Download Fisher 2002 - Salamander Genome Project

PERSPECTIVES
OPI N ION-DECIS ION MAK I NG I N TH E I MMU N E SYSTE M
Cellular identity and lineage choice
Amanda G. Fisher
In multicellular organisms, cells usually
respond to signals that they encounter in
a manner that depends on their particular
lineage ‘identity’. In other words, cells
that have identical genomes can
respond in markedly different ways to the
same stimulus, with the outcome being
determined largely by the previous
developmental history of the cell. This
general observation implies that individual
somatic cells retain a ‘working memory’ of
their ancestry and that this epigenetic
information can be passed through
successive rounds of DNA replication and
cell division. Here, I discuss whether recent
advances in our knowledge of chromatin
biology and gene silencing can provide new
insights into how cell fate is chosen and
maintained during development.
The formation of blood cells from haematopoietic stem cells (HSCs) has been studied
intensively and remains a model for understanding cell commitment. Although standard texts illustrate a hierarchical relationship
between HSCs and common myeloid and
common lymphoid precursors, other haematopoietic ‘fate-maps’ have been proposed
also. Each model is supported by different
types of evidence and each offers a different
interpretation of the precursor–product relationships between haematopoietic cells. For
example, in the hierachical model (FIG. 1a), the
first step proposes the generation of a progenitor that gives rise to megakaryocytes, erythrocytes, granulocytes and monocytes (a
common myeloid precursor, CMP), as well
as a precursor for T and B cells (a common
lymphoid precursor, CLP). Evidence for this
model comes from early demonstrations that
chromosome-marked bone marrow can
repopulate the myeloid or lymphoid compartments of transplant recipients1,2. The
subsequent identification of progenitors that
have the functional and phenotypic characteristics of CMPs3 and of CLPs4, and a comparison of their transcriptional profiles5, have
added weight to the argument that myeloid
versus lymphoid choice is a primary decision.
In the second scheme (FIG. 1b), the commitment decision of haematopoietic precursors is considered to be stochastic — that is, at
any one time, the choice is essentially random, with success being determined only
later by the availability of essential growth
and survival factors (a stochastic–selective
model). Some support for this model can be
taken from studies showing the co-expression
of several distinct lineage-affiliated programmes by multipotent haematopoietic
stem cells and progenitor cells6,7. In addition,
assuming that cellular identity is determined
ultimately at the level of many individual
genes, there is growing evidence that gene
activation has a stochastic basis, from which
‘all or nothing’ outcomes of gene expression
can result. Examples of this include the
expression of globin genes in developing erythroblasts8 and genetic studies of the mechanisms that underlie position-effect variegation
(discussed in REF. 9).
A third, radically different, model was put
forward by Brown and colleagues in 1985
(REF. 10). The sequential lineage-determination
model proposed that there is a predetermined
(or prescribed) order of developmental choices
(FIG. 1c). This model was based on experimental
data showing the common origins and shared
NATURE REVIEWS | IMMUNOLOGY
features of different haematopoietic cells. The
model proposes that HSCs undergo an
intrinsic programme of decisions to generate
cells that can differentiate along one or, at
most, two discrete pathways. Support for the
model was provided by quantitative studies
of the rate of production of various haematopoietic cells11 and, more interestingly, by
detailed analysis of the cell types that are
absent from mice lacking key transcriptional
regulators12.
To illustrate the fundamental differences
between these three models, it is perhaps
worth using as an analogy the selection of
careers by students during their secondary
(high-school) education. According to the
first model, students would be encouraged
to make a fundamental decision early on,
either to study arts or sciences. Within these
disciplines, different career options are then
offered (for example, medicine or fine art),
as well as similar ones that differ only in
their particular speciality (for example, to
become a teacher). According to the second
model, career choice is mainly vocational,
the outcome depending on the individual
student and whether there is a large enough
job market for the skills on offer. The third
model proposes a progressive education, in
which candidates are offered career options
in a defined order. If the offer is declined,
then students continue their education
(according to a defined syllabus) until they
are qualified for the next round of recruitment. However, when reviewing these
contrasting descriptions of lineage (or
career) choice, it is important to understand
that alternative schemes should be considered13 and that hierarchical models could
also involve stochastic events or a sequential requirement for transcription factors.
According to the career-choice analogy, one
could imagine that a chance encounter with
an exceptionally charismatic art or chemistry
teacher might determine the career choice of
a student, which is an example of randomness of choice in a prescribed educational
structure.
VOLUME 2 | DECEMBER 2002 | 9 7 7
© 2002 Nature Publishing Group
PERSPECTIVES
differentiation. These protein complexes are
postulated to undergo successive changes in
composition, a feature that might allow
them to acquire and carry out new functions
Transcription and cell identity
During haematopoiesis, transcription factors are involved in many protein–protein
interactions that define lineage and stage of
a
Stem cell
CLP
T cell
CMP
B cell
Monocyte
Neutrophil
Erythrocytes
Megakaryocyte
b
T cell
B cell
Monocyte
Stem cell
c
Stem cell
Megakaryocyte
Neutrophil
Me
E
Erythrocytes
Megakaryocyte
G
Erythrocytes
M
Neutrophil
B
Monocyte
T
B cell
T cell
Figure 1 | Contrasting models of haematopoiesis. Three alternative schemes describing the
generation of blood cells from primitive haematopoietic stem cells are shown. a | In the hierachical model,
the first decision is to become a progenitor committed to either a lymphoid fate (a common lymphoid
progenitor; CLP) or a myeloid fate (a common myeloid progenitor; CMP). Thereafter, CLPs generate
T or B cells, whereas CMPs give rise to a range of other cell types, including monocytes, neutrophils,
erythrocytes and megakaryocytes. b | The stochastic–selective model proposes that cell fate is chosen
randomly, with success being determined by the availability of appropriate survival and differentiation
factors. c | In the sequential-determination model, the order of developmental decisions is pre-determined,
with cells making successive choices to differentiate along one or, at most, two discrete pathways.
Precursors derived from haematopoietic stem cells progressively express the potential for megakaryocyte
(Me), erythrocyte (E), granulocyte (G), monocyte (M), B-cell and T-cell development. At each developmental
stage, a decision is made to either execute the differentiation programme that is on offer or proceed to the
next cell-fate option.
978
| DECEMBER 2002 | VOLUME 2
(discussed in REF. 14). One of the best-characterized examples of this is the interaction
of the transcription factor GATA1 with many
partners, including FOG1 (friend of GATA1),
EKLF (erythroid Kruppel-like factor), SP1,
CBP (CREB-binding protein)/p300 and PU.1
(REF. 15). There is compelling evidence that by
manipulating the composition of transcription-factor complexes experimentally, different lineage outcomes can be favoured. This is
shown most convincingly by examples of the
re-specification of lineage fate induced by
transcriptional regulators. For example,
Heyworth, Enver and colleagues16 showed
recently that in response to ectopic expression
of GATA1, neutrophil/monocyte precursors
are reprogrammed to take on erythroid,
eosinophil and basophil-like cell fates. This
demonstration of the re-specification of primary cells was based on early studies showing
the antagonistic roles of PU.1 and GATA1 in
myeloid and erythroid differentiation17,18,19,
and the ability of GATA1 to reprogram avian
myelomonocytic precursors20.
Classical studies have shown also that single transcription factors, such as MyoD
(myogenic differentiation)21, can specify cell
lineage actively. In these experiments, the
expression of MyoD was shown to be sufficient to generate muscle cells from a range of
cell types, including fibroblasts and pigmented epithelial cells of the retina21,22. The
importance of single transcription factors
for dictating lineage fate is also clear from
studies of the transcriptional regulator PAX5
(paired box gene 5)23. In this case, PAX5
functions not only by activation of the
B-cell programme, but also by repression of
other lineage fates — a fact that is evident at
the molecular level by the activation or
repression of the respective lineage-specific
genes. For example, PAX5 is required for
B-cell differentiation as its removal favours
the development of alternative haematopoietic cells from haematopoietic precursors24.
Attempts to restore B-cell development by
supplying exogenous PAX5 to genetically
deficient cells have indicated that PAX5
is required more or less continuously to
maintain B-cell identity25,26. PAX5 can function as both a conventional activator of
transcription (for example, in the case of
the CD19 gene27) and a potent transcriptional repressor (through interaction with
Groucho-like co-repressor proteins28).
Together, these data highlight the importance of transcriptional activators and
repressors for regulating lineage decisions,
and they might indicate that commitment
to a particular lineage is more flexible than
was thought previously.
www.nature.com/reviews/immunol
© 2002 Nature Publishing Group
PERSPECTIVES
Stable gene silencing
•
•
•
•
HAT activity
Active gene
transcription
HDAC activity
Chromatin compaction
Locus repositioning
CpG methylation
Polycomb-group recruitment
HMT activity
SUV39H
E(z)–Esc
SUV39H
Ac
Ac
Ac
P
P
P
Me
Me
P
P
Me HP1
Me
P
Priming
Active euchromatin
E(z)–Esc
Me Me PC
Me
Neutral or permissive
chromatin
Repressed chromatin
Figure 2 | A chromatin-based model of gene activation and silencing. DNA complexed with core histones and other chromosomal proteins forms
chromatin. Gene activation and silencing are associated with characteristic changes in chromatin structure, which include specific modifications to core histone
tails — such as acetylation (Ac), phosphorylation (P), methylation (Me) and ubiquitylation (not shown) — as well as changes in the degree of condensation of
nucleosome fibres. In this example, an actively transcribed (euchromatic) locus is shown, in which the nucleosome fibre is accessible and acetylated (by histone
acetyltransferases, HATs). This state might be stabilized further by interaction with Trithorax-group proteins and the selective methylation of, for example, histone 3
at lysine-4 (H3-K4; not shown). Gene silencing is shown as a sequential multi-step process, in which the degree of acetylation is reduced by histone deacetylases
(HDACs) and site-specific methylation is achieved by recruitment of various histone methyltransferases (HMTs), including SUV39H (which methylates H3-K9) and
the extra sex combs (Esc)–enhancer of Zeste (E(z)) complex (which methylates H3-K9 and H3-K27). Methyl-docking partners, such as heterochromatin protein 1
(HP1) and Polycomb (PC), that recognize these specific modifications, stabilize heterochromatin structure and condensation, and reduce accessibility. Silencing or
‘marking’ of an inactive locus might be enhanced further by the recruitment of additional PC-group proteins, CpG DNA methylation and locus repositioning in the
nucleus. SUV39H, suppressor of variegation 3-9 homologue.
Conventionally, cellular identity has been
described in positive terms — that is, in terms
of the set of expressed genes that give a cell its
distinguishing character. It is worth pointing
out that an equally valid, if less intuitive,
description of identity could be based on the
extent of the genome that is not expressed or
is actively silenced. In fact, this ‘negative’
description of cell identity pre-dates much of
our present fascination with molecular signatures of gene expression (discussed in REF. 29).
The idea that lineage-specific transcription
factors function not only by activating gene
expression, but also by repressing inappropriate genes, has several important precedents, particularly with regard to functional
studies of the transcription factors MAFB
(musculoaponeurotic fibrosarcoma oncogene
homologue B)30, PU.1 (REF. 31) and GATA1
(REF. 32). Despite accumulating evidence that
gene repression is crucial for determining
haematopoietic and neural cell fates33,34, less
attention has been focused on gene silencing
that on gene activation.
Access to accurate gene-expression profiles
for HSCs, lineage-committed precursors35,36,
neural stem cells and embryonic stem (ES)
cells might allow us to distinguish overlapping sets of genes that are likely to be concerned with, for example, stem-cell identity
(self renewal and differentiation). Recently,
Ivanova and Lemischka37 identified a group of
genes that are enriched in HSCs, ES cells and
neural stem cells, and they showed that these
genes encode several transcription factors that
were known previously to sustain the activity
of HSCs (such as early developmental regulator 1, EDR1) and epidermal stem cells (such
as transcription factor 3, TCF3), and to regulate the proliferation of neural stem cells
(such as ephrin-B2, EFNB2; and hairy and
enhancer of split 1, HES1). This kind of
analysis, although in its infancy, offers some
promise. If it can be applied to mammalian
cells with the same success as has been
achieved for Saccharomyces cerevisiae (a simple eukaryote), then we will have an ‘information framework’ with which to begin to define
the ‘molecular circuitry’ of haematopoietic
cells during differentiation. The complexity
of this task should not be underestimated
(discussed in REF. 38), and complementary
approaches are likely to be required to advance
our understanding of haematopoiesis in the
interim.
Chromatin and lineage restriction
Gene expression is determined not only by
the availability of combinations of transcription factors, but also by chromatin context.
The interactions between transcription factors that occur during haematopoietic differentiation have been likened to a cocktail
party, in which the introduction of new
guests (and exit of old guests) encourages
new topics for discussion14. According to this
NATURE REVIEWS | IMMUNOLOGY
analogy, chromatin structure can be thought
of as the venue for the cocktail party. Just as
one might anticipate that at a gentlemen’s
club (to which women are not admitted) or at
a venue where alcohol is prohibited, the topics
of conversation might be more restricted (and
less cosmopolitan), so the chromatin context
of specific genes might restrict or encourage
particular transcriptional outcomes. Formal
proof that gene expression is not dictated by
transcription factors alone is provided by several well-characterized examples of imprinted
genes, for which expression of a single allele is
predetermined according to the parent of origin39. It is probable also that chromatin-based
self-templating mechanisms are involved in
conveying transcriptional states through the
cell cycle. This might be particularly important for maintaining lineage identity in
haematopoietic cells, such as HSCs and lymphocytes, for which proliferation is an essential part of the function of each cell type.
Several epigenetic modifications that are
associated with active and inactive chromatin
states have provided clues as to how such selfpropagating mechanisms might operate. For
example, DNA methylation — which is
important for the transcriptional repression
of imprinted genes, as well as genes on the
inactive X-chromosome and candidate tissuespecific genes40 — is mediated by specific
DNA methyltransferases (DNMTs). In the
case of DNMT1, this enzyme is recruited to
VOLUME 2 | DECEMBER 2002 | 9 7 9
© 2002 Nature Publishing Group
PERSPECTIVES
replication forks at S-phase41, where it establishes symmetrical CpG methylation of hemimethylated substrates, effectively duplicating
DNA-methylation patterns on newly synthesized DNA strands. This property, together
with the action of methyl-CpG-bindingdomain proteins (such as MECP2), acts to
repress transcription and recruit histone
deacetylases, providing a possible mechanism
to propagate transcriptionally inactive
states42,43. Other features of active and inactive
chromatin that might be relevant to understanding how transcriptional states are effectively ‘locked-in’ in differentiated cells include
the covalent modification of histone tails44
and the spatial restriction of loci to certain
nuclear domains (reviewed in REF. 45). It is
an interesting consideration how specific
genes are targeted for this type of regulation. It is assumed that sequence-specific
DNA-binding factors that activate or repress
transcription either can recruit protein partners that are capable of further chromatin
modifications and locus recruitment, or can
carry out such functions themselves (discussed in REFS 45, 46).
On the basis of several recent reports, it
is possible to construct a hierarchy of interrelated epigenetic changes that provide a
plausible mechanistic bridge between transcriptionally active, permissive, repressive
and permanently silent chromatin states.
Although the scheme that is illustrated in
FIG. 2 is far from complete, these new studies
provide three important clues to understanding chromatin-based transcriptional regulation. First, histone methylation at specific
residues can recruit additional components
that further stabilize a repressive chromatin
state. For example, the methylation of
lysine-9 of histone 3 (H3) by SUV39H (suppressor of variegation 3-9 homologue)
recruits the structural heterochromatin protein HP1 (REFS 47,48). Similarly, the methylation of H3 lysine-9 and lysine-27 by extra
sex combs (Esc) and enhancer of Zeste
(E(z)) is postulated to recruit Polycomb
(Pc), a component of the Polycomb repressor complex 1 (PRC1), thereby enhancing
repression and permanently ‘marking’ the
silent state49,50. Second, it has been shown
that histone methylation and DNA methylation are inter-dependent51, which provides a
mechanism by which protein and DNA
modifications can ‘converse’52. Third, recent
studies highlight an increasingly important
role for intergenic and sterile (non-coding
transcripts) transcription for the recruitment of chromatin modifiers (in particular,
histone acetyltransferases) before bona fide
gene transcription53,54.
980
Pro-T cell
IgH locus
Constitutive heterochromatin
comprising 'clusters' of centromeric
DNA and associated proteins
ES cell
Relocation
Locus
Pro-B cell
compaction
Gene
rearrangement
Immunoglobulin
alleles are peripheral
Mature B cell
Allelic
choice
Immunoglobulin alleles
are poised and central
Monoallelic
expression
Immunoglobulin alleles
are non-equivalent
Figure 3 | Repositioning of immunoglobulin alleles in developing B cells. Immunoglobulin heavychain (IgH) and light-chain (IgL) loci undergo large-scale changes in their sub-nuclear position and
compaction during B-cell development. In embryonic stem (ES) cells and T-cell progenitors (pro-T cells),
immunoglobulin alleles (red circles) are located adjacent to the nuclear periphery and away from clusters of
centromeric heterochromatin (shown in grey). In pro-B cells, before the onset of IgH rearrangement, IgH
alleles are repositioned in the interior of the nucleus and undergo locus compaction. After immunoglobulin
gene rearrangement and allelic choice, a single productively rearranged and transcribed IgH allele (and a
single IgL allele) is positioned away from heterochromatin clusters. Unrearranged and non-productively
rearranged immunoglobulin alleles are recruited close to centromeric heterochromatin domains in the
nucleus of mature B cells.
Interestingly, the low levels of transcription of lineage-affiliated genes that are seen in
HSCs, CLPs and MLPs4,16 might be owing to
the selective ‘opening-up’ of chromatin (or
priming) in precursor populations. An alternative interpretation of these observations is
that the apparently ‘promiscuous’ transcription indicates that changes from permissive to
repressive chromatin states occur in committed progenitors once a particular fate or
option has been excluded. These contrasting
possibilities illustrate differing viewpoints on
the probable contribution of sequential gene
activation versus progressive gene silencing to
commitment and lineage restriction29. An
additional feature of the multi-layered model
of chromatin events that promote either gene
activation or heritable silencing (FIG. 2) is that
it might offer some clues about cell plasticity.
According to such a model, it might be relatively easy to reprogram cells that retain a permissive chromatin configuration, but more
difficult to re-specify cells that have already
shut down expression of key genes.
| DECEMBER 2002 | VOLUME 2
Chromatin and nuclear location
The combined role of chromatin-based
mechanisms and transcription factors in
controlling lineage fate is exemplified by the
development of T helper (TH)-cell subsets.
Naive CD4+ progenitors can be induced by
different stimuli to differentiate to two functionally different types of cell that express
either interferon-γ (IFN-γ; TH1 cells) or interleukin-4 (IL-4; TH2 cells). Differentiation
results in modifications to the chromatin
structure of the effector-cytokine genes55,
including changes in DNA demethylation
and methylation56, histone acetylation and
accessibility to NFAT1 (nuclear factor of activated T cells 1) at locus-regulatory regions57.
These modifications, which are progressive,
help to polarize the developing cells, so that
expression of IL-4 and IFN-γ is mutually
exclusive. The silencing of expression of
IL-4 or IFN-γ might be stabilized further by
the large-scale recruitment of these loci to
specific heterochromatin domains in the
nucleus 58. Interestingly, with successive
www.nature.com/reviews/immunol
© 2002 Nature Publishing Group
PERSPECTIVES
division, progeny cells loose the ability to
revert to the alternative lineage, an observation that is consistent with lineage restriction
being underpinned by progressive gene
silencing.
Although the role of nuclear ‘architecture’
in regulating cellular gene expression is not
well understood, there are specific and compelling examples in which gene recruitment
to heterochromatin can cause silencing59 and
gene movement away from heterochromatin
(or from chromosome territories) is a feature
of expression60–62. In this respect, several socalled ‘nuclear compartments’ have been
described, and two of these — the nuclear
periphery and centromeric heterochromatin
— seem to be important for propagating or
restricting expression of certain genes in
haematopoietic cells. One example, which is
shown in FIG. 3, is the relocation of immunoglobulin alleles in developing B cells. Kosak
et al.63 showed recently that in ES cells, multipotent bone-marrow precursors and pro-T
cells, immunoglobulin heavy-chain (IgH)
and immunoglobulin κ-light-chain (Igκ)
alleles occupy perinuclear positions close to
the nuclear lamina. In committed pro-B cells,
these loci are selectively repositioned away
from this compartment immediately before
immunoglobulin-gene rearrangement — an
observation that indicates that relocation
might regulate the lineage- and stage-specific
accessibility of immunoglobulin loci to chromatin modifiers and the recombination
machinery. Later, after the productive rearrangement of single immunoglobulin
heavy- and light-chain genes, Skok and colleagues64 showed that individual immunoglobulin alleles that are ‘selected against’
(germline or non-productive rearrangements) are repositioned close to repressive
heterochromatin clusters, resulting in expression from a single IgH and Igκ (or Igλ) locus.
Amanda G. Fisher is at the Lymphocyte
Development Group, Medical Research Council
Clinical Sciences Centre, Faculty of Medicine,
Imperial College of Science,
Technology and Medicine, Hammersmith Campus,
Du Cane Road, London W12 0NN, UK.
e-mail: [email protected]
doi:10.1038/nri958
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
Concluding remarks
Our knowledge of the relationship between
chromatin structure, chromosome movement and cell-fate choice is still fragmentary. However, there is increasing evidence
that epigenetic modifications — including
chromatin remodelling, large-scale locus
repositioning and locus condensation —
leave traceable ‘marks’ on the genome. If
these marks can be interpreted, then it
might allow a unique opportunity to ‘glance
backwards’ at the previous developmental
history of a cell. Such information might be
useful to validate or disprove the various
haematopoietic cell-fate maps and to examine how lineage decisions are made from a
fresh viewpoint.
18.
19.
20.
21.
22.
23.
Becket, A. J., MucCulloch, A. E. & Till, J. E. Cytological
demonstration of the clonal nature of spleen colonies
derived from transplanted mouse marrow cells. Nature
197, 452 (1963).
Abramson, S., Miler, R. G. & Phillips, R. A. The
identification in adult bone marrow of pluripotent and
restricted stem cells of the myeloid and lymphoid
systems. J. Exp. Med. 145, 1567 (1978).
Akashi, K., Traver, D., Miyamoto, T. & Weissman, I. L.
A clonogenic common myeloid progenitor that gives rise
to all myeloid lineages. Nature 404, 193–197 (2000).
Kondo, M., Weissman, I. L. & Akashi, K. Clonogenic
common lymphoid progenitors in mouse bone marrow.
Cell 91, 661–672 (1997).
Miyamoto, T. et al. Myeloid or lymphoid promiscuity as a
critical step in hematopoietic lineage commitment. Dev.
Cell 3, 137–147 (2002).
Hu, M. et al. Multilineage gene expression precedes
commitment in the hemopoietic system. Genes Dev.
11, 774–785 (1997).
Enver, T., Heyworth, C. M. & Dexter, T. M. Do stem cells
play dice? Blood 2, 348–351 (1998).
De Krom, M., van de Corput, M., von Lindern, M.,
Grosveld, F. & Stouboulis, J. Stochastic patterns in
globin gene expression are established prior to
transcriptional activation and are clonally inheritied
Mol. Cell 9, 1319–1326 (2002).
Festenstein, R. & Kioussis, D. Locus-control regions and
epigenetic chromatin modifiers. Curr. Opin. Genet. Dev.
10, 199–203 (2000).
Brown, G., Bunce, C. M. & Guy, G. R. Sequential
determination of lineage potentials during haemopoiesis.
Br. J. Cancer 52, 681–686 (1985).
Novak, J. P. & Stewart, C. C. Stochastic versus
deterministic in haemopoiesis: what is what?
Br. J. Haematol. 78, 149–154 (1991).
Singh, H. Gene targeting reveals a hierarchy of
transcription factors regulating specification of lymphoid
cell fates. Curr. Opin. Immunol. 8, 160–165 (1996).
Lu, M., Kawamoto, H., Katsube, Y., Ikawa, T. & Katsura, Y.
The myelolymphoid progenitor: a key intermediate stage
in hemopoiesis generating T and B cells. J. Immunol.
169, 3519–3525 (2002).
Sieweke, M. H. & Graf, T. A transcription factor party
during blood-cell differentiation. Curr. Opin. Genet. Dev.
8, 545–551 (1998).
Cantor, A. B. & Orkin, S. H. Transcriptional regulation of
erythropoiesis: an affair involving multiple partners.
Oncogene 21, 3368–3376 (2002).
Heyworth, C., Pearson, S., May, G. & Enver, T.
Transcription factor-mediated lineage switching reveals
plasticity in primary committed progenitor cells. EMBO J.
21, 3770–3781 (2002).
Rekhtman, N., Radparvar, F., Evans, T. & Skoultchi, A. I.
Direct interaction of haemopoietic transcription factors
PU.1 and GATA-1: functional antagonism in erythroid
cells. Genes Dev. 13, 1398–1411 (1999).
Zhang, P. et al. PU.1 inhibits GATA-1 function and
erythroid differentiation by blocking GATA-1 DNA binding.
Blood 96, 2641–2648 (2000).
Visvader, J. E., Crossley, M., Hill, J., Orkin, S. H. &
Adams, J. M. The C-terminal zinc finger of GATA-1
or GATA-2 is sufficient to induce megakaryocytic
differentiation of an early myeloid cell line. Mol. Cell. Biol.
15, 634–641 (1995).
Kulessa, H., Frampton, J. & Graf, T. GATA-1 reprograms
avian myelomonocytic cell lines into eosinophils,
thromboblasts and erythrocytes. Genes Dev.
9, 1250–1262 (1995).
Lassar, A. B., Paterson, B. M. & Weintrub, H. Transfection
of a DNA locus that mediates the conversion of 10T1/2
fibroblasts to myoblasts. Cell 47, 649–656 (1986).
Choi, J. et al. MyoD converts primary dermal fibroblasts,
chondroblasts, smooth muscle and retinal pigmented
epithelial cells into striated mononucleated myoblasts
and multinucleated myotubes. Proc. Natl Acad. Sci. USA
87, 7988–7992 (1990).
Nutt, S. L., Heavey, B., Rolink, A. G. & Busslinger, M.
Commitment to the B-lymphoid lineage depends on the
transcription factor Pax5. Nature 401, 556–562 (1999).
NATURE REVIEWS | IMMUNOLOGY
24. Busslinger, M., Nutt, S. L. & Rolink, A. G. Lineage
commitment in lymphopoiesis. Curr. Opin. Immunol.
2, 151–158 (2000).
25. Mikkola, I., Heavey, B., Horcher, M. & Busslinger, M.
Reversion of B-cell commitment upon loss of Pax5
expression. Science 297, 110–113 (2002).
26. Horcher, M., Souabni, A. & Busslinger, M. Pax5/BSAP
maintains the identity of B cells in late B lymphopoiesis.
Immunity 6, 779–790 (2001).
27. Kosmik, Z., Wang, S., Dorfler, P., Adams, B. &
Busslinger, M. The promoter of the CD19 gene is a target
for the B-cell-specific transcription factor BSAP. Mol. Cell.
Biol. 6, 2662–2672 (1992).
28. Eberhard, D., Jimenez, G., Heavey, B. & Busslinger, M.
Transcriptional repression by Pax5 (BSAP) through
interaction with corepressors of the Groucho family.
EMBO J. 10, 2292–2303 (2000).
29. Fisher, A. G. & Merkenschlager, M. Gene silencing, cell
fate and nuclear organisation. Curr. Opin. Genet. Dev.
12, 193–197 (2002).
30. Sieweke, M. H., Tekotte, H., Frampton, J. & Graf, T. MafB
is an interaction partner and repressor of Ets-1 that
inhibits erythroid differentiation. Cell 85, 49–60 (1996).
31. Zhang, P. et al. Negative cross-talk between
hematopoietic regulators: GATA proteins repress PU.1.
Proc. Natl Acad. Sci. USA 96, 8705–8710 (1996).
32. Nerlov, C., Querfurth, E., Kulessa, H. & Graf, T. GATA-1
interacts with the myeloid PU.1 transcription factor and
represses PU.1-dependent transcription. Blood 95,
2543–2551 (2000).
33. Muhr, J., Anderson, E., Persson, M., Jessel, T. M. &
Ericson, J. Groucho-mediated transcriptional repression
establishes progenitor-cell pattern and neuronal fate in
the ventral neural tube. Cell 104, 861–873 (2001).
34. Marquardt, T. & Pfaff, S. L. Cracking the transcriptional
code for cell specification in the neural tube. Cell 106,
651–654 (2001).
35. Phillips, R. L. et al. The genetic program of hematopoietic
stem cells. Science 288, 1635–1640 (2000).
36. Muschen, M. et al. Molecular portraits of B-cell lineage
commitment. Proc. Natl Acad. Sci. USA 99,
10014–10019 (2002).
37. Ivanova, N. B. et al. A stem-cell molecular signature.
Science 298, 601–604 (2002).
38. Holstege, F. C. & Young, R. A. Transcriptional regulation:
contending with complexity. Proc. Natl Acad. Sci. USA
96, 2–4 (1999).
39. Ohlsson, R., Tycko, B. & Sapienza, C. Monoallelic
expression: ‘there can only be one’. Trends Genet.
11, 435–438 (1998).
40. Bird, A. The essentials of DNA methylation. Cell 10, 5–8
(1992).
41. Leonhardt, H., Page, A. W., Weier, H. U. & Bestor, T. H.
A targeting sequence directs DNA methyltransferase to
sites of DNA replication in mammalian nuclei. Cell 71,
865–873 (1992).
42. Bird, A. P. & Wolffe, A. P. Methylation-induced repression
— belts, braces and chromatin. Cell 99, 451–454 (1999).
43. Allshire, R. & Bickmore, W. Pausing for thought on the
boundaries of imprinting. Cell 102, 705–708 (2000).
44. Jenuwein, T. & Allis, C. D. Translating the histone code.
Science 293, 1074–1080 (2001).
45. Gasser, S. M. Positions of potential: nuclear organization
and gene expression. Cell 104, 639–642 (2001).
46. Zhang,Y. & Reinberg, D. Transcription regulation by
histone methylation: interplay between different covalent
modifications of the core histone tails. Genes Dev. 15,
2343–2360 (2002).
47. Lachner, M., O’Carroll, D., Rea, S., Mechtler, K. &
Jenuwein, T. Methylation of histone H3 lysine-9 creates a
binding site for HP1 proteins. Nature 410, 116–120 (2001).
48. Bannister, A. J. et al. Selective recognition of methylated
lysine-9 on histone H3 by the HP1 chromo domain.
Nature 410, 120–124 (2001).
49. Müller, J. et al. Histone methyltransferase activity of a
Drosophila polycomb-group repressor complex. Cell
111, 197–208 (2002)
50. Czermin, B. et al. Drosophila enhancer of Zeste/ESC
complexes have a histone H3 methyltransferase activity
that marks chromosomal polycomb sites. Cell 111,
185–196 (2002).
51. Tamaru, H. & Selker, E. U. A histone H3
methyltransferase controls DNA methylation in
Neurospora crassa. Nature 414, 277–283 (2001).
52. Bird, A. Molecular biology. Methylation talk between
histones and DNA. Science 294, 2113–2115 (2002).
53. Gribnau, J., Diderich, K., Pruzina, S., Calzolari, R. &
Fraser, P. Intergenic transcription and developmental
remodeling of chromatin subdomains in the human
β-globin locus. Mol. Cell 2, 377–386 (2000).
VOLUME 2 | DECEMBER 2002 | 9 8 1
© 2002 Nature Publishing Group
PERSPECTIVES
54. Schubeler, D. et al. Nuclear localization and histone
acetylation: a pathway for chromatin opening and
transcriptional activation of the human β-globin locus.
Genes Dev. 14, 940–950 (2000).
55. Agarwal, S. & Rao, A. Modulation of chromatin structure
regulates cytokine gene expression during T-cell
differentiation. Immunity 9, 765–775 (1998).
56. Lee, D. U., Agarwal, S. & Rao, A. TH2-lineage
commitment and efficient IL-4 production involves
extended demethylation of the IL-4 gene. Immunity
5, 649–660 (2002).
57. Avni, O. et al. TH-cell differentiation is accompanied by
dynamic changes in histone acetylation of cytokine
genes. Nature Immunol. 7, 643–651 (2002).
58. Grogan, J. L. & Locksley, R. M. T-helper cell
differentiation: on again, off again. Curr. Opin. Immunol.
3, 366–372 (2002).
59. Dernburg, A. F. et al. Perturbation of nuclear architecture
by long-distance chromosome interactions. Cell 85,
745–759 (1996).
60. Francastel, C., Walters, M. C., Groudine, M. & Martin D. I.
A functional enhancer suppresses silencing of a
transgene and prevents its localization close to
centromeric heterochromatin. Cell 99, 259–269 (1999).
61. Volpi, E. V. et al. Large-scale chromatin organization of
the major histocompatibility complex and other regions
of human chromosome 6 and its response to interferon
in interphase nuclei. J. Cell Sci. 113, 1565–1576
(2000).
62. Williams, R. E., Broad, S., Sheer, D. & Ragoussis, J.
Subchromosomal positioning of the epidermal
differentiation complex (EDC) in keratinocyte and
lymphoblast interphase nuclei. Exp. Cell Res. 272,
163–175 (2002).
63. Kosak, S. et al. Subnuclear compartmentalization of
immunoglobulin loci during lymphocyte development.
Science 296, 158–162 (2002).
64. Skok, J. A. et al. Nonequivalent nuclear location of
immunoglobulin alleles in B lymphocytes. Nature
Immunol. 2, 848–854 (2002).
Acknowledgements
I would like to thank my colleagues, in particular members of the
Lymphocyte Development Group, for helpful discussions.
Online links
DATABASES
The following terms in this article are linked online to:
LocusLink: http://www.ncbi.nlm.nih.gov/LocusLink/
CBP | CD19 | DNMT1 | EDR1 | EFNB2 | EKLF | FOG1 | GATA1 |
H3 | HES1 | IFN-γ | IL-4 | MAFB | MECP2 | MyoD | NFAT1 |
p300 | PAX5 | PRC1 | PU.1 | SP1 | SUV39H | TCF3
Access to this interactive links box is free online.
OPI N ION-DECIS ION MAK I NG I N TH E I MMU N E SYSTE M
Progressive differentiation and
selection of the fittest in the
immune response
Antonio Lanzavecchia and Federica Sallusto
T cells are stimulated by stochastic
exposure to antigen-presenting cells and
cytokines. We review evidence that the level
of signal that is accumulated determines
progression through hierarchical thresholds
for proliferation and differentiation, leading to
the generation of various intermediates and
effector T cells. These cells are then
selected to enter the memory pool
according to their fitness — that is, their
capacity to access and use survival signals.
We suggest that the intermediates that are
generated by antigenic stimulation of T and
B cells persist as central memory cells,
which can mount secondary responses to
antigen and maintain appropriate levels of
effector cells and antibodies throughout the
lifetime of an individual.
Antigenic stimulation can lead to divergent
responses that range from the deletion of
antigen-specific lymphocytes and tolerance to
the generation of a large number of effector
cells, followed by establishment of immunological memory. The mechanisms that account
for these divergent responses remain uncertain. In this article, we elaborate further on a
982
progressive-differentiation model that we
have proposed previously for T cells1, and we
discuss this model in the context of present
experiments and competing models. The
progressive-differentiation model (BOX 1) proposes that, as a function of the level of signal
that is accumulated, T cells progress through
hierarchical thresholds for proliferation and
DIFFERENTIATION. Stochastic exposure to antigenpresenting cells (APCs) and cytokines, owing
to random encounters of variable duration,
results in the generation of different fates that
are then selected on the basis of their capacity
to survive, in the absence of antigen, as distinct
populations of memory T cells. This model,
which can be extended to B-cell differentiation, proposes that the INTERMEDIATES of the
differentiation process form a pool of CENTRALMEMORY CELLS with stem-cell-like properties.
Signal strength and T-cell fate
The amount of signal (SIGNAL STRENGTH) that
T cells receive by interacting with APCs is
determined by three factors: the concentration
of peptide–MHC complexes, which determines the rate of T-cell receptor (TCR) triggering2; the concentrations of co-stimulatory
| DECEMBER 2002 | VOLUME 2
molecules, which determine the extent of signal amplification3; and the duration of the
interaction between T cells and APCs, which
determines for how long signal is accumulated. The amount of signal that T cells receive
can vary over several orders of magnitude.
The number of peptide–MHC complexes displayed by an APC can vary by a factor of 103,
and the degree of signal amplification provided by co-stimulation can vary by a factor
of 102 (REFS 3,4). In addition, T cells can engage
APCs in single or multiple contacts of variable duration5,6. The concept of signal accumulation is supported by the finding that the
commitment of naive T cells to proliferation
can be reached in ~6 hours if T cells are
stimulated by a high dose of antigen plus costimulation, but requires as long as 40 hours
if T cells are exposed to low doses of antigen
and co-stimulation7.
T-cell differentiation involves the regulation of transcriptional programmes that control the cell cycle, response to cytokines,
migratory capacity, effector function and susceptibility to activation-induced cell death
(AICD)8–11. In general, T-cell differentiation is
a slow and progressive process that is mediated by EPIGENETIC changes12. After primary
stimulation, effector function is acquired by
only a fraction of the proliferating T cells, and
this fraction increases after subsequent restimulations. The main hypothesis on which
the progressive-differentiation model is based
is that some transcriptional programmes are
activated at a low strength of stimulation,
whereas others require a higher strength of
stimulation, as well as additional signals delivered by cytokines. By exposing CD4+ T cells to
plastic-bound TCR ligands and co-stimulatory
antibodies for different periods of time, it has
been possible to determine a precise relationship between signal strength and T-cell fate.
Hierarchical thresholds of stimulation have
been defined for cell proliferation, differentiation and death7,13–15 (FIG. 1a). At low signal
strength, naive T cells proliferate but do not
acquire effector function, and they retain
lymph-node-homing capacity. By contrast, at
high signal strength and in the presence of
cytokines that polarize differentiation to
T helper 1 (TH1) or TH2 effector cells, CD4+
T cells lose lymph-node-homing capacity, but
acquire effector function and the capacity to
migrate to inflamed peripheral tissues. Finally,
at even higher levels of stimulation, T cells are
deleted through AICD.
The role of cytokines. Cytokines are the prototypical external cues that determine the quality of T-cell responses. Interleukin-12 (IL-12)
and IL-4, which are produced by APCs and
www.nature.com/reviews/immunol
© 2002 Nature Publishing Group