MLL1 is required for maintenance of intestinal stem cells

Epigenetic mechanisms are gatekeepers for the gene expression patterns that establish and maintain cellular identity in mammalian development, stem cells and adult homeostasis. Amongst many epigenetic marks, methylation of histone 3 lysine 4 (H3K4) is one of the most widely conserved and occupies a central position in gene expression. Mixed lineage leukemia 1 (MLL1/KMT2A) is the founding mammalian H3K4 methyltransferase. It was discovered as the causative mutation in early onset leukemia and subsequently found to be required for the establishment of definitive hematopoiesis and the maintenance of adult hematopoietic stem cells. Despite wide expression, the roles of MLL1 in non-hematopoietic tissues remain largely unexplored. To bypass hematopoietic lethality, we used bone marrow transplantation and conditional mutagenesis to discover that the most overt phenotype in adult Mll1-mutant mice is intestinal failure. MLL1 is expressed in intestinal stem cells (ISCs) and transit amplifying (TA) cells but not in the villus. Loss of MLL1 is accompanied by loss of ISCs and a differentiation bias towards the secretory lineage with increased numbers and enlargement of goblet cells. Expression profiling of sorted ISCs revealed that MLL1 is required to promote expression of several definitive intestinal transcription factors including Pitx1, Pitx2, Foxa1, Gata4, Zfp503 and Onecut2, as well as the H3K27me3 binder, Bahcc1. These results were recapitulated using conditional mutagenesis in intestinal organoids. The stem cell niche in the crypt includes ISCs in close association with Paneth cells. Loss of MLL1 from ISCs promoted transcriptional changes in Paneth cells involving metabolic and stress responses. Here we add ISCs to the MLL1 repertoire and observe that all known functions of MLL1 relate to the properties of somatic stem cells, thereby highlighting the suggestion that MLL1 is a master somatic stem cell regulator.


Introduction
Stem cells are cornerstones of tissue biology, ensuring homeostasis and regeneration in many organs, including epithelial tissues such as skin, intestine and mammary gland [1]. Stem cells are characterized by multipotency, which is the ability to differentiate into a restricted number of defined cell types, and self-renewal, which is the capacity to undergo infinite replicative cycles without losing stem cell identity [2]. The remarkable capacities of stem cells, particularly the restricted specificities of multipotency, rely on interplays between specific transcription factors and distinct epigenetic landscapes. Whereas many transcription factors involved in stem cell maintenance and differentiation have been defined, epigenetic contributions are proving more elusive. For example, the transcription factor hierarchies in the stem cell paradigm, hematopoiesis, have been elegantly dissected [3]. However the contributions of DNA and histone methyltransferases to hematopoiesis are still emerging and indicate both specificities and the deeper complexities of epigenetic regulation [4][5][6][7][8].
Methylation of histone 3 on lysine 4 (H3K4) is one of the most conserved and widespread epigenetic systems [9]. H3K4 is methylated in euchromatic regions, with trimethylated H3K4 (H3K4me3) on nucleosomes surrounding active promoters, H3K4me2 marking transcribed regions and H3K4me1 relating to enhancers and active chromatin in general [10][11][12][13][14]. Mammals have six Set1/Trithorax-related H3K4 methyltransferases (H3K4MTs) that are encoded by three pairs of paralogous sister genes namely, Mll1 (Kmt2a) and Mll2 (Kmt2b), Mll3 (Kmt2c) and Mll4 (Kmt2d), Setd1a (Kmt2f) and Setd1b (Kmt2g). Each of the six H3K4MTs resides in their own, large, protein complex. However all six complexes are based on a four membered scaffold termed WRAD for the subunits WDR5, RBBP5, ASH2L and DPY30 [15] or sometimes called COMPASS [9]. Functional differences between the six complexes potentially arise from the presence of additional subunits, which are usually shared by paralogous pairs or sometimes uniquely found in one of the six complexes. For example, MENIN is a subunit of both MLL1 and 2 complexes [16] but so far, no subunit unique to either paralogous complex has been described.
Due to the high turnover and hierarchical architecture of the intestinal epithelium, intestinal stem cells (ISCs) have become an adult stem cell paradigm. ISCs have been identified as either actively cycling crypt base columnar cells (CBCs) or quiescent label-retaining cells (LRCs) located at the +4 position from the crypt base [29]. LGR5 (Leucine-rich repeat-containing G protein-coupled receptor 5) is the characteristic marker for the CBC class of ISCs [30], which generates the transit amplifying (TA) daughter cells. Proliferation of the TA cells generates the cells that line the villi: absorptive enterocytes, secretory goblet, enteroendocrine; as well as Paneth cells [31]. Except for Paneth cells, these cell types take 3 to 5 days to migrate out of the crypt up the villi, where they are shed into the intestinal lumen. Paneth cells stay in the crypt and interweave with ISCs to form the intestinal stem cell niche that is sustained by Wnt signaling [31,32]. Quiescent LRCs serve as a stem cell reserve to replace damaged ISCs and are active only during stress or injury [33]. Enterocytes, preterminal enteroendocrine cells, goblet cell precursors and Dll1 + secretory progenitors are also notable for their plasticity and can serve as a reservoir for lost stem cells [34][35][36]. Mature Paneth cells also show an injury-activated conversion to a stem cell like state [37]. Together with new perceptions in hematopoiesis [4,6,38], the dynamic plasticity of the intestinal crypt has expanded the stem cell paradigm [39], especially during replenishment after damage or inflammation.
To examine whether MLL1 plays non-hematopoietic roles in the adult, we employed ligand-induced conditional mutagenesis [40,41] using a tamoxifen-inducible Rosa26-CreERT2 line (RC) [42] for near-ubiquitous Cre recombination to discover that MLL1 is required for the maintenance of the ISC compartment and the balance between secretory and absorptive cell lineages in the adult intestine.

Intestinal functions collapse after loss of MLL1 in adult mice
To explore MLL1 functions, we utilized a multipurpose allele that can be converted from one state to another using FLP and Cre recombination (S1A-S1C Fig) [43,44]. Homozygous embryos carrying the targeted gene trap allele, Mll1 A/A , developed normally until E12.5 when they displayed pallor of the liver and were smaller (S1D Fig). After E13 no live Mll1 A/A embryos were found (S1 Table). After FLP and Cre recombination, Mll1 FC/FC embryos carrying the frame-shifted allele displayed the same phenotype indicating that both A and FC are true null alleles as expected from the loss of MLL1 protein [45]. Furthermore the Mll1 A/A and Mll1 FC/FC phenotype recapitulated another likely null allele [6]. Homozygous mice from these three mutant alleles present the same embryonic lethality due to the failure to engage definitive hematopoiesis [21]. Consequently we suggest that the diverse early developmental phenotypes reported previously for mutations in Mll1 [46,47] were likely due to dominant negative Mll1 alleles.
To identify postnatal roles of MLL1, ligand-induced conditional mutagenesis using Rosa26-CreERT2 (RC) was applied to two month old adults. Mice lacking MLL1 developed severe bone marrow cytopenia and died or had to be sacrificed on average within two weeks (Fig 1A and 1B). As previously reported using the same conditional allele, in vitro deletion of Mll1 in KSL-enriched HSCs from Mll1 F/F; RC/+ mice resulted in significant downregulation of Hoxa9, Meis1, Mecom/Evi1 and Prdm16 [6].
To bypass the bone marrow related lethality and thereby uncover non-hematopoietic phenotypes, bone marrow from wild type (wt) B6.SJL mice was transplanted into lethally irradiated Mll1 F/+; RC/+ or Mll1 F/F; RC/+ mice. After stable engraftment tamoxifen gavage induced widespread Cre-mediated excision of Mll1 (Fig 1C). Examination of the bone marrow confirmed the successful and near-complete reconstitution of the hematopoietic stem cell compartment by wt donor cells. FACS analysis for KSL-Slam enriched HSCs showed comparable frequencies in bone marrow transplanted (BMTx) Mll1 FC/+; RC/+ and Mll1 FC/FC; RC/+ mice with the hematopoietic compartment comprised only of wt cells of CD45.1 origin ( Fig 1C). Notably, the BMTx Mll1 FC/FC; RC/+ mice suffered from diarrhea and wasting but heterozygous BMTx Mll1 FC/+; RC/+ did not ( Fig 1D). These data indicate that MLL1 is not only required in the hematopoietic compartment but is similarly critical elsewhere.
The small and large intestine harbor stem cell compartments at the base of the intestinal crypts. In the crypt, MLL1 is expressed in ISCs and the TA compartment, whereas it is absent in differentiated cells above the TA compartment (Figs 1E and S1E) [48]. RNA profiling of sorted ISCs and Paneth cells confirmed expression of Mll1 and the other five H3K4MTs in both cell types however all are more strongly expressed in ISCs (S1F Fig). In the mutant small intestine of BMTx Mll1 FC/FC; RC/+ mice, expression of MLL1 was efficiently ablated and ISC markers OLFM4 (olfactomedin 4) and SOX9 were lost (Fig 2A). Concordantly, proliferation in the crypt was reduced as revealed by strongly reduced expression of the mitotic marker Ki67 (Fig 2A). However, no apparent changes in global H3K4 mono-, di-and trimethylation in the intestinal epithelium of Mll1 FC/FC; RC/+ mice were observed (S1G Fig).
Shortened villi with distorted morphology including vacuolar structures at the tip indicated diminished replenishment of cells into the villus (Fig 2B). Furthermore we observed increased numbers of enlarged goblet cells distributed irregularly along the villus and also ectopically in the crypt (Fig 2B). However, Paneth cells appeared unchanged possibly due to their longer life span (Fig 2C). Similarly, as evaluated by chromogranin A and alkaline phosphatase, the enteroendocrine and absorptive lineages appeared to be unaffected ( Fig 2C).
Without bone marrow rescue, tamoxifen-induced conditional mutagenesis of Mll1 (Mll1 FC/FC; RC/+ mice) showed the same defects with depletion of ISCs, decreased proliferation and a distortion of the secretory lineage (S2A- S2D Fig). Differentiation into the enteroendocrine and absorptive lineage was also apparently unaffected (S2E Fig). Notably the hallmark of Wnt signaling, nuclear β-catenin, was also unaffected (S2F Fig).

Intestinal epithelium-specific Mll1 conditional mutagenesis recapitulates ubiquitous deletion
To delete Mll1 exclusively in the adult intestine we employed the tamoxifen-inducible gut epithelium-specific Villin-CreER T2 line (Vil-Cre-ERT2) [49]. After tamoxifen administration about half (10/18) of the Mll1 FC/FC; Vil-Cre-ERT2/+ mice lost weight compared to control mice. These mice displayed wasting and were mostly analyzed three to four weeks after tamoxifen induction (Fig 3A and 3B). Compared to Mll1 FC/+; Vil-Cre-ERT2/+ littermates, OLFM4 and SOX9 expression, and proliferation were markedly decreased, goblet cells were enlarged and increased in number whereas Paneth and enteroendocrine cell numbers were unchanged ( Fig  3C and 3D). However the rest (8/18) of the Mll1 FC/FC; Vil-Cre-ERT2/+ mice did not display wasting and survived (Fig 3A), which is a result we attribute to insufficient Cre recombination by Vil-Cre-ERT2 as noted before [49,50], and the regenerative capacities of the intestinal crypt   To gather more evidence regarding mechanism and the temporal order of events after loss of MLL1, we analyzed intestinal sections shortly after tamoxifen induction. One to two weeks after the first tamoxifen gavage, the stem cell marker, OLFM4 and the proliferation indicator, Ki67, were markedly decreased. However only a slight increase in the goblet cell population in Mll1 FC/FC; Vil-Cre-ERT2/+ intestinal sections was observed ( Fig 3E) indicating that changes in ISCs preceded other alterations. We applied DESeq2 to analyze differentially expressed genes (DEGs). In the 4 day Lgr5 + stem cell profile, only 87 and 49 genes were up-or downregulated after removal of MLL1 at 5% FDR (false discovery rate) ( Fig 4A and S1 Data). However, none of the upregulated transcripts were increased by more than 2-fold and by DAVID analysis were mainly related to diverse terms such as 'response to metal ion', 'organic acid metabolic process' and 'regulation of lipid metabolic process' (Fig 4B). In contrast the most significant terms associated with the downregulated mRNAs were 'regulation of gene expression', 'epithelial cell differentiation' and 'cell proliferation' (Fig 4B). The 10 day profile revealed 179 DEGs, of which 105 were upregulated, 74 were downregulated with significant overlaps to the 4 day profile (S1 Data).

Transcriptome analysis identifies key intestinal transcription factors as central to the crypt stem cell niche
Gene set enrichment analysis (GSEA) of both 4 and 10 day ISC profiles revealed that overall ISC signature genes were downregulated whereas goblet cell signature genes were upregulated (Figs 4C, 4D, 4F and S5A and S5B), which concords with our immunohistochemical analyses. We focused on the overlap between the 4 and 10 day downregulated mRNAs. The transcription factors Pitx2, Foxa1, Zfp503/Nolz1, Pitx1, Onecut2 and Gata4 are prominent (Table 1 and Fig 4E). The expression of Foxa1 and Pitx2 was evaluated by qRT-PCR with good agreement    4G). Concordant with RNA-sequencing data from Mll1-deficient forebrain [52,53], mRNA for the chromatin co-factor, Bahcc1 was downregulated (Table 1). BAHCC1 binds to H3K27me3 and sustains repression of target genes [54]. These data suggest that MLL1 maintains the transcriptional identity of ISCs. H3K4me3 chromatin immunoprecipitation indicated that Foxa1, Pitx1, Pitx2, Gata4, One-cut2, Zfp503/Nolz1 and Bahcc1 promoters are direct targets for MLL1. However similar indications for Far1, Hspa8 or Ces2g were not obtained (Fig 4H). Notably two long non-coding RNAs featured prominently in the list of downregulated transcripts ( Table 1). One of these, ENSMUST00000215098.2, overlaps with the last two exons of Jaml/Amica1 and was very strongly downregulated. However this transcript appears to play no role in the regulation of JAML expression, which remained unchanged after loss of MLL1 as evaluated by Western blot analysis (S5D Fig).

Loss of MLL1 in ISCs provokes transcriptional changes in Paneth cells
ISCs are anchored in the crypt in close association with Paneth cells. In order to elucidate whether the transcriptional changes in ISCs influenced the neighboring Paneth cells, we analyzed Paneth cell transcriptional profiles 4 days after deletion of Mll1 in Lgr5 + ISCs, to find 198 and 72 transcripts significantly up-or downregulated respectively (Fig 5A and S1 Data). The most significant terms associated with downregulated mRNAs relate to perturbation of protein folding and homeostasis in the endoplasmic reticulum (Fig 5B-5D). In contrast upregulated mRNAs associate with metabolic changes. Strikingly, transcripts of genes belonging to all five respiratory chain complexes were upregulated (Fig 5B-5D). Paneth cells normally utilize glycolysis with lactate as the end product whereas ISCs depend on mitochondrial oxidative phosphorylation [55]. These transcriptional changes suggest that loss of MLL1 in the stem cell compartment provokes changes in Paneth cells and indeed expression of Paneth cell marker genes was downregulated (Fig 5C and 5D).

Skewed differentiation of organoids after loss of MLL1
To evaluate the cell-intrinsic requirement of MLL1 in the small intestine, we isolated crypts from Mll1 F/+; RC/+ and Mll1 F/F; RC/+ mice and cultured organoids [56]. After passaging, the organoids were induced with 4-hydroxy tamoxifen for 24 hours on day 2. After further passages, the Mll1 FC/FC; RC/+ organoids showed a gradual decrease in budding and increasingly formed round, less differentiated cyst-like spheres (Fig 6A and 6B).
To analyze DEGs, we applied DESeq2 analysis to the RNA-seq dataset. At P1 only 24 genes were up-or downregulated at a 5% FDR (Fig 6C) including an exceptional concordance with downregulated genes in the ISC RNA-seq (Table 1), namely the transcription factors Pitx2, Foxa1, Zfp503, Pitx1 and Gata4 (Table 1 and S1 Data). Other commonly downregulated genes include Bahcc1, Hspa8 and Prss12. In contrast, no significantly upregulated genes in common between 4 day ISC and P1 organoid transcriptomes were observed, potentially reflecting the different cellular compositions and circumstances between sorted ISCs and organoids in culture. DAVID analysis of the P1 organoid data showed that the upregulated transcripts mainly related to diverse terms including 'response to external stimulus', 'leukocyte cell-cell adhesion' and 'cell response to chemical stimulus' (Fig 6D). In contrast, the most significant terms associated with downregulated transcripts after MLL1 removal were comparable to the downregulated terms of the sorted ISCs (Fig 4B) including 'cell fate commitment', 'negative regulation of gene expression' and 'intestinal epithelial cell differentiation' (Fig 6E). After P3, many more (1629 and 434) genes were up-and downregulated at a 5% FDR (S6E Fig). The downregulated transcripts primarily associated with metabolic changes potentially reflecting the transition to spheroids (S6F Fig). Nevertheless, most of the prominent downregulated mRNAs common to ISCs and P1 organoids were also downregulated at P3 (Table 1 and S7  Fig). The most significant terms associated with upregulated transcripts relate to 'cell surface  requirement for MLL1 is cell intrinsic and essential for the maintenance of crypt morphology and budding in organoid cultures.
Overall the data establish that MLL1 sustains the expression of key ISC genes. Their reduced expression after loss of MLL1 leads to a loss of ISCs, consequent loss of TA cells and aberrant secretory cell differentiation biased towards goblet cells. These perturbations in intestinal homeostasis lead to wasting and death of the mouse.

Discussion
Here we add a fourth stem cell to the known MLL1 repertoire of (i) HSCs [21,22], (ii) skeletal muscle satellite cells [27] and (iii) postnatal neural stem cells (NSCs) [28]. Additionally, MLL1 is required to sustain salivary gland and intestinal cancer stem cells mutated in the Wnt pathway [48,57,58]. Consequently, all known MLL1 functions relate to action in stem cells, which suggests that MLL1 conveys an essential stem cell property. This possibility is enhanced by comparison to the MLL1 paralogue, MLL2, whose known functions in adult mice do not relate to stem cells rather to macrophages (for lipopolysaccharide response) or fertility [59][60][61][62].
To explore the idea that MLL1 conveys a key stem cell property, we inspected the transcriptome profiles after conditional loss of MLL1 in the four adult/postnatal stem cell populations [27,28,63] (S1 Data). Some overlap between two stem cell types was observed, however no fully shared candidate regulators or gene expression programs were identified. Although deeper, more systematic, transcriptome or cell biology approaches may reveal a shared MLL1 stem cell property, the lack of concordance between MLL1 regulation of these four stem cell transcriptomes is not unexpected. Previous work with MLL1 noted that direct target genes are not shared between different cell types [63] and a similar observation was made for MLL2 [59]. That is, the regulation of gene expression by the Trithorax homologues, MLL1 and MLL2, varies depending on the cell type and is not universal.
How does MLL1 regulate key lineage specific transcription factors differently in different lineages? MLL1 and MLL2 are amongst the few proteins that include the CxxC zinc finger that binds unmethylated CpG dinucleotides [24,64] as well as a PHD finger that binds H3K4me3 [65]. Hence, as suggested before [59], MLL1 and 2 have the potential ability to bind CpG island promoters without the need for recruitment by sequence specific DNA binding transcription factors. This potential accords with the observation that both MLL1 and 2 appear to be bound at almost all active promoters [45,66]. Consequently additional factors are required to explain the restricted transcriptional specificities of the MLLs. Notable in this regard, PAX7 is bound to MLL1 when satellite cells are activated, and enhanced transcriptional activation from both the Myf5 promoter, to initiate skeletal muscle replenishment, and the Pax7 promoter itself, depends on MLL1 [27]. This suggests that key transcription factors can either acquire the ability to interact with MLL1 bound at target promoters or recruit MLL1 to target promoters, or both.
Amongst the transcription factor mRNAs identified after loss of MLL1 in ISCs, Pitx2 is prominent. Pitx2 was previously identified as a direct target of MLL1 in ESCs and HSCs/ hematopoietic progenitor cells [67,68]. PITX2 is a homeodomain protein responsible for leftright asymmetric morphogenesis in the gut and proper positioning of the small intestine in the body cavity [69]. Also notably identified in ISCs are Foxa1 and Onecut2. Both genes, previously known as Hnf3α and Hnf6α, are expressed in all epithelia of the gastrointestinal tract from its embryonic origin into adulthood. Together with Math1, they are critical for goblet cell differentiation and function [70,71]. Gata4, which has previously been described as an MLL1 target gene [48,72], is also amongst the top downregulated mRNAs after loss of MLL1. Some aspects of the intestine specific deletion of Gata4 in the adult mouse resemble the MLL1 phenotype described here including decreased proliferation in the crypt with increased numbers of goblet cells [73]. Also notable is downregulation of the chromatin accessory factor, BAHCC1, which has recently been shown to bind H3K27me3 and loss of Bahcc1 expression led to loss of H3K27me3-mediated repression of target genes, thereby defining an H3K27me3 mechanism for repression that does not rely on Polycomb Repressive Complexes (PRCs) [54]. Hence maintenance of Bahcc1 expression provides a further component for the maintenance of ISC identity by MLL1 through repression of gene expression, potentially those involved in the MAPK and Wnt signaling pathways [48].
Upon loss of MLL1, the close association between ISCs and Paneth cells in the crypt and organoids is destabilized (Fig 7). We suggest that loss of niche integrity leads to secondary alterations particularly in the Paneth cell transcriptome but also possible changes in niche properties especially to the central ISC at the bottom of the crypt, which is flanked by two Paneth cells and has long-term self-renewal potential compared to border ISCs [74]. Notably, MLL1 regulates the positional identity of NSCs [28], which may be relevant to the positional identity of the central ISC.
Deletion of Mll1 in organoid cultures resulted in a gradual transition from branched organoids to spheroids. Under normal conditions, ISCs and Paneth cells reside in the tightly curved organoid budding regions where adherence junctions and high curvature limits stem cell expansion [75]. Loss of MLL1 provokes cellular changes that leads to the formation of spheroids. Notably, the transition to spheroids was not accompanied by a loss of proliferation, which stands in contrast to the loss of MLL1 in the crypt. This difference may be due to the absence of mesenchymal tissue in organoids, which exhibit a complex dynamic in intestinal crypts, or possibly de-differentiation of the ISCs in the organoids to an MLL1-independent state.
Once again we report the remarkable specificity of regulated gene expression conveyed by a Trithorax homologue. As described here, loss of MLL1 resulted in loss of transcriptional regulation with subsequent significant functional consequences in the crypt. In ESCs, the expression of only one gene, Magoh2, entirely depends on MLL2. Removal of MLL2 resulted in the suppression of Magoh2 expression by H3K27 methylation, followed by DNA methylation [59,76] thereby providing more evidence supporting the conclusion that a primary function of Trithorax action is to prevent Polycomb-Group repression [77]. These observations and conclusions with Magoh2 in ESCs were recently confirmed [78]. Selective gene specific antirepression could also explain the action of MLL2 on Pigp in macrophages [60] and MLL1 on Hoxa9 in HSCs [26].
Despite a high degree of evolutionary conservation and near-ubiquitous expression, the Trithorax homologues appear to regulate only a small number of genes in cell-type specific patterns. As observed in other adult stem cells, MLL1 regulates the expression of transcription factors in ISCs that maintain stem cell identity and likely influence lineage commitment decisions. These observations lead to the attractive proposition that MLL1 is a master stem cell regulator. However, a unifying mechanism for this proposition remains to be identified.

Ethics statement
Experiments were performed in accordance with German animal welfare legislation, and were approved by the relevant authorities, the Landesdirektion Dresden.

Targeting constructs
The targeting construct for Mll1 was generated using recombineering (S1A Fig) employing an engrailed-intron-splice-acceptor-IRES-LacZ-Neomycin-polyA cassette flanked by FRT sites [44]. The critical exon 2, which upon deletion results in a frameshift and a premature stop codon in exon 3, was flanked by loxP sites.

Gene targeting and generation of conditional knockout mice
Gene targeting in R1 embryonic stem cells (ESCs) was performed as described [79]. Correct integration in the Mll1 locus was confirmed by Southern blot analysis using 5' and 3' external probes and a LacZ internal probe (S1B and S1C Fig). Two correctly targeted ESC clones were injected into blastocysts and gave rise to several chimeras, which were able to establish germ line transmission. Mll1 A/+ mice were crossed to the hACTB-Flpe line to generate Mll1 F/+ mice. Mll1 A/+ and Mll1 F/+ mice were backcrossed at least six generations to C57BL/6JOlaHsd mice. Subsequently, those mice were crossed to the Rosa26-Cre-ERT2 (RC) line [42] to generate conditional, tamoxifen-inducible Mll1 F/+; RC/+ mice. Mll1 F/+ mice were bred with Lgr5-eGFP--CreERT2 and Villin-CreER T2 (Vil-Cre-ERT2) [30,49] mice. Primers for genotyping are provided in S2 Table. Tamoxifen Tamoxifen (Sigma Aldrich, T5648) was given to at least 10-week old mice by gavage (4.5 mg per day) for six days with three days break in between [80]. For RNA-Sequencing experiments, Mll1 F/+; Lgr5-eGFP-CreERT2/+ and Mll1 F/F; Lgr5-eGFP-CreERT2/+ mice received 1 mg tamoxifen via intraperitoneal (IP) injection for 3 consecutive days. To obtain Mll1-deficient crypts Mll1 F/F; RC/+ and Mll1 F/+; RC/+ mice received tamoxifen (4.5 mg per day) via gavage for four days and were sacrificed seven days after the final gavage. Intestinal organoids were induced on day 2 after splitting using 800 nM 4-hydroxy tamoxifen (Sigma Aldrich, H7904) for 24 h.

Bone marrow transplantation
Mll1 F/+; RC/+ and Mll1 F/F; RC/+ recipient (CD45.2) mice were lethally irradiated with 8.5 Gy (Xray source MaxiShot from Yxlon). Bone marrow cells from B6.SJL mice (CD45.1) were prepared by crushing femurs and tibiae with a mortar and pestle in ice-cold PBS supplemented with 5% fetal bovine serum. Red blood cells were removed with ACK lysis buffer (Thermo Fisher Scientific). 1 x 10 6 lineage depleted (Lin -) bone marrow cells were injected into the retro-orbital venous plexus. Animals were maintained on water containing 1.17 mg/ml neomycin (Merck) for three weeks after irradiation. Complete donor cell engraftment of wt CD45.1 + cells was confirmed by flow cytometry on peripheral blood with antibodies directed against the following murine antigens (clones given in brackets): CD45.1 (A20), CD45.

5-Bromo-2-deoxyuridine (BrdU) assay
Mice were injected IP with BrdU (0.6 mg/10 g body weight in sterile PBS) and sacrificed 2 h later. The jejunum was dissected in cold PBS and processed for immunohistochemistry.

Histochemistry and immunohistochemistry
Mouse intestine was flushed gently with cold PBS. Embryos were dissected from plugged mice on the respective gestational stage and placed in PBS. Intestine and embryos were fixed in 4% paraformaldehyde overnight. Dehydration and paraffin infiltration utilized the Paraffin-Infiltration-Processor (STP 420, Zeiss). Dehydrated tissues were embedded in paraffin (Paraffin Embedding Center EG1160, Leica) and 5 μm sections were prepared. Sections were deparaffinized in xylene, rehydrated through a series of alcohols, stained, dehydrated, and mounted. Femur sections were stained with Giemsa according to standard protocols. For the basic evaluation of the intestine hematoxylin and eosin (H&E) stain was performed. Periodic acid-Schiff (PAS) stain and alcian blue were used to identify goblet cells. For enterocytes, alkaline phosphatase stain was performed using two different methods. Sections were either incubated with Red Alkaline Phosphatase Substrate (Vector red, Vector Laboratories) for 10 minutes (min) or with nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate solution for 30 min at room temperature (RT). For immunohistochemistry, antigen retrieval was performed by microwaving slides in 10 mM citrate buffer (pH 6.0) for 12 min (Microwave RHS 30, Diapath).
Endogenous peroxidases were quenched with 0.3% H 2 O 2 in methanol. Sections were incubated in blocking serum (5% goat serum) for 1 hr at RT followed by overnight incubation with primary antibodies (S3 Table) at 4˚C. Following incubation with the secondary antibody (S3 Table) the immune peroxidase was detected using a Vectastain ELITE ABC kit (Vector) and visualized with a solution of diaminobenzidine (Sigma Aldrich) in the presence of 0.01% H 2 O 2 . All sections were counterstained with hematoxylin or alcian blue. Images were collected with an Olympus WF upright microscope and analyzed using the MetaMorph Microscopy Automation and Image Analysis Software.

In situ hybridization
The protocol for in situ hybridization was modified from [81]. Briefly, 8 μm thick sections were rehydrated. The sections were treated with 0.1 N HCl and proteinase K. Slides were postfixed and sections were then demethylated with acetic anhydride and prehybridized. Hybridization was done with 2 μg/ml digoxigenin (DIG)-labeled Olfm4 RNA probe for 24 h at 65˚C. Slides were washed and incubated with blocking solution for 1 hr. The sections were incubated with anti-DIG-alkaline phosphatase conjugate overnight at 4˚C. Slides were washed and developed with BM purple.

Crypt isolation
The jejunum was harvested from mice, flushed with ice cold PBS to remove any faecal content and cut open longitudinally. The tissue was placed lumen side up on a petri dish and the villi were removed by gently scraping the tissue using a glass cover slip. The tissue was cut into 2-4 cm pieces and was washed several times with ice-cold PBS to remove residual villi fragments. Tissues were transferred into a fresh tube containing 15 ml of 2 mM EDTA/PBS chelation buffer and placed on a rotating wheel for 30 min at 4˚C. The crypts were then detached from the basal membrane by vigorous shaking in 5% FCS/PBS solution. The suspension was filtered with a 100 μm cell strainer followed by a 70 μm cell strainer. Isolated crypts were centrifuged at 800 rpm for 5 min at 4˚C. The final fraction consisted of pure crypts and was used for organoid culture or single cell dissociation.

Organoid culture
Purified crypts were resuspended in 10 ml DMEM/F12 (Life Technologies). 10 μl of the crypt suspension was used to count the number of crypts under the microscope. The pelleted crypts were resuspended in Matrigel matrix (Corning) at desired crypt density. Approximately, 400 crypts in 25 μl Matrigel matrix were seeded per well in a pre-warmed 24-well plate and incubated for 15 min at 37˚C until the Matrigel matrix solidified. Then, 400 μl of IntestiCult Organoid Growth Medium (STEMCELL Technologies) was added to each well. Organoids were cultured at 37˚C in a 5% CO 2 incubator and maintained in culture for 5 days before being passaged and split for experimental procedures. The growth medium was replaced every 2-3 days. Organoids at passage 1 and 3 on day 5 were washed with PBS and total RNA was extracted using Trizol.

Flow cytometry
ISCs were initially characterized and identified with the use of an Lgr5-eGFP-CreERT2 knockin allele [30]. Mll1 F/+; Lgr5-eGFP-CreERT2/+ and Mll1 F/F; Lgr5-eGFP-CreERT2/+ littermates (n = 4) were given tamoxifen via IP injections for 3 consecutive days and were dissected 4 and 10 days later. Crypts were dissociated into single cells with TrypLE Express (Thermo Fisher Scientific) for 30 min at 37˚C. Dissociated cells were passed through 70 μm cell strainer and washed with 5% FCS/PBS. Cells were stained with antibodies with the following antibodies for 45 min on ice: anti-mouse CD24 PE (clone M1/69), anti-mouse CD326 (EpCAM) APC (clone G8.8), antimouse CD45 Alexa-Fluor 700 (clone 104). Sorting was performed on a FACS Aria III cell sorter (BD). After scatter discrimination to remove doublets the cell suspension was negatively selected with SYTOX blue dead cell stain and anti-CD45 to remove dead and hematopoietic cells, respectively. The cells where then positively selected with anti-EpCAM to enrich for intestinal epithelial cells. According to [82] CD45 -, EpCAM high , CD24 med and GFP high characterized ISCs and CD45 -, EpCAM high , CD24 high and GFPcharacterized Paneth cells (S3A and RNA sequencing 300 intestinal stem and Paneth cells were sorted into 2 μl of nuclease free water with 0.2% Triton-X 100 and 4 U murine RNase Inhibitor (NEB). RNA was reverse transcribed (Invitrogen) and cDNA amplified using Kapa HiFi HotStart Readymix (Roche). The cDNA quality and concentration was determined with the Fragment Analyzer (Agilent). Samples were subjected to library preparation (TruePrep DNA library Prep Kit V2 for Illumina, Vazyme). Libraries were purified followed by Illumina sequencing on a Nextseq500 with a sample sequencing depth of 30 million reads on average. mRNA was enriched from 500 ng total RNA from organoids of the respective genotype by poly-dT enrichment using the NEBNext Poly(A) mRNA Magnetic Isolation Module according to the manufacturer's instructions. The polyadenlyated RNA fraction was eluted in first strand cDNA synthesis buffer (NEBnext, NEB). After chemical fragmentation samples were directly subjected to the workflow for strand specific RNA-seq library preparation (Ultra II Directional RNA Library Prep, NEB). Libraries were purified, quantified using the Fragment Analyzer (Agilent) and sequenced on an Illumina Novaseq 6000 with 2x 100 bp reads using a S4 flowcell to an average depth of 50 million read pairs.
The short reads were aligned to the mm10 transcriptome with GSNAP (2018-07-04 and 2020-12-16) and a table of read counts per gene was created based on the overlap of the uniquely mapped reads with the Ensembl Gene annotation (version 92 and 98), using feature-Counts (version 1.6.3 and 2.0.1). Normalization of the raw read counts based on the library size and testing for differential gene expression between the different genotypes was performed using the DESeq2 R package (version 1.24.0 and 1.20.0). Genes with an adjusted p-value (padj)� 0.05 were considered as significantly differentially expressed accepting a 5% FDR. To identify enrichment for particular biological processes and pathways associated with the DEGs, the DAVID GO/BP/FAT and KEGG database [83] was used. Gene set enrichment analysis was performed using GSEA software from the Broad Institute [84].

Reverse transcription and quantitative PCR (qRT-PCR) analysis
RNA from sorted cells was extracted using Trizol (Sigma-Aldrich) and reverse transcribed using AffinityScript Multiple Temperature cDNA Synthesis Kit (Agilent Technologies). Realtime quantitative PCR was performed with GoTaq qPCR Master Mix (Promega) by Mx3000P QPCR System (Agilent Technologies). Ct values were normalized against Rpl19. Primer sequences and length of the amplified products are given in S2 Table. Fold differences in expression levels were calculated according to the 2 -ΔCt method [85].

ChIP-qPCR
ChIP-qPCR was performed as previously described [4,45,86]. In brief, equal amounts of crypts were crosslinked by adding formaldehyde followed by addition of glycine to quench the formaldehyde. After a wash with ice cold PBS the cells were harvested and lysed. Sonication was performed with a water bath sonicator (BioRuptor, Diagenode) for 20 min at high power, 30 seconds ON, 30 seconds OFF resulting in DNA fragments of 200-600 bp. Immunoprecipitation was performed using an antibody against H3K4me3 (ab8580; Abcam). A mock control without addition of antibody was always performed in parallel. After washing the immune complexes were eluted with 500 μl 1% SDS, 0.1 M NaHCO 3 for 30 min, NaCl was added to a final concentration of 200 mM and samples were de-crosslinked overnight at 65˚C. Samples were extracted with phenol/chloroform/isoamylalcohol, isopropanol-precipitated in the presence of 20 μg glycogen and resuspended in water. The immunoprecipitated DNA was amplified by qPCR using primers flanking the promoter region (S2 Table). ChIP-qPCR data were normalized with percentage of input analysis. This method represents the amount of DNA pulled down by using the antibody of interest in the ChIP reaction, relative to the amount of starting material (input sample).

Western blot analysis
Crypt and organoid pellets were resuspended in ice-cold RIPA buffer supplemented with 1x cOmplete Protease Inhibitor Cocktail (Roche) and PMSF for 15 min on ice, centrifuged (13200 rpm, 4˚C, 15 min) and supernatant was collected. Whole cell extracts were separated by SDS-PAGE (10% Tris-glycine gel) and transferred to a PVDF membrane. Blots were probed with primary antibodies (S3 Table).

Quantification and statistical analysis
Data is presented as mean and error bars indicate standard deviation (s.d.) unless otherwise indicated. Statistical details of the experiments can be found in the figure legends. Graphs and statistics were generated with GraphPad Prism software (v6.0) and Microsoft Excel. Significance (p-values) for Kaplan-Meier graphs was determined by Mantel-Cox test. Significance for differential gene expression (p-values < 0.05) was determined with Wald test. Significance for the breeding statistics was determined with Chi-square test. All other graphs are computed with two-tailed Student's t test. N indicates the numbers of independent biological replicates per experiment unless otherwise indicated.