SETD1A protects HSCs from activation-induced functional decline in vivo

Lifelong maintenance of the immune system depends on self-renewing hematopoietic stem and progenitor cells (HSPCs) that are activated under stress conditions such as infection to replenish blood cells during demand-adapted hematopoiesis.¹  The accumulation of DNA damage limits the regeneration of blood cells and is a hallmark of aging and age-associated disorders, including blood cancers.^2,3  The activation of hematopoietic stem cells (HSCs) through interferon signaling during inflammation⁴  provides a physiological source for DNA damage, and lack of a functional Fanconi anemia DNA repair pathway results in the attrition of activated HSCs in situ.⁵  It remains unknown whether other DNA damage repair pathways are involved in this process. The fundamental dilemma between the maintenance of genomic integrity and the need to replenish tissue through replication is now a central issue in stem cell research,^6,7  highlighted by the counterintuitive finding that long-term (LT) HSCs are normally quiescent and thereby maintain genomic integrity by not replicating.^8,9  In this context, information on the molecular nature of physiological regulators orchestrating the expression of components maintaining genomic integrity of dividing LT-HSCs and hematopoietic precursors (HPCs) is missing.

The first indication that epigenetic mechanisms contribute to the regulation of hematopoiesis was provoked by the realization that the major leukemia gene, mixed-lineage leukemia (Mll1¹⁰ ), encodes a histone 3 lysine 4 methyltransferase¹¹  of the Set1/Trithorax type.¹²  Knockout studies in mice revealed that Mll1 is required for definitive hematopoiesis and that embryos lacking MLL1 die at E13.5.^13,14  MLL1 is 1 of 6 Set1/Trithorax-type histone 3 lysine 4 (H3K4) methyltransferases (Mll1-4; Setd1a and -b) in mammals. Gathering evidence indicates that the other 5 are also involved in hematopoiesis and various leukemic transformations.¹⁵  However, a unifying understanding of how the 6 H3K4 methyltransferases relate to each other and regulate hematopoiesis remains elusive.

A comprehensive evaluation of epigenetic regulation in hematopoiesis, in particular the H3K4 methyltransferase system, using constitutive and conditional knockouts in the mouse, is emerging. Conditional ubiquitous depletion in adults of MLL1 (Kmt2a¹⁶ ) or SETD1A (Kmt2f¹⁷ ), but not MLL2 (Kmt2b¹⁸ ) or MLL4 (Kmt2d¹⁹ ), is lethal. In hematopoiesis, lack of MLL1 or MLL4 resulted in loss of HSC maintenance in G₀ and in impaired or complete absence of repopulation activity after transplantation, which is a test that imposes replication stress through the requirement for extensive proliferation. The other 2 of the 6, Setd1b (Kmt2g) and Mll3 (Kmt2c), have not yet been functionally examined by conditional mutagenesis. However, 2 common subunits of H3K4 methyltransferase complexes (Dpy30, Cfp1) have been investigated, and recent conditional mutagenesis of either complex member revealed an essential role in adult hematopoiesis.^20,21  In both cases, loss of function in mice was lethal, most likely because of the ablation-induced stark pancytopenia. It remains unknown as to which defective H3K4 methylation complex caused the phenotype or whether the severe defects were due to summated effects from dysfunction of more than one H3K4 methylation complex.

SETD1A is identified as the major H3K4 methyltransferase in mouse embryonic stem cells, and knockout embryos die shortly before gastrulation.²²  Conditional ablation of Setd1a in hematopoietic cells revealed that it is a regulator of B-cell differentiation in the bone marrow (BM) and spleen,¹⁷  with an impact on erythrocyte differentiation,²³  suggesting that SETD1A regulates adult hematopoiesis. However, a systematic analysis of SETD1A in HSCs and progenitors has been lacking, and a clear functional assignment has not been made. Here we used various Cre drivers and assays to dissect SETD1A function in vivo and in vitro. Thereby we defined critical functions of SETD1A in HSCs and committed progenitors.

C57BL/6J (B6) and B6.SJL-Ptprc^aPep3^b/BoyJ (B6.SJL) mice were purchased from the Jackson Laboratory. Setd1a^F/F mice were described previously²²  and crossed with Rosa26-CreER^T2 (R.26-CreER^T2 24 ), SCL-CreER^T,²⁵ Vav-Cre,²⁶  and CD4-Cre²⁷  mice. LT-HSC in vivo proliferation was analyzed using SCL-CreER^T+;Setd1a^F/F;R.26-rtTA^ki/ki;Col1a1-tetO-H2B-mCherry^ki/ki 28  mice. Details on tamoxifen, 5-fluorouracil (5-FU), doxycyclin, and poly I:C (pI:C) treatment, irradiation, and transplantation are provided in the supplemental Material and methods, available on the Blood Web site. Flow cytometry was performed, as has been previously described.²⁹ 

Colony assays were performed according to the manufacturer’s instructions (Methocult M3434; StemCell Technologies). For single-cell cultures, LT-HSCs (KSL = lineage, Lin⁻ [CD3, CD11b, CD19, B220, Gr1, NK1.1, Ter119]⁻ Kit⁺, Sca-1⁺; CD34⁻ CD135⁻), short-term (ST) HSCs (KSL CD34⁺ CD135⁻), and multipotent progenitors (MPPs; KSL CD34⁺ CD135⁺) were sorter purified. For bulk analysis, 3000 KSL or myeloid progenitors (MPs; Lin⁻ Kit⁺ Sca-1⁻) or 600 LT-HSCs and ST-HSCs were cultivated for 3 or 8 days, respectively.

Molecular analysis, including next-generation sequencing,³⁰  quantitative polymerase chain reaction (qPCR), recombination analysis, and alkaline comet assays (OxiSelect, Cell Biolabs, Inc.), was performed according to manufacturer’s instructions. H3K4 methylation was tested by Western blot (Lin⁻ cells) and immunofluorescence (KSL cells). Details for all procedures are given in the supplemental Material and methods.

Significant differences of survival were determined using the log-rank test (Prism 5). Two-tailed Student t tests were performed for all other analyses (*P < .05; **P < .01; ***P < .001). Error bars show standard deviations.

The histone lysine methyltransferase Setd1a is expressed in mouse HSPCs (Figure 1A-B). Complete loss of Setd1a in the embryo using either a targeted gene trap allele or an allele lacking exon 4 provoked lethality shortly after implantation.²²  To examine SETD1A function in adult mice, we bypassed embryonic lethality using tamoxifen (TAM)–inducible site-specific recombination,³¹  initially from the Rosa26 locus.^24,32  In 2-month-old adults, deletion of Setd1a exon 4 led to the death of R.26-CreER^T2+;Setd1a^F/F mice 12 to 26 days after TAM induction (Figure 1C). Because of the near-ubiquitous expression of CreERT2 in these mice, it was not clear in which cell types SETD1A was required, but the speed of death suggested that defects in hematopoiesis could be involved. This idea was tested using Vav-Cre to delete Setd1a, which resulted in lethality 7 to 20 days after birth (Figure 1D). Vav-Cre-mediated cell type–specific deletion of Setd1a results in the lack of expression of SETD1A in definitive HSCs from the time point of their emergence during embryonic development.²⁶ Vav-Cre⁺;Setd1a^F/F mice were born at normal Mendelian frequencies (supplemental Figure 1A-C) but lacked Lin⁻ Kit⁺ Sca-1⁺ (Lin = CD3, CD11b, CD19, B220, Gr1, NK1.1, Ter119) HSPCs in the BM 9 days after birth (supplemental Figure 1D-E), indicating that these cells are not generated or expanded during pre- and neonatal development in the absence of SETD1A. To confirm this indication, TAM induction of transplanted R.26-CreER^T2+;Setd1a^F/F BM chimeric mice (scheme; Figure 1E) resulted in pancytopenia characterized by a severe drop of hematological blood parameters. After the mice had recovered from the impact of tamoxifen, red and white blood cells were reduced in comparison with the transplanted control (R.26-CreER^T2+;Setd1a^+/+) chimeras (Figure 1F). Furthermore, SETD1A-deficient HSPCs exhibited reduced capacity to form colonies in vitro (Figure 1G). In Lin⁻ cells, lack of SETD1A led to global reductions in mono-, di-, and trimethylation on lysine 4 of histone 3 (H3K4me1, me2, me3; Figure 1H). Taken together, the data indicate that SETD1A is an important cell-intrinsic regulator for adult HSC differentiation and is the major H3K4 methyltransferase in blood stem and progenitor cells.

Notably, using CD4-Cre, we found that SETD1A is not required in thymocytes (Figure 1I-J), despite efficient recombination (supplemental Figure 1F), thereby indicating that SETD1A is required earlier in hematopoiesis. This observation also provides further evidence that SETD1A is not essential in all cell types.²² 

Mixed BM chimeras were established between R.26-CreER^T2+;Setd1a^F/F and wild-type donor cells. Three months after transplantation, recombination of the Setd1a^F/F alleles was induced by TAM treatment. Survival of the mice was ensured by wild-type competitor cells (scheme; Figure 2A). Consistent with the pancytopenia in plain BM chimeric mice (Figure 1F), the contribution of R.26-CreER^T2+;Setd1a^F/F donor cells to blood neutrophils (PMN), other myeloid cells, and T and B lymphocytes severely declined after TAM treatment of competitively transplanted chimeric mice (Figure 2B). In contrast, LT-HSC numbers were unaffected (Figure 2C-D). The KSL compartment was enlarged (Figure 2E) owing to mild expansions of ST-HSCs and MPPs upon deletion of SETD1A (Figure 2D). However, more downstream HPCs, including CMPs and MEPs, were diminished (Figure 2D). The strong effect on MEPs is consistent with a recent report on the role of SETD1A in red blood cell differentiation.²³  Efficient recombination of the loxP-flanked Setd1a allele in LT-HSCs was confirmed by PCR genotyping of 165 single LT-HSCs (Figure 2F), indicating that LT-HSCs can survive without SETD1A, which is, however, required for later steps of hematopoietic differentiation.

Because a function for SETD1A in adult LT-HSCs was not revealed by the above investigations, LT-HSCs lacking SETD1A were challenged under conditions of replicative stress. Transplantation of BM cells from TAM-induced primary mixed BM chimeras into secondary recipients (scheme; Figure 2A) revealed no detectable contribution of R.26-CreER^T2+;Setd1a^F/F donor cells to myeloid or lymphoid blood compartments (Figure 2G). Importantly, however, low levels of LT-HSCs, ST-HSCs, and MPPs were found (Figure 2H), suggesting that SETD1A contributes to HSPC expansion.

A further stress condition was examined by injecting 5-FU into TAM-treated BM chimeric mice. The injection abolished the contribution of Setd1a-deleted cells to blood neutrophils, whereas controls recovered from the 5-FU assault (Figure 2I). Additionally, R.26-CreER^T2+;Setd1a^F/F HSCs were severely decreased 14 days after 5-FU administration (Figure 2J). We conclude that SETD1A is an important regulator of demand-adapted hematopoiesis. Although it is not strictly required for survival of stem and progenitor cells, it may contribute to their expansion.

To investigate the effect of Setd1a deletion in LT-HSCs, we generated SCL-CreER^T+;Setd1a^F/F mice (scheme; Figure 3A). In contrast to ubiquitous Setd1a deletion using R.26-CreER^T2+, and deletion of Setd1a in definitive HSCs at the time point of their emergence during development (Vav-Cre), deletion mediated by SCL-CreER^T is largely confined to LT-HSCs with little recombination activity in hematopoietic progenitors.²⁵  The use of this Cre deleter delivered an important insight because the SCL-CreER^T+;Setd1a^F/F mice survived for at least 24 weeks after TAM induction (Figure 3B). As was expected, the efficiency of SCL-CreER^T recombination was high in LT-HSCs but lower in KSL, MP, and PMN cell fractions (supplemental Figure 2A), suggesting that hematopoietic activity of Setd1a wild-type progenitor cells suffices for survival.^8,9  Also in SCL-CreER^T+;Setd1a^F/F HSPCs, mono-, di-, and trimethylation of H3K4 was significantly decreased in the absence of SETD1A (supplemental Figure 2B-C). Concordantly, only modest perturbations in blood erythrocytes (Figure 3C), spleen T, B, and myeloid cells (Figure 3D; supplemental Figure 2D-G) were observed 7 or 24 weeks after TAM induction in SCL-CreER^T+;Setd1a^F/F mice. Despite efficient recombination, LT-HSC numbers in the BM were unaltered (Figure 3E), whereas the KSL population was expanded (Figure 3F) with increased ST-HSCs and MPP populations (Figure 3E). Notably, colony formation from BM was strongly reduced (Figure 3G), and competitive transplantation revealed no SCL-CreER^T+;Setd1a^F/F contribution to blood neutrophils, other myeloid cells, or T or B lymphocytes (Figure 3H; supplemental Figure 2H-I). However, there was a small but detectable population of SCL-CreER^T+;Setd1a^F/F donor LT-HSCs cells in the BM of the recipient mice 16 to 40 weeks after transplantation (Figure 3I; supplemental Figure 2J), suggesting that homing into the BM occurs independently of the presence of SETD1A (supplemental Figure 2K). The data from the SCL-CreER^T experiments therefore concord with and focus the conclusions from the R.26-CreER^T2 competitive transplantation experiments. SETD1A is not essential in LT-HSCs but contributes when stress is applied. Furthermore, it also appears to be required during differentiation.

Most immature LT-HSCs express intermediate levels of the receptor KIT, which is transiently elevated upon activation.^30,33  5-FU stimulated HSCs also have reduced KIT expression.^29,34  LT-HSCs from SCL-CreER^T+;Setd1a^F/F mice expressed low levels of KIT (Figure 4A), suggesting either that they are potent LT-HSCs or that they are stimulated and proliferating HSCs. To discriminate between these possibilities, we compared the proliferative capacity of SETD1A-depleted individual LT-HSCs and ST-HSCs from TAM-treated SCL-CreER^T+;Setd1a^F/F mice. In vitro, both SETD1A-deficient LT-HSCs and ST-HSCs showed a significantly reduced division rate (supplemental Figure 2L). However, in vivo, SETD1A-deficient LT-HSCs seem to replicate more than do wild-type controls, as is shown by an in situ pulse chase experiment using SCL-CreER^T+;Setd1a^F/F;R.26-rtTA^ki/ki;Col1a1-tetO-H2B-mCherry^ki/ki mice (Figure 4B-C) and ex vivo cell cycle analysis (Figure 4D). Cycling activity of SCL-CreER^T+;Setd1a^F/F LT-HSCs in situ was not accompanied by increased apoptosis, as was evaluated by the lack of accumulation of cells containing fragmented DNA (subG1; Figure 4E-F) or annexin V positivity (Figure 4G). Also, there were no differences in γH2A.X levels, which are indicative for double-strand DNA breaks (Figure 4H). However, the production of reactive oxygen species (ROS), which usually accumulate in metabolically active cells and provoke oxidative stress, limiting the life span of HSPCs in vivo,³⁵  was reduced in the absence of SETD1A (Figure 4I). Together, these data suggest that Setd1a-deficient LT-HSCs and progenitor cells can contribute to hematopoiesis in vivo but have reduced expansion potential under experimental conditions after isolation from the mouse.

Key insights were revealed by transcriptome profiling of R.26-CreER^T2+;Setd1a^F/F KSLs from BM chimeras (scheme; Figure 5A). Unsupervised clustering confirmed a high data quality (supplemental Figure 3A), and loss of loxP-flanked exon 4 transcripts again revealed efficient recombination of the Setd1a^F alleles (supplemental Figure 3B). DAVID comparison with controls revealed that pathways related to DNA repair and stress response were the most affected by SETD1A ablation (Figure 5B; supplemental Table 1). Also strongly downregulated were genes involved in transfer RNA and noncoding RNA metabolism. In comparison with the number of downregulated genes, many fewer genes were upregulated (Figure 5C). This asymmetry concords with the expectation that H3K4 methyltransferases sustain gene expression, and therefore removal should reduce expression of direct targets. Consequently, genes upregulated after SETD1A removal may be indirect targets. The upregulated genes included terms related to protein localization, phosphorylation, and energy metabolism, which may be a consequence of the expansion of ST-HSCs and MPPs within the KSL population (Figure 2D-E; supplemental Table 1).

Gene set enrichment analysis (GSEA) revealed that LT-HSC signature genes³⁶  were globally downregulated (supplemental Figure 3C), whereas LSK CD48⁺ (ST-HSC) signature genes were globally upregulated (supplemental Figure 3D, supplemental Table 2). These data can be due to the increase in the frequency of ST-HSCs and MPPs in R.26-CreER^T2+;Setd1a^F/F BM chimeric mice (Figure 2D-E) but may also suggest a role for SETD1A in maintaining the transcriptional identity of LT-HSCs.

As for the DAVID analysis, GSEA also revealed significant underrepresentation (false discovery rate < 0.25) of multiple gene sets related to DNA repair and DNA damage response (Figure 5D; supplemental Table 3). In fact, differential expression analysis revealed that transcripts important for virtually every DNA damage detection and repair pathway were significantly downregulated (P_adj ≤ .05), including homologous recombination and replication fork repair (Rad50, Rad51b, Rad51c, Zswim7, Recql, Rpain, Bre, Fam175a, Emsy, Eme1, Fbxo18), nonhomologous end joining (NHEJ; Prkdc, Nhej1), nucleotide excision repair (Gtf2h3, Gtf2h4, Rad23a), base excision repair (Parp2), DNA interstrand cross-link repair (Dclre1a, Fancg, Fancd2, Fanci), direct damage reversal (Mgmt), and protection from oxidative stress (Ogg1, Nudt1), mismatch repair system (Mlh1), translesion DNA synthesis (Poli), R-loop processing (Rnaseh1), telomere homeostasis (Pot1a), and cell cycle progression and checkpoint activation (Rad1, Mcts1, Tbrg1, Rbx1, Fbox6). Accordingly, deletion of Setd1a in Lin⁻ cells causes downregulation of H3K4me3 on these promoters, but not on an intergenic region on chromosome 9 (Figure 5E). These transcriptome and chromatin immunoprecipitation (ChIP) qPCR data suggest that SETD1A ablation in HSPCs impairs DNA damage responses and repair, which in turn may perturb the cycling activity of SETD1A-deficient HSPCs in vitro and in vivo.

These conclusions were first tested using in vitro cultivation of LT-HSCs and ST-HSCs from R.26CreER^T2+;Setd1a^F/F and control mice (Figure 5F). Setd1a was efficiently deleted by TAM induction in vitro (Figure 5G), resulting in significantly reduced cell expansion (Figure 5F,H), likely due to increased apoptosis (Figure 5I) and DNA damage (Figure 5J). Notably, in agreement with the in vivo data, ROS levels were reduced in R.26-CreER^T2+;Setd1a^F/F LT-HSCs and ST-HSCs in culture (Figure 5K), indicating that ROS production is not the cause of DNA damage.

Taken together, SETD1A regulates the expression of many components of DNA damage detection and repair pathways, and the specific ablation of SETD1A expression in HSPCs results in impaired colony growth and in the accumulation of DNA breaks in LT- and ST-HSCs.

We then used SCL-CreER^T+;Setd1a^F/F mice to examine responses to DNA damage in vivo. Directly after irradiation (10 Gy, 0 hour), SETD1A-deficient KSL showed reduced levels of γH2A.X (Figure 6A-B), indicating reduced activation of the DNA break-sensing machinery. Also, the subsequent disappearance of γH2A.X was delayed, suggesting that DNA repair was impaired (Figure 6A-B). Notably, SETD1A-deficient KSL cells showed constitutively increased DNA breaks as detected by terminal deoxynucleotidyltransferase-mediated deoxyuridine triphosphate nick end labeling (TUNEL), and repair was again delayed (Figure 6C-D). Both constitutively elevated levels of DNA breaks and delayed DNA break repair were confirmed by the comet assay (Figure 6E-F).

Furthermore, SCL-CreER^T+;Setd1a^F/F mice were significantly more radiosensitive than were controls (Figure 6G). Moreover, wild-type donor HSPCs transplanted into nonconditioned TAM-induced congenic SCL-CreER^T+;Setd1a^F/F mice engrafted efficiently and contributed to blood neutrophil production over 16 weeks (Figure 6H), suggesting that SETD1A contributes to the competitiveness of HSCs in situ. Consistently, the HSC compartment was chimeric, suggesting that SCL-CreER^T+;Setd1a^F/F LT-HSCs were out-competed by wild-type LT-HSCs (Figure 6I). We conclude that SETD1A orchestrates the expression of DNA damage-sensing and repair pathways, and these deficiencies mediate hypersensitivity of ablated HSPCs to irradiation insult and loss of competitiveness in comparison with wild-type LT-HSCs.

To further test for the physiological relevance of SETD1A-mediated defects in vivo, we activated HSCs of TAM-induced SCL-CreER^T+;Setd1a^F/F and control mice by injecting pI:C, which activates toll-like receptor 3 (scheme; Figure 7A). HSC activation is characterized by cell cycle entry, which makes mice highly sensitive to 5-fluorouracil (5-FU ⁴ ). Furthermore, resolution of inflammation-mediated activation of HSPCs depends on a functional Fanconi anemia DNA damage repair pathway.⁵  Without additional challenge, SCL-CreER^T+;Setd1a^F/F mice were as sensitive to 5-FU treatment as were controls (supplemental Figure 3E). In contrast, SCL-CreER^T+;Setd1a^F/F mice were more sensitive to activation-induced attrition than were wild-type mice. We first confirmed that loss of SETD1A in LT-HSCs resulted in G₀-to-G₁ transition and progression through the cell cycle (Figure 7B-C). pI:C treatment also resulted in increased cycling activity of LT-HSCs (Figure 7B-C). As a consequence, cell numbers of LT-HSCs and ST-HSCs increased (Figure 7D), and increased γH2A.X labeling indicated accumulating DNA damage (Figure 7E). Half of SCL-CreER^T+;Setd1a^F/F mice succumbed to pI:C treatment, suggesting increased sensitivity to type I interferon-induced inflammation (Figure 7F). The mice that survived pI:C treatment were found exquisitely sensitive to 5-FU treatment in comparison with wild-type mice (Figure 7F). We suggest that the functional decline of SETD1A-depleted HSPCs under physiologic inflammatory conditions may be based on receding genomic integrity and that SETD1A coordinates the expression of DNA damage recognition and repair components that help rescuing HSPCs from activation-induced attrition in vivo.

We observed that SETD1A prevented the accumulation of DNA lesions through both the speed of DNA damage signaling and DNA repair. Both processes were delayed by the absence of SETD1A. These delays suggest that SETD1A may be a direct and integral coordinator of DNA damage response in HSPCs. Furthermore, loss of Setd1a in HSCs rendered mice hypersensitive to inflammatory HSC activation and subsequent antiproliferative treatment, again placing SETD1A at the center of the orchestrated expression of DNA damage-signaling and repair components.

Vav-Cre-mediated deletion of Setd1a results in a constitutive lack of expression of SETD1A in definitive HSCs from the time point of their emergence during embryonic development. The severe reduction of KSL cells suggests that definitive HSCs are not generated or expanded during pre- and neonatal development. In contrast, HSCs in adult mice are largely quiescent and phenotypically maintained after depletion of Setd1a, suggesting that Setd1a is important for HSC division under conditions of proliferative pressure. In line with this interpretation is the finding that cell growth in vitro is blunted but not completely abrogated in the absence of Setd1a, suggesting that HSCs have the capacity to divide in the absence of Setd1a but fail to do so excessively.

Ubiquitous loss of Setd1a in mice led to rapid death concomitant with a severe drop in blood cell counts as well as a reduced frequency of precursor cells in the BM. However, upon HSC-specific Setd1a depletion, phenotypically defined HSCs were maintained but were functionally impaired. Setd1a-deleted HSCs lost their ability to reconstitute multiple lineages and displayed a competitive disadvantage in comparison with wild-type HSCs. As is shown by expression profiling and functional assays, the role of SETD1A can be primarily attributed to the expression of genes crucial for DNA damage detection and repair. HSCs lacking SETD1A accumulate DNA breaks and consequently fail to recover under conditions of replicative stress such as transplantation or irradiation-induced regeneration. We conclude that SETD1A protects HSCs from DNA damage-induced loss of functionality and productivity. The lethality provoked by R.26-CreERT2-mediated conditional deletion indicates that SETD1A is also required later in hematopoiesis. However, SETD1A is not simply required in all cell types, because conditional deletion using CD4-Cre had no impact.

SETD1A is not required for maintaining phenotypic LT-HSCs in situ. However, it is critically important for the maintenance of HSC identity characterized by gene expression and function, and viability in the absence of LT-HSCs is consistent with the survival of mice after loss of most LT-HSCs under steady-state conditions.³⁷  LT-HSC-specific loss of SETD1A was accompanied by the absence of HSPC function under stress situations in situ and after transplantation. This phenotype is similar to the loss of DNA repair genes in hematopoiesis. In particular, HSCs lacking DNA damage-related genes involved in homologous recombination (eg, Fancd2, Fancc), nonhomologous end joining (eg, Ku80, Lig4), mismatch repair (eg, Ercc1, Xpd), or DNA damage sensing (eg, ATM, ATR) display a dramatic loss of their reconstitution potential. However, young mutant mice contain HSCs under homeostatic conditions.^38-40  Indeed, the balance between the imperative to preserve the genomic integrity of stem cells and the need to retain stem cell pools has promoted studies on the role of the DNA damage response in stem cell decisions.^6,41  To this discussion we now add the critical role that SETD1A plays in HSCs to sustain the expression of genes involved in the maintenance of genomic integrity and the possibility that it also regulates aspects of stem cell proliferation. Possibly, SETD1A sits at the center of this balance and integrates the outcome.

Hematopoiesis-specific depletion of MLL1,¹⁶  MLL4,¹⁹  DPY30,²⁰  or CFP1²¹  resulted in comparable HSC phenotypes characterized by loss of stem cell identity and accumulation of nonfunctional KSL cells, together with lack of repopulation activity. Hematopoiesis in the absence of MLL5 (Kmt2e), a methyltransferase-incompetent SET domain protein, also showed a comparable stem cell phenotype.⁴²  Detailed analysis of these conditional knockouts assigned the defects to deregulated expression of specific self-renewal genes, including Mecom and Prdm16 in the absence of the H3K4 HMT Mll1, lineage-determining genes such as Lmo-2 in the absence of Dpy30, or genes that protect from oxidative stress, as has been described in Mll4- and Mll5-deficient mice.^19,20,43  In Mll4- and Mll5-deficient HSCs the accumulation of ROS and resulting DNA damage was suggested as the underlying reason for the decline of HSC function. In contrast, HSC dysfunction in the absence of SETD1A occurred independently from the accumulation of mitochondrial ROS. Many genes that are important for virtually all DNA repair pathways were downregulated, and consequently SETD1A-deficient HSPCs accumulated DNA breaks.

The functional relevance of the 6 Set1/Trithorax-type methyltransferases has usually been equated with H3K4 methylation, which is the principal epigenetic characteristic of active chromatin. However, the H3K4 methyltransferase activity of MLL1 has no relevance to its essential role in definitive hematopoiesis and leukemiogenesis.⁴⁴  Similarly, MLL3 and MLL4 do not rely upon H3K4 methylation to act as coactivators.⁴⁵  In contrast, the H3K4 methyltransferase activity of SETD1A may be integral to its function. In cancer cell lines⁴⁶  and ES cells,²²  SETD1A is the major H3K4 methyltransferase, as is the sole Set1 homolog in flies.^47,48  SETD1A is also the major H3K4 methyltransferase in KSL cells (Figure 1H; supplemental Figure 2B). However, it is important to note that it is not essential in all cell types, as we show here for thymocytes and was shown previously in the early embryo.²²  DPY30 is a common component of all 6 Set1/Mll complexes, and its conditional knockout in hematopoiesis resulted in reduced H3K4 mono-, di-, and trimethylation.²⁰  Our results imply that DPY30-mediated global loss of H3K4 methylation can be assigned to the SETD1A complex. However, whether the methyltransferase activity is decisive for SETD1A action in HSCs and later hematopoiesis requires further studies with SETD1A SET domain mutations. Recently, we reported that SETD1B regulates gene expression of a specific aspect of late oogenesis.⁴⁹  Here we report that SETD1A regulates the expression of many genes involved in DNA repair in HSCs. Among the enigmatic functions of the H3K4 methyltransferases, these findings indicate that they can serve as regulators of specific gene-expression programs.

The authors thank the staff of the animal houses at the Max Planck Institute and Medical Theoretical Center for great care of mouse husbandry and the Core Facility Cellular Imaging for their help. The authors are grateful to Joachim Göthert for providing SCL-CreER^T mice. The authors thank Patricia Ernst for advice on the experiments and critical reading of the manuscript.

This work was supported by the German Research Foundation (WA2837, FOR2033-A03, TRR127-A5) (C.W.) and by a grant from the Else Kröner-Fresenius-Stiftung (2013_A262) (C.W., A.K., and A.F.S.). M.D.V. is a Helmholtz Young Investigators Group leader (Helmholtz Association).

Contribution: K.A., A.K., J.F., and A.S.B. performed experiments and analyzed data; K.A., A.K., A.F.S., and C.W. designed experiments and interpreted data; M.D.V. interpreted data; M.L. and A.D. performed next-generation sequencing and data analysis; A.J. and T.H. conducted gene set enrichment analysis; A.K. and C.W. conceived of the study; and K.A., A.F.S., and C.W. wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Claudia Waskow, Regeneration in Hematopoiesis, Leibniz Institute on Aging, Fritz Lipmann Institute, Beutenbergstr 11, 07745 Jena, Germany; e-mail: claudia.waskow@leibniz-fli.de.