The chromatin regulator Mll1 supports T follicular helper cell differentiation by controlling expression of Bcl6, LEF-1, and TCF-1

Abstract

T follicular helper (T_FH) cells are essential for developing protective antibody responses following vaccination. Greater understanding of the genetic program leading to T_FH differentiation is needed. Chromatin modifications are central in the control of gene expression. However, detailed knowledge of how chromatin regulators (CRs) regulate differentiation of T_FH cells is limited. We screened a large shRNA library targeting all known CRs in mouse and identified the histone methyltransferase Mll1 as a positive regulator of T_FH differentiation. Loss of Mll1 expression reduced formation of T_FH cells following acute viral infection or protein immunization. In addition, expression of the T_FH lineage-defining transcription factor Bcl6 was reduced in the absence of Mll1. Transcriptomics analysis identified Lef1 and Tcf7 as genes dependent on Mll1 for their expression, which provides one mechanism for the regulation of T_FH differentiation by Mll1. Taken together, CRs such as Mll1 substantially influence T_FH differentiation.

INTRODUCTION

High-affinity antibodies are central to immune protection elicited following vaccination. Germinal center (GCs) are essential for the production of such antibodies (1). T follicular helper CD4 T cells (T_FH) cells provide essential help signals to antigen-specific B cells for the formation of germinal center B cells (B_GC) that culminate in the formation of long-lived antibody-secreting plasma cells and memory B cells (1). Differentiation of CD4 T cells into T_FH cells is initiated upon interaction of activated CD4 T cell with antigen-presenting dendritic cells (1). Expression of the transcriptional repressor Bcl6 in CD4 T cells induces the T_FH transcriptional program. Bcl6 directs T_FH differentiation by suppressing alternative T helper cell fates and by promoting expression of gene products associated with T_FH cells and function (1–5). Other transcription factors such as LEF-1/TCF-1 (Lef1/Tcf7) (6–8), Batf (9, 10), Maf (11, 12) or E proteins (13, 14) play critical roles in supporting T_FH differentiation. These transcription factors are balanced by inhibitors of T_FH differentiation such as Blimp-1 (5) and Klf2 (2, 15).

Differentiation of CD4 T cells into a wide variety of helper subsets is tightly controlled by chromatin modifications (16). Transcription factors driving T_FH differentiation have been thoroughly characterized, but the role of epigenetics in controlling this process remains mostly unexplored. T_FH cells possess cell type-specific enhancers identified by monomethylation of histone 3 lysine 4 (H3K4Me) and acetylation at lysine 27 of histone 3 (H3K27Ac) (17, 18). Additionally, T_FH cells possess unique patterns of chromatin accessibility (2, 17). All of the above suggest an epigenetic program specific for T_FH.

Bcl6 plays a central role in establishing the T_FH-specific epigenetic program. The middle domain of Bcl6, which recruits the corepressor MTA3, is necessary for full activity of Bcl6 and proper formation of T_FH cells (19). Additionally, depletion of the corepressor NCoR1, or deletion of BCoR, impairs T_FH formation (2, 20). TCF-1 recruits the histone methyltransferase EZH2 to activate Bcl6 expression (21), and absence of EZH2 abrogates T_FH differentiation (21, 22). The histone methyltransferase Nsd2 influences T_FH differentiation by promoting expression of Bcl6 in the early stages of CD4 T cell activation (23). Bcl6, of course, is not the only T_FH-associated target of CRs. The histone demethylase UTX impacts T_FH differentiation by regulating Il6ra expression (24). We sought to identify novel CRs essential for T_FH formation. To this end, we adapted our in vivo shRNA-based screening approach (25) to survey a library of shRNAmir targeting all known CRs in mouse. We identified Mll1 (Mixed Lineage Leukemia 1) as a positive regulator of T_FH differentiation and function.

MATERIALS AND METHOD

Mice, LCMV infection and protein immunization.

SMARTA mice (transgenic expression of an I-A^b-restricted TCR specific for LCMV glycoprotein amino acids 66–77) (26) and CD45.1⁺ mice were on a full C57BL/6 background and bred in-house. C57BL/6 and Cas9^Tg mice (JAX stock #028555) were purchased from the Jackson laboratory. Both male and female mice were used throughout the study, with sex and age matched T cell donors and recipients. All mice were maintained in specific-pathogen-free facilities and used according to protocols approved by the animal care and use committees of LJI. Mice were infected via the intraperitoneal route with 2×10⁵, 4×10⁵ or 5×10⁵ plaque-forming units of LCMV-Armstrong for days 6, 4 or 3 analysis, respectively. 10 μg of KLH conjugated to gp_66–77 peptide of LCMV mixed with 2.5 μg of SMNP (27) in 100 μL PBS was injected at the base of tail. 1 mg BrdU in PBS was injected via intraperitoneal route 16 hours before analysis.

In vivo shRNA screen.

We included a set of 42 control shRNAmir-RV (Supplemental Table 1). 322 CRs genes expressed in mouse are targeted by 3 or 4 shRNAmir each for a total of 1230 shRNAmir-RV in the library (Supplemental Table 1). The initial screen was performed twice as previously described (25). Subsequent screens were modified as follows. Equal amounts of every shRNAmir were pooled to create 22 tubes. PLAT-E cells were transfected with 2 μg of pooled shRNAmir (1 pool/well) and 1 μg pCL-Eco plasmids. Anti-CD3/CD28 activated SMARTA CD4 T cells were transduced once with RV supernatant. 5×10⁵ sorted cells were transferred into 20 recipient mice by intravenous injection. An aliquot of 1.1–1.5×10⁶ sorted cells was saved for RNA isolation (Input). Mice were infected with 2×10⁶ PFU LCMV-Armstrong 4 days after transfer.

3 days after infection, spleens were isolated and pooled. CD45.1⁺ SMARTA cells were pre-enriched and sorted into CXCR5⁺SLAM^lo T_FH and CXCR5⁻SLAM⁺ T_H1 SMARTA cells. 1.7×10⁵ T_FH and 1.6×10⁶ T_H1 cells were sorted for the first screen. 7.2×10⁵ T_FH and 3.2×10⁶ T_H1 cells were sorted for the second screen. Total RNA was isolated using mirVana miRNA Isolation Kit (Invitrogen). RNA was quantified with High Sensitivity RNA ScreenTape kit and TapeStation (Agilent). 70 ng of RNA was used to synthesize cDNA using random oligonucleotides in triplicate reactions that were subsequently pooled.

Sequencing the shRNAmir library.

The amount of shRNAmir-derived cDNA template in each sample preparation was quantified using qPCR and a standard curve prepared from known molar amounts of gel-purified shRNAmir amplicons. To build shRNAmir amplicon libraries capturing passenger strand sequences, an amount of cDNA template in each sample resulting in equivalent amplification to previously analyzed libraries amplified from 100 ng of total genomic DNA template from transduced T cells was amplified using primers including Illumina P5 adapter (5’-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTCTCGAGAAGGTATATTGCT-3’) and P7 (5’-CAAGCAGAAGACGGCATACGAGATNNNNNNGTGACTGGAGTTCAGACGTGTGCTCTTCGATCTACATCTGTGGCTTCACTA-3’) adapter sequences. The miR30 (P5) and shRNAmir-loop (P7) annealing sequences are underlined and the custom barcode is indicated (Ns). Sequencing libraries were amplified in triplicate reactions using Phusion High-Fidelity DNA polymerase in 1X HF buffer for 24 PCR cycles. Replicate reactions were pooled, purified with 1.6X AMPure XP beads (Beckman Coulter), and quantified with Qubit DNA assay and Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.). The individually barcoded libraries were pooled in equimolar ratios, mixed with separate base-balanced libraries to increase diversity and sequenced on the NextSeq 500 at 1.8 pM concentration with paired-end 75bp reads.

A FASTA formatted reference database comprising 97mer sequences of all analyzed shRNAmirs in the library was built. Raw sequencing reads were merged using FLASH and the merged reads were compared to the reference database using nucleotide BLAST. Only the top alignment for each read was retained and the number of times each reference sequence was aligned in each sample library was counted and reported. Normalized reads for all shRNAmir targeting the same gene were averaged to create a single gene read number and gene log2 T_FH/T_H1 ratios were calculated.

Retroviral vectors, transductions and cell transfers.

pMIG, Lef1, Tcf7-p45 and Bcl6 pMIG were described previously (6, 7, 19). Guide RNAs were selected by CHOP-CHOP (https://chopchop.cbu.uib.no) and cloned into the BbsI site of a modified LMPd-Ametrine vector. Transduction and transfer of CD4 T cells were performed as previously detailed (25). Transferred cells were allowed to rest in host mice for 3–4 days before infection or immunization. 2×10⁴, 5×10⁴ or 4×10⁵ transduced CD4⁺ T cells were transferred for day 6, 4 or 3 analysis, respectively. For protein immunization, 2×10⁵ transduced cells were transferred. For T_H1 cultures, cells were cultured in the presence of 10 ng/mL IL-12 (Peprotech) and 10 μg/mL anti-IL-4 (R&D Systems).

Flow cytometry and antibodies.

Single-cell suspensions of spleen or draining inguinal lymph nodes were prepared by standard gentle mechanical disruption. Surface staining for flow cytometry was done with monoclonal antibodies against CD4 (RM4–5, 1:400), CD8 (53–6.7, 1:400), CD45.1 (A20, 1:400), B220 (RA3–6B2, 1:400) (eBiosciences); PSGL-1 (2PH1, 1:400), (from BD Biosciences); SLAM (TC15–12F12.2, 1:400), CD25 (PC61, 1:400), CD44 (IM7, 1:400) (BioLegend). CXCR5 staining was done using biotinylated anti-CXCR5 (SPRCL5, eBioscience), followed by PE-Cy7-, BV421- or APC-labeled streptavidin (BioLegend). Intranuclear staining was performed with a monoclonal antibody to Bcl6 (K112–91, BD Biosciences), TCF1 (C63D9), LEF-1 (C12A5, Cell Signaling) or Blimp-1 (5E7, BioLegend), and the Intracellular Fixation & Permeabilization Buffer Set (Invitrogen). For measurement of cytokines, cells from spleen were cultured with 10 μg/ml gp₆₆ peptide and Brefeldin A. Intracellular staining for cytokines was performed with monoclonal antibody to IFN-γ (XMG1.2), CD40L (MR1), IL-2 (JES6–5H4, Invitrogen) and recombinant mouse IL-21 receptor Fc (R&D Systems), followed by anti-human IgG (Invitrogen), using the Fixation/Permeabilization buffer kit (BD Biosciences). BrdU staining was performed with the BrdU Flow Kit (BD Biosciences). Stained cells were analyzed using LSRII or Celesta (BD Biosciences) and FlowJo software (TreeStar). Anti-TBP (ab63766, Abcam) and anti-Mll1 (14689, Cell Signaling Technologies) were used in western blots.

RNA-Seq.

CD45.1⁺ gRNA⁺ Cas9^Tg SMARTA cells were pre-enriched from spleens pooled from 4 mice per group. 5,000–150,000 CXCR5⁺SLAM^lo T_FH, CXCR5⁻SLAM⁺ T_H1 SMARTA cells, or naïve SMARTA cells were sorted into Trizol LS (Invitrogen). RNA-Seq was performed as previously described (2). 0.5 ng cDNA was used to prepare a standard Nextera XT sequencing library (Nextera XT DNA library preparation kit and index kit; Illumina). Libraries were sequenced using a HiSeq2500 (Illumina) to generate 50-bp single-end reads, generating median of >17 million mapped reads per sample. Three biological replicates were generated for T_FH and T_H1 samples. RNA-Seq was analyzed as previously described (2) except for the following modifications. The reads were aligned with the mm10 reference genome using STAR (v 2.6.1) and the RefSeq gene annotation downloaded from the UCSC Genome Bioinformatics site. Read counts to each genomic feature were obtained with the featureCounts (v 1.6.5) using the default option along with a minimum quality cut off (Phred > 10). We considered genes differentially expressed between two groups of samples when the DESeq2 analysis resulted in an adjusted p-value of <0.05 and the difference in gene expression was 1.4-fold. TPMs were calculated. Gene Set Enrichment Analysis (GSEA) was run on gene lists with GSEA v3.0 (Broad Institute, Inc.). Gene signatures from in vitro T_H2, T_REG, Bcl6-bound gene list from human tonsil GC-T_FH (18), Blimp-1-bound genes from activated CD8 T cells (28), or from in vitro-differentiated T_H17 (29) and genes signatures from T_FH, T_H1 and Tcf7^−/−Lef1^−/− GC-T_FH isolated after LCMV-Arm infection and from in vitro-differentiated Lef1⁺-RV T_H1 (6) were previously published.

Statistical Methods.

Statistical tests were performed using Prism 8.0 (GraphPad). Significance was determined by unpaired Student’s t-test with a 95% confidence interval, by one-way ANOVA with Tukey’s multiple comparisons test or by two-way ANOVA with Sidak’s multiple comparisons test. *p<0.05, **p<0.01, ***p<0.001 and ****p<0.0001.

RESULTS

shRNA screen for chromatin regulators of early T_FH differentiation

To discover novel CRs regulating T_FH differentiation, we generated an shRNAmir retroviral vector (RV) library targeting all known mouse CRs, 322 genes, each targeted by 3–4 shRNAmir for a total of 1230 shRNAmir-RVs. (Fig. 1a). Briefly, LCMV gp66–77-specific TCR transgenic SMARTA CD4 T cells were transduced once with shRNAmir-RV-containing supernatant. shRNAmir-RV⁺ SMARTA CD4 T cells were sorted based on Ametrine marker expression and adoptively transferred into recipient C57BL/6 mice (Fig. 1b). Total RNA was isolated from an aliquot of CD4 T cells for the Input sample. Recipient mice were infected with LCMV-Arm and, three days after infection, SMARTA CD4 T cells were sorted into CXCR5⁺SLAM^lo T_FH and CXCR5⁻SLAM⁺ T_H1 populations (Fig. 1c). cDNA sequencing libraries were prepared for the SMARTA Input, T_FH and T_H1 samples. Sequenced cDNA was aligned to a list of all shRNAmir sequences in the library to determine read counts for every shRNAmir in the library. The screen was performed twice in independent experiments.

Figure 1: An in vivo RNAi screen identifies Mll1 as a novel regulator of T_FH differentiation.

(A) Scheme for screening the shRNAmir library targeting CRs. (B) Transduction efficiency of SMARTA CD4 T cells infected with pooled shRNAmir-RV from the CR library. (C) Sorting strategy of shRNAmir⁺ SMARTA CXCR5⁺SLAM^lo T_FH and CXCR5⁻SLAM⁺ T_H1 cells 3 days after LCMV infection. (D) Construct T_FH/T_H1 ratio Z scores for all shRNAmir in the library in two independent experiments. (E) Gene T_FH/T_H1 ratio Z scores for all genes in the library in two independent experiments.

We calculated Z score values of the T_FH/T_H1 ratio for every shRNAmir in the library. Direct comparison of Z scores per shRNAmir construct was not informative (Fig. 1d). It is possible that differences in shRNA knockdown efficacy was the source of the variability. We averaged the reads numbers for each shRNA targeting the same gene to create a single Z score for each targeted gene. We hypothesized that calculating a single Z score value for each targeted gene would reduce variability and be more informative. This analysis method showed improved reproducibility between experiments (Fig. 1e, Supplemental Table 1). We assessed the gene Z score distribution of known positive and negative regulators of T_FH differentiation. Bcl6 and Itch had negative Z score values in both experiments, indicating their shRNAmir constructs were depleted from the T_FH population (Fig. 1e), consistent with their known roles as positive T_FH regulators (3–5, 30). Conversely, Prdm1, Tbx21 and Id2 had positive Z score values, indicative of enrichment in T_FH cells (Fig. 1e), consistent with their known roles in directly inhibiting T_FH differentiation or promoting T_H1 cells (5, 13, 31). We then filtered the gene Z score data for novel regulators of T_FH differentiation by ranking genes depleted from T_FH cells. The gene coding for the histone methyltransferase Mll1 was one of the most depleted genes from T_FH cells, with Z score values of −1.4 and −4.8, suggesting Mll1 is a positive regulator of T_FH differentiation (Fig. 1e). Mll1 methylates lysine 4 of histone 3 (H3K4), a mark associated with active transcription, and assembles with numerous regulatory proteins such as RbBP5, Wdr5, Ash2l, Dpy30 and Menin-1 (32). Mll1 regulates the expression of T_H1 and T_H2 cytokines in CD4 T cells (33, 34) but a role for Mll1 in T_FH was unknown. Our shRNA library included shRNA for the Ezh2 and NCoR1, two CR known to regulate Tfh differentiation (2, 23). NCoR1 had Z score values of −0.50 and −2.46 while Ezh2 had Z score values of −1.29 and −0.91 suggesting our screen can detect CRs known to regulate T_FH differentiation.

Loss of Mll1 impairs early T_FH differentiation

In published RNA-Seq data (2, 6), we observed that Mll1 is expressed at similar levels between naive CD4, T_FH and T_H1 cells (Supplemental Fig. 1a). Two shRNAmirs that efficiently inhibited Mll1 expression in vitro were tested individually for their effect on T_FH differentiation in vivo (Supplemental Fig. 1b). We transduced SMARTA CD4 T cells with shMll1-RV or shCd8-RV, transferred the SMARTA CD4 T cells into C57BL/6 mice, and analyzed T_FH differentiation 3 days after infection with LCMV-Arm. shMll1⁺ SMARTA CD4 T cells exhibited a significantly smaller early T_FH cell population (CXCR5⁺CD25⁻ or CXCR5⁺Bcl6⁺) when compared to shCd8⁺ SMARTA cells (Fig. 2a, b, Supplemental Fig. 1c, d). shRNAmirs targeting Mll1 had a small but significant impact on the accumulation of SMARTA CD4 T cells in vivo (Fig. 2c, Supplemental Fig. 1e, f). Importantly, we observed reduced expression of Bcl6 in Mll1-deficient CXCR5⁺CD25⁻ early T_FH cells (Fig. 2d). These results confirmed the identification of Mll1 as a novel positive regulator of T_FH differentiation after LCMV infection.

Figure 2: Mll1 is required for proper differentiation of early T_FH after LCMV infection.

shRNA⁺ SMARTA (A-D) or gRNA⁺ Cas9^Tg SMARTA (E-H) CD4 T cells were transferred into C57BL/6 mice and analyzed 3 days after infection with LCMV-Arm. (A) CXCR5⁺CD25⁻ T_FH and CXCR5⁻CD25⁺ T_H1 and (B) CXCR5⁺Bcl6⁺ T_FH in SMARTA CD4 T cells. (C) Quantitation of SMARTA CD4 T cells. (D) Bcl6 expression in CXCR5⁺CD25⁻ SMARTA CD4 T cells. Numbers in the histogram indicate geometric MFI values. (E) CXCR5⁺CD25⁻ T_FH and CXCR5⁻CD25⁺ T_H1 and (F) CXCR5⁺Bcl6⁺ T_FH in Cas^Tg SMARTA CD4 T cells. (G) Quantitation of Cas9^Tg SMARTA CD4 T cells. (H) Bcl6 expression in CXCR5⁺CD25⁻ Cas9^Tg SMARTA CD4 T cells. Data shown are from one experiment representative of 5 (A-D) or 3 (E-H) independent experiments with 3–4 mice per group. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001 (unpaired two-tailed Student’s t-test).

We next sought to confirm the role of Mll1 using an independent experimental approach. Cas9^Tg SMARTA CD4 T cells transduced with a control gCd8-RV or gMll1-RV (Supplemental Fig. 1g) were transferred into C57BL/6 mice and analyzed 3 days after LCMV-Arm infection. gMll1⁺ Cas9^Tg SMARTA CD4 T cells had substantially reduced potential to differentiate into CXCR5⁺CD25⁻ or CXCR5⁺Bcl6⁺ T_FH cells (Fig. 2e, f, Supplemental Fig. 1h, i). Accumulation of gMll1⁺ Cas9^Tg SMARTA CD4 T cells was normal after infection (Fig. 2g, Supplemental Fig. 1j). Bcl6 expression was also significantly reduced in gMll1⁺ CXCR5⁺CD25⁻ cells (Fig. 2h). Thus, disruption of Mll1 expression by gRNA phenocopied shRNA-mediated knockdown of Mll1, establishing Mll1 as a novel positive regulator of T_FH differentiation.

Reduced peak T_FH differentiation in Mll1-deficient CD4 T cells

To assess if Mll1 is dispensable for later stages of T_FH differentiation, we analyzed T_FH differentiation of Mll1-deficient CD4 T cells 6 days after infection. Compared to control shCd8⁺ cells, shMll1⁺ SMARTA CD4 T cells showed impaired formation of T_FH cells (CXCR5⁺SLAM^lo, Fig. 3a). We next analyzed formation of GC-T_FH. The percentages of CXCR5⁺Bcl6⁺ GC-T_FH cells in shMll1⁺ SMARTA CD4 T cells were comparable to those of control shCd8⁺ SMARTA (Fig. 3b). Unlike what we observed at 3 days post infection, the reduction in T_FH differentiation was coupled to a significant decrease in the accumulation of shMll1⁺ SMARTA CD4 T cells by day 6 (Fig. 3c, Supplemental Fig. 1k).

Figure 3: Mll1 is required for proper differentiation of T_FH.

shRNA⁺ SMARTA (A-C) or gRNA⁺ Cas9^Tg SMARTA (D-H) CD4 T cells were transferred into C57BL/6 mice and analyzed 6 days after infection with LCMV-Arm (A-F) or 8 days after immunization with GP-KLH mixed with SMNP (G-H). (A) CXCR5⁺SLAM^lo T_FH or (B) CXCR5⁻Bcl6⁻ T_H1, CXCR5⁺Bcl6⁻ T_FH and CXCR5⁺Bcl6⁺ GC-T_FH in SMARTA CD4 T cells. (C) Quantitation of SMARTA CD4 T cells. (D) CXCR5⁺SLAM^lo T_FH or (E) CXCR5⁻Bcl6⁻ T_H1, CXCR5⁺Bcl6⁻ T_FH and CXCR5⁺Bcl6⁺ GC-T_FH in Cas9^Tg SMARTA CD4 T cells. (F) Quantitation of Cas9^Tg SMARTA CD4 T cells. (G) CXCR5⁻PSGL-1⁺ T_H1 and CXCR5⁺PSGL-1^lo T_FH in Cas9^Tg SMARTA CD4 T cells. (H) Quantitation of Cas9^Tg SMARTA CD4 T cells. (I-K) shRNA⁺ SMARTA CD4 T cells were transferred into C57BL/6 mice and mice were infected with LCMV-Arm. Mice were injected with BrdU 3 days after infection and analyzed 16 hours later. (I) BrdU⁺ cells in SMARTA CD4 T cells. (J) CXCR5⁺SLAM^lo T_FH SMARTA CD4 T cells and (K) CXCR5⁻SLAM⁺ T_H1 SMARTA CD4 T cells. Data shown are from one experiment representative of 5 (A-C), 1 (D-G) or 2 (I-K) independent experiments with 4 mice per group. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001 (unpaired two-tailed Student’s t-test).

gRNA-mediated disruption of Mll1 also significantly reduced CXCR5⁺SLAM^lo T_FH differentiation 6 days after LCMV infection (Fig. 3d). gMll1⁺ Cas9^Tg SMARTA CXCR5⁺Bcl6⁺ GC-T_FH cells were reduced compared to gCd8⁺ (Fig. 3e). These results suggest that Mll1 mostly regulates T_FH differentiation with less impact on GC-T_FH differentiation because a reduction in CXCR5⁺Bcl6⁺ GC-T_FH was observed only in gMll1⁺ Cas9^Tg SMARTA CD4 T cells but not in shMll1⁺ SMARTA CD4 T cells. gMll1⁺ Cas9^Tg SMARTA CD4 T cells also accumulated significantly less than gCd8⁺ Cas9^Tg SMARTA CD4 T cells (Fig. 3f). These results showed that Mll1 is a positive regulator of T_FH differentiation after an acute viral infection.

We next tested Mll1 disruption in the context of protein immunization. C57BL/6 mice that received gCd8⁺ or gMll1⁺ Cas9^Tg SMARTA CD4 T cells were immunized subcutaneously with KLH conjugated to GP_66–77 peptide from LCMV (GP-KLH) mixed with the SMNP adjuvant (saponin monophosphoryl lipid A nanoparticle). Draining lymph nodes were harvested for analysis 8 days later. Differentiation of gMll1⁺ Cas9^Tg SMARTA CD4 T cells into CXCR5⁺PSGL-1^lo T_FH cell was reduced following immunization (Fig. 3g) while accumulation of Mll1-deficient SMARTA CD4 T cells was not compromised (Fig. 3h). These experiments demonstrate a role for Mll1 in T_FH cell differentiation in different contexts.

We tested if reduced proliferation of Mll1-deficient CD4 T cells was causing their reduced accumulation after LCMV infection. Four days after infection, total shMll1⁺ SMARTA CD4 T cells showed significantly reduced incorporation of BrdU when compared to shCd8⁺ SMARTA cells (Fig. 3i). The decrease in BrdU incorporation was observed for both CXCR5⁺SLAM^lo T_FH and CXCR5⁻SLAM⁺ T_H1 cells (Fig. 3j, k). These results indicate that Mll1 is required for optimal proliferation of CD4 T cells following infection and suggest that the reduced T_FH differentiation of shMll1⁺ SMARTA CD4 T cells was not caused by a selective proliferation defect of T_FH cells. These experiments demonstrated the importance for Mll1 in supporting T_FH differentiation.

Mll1 is required for proper IL-21 expression

To test the functionality of Mll1-deficient T_FH cells, we assessed cytokine production in shCd8⁺ and shMll1⁺ SMARTA CD4 T cells 6 days after infection. A substantial fraction of IL-21-producing cells were observed in control shCd8⁺ SMARTA, but IL-21 expression was greatly reduced in shMll1⁺ SMARTA (Fig. 4a, Supplemental Fig. 2a). Expression of CD40L or IFN-γ was not affected by loss of Mll1 (Fig. 4b, Supplemental Fig. 2b). Production of IL-2, whose gene is immediately adjacent to Il21, was mostly unaffected by Mll1-deficiency (Supplemental Fig. 2c). IL-21 expression was similarly disrupted in gMll1⁺ Cas9^Tg SMARTA CD4 T cells (Fig. 4c). A small but significant reduction in CD40L was observed while IFN-γ production was not significantly impacted (Fig. 4d, Supplemental Fig. 2d). IFN-γ expression in T_H1 cells in response to LCMV infection was also reduced after gRNA-mediated abrogation of Mll1 expression (Supplemental Fig. 2e, f) similar to what was observed in Mll1^+/− CD4 T cells responding to tuberculosis antigens (34).

Figure 4: Mll1 is required for production of IL-21.

shRNA⁺ SMARTA (A-B, E) or gRNA⁺ Cas9^Tg SMARTA (C-D) CD4 T cells were transferred into C57BL/6 mice. 6 days after infection with LCMV-Arm, splenocytes were restimulated with LCMV gp_66–77 peptide. Expression of IL-21 (A) and CD40L (B) in SMARTA CD4 T cells or T_FH and T_H1 SMARTA CD4 T cells (E). Expression of IL-21 (C) and CD40L (D) in Cas9^Tg SMARTA CD4 T cells. (F) shRNA⁺ CD4 T cells were cultured in T_H1 conditions. Expression of IL-21 in shRNA⁺ CD4 T cells restimulated with anti-CD3. (G) gRNA⁺ Cas9^Tg SMARTA CD4 T cells were cultured in T_H1 conditions. Expression of IL-21 in gRNA⁺ Cas9^Tg SMARTA CD4 T cells restimulated with anti-CD3. Data shown are from one experiment representative of 5 (A-B, E), 3 (C-D) or 2 (F-G) independent experiments with 3–4 mice per group. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001 (unpaired two-tailed Student’s t-test).

Notably, we observed reduction of IL-21 production by both T_H1 and T_FH cells (Fig. 4e), suggesting that Mll1 regulates IL-21 production independently of T_FH differentiation. We next tested if Mll1 is required for IL-21 production in a system independent of T_FH differentiation. In vitro shMll1⁺ T_H1 cells and gMll1⁺ Cas9^Tg T_H1 cells exhibited impaired IL-21 expression (Fig. 4f, g). Impacts on IFN-γ in vitro were variable with shMll1⁺ T_H1 producing normal levels of IFN-γ while IFN-γ production was reduced in gMll1⁺ T_H1 (Supplemental Fig. 2g, h). Altogether, these results demonstrate that IL-21 production by CD4 T cells depends on Mll1.

Mll1-deficient T_FH cells have a dysregulated T_H1 genetic program

To gain further understanding of the function of Mll1, we performed a transcriptomics analysis of Mll1-deficient T_FH and T_H1 cells (Fig. 5a). Mll1-deficient T_FH cells had expression changes of a select number of genes in comparison to control T_FH cells (89 upregulated and 29 downregulated genes, |FC| > 1.4, padj < 0.05, Fig. 5b, c). In T_H1 cells, loss of Mll1 resulted in the downregulation of 91 and the upregulation of 72 genes (Fig. 5b, c). We next asked how different gene expression signatures associated with CD4 T cell subsets were affected by loss of MLL1 expression. We performed gene set enrichment analysis of T_FH, T_H1, T_H17, T_H2 and T_reg gene signatures against control and Mll1-deficient T_H1 and T_FH populations. In these analyses, a negative normalized enrichment score (NES) denotes enrichment of the indicated gene signature in the gMll1⁺ populations while a positive NES indicates enrichment of the specified gene signature in gCd8⁺ cells. T_FH-associated gene expression was depleted from Mll1-deficient T_FH cells (Fig. 5d). Conversely, a T_H1 signature was enriched in Mll1-deficient T_FH cells (Fig. 5d). Enrichment of a T_H1 gene signature was also observed in Mll1-deficient T_H1 cells (Fig. 5d). Genetic signatures associated with T_H2, T_H17 and T_reg cells were slightly enriched in Mll1-deficient T_FH cells (Fig. 5d). In a separate analysis, we observed that wild-type T_H1 cells were enriched for genes upregulated in Mll1-deficient cells (Fig. 5e). It was possible that Mll1 controls T_FH differentiation by acting in concert with Bcl6 or Blimp-1. We tested the enrichment of Bcl6-bound and Blimp-1 genes in control and Mll1-deficient T_H1 and T_FH populations but found no strong evidence of Mll1 involvement in the regulation of Bcl6-bound genes (Fig. 5f). Expression of Blimp-1-bound genes was decreased in Mll1-deficient TH1 cells suggesting dysregulation of Blimp-1 activity in the absence of MLL1 (Fig. 5f). Together, these GSEA analyses suggest that in the absence of Mll1, T_FH cells gain expression of genes associated with the T_H1 program.

Figure 5: Transcriptomics analysis of Mll1-deficient T_FH.

(A) gRNA⁺ Cas^Tg SMARTA CD4 T cells were transferred into C57BL/6 mice and sorted into CXCR5⁺CD25⁻ T_FH and CXCR5⁻CD25⁺ T_H1 3 days after infection with LCMV-Arm. (B) Quantitation of differentially expressed genes in gCd8⁺ and gMll1⁺ T_FH and T_H1. The numbers of DEGs are indicated in the columns. (C) Heatmaps of differentially expressed genes in gCd8 and gMll1 T_FH and T_H1. (D, F) Bubblegum plots of GSEA analyses of genes differentially expressed between gCd8⁺ and gMll1⁺ T_FH or T_H1 against the indicated gene signatures. (E) GSEA of genes down- or up-regulated in gMll1⁺ T_FH compared to genes differentially expressed in gCd8⁺ T_FH and T_H1.

Dysregulated expression of LEF-1 and TCF-1 in Mll1-deficient T_FH

We searched a curated list of known T_FH regulators for genes with dysregulated expression in Mll1-deficient T_FH cells. Mll1-deficient T_FH expressed significantly less Lef1 and Tcf7, (Fig. 6a, b). We also observed that Prdm1, the gene encoding Blimp-1 had higher expression in Mll1-deficient T_FH in comparison to control T_FH (Fig. 6a). Id2 and Il2ra, two genes associated with suppression of T_FH fate (13, 35, 36), were also upregulated in Mll1-deficient T_FH cells (Fig. 6a). Interestingly, Lef1 also showed reduced expression in Mll1-deficient T_H1 cells (Fig. 6a, b).

Figure 6: Mll1 regulates expression of LEF-1 and TCF-1 in T_FH cells.

(A) Volcano plots of gene expression in gCd8⁺ and gMll1⁺ T_FH or T_H1 plotted against adjusted p-value. Genes of interest are indicated. (B) TPM values for Tcf7 and Lef1 in the indicated populations. (C) GSEA of genes down- or up-regulated in Lef-RV⁺ T_H1 compared to genes differentially expressed in gCd8⁺ T_FH and gMll1⁺ T_FH. (D) GSEA analysis of genes down- or up-regulated in Lef1^−/−Tcf7^−/− GC-T_FH compared to genes differentially expressed in gCd8⁺ T_FH and gMll1⁺ T_FH. Selected genes that form the core enrichment signatures are listed. (E-F) shRNA⁺ SMARTA CD4 T cells were transferred into C57BL/6 mice and analyzed 3 days after infection with LCMV-Arm. (E) LEF-1 or (F) TCF-1 expression in CXCR5⁺CD25⁻ SMARTA CD4 T cells. Numbers in the histograms indicate geometric MFI values. Data shown are from one experiment representative of 3 (E-F) independent experiments with 4 mice per group. (B) **p ≤ 0.01, ***p ≤ 0.001 (two-way ANOVA with Sidak’s multiple comparison test), (E-F) *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 (one way ANOVA with Tukey’s multiple comparisons test).

GSEA analysis revealed that genes upregulated following ectopic expression of Lef1 in CD4 T cells are depleted from Mll1-deficient T_FH cells (Fig. 6c). Conversely, genes suppressed when Lef1 is overexpressed are over-represented in T_FH lacking Mll1 (Fig. 6c). In a separate GSEA analysis, Mll1-deficient T_FH cells were enriched for genes upregulated in Lef1^−/−Tcf7^−/− GC-T_FH (Fig. 6d). These findings indicated that LEF-1- and TCF-1-regulated genes were dysregulated, implicating roles for LEF-1 and TCF-1 downstream of Mll1. We measured the expression of LEF-1 and TCF-1 in Mll1-deficient T_FH cells by flow cytometry. Disruption of Mll1 reduced expression of both LEF-1 and TCF-1 in early T_FH cells (CXCR5⁺CD25⁻, Fig. 6e, f, Supplemental Fig. 3a–c) These experiments demonstrate that in the absence of Mll1, LEF-1 and TCF-1 expression is impaired.

We sought to confirm the increased Prdm1 and Il2ra mRNA level in Mll1-deficient T_FH by analyzing Blimp-1 and CD25 expression by flow cytometry. We first validated the staining of Blimp-1 by showing that CXCR5⁻CD25⁺ early T_H1 cells express more Blimp-1 than CXCR5⁺CD25⁻ early T_FH cells (Supplemental Fig. 3d). We observed sporadic increases in Blimp-1 expression in Mll1-deficient CXCR5⁺Bcl6⁺ T_FH 3 days after LCMV infection (Supplemental Fig. 3e). Similarly, CD25 expression was not reproducibly increased in Mll1-deficient T_FH (Supplemental Fig. 3f).

TCF-1 partially rescues T_FH differentiation of Mll1-deficient CD4 T cells

We hypothesized that ectopic expression of Bcl6, TCF-1 or LEF-1 in Mll1-deficient CD4 T cells may rescue T_FH differentiation, as the expression of all three is decreased in the absence of Mll1. To test this hypothesis, we transduced shCd8⁺ and shMll1⁺ SMARTA CD4 T cells with RV-Lef1, RV-Tcf7-p45 (full-length isoform of TCF-1), RV-Bcl6 or control RV expressing GFP alone. RV⁺shRNAmir⁺ SMARTA CD4 T cells were transferred into C57BL/6 recipient mice and T_FH differentiation was analyzed 3 days after LCMV-Arm infection. The expression of LEF-1 in shMll1⁺Lef1-RV⁺ T_FH cells was equivalent to that of shCd8⁺GFP⁺ T_FH cells (Fig. 7a). As expected, forced expression of LEF-1 resulted in increased T_FH differentiation of shCd8⁺ SMARTA CD4 T cells (Fig. 7b). However, ectopic expression of wild-type levels of LEF-1 in shMll1⁺ SMARTA CD4 T cells did not rescue T_FH differentiation (Fig. 7b). The levels of LEF-1 may not have quantitatively compensated for the reduced levels of TCF-1 (Supplemental Fig. 3g), as Tcf7 is more highly expressed than Lef1 at the mRNA level in mouse CD4 T cells (Fig. 6b), supporting a speculation that TCF1 protein is more highly expressed than LEF1.

Figure 7: TCF-1 partially rescues T_FH differentiation of Mll1-deficient CD4 T cells.

SMARTA CD4 T cells transduced with the indicated RV were transferred into C57BL/6 mice and analyzed 3 days after LCMV-Arm infection. (A) LEF-1 or (C) TCF-1 expression in CXCR5⁺CD25⁻ SMARTA CD4 T cells. Numbers in the histogram indicate geometric MFI values. (B,D) CXCR5⁺CD25⁻ T_FH or (E) CXCR5⁺Bcl6⁺ T_FH in SMARTA CD4 T cells. Data shown are from one experiment representative of 2–3 independent experiments with 4 mice per group. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.001 (one way ANOVA with Tukey’s multiple comparisons test).

Expression of Tcf7-p45 in Mll1-deficient CD4 T cells resulted in TCF-1 expression equivalent to control shCd8⁺GFP⁺ cells (Fig. 7c). TCF-1 partially rescued the differentiation of early T_FH differentiation by Mll1-deficient SMARTA CD4 T cells (Fig. 7d), including minimal changes in expression of Bcl6 and LEF-1 (Supplemental Fig. 3h). Overall, the results indicate Mll1 regulates multiple aspects of T_FH differentiation, which can only be partially restored by TCF-1 expression.

If Mll1 primarily functions upstream of Bcl6 in T_FH differentiation, ectopic expression of Bcl6 may bypass the loss of Mll1 and support T_FH differentiation. T_FH differentiation of shMll1⁺ SMARTA CD4 T cells was rescued by ectopic expression of Bcl6 (Fig. 7e) but did not increase LEF-1 and TCF-1 expression in T_FH cells (Supplemental Fig. 3i). Overall, these data indicate that Mll1 regulates T_FH differentiation by regulating expression of a network of transcription factors that includes, but is most likely not limited to, Bcl6, LEF-1 and TCF-1.

DISCUSSION

Here, we identified Mll1 as a critical regulator of T_FH cells. Our work shows that Mll1 is required for optimal T_FH cell differentiation and production of IL-21. We demonstrate that absence of Mll1 results in impaired expression of Bcl6, LEF-1 and TCF-1, each of which are important regulators of T_FH differentiation. Our findings add Mll1 to the CRs that regulate T_FH differentiation (19–24). The results here demonstrate that Mll1 is required for T_FH differentiation. Mll1 is also involved in the expression of T_H1 and T_H2 cytokines (33, 34). Thus, Mll1 has roles in multiple CD4 T cell subsets similar to other CRs and transcription factors. The histone methyltransferase EZH2 is required for both T_FH and T_reg differentiation (21, 22, 37, 38). Similarly, the transcription factors Maf, IRF4 and Batf are all required for both T_H17 and T_FH differentiation (9–12, 39, 40).

Accumulation of Mll1-deficient CD4 T cells was mildly impaired 3 days after LCMV infection but was more dramatically impaired 3 days later. Loss of Mll1 expression resulted in decreased T_FH population at both times analyzed after LCMV infection. It is possible that Mll1 plays a more prominent role in the early stages of T_FH differentiation, at a time when the impact of Mll1-deficiency on CD4 T cell accumulation is marginal. Mll1 impact on CD4 T cell accumulation may become more prominent at later times as evidenced by decreased BrdU labelling observed 4 days after LCMV infection and the substantial loss in Mll1-deficient CD4 T cells 2 days later. Six days after LCMV infection, the decrease in T_FH cell frequencies may be secondary to the decreased accumulation of Mll1-deficient CD4 T cells. Nevertheless, in a protein immunization system, T_FH cells were decreased 8 days after immunization even when accumulation of CD4 T cells was not impaired in the absence of Mll1.

Our study provides a mechanistic explanation for the defective T_FH differentiation in the absence of Mll1. First and foremost, expression of the critical T_FH regulator Bcl6 is reduced in Mll1-deficient CD4 T cells. A small reduction in Bcl6 expression is sufficient to negatively impact T_FH differentiation, as seen in CD4 T cells lacking a single Bcl6 allele (13). The reduced expression of Bcl6 in the absence of Mll1 most likely impairs suppression of alternative cell fates. Partial loss of Bcl6 is probably not the only cause for defective T_FH differentiation in the absence of Mll1. LEF-1 and TCF-1 have largely redundant functions to each other in CD4 T cells, and reduced expression of both is likely a key contributing factor to the observed phenotype of Mll1-deficient CD4 T cells. Ectopic expression of LEF-1 did not rescue the defective T_FH in Mll1-deficient CD4 T cells. It is plausible that forced expression of LEF-1 was not elevated enough to compensate for both LEF-1 and TCF-1. TCF-1 is expressed at substantially higher levels than LEF-1 and may explain why ectopic expression of TCF-1 partially compensated for Mll1 deficiency. We suggest that Mll1 controls a network of genes that includes Bcl6, LEF-1 and TCF-1 to promote T_FH differentiation. We did not observe any effect of Mll1 on the expression of Bcl6- or Blimp-1-bound genes, suggesting Mll1 does not influence their transcriptional activities. Mll1 may still act on a small subset of such genes and this influence is not detected in our bioinformatics analysis. It is also possible that Mll1 is required for the function of other transcription factors controlling formation of T_FH cells.

In addition to driving T_FH differentiation, Mll1 promotes production of IL-21. Loss of Mll1 does not impact expression of Maf (data not shown), a transcription factor essential for IL-21 production (11, 12). Deletion of Mll1 could limit H3K4 methylation at the Il21 locus reducing its transcription leading to impaired production of IL-21. Mll1 is required for normal hematopoiesis (41, 42), but this function does not require the histone methyltransferase domain of Mll1 (43). It is conceivable that the enzymatic activity of Mll1 is not involved in promoting IL-21 production and T_FH differentiation.

Chromatin modification is a complex process, and absence of a single CR is often compensated by other enzymes. Reduction in T_FH differentiation of Mll1-deficient CD4 T cells may be partial due to compensatory mechanisms by other Mll1 family members (Mll2, Mll3 or Mll5), all of which form complexes with Wdr5, Ash2l and RbBp5 (32). Expression of Mll family members and genes coding for components of the Mll1 complex (Wdr5, Ash2l, Dpy30, RbBp5, Menin-1) was normal in the absence of Mll1 (data not shown). The phenotype of Mll1-deficient CD4 T cells is similar to the incomplete loss of T_FH cells when other CRs are deleted (21–24). Further studies are required to obtain a detailed understanding the role of Mll1 in the control of epigenetic modifications of enhancers and promoters and how this impacts general chromatin accessibility. The roles of members of the Mll1 complex in regulating Mll1 function in the context of T_FH differentiation should also be studied.

Disruption of the interaction between Mll1 and the core complex protein Wdr5 with a small molecule drug inhibits the growth of Mll1 leukemic cells (44). Similarly, a small-molecule inhibitor blocking the interaction between Mll1 and Menin-1 reduces tumor burden in mouse (45). Targeting Mll1 to inhibit T_FH differentiation or function could be a promising therapeutic approach for autoimmune disorders characterized by elevated T_FH activity. Alternatively, targeting the Mll1 complex to enhance its activity may be used in the design of better vaccines.

KEY POINTS.

• MLL1 is a positive regulator of T_FH differentiation

• MLL1 regulates expression of LEF-1, TCF-1 and Bcl6