Histone H3 Lysine 27 demethylases Jmjd3 and Utx are required for T-cell differentiation

Although histone H3 lysine 27 trimethylation (H3K27Me3) is associated with gene silencing, whether H3K27Me3 demethylation affects transcription and cell differentiation in vivo has remained elusive. To investigate this, we conditionally inactivated the two H3K27Me3 demethylases, Jmjd3 and Utx, in non-dividing intrathymic CD4+ T-cell precursors. Here we show that both enzymes redundantly promote H3K27Me3 removal at, and expression of, a specific subset of genes involved in terminal thymocyte differentiation, especially S1pr1, encoding a sphingosine-phosphate receptor required for thymocyte egress. Thymocyte expression of S1pr1 was not rescued in Jmjd3- and Utx-deficient male mice, which carry the catalytically inactive Utx homolog Uty, supporting the conclusion that it requires H3K27Me3 demethylase activity. These findings demonstrate that Jmjd3 and Utx are required for T-cell development, and point to a requirement for their H3K27Me3 demethylase activity in cell differentiation.


Supplementary Figures
Supplementary Figure 1 Targeting strategy for the generation of conditional Jmjd3-or Utx conditional mice.
(a, b) Structure of wild-type Jmjd3 (a) or Utx (b) loci and targeting vectors. The targeting vector for each gene contains an inserted neomycin (Neo) cassette flanked by Frt (white arrowheads), and three LoxP sites (black arrowheads) flanking the Neo cassette and exon(s) encoding the catalytic JmjC domain for each enzyme. Homologous recombination in embryonic stem cells results in the targeted allele (Jmjd3 t or Utx t ).
Subsequent deletion of Frt-flanked Neo cassette by Flpe recombinase gives rise to the floxed allele (Jmjd3 f or Utx f ), which can be further modified to the deleted allele (Jmjd3 or Utx by the Cre recombinase.

Supplementary Figure 2
Deletion of Jmjd3 and Utx alleles.
(a) Depletion of Jmjd3 protein in Jmjd3 f/f Cd4-Cre (Jmjd3 KO) thymocytes as determined by immunoprecipitation (IP) and immunoblot analyses with a Jmjd3 antibody. Numbers below the image are the relative intensity ratio of the Jmjd3 bands measured by densitometric analyses. NS: non-specific band.
(b) Deletion of the floxed region in Utx allele as monitored by genomic PCR. Two biological replicates are shown for wild-type (WT) and Utx f/f Cd4-Cre mice (Utx KO), using genomic DNA purified from total thymocytes of each mouse.
(c) Depletion of both Jmjd3 and Utx proteins in double-deficient (dKO) thymocytes as determined by IP/immunoblot analyses with a Jmjd3 or Utx antibody.

Supplementary Figure 3
Impaired T cell development in H3K27 demethylase-deficient mice.
(a-c ) Plots indicate numbers of spleen CD44 lo CD4 + T cells (a), mature (TCR hi CD24 lo ) CD8 SP thymocytes (b) and spleen CD8 + T cells (c) from indicated mice. Each symbol represents one 6-8 week-old mouse. Data is from the same mice as in Fig. 1b.
(e) Dot plots show absolute numbers of total or mature thymocytes, or of CD4 + T cells from mice in (d). Each symbol represents one individual mouse. Data are representative of or combined from three independent experiments (d, e).

Supplementary Figure 4 T cell development in bone marrow chimeric mice.
Contour plots show expression of CD4 and CD8 in thymocytes (top) or splenocytes (bottom) and TCRβ vs. CD24 (middle) expression in CD4 SP thymocytes from control chimera prepared from littermate control (Ltm, CD45.2) and wild-type (CD45.1) bone marrow. Experiment and mouse numbers indicated in Figs. 2a, b.

Supplementary Figure 5
Development of Utx-deficient OT-II transgenic T cells.
(a) Contour plots show expression of S1pr1 and Vα2 on mature thymocytes and dot plot shows the frequency of S1pr1 hi CD4 SP among mature thymocytes from OT-II transgenic wild-type (WT) and Utx KO mice as shown in Fig. 3b. (b, c) Contour plots of CD4 and CD8 expression on total thymocytes or splenocytes (b) and absolute numbers of mature thymocytes or spleen Vα2 hi CD4 + T cells (c) from wildtype (WT) and Utx KO OT-II mice.
Data are representative of or combined from five independent experiments. * P < 0.05, *** P < 0.001, **** P < 0.0001 (unpaired t-test). Error bars indicate s.d. 8 Supplementary Figure 6 Impact of Jmjd3 and Utx on thymocyte development (a) Real-time RT-PCR analysis of S1pr1 and Klf2 expression in CD24 lo CD4 SP thymocytes from indicated mice. Bar graphs show fold change relative to wild type values (set as 1) after normalization with 18S rRNA. Data is from three independently sorted sample sets.
(b) Contour plots show expression of TCRβ and S1pr1 on TCR hi CD24 lo CD4 SP thymocytes of indicated mice.
(c) Contour plots show expression of S1pr1 and Vα2 on Vα2 hi CD24 lo CD8 SP thymocytes among mature thymocytes from P14 transgenic wild-type (WT) and dKO mice.
(d) Vα2 expression in wild-type (top) and dKO (bottom) thymocytes carrying the P14 TCR transgene (plain line) or in DP thymocytes from non-transgenic wild-type mice (grey shaded). Brackets define the Vα2 hi subset analyzed for CD4 and CD8 expression (contour plots, right). Numbers on plot indicate percent of cells within nearby box or bracket.
(e) Contour plots of CD4 and CD8 expression on total thymocytes (left) or splenocytes (right), and Vα2 vs. CD24 expression on CD8 SP thymocytes (middle) from wild-type (WT) and dKO mice carrying the P14 transgene. Numbers of mice for each genotype indicated in (f).
(f) Dot plots show absolute numbers of total thymocytes, mature Vα2 hi CD24 lo CD8 SP thymocytes, total splenocytes or CD8 + T cells from mice in (e). Each symbol represents one individual mouse.
Relative expression of S1pr1, Klf2 and Nr4a1 mRNA in Vα2 hi CD24 lo CD4 SP thymocytes, from WT OTII mice, after overnight culture with or without anti-CD3. Bar graphs show fold change relative to unstimulated values (set as 1 for each gene) after normalization with 18S rRNA. Data is from three independently sorted sample sets. **P < 0.01, and ***P < 0.001 (unpaired t-test). Error bars indicate s.d.

Supplementary Figure 8
Impact of Jmjd3 and Utx on H3K27 trimethylation and gene expression.
(a) Scatter plot show H3K27Me3 signals on the 7730 'peak' gene set. Each symbol depicts signal (log 2 value) at a given promoter in wild-type (x-axis) vs. dKO (y-axis) mature CD4 SP OT-II thymocytes. Dotted lines indicate 2-fold changes. Filled red and blue red symbols depict over-and under-decorated genes defined in Fig. 5b. (b, d) IGV browser tracks show the distribution (normalized sequence reads) of H3K27Me3 ChIPseq signals at indicated loci in wild-type and dKO OT-II transgenic CD4 SP thymocytes (top two tracks), and indicated thymocyte subsets of mice expressing a diverse endogenous TCR repertoire. Arrows indicate transcription boundaries.
(c) Scatter plot show H3K27Me3 signals, as in (a) on genes within the Hoxa-d clusters in wild-type (x-axis) vs. dKO (y-axis) mature CD4 SP OT-II thymocytes.
(e) Heat map indicates ChIPseq signals (log 2 values, color-coding scale at bottom) on over-and under-decorated promoters (defined in Fig. 5b and ranked in the same order as in Fig. 5c) in mature CD4 SP thymocyte from indicated mice expressing an endogenous diverse TCR repertoire. Each lane represents a separate ChIPseq experiment.
(f) IGV browser tracks show the distribution (normalized sequence reads) of H3K27Me3 ChIPseq signals at S1pr1, Klf2, Zbtb7b and Ccnd2 in thymocytes of mice expressing a diverse endogenous TCR repertoire. Arrows indicate transcription boundaries.

Supplementary Figure 9
Demethylase-independent functions of Jmjd3 and Utx.