Introduction

The immune system comprises a collection of functionally distinct cell types which act in concert to protect the body from infection and malignancy. These diverse cell populations develop from common haematopoietic progenitors in a highly controlled fashion driven by a network of transcription factors that control lineage-specific gene expression programs. GATA-binding protein 3 (Gata3) is one of the master transcription factors that regulates the development of immune cells such as T cells, Natural Killer (NK) cells, and innate lymphoid cells (ILCs) (1, 2). Gata3 is also critical for the development of CD4⁺ T helper 2 (Th2) cells (3) which produce cytokines such as IL4, 5 and 13 that are critical for protection against extracellular pathogens. It has recently emerged that a population of ILCs known as group 2 ILCs (ILC2) with similar characteristics to Th2 cells reside in non-lymphoid tissues such as the small intestine, visceral adipose tissue and lung and have been shown to play a non-redundant role in immunity to helminths and drive allergic immune responses(4). ILC2s develop in the bone marrow via common ILC progenitor (CILP)(5), common helper ILC progenitor (CHILP)(6), innate lymphoid cell progenitor (ILCP) (7) and finally ILC2 progenitor (ILC2P)(8). Gata3 is critical to the development of ILCs (8, 9) and high levels are required for the development of ILC2P in the bone marrow (8, 10).

Dosage of Gata3 plays a critical role in T, NK and ILC development and peripheral immunity (8, 10, 11). To support this Gata3 expression is spatiotemporally tuned by several distal enhancers that dictate Gata3 activity in these individual lineages. Previous research identified a 7.1kb region ∼280kb from the Gata3 gene that is important for its expression in NK and T cell development, known as T cell enhancer 1 (Tce1) (12, 13). Subsequent work found that this region (also termed Gata3 +278/285) also impacts ILC2 development but not Th2 cell differentiation (14). ILC2-specific effects have been attributed to additional distal regulatory regions, such as Gata3 +674/762 (14) and the Gata3 tandem super enhancers (+500/764)(15). In addition, a Th2 specific enhancer (Gata3 +906/935) has recently been identified (16). Overall, this complex regulatory landscape mechanistically fit with a dosage-threshold model controlling Gata3 expression.

Recently we discovered a long non-coding RNA gene that we named Distal regulatory enhancer of Gata3 1 (Dreg1) which lies ∼0.6kb upstream of Tce1 enhancer (17, 18). Expression of Dreg1 is high in early T cell progenitors and Th2 cells, mirroring Gata3 expression (17). Overexpression studies indicated that it may be involved in the establishment but not maintenance of Gata3 expression (17, 18), but it remains unclear how the loss of Dreg1 affects immune system development. Here, we explore the role of Dreg1 in the regulation of immune cell differentiation. We find that deletion of the Dreg1 locus leads to a specific loss of ILC2 progenitors and their downstream progeny in non-lymphoid tissues but not T or NK cells. Chromatin profiling reveals that the regulatory regions around the Dreg1 gene are open in haematopoietic progenitors and are bound by the transcription factor TCF1 in early innate lymphoid progenitors and TCF1-deficient mice lack Dreg1 expression. Finally, we discovered potential homologues of Dreg1 that highly expressed in human ILC2 emanating from a syntenic enhancer of GATA3.

Results

Deletion of the Dreg1 locus results in a specific reduction of ILC2

To understand how Dreg1 contributes to the development of the immune system we generated Dreg1-deficient mice by CRISPR-Cas9-mediated excision of the complete 3kb of the Dreg1 gene (Figure 1A). We confirmed that the Dreg1 gene was removed from the germline and found that these mice reproduced at Mendelian frequencies and exhibited no overt developmental defects suggesting that Dreg1 is not required for embryogenesis.

Figure 1:

Given our previous finding of the relatively high levels of Dreg1 expression in T cells (17), and of a critical role of the Tce1 (Gata3 +278/285) enhancer (12, 14), we expected that the deletion of Dreg1 would impact T cell and NK cell development. We therefore performed an analysis of the different populations of T cells. Mature CD4⁺ and CD8⁺ T cells in the spleen appeared unaltered in frequency (Supplementary Figure 1). Moreover, both populations had similar proportions of naïve, effector, and memory cells (Supplementary Figure 1B &C). This suggests that Dreg1 is not required for conventional T cell development or maintenance. In addition, the loss of Dreg1 did not affect the development of NK cells and γδ T cells as normal proportions were observed in the spleen (Supplementary Figure 1D). Overall, this analysis suggests that Dreg1 is dispensable for NK and T cell development.

Gata3 is also critical for the development of all innate lymphoid cells (ILCs)(9) and the levels of Gata3 are critical in directing ILC2 lineage development (8, 19). Therefore, we next explored peripheral sites known to harbour substantial numbers of these cell types such as visceral adipose tissue (VAT), the small intestine lamina propria (SI LP) and the lung. While these tissues had relatively normal proportions of NK, ILC1 and ILC3 (Figure 1B, C & E) we noted a specific reduction in the population of ILC2 in each of the tissues we examined (Figure 1D). Taken together these results show that the loss of Dreg1 results in the specific loss of ILC2 in different peripheral tissues.

Dreg1 deficiency results in the reduction of ILC2 progenitors

We next investigated the origins of the ILC2 defect in the Dreg1-deficient mice. It is known that expression of Gata3 in the early ILC progenitor stage is required for development of mature ILC2s (8, 10, 19). Indeed, analysis of publicly available RNAseq data (20) demonstrated that Gata3 expression was very high in ILC2P and ILC2 compared to other populations, whereas the highest expression of Dreg1 was found in ILC2P (Figure 2A). We examined ILC development in the bone marrow (Figure 2B-G, gating strategy in Supplementary Figure 2A). We began by examining the common lymphoid (CLP) and lymph-myeloid primed progenitors (LMPP) and found no overt changes in proportions or total number of these populations (Figure 2B & C). However, we found increased CILP, CHILP and ILCP in the Dreg1-deficient mice (reaching significance in the CILPs) (Figure 2D-F). We observed a substantial reduction of the ILC2P population suggesting that the loss of Dreg1 leads to a block in ILC development (Figure 2G). This suggests that the loss of Dreg1 results in a bottleneck at the early stages of ILC development. We next examined the levels of Gata3 in these populations (Figure 2H). We observed a reduction in Gata3 signal in the upstream CHILP and the ILCP from the Dreg1-deficient mice, but interestingly Gata3 levels were normal in the ILC2P even though this population is reduced in the Dreg1-deficient mice. This suggests that Dreg1 acts to regulate Gata3 prior to the ILC2P stage which leads to a block in their development, however, if cells can overcome this perturbation, they express relatively normal levels of Gata3.

Figure 2:

Dynamic epigenetic regulation of the Dreg1 locus in a Tcf1-dependent manner during ILC development

To gain a greater understanding of how the expression of Dreg1 is regulated we next studied the chromatin state of the Dreg1 locus during early ILC development via analysis of ATAC-Seq data generated on progenitors and their downstream ILC progeny (Figure 3A). Of note, the original Refseq annotation of Dreg1 was missing the first exon and TSS which we identified (17) and show in Figure 3A. We observed three significant accessible chromatin “peaks”, with the predominant one at the promoter which was open at the CLP stage remained so in ILCP and ILC2P. One region downstream became more accessible in ILCP and ILC2P in line with that observed by others (14). Overall, we find the promoter of Dreg1 is open in early lymphoid progenitors and the acquisition of chromatin accessibility downstream correlates with increased Dreg1 expression in ILC2 progenitors.

Figure 3:

To understand the transcription factors that might be involved in regulating Dreg1 expression we examined the motifs at the Dreg1 promoter and downstream accessible region. We observed a specific enrichment of motifs for Tcf1, Lef1 and Ets1 suggesting that these transcription factors may be involved in regulating Dreg1 expression. As Tcf1 (encoded by the Tcf7 gene) has been implicated in ILC2 development (21, 22) we focussed on its role in regulating Dreg1 expression. Firstly, we examined publicly available Tcf1 CUT&Run data (23) and found that indeed Tcf1 was specifically bound to the accessible sites of the Dreg1 locus in early innate lymphoid progenitors (EILP, which are a population that sit between CILP and CHILP) (Figure 3B). We next examined ChIC-Seq data (24) of active (H3K27ac) or repressive (H3K27me3) histone modifications at different stages of ILC development (Figure 3D). Firstly, we found no evidence of H3K27me3 in this region in any of the cell types examined. However, there was a modest accumulation of H3K27ac around the accessible regions in the Dreg1 locus at the EILPs which was noticeably increased at the ILCP stage. Of note, Tcf1-deficient (via deletion of the Tcf7 gene) EILP showed lower levels of H3K27ac suggesting that Tcf1 is likely involved in creating a permissive chromatin landscape at the Dreg1 locus during ILC development. Indeed, the expression of Dreg1 was dependent on the presence of Tcf1 as Tcf7-/- EILPs lacked Dreg1 expression (Figure 3E). Taken together this data suggests that the alterations to the chromatin state Dreg1 locus and its subsequent transcriptional activity during ILC development is dependent on Tcf1.

Identification of a human GATA3 enhancer harbouring Dreg1 homologues that are active in ILC2s

Previously we identified a region syntenic to the mouse Dreg1 locus on human chromosome 10 that is a similar distance from the GATA3 gene and forms an 3D interaction in T cells but not B cells suggesting it may a represent an enhancer element (17). Interestingly, bidirectional transcription initiates from the sequence-conserved cis element to produce two potential human homologs of Dreg1, the lncRNAs CAT00000105356.1 (we named DREG1.1) and CAT00000117261.1 (DREG1.2). To explore whether these transcripts are also expressed in ILC populations we examined publicly available RNAseq data from healthy human blood (25)(Figure 4A). We found that GATA3 was highly expressed in human ILCP, ILC2 and TH2 cells. We also observed this pattern for DREG1.1 and to an even greater extent for DREG1.2 suggesting that the syntenic region harbours transcripts which reflect the same expression pattern as murine Dreg1.

Figure 4:

Finally, we investigated the functional role of this potential enhancer of GATA3 by examining a recently published tiling CRISPR deletion screen searching for regulatory elements that control GATA3 expression in human Th2 cells (26). We focused on the syntenic region which was defined as functional sequence (FS) 23 (Figure 4B). Specifically, guides that deleted conserved sequences FS23-4 and FS23-5 led to a significant downregulation of GATA3 expression (26) suggesting that this region represents a distal enhancer of GATA3 (Figure 4B). ATACseq data from either cultured (pcILC2) or fresh ILC2 from healthy human blood (27) revealed a clear accessible region in these cells that overlaps the DREG1.1/2 and FS23 (Figure 4C). Overall, we find that this syntenic region in humans represents an enhancer that is important for high levels of GATA3 expression and contains two non-coding RNA genes that show a similar expression pattern to murine Dreg1.

Discussion

Here we show that germline excision of the Dreg1 locus resulted in a specific reduction in ILC2 cells owing to a developmental bottleneck during ILC development in the bone marrow. These results are in line with the fact that very high levels of Gata3 are required for ILC2 development (8, 10). Given its proximity to the Tce1 enhancer, a surprising finding was that the deletion of Dreg1 did not affect T or NK cells lineages. While it is tempting to speculate that the Dreg1 locus represents a ILC2-specific enhancer element we favour the idea that Dreg1 likely contributes to dosage tuning of Gata3 during ILC2 commitment and that ILC2s because of their higher Gata3 threshold are more sensitive to modest reductions, whereas T and NK cells are buffered by redundant modules.

How Dreg1 is involved in regulating Gata3 levels mechanistically is yet to be determined. Dreg1 may act as an enhancer RNA that cooperates with chromatin modifying proteins (28) to reinforce local H3K27ac and enhancer-promoter communication. Future experiments that selectively suppress Dreg1 transcription for example via antisense oligonucleotides or CRISPRi at the Dreg1 promoter will enable the discrimination of RNA-dependent from DNA element-dependent effects.

In humans, we identify two noncoding transcripts within the syntenic GATA3 distal enhancer that are highly expressed in ILCP, ILC2 and TH2 cells. Perturbations within this conserved block in an independent tiling-deletion screen resulted in reduced GATA3 expression in T cells. While these experiments were performed in T cells, we expect that this enhancer region will likely also play an important role in the regulation of ILC2 development. Given that ILC2 have been implicated in the development of allergic reactions (4) the noncoding transcripts we identified may represent targets for the modulation of GATA3 levels and ILC2 function in disease.

Methods

Mice

The Dreg1 knockout mouse line was generated at the Melbourne Advanced Genome Editing Centre (MAGEC) laboratory using CRISPR-Cas9 through microinjection of 2 sgRNAs (CTACTTGCTGACAAGTCGTC and TCTAGTAAGTCCAGTTGCTT) to target the 3,594 bp genomic region around Dreg1. The F0 offspring were validated for Dreg1 deletion with forward primer GACCAGATATGGAGACGTGCA and reverse primer TCTTTGCCATCTTCTGTGTGC and were backcrossed with wild-type C57BL6/J mice.

Cells and tissues were obtained from littermate control mice aged at least 6 weeks unless otherwise specified. Due to sex differences in ILC populations (29), only female mice were used. All mice were maintained at the WEHI Animal Facility under specific pathogen-free conditions.

Cell isolation

Thymus and spleen

The spleen and thymus were made into single-cell suspension by mashing through a 70µm mesh. Splenic cells were treated with red cell lysis buffer.

Bone marrow

Both pairs of hips, femurs, and tibia bones were grounded in a mortar and pestle to collect the bone marrow. The cell suspension was filtered through a 70µm cell strainer and negatively enriched for ILC cells with the mouse lineage cell depletion kit (Miltenyi).

Extraction of lymphocytes from non-lymphoid tissues

Lymphocytes from the VAT and lung were finely chopped in 3-5mL digestion buffer (RPMI and 2% FCS with 2mg/mL Collagenase IV). The suspension was digested for 40 min at 37°C and shaken at 180 rpm. Digested VAT samples were topped up to 30ML with FACS buffer (1xPBS with 2% FCS) and spun at 800g for 15 min at 4°C. Lungs were mashed through a 70mM filter and topped to 30mL FACS buffer and pelleted at 1700rpm for 5min. Both VAT and lung pellets were treated with RBC lysis buffer.

Isolation of lymphocytes from small intestinal lamina propria

Mesentery and Peyer’s patches were removed longitudinally, and faeces removed. Intestine was chopped into 0.5cm pieces into 1xPBS and vortexed. Intestinal pieces were placed into 20mLs of dissociation buffer (1xHBSS without Ca2+ and Mg2+ with 2% FCS, 10mM HEPES and 5 mM EDTA) and digested for 40min at 37°C at 180rpm. Digested pieces were washed with FACS buffer and then placed in 5-8mL of digestion buffer (1xHBSS with indicator containing 2% FCS, 10mM HEPES, 2mg/mL Collagenase IV and 2mg/mL DNAse I) and incubated for 30-40min at 37°C at 180rpm. Digested tissue was mashed through a 70 mM cell strainer and pelleted. A 40/80% percoll gradient was performed to enrich for lymphocytes.

Flow cytometry and antibodies

Fluorophore conjugated antibodies against mouse antigens were used for flow cytometry. Antibodies, clone names and manufactures: Biolegend- CD4 (GK1.5) APCFire750, CD8a (53- 6.7) BV570, NKp46 (29A1.4) PEcy7, ICOS (7E.17G9) BV421, Ki67 (11F6) BV650, CD62L (MEL-14) APC and T-bet (4B10) APC. BD- CD19 (1D3) BV786, TCRb (H57-597) BV750, TCRgd (GL3) BV711, B220 (RA3-6B2) BV786, GATA3 (L50-823) BUV395, RORgt (Q31-378) BV421, RORgt (Q31-378) PerCPcy5.5, KLRG1 (2F1) BUV615, CD49a (Ha31/8), CD25 (PC61) BV480, NK1.1 (PK136) BUV805, NK1.1 (PK136) BB700, CD44 (IM7) BUV496. Invitrogen/bioscience- CD11c (N418) FITC, FOXP3 (FJK-16s) PE, IL-33R (ST2) (RMST2-2) PerCP-eFLuor710, Streptavidin BUV563.WEHI-CD45.2 (104) AF700. Live dead staining was performed using Fixable viability stain 440UV diluted in 1xPB at 1:1000 dilution. Surface staining of antibodies was for 30min on ice at 1/200, 1/300 or 1/400 in FACS buffer. Intracellular staining of transcription factors was overnight at 4°C using the Foxp3/Transcription factor staining buffer set from eBioscience overnight. Cell numbers for specific cell populations were calculated using the volume of the sample that was analyzed, the cell counts per sample, and the percentage of the specific cell population.

RNA-Seq Analysis

Publicly available RNA-Seq data were downloaded from SRA and re-analysed. Briefly fastq files were analyzed with FASTQC (0.12.1) and adapters were removed using trimmomatic(30) (0.36). The reads were aligned to GRCm39 for mouse samples and GRCh38 for human samples using hisat2(31) (2.0.5). Gene counts were quantified using Rsubread(32) featureCounts (1.6.3) with annotation from gencode(33) (vM33 for mouse and v42 for human), which had the annotation for the Dreg1 gene or human homologs of the Dreg1 gene added, respectively. Samples were processed using edgeR(34) (3.42.4). Lowly expressed genes were removed, and samples were normalized according to library size. Gene expression was plotted as logCPM using ggplot2 (3.4.4).

ATAC-Seq Analysis

Publicly available ATAC-Seq data were downloaded from SRA and re-analysed. Processing was based on the ENCODE ATAC-Seq pipeline. Briefly fastq files were analyzed with FASTQC and adapters were removed using trimmomatic (0.36). The reads were aligned to the GRCm39 gnome for mouse samples and GRCh38 genome for human samples using bowtie2 (2.4.4). Duplicate reads were marked using Picard tools (2.26.11) and reads were filtered using samtools(35) (1.9) with -F 1804 -f 2. Fold-change coverage tracks were created using macs2 (2.2.7.1) callpeak with a smooth window of 150 and a shift size of -75 and bdgcmp. The bedgraph was converted to a bigwig using ucsc tools (331) bedGraphToBigWig(36) and visualized using IGV(37) (2.17.4) with group scaling among samples from the same experiment.

Motif Scanning

The regions of interest as identified through peak calling of ATAC-Seq data were scanned for motifs using FIMO(38) from the MEME suite (5.0.5) using the Hocomoco(39) (H12) core motif database. The resulting list was filtered to remove factors that were not expressed in ILC2P cells as determined through bulk RNA-Seq (GSE77695) with a cutoff of >= 1 LogCPM.

ChIC and CUT&Run Analysis

Publicly available ChIP-Seq or CUT&Run data were downloaded from SRA and re-analysed. Briefly fastq files were analyzed with FASTQC and adapters were removed using trimmomatic (0.36). The reads were aligned to the GRCm39 gnome for mouse samples using bowtie2 (2.4.4) with the options -q -5 0 -3 0. Duplicate reads were marked using Picard tools (2.26.11) and reads were filtered using samtools (1.9) with -F 1804 -f 2. Fold-change coverage tracks were created using macs2 (2.2.7.1) callpeak with a smooth window of 50 and a shift size of -25 for transcription factors and 150 and -75 for histone marks along with macs2 bdgcmp. The bedgraph was converted to a bigwig using ucsc tools (331) bedGraphToBigWig and visualized using IGV (2.17.4) with group scaling among samples from the same experiment.

CRISPR Guides

The guide counts were downloaded from GSE190860 and were used to create a bed file to identify the target regions. The bed file was visualized in IGV (2.17.4).

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files; source data files have been provided for all figures.

Acknowledgements

The authors thank the members of the Allan and Vasanthakumar labs for their input into this project. The authors also gratefully acknowledge the WEHI Flow Cytometry and Bioservices Facilities for their support and assistance in this work. We also thank the Melbourne Advanced Genome Editing Centre (MAGEC) laboratory for generation of the Dreg1^-/- mice.

Additional information

Funding

This work was supported by the NHMRC, Australia Ideas Grant 2001131 and a Stafford Fox Medical Foundation Research Grant to R.A and an NHMRC Investigator Grant 2009336 to A.V. This work was made possible through Victorian State Government Operational Support Program and the Australian Government NHMRC IRIISS.

Authorship contributions

Sara Quon: Conceptualization, Investigation, Formal Analysis, Writing-Original Draft, Review & Editing. Adelynn Tang: Conceptualization, Investigation, Formal Analysis, Writing-Review & Editing. Nadia Iannarella: Investigation. Kael Schoffer: Investigation. Wing Fuk Chan: Conceptualization, Methodology. Timothy Johanson: Investigation, Supervision. Ajith Vasanthakumar: Conceptualization, Supervision, Funding Acquisition, Writing-Review & Editing. Rhys Allan: Conceptualization, Supervision, Funding Acquisition, Writing-Original Draft, Review & Editing.

Funding

Australian NHMRC (2001131)

• Rhys Allan

Stafford Fox Medical Research Foundation

• Rhys Allan

Australian NHMRC (2009336)

• Ajithkumar Vasanthakumar