A Shared Regulatory Element Controls the Initiation of Tcf7 Expression During Early T Cell and Innate Lymphoid Cell Developments

The transcription factor TCF-1 (encoded by Tcf7) plays critical roles in several lineages of hematopoietic cells. In this study, we examined the molecular basis for Tcf7 regulation in T cells, innate lymphoid cells, and migratory conventional dendritic cells that we find express Tcf7. We identified a 1 kb regulatory element crucial for the initiation of Tcf7 expression in T cells and innate lymphoid cells, but dispensable for Tcf7 expression in Tcf7-expressing dendritic cells. Within this region, we identified a Notch binding site important for the initiation of Tcf7 expression in T cells but not in innate lymphoid cells. Our work establishes that the same regulatory element is used by distinct transcriptional controllers to initiate Tcf7 expression in T cells and ILCs.


INTRODUCTION
Regulation of gene expression is central to development and cellular differentiation, and erroneous gene expression is linked to many diseases such as cancer. During development, the coordinated and combinatorial action of transcription factors and enhancers results in gene expression patterns that can be distinct between cell lineages and developmental stages. Understanding gene regulation through the study of enhancers is a strategy relevant to understand development and target diseases.
TCF-1 (encoded by the gene Tcf7) is an HMG-box transcription factor that plays important functions in development and homeostasis. In particular, TCF-1 is expressed and required at early steps of development of T cells and innate lymphoid cells (ILC) (1)(2)(3)(4)(5)(6)(7)(8)(9). The mechanism by which TCF-1 expression is regulated is not understood. Several regulators have been proposed to act upstream of Tcf7 gene expression, however, whether and how they directly regulate Tcf7 is unknown.
In this study, we aimed to better understand Tcf7 gene regulation in hematopoietic lineages. We analyzed publicly available ATAC-seq and ChIP-seq datasets, together with genomic deletions using CRISPR/Cas9 technology, to identify regulatory regions controlling Tcf7 expression in early stages of T cells and ILC development. We identified a 1kb regulatory element upstream of Tcf7 that controls the initiation of Tcf7 expression in T cells and ILCs, but it is dispensable for Tcf7 expression in conventional Dendritic Cells (cDCs). Within this region, we identified a Notch binding site that contributes to Tcf7 initiation in T cells but not in ILCs. Our results establish that the many transcriptional similarities between T cells and ILCs include control of Tcf7 through a shared regulatory element, and further establish that lymphocytes and cDCs differ in the regulatory elements they use to control expression of Tcf7.

Statistical Analysis
Statistical analysis was performed on groups with limited variance using Excel or Prism. Differences between groups of mice or wells were determined by a two-tailed unpaired Student's t-test. A p of p < 0.05 was considered statistically significant. Sample sizes were empirically determined, no samples or animals were excluded from the analysis, no randomization, or blinding was used.

Identification of Candidate Regulatory Regions Upstream of Tcf7
We wished to identify candidate regulatory regions that might be required for Tcf7 expression in one or several hematopoietic lineages. Assay for transposase accessible chromatin followed by next generation sequencing (ATAC-seq) reveals regions of open chromatin, and thus can be used to identify putative enhancers. Using publicly available ATAC-seq data, we investigated the presence of open chromatin regions in T cell precursors (ETP, DN3), mature T cells (CD4), and ILC precursors (ILCP), which all express Tcf7; and B cells, that lack Tcf7. We additionally examined common lymphoid progenitors (CLP), which lack Tcf7 expression, and can give rise to T cells, B cells, and ILCs. This analysis identified a 20 kb region upstream of the Tcf7 promoter that showed peaks of open chromatin shared by all Tcf7expressing populations and in a lesser extent CLP, but not B cells ( Figure 1A). On the other hand, the Tcf7 super-enhancer previously identified in T cells (25) and located downstream of the region 1-2, presented numerous ATAC-seq peaks that were not shared between Tcf7-expressing populations, and were not specifically present in Tcf7-expressing populations compared to B cells (Supplementary Figure 1). Importantly, most of the ATAC-seq peaks present in region 1-2 were located in highly conserved regions between mouse and human and contained many binding motifs for factors involved in hematopoiesis ( Figure 1A, and not shown). Interestingly, one of the conserved ATAC-seq peaks within the 20 kb region contained a single nucleotide polymorphism (rs244689), that is linked to systemic lupus erythematosus (SLE) in Asian populations ( Figure 1B) (26).
We further examined whether candidate controllers of Tcf7 bind in the vicinity of the Tcf7 locus. Although Tcf7 controllers are yet to be identified in ILC, many transcription factors have been proposed to act upstream of Tcf7 expression in developing T cells. Tcf7 expression is thought to initiate downstream of Notch1 signaling in T cell precursors (1,8,27), and this initiation requires RUNX factors (28). GATA-3 and TCF-1 itself might further contribute to optimal expression of Tcf7 (8,29). We used publicly available ChIP-seq data for these transcriptional controllers in T-lineage cells, and found that they all bind the 20 kb candidate regulatory region we identified (Figures 1A,B). On the other hand, we did not observe binding for these factors in the Tcf7 super-enhancer (Supplementary Figure 1). Importantly, ChIP-seq peaks for Notch1, TCF-1, RUNX, and GATA-3 ( Figure 1B) all colocalized with binding motifs for these factors that were conserved between mouse and human (not shown). We therefore hypothesized that the 20 kb 1-2 region contains regulatory elements important for Tcf7 expression in one or several hematopoietic lineages.

The Region 1-2 Is Required for Tcf7 Expression in the T Cell Lineage
Using CRISPR/Cas9, we generated mice lacking the 20 kb region 1-2, called the Tcf7 1−2 strain. To examine whether the deletion of the region 1-2 affected Tcf7 expression in thymocytes, we first examined T cell development in the Tcf7 1−2/ 1−2 mice compared to Tcf7 −/− mice that lack TCF-1 functional protein (7). Similarly to Tcf7 −/− mice, Tcf7 1−2/ 1−2 mice presented a reduced frequency of CD4 + CD8α + DP cells, and total thymocyte numbers were decreased more than 10-fold compared to WT mice (Figure 2A). Early T cell development was particularly disrupted compared to WT cells, as shown by the absence of recognizable Lin T− Kit hi CD25 − ETP and Lin T− Kit hi CD25 hi DN2 subsets and development of aberrant Lin T− Kit int CD25 int subsets (Figure 2A).
We further wished to examine Tcf7 initiation during T cell development. Dramatic defects in the generation of ETP and DN2 cells in Tcf7 1−2/ 1−2 mice prevented direct assessment of TCF-1 expression at these early developmental stages. We therefore examined TCF-1 initiation in Tcf7 1−2/ 1−2 T cell precursors in vitro, after culture on stromal layers expressing Notch ligand (OP9-DL1). In this system, WT BM progenitors upregulated TCF-1 by day 5 (Figure 2C) and generated Thy1 hi CD25 hi DN2/DN3 stages T-lineage cells by day 7 (Figure 2D). On the other hand, Tcf7 1−2/ 1−2 and Tcf7 −/− progenitors failed to upregulate TCF-1 and to generate DN2/DN3 T-lineage cells, even at later time points in the culture (Figures 2C,D). Altogether, these results indicate that the 20 kb region deleted in Tcf7 1−2/ 1−2 mice is required for initiation of Tcf7 expression at early stages of T cell development.

The Region 1-2 Is Dispensable for Tcf7 Expression in Migratory cDC
We additionally examined TCF-1 expression in other hematopoietic lineages. Using a Tcf7-reporter mouse that expresses the Yellow Fluorescent Protein (YFP) downstream of the endogenous Tcf7 gene (2), we found that Tcf7 expression was evident in cDCs present in lymph nodes (LN) but very low in spleen cDCs. Tcf7-expressing cDCs present in LN corresponded to migratory DCs as defined using CD11c and MHC-II expression (33), whereas such cells were almost absent from the spleen ( Figure 4A). To examine whether TCF-1 played a role in the development of migratory cDCs, we reconstituted lethally irradiated mice with Lin BM -depleted BM cells isolated from Tcf7 −/− CD45.2 + mice or littermate control, mixed with CD45.1 + Lin BM -depleted BM cells. The competitive setting of this experiment should reveal even mild defects in hematopoietic development. After 12 weeks of reconstitution, migratory and resident cDC developed from Tcf7 −/− hematopoietic precursors as efficiently as WT cells, similarly to B cells that do not express or require TCF-1 ( Figure 4B). In contrast, T cell reconstitution from Tcf7 −/− hematopoietic precursors was greatly deficient as expected ( Figure 4B). We next examined whether the region 1-2 was required for Tcf7 expression in LN cDCs. TCF-1 expression in LN cDCs as well as cDC numbers appeared unaffected by deletion of the region 1-2 (Figures 4C,D). We did not observe ectopic expression of TCF-1 in Tcf7 1−2/ 1−2 mice, in lineages normally lacking TCF-1, like B cells ( Figure 4C). The region 1-2 appears therefore dispensable for Tcf7 expression in migratory cDCs, and for Tcf7 lack of expression in B cells. Consistently, regions of open chromatin were absent from the region 1-2 in cDCs (Supplementary Figure 1).

A 1kb Region Is Required for Tcf7 Initiation in Early T Cell and ILC Precursors
To refine the region responsible for Tcf7 initiation in T cell and ILC lineages, we subdivided the 20 kb 1-2 region into a 8kb (region 1) and a 12 kb (region 2) regions ( Figure 1A). We assessed mice deficient for each of these regions (Tcf7 1/ 1 and Tcf7 2/ 2 mice, respectively) for defects in T cell and ILC development. No apparent defect in T cell development was seen in Tcf7 1/ 1 mice (Supplementary Figures 3A,C). Furthermore, although region 1 included a region previously proposed to play key functions for Tcf7 expression specifically in naïve T cells (19) (Supplementary Figure 1), no defect was seen for TCF-1 expression in Tcf7 1/ 1 mice (Supplementary Figure 3D). In EILP, Tcf7 1/ 1 mice did not show a significant defect in TCF-1 initiation (Supplementary Figures 4A,D). On the other hand Tcf7 2/ 2 mice showed major defects in T cell and ILC development, similar to Tcf7 1−2/ 1−2 mice (Supplementary Figures 3A,C, 4A,D). These data indicate that a key regulatory element controlling Tcf7 initiation during early T cell and ILC development falls within region 2, and that region 1 is largely dispensable for Tcf7 initiation in these two lineages.
Interestingly, although the Tcf7 1/ 1 and Tcf7 4/ 4 mice did not show significant defects in TCF-1 expression at EILP stage, defects in TCF-1 expression were evident at later stages of ILC development, in ILCP and ILC2P (Supplementary Figures 4E,F). This result indicated that regions 1 and 4 might regulate Tcf7 expression after initiation. On the other hand, in Tcf7 1−2/ 1−2 , Tcf7 2/ 2 , and Tcf7 3/ 3 mice that presented dramatic defects in TCF-1 initiation of expression at EILP stage, TCF-1 expression showed upregulation at later developmental stages and almost reached WT levels (Figures 3G,H, 6F,G and Supplementary Figures 4E,F). Consistently, despite the dramatic defects in EILP and ILCP numbers (Figures 3A-C, 4A-C, Supplementary Figures 4A,B), ILC2P still developed in Tcf7 1−2/ 1−2 , Tcf7 2/ 2 , and Tcf7 3/ 3 mice (Figures 3D, 4D, Supplementary Figure 4C). This upregulation could indicate that region 3 is dispensable for TCF-1 regulation of expression after initiation. Alternatively, TCF-1 upregulation could be the result of compensatory mechanisms that are secondary to the defect in Tcf7 initiation. Altogether, the data indicate that the 1 kb region 3 contains regulatory elements critical for the initiation of Tcf7 expression in the T cell and ILC lineages. On the other hand, regions 1 and 4 are largely dispensable for Tcf7 initiation of expression in T cells and ILCs, but may play important functions at later developmental stages in ILC.

Disruption of a Notch Binding Site Upstream of Tcf7 Impact TCF-1 Expression in T Cells
The 1 kb region 3 we identified as crucial for Tcf7 expression during T cell and ILC development was bound by several transcription factors previously proposed to play important functions in regulating Tcf7 expression in T cells, namely Notch, TCF-1, RUNX, and GATA-3 ( Figure 7A). This region further includes binding motifs for these factors that are conserved between mouse and human ( Figure 7B). In particular, a Notch/Rbpjk binding motif was previously proposed to control Tcf7 initiation in T cells (1,8) (Figure 7A). Using CRISPR/Cas9 technology, we generated a mouse selectively lacking this Notch binding site (NBS, Tcf7 NBS/NBS ) (Figure 7B). Thymocyte populations and cellularity appeared unaffected in these mutant mice (Supplementary Figure 5A). However, TCF-1 expression was decreased by about 2-fold at every stage of T cell development, to levels similar to Tcf7 +/− heterozygous mice (Figure 7C). We further examined T cell development from CD45.2 + Tcf7 NBS/NBS hematopoietic progenitors (Lin BM− Kit hi Sca1 + ; LSK) in the presence of WT CD45.1 + competitor cells in long-term hematopoietic chimeras. After 10 weeks of reconstitution, we quantified the percentage of CD45.2 + donor cells in thymocyte populations compared to LSK. The percentage of cells generated by Tcf7 NBS/NBS hematopoietic progenitors significantly decreased from LSK to ETP compared to WT cells, similarly to Tcf7 +/− hematopoietic progenitors ( Figure 7D). Interestingly, unlike the larger 1 kb region 3, the NBS deletion affected TCF-1 expression at all stages of T cell development evenly, without particularly impacting TCF-1 upregulation at β-selection (Figure 7C and Supplementary Figure 5B). This result suggested that Notch signaling is not responsible for this upregulation, and that other factors are controlling Tcf7 upregulation at this transition. Because pre-TCR signaling is a major driver of changes occurring at β-selection, pre-TCR signaling could be important in inducing TCF-1 upregulation at the TCRβ-selection checkpoint. Consistent with this hypothesis, DN3 thymocytes expressing  intracellular TCRβ did not upregulate TCF-1 in the absence of LAT, a signaling molecule required downstream of the pre-TCR (Supplementary Figure 5C). Finally, we examined whether the NBS played some function in Tcf7 expression in the ILC lineage. ILC precursors quantification showed no significant differences between Tcf7 NBS/NBS , Tcf7 +/− , and WT mice (Supplementary Figures 5D-F). We next examined TCF-1 expression in Tcf7 NBS/NBS ILC precursors compared to Tcf7 +/− and WT. Although TCF-1 expression was detectably lower in Tcf7 +/− EILP and ILC2P compared to WT, TCF-1 expression by Tcf7 NBS/NBS ILC precursors was similar to WT (Figures 7E-G).
Our data therefore identify a significant contribution for a single Notch binding site in Tcf7 initiation during T cell development, but not ILC development.
We further wished to examine the contribution of other candidate factors for the regulation of Tcf7 expression through the 1 kb region 3. Because of the lack of early T cell precursors in mice deficient for GATA-3 (34), and RUNX (28), we were unable to examine whether these factors contribute to Tcf7 expression in T cells. Additionally, RUNX is required for the development of ALP, that are upstream of ILC precursors; thus TCF-1 expression during early ILC development cannot be assessed in RUNX deficient mice (30). On the other hand, some EILP are still present in GATA-3 deficient mice (30), which enable us to directly assess TCF-1 expression in these cells. This analysis showed that TCF-1 expression was greatly decreased in GATA-3 deficient TOX + EILP (defined using TOX instead of TCF-1 as before) compared to littermate controls ( Figure 7H).
Our data therefore identify a significant contribution for a single Notch binding site in Tcf7 initiation during T cells development, but not ILC development. Furthermore, GATA-3 significantly contributes to levels of Tcf7 during initiation of expression in early ILC development.

DISCUSSION
In this study, we investigated which regulatory elements upstream of Tcf7 are required for Tcf7 expression in hematopoietic lineages. We identified a 1 kb region (region 3) located upstream of the Tcf7 promoter, which contains regulatory elements required for Tcf7 initiation in T cell and ILC lineages, but not in migratory cDC. We further identified a significant contribution for a conserved Notch binding site located in this region for initiating Tcf7 expression during early T cell development, but not ILC development.
The 1 kb regulatory element we identified to be required for Tcf7 initiation in T cells and ILCs was not part of the super-enhancer previously identified for Tcf7 (25). However, this region contained a high density of transcription factors binding sites, and could correspond to a region bound by a cluster of transcription factors that cooperate to regulate gene expression (35,36). In particular, this region was bound by many transcription factors required for normal T cell and/or ILC development, such as GATA-3, RUNX, TCF-1, and Notch, which are suggested to be involved in Tcf7 expression. Notch signaling was thought to be a major controller of Tcf7 initiation during T cell development (1,8,27). By mutating a single Notch binding site, we provide the first evidence that Notch directly contributes to Tcf7 initiation in T cells. However, contribution of this binding was unexpectedly minor, thus other elements are required to control Tcf7 initiation in T cells. Notch signaling might further control Tcf7 initiation in T cells through binding to additional sites, or by inducing expression of other transcription factors such as GATA-3 (29). Because multiple GATA-3 putative binding sites are present in the 1kb Tcf7 enhancer, examining the role for GATA-3 in directly regulating Tcf7 expression is challenging. Furthermore, the absence of early T cell precursors in GATA-3 deficient mice (34) prevents the analysis of Tcf7 expression in ETP in such mice. Interestingly, we found that TCF-1 failed to be properly expressed in GATA-3 deficient early ILC precursors, which are transcriptionally similar to T cell precursors (30). This result may suggest a role for GATA-3 in Tcf7 initiation or amplification of expression in T cells (29). Consistent with previous work indicating that Notch is not required for early ILC development (37), we did not find a contribution for the Notch binding site in TCF-1 expression in ILC precursors. Future work will aim to identify additional transcriptional controllers involved in Tcf7 initiation during early T and ILC development, and to understand how they collaborate in this process.
This work additionally identified two regulatory regions (region 1 and region 4) that did not significantly contribute to Tcf7 initiation in T cells and ILCs, but regulated Tcf7 expression at later stages of ILC development. This result indicated that distinct regulatory regions may cooperatively regulate Tcf7 expression at specific developmental stages. The dramatic defect in early T cell and ILC development seen in mice lacking the 1 kb region 3 precluded conclusions on the role of this region beyond Tcf7 initiation. Indeed, although Tcf7 was upregulated after the initial defect in expression in both T cells and ILCs, this upregulation could conceivably be the result of compensatory mechanisms that are secondary to the defect in Tcf7 initiation, and that might not be present in normal (unmutated) cells. Assessing the role of this region after Tcf7 initiation would thus require the generation of new mouse models allowing conditional deletion after ETP and EILP stages. Importantly, the region 1 we deleted included a previously identified enhancer candidate (19). Although this enhancer was predicted to control Tcf7 expression in naïve T cells (19), our study did not find support for such a function. Hence our data highlight the difficulty of enhancer prediction, and the necessity to experimentally test enhancer candidates.
By examining TCF-1 expression in various hematopoietic lineages, we found that migratory cDCs expressed high levels of TCF-1. We have not found a role for TCF-1 in the development of cDCs, but it could be involved in cDC function. Interestingly, the regulatory regions required for TCF-1 expression in T cells and ILCs were not important for expression in cDCs. This result indicates the existence of additional regulatory regions, outside the 20 kb region we deleted, that could be required for TCF-1 expression in DCs. Consistently, in cDCs, regions of open chromatin found in T cells and ILCs were absent from the 20 kb region we deleted, but present in the Tcf7 super-enhancer (25). TCF-1 has been involved in bone formation (38), thermogenesis by brown adipocytes (39), pancreatic β-cell survival and glucose tolerance (40), although the cells in which TCF-1 is expressed are not well defined. Finally, aberrant expression and function of TCF-1 is associated with various diseases, and many cancers (41)(42)(43)(44)(45). Understanding the mechanisms by which TCF1 is regulated in these various cell type could enable us to target TCF-1 expression in specific cell types. Such knowledge could be useful to target malignant cells while leaving essential functions intact.
The identification of a shared regulatory element controlling Tcf7 initiation specifically in T cells and ILC adds to the striking developmental and functional similarities noted between the two lineages (2,30,46). Our results additionally indicate that the same regulatory element is used by distinct transcriptional controllers to initiate Tcf7 expression in T cells and ILCs.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

ETHICS STATEMENT
The animal study was reviewed and approved by NIH Animal Care and Use Committees.