Positive selection and lineage commitment during peripheral B-lymphocyte development.

Abstract

Although it is appreciated that the antigen receptor on B cells is required for peripheral B-lymphocyte development and survival, it has been unclear whether this receptor interacts with self-antigens during development or if it signals constitutively in an antigen-independent fashion. The analysis of mutant mice in which antigen receptor signaling in B cells is either attenuated or enhanced has revealed the existence of a follicular versus marginal zone B-lymphocyte cell-fate decision. These analyses indicate that weak antigen receptor-derived signals favor marginal zone B-cell generation, and relatively strong signals favor the development of mature follicular B cells. Even stronger signals derived from the antigen receptor favor the generation of B1 B cells. This signal strength model for B-cell development supports the notion that self-antigens of varying affinity may mediate positive selection and lineage commitment. Direct evidence supporting such a view has been obtained from the analysis of antigen receptor knockin mice. Specific antigen receptors guide B cells to develop into specific lineages. Although Notch-2, nuclear factor-kappaBp50, and other genes are essential for marginal zone B-cell development, instructive signals delivered by the antigen receptor represent the primary force driving positive selection and lineage commitment in B lymphocytes.

Full text

Page 1

Positive selection and lineage
commitment during peripheral
B-lymphocyte development
Shiv Pillai
Annaiah Cariappa
Stewart T. Moran
Authors’ address
Shiv Pillai, Annaiah Cariappa, Stewart T. Moran,
Massachusetts General Hospital Cancer Center,
Harvard Medical School, Boston, MA, USA.
Correspondence to:
Shiv Pillai
Massachusetts General Hospital Cancer Center
Harvard Medical School
149, 13th Street, Charlestown Navy Yard
Boston, MA 02129
USA
Tel.: þ16177265619
Fax: þ16177249648
E-mail: pillai@helix.mgh.harvard.edu
Acknowledgements
This work was supported by grants AI 33507, CA
102793, and AI 54917 from the National Institutes of
Health.
Summary: Although it is appreciated that the antigen receptor on B cells
is required for peripheral B-lymphocyte development and survival, it has
been unclear whether this receptor interacts with self-antigens during
development or if it signals constitutively in an antigen-independent
fashion. The analysis of mutant mice in which antigen receptor signaling
in B cells is either attenuated or enhanced has revealed the existence of a
follicular versus marginal zone B-lymphocyte cell-fate decision. These
analyses indicate that weak antigen receptor-derived signals favor
marginal zone B-cell generation, and relatively strong signals favor the
development of mature follicular B cells. Even stronger signals derived
from the antigen receptor favor the generation of B1 B cells. This signal
strength model for B-cell development supports the notion that self-
antigens of varying affinity may mediate positive selection and lineage
commitment. Direct evidence supporting such a view has been obtained
from the analysis of antigen receptor knockin mice. Specific antigen
receptors guide B cells to develop into specific lineages. Although
Notch-2, nuclear factor-kBp50, and other genes are essential for marginal
zone B-cell development, instructive signals delivered by the antigen
receptor represent the primary force driving positive selection and lineage
commitment in B lymphocytes.
Introduction
There are numerous close parallels that can be drawn between
B- and T-lymphocyte development. In both developing B and
T cells, immunoglobulin (Ig) and T-cell receptor (TCR) genes
undergo rearrangement in a virtually identical fashion. During
B-lymphocyte development, pro-B cells that make in-frame
rearrangements at the Ig heavy chain locus can assemble and
receive signals from the pre-B-cell receptor (BCR). This pre-
antigen receptor not only mediates survival and proliferation
but also provides signals for allelic exclusion and further
developmental progression. A virtually identical positive selec-
tion event occurs in developing ab T cells. Pro-T cells that
make in-frame rearrangements at the TCR-b locus are able to
generate a pre-TCR and receive signals through this receptor.
This reading-frame-based positive selection event in the
Immunological Reviews 2004
Vol. 197: 206–218
Printed in Denmark. All rights reserved
Copyright ? Blackwell Munksgaard 2004
Immunological Reviews
0105-2896
206

Page 2

T lineage is commonly referred to as ‘beta selection’ or ‘pre-
TCR-mediated selection,’ in order to distinguish it from a
distinct antigen-driven selection event that occurs a little later
in development and which is linked to commitment to either
the CD4þor CD8þT lineages. This latter self-antigen-based
thymic selection event has historically been called ‘positive
selection’. Does B-cell development also depend on a self-
antigen-based positive selection event? Are all peripheral
lymphocytes, in both the B and T lineages, created in the
image of self-structures? This review addresses questions that
relate to the existence of, the teleological rationale for, and the
mechanisms that drive what we consider to be a ‘positive
selection’ and lineage commitment event during B-lymphocyte
maturationthat broadlyresembles
counterpart in the T lineage.
itswell-established
A role for the BCR during the development and
maturation of B lymphocytes
In a landmark study, Rajewsky and colleagues (1) used gene-
targeting approaches to knockin rearranged Ig heavy chain and
light-chain genes flanked by lox P sites, and they induced a
cytokine-regulated Cre transgene to generate the first condi-
tional knockout of the BCR. This study revealed a critical
requirement for the BCR in the survival of all B cells. In the
absence of the antigen receptor, all B cells, including naı ¨ve
peripheral B lymphocytes, were rapidly lost. The inability of B
cells to survive in the absence of the BCR suggested that this
receptor either provides constitutive or tonic signals in a
ligand-independent manner (2, 3) or that self-antigens recog-
nize cognate BCRs on immature or transitional B cells and,
thus, deliver signals that are required for B-cell survival. In
similar conditional deletion experiments performed later on T
cells (4), induced loss of the T-cell receptor resulted in a
relatively gradual disappearance of peripheral T cells, tempor-
ally distinguishable from the fairly dramatic loss of B cells
seen in the original study (1). This difference in the rate of
cell loss has been used as an argument to suggest that while the
TCR mediates antigen-driven positive selection, the BCR may
be required to generate antigen-independent ‘tonic’ signals for
peripheral B-cell survival.
Clearly, a number of other unexplored factors could account
for the distinguishable rates at which B and T cells disappear
after antigen receptor genes are inducibly deleted. These
factors range from possible differences in the rate of protein
turnover of the BCR complex versus the TCR, specific vari-
ations that might exist in the pathways that influence cell
survival in B and T cells, or differences in the availability of
growth or survival regulators other than antigen receptors to B
versus T cells.
The concept of constitutive or ligand-independent signaling
may have had a more easily supportable rationale, earlier in
lymphocyte development, as discussed below. The reason why
such a modality was deemed worthy of consideration in the
peripheral B-cell but not the peripheral T-cell context begs the
exploration of some historical biases regarding positive
selection. We first discuss issues that relate to tonic signaling.
Constitutive or tonic signaling during lymphocyte
development
Over a decade ago, we first suggested that the pre-B receptor
might signalconstitutively, ortonically,ina ligand-independent
fashion (5). A line of teleological reasoning that lent support to
the ‘LIAR’ (ligand-independent activation of receptor) hypoth-
esis was that pre-antigen receptors are designed not to monitor
the environment, but to determine whether in a given develop-
ing lymphocyte a productive in-frame rearrangement has
occurred at the Ig heavy chain locus or at the TCR-b chain
locus. Our subsequent studies, as well as those of others, on
pre-BCR signaling (6–8), and the studies of von Boehmer and
colleagues that followed on the pre-T receptor (9) have lent
support to the view that pre-antigen receptors probably signal
immediately after assembly in the absence of any known ligand.
The failure to conclusively identify specific and functionally
definable ligands for the pre-BCR and pre-TCR remain the
most compelling pieces of (albeit negative) evidence in support
of the LIAR hypothesis. The recognition that the BCR is required
for peripheral B-cell development and survival has, as discussed
above, led to a renewed interest in the notion of constitutive
signaling not just in pre-B and pre-T cells but also in mature B
cells. It is clear from some studies (10, 11) that when develop-
ing B cells are engineered to generate constitutive signals that
mimic those that can be generated by the BCR, these tonic
signals can mediate the maturation of B cells. Although these
studies suggest that constitutive signals are capable of facilitating
B-cell survival and maturation, they do not clarify whether
similar signals are actually generated in B cells in vivo in an
antigen-independent fashion. Whilst the term ‘constitutive anti-
gen-independent signal’ appears to suggest an event that does
not involve cell selection, antigen-independent signaling could
potentially be invoked as a means for positively selecting cells in
which both chains of an antigen receptor are ‘compatible’ at the
level of association (3). Tonic signaling might, therefore, iden-
tify and select cells that have been tested after rearrangement of
both receptor chains for efficient antigen receptor assembly and
Pillai et al?B-cell selection and lineage commitment
Immunological Reviews 197/2004
207

Page 3

transport. In contrast to the debate regarding this issue in the
B lineage, constitutive antigen-independent signaling has never
been considered necessary in the ab T-cell context.
The ‘need’ for positive selection during T-cell
development and the rationale for invoking a parallel
event in B cells
During T-cell development, it has long been accepted that
major histocompatibility complex (MHC) antigens mediate
positive selection at the double-positive T-cell stage, and
there has been no reason to assume that the TCR in double-
positive T cells, or in mature T cells, generates signals in the
absence of ligand (12). There is a deeply rooted view that
there is a ‘need’ in T cells, but not in B cells, for self-antigen-
mediated positive selection, and as a result, it has been rela-
tively difficult for many immunologists to seriously consider
the possibility that self-antigen-based positive selection might
also occur in the B lineage. The concept of a need for positive
selection in T cells evolved primarily from the view that T cells
bearing TCRs that recognize self-MHC molecules with low
affinity must be prevented from dying a default death, pri-
marily, because such a selection event is crucial for MHC
restriction.
It is, however, possible that both B and T cells need to be
positively selected for identical reasons – primarily to test for
the generation of functional receptors and to simultaneously
generate instructive signals for lineage commitment (3, 13).
Evidence that antigen specificity of the BCR is the prime arbiter
of lineage commitment in B lymphocytes are discussed in this
review. This evidence demands a re-examination of the pre-
sumed underlying reasons for positive selection in the T lin-
eage. There is considerable evidence to support the view that
TCRs are intrinsically designed to recognize MHC molecules
(14). It might, therefore, be inferred that there may be no
need to select T cells that see MHC-like shapes, because all
TCRs appear to start out with an MHC bias. Positive selection
in T cells may primarily be designed to ensure that only cells
that have rearranged compatible a- and b-chains, and there-
fore express detectable levels of the ab-TCR on the cell surface,
receive survival signals. The very same MHC self-peptide com-
plexes that mediate positive selection and facilitate lineage
commitment in the thymus also presumably provide trophic
signals for the survival of single-positive T cells in the periph-
ery. MHC restriction is achieved, because it is self-MHC mole-
culesthat delivertrophic
development. In the light of evidence presented in this review
supporting the existence of a positive selection event in B cells,
signalsduringthymocyte
we consider it extremely likely that self-antigens mediate
positive selection and participate in lineage commitment in
both B and T lymphocytes.
Peripheral B-lineage cells
Because one rationale for self-antigen-mediated positive selec-
tion is that this event couples survival signals to lineage com-
mitment, it is necessary in the B-cell context to briefly review
some of the features of cells that belong to defined peripheral
B lineages. Although it has long been considered adequate to
divide peripheral B cells into B1 and B2 lineages, it is now
clear that it would be more useful to classify naı ¨ve peripheral B
cells into at least three distinct lineages (Fig.1), namely mature
follicular (FO), marginal zone (MZ) and B1 B cells. It is also
critical to clearly distinguish between mature cells of these
lineages and their definable and phenotypically distinguishable
precursors. At least five naı ¨ve splenic B-cell populations can
be distinguished, including mature FO B cells, MZ B cells,
newly formed or transitional type-1 B cells, and two other
distinct categories of transitional B cells, follicular precursors
(FP cells) and marginal zone precursors (MZP), which we
discuss in a later section. The immunophenotypic character-
istics of all these populations and issues of nomenclature
are outlined in Table1. Although our focus in this review is
primarily on FO and MZ B lymphocytes because of the cell-fate
decision that links them developmentally, we also briefly
touch upon aspects of the development and biology of B1
cells.
Mature FO B cells (IgDhiIgMlo) are relatively short-lived cells
with a half-life of around 2–3 months. Following the condi-
tional knockout of recombination-activating gene-1 (Rag-1), a
fairly rapid decline in FO B-cell number was noted, whereas
the number of B1 and MZ B cells remained virtually
unchanged up to a year later (15). It is presumed that MZ B
cells (IgMhiIgDlo/–CD23–CD21hiCD1dhi), like B1 cells, are self-
renewing. Whilst FO B cells probably contribute to most
T-dependent germinal center-based responses that originate
in lymph nodes and Peyer’s patches, MZ B cells are presumed
to be critical for T-independent responses to blood-borne
pathogens (16). Because MZ B cells express high levels of
B7-1 and B7-2, they might also present blood-borne antigens
to naı ¨ve T cells and, thus, possibly contribute to T-dependent
responses (17). It has, however, not been formally established
whether MZ B cells can actually contribute to the generation of
germinal center B cells.
B1 cells appear to be primarily involved in the generation of
T-independent responses to antigens that are encountered at
Pillai et al?B-cell selection and lineage commitment
208
Immunological Reviews 197/2004

Page 4

mucosal sites. B1a cells are of fetal liver origin, are abundant in
the peritoneal cavity, and can be defined phenotypically as
IgMhiIgDlo/–CD5þMac-1þB cells (18). B1b B cells are less well
understood, do not express CD5, and are considered later in
this review.
In contrast to the limited understanding we have of B1 B-cell
development, considerable progress has been made fairly
recently in our understanding of how FO and MZ B cells
develop. Some of this knowledge has been obtained by the
manipulation of the BCR/Bruton’s tyrosine kinase (Btk)/
Protein kinase Cb (PKC-b) pathway as described below.
The BCR/Btk/PKC-b pathway is required for the
development of two peripheral B lineages
We reasoned that if the BCR in developing B cells initiates
commitment to more than one type of peripheral B cell, this
receptor must be capable of transducing a range of qualita-
tively or quantitatively distinct signal outputs. Each different
subset of signal outputs would presumably be induced by
groups of self-antigens of differing avidity for their cognate
BCRs. We surmised that if such a view was correct, altering
BCR-signal strength should influence peripheral B-cell fate. An
obvious pathway to modulate downstream of the BCR was the
Btk pathway, because Btk is a key player in the development of
at least two types of naı ¨ve peripheral B cells.
Btk is a tyrosine kinase that is defective in patients with
X-linked agammaglobulinemia and is also mutated in X-linked
immunodeficient mice (19–21). This TEC kinase is activated
downstream of the BCR and the pre-BCR (6, 22, 23). Btk is
required in the mouse for the development or survival of
mature FO B cells and of B1 cells (24–26). Crosslinking of
the BCR induces Src family kinase-mediated tyrosine phos-
phorylation of Iga and Igb, followed by the recruitment and
activation of the Syk tyrosine kinase. Syk in turn contributes to
the tyrosine phosphorylation of specific integral membrane
adapters, the most prominent being B-cell adapter for phos-
phoinositide 3-kinase (BCAP), CD19, and linker for activation
of B cells (LAB) (27–31). BCAP and, probably to a lesser
extent, CD19 mediate phosphoinositide 3-kinase (PI3K) acti-
vation by recruiting the p85a subunit of PI3K via an Src
homology 2 (SH2) domain–phopshotyrosine interaction.
Activated PI3K in turn induces the generation of phosphati-
dylinositol-3,4,5-triphosphate (PIP3) on the inner phase of the
plasma membrane, which recruits Btk to the membrane via the
pleckstrin homology (PH) domain of the latter. The conse-
quent translocation of Btk to the membrane facilitates the Src
family kinase-mediated phosphorylation of tyrosine 551, a
critical residue in the activation segment of the catalytic
domain of Btk (32). Activated Btk cooperates with an adapter
protein called B-cell linker protein (BLNK)/SH2 domain-con-
taining leukocyte-specific phosphoprotein of 65kDa (SLP-65),
with the integral membrane adaptor LAB, and with Syk
to phosphorylate and activate phospholipase C-g (PLC-g)
(33–36). The latter, by virtue of its ability to cleave phos-
phatidylinositol-4,5-bisphosphate(PIP2)into inositol-1,4,
Y
CD5
B1
B CELLs
IgM
YY
FOLLICULAR
B CELLS
CD1
CD21Y
MARGINAL ZONE
B CELLS
IgM
IgM
IgD
CD21
T-independent responses
?T-dependent responses
Immunity to blood borne
pathogens
Self-renewing?
T-independent responses
Mucosal Immunity
B1a cells primarily of
fetal liver origin
Self-renewing
T-dependent responses
T-independent responses
Immune responses in most
secondary lymphoid organs
CD23
Mac-1
Fig.1. A summary of the characteristics of
the three major naı ¨ve peripheral B-cell
populations.
Table1. Categorization of splenic B2 cells
MarkersCurrent Loder et al. (48) Sites
IgMhiIgD–/loCD21–CD23–
IgMhiIgDhiCD21intCD23þ
IgMhiIgDhiCD21hiCD1dhiCD23þ
IgDhiIgMloCD21intCD23þ
IgMhiIgDloCD21hiCD1dhiCD23–
BM, bone marrow; B, blood; S, spleen; LN, lymph nodes.
Newly formed (NF)
FO precursors (T2-FP)
MZ precursors (T2-MZP)
Mature follicular (FO)
Marginal zone (MZ)
T1
T2
T2
M
MZ
BM, B, S
BM, B, S, LN
S
BM, B, S, LN
S
Pillai et al?B-cell selection and lineage commitment
Immunological Reviews 197/2004
209

Page 5

5-triphosphate (IP3) and diacylglycerol, contributes to the activ-
ation of PKCs including PKC-b (Fig.2).
The analysis of the phenotypes of mutant mice has revealed
a surprisingly linear pathway involving the activation of Btk
and PLC-g downstream of the BCR (Fig.2). Knockouts of
CD45, BCAP, Btk, BLNK, and PLC-g, all reveal the loss of B1
cells and of IgDhiIgMlomature FO B cells (24–26, 37–42). MZ
B cells are, however, spared. Downstream of PLC-g, PKC-b
appears to be critical for B1-cell generation or survival, but the
loss of FO B cells is not prominent in this mutant (43). We
conclude, therefore, that there is a branchpoint in the Btk
pathway downstream of PLC-g. One fork at this branchpoint
involves the activation of PKC-b and the subsequent activation
of nuclear factor kB (NF-kB), and this branch of the pathway
is required for efficient BCR-induced B-cell proliferation and
for the generation of B1 B cells. The other branch downstream
of PLC-g is required for the survival of FO B cells and the
biochemical events that may be required for FO B-cell survival
are discussed later in this review.
Semantic issues regarding descriptions of antigen
receptor-signal strength
One of the unique features about signal transduction mediated
by antigen receptors is that these complexes can receive inputs
from a range of physiological ligands, which may possess
widely varying affinities and which may also exist in forms
of considerably divergent valency. The avidity of a ligand–
receptor interaction may not only potentially influence the
degree of receptor crosslinking, but, also, if conformational
change is an option, may lead to varying magnitudes of such
an alteration in receptor or receptor-linked chains. The avidity
of the interaction may also lead to differences in the duration
of signaling. Signal output from these receptors, therefore,
may widely vary in qualitative and quantitative terms. We,
and in general most investigators in the field, use the term
‘signal strength’ in a very broad way. ‘Weak’ BCR signaling,
for instance, might represent the induction of only a few
downstream pathways rather than the entire spectrum of
possible signaling events. It could represent the generation of
a fraction of the total possible output of a second messenger in
a given temporal quantum following receptor ligation. It
might reflect signaling for a very short time through the
BCR. When we, therefore, refer to weak BCR signals initiated
by some subset of antigens, we might more appropriately be
referring to ‘incomplete’ BCR signaling or to ‘a short duration
of’ BCR signaling. It is likely that in the coming years it will
become possible to more precisely and functionally resolve
these differences, but for the time being and for the sake of
simplicity we will use the terms weak, intermediate, and
strong BCR signaling (Fig.3).
The follicular versus marginal zone B-lymphocyte cell-fate
decision: a role for BCR-signal strength in peripheral B-
cell fate
As discussed above, manipulating the signal output from the
BCR/Btk pathway could provide an approach to determine
whether constitutive ligand-independent BCR signaling or
self-antigen-mediated ‘tickling’ of the BCR was crucial during
B-cell development. The rationale for assuming that self-
antigen-mediated tickling might be relevant during peripheral
B-cell development was developed as follows. We reasoned
Btk
PLC-γ
PKC-β
PI3-K
Src-family kinases
NF-κB2 p100
processing
BLNK
NIK/IKK-α
NF-κB activation
NF-κB2 (p52)
BCR-dependent
proliferation and B1
cell generation
Survival of mature
follicular fraction I
B cells
? BAFF-R
CARD-11/Bcl-10/MALT-1
?
BCAP CD45Aiolos
?
Fig.2. The B-cell receptor (BCR)/Bruton’s tyrosine kinase (Btk)/
phospholipase C-g (PLC-g) pathway is required for the development
or survival of mature follicular (FO) and B1 B cells.
Pillai et al?B-cell selection and lineage commitment
210
Immunological Reviews 197/2004