Distinct super-enhancer elements differentially control
Il2ra
gene expression in a cell-type specific fashion
Rosanne Spolski
1*
, Peng Li
1*
, Vivek Chandra
2
, Boyoung Shin
3
,
Chengyu
Liu
1
, Jangsuk Oh
1
, Min
Ren
1
, Yutaka Enomoto
1
, Erin E. West
1
, Stephen Christensen
1
, Edwin C.K. Wan
1
, Meili Ge
1
,
Jian-Xin Lin
1
, Pandurangan Vijayanand
2
, Ellen V. Rothenberg
3
, Warren J. Leonard
1
1
Laboratory of Molecular Immunology
Immunology Center
National Heart, Lung and Blood Institute
National Institutes of Health
Bethesda, MD 20892-1674
2
La Jolla Institute for Immunology
La Jolla, CA
3
Division of Biology and Biological Engineering
California Institute of Technology
Pasadena, CA
*These authors contributed equally.
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
2
Summary
The IL-2 receptor
α
-chain (IL-2R
α
/CD25) is constitutively expressed on DN2/DN3 thymocytes
and Treg cells but induced by IL-2 on mature T and NK cells.
Il2ra
expression is regulated by a
super-enhancer extensively bound by STAT5 in mature T cells. Here, we demonstrate that
STAT5 cooperates with Notch to induce/maintain
Il2ra/
CD25
expression in DN2/DN3 cells.
Moreover, we systematically investigated CD25 regulation using a series of mice with deletions
spanning STAT5 binding elements. Deleting the upstream super-enhancer region mainly affected
constitutive CD25 expression on DN2/DN3 thymocytes and Tregs, whereas deleting an intronic
region primarily decreased IL-2-induced CD25 on peripheral T and NK cells. Thus, distinct
elements preferentially control constitutive versus inducible expression in a cell-type-specific
manner, with the MED1 coactivator co-localizing with specific STAT5 binding sites. Moreover,
the intronic region was a dominant element whose deletion altered the structure throughout the
super-enhancer in mature T cells. These results demonstrate differential functions for distinct
super-enhancer elements, thereby indicating ways to manipulate CD25 expression in a cell-type
specific fashion.
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
3
Introduction
Super-enhancers are extended regions of chromatin that bind transcription factors and
coactivators at high density and orchestrate the formation of nuclear condensates, driving
transcription of key genes involved in lineage establishment or maintenance (Hnisz et al., 2013;
Whyte et al., 2013). Super-enhancers are variably considered to represent distinctive structures
or to be comprised of multiple individual typical enhancers (Hay et al., 2016; Hnisz
et al.
, 2013;
Pott and Lieb, 2015), and they are dynamic structures that can be remodeled during
differentiation or in response to external signals (Adam et al., 2015), with context-dependent
roles. Coactivators such as MED1 and BRD4 can form phase-separated condensates at super-
enhancers, thereby concentrating the transcription apparatus at key cell-identity genes and
influencing their expression. Moreover, studies of super-enhancers have provided insights into
the mechanisms underlying gene control in normal and pathologic states (Chong et al., 2018;
Sabari et al., 2018). Super-enhancer activity can be modulated by 3-dimensional chromatin
interactions organized into compartments denoted as topologically-associating domains (TADs)
(Rowley and Corces, 2018).
The human and mouse genes encoding IL-2R
α
(CD25) have been extensively studied
over the years to explain the responsiveness of this gene to both activation via the T cell receptor
and IL-2. Our lab and others identified a series of upstream positive regulatory regions (PRRs),
including PRRI, PRRII, PRRIII, where PRRI and PRRII were required for the mitogenic
induction of the
IL2RA
gene and PRRIII was an IL-2 response element, as well as a CD28
response element (Cross et al., 1987; Cross et al., 1989; John et al., 1995; John et al., 1996; Kim
et al., 2001; Kim and Leonard, 2002; Rameil et al., 2000; Soldaini et al., 1995; Yeh et al., 2001;
Yeh et al., 2002). PRRIII bound STAT5A and STAT5B proteins as well as ELF-1, HMG-I(Y),
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
4
and a GATA-1-like protein.
Our lab then characterized another IL-2-regulated element, PRRIV,
that could bind STAT5 proteins and HMG-I(Y) (Kim et al., 2006; Liao et al., 2013). We
subsequently showed that the
Il2ra
super-enhancer is the top-ranked IL-2/STAT5-dependent
super-enhancer in mouse pre-activated T cells, comprising approximately 13 STAT5 binding
sites and exhibiting potent activity upon IL-2 stimulation (Li et al., 2017). The corresponding
super-enhancer in the human
IL2RA
locus has a similar overall organization (Li
et al.
, 2017). We
previously deleted three individual STAT5 binding sites (one upstream and two intronic) in mice
using CRISPR/Cas9 technology and showed that each deletion partially contributed to IL-2-
induced
Il2ra
expression in mature CD8
+
T cells. Moreover, the mutation in mice of an
autoimmunity risk variant in an intronic enhancer delayed but did not abrogate
Il2ra
gene
expression upon TCR stimulation in mature T cells (Simeonov et al., 2017). However, it is not
clear how individual super-enhancer elements throughout the gene contribute to cell-type
specific
Il2ra
expression during development, in different lineages, and in response to stimuli.
To further understand
Il2ra
regulation, here we have generated a range of individual or
combinatory deletions of STAT5-bound regulatory elements in the
Il2ra
super-enhancer
in vivo
and analyzed them in multiple cell types, including in response to cytokine stimulation. CD25
expression varies at different stages of mouse lymphoid development. It is constitutively
expressed in subsets of double negative (DN) thymocytes (DN2/DN3) and in regulatory T cells
(Treg cells); in contrast, CD25 is not expressed in resting T or NK cells but is induced in
response to appropriate antigen or cytokine signals (Leonard et al., 2019; Liao
et al.
, 2013). We
show that STAT5 is important for normal CD25 expression at specific DN stages of T cell
development as well as in mature CD8
+
T cells. We reasoned that focusing on the STAT5
binding sites would allow us to identify important enhancer elements that regulated expression
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
5
during development or in response to T cell receptor or cytokine signals. Strikingly, we found
that distinct super-enhancer regions preferentially control constitutive versus inducible CD25
expression at different stages of T cell development and in other cell types as well. Furthermore,
we show extensive looping within this super-enhancer, with intronic regions dominantly
affecting the overall chromatin structure and activity of the gene, and moreover, we show the co-
localization of the MED1 coactivator and of transcription factors NFATc1, FOXP3, TCF1, and
SMAD4 at the super-enhancer. This detailed analysis provides mechanistic insights into CD25
regulation not only by IL-2 but also by TCR and TGF
β
signals throughout early T cell
development as well as in specific functional T cell subsets.
Results
Lineage- and developmental-specific STAT5 binding and open chromatin structure at the
Il2ra
locus.
We initially examined the STAT5 binding profile at the
Il2ra
locus in multiple immune
cell populations using ChIP-Seq
(Figure 1A;
the top enriched STAT5-binding GAS (gamma-
activated sequence) motifs in each cell type are shown in
Table S1
, and the GAS motif at each
upstream and intronic site is shown in
Table S2)
. TCR-activated CD8
+
T cells and inducible
Treg (iTreg) cells had similar STAT5 binding profiles in response to IL-2 stimulation, with most
of the upstream and intronic sites occupied, whereas little STAT5 binding was observed in these
cells in the absence of IL-2 stimulation (
Figure 1A
). Distinctive STAT5 binding patterns were
found in natural Treg cells isolated ex vivo, as well as in natural killer (NK) cells stimulated with
IL-15, splenic dendritic cells (DCs) stimulated with GM-CSF, and in mast cells stimulated with
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
6
anti-IgE (
Figure 1A
), suggesting that distinct
Il2ra
super-enhancer elements might be
differentially utilized in a cell-type-specific fashion.
During early mouse T cell development, CD25 is a key marker of DN thymocytes that is
expressed in the DN2 (cKit
+
CD44
+
CD25
+
) and DN3 (cKit
−
CD44
−
CD25
+
) stages bridging T-cell
lineage commitment (Rothenberg et al., 2008). These cells upregulate IL-7R
α
as they enter the
DN2 stage, and the natural thymic microenvironmental signal, IL-7, stimulates these cells,
activating STAT5A and/or STAT5B. Although
Stat5a
and
Stat5b
deletion severely compromises
early T cell development (Yao et al., 2006) and in the individual absence of either STAT5A or
STAT5B, there is a partial defect in T cell numbers and/or IL-2-induced CD25 expression
(Imada et al., 1998; Nakajima et al., 1997), whether CD25 expression in DN thymocytes is
dependent on STAT5 activation and binding to the
Il2ra
gene has remained unclear. To
investigate this, we first identified the stage at which STAT5 is activated by IL-7 by using
B6.Bcl11b-mCitrine reporter mice (
Figure S1A
) and found that STAT5 phosphorylation
approximately corresponded to when CD25 is first expressed in DN thymocytes—namely in
DN2a/DN2b cells and then extending into DN3 cells but declining in DN4 cells (
Figures S1B-
S1D
). Next, we used sgRNAs to acutely delete both
Stat5a
and
Stat5b
in normal Cas9;
Bcl2
-
transgenic hematopoietic precursors, starting one stage before CD25 is normally expressed, and
we examined the early stages of their T-lineage development in artificial thymic organoid (ATO)
(
Figures S1E-S1G
) and OP9-DLL1 coculture (
Figures S1H-S1J
) in vitro systems. The presence
of a
Bcl2
transgene in each system prevented the diminished survival that otherwise would have
resulted from defective IL-7-STAT5 mediated induction of
Bcl2
in these DN cells. In the ATO
system, deletion of either
Stat5a
or
Stat5b
partially lowered pSTAT5 in DN2 cells, and DN2
cells generated from
Stat5a
/
Stat5b
-deleted precursors showed greatly diminished IL-7-induced
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
7
pSTAT5 (
Figures S1F and S1G
). To measure the impact of STAT5 loss on CD25 expression,
we generated DN2 cells using the OP9-DLL1 co-culture T-cell differentiation system (Romero-
Wolf et al., 2020; Shin et al., 2021) (
Figure S1H
) and found that
Stat5a
/
Stat5b
-deletion reduced
CD25 expression intensity across the stages (
Figure S1I
). Importantly, this effect on CD25
expression did not reflect a developmental block given that
Stat5a/Stat5b
-deleted cells still
underwent normally timed commitment to the T-cell lineage, as shown by upregulation of a
Bcl11b
-mCherry reporter allele. By day 7, the control cells had progressed to the DN2 and early
DN3 stages, with strong surface expression of CD25; however, when both
Stat5a
and
Stat5b
were deleted, both surface CD25 (
Figures S1I and S1J)
and
Il2ra
RNA
(Figures S1K and S1L)
expression were significantly reduced in both pre-commitment (BCL11b
-
) and post-commitment
(BCL11b
+
) DN2 stages. A global transcriptomic analysis showed that
Stat5a/Stat5b-
deficient
DN2/3 cells had lower expression of growth and viability control genes (e.g.,
Bcl2, Bcl2l1,
Socs2, Cdkn1a, Eno1, Xbp1,
and
Bhlhe40
), as well as of
Il2ra
(
Table S3
), but the cells were not
altered in their developmental progression (
Table S3)
as defined by a curated panel of markers,
including
Tcf7, Bcl11b, Lck, Thy1, Cd3
clusters
,
Ets1
,
Tcf12,
and
Rag
genes
(Zhou et al., 2019).
This effect of STAT5 was in addition to the known requirement for Notch signaling to induce
and sustain CD25 expression in DN thymocytes (
Figure S1M;
this panel is from (Romero-Wolf
et al.
, 2020)). Thus, STAT5 activation cooperated with Notch signaling to induce and maintain
normal levels of
Il2ra/
CD25
expression in DN2-DN3 pro-T cells.
In vitro generated DN2 (CD25
+
cKit
+
) thymocytes responded to IL-7 with a STAT5
binding pattern similar to that observed in IL-2-stimulated CD8
+
T cells (
Figure 1A
), but
interestingly, steady-state thymic DN3 cells isolated
ex vivo
had their greatest STAT5 binding at
the upstream region of the super-enhancer, primarily at the UP3 element (
Figure 1A
), a pattern
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
8
distinct from that observed in IL-2-induced CD8
+
T cells or iTregs. Other factors including
Notch/RBPJ
(Chen et al., 2019; Del Real and Rothenberg, 2013; Liu et al., 2010a; Romero-Wolf
et al.
, 2020), RUNX1 (Guo et al., 2008; Shin
et al.
, 2021), BCL11b (Liu et al., 2010b), and TCF-
1 (Weber et al., 2011) are known to contribute to lineage progression to the DN2b-DN3 stages,
and these factors bound in proximity to the STAT5 binding sites in DN2b-DN3 cells
(Figure
S2
), suggesting that they might cooperate to control the developmental regulation of
Il2ra
expression in DN cells.
We next analyzed chromatin accessibility at the
Il2ra
locus throughout lymphoid
development using ATAC-Seq (Assay for Transposase-Accessible Chromatin with high-
throughput sequencing) datasets from the ImmGen database (Yoshida et al., 2019). Interestingly,
an open chromatin region at upstream element UP3 (
Figure 1B
; left red box) was detected in
DN1 thymocytes, modestly increased in DN2a, DN2b, and DN3 thymocytes, but then was
essentially absent in DN4 thymocytes; thus, the loss of chromatin accessibility at UP3
approximately correlated with the loss of CD25 expression. Accessibility at the intronic IN1m
STAT5 binding site was relatively low in thymocyte populations, was greater in naïve CD4 and
CD8 cells, Treg cells, NK cells, and ILC2s, but then was weak in splenic conventional DC and
plasmacytoid DC (pDC) populations
(Figure 1B,
right red box
)
, suggesting differential use of
upstream and intronic elements in different cellular populations.
Upstream super-enhancer elements are required for CD25 expression in DN thymocytes
To investigate the roles of the upstream and intronic enhancer elements in lymphoid
development and function, we next deleted a range of individual elements or combinations of
elements within the
Il2ra
super-enhancer (
Figure 2A
)
in vivo
using CRISPR-Cas9 methodology.
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
9
No individual upstream element deletion nor even deletion of the entire UP1-6 upstream region
(
Δ
UP1-6) or intron region (
Δ
Intron) or both together (
Δ
UP1-6/
Δ
Intron) significantly affected
either thymic or splenic cellularity
(Figures S3A and S3B),
nor did they affect the percentage of
DN thymocytes
(Figure S3C)
. However, mice lacking the UP1-6 region had a profound
decrease in the percentage of CD25
+
CD44
+
DN2 and CD25
+
CD44
-
DN3 thymocytes, whereas
loss of the intron region had little effect
(Figure 2B)
. RNA-Seq analysis of DN thymocytes from
WT,
Δ
UP1-6, or
Δ
UP1-6/
Δ
Intron mutant mice showed that
Il2ra
mRNA was essentially absent
in the mutant mice, but
Bcl11b,
Tcf7,
and
Il7ra
mRNAs, which are normally expressed in
DN2/DN3 cells (Hosokawa and Rothenberg, 2018), were still expressed
(Figure S3D)
,
indicating that pro-T cells were indeed still present but lacked CD25 expression due to deletion
of essential
Il2ra
regulatory elements. Separate deletions of UP1, UP2, UP3, and UP5-6 did not
affect CD25 expression in DN thymocytes
(Figures 2C and 2D).
Interestingly, however,
deletion of the
Δ
UP2-3 region, which extends downstream of UP3 to include a putative RBPJ
binding site (TGACTAATG) (Romero-Wolf
et al.
, 2020), essentially eliminated CD25
expression on DN cells. We therefore generated mice lacking UP3 extending through the RBPJ
site (
Δ
UP3-RBPJ) and found that this deletion also abrogated CD25 expression on DN
thymocytes, phenocopying the
Δ
UP1-6 mice (
Figures 2C and 2D
). Thus, the UP3-RBPJ region
was indispensable for
Il2ra
/CD25 expression in DN thymocytes. To further analyze the role of
the upstream elements, we cloned individual upstream enhancer elements in a reporter construct
and analyzed reporter activity in a DN3-like cell line (SCID-ADH-2C2)
(Anderson et al., 2002;
Dionne et al., 2005). The UP3-RBPJ region was sufficient to drive reporter activity, but reporter
activity was not augmented by stimulating the cells with IL-2, IL-7, or the combination of IL-2
with PMA + ionomycin (PI), which can mimic TCR stimulation
(Figure 2E)
. Strikingly
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
10
however, activity was significantly reduced when the STAT5 binding site (GAS motif), the
RBPJ binding site, or both sites were mutated
(Figure 2F)
, or when a Notch inhibitor (
γ
-
secretase inhibitor, GSI) was used
(Figure 2G)
. Even though
in vivo
deletion of the UP5-UP6
region had little if any effect, the UP5 element also exhibited modest reporter activity in these
cells, with enhanced activity induced by IL-2, IL-7, PI, and PI/IL-2
(Figure 2E)
, but inhibiting
Notch had little effect on its activity, indicating the specificity of the Notch inhibitor for the
UP3-RBPJ region
(Figure 2G)
. These results indicate a key role for the UP3-RBPJ element in
regulating CD25 expression in DN2 and DN3 cells, with its activity requiring both STAT5 and
Notch. In addition to the co-localization of RBPJ with STAT5 at the UP3 site, RUNX1 and
BCL11b, which are both involved in progression through DN thymocyte development, also
bound at the UP3 region
(Figure S2)
, suggesting that multiple signals may be involved in the
regulation of CD25 expression in DN2/DN3 thymocytes.
Differential control of CD25 expression on thymic and peripheral Treg cells
Because CD25 is also constitutively expressed on Treg cells, we next investigated
whether
Il2ra
gene regulation in these cells was similar to what we observed in DN thymocytes.
Thymic and splenic Treg cell numbers were not significantly affected by any of the deletions
(Figures S4A and S4B)
but deleting UP1-6 or both UP1-6 and intronic regions markedly
reduced CD25 expression on both thymic and splenic Tregs
(Figures 3A and 3B)
. Strikingly,
whereas the
Δ
UP2-3 or
Δ
UP3-RBPJ deletions essentially abrogated CD25 expression on
DN2/DN3 thymocytes (
Figure 2C
), these deletions did not significantly affect CD25 expression
on thymic Tregs
(Figure 3A)
and only modestly lowered CD25 expression on splenic Tregs
(
Figure 3B
), indicating distinctive regulation of the
Il2ra
gene in DN thymocytes and Tregs.
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint
Spolski et a l.
11
Although the intron deletion did not significantly affect CD25 expression on thymic DN2/DN3
cells
(Figure 2B)
,
Δ
Intron thymic FoxP3
+
Tregs include a particularly strong CD25-negative
population
(Figure 3C
left and middle panels
and 3D)
, reminiscent of FoxP3
+
CD25
-
thymic
Treg precursor cells that normally express CD25 in response to IL-2, IL-7, or IL-15 (Cheng et
al., 2013; Owen et al., 2018). This suggests that cytokine-mediated induction of CD25 in
precursor Tregs may be particularly dependent on the intronic region, although a population of
CD25
low
thymocytes was also present in the
Δ
UP1-6 mice (
Figure 3C
). Consistent with this,
when the total thymic CD4
+
cells were examined for Treg precursor populations, it was apparent
that deletion of either the intron or UP1-6 region led to a decreased percentage of
CD25
+
FoxP3
neg
cells but increased percentage of CD25
neg
FoxP3
+
cells
(Figures S4C and S4D)
.
This CD25
neg
FoxP3
+
precursor population was increased in
Δ
Intron and
Δ
UP1-6/
Δ
Intron Tregs
(Figures 3D and S4B)
. In
ΔΙ
ntron splenic Tregs, such a bimodal distribution was not as readily
observed (
Figure 3C
, right panel), indicating differences in CD25 regulation in thymic and
splenic Tregs and suggesting that CD25
neg
FoxP3
+
Tregs in the thymus either do not exit to the
periphery or are eliminated.
Interestingly,
Δ
Intron mice expressed lower levels of FoxP3 in thymic Tregs, while
Δ
UP1-6 mice expressed lower levels of FoxP3 mainly in splenic Tregs
(Figure S4E)
. This
suggests that STAT5-activating cytokines such as IL-2 can regulate expression of FoxP3 in
Tregs, consistent with evidence that a FoxP3 enhancer element containing STAT5 binding sites
can mediate IL-2-regulated Treg function (Dikiy et al., 2021). In addition to the regulation of
FoxP3 transcription by IL-2-induced STAT5 activation, FoxP3 binding sites colocalized with the
STAT5 sites in the
Il2ra
intronic and proximal promoter regions (
Figure S5
), suggesting a
feedback regulation where IL-2 via STAT5 regulates expression of FoxP3, which in turn
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC
The copyright holder for this preprint
this version posted November 18, 2022.
;
https://doi.org/10.1101/2022.11.18.516445
doi:
bioRxiv preprint