screening stimulated
human blood T cell libraries with the murine CD40-L probe ( 1 17 ,
1 1 9,1 27, 128). Another group had independently isolated a
TNF-related activation protein (TRAP) from activated human T cells,
which turned out to be the human homolog to murine CD40-L ( 1 29).
The cDNA for human CD40-L encodes a polypeptide of 261 aa. The
human CD40-L has a 22 aa cytoplasmic domain, a 24 aa transmembrane
domain, and a 2 1 5 aa extracellular domain with five cysteines.
The murine and human CD40-L display a conserved N-linked
glycosylation site in the extracellular domain and the human CD40-L
displays an additional, but probably not utilized, glycosylation
site in the cytoplasmic domain. The two sequences exhibit 78% aa
identity. There is 75% identity in the extracellular domain, 96%
between the transmembrane region, and 81 % between the cytoplasmic
domains.
Northern blot analysis of activated T cells demonstrates the
presence of two mRNA species of 2 .1 and 1.4 kb (117, 127) that
differ in the length of the 3' untranslated ends. CD40-L mRNA has
been detected in activated CD4+ and CD8+T cells. RT-PCR analysis
also indicates the presence of
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
R __ IU m ... I ..... U m_ hUb 13 .. S me ........ 108 h(lD.H
,."' ..... $7 mCD3O-< .. "'498-1. lOG --
-.
R __ hU .a mL1 .,.., 15' mTNI' ... h"b 147 -, , ..
"""""' ... hCD2T..t ". """'lO-l ". -"- ,..
ll)4 .. t80.-l ... --
-....
J.t<
tPQVGLTPS AAU(ARWHPKMHlAHSTt KON
L$OVfH _ . S AARIAHPl POKHI. THOll KON . , 'iRSSSRf PSfl 'EO
. t.RSSSQNSSO VEE
LPE EEPE fQL SPOt "GO EN FfMOKODQNf"
, Cf N f EOP SNAAS
LPt 51 lC!t. KA!\PF
, Fe 11. K S r f> S GCPNfrQQG
,H ;:'.j. i\J., I , I ,a A! Ac.al* J." AjAA.aa I, .
A TH S ' S's SGVQ
_ LfPCSMf: a,(\AAA1A,l. t, t t t t t t. 1 I t 4\4A
bbbbbbbb F
AE
A
!a::JI:1 :11
..... &&A GtPHlVl..,SPS
GtSHlHf$PS,
npOVlDfAES6
tP'I(YlOfAESG
HPOMyf) fARO.
OPSQYSHaTG . Ii: A t HR .... G_ G T L l RM r I)
A*4$. ' AAAt I
Figure 5 Sequence and structural alignment ofTNF superfamily
cytokines. The known X- ray structures ofTNF-a, LT and the model
construction of CD40-L (135, 136) were used to accurately align the
available sequences of human (h) and mouse (m) TNF-like cytokines.
J3-strands derived from the TNF-(l( and LT structures are boxed and
labeled according to standard viral capsid nomenclature: B- C-
D-E-F-G-H-I, in such a manner that strands B-I- D-G and C-H-E-F
form two opposing J3-sheets of a flattened barrel structure. Note a
small insertion of sequence (large boxed region) between strands B
and C that forms threc short J3-strands labeled 8', B, and C. The
line beneath the sequences denotes the residue environment from
TNF-a and LT X-ray structures: d marks highly exposed (i.e. solvent
accessible) residues, are moderately exposed amino acids, *residues
are buried in the hydrophobic core of the fold while t denotes
residues buried in the trimer interface. Above the sequences are
noted residues potentially in contact with receptors by homology to
the determined L T- TNFR I receptor complex: & and $ symbols
separate these interactions to the two receptor subunits that
contact each ligand subunit in the trimer.
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
898 BANCHEREAU ET AL
( 140) that was revealed to be LT-f3 ( 1 37). These
membrane-tethered homoor heteromeric cytokines bind to specific
TNFR superfamily receptors that are displayed on the surface of
neighboring cells, signaling the direct, contact-mediated
immunoregulation of one cell type by another. However, as aptly
demonstrated by the X-ray complex of LT and TNFRI proteins, soluble
forms of these cytokines are perfectly capable of inducing receptor
association, and subsequent intracellular signalling.
A comprehensive alignment of representative TNF-C( and LT
molecules with the LT -13 chain and the ligands for CD27, CD30, and
CD40 illustrates the great divergence of amino acid sequences
within the TNF superfamily and is a graphic measure of the
plasticity of the TNF fold (Figure 2). How can we decide which
sequence variations are covariantly absorbed by the core fold, and
which contribute to changes in receptor specificity? Several groups
have tried to answer this question by exhaustive mutagenic studies
of the structure and function of TNF-C( and LT ( 14 1-144) prior to
the determination of the complex structure. Van Ostade et al ( 145)
have remarkably arrived at the identification of a single residue
change in human TNF-C( (Arg32 ---t Trp) that drastically lowers the
binding affinity for TNFR2 but retains wild-type binding to TNFR I
. As each receptor has distinct signalling pathways (23), this
finding has important therapeutic potential ( 146). In addition,
these studies may reveal the structural basis of receptor
promiscuity for TNF-like ligands: clearly the distantly related LT
and LT-f3 molecules can bind to the same receptor subunits ( 1 37,
1 40). Similar situations may exist with other TNF-like cytokines,
reminiscent of the multitude of promiscuous couplings that are
found between hematopoietic cytokines and their receptors ( 147,
148).
FUNCTIONAL CONSEQUENCES OF CD40 ENGAGEMENT Mature B Lymphocytes
For simplifying the description of the effects ofCD40 ligation on
human B cells, we use the classical model where B cells, following
antigen encounter, undergo activation, proliferation, and
differentiation in a stepwise fashion ( 149).
ACny A nON AND PROLIFERA nON OF B LYMPHOCYTES Resting B cells
cultured in the presence of anti-CD40 antibody increase their size
(72) and form homotypic aggregates ( 1 1 2, 1 50-1 53). Studies
with antibodies have shown that this aggregation involves not only
LFAI-ICAMl , but also the recently identified pair CD23/CD21 ( 1
54, 1 55). This latter interaction is inhibited by the anti-CD23
antibody MHM6 and the anti-CD2 1 antibody
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 899
BU32, while antibodies directed to other epitopes of CD23 and
CD2 1 are inefficient. Interestingly, fucose-2-P04 also blocks CD40
induced homotypic aggregation in accordance with the lectin-like
nature of CD23. Activation of B lymphocytes through CD40 also
stimulates their adhesiveness to endothelial cells in a VLA-4
dependent fashion ( 1 56). Soluble antiCD40 ( 1 57) and CD40-L
transfected cells ( 1 58) are able to prevent apopto tic death of
germinal center B cells. Triggering B cell CD40, either by
cross-linking with monoclonal anti-CD40 antibody presented by a
murine fibroblastic Ltk-cell iine transfected with FcyRIIjCDw32
(CD40 system) (4, 5) or with CD40-L transfected cells, results in
an increased expression of CD23, class II antigens and B7jBB l ( 1
2 1 , 1 24, 1 59-161 ) . CD38+ CD44-germinal center B cells
cultured in the CD40 system (CDw32 + L cells + anti-CD40 antibody)
in the presence of IL-4, acquire the phenotype of memory CD3S-
CD44+ B cells (YJ Liu et aI, manuscript in preparation). Finally,
CD40 cross-linking induces B cells to produce IL-6 and IL- l O (
162, 1 63).
Anti-CD40 antibodies have been isolated for their ability to co
stimulate with either anti-IgM antibodies or phorbol esters (2, 7 1
, 72). In a soluble form, some anti-CD40 antibodies can induce DNA
replication in resting B cells although this does not result in
sustained proliferation ( 1 50, 1 64, 1 65). B cells cultured with
soluble CD40-1igand in a monomeric form enter into limited DNA
synthesis, but a trimeric form of soluble CD40 ligand obtained
through various constructions results in quite significant DNA
synthesis particularly in combination with anti-Ig ( 1 1 5, 1 1 9,
1 66). Culture of resting B cells in the CD40 system results in
strong and long lasting B cell DNA replication (4). At least 70-80%
of B cells enter into the G 1 phase of cycle and 50--60% into the S
phase, permitting B cell numbers to increase by three- to four-fold
over two weeks. These culture conditions allow the proliferation of
various B cell subpopulations including mantle zone sIgD+ sIgM + B
cells, sIgD- sIgM + /- B cells, CD5+ and CD5- B cells ( 167).
Furthermore, quite significant DNA synthesis is observed in
leukemic B cells, such as non-Hodgkin B cell lymphomas and chronic
lymphocytic leukemia B cells ( 1 68). Transient expression of
transfected murine and human CD40-L into various cell lines allows
short-term proliferation of murine and human B lymphocytes ( 1 1 7,
1 19, 1 2 1 , 1 27, 1 28, 1 59, 1 61) . L cells stably expressing
human CD40-L can induce a proliferation of B lymphocytes at least
as important as that obtained using CDw32 L cells and anti-CD40
(Figure 6).
DNA synthesis of B cells in response to soluble anti-CD40
antibodies and either anti-IgM or phorbol esters is further boosted
by IL-4 (72, 1 50, 1 69). Addition of IL-4 to B cells cultured in
the CD40 system results in their sustained proliferation (4), and
the total B cell population can expand
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
900 BANCHEREAU ET AL
A
B
cpm
80000
60000
40000
20000
cpm 100000
80000
60000
40000
20000
CDw32 L-celJs
none IL2 IL4 IUO none IL2 IL4 IL 1 0 mAb89
CD40-ligand L-celJs
none IL2 IL4 IL 10 none IL2 IL4 IL 1 0 mAb89
Figure 6 B cell proliferation induced either by CDw32-L cells
and anti-CD40 mAb 89 (CD40 system) or by CD40-Ligand-L cells.
Purified total tonsil B cells (20 x 1 031200 III well) were
cultured with irradiated L cells (8000 Rad; 4 x 103lwell) which
were stably-transfected with either FcyRIIICDw32 (Panel A) or with
CD40-L (Panel B). Cultures were performed without or with IL-2 (40
Vlml), IL-4 ( 100 Vlml) or IL- IO ( 1 00 nglml) in the absence or
presence of anti-CD40 antibody mAb89, as indicated. B cell
proliferation was determined at day 5 of culture by the addition of
3H-thymidine, present during the last 16 hr of the culture period.
Indicated is the mean cpm observed in triplicate cultures. Note
that the Mab89, in a soluble form, is able to inhibit CD40-L
induced B cell proliferation while it is a powerful stimulator in
an immobilized form.
up to lOOO-fold. This results in the generation of
factor-dependent longterm normal B cell lines which are negative
for Epstein-Barr viral infection. B cell clones can be generated
that contain several hundred cells. IL-4 also strongly enhances the
proliferation of B cells cultured in the presence of cells
expressing CD40-L (Figure 6) ( 1 1 7, 1 59) . In recent studies, we
have been able to show that L cells stably expressing CD40-L are
able to induce
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 90 1
the multiplication of B cells over two weeks provided IL-4 is
added to cultures. IL- 1 3, a cytokine which shares homology with
IL-4 ( 1 70, 1 7 1 ) can also induce a strong and long lasting
proliferation o f B cells stimulated in the CD40 system or with
CD40-L transfected cells (1 27).
IL- l and IFNy enhance the DNA synthesis observed in the CD40
system or with CD40-L transfected cells, with or without IL-4 ( 1
59, 1 72). In our hands, IL-2 poorly enhances the proliferation of
B cells cultured in the CD40 system or with CD40-L transfected
cells, while another study indicates that IL-2 is able to stimulate
the proliferation of B cells activated with CD40-L transfected
cells ( 1 59). Cells cultured in the CD40 system with IL-4 express
CD19 , CD20, CD40, sIg, high levels of CD23, and HLA class II
antigens. Surprisingly, a quite significant proportion of cells
cultured for three weeks still express sIgD ( l 72a). Addition of
anti-IgM antibody or SAC particles, which enhance DNA synthesis (
173), does not down-regulate sIgD expression (our unpublished
results).
Both viral and human IL- l O enhance the proliferation ofB cells
cultured in the CD40 system or with CD40-L transfected cells, as
determined both by tritiated thymidine incorporation and increased
viable cell numbers ( 1 59, 1 74) (Figure 6). IL- l 0 appears to be
almost as efficient as IL-4 during the first week of culture, but
proliferation slows down thereafter and eventually ceases after 1 4
days. The combination of IL-4 and IL- l O is additive and results
in a 60-100 fold expansion of viable B cells over two weeks. IL- IO
upregulates the expression of CD25jTac on anti-CD40 activated B
cells, and accordingly addition of IL-2 strongly enhances B cell
proliferation ( 1 75). In fact, the production of IL-I 0 by CD40-L
transfected CVl cells could explain the response of B cells to IL-2
( 1 59) which is only marginal in our studies. B cells cultured in
IL- IO differ microscopically from those cultured in IL-4 in that
loose aggregates are observed early on, which then yield cultures
mostly composed of single large cells.
DIFFERENTIATION OF B LYMPHOCYTES Human and murine B cells
cultured with anti-CD40 antibody, with or without CDw32 L cells or
with CD40-L transfected cells, produce marginal amounts of
immunoglobulins ( 1 2 1 , 1 59, 1 6 1 , 1 72). However, coculture o
f human B cells with SAC particles, anti-CD40 antibody, and CDw32 L
cells results in the production of very large amounts of IgM, IgG,
and IgA without IgE ( 1 76). Mantle zone sIgD + sIgM + B cells
secrete only IgM, whereas sIgD - sIgM + j - B cells secrete IgG and
IgA and lower amounts of IgM. This indicates that the concommitant
triggering of sIg and CD40 results in differentiation of human B
lymphocytes independently of exogenous cytokines.
Addition of IL-4 to cells cultured in the CD40 system only
results in a
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
902 BANCHEREAU ET AL
slight increase in the production of IgM and IgG and in the
secretion of large amounts of IgE ( 1 72). In fact, IgE production
results from isotype switching as highly purified naive slgD+ B
cells produce as much IgE as isotype committed slgD - B cells. In
contrast to long-term B cell proliferation, the production of IgE
does not require the presence of CDw32 L cells (6, 1 77-1 79).
Cells transfected with CD40-L can also induce murine and human B
lymphocytes to secrete IgE in response to IL-4 ( 1 17 , 1 2 1 , 1
59, 1 6 1) . Addition of lFNy or IFNa: to CD40 activated B cells
fails to inhibit IL-4-induced IgE production. Indeed the inhibitory
effects of interferons on IL-4-induced IgE production by
mononuclear cells ( 1 80) may to be due to downregulation of CD40-L
on IL-4 activated T cells ( 1 23, 1 28). TGFp and TNF!X are able
respectively to block and to stimulate the production of IgE by
CD40-activated B cells ( 1 8 1 ) . B cells stimulated through their
CD40 antigen secrete IgE and IgG4 in response to IL- 1 3 , as a
result of isotype switching ( 1 27, 1 82).
Addition of lL- l O to CD40-activated B lymphocytes results in
the production of considerable amounts of IgM, IgG, and IgA without
any IgE ( 1 59, 1 74). In fact, cells cultured in the CD40 system
in the presence of lL-10 differentiate into plasma cells expressing
large amounts of intracytoplasmic immunoglobulins. IL- l O induces
anti-CD40 activated tonsil B cells to secrete IgG I , IgG2, and
IgG3. Purified human B lymphocytes cultured in the CD40 system in
the presence of IL- I 0 produce IgG and IgM antibodies able to bind
to antigens such as tetanus toxoid (Table 2). The combination of
soluble anti-CD40 or soluble CD40-L and IL- l O is, however,
insufficient to allow production of bacteriophage-specific antibody
by memory slgD- B cells from immunized individuals (I 82a).
Specific antibody can however be detected provided bacteriophages
are added to cultures, a finding consistent with a preferential
expansion of antigen specific B cells and reminiscent of the comito
genic effect of anti-CD40 and anti-Ig antibody (2, 72, 1 73).
Interestingly, CD40 activated slgD+ JslgM+ B cells were found
essentially to secrete IgM but also IgG 1 and IgG3 in response to
IL- l O, indicating that this cytokine may act as a switch factor
for certain IgG subclasses ( 1 83a). CD40 activated naive slgD+,
slgM+ B cells cultured with IL- l O also produce low levels of lgA
l , and addition of TGFp induces large amounts of both IgAI and
IgA2 subtypes, while inhibiting IgM and IgG production ( 1 76) (F.
Briere, manuscript in preparation). In contrast, TGFfl inhibits the
production of lgM, IgG, and IgA by isotype committed sIgD - B cells
stimulated by IL- l O in the CD40 system. This strongly suggests
that TGFfl may represent an IgA-switch factor both in human and in
mouse ( 1 83, 1 84). In line with these findings, recent studies
have indicated that TGFfl is able to induce !X I and !X2 germline
transcripts in activated human B cells ( 1 85).
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 903
Table 2 Production of tetanus toxoid specific antibody by B
lymphocytes activated in the CD40 system with cytokines'
% Wells with IgG anti-tetanus toxoid
5000 B cells per well 500 B cells per well
No cytokine 0 0 IL2 0 0 IL4 3 0 ILIO 9 0 ILIO + IL4 19 8 ILIO+
IL2 29 1 1
' Purified tonsillar B lymphocytes were cultured over irradiated
CDw32-L cells (8000 Rad; 4 x I O'/well) with 0.5 Jig/ml anti-CD40
Mab 89. Cultures were performed without or with IL2 (40 V/ml) or
ILIO (100 ng/ml) or their combination. Wells were harvested after
10 days and the presence of antitentanus toxoid antibody was
determined by standard ELISA. Positive wells yielded an optical
density equal to at least three times the background observed
without addition of culture supernatants.
IL-5 acts synergistically with IL-4 to induce CD40 activated
murine B cells to secrete IgG I and IgE, and to promote IgM and
IgG3 secretion. Surprisingly, the induction of antigen specific
antibody responses of CD40-L activated murine B cells requires the
presence of antigen and IL-2, while IL-4 and IL-5 are only poorly
efficient ( 1 2 1 ). This suggests that triggering of the antigen
receptor may skew the cytokine response of CD40 activated B cells.
Thus the engagement of CD40 on B cells turns on their isotype
switching machinery, the specificity of which is subsequently
provided by cytokines.
ROLE OF CD40 IN T CELL-DEPENDENT B CELL ACTIVATION Many
different culture systems have been set up to elucidate the cell
surface and soluble molecules involved in the growth and
differentiation of B lymphocytes. Activated T cells can induce
resting B cells to proliferate and differentiate into Ig secreting
cells ( 1 86, for a review). In particular, the mouse thymoma EL-4
can activate resting human B cells to proliferate ( 1 26), an
observation which allowed the cloning of the ligand for CD40 (7). T
cells activated with immobilized anti-CD3 antibodies, which mimic T
cell receptor engagement, are able to induce normal B cells to
proliferate and secrete Igs in the complete absence of accessory
cells or lectins that might favor cellular interactions ( 1 87, 1
88). Cytokines produced by activated T cells are involved in the
growth and differentiation of B cells, but the cell contact cannot
be replaced by T cell supernatants. In contrast, fixed activated T
cells or their membrane enriched fraction can induce B cell
proliferation, and addition of cytokines can further enhance growth
and can induce Ig
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
904 BANCHEREAU ET AL
secretion ( 1 88-195). The CD40-Fc fusion protein inhibits B
cell stimulation induced by activated T cells and their membranes (
1 1 6, 1 2 1) , as well as IL-4-induced IgE production by
mononuclear cells ( 1 96). Monoclonal antibodies blocking T
cell---dependent B cell activation against murine ( 1 1 6) and
human (1 24) activated T cells were in fact specific for CD40-L. In
line with these findings, anti-CD40 antibody strongly block both
proliferation and Ig secretion of tonsillar B cells induced by T
cells stimulated with immobilized anti-CD3 ( 197, 1 98).
Interestingly, activation of naive sIgD+ B cells is significantly
less inhibited by anti-CD40 than that of sIgD - B cells, suggesting
that the differentiation of naive B cells may also occur
independently of CD40-CD40-L interactions. Accordingly, patients
with defective CD40-L display increased circulating IgM levels (see
below).
B Lymphocyte Precursors Soluble anti-CD40 antibodies neither
stimulate the proliferation of normal B cell precursors (BCP) nor
alter the effect of known growth signals for such cells ( 12, 82).
However, they enhance tyrosine phosphorylation and inositol 1
,4,5,-trisphosphate production in fetal liver pro-B cells (84). CD
I9+ CDlO+ BCP can grow in the CD40 system provided IL-3 is added (
12). This proliferation is further potentiated by IL-7 and IL- IO.
However, at variance with mature B cells, IL-4 does not induce
proliferation of BCP cultured in the CD40 system. BCP can also be
induced to proliferate by triggering their CD40 with CD40-L present
on activated CD4+ T cell clones ( l 98a). The signal provided by
CD40-L is essential, as is demonstrated by the blocking of T
cell---dependent BCP proliferation with the anti-CD40 mAb 89 and
the lack of stimulatory effect of T cell clones lacking functional
CD40-L obtained from a hyper-IgM patient. Activation of BCP through
CD40, either in the CD40 system or in the presence of T cells, does
not promote differentiation into mature sIg+ B cells. A small
proportion of immature sIgM+ cells emerge in CD40-dependent
cultures, but cells bearing other isotypes are not observed ( 1 2)
. Of interest, triggering of CD40 can induce high-level surface
membrane expression of CD23 on BCP ( 1 2). As CD23 is involved in
myelopoiesis ( 1 99) and T cell ontogeny (200), it appears of
interest to evaluate the role of this molecule in B
lymphopoiesis.
Other Cell Types
Recent studies indicate that CD40 is functional on cells other
than B lineage cells. Thymic epithelial cells are induced to
secrete GM-CSF when triggered with soluble anti-CD40 in conjunction
with IL- l and most notably IL- l and IFNI', the latter
upregulating CD40 expression ( 10). This effect occurs in the
absence of cell proliferation.
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 905
Monocytes stimulated by CD40-L transfected cells secrete low
amounts of IL-6 and IL-8 ( 1 1 ) . Addition of GM-CSF, IL-3, or
IFNy, which upregulate CD40 expression on monocytes, further boosts
CD40-L-induced secretion of IL-6 and IL-8 and allows secretion
ofTNFcx. Cross-linking of monocyte CD40 using CD40-L transfected
cells also results in the activation of tumoricidal activity
against a melanoma cell line.
T lymphocytes also appear to respond to CD40-L (20 1) . CD40-L,
on transfected cells or in a soluble trimeric form, induces (i)
resting T cells to express CD25/Tac and CD40-L; (ii) activated T
cells to secrete IFNy, TNFa, and IL-2, and (iii) activated T cells
to proliferate. Both CD4 + and CD8+ T cells appear to proliferate
in an IL-2-independent fashion. It is proposed that CD40-L binds to
CD40 antigen which is expressed at very low density on T cells (201
).
ROLE OF CD40 IN ANTIGEN-DRIVEN IMMUNE RESPONSES Mutated
CD40-Ligand in the X-Linked Hyper-IgM Syndrome Several
immunodeficiencies mapped to loci distributed throughout the X
chromosome include: X-linked agammaglobulinemia, X-linked severe
combined immunodeficiency, Wiskott-Aldrich syndrome, X-linked
lymphoproliferative syndrome, and X-linked hyper IgM syndrome (202,
203, for a review) . Males affected with hyper-IgM syndrome are
susceptible to various infections (204). These patients do not make
antibodies to exogenous antigens but make a variety of
autoantibodies. Their serum has slightly or vastly elevated
concentrations of polyclonal IgM and IgD, but no detectable IgA or
19B and very low levels of IgG. The secondary lymphoid organs of
these patients display no germinal centers although they have
normal levels of circulating B cells and plasma cells producing IgM
and IgD in lymphoid tissues and the gastrointestinal tract.
Immunization of these patients with bacteriophage 1 74, a
thymus-dependent antigen, results in a poor humoral immune response
that is restricted to the IgM isotype (205). These features
indicate a defect in the generation of memory B cells. It is
important to note that these patients often also suffer from
neutropenia.
The gene mutated in the hyper-IgM syndrome was mapped to
chromosomal region Xq24-27 close to HPRT (206). The localization of
the CD40-L gene in region Xq26-3-Xq27- 1 ( 129, 1 36, 207) led to
the demonstration that the defect in the hyper-IgM syndrome is due
to point mutations or deletions in the gene encoding the CD40-L (9,
1 36, 207-2 10) . Activated T cells from these patients do not bind
CD40 fusion proteins, although cells from some of these patients
stain with polyclonal antibody
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
906 BANCHEREAU ET AL
specific for the CD40-L. This suggests either expression of the
truncated CD40-L on the cell surface or expression of a
conformationally altered protein unable to bind CD40. mRNA
transcripts for the CD40-L have been sequenced in 1 3 patients, and
1 2 displayed either point mutations (9 cases) or deletions (3
cases) (Figure 7). Of note, one of the patients displayed a CD40-L
cDNA without any nucleotide change within the coding region (207).
One study on four patients showed that the CD40-L of three of the
patients' mothers had the same mutation, while a fourth patient
appeared to have had a de novo alteration (21 0) . Expressed
mutated CD40-L are unable to activate B lymphocytes from normal
individuals (1 36, 207), whereas the B lymphocytes from hyper-IgM
patients can be stimulated either with activated T cells (2 1 1 )
orwith anti-CD40 and cytokines ( l 36, 207-209, 2 1 2, 2 1 3). In
line with the Ig status of hyper-1gM patients, mice develop an
hyper-IgM response to the thymus-dependent antigen DNP-Ovalbumin
when CD40-CD40-L interactions are blocked in vivo by injecting
CD40-Fc chimeric molecules (D Gray, personal communication).
A Tentative Synthesis on Where, When, and How CD40 Is
Triggered
The key role of CD40 for T cell-dependent B cell activation in
vitro was firmly supported by the in vivo finding that mutations in
CD40-L gene result in hyper-lgM syndrome. Based on the kinetics and
in vivo expression of CD40 and CD40-L, the following model for the
timing and sites of
Figure 7 Schematic representation of the coding region of
CD40-Ligand and localization
of the mutations, deletions found in 12 hyper-IgM patients.
Patients A.T., B.W., T.G. (209),
P I-P4 (210), P5-P7 (207) and CD., J.W. ( 1 36). IC,
intracellular domain; TM, transmembrane
domain; EC, extracellular domain.
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 907
CD40/CD40-L interactions during T cell-dependent immune
responses is proposed (Figure 8), based on earlier versions (21 4-2
1 8) . The entry of pathogen/antigen into the mammalian organism
elicits specific and nonspecific immune responses. Dendritic cells
from skin or mucosa
( differentiation) memory
B cell
l -, - t follicular dendritic apoptosisl cell (FOC)
'-c-.o-na-I-ex-p-a-ns-;o-n-, , ? somatic mutations
centroblast ,
-m-ig-ra-ti-on-in-to- - - - - - - t - - - -primary follicle
naive or , . "., ,
/-0. interdigitating
APICAL LIGHT ZONE
BASAL UGHT ZONE
DARK ZONE
SECONDARY LYMPHOID QBM!'
PARACORTICAL . AREA
_ eric:I - -t- ,- '- _ - _
II --.... Slillfl M!,/,!;QA
dendritic I Langerhans cell
Figure 8 Model for CD40/CD40-Ligand interactions in antigen
dependent immune responses (See section: A tentative synthesis on
where, when and how CD40 is triggered), = Antigen,
-
908 BANCHEREAU ET AL
initiate the specific reaction by capturing antigen and
migrating into the paracortical T cell-rich areas of secondary
lymphoid organs such as lymph nodes. In these sites, these APC are
called interdigitating dendritic cells (IDC). During the antigen
loading and migration phases, dendritic cells acquire surface CD40,
possibly following interaction with GM-CSF released by various cell
types (keratinocytes, neutrophils, mast cells) at the site of
antigen entry. In the T cell-rich areas, the IDCs present
antigenderived peptides bound to MHC class II antigens, to naive or
memory antigen-specific T cells initiating the extrafollicular
reaction. Cross-linking of the T cell receptor readily turns on
CD40-L expression (Figure 9). Meanwhile various sets of adhesion
molecules strengthen the interactions between T cells and the IDe.
The cross-linking of CD40 on IDC by T cell CD40-L enhances IDC
cytokine production which subsequently potentiates T cell
activation, proliferation and differentiation. In a mirror fashion
(Figure 9), the cross-linking of CD28/CTLA4 on T cells by B7
antigen on IDC results in increased cytokine production by T cells
which may then act (i) in an autocrine fashion as T cell growth and
differentiation factors, (ii) to further activate the IDC, and
(iii) to induce the proliferation and differentiation of recruited
antigen-specific B cells. It is possible that IDC present antigen
to B cells. The antigen activated B cells interact with T cells
through various surface molecules. In particular, CD40
cross-linking boosts B cell activation and differentiation and may
represent a key signal for the migration of B cells into primary
follicles composed of a network of follicular dendritic cells (FDC)
(2 19, 220). Some antigen-specific activated T cells may also
migrate at that time. These interactions result in massive B cell
proliferation which starts the germinal center (GC) reaction (221 ,
222). When a full germinal center is developed, proliferating
centroblasts can be identified in the dark zone. This is the
anatomic site where high-rate somatic mutations occur within the
variable Ig regions (223, 224). There is no evidence, at the
present time, that CD40/CD40-L interactions are operating at that
stage. The centro blasts then mature into nonproliferating
centrocytes that compose the light zone. In the basal light zone,
the somatic mutants undergo selection based on their ability to
bind to antigen deposited on FOCs in the form of immune complexes.
Antigen receptor triggering will permit the survival of these cells
while others die from apoptosis. Subsequently, in the apical light
zone, the antigen on FDC will be processed by selected B cells and
presented to antigen-specific activated T cells that produce IL-2,
IL-4, and IL- l 0 (225, 226). It is possible that the CD40 on FDC
may engage the CD40-L on activated T cells. At that stage,
C040/C040-L interactions may play a key role in (i) inducing
multiplication of the rare selected B cells, (ii) turning on the
isotype switch-
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 909
ing machinery, and (iii) inducing the differentiation of
selected B cells either into memory cells which will recirculate or
into plasmablasts that will leave the germinal centers to migrate
either to the bone marrow or to the mucosal lamina propria where
they become plasma cells.
The clinical and biological status of patients suffering from
the hyperIgM syndrome permits us to conclude that the role of CD40
on APC to boost T cell responses is either of secondary importance
or can be replaced by other molecular pairs as suggested by normal
T cell numbers and lack of viral infections. However, the
susceptibility of these patients to opportunistic agents such as
Pneumocystis yet indicate a partially altered repertoire of T cell
responses. The presence of normal or elevated circulating IgM
levels of polyclonal origin indicates that primary B cell
Figure 9 CD40/CD40-Ligand in cellular interactions. T cells
recognize peptide presented by MHC Class II on antigen presenting
cells (APC) (dendritic cells or B cells). Adhesion molecules
strengthen the interaction. This results in upregulation of
CD40-ligand on T cells and B7/BBI on APC. Triggering of CD40 on APC
permits activation of B cells and/or cytokine production by B cells
or dendritic cells. The produced cytokines further activate T cells
and allow their proliferation. The increased B7/BB I expression on
APCfB cells triggers CD28 or CTLA-4 on T cells which then secrete
cytokines which will either further activate the
proliferation/differentiation of B cells or act as autocrine T cell
growth factor. Thus, the interaction between CD40/CD40-L signals B
cells and APC, while the triggering of CD28/CTLA-4 signals T cells
to produce cytokines, which results in T cell proliferation, B cell
growth and differentiation, and APC activation.
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
9 1 0 BANCHEREAU ET AL
reactions are not affected by lack of CD40jCD40-L interactions
either because this interaction is not involved in this process or
because other molecules can substitute at that level. In this
respect, the B cell stimulatory effect of membrane TNF on activated
T cells may play a significant role (227, 228). The lack of
germinal centers in the secondary lymphoid organs of hyper-IgM
patients suggest either an altered migration of activated T and B
cells into primary follicles or an altered proliferation of
centroblasts. The mechanisms leading to the proliferation of centro
blasts still remain unclear and may be CD40-L independent, because
no CD40-L + T cells can be detected in the GC dark zone and because
B cells do not appear to undergo somatic mutations at a high rate
when cultured in the CD40 system in the presence of IL-4 (229).
PERSPECTIVES
As CD40 belongs to a family of molecules that bind several
ligands as exemplified by the two TNFRs, the LNGFR, and 4-1 BB, it
is plausible that CD40 may bind to other presently uncharacterized
ligands. Indeed, as 4- l BB binds both a TNF-like molecule ( l 39a)
and extracellular matrix proteins (56), it is possible that CD40
may also bind to such a matrix. This would be in agreement with the
high levels of binding of soluble CD40 observed on murine B cells
(229a). The function of CD40 on cells other than B cells and
monocytes remains to be determined. In particular the broad
expression of CD40 on CD34 + progenitors raises the possibility
that CD40 plays an important role in hematopoiesis. CD40 activation
could represent a pathway through which CD40-L + T cells may
regulate hematopoietic production levels in the bone marrow during
acute situation. Such a regulatory function would be compatible
with the fact that hyper-IgM patients have a normal output of newly
formed B cells despite alteration of their CD40-L. However, CD40
may also participate in constitutive hematopoiesis, as might be
suggested from the frequent neutropenia, as well as the less common
anemia and thrombocytopenia observed in hyper-IgM patients. Also,
the existence of an alternative ligand for CD40, as evoked above,
should be considered within the bone marrow, as it could point to a
more central function of CD40 in constitutive hematopoiesis.
Obviously, further studies are necessary to define the role of CD40
in hematopoietic development. In this context, evaluation of mice
in which the CD40 gene has been disrupted by homologous
recombination appears of importance.
The apparently crucial role of CD40 triggering in isotype switch
will permit us to dissect the mechanisms leading to the
intracellular assembly
Ann
u. R
ev. I
mm
unol
. 199
4.12
:881
-926
. Dow
nloa
ded
from
ww
w.a
nnua
lrev
iew
s.or
g A
cces
s pr
ovid
ed b
y U
nive
rsity
of
Cal
ifor
nia
- D
avis
on
01/2
5/15
. For
per
sona
l use
onl
y.
-
CD40 AND ITS LIGAND 9 1 1
of a functional isotype switch machinery. The elucidation of
these intracellular pathways may ultimately allow the construction
of antagonists blocking isotype switch. Finally, the lack of memory
cells in patients suffering from CD40-L mutations suggests that
pharmacological targeting of the CD40/CD40-L interactions may make
it possible to down-regulate undesired humoral responses such as
antibody-mediated autoimmune diseases. The administration of
anti-CD40-L antibody to animals was recently found to interfere
with the development of primary and secondary humoral immune
responses (230). Furthermore, such antibody treatment also
prevented the arthritis induced in mice by immunization with type
II collagen (23 1) . The available structural information on CD40
and CD40-L will be refined in the near future and will facilitate
the construction of mutants acting as antagonists.
ACKNOWLEDGMENTS
The authors wish to thank Nicole Courbiere for her invaluable
editorial assistance and Drs. R. Armitage, R. Geha, J. Gordon, D.
Gray and M. Howard for communicating preprints before
publication.
Any Annual Review chapter, as well as any article cited in an
Annual Review chapter, may be purchased from the Annual Reviews
Preprints and Reprints service.
1-800-347-8007; 415-259-5017; email: [email protected]
Literature Cited
1 . Paulie S, Ehlin-Henriksson B, Mellstedt H, Koho H, Ben-Aissa
H, Perlmann P. 1985. A p50 surface antigen restricted to human
urinary bladder carcinomas and B lymphocytes. Cancer Immunol.
Immunother. 20: 23-28
2. Clark EA, Ledbetter JA. 1986. Activation of human B cells
mediated through two distinct cell surface differentiation
antigens, Bp35 and Bp50. Proc. Natl. Acad. Sci. USA 83: 4494-98
3. Stamenkovic I, Clark EA, Seed B. 1989. A B-Iymphocyte
activation molecule related to the nerve growth factor receptor and
induced by cytokines in carcinomas. EMBO J. 8: 1403-10
4. Banchereau J, de Paoli P, Valle A, Garcia E, Rousset F. 199
1. Long term human B cell lines dependent on interleukin 4 and
anti-CD40. Science 251: 70-72
5. Banchereau J, Rousset F. 1991. Growing human B lymphocytes in
the CD40 system. Nature 353: 678-79
6. Jabara HH, Fu SM, Geha RS, Vercelli D. 1990. CD40 and IgE:
synergism between anti-CD40 monoclonal antibody and interleukin 4
in the induction of IgE synthesis by highly purified human B cells.
J. Exp. Med. 172: 1861-64
7. Armitage RJ, Fanslow WC, Strockbine L,