TET proteins regulate Drosha expression and impact microRNAs in iNKT cells

DNA demethylases TET2 and TET3 play a fundamental role in thymic invariant natural killer T (iNKT) cell differentiation by mediating DNA demethylation of genes encoding for lineage specifying factors. Paradoxically, differential gene expression analysis revealed that significant number of genes were upregulated upon TET2 and TET3 loss in iNKT cells. This unexpected finding could be potentially explained if loss of TET proteins was reducing the expression of proteins that suppress gene expression. In this study, we discover that TET2 and TET3 synergistically regulate Drosha expression, by generating 5hmC across the gene body and by impacting chromatin accessibility. As DROSHA is involved in microRNA biogenesis, we proceed to investigate the impact of TET2/3 loss on microRNAs in iNKT cells. We report that among the downregulated microRNAs are members of the Let-7 family that downregulate in vivo the expression of the iNKT cell lineage specifying factor PLZF. Our data link TET proteins with microRNA expression and reveal an additional layer of TET mediated regulation of gene expression.


Introduction
Ten Eleven Translocation (TET) proteins are enzymes that regulate the process of DNA demethylation by oxidizing 5 methylcytosine (5mC) to generate 5 hydroxymethylcytosine (5hmC) also known as the sixth base of our genome (1).In addition, TET proteins can oxidize 5hmC to generate additional modified cytosines, namely 5 formylcytosine (5fC) and 5 carboxylcytosine (5caC) (2,3).The TET family of proteins consists of three members: TET1, that is most highly expressed in embryonic stem cells (ESCs), TET2, which is broadly expressed in various cell types and developmental stages, and TET3 that is more highly expressed as cells differentiate (4).All three TET proteins exert critical roles in shaping the development and function of a vast array of cells (5,6).We have previously demonstrated that 5hmC is dynamically distributed across the genome of thymic T cell subsets (7).During the process of T cell lineage specification, 5hmC was shown to be increased in the gene body of very highly expressed genes and in active enhancers (7).Similar findings have been reported for murine and human peripheral T cells (7)(8)(9)(10), indicating the critical role of TET proteins and 5hmC in regulating gene expression in T cells (5).
To dissect the in vivo roles of TET proteins in T cell development we generated Tet2-/-Tet3flx/flx CD4 cre (Tet2/3 DKO) mice (11).We focused our analysis on concomitant deletion of TET2 and TET3 since our data indicated redundancy between TET proteins (11).The phenotype of the Tet2/3 DKO mice was complex, revealing that TET proteins are critical regulators of various T cell types.Specifically, TET2 and TET3 are fundamental for the stability of the transcription factor (TF) FOXP3 and thus the functionality and stability of regulatory T cells (Tregs) (12).In addition, Tet2/3 DKO mice exhibit a striking expansion of invariant natural killer (iNKT) T cells (11).
iNKT cells are an unconventional type of T cells that express an invariant TCR Vα14 chain and recognize lipids instead of peptides (13).iNKT cells develop in the thymus endowed already with effector properties and they have the ability to generate significant amount of cytokines, immediately upon antigen encounter (14).iNKT cell lineage commitment is orchestrated by the TF Promyelocytic leukaemia zinc finger (PLZF) protein, which endows iNKT cells with effector properties (15,16).In the thymus, we can discern 3 subsets based on the expression of TFs and their functional properties: NKT2 express GATA3, NKT17 express RORt and NKT1 express T-bet (17)(18)(19)(20).iNKT cells exert important roles in recognition of bacterial pathogens and have been shown to be of clinical value in the context of cancer immunotherapy (14,(21)(22)(23)(24). Thus, deciphering the molecular mechanisms that shape their differentiation and functionality is of outmost importance in order to take full advantage of their effector properties (25,26).
We have previously demonstrated that Tet2/3 DKO iNKT cells show increased expression of the TF RORt (11,27,28).Integration of our genome wide datasets evaluating gene expression, whole genome methylation, whole genome hydroxymethylation and chromatin accessibility analysis in control and Tet2/3 DKO iNKT cells revealed that TET2 and TET3, by regulating DNA demethylation, upregulate lineage specifying TFs such as T-bet and Th-POK that are critical for iNKT cell lineage diversification and for suppression of RORt (11), in a TET2 dependent catalytic manner (29).However, not all the observed differences in the gene expression program of Tet2/3 DKO iNKT cells can be attributed to gain of methylation in promoters or enhancers of the differentially expressed genes (11).That was particular true in the context of genes that were gaining expression upon loss of TET proteins (11).One possibility is that deletion of TET proteins can result in downregulation of repressors, allowing the upregulation of the targeted genes (30,31).Repression of genes can occur by small RNAs that target mRNAs and can mediate their degradation (32).
DROSHA regulates the generation of precursor miRNAs in the nucleus and then further processing occurs in the cytoplasm by DICER and the ARGONAUTE complex (33,34).
Notably, miRNAs are important for iNKT cell development as indicated by Dicer deficient mice (35,36).In this study, we report that TET proteins regulate expression of Drosha in iNKT cells.We demonstrate that Tet2/3 DKO iNKT cells show altered expression of precursor and mature miRNAs.Among the identified downregulated miRNAs are members of the Let-7 family that has been demonstrated in vivo to target and downregulate the transcription factor PLZF in iNKT cells (37).

Results and Discussion
Analysis of our previously published RNA-seq datasets (11) revealed that Drosha was downregulated in Tet2/3 DKO iNKT cells (Figure 1A).To further dissect the molecular mechanisms by which TET2 and TET3 can impact expression of Drosha we assessed 5hmC distribution across the gene body.5hmC upon treatment with bisulfite sequencing is converted to cytosine-5-methylenesulfonate (CMS) (38).Analysis of CMS immunoprecipitation with sequencing (CMS-IP seq) (39,40) datasets (11) revealed that in wild type iNKT cells 5hmC is distributed across the gene body of Drosha (Figure 1B).We have previously demonstrated that 5hmC is enriched in the gene body of highly expressed genes, whereas the promoters of these genes are devoid of 5hmC, in conventional T cells and unconventional iNKT cells (7,11).Similar findings have been reported for naïve and helper T cell subsets (8-10) as well as for regulatory T cells (41).In addition, we have previously shown that 5hmC correlates with chromatin accessibility in both conventional and unconventional T cells (11,29).We then investigated how loss of TET proteins may impact chromatin accessibility in the Drosha locus.Thus, we compared our datasets (11) for assay for transposase accessible chromatin with sequencing (ATAC-seq) (42) for wild type and Tet2/3 DKO iNKT cells.We demonstrate that in Tet2/3 DKO iNKT cells there is reduced accessibility in an intragenic genomic region (mm10: chr15:12,894,551-12,896,829) that has increased accessibility and enrichment of 5hmC in wild type iNKT cells (Figure 1B).Due to the low abundance of 5hmC in Tet2/3 DKO thymic T cell subsets we were not able to perform CMS-IP seq for the Tet2/3 DKO iNKT cells (11).However, we performed whole genome bisulfite sequencing (WGBS) in order to assess at single-nucleotide resolution the modification status of cytosine.Our analysis revealed a gain of methylation at this intragenic region in the Drosha locus at the Tet2/3 DKO iNKT cells (Figure 1B).
We hypothesize that this intragenic region may exert regulatory function to promote the expression of Drosha.We have previously shown that 5hmC decorates active enhancers (7).Additional studies have demonstrated a strong correlation of 5hmC with active enhancers in various T cell subsets (8,41).In many cases these regulatory elements that require 5hmC enrichment in order to be active are intragenic, such as the CNS2 enhancer in the Foxp3 locus (12,43,44), an intragenic enhancer that regulates stable expression of Cd4 gene in CD4 cells (45) as well as the proximal enhancer of Zbtb7b gene that encodes Th-POK (29).
We have also demonstrated that 5hmC decorates intragenic cite A at the Zbtb7b locus to regulate the accessibility and the binding of the transcription factor GATA3 (29).It has been previously suggested that the binding of GATA3 to cite A promotes Th-POK expression (46).
We have previously discovered a shared gene expression program between Tet2/3 DKO thymic iNKT cells and CD4 single positive (SP) cells (29).As DROSHA is expressed in both subsets we investigated if its expression was also affected in CD4 SP cells.We report that Drosha is downregulated in Tet2/3 DKO CD4 SP cells (Supplementary Figure 1).
Interestingly, there is 5hmC enrichment at the same intragenic site of the locus in WT CD4 SP cells (Figure 1B).In addition, we looked into our data assessing recruitment of GATA3 (by CUT&RUN) in WT and Tet2/3 DKO CD4 SP cells (29).We discover that GATA3 binds in this region in WT CD4 SP cells, whereas no binding was detected in Tet2/3 DKO CD4 SP cells.Moreover, we looked into the binding of Th-POK by using publicly available ChIP-seq datasets (47) and we demonstrate binding of Th-POK in this potentially regulatory region in CD4 SP cells.Collectively, our findings suggest that TET2 and TET3 generate 5hmC and regulate chromatin accessibility in the Drosha locus to promote the expression of the gene (Figure 1).Further studies are required to elucidate the precise regulatory elements that control the expression of Drosha.However, as TET2 and TET3 deletion results in partial reduction of Drosha expression and not complete loss it becomes apparent that additional mechanisms are in place to control the expression of this gene.
As DROSHA is involved in regulating the pathway of microRNAs (miRNAs) we asked whether the reduced expression of Drosha has an impact on the miRNAs that are expressed in Tet2/3 DKO iNKT cells.To identify small RNAs that are impacted we isolated thymic iNKT cells by FACS sorting (Supplementary Figure 2) from wild type or Tet2/3 DKO mice and we performed small RNA-seq (Figure 2A).Comparison of precursor and mature miRNAs in the WT and the Tet2/3 DKO iNKT samples confirmed that samples of the same genotype were similar to each other (Supplementary Figure 3).Our analysis compared expression of precursor (Supplementary Table 1) and mature miRNAs (Supplementary Table 2) and we found that among those that were differentially expressed the majority were downregulated (Figure 2B, C).This could be due to the downregulation of Drosha expression.We then focused on the affected mature miRNAs (Figure 2C).The vast majority of the differentially expressed mature miRNAs were downregulated in the Tet2/3 DKO iNKT cells (Figure 2C).
An additional mechanism could be that in the absence of TET proteins at least some miRNAs could gain cytosine methylation, resulting in their downregulated expression.However, when we looked into our previously generated WGBS data (11) we did not notice significant changes in methylation for the vast majority of the miRNAs that were differentially expressed in thymic iNKT cells.We only detected some gain of methylation in mir199b and mir7058 (Supplementary Figure 4).Our analysis demonstrated that among the downregulated miRNAs were members of the Let-7 family.Specifically, we observed downregulation of Let-7c, Let-7b and Let-7k (Figure 2C).Interestingly, Let-7 miRNAs have been previously shown to target Zbtb16 mRNA, which encodes for PLZF, for degradation in murine iNKT cells in vivo (37).Thus, we hypothesize that the downregulation of some of the members of the Let-7 family could result in increased expression of PLZF.We evaluated PLZF levels in WT and Tet2/3 DKO iNKT cells by Flow cytometry (Figure 3A).Our data indicates that Tet2/3 DKO iNKT cells exhibit upregulation of PLZF (Figure 3A, B).Thus, we propose that in Tet2/3 DKO iNKT cells downregulation of some of the Let-7 miRNAs results in reduced targeting for degradation of Zbtb16 mRNA, resulting in increased expression of PLZF (Figure 3C).

Conclusions
In this study, we report that TET2 and TET3 regulate the expression of Drosha.We also discover various miRNAs that are differentially expressed including downregulation of Let-7 miRNAs.However, we must emphasize that the NKT17 skewing of the Tet2/3 DKO iNKT cells can be fully rescued by deletion of ThPOK and partially rescued by deletion of T-bet as we have previously shown (11).Importantly, our unbiased, integrative analysis of genome wide datasets indicated that both ThPOK and Tbet are targets of TET proteins based on 5hmC enrichment and gain of methylation upon concomitant TET2 and TET3 loss (11).Thus, in support of our previous findings that TET proteins exert multifaceted roles in regulating gene expression (30,31,48), we propose an additional layer of TET-mediated regulation of lineage specification by affecting expression of miRNAs.

Mice
Mice were housed in pathogen free conditions in the Genetic Medicine Building at University of North Carolina (UNC) Chapel Hill in a facility managed by the Division of Comparative Medicine at UNC Chapel Hill.All the experiments using mice in this study were performed according to our approved protocol by the UNC Institutional Animal Care and Use Committee (protocol no: 22-252).Age and sex-matched mice were analyzed.Male and female mice were used for our experiments.Control (C57BL/6 (B6), strain number: 000664), RRID: IMSR_JAX: 000664) mice were purchased from Jackson (Jax) laboratories and were bred in our facility at UNC. Tet2-/-Tet3flx/flx CD4 cre mice have been previously described (11,29).Briefly, Tet2-/-mice (49) (Jax strain no; 023359, RRID: IMSR_JAX:023359) were crossed with Tet3flx/flx (50, 51) (Jax strain no: 031015, RRID: IMSR_JAX:031015) CD4cre mice (52).To determine the genotype of the mice, tissue was isolated and genomic DNA was extracted using Phire Animal Tissue Direct PCR kit (Thermo scientific, cat no F-140WH) following the manufacturer's protocol.Then DNA fragments were amplified by PCR using the Phire DNA polymerase (Thermo scientific, cat no F-140WH) and specific primers using Biorad T100 or Biorad C1000 Touch thermocyclers.

Cell preparation
Thymocytes were isolated from young mice 21-25 days old.Thymocytes were dissociated to prepare single cell suspensions as previously described (53,54).

Statistical Analysis
For the statistical analysis we used Prism software (Graphpad).We applied unpaired student's t test.In the relevant figure legends, we indicated p-values for statistically significant differences (p < 0.05).Data are mean ± s.e.m.In the graphs, each dot represents a mouse.Unless otherwise indicated the p-value was not statistically significant.Differences were considered significant when p < 0.001 ( * * * ); < 0.0001 ( * * * * ).Both male and female mice from different litters were evaluated, with reproducible results.

RNA isolation, Library preparation of small RNAs and sequencing
FACS sorted iNKT cells were lysed in RLT plus lysis buffer from the miRNeasy plus kit (Qiagen, cat no: 217084).Total RNA was isolated following the instructions provided by the manufacturer and was quantified using Qubit RNA High Sensitivity assay (Invitrogen) in Qubit 4 Fluorometer (Invitrogen).Total RNA was provided to the UNC High Throughput Sequencing Facility (HTSF).RNA integrity was evaluated with a Tapestation (Agilent) using High Sensitivity RNA ScreenTape (Agilent).RNA with RIN value>9 was used for library preparation.Small RNA libraries were generated using the Revvity NETFLEX small RNA sequencing kit V4.Libraries were pooled and sequenced in an Illumina NextSeq2000 P1 Single End 1x50 to obtain 100 million single end reads.6 biological replicates for wild type and 7 biological replicates for DKO samples were analyzed.Both male and female mice were evaluated.The small RNA samples were processed using nf-core/smrnaseq (2.2.4) using default parameters (55,56).The differential expression analysis was done using nfcore/differentialabundance (1.4.0) using default parameters (57).

CMS-seq data analysis
The CMS-IP and input reads from 2 biological replicates of WT iNKT cells were mapped against mm10 using Bismark (0.22.3) (58).The mapping was done using the Bowtie 2 (2.4.1) (59) backend in the paired-end mode with the following parameter values: -I 0 -X 600 -N 0.

ATAC-seq data analysis
Adapter trimming and quality filtering of the sequencing libraries (3 biological replicates per genotype, 6 samples in total) was done using fastp (0.21.0) (61) with the default parameters.
Reads with identical sequences were filtered and only one was retained for subsequent analysis.The coverage tracks were generated from the samples obtained by pooling the biological replicates using HOMER (4.10) (makeBigWig.pl-norm 1e6) (60).

Figure 1 .
Figure 1.TET2 and TET3 regulate expression of Drosha in thymic iNKT cells.A. Gene expression of Drosha in WT (in black) and Tet2/3 DKO thymic iNKT cells (in purple), evaluated by RNA-seq. 3 biological replicates per genotype were assessed.*** (p =0.0004), unpaired t test.Each dot represents an individual biological replicate.Horizontal lines indicate the mean (s.e.m.).B. Portraits of epigenetic regulation (determined by 5hmC, 5mC and chromatin accessibility) in thymic iNKT cells and transcriptional regulation in CD4 SP cells of the Drosha locus.Genome browser view of 5hmC distribution (by CMS-IP seq) in the gene body of Drosha in WT iNKT cells reveals enrichment of this modification indicating TET activity.2 biological replicates were analyzed.

Figure 2 .
Figure 2. Differential expression of precursor (hairpin) and mature miRNAs in WT and Tet2/3 DKO thymic iNKT cells.A. Experimental outline.B. Heatmap indicating hairpin miRNAs whose adjusted p-value < 0.05 and absolute log2 fold-change > 2. The z-score normalized expression values are shown.C. Heatmap indicating mature miRNAs whose adjusted p-value < 0.05 and absolute log2 fold-change > 2. The z-score normalized expression values are shown.Both male and female mice were used for each genotype.N=6 WT mice and N=7 Tet2/3 DKO mice were used.

Figure 3 . 1 A
Figure 3. Let-7 miRNAs downregulation in Tet2/3 DKO thymic iNKT cells contributes in upregulation of PLZF. A. Representative flow cytometry plots of thymocytes isolated from wild type and Tet2/3 DKO mice identify iNKT cells as aGalCer-loaded tetramer + and TCR intermediate cells.Representative histogram for the lineage specifying transcription factor PLZF indicates increased expression, determined by intracellular staining and Flow cytometry, in the Tet2/3 DKO thymic iNKT samples (in purple) compared to WT (in black) counterparts.B. Plot