You are on page 1of 3

news & views

DENDRITIC CELLS

Decrypting DC development
The NFIL3–ZEB2–ID2 transcription-factor regulatory circuit switches the E protein–dependent +41 kb Irf8 enhancer
in DC progenitors to the BATF3-dependent +32 kb Irf8 enhancer in mature cDC1s. Deletion of the cryptic +41 kb
Irf8 enhancer impedes cDC1 development.

Martin Guilliams and Tomohiko Tamura

T
he decryption of the German Enigma Irf8 is regulated during DC development absolutely required for cDC1 development1.
cipher by the Allies during World War has remained obscure. Irf8 expression is Although the number of Irf8 +32–/– pre-
II was one of the most decisive code- very low in HSCs but increases at the MDP cDC1s remains normal, IRF8 expression
breaking successes in history. In this issue of stage, is further upregulated in CDPs and is reduced in these cells. Thus, the +32 kb
Nature Immunology, two papers by Murphy is maintained in cDC1s and plasmacytoid Irf8 enhancer directs the transition from
and colleagues (Durai et al.1, and Bagadia, DCs (pDCs), but becomes modest in pre-cDC1 to cDC1 by supporting IRF8
Huang, Liu et al.2) report that an NFIL3– monocytes6–8 (Fig. 1a). Interestingly, Irf8 expression.
ZEB2–ID2 transcription-factor regulatory expression is at first independent of BATF3 Durai et al. next seek to understand how
circuit controls the switch in enhancers of in DC progenitors and then becomes pre-cDC1s are specified from CDPs1. The
the gene encoding the transcription factor dependent on BATF3 in mature cDC1s. The authors unexpectedly find that the +41 Irf8
IRF8 to ensure the development of classical mechanisms that control this differential enhancer, which they previously showed
type 1 dendritic cells (cDC1s). These two dependency of Irf8 expression is unknown. is active in pDCs but not in cDC1s8, is at
studies by the Murphy laboratory will no Irf8 has been shown to carry three distinct an open chromatin configuration not only
doubt be recognized as a decisive step in the enhancers at 50 kb upstream (–50 kb) and in pDCs but also in CDPs and pre-cDC1s.
decryption of the central mechanisms that 32 kb and 41 kb downstream (+32 kb and The authors confirm that deletion of the
control DC development. +41 kb, respectively) of the transcription +41 Irf8 enhancer reduces the expression of
The development of various immune start site (Fig. 1a). However, the functional IRF8 in pDCs but, surprisingly, also reveal
cells relies on the establishment of requirement for the endogenous Irf8 that these ‘Irf8 +41–/–’ mice lack pre-cDC1s
appropriate gene-expression patterns, enhancers during in vivo DC development and cDC1s. In addition, the rise in IRF8
which is controlled mainly by regulatory cis has not been assessed. expression after the transition from MDP
elements called ‘enhancers’ that are bound Durai et al. clarify the lineage- and to CDP is abolished in these mice. As for
by lineage-specific transcription factors. A differentiation stage–specific roles of the possible transcription factors that act
single gene often has multiple enhancers the three Irf8 enhancers in vivo by using on this enhancer carrying six E-box motifs,
that act either in a cellular context– CRISPR–Cas9 genome editing to generate the authors show that the E protein E2A is
dependent manner to allow expression mouse strains that lack each enhancer1. The expressed in MDPs and CDPs and that its
at appropriate times or in a redundant –50 kb Irf8 enhancer, which contains two expression is reduced in pre-cDC1s and
manner to ensure robustness of expression3. elements that bind the transcription factor is absent in cDC1s and pDCs. Strikingly,
cDC1s serve pivotal roles, particularly PU.1, had been proposed to induce Irf8 in bone marrow progenitor cells from mice
in the host defense against tumors and MDPs by a study of mice with transgenic deficient in Tcf3 (which encodes E2A) have
viruses, due to their unique ability to expression of a phage artificial chromosome diminished potential to give rise to cDC1s
‘cross-present’ exogenous antigens to CD8+ reporter9. Durai et al. find, however, that and also pDCs in vivo and in vitro. Thus, the
cytotoxic T cells4. The mechanism of cDC1 deletion of this enhancer in vivo does not +41 kb Irf8 enhancer, probably activated by
differentiation is thus an important theme affect IRF8 expression in MDPs and the E2A, directs the transition from CDP to pre-
for understanding the nature of immunity subsequent DC lineage and instead reduces cDC1 by boosting IRF8 expression before
and for establishing effective therapies IRF8 expression in Ly6C+ monocytes and the +32 kb Irf8 enhancer becomes active. In
against cancers and infections. F4/80+ peritoneal macrophages but not in pDCs, the +41 kb Irf8 enhancer may bind
cDC1s are derived from bone marrow red-pulp macrophages1. Thus, the –50 kb another E protein, E2-2, that is essential for
hematopoietic stem cells (HSCs) via Irf8 enhancer functions in monocytes and pDC development.
intermediate progenitor cells such as certain types of macrophages. In the accompanying paper by Bagadia,
monocyte–DC progenitors (MDPs), The Murphy laboratory previously Huang, Liu et al., the authors try to delineate
common DC progenitors (CDPs) and pre- suggested that the +32 kb Irf8 enhancer, the regulatory circuits controlling that
cDC1s5. The transcription factors IRF8, which contains four transcription factor essential switch from the +41 kb Irf8
BATF3, NFIL3 and ID2 have been shown to AP1–IRF composite elements that bind enhancer in DC progenitors to the +32 kb
be required for the development of cDC1s4,6. BATF3 and IRF8, maintains high expression Irf8 enhancer in mature cDC1s2. NFIL3,
Among these, IRF8 appears to be the of Irf8 via auto-activation8. By performing ID2 and ZEB2 are transcription factors
master lineage-determining transcription ATAC-seq chromatin profiling and assessing essential for DC development10–12. The
factor, and Irf8–/– mice lack cDC1s and their mice that lack all four such elements (called authors use Zbtb46gfp, Id2gfp and Zeb2egfp
precursors, including pre-cDC1s. While ‘Irf8 +32–/–’ mice), Durai et al. prove that reporter mice (with expression of each
the function of IRF8 has been well studied, this +32 kb Irf8 enhancer is active only gene product reported by green fluorescent
the mechanism by which the expression of in pre-cDC1s and cDC1s and is indeed protein (GFP) or enhanced GFP (EGFP))
Nature Immunology | www.nature.com/natureimmunology
news & views

a genes (such as Zeb2 and Sirpa) in the


Irf8 expression HSC transition from ZEB2-EGFPhi, ID2-GFPlo or
ZBTB46-GFPlo CDP to ZEB2-EGFPlo, ID2-
pDC GFPhi or ZBTB46-GFPhi cDC1-committed
Low High
CDP to previously defined pre-cDC18.
Subsequently, the authors demonstrate that
this ZEB2-EGFPlo, ID2-GFPhi, ZBTB46-
MDP
GFPhi cDC1-committed CDP subset is
? severely diminished in the absence of Id2
or Nfil3 but is greatly increased in absence
cDC1
of Zeb2.
MФ Mo
Pre-cDC1 CDP But how does one define the precise
regulatory relationship among NFIL3,
ID2, and ZEB2 in the circuit controlling
cDC1 development? To crack this ‘cipher’,
Murphy and colleagues apply an ingenious
genetic epistasis strategy they achieve by
–50 Irf8 +32 +41 crossing Nfil3-, Id2- or Zeb2-deficient mice
to one another to determine the regulatory
PU.1 BATF3 E2A
IRF8 E2-2
hierarchy of these transcription factors
(Fig. 1b). The result is a regulatory circuit
in which Id2 is repressed by ZEB2, and
Zeb2 in turn is repressed by NFIL3. The
b Which genetic regulatory circuit controls Irf8 expression during cDC1 development? expression of NFIL3 therefore enforces Id2
Conventional DC expression by suppressing Zeb2. Finally,
+32 +41
development NFIL3 CDP the authors demonstrate that ID2
cipher E2A
ZEB2
E2-2
terminates the E2A activity at the +41 kb
ID2 Irf8 enhancer, which indicates that NFIL3 is
involved in the switch of the E2A-controlled
A
2

E2A
ID2
E2
ZEB

Zeb2 hiId2 lo E2A/E2-2 NFIL3


NF
3

+41 kb Irf8 enhancer to the BATF3-


BA
NFIL

IL3
BATF

ZEB

TF3

controlled +32 kb Irf8 enhancer during


ID2 ZEB2
2

cDC1 development (Fig. 1b).


Collectively, these two manuscripts
NF
ID2 IL3
E

Pre-cDC1
2A
2

elegantly illustrate the non-redundant roles


A

BA

B
E2

ZE use of
ID E2A/E2-2 NFIL3
TF

of Irf8 enhancers during cDC1 development


2
genetic
3

BATF3
epistasis* and reveal the regulatory circuits that
to decrypt ID2 ZEB2
the cDC cipher Zeb2 loId2 hi control the switch from one enhancer to
(*use double KOs) the other during this developmental
E2A/E2-2 NFIL3
pathway. At the same time, these findings

✓ cDC1
ID2 ZEB2
raise several subjects worthy of future
investigation. First, the finding that F4/80+
E2A/E2-2 NFIL3 E2A/E2-2 ZEB2 peritoneal macrophages require the –50 kb
Irf8 enhancer for IRF8 expression, but
red-pulp macrophages do not, raises the
ID2 ZEB2 ID2 NFIL3
+32 +41 question of whether the enhancer usage
Zeb2 loId2 hi BATF3 depends on the cellular origin (i.e., mainly
IRF8 bone marrow HSC origin for peritoneal
macrophages versus mainly embryonic
Fig. 1 | Decrypting DC development through epistasis. a, Overview of the distinct Irf8 enhancers and origin for red-pulp macrophages13) or
their use in myeloid precursors, monocytes (Mo), macrophages (MΦ) and cDC1s. Irf8 expression (key): on the differential tissue imprinting of
HSC <<< MΦ, Mo < MDP < CDP, pre-cDC1, cDC1, pDC. b, Ancient cipher (top left) consisting of four macrophages. Second, identification of
circles from which multiple transcription factors can be chosen: turning each circle results in many the enhancers and transcription factors
combinations of the transcription factors. BATF3, NFIL3, ZEB2, E2A and ID2 were known to be involved that direct Irf8 expression in progenitor
in cDC1 development, but it was not known in which order of hierarchy these transcription factors would cells more upstream than CDPs will be
work. Through their epistasis strategy, Murphy and colleagues have solved the puzzle: NFIL3 inhibits important. This is because IRF8 starts to be
Zeb2, ZEB2 inhibits Id2, and ID2 inhibits the activity of E2A (encoded by Tcf3). This explains the switch expressed at a low level near the HSC stage,
from the E2A-dependent +41 kb Irf8 enhancer to the BATF3-dependent +32 kb Irf8 enhancer. such as in a fraction of lymphoid-primed
multipotent progenitors14,15, and thereby
epigenetically induces early specification
and identify ZBTB46-GFPhi cells within the The authors next perform transcriptomics toward cDC1s15. Moreover, IRF8 has high
CDP population that are already committed analysis and find a stepwise increase in expression in MDPs and common monocyte
to the cDC1 lineage. They also identify the expression of cDC1-associated genes progenitors, establishes their enhancer
very similar cells as ZEB2-EGFPlo or ID2- (including Id2 and Batf3) and a drop in the landscapes and is also required for MDP-to-
GFPhi cells within the CDP population. expression of cDC2- and pDC-associated CDP and common monocyte progenitor–

Nature Immunology | www.nature.com/natureimmunology


news & views

to–Ly6C+ monocyte transitions6,7,16. Third, also been shown to be central players in ILC References
1. Durai, V. et al. Nat. Immunol. https://doi.org/10.1038/s41590-019-
it is still unknown whether the suppressive development. The same regulatory circuit
0450-x (2019).
effects of NFIL3 on Zeb2 and of ZEB2 on could therefore control both DC fate and 2. Bagadia, P. et al. Nat. Immunol. https://doi.org/10.1038/s41590-
Id2 are direct or indirect. Finally, the authors ILC fate without the need for a common 019-0449-3 (2019).
discuss the possibility that the cDC1–pDC cDC1–pDC progenitor. ❐ 3. Schoenfelder, S. & Fraser, P. Nat. Rev. Genet. 20,
437–455 (2019).
bifurcation occurs within the CDP or MDP 4. Anderson, D. A. I. III, Murphy, K. M. & Briseño, C. G. Cold
population via the mutual repression of Id2 Martin Guilliams   1,2 and Spring Harb. Perspect. Biol. 10, a028613 (2018).
(which promotes cDC1 development) and Tomohiko Tamura   3,4 5. Guilliams, M. et al. Nat. Rev. Immunol. 14, 571–578 (2014).
6. Kurotaki, D. & Tamura, T. J. Interferon Cytokine Res. 36,
Zeb2 (which promotes pDC development), 1
Laboratory of Myeloid Cell Ontogeny and 433–441 (2016).
but this appears to not fit with a recent Functional Specialization, VIB Center for 7. Sichien, D. et al. Immunity 45, 626–640 (2016).
model17, in which it was proposed that pDCs Inflammation Research, Ghent, Belgium. 8. Grajales-Reyes, G. E. et al. Nat. Immunol. 16, 708–717 (2015).
9. Schönheit, J. et al. Cell Reports 3, 1617–1628 (2013).
develop from lymphoid progenitor cells 2
Department of Biomedical Molecular Biology, 10. Scott, C. L. et al. J. Exp. Med. 213, 897–911 (2016).
independently of the myeloid cDC lineage Ghent University, Ghent, Belgium. 3Department of 11. Wu, X. et al. Proc. Natl Acad. Sci. USA 113,
and that pDCs could be a subset of innate Immunology, Yokohama City University Graduate 14775–14780 (2016).
12. Kashiwada, M., Pham, N. L., Pewe, L. L., Harty, J. T. & Rothman,
lymphoid cells (ILCs)17. Such a stricter and School of Medicine, Yokohama, Japan. 4Advanced
P. B. Blood 117, 6193–6197 (2011).
much earlier split between the lymphoid Medical Research Center, Yokohama City University, 13. Ginhoux, F. & Guilliams, M. Immunity 44, 439–449 (2016).
pDC lineage and myeloid cDC1 lineage, Yokohama, Japan. 14. Lee, J. et al. Nat. Immunol. 18, 877–888 (2017).
15. Kurotaki, D. et al. Blood 133, 1803–1813 (2019).
however, does not in any way invalidate the e-mail: martin.guilliams@irc.vib-ugent.be;
16. Kurotaki, D. et al. Cell Rep. 22, 2628–2641 (2018).
findings about the dependence of cDC1s tamurat@yokohama-cu.ac.jp 17. Dress, R. J. et al. Nat. Immunol. 20, 852–864 (2019).
and pDCs on specific Irf8 enhancers or the
NFIL3–ZEB2–ID2–E2A circuit featured Published: xx xx xxxx Competing interests
here. In fact, NFIL3, ZEB2 and ID2 have https://doi.org/10.1038/s41590-019-0457-3 The authors declare no competing interests.

Nature Immunology | www.nature.com/natureimmunology

You might also like