KMT2D Deficiency Promotes Myeloid Leukemias which Is Vulnerable to Ribosome Biogenesis Inhibition

Abstract KMT2C and KMT2D are the most frequently mutated epigenetic genes in human cancers. While KMT2C is identified as a tumor suppressor in acute myeloid leukemia (AML), the role of KMT2D remains unclear in this disease, though its loss promotes B cell lymphoma and various solid cancers. Here, it is reported that KMT2D is downregulated or mutated in AML and its deficiency, through shRNA knockdown or CRISPR/Cas9 editing, accelerates leukemogenesis in mice. Hematopoietic stem and progenitor cells and AML cells with Kmt2d loss have significantly enhanced ribosome biogenesis and consistently, enlarged nucleolus, increased rRNA and protein synthesis rates. Mechanistically, it is found that KMT2D deficiency leads to the activation of the mTOR pathway in both mouse and human AML cells. Kmt2d directly regulates the expression of Ddit4, a negative regulator of the mTOR pathway. Consistent with the abnormal ribosome biogenesis, it is shown that CX‐5461, an inhibitor of RNA polymerase I, significantly restrains the growth of AML with Kmt2d loss in vivo and extends the survival of leukemic mice. These studies validate KMT2D as a de facto tumor suppressor in AML and reveal an unprecedented vulnerability to ribosome biogenesis inhibition.


Kmt2d Downregulation Promotes Acute Myeloid Leukemogenesis
Investigating the expression profiles of KMT2D in AML patients, we found that compared to normal control (CD34 + cord blood samples, n = 17), AML samples (n = 43) contained significantly lower KMT2D expression (GSE48173 [28] ). A similar observation was obtained from another cohort (GSE1159: [29] 285 AML samples vs five normal bone marrow and three CD34 + cell samples; Figure 1A and Table S1, Supporting Information). Moreover, AML patients with lower KMT2D expression levels were associated with shorter overall survival, according to the analysis from the TCGA [4] or Beat [30] AML cohort ( Figure 1B and Table S1, Supporting Information). These seemingly generally reduced expressions of KMT2D in AML suggest that KMT2D deficiency may contribute to AML development.
To investigate the role of KMT2D in leukemogenesis, we developed a transplantation-based mouse model with RNA interference approach. Both shRNAs effectively repressing Kmt2d (shKmt2d) in mouse cells were cloned and confirmed by quantita-tive real-time polymerase chain reaction (qRT-PCR, Figure S1A, Supporting Information). Since KMT2D mutations in patients with leukemia co-occur with TP53 and NF1 mutations, we tested the effects of Kmt2d deficiency in mice in the context of Trp53 and Nf1 deletion ( Figure S1B and Table S1, Supporting Information). We transplanted c-Kit + HSPCs infected with GFP-linked shKmt2d or control Renilla shRNA (shRen), together with Trp53 and Nf1 loss, into sub-lethally irradiated syngeneic wild-type C57BL/6 recipient mice ( Figure 1C and Figure S1C, Supporting Information). After transplantation, recipient mice were monitored weekly. Compared to control recipient mice, mice transplanted with Trp53 −/− ; shNf1; shKmt2d HSPCs (hereafter referred to as TNK) developed AML significantly faster with shorter overall survival (shKmt2d_#1: median 47 days after transplantation; p = 0.0004 and shKmt2d_#2: median 74 days after transplantation; p = 0.0094; Figure 1D).
Recipients with shKmt2d HSPCs displayed significantly increased peripheral white blood cell (WBC) counts compared to controls 7 weeks after transplantation, while the former had variably decreased hemoglobin (Hb) and platelet (PLT) levels, indicative of leukemic outgrowth with the suppression of normal hematopoiesis ( Figure 1E). Flow cytometry results showed that GFP-linked shKmt2d was enriched in leukemia cells, indicating a selective advantage of cells with Kmt2d knockdown during leukemia development ( Figure 1F). All recipients with Kmt2d knockdown developed AML with neoplastic cells expressing stem and progenitor marker c-Kit as well as myeloid surface markers CD11b/Gr-1 ( Figure 1F), and peripheral blood smears showing leukocytosis with increased numbers of neutrophils, monocytes, and blasts, except three mice in one hairpin shKmt2d (shKmt2d_#2) who had mixed lineage leukemia ( Figure 1G). Sacrificed mice showed significant hepatomegaly (mean liver weight, shKmt2d_#1: 3.387 g; shKmt2d_#2: 2.116 g) and splenomegaly (mean spleen weight, shKmt2d_#1: 1.187 g; shKmt2d_#2: 0.618 g) because of extramedullary hematopoiesis and leukemia infiltration. Hematoxylin and eosin (H&E) staining of the spleen, liver, and bone marrow revealed prominent leukemia with the disruption of normal architecture ( Figure 1G). Harvested bone marrow cells enriched with leukemia cells could generate AML in secondary recipient mice ( Figure 1H and Figure S1D,E, Supporting Information). Control shRen recipients developed AML eventually with similar immunophenotype and histopathology, despite a longer tumor formation time.
We confirmed that AML generated above is driven by Kmt2d suppression. First, Kmt2d mRNA levels in harvested leukemia cells were dramatically reduced ( Figure 1I). Then, since KMT2D is a writer of H3K4 mono-and di-methylation, we also noticed the consistent reduction of H3K4me1 and H3K4me2 levels in leukemia cells with Kmt2d knockdown, indicating on-target effects of shKmt2d ( Figure 1J). Hence, our results demonstrated that suppression of Kmt2d promotes AML genesis in mice.

Figure 1.
Kmt2d deficiency by shRNAs promotes AML in mice. A) Expression levels of KMT2D in AML and normal samples. Left, data were analyzed from RNA-seq data (GSE48173) with 17 CD34 + cord blood and 43 AML samples; Right, data were analyzed from microarray data (GSE1159) with eight healthy donors (five normal bone marrow and three CD34 + cell samples) and 285 AML samples; *p < 0.05, ***p < 0.001 (two-tailed Wilcoxon rank-sum test). B) Survival curves of AML patients stratified by high and low expression of KMT2D in the TCGA-LAML (left) and Beat AML (right) cohort, respectively. The cut-off values were determined by maximally selected rank statistics. p Values were determined by the log-rank test. C) Schematic experimental design for mouse modeling using shRNA technique. Trp53 −/− mouse HSPCs were transduced with GFP-linked shKmt2d/shRen and mCherry-linked shNf1, and then transplanted into sub-lethally irradiated syngeneic mice. D) Kaplan-Meier survival curves of mice transplanted with Trp53 −/− HSPCs transduced with shNf1 and shRen (blue; n = 10), shKmt2d_#1 (red; n = 10), or shKmt2d_#2 (orange; n = 10). **p < 0.01, ***p < 0.001 (log-rank test). The results were the combination of two independent trials. E) White blood cell (WBC), hemoglobin (Hb), and platelet (PLT) counts of shKmt2d and shRen mice 7 weeks post-transplant. F) Representative flow cytometric profiles showing the expression of fluorescent markers (GFP and mCherry), myeloid lineage markers (CD11b and Gr-1), lymphoid lineage markers (B220 and CD3 ), and stem cell marker (c-Kit) in bone marrow cells of sacrificed TNK (Trp53 −/− ; shNf1; shKmt2d) mice. G) Representative images of histological analyses of blood, spleen, liver, and bone marrow of sacrificed TNK mice. H) Kaplan-Meier survival curves of secondary transplants from two independent primary leukemia cells of each TNK mouse (n = 4 per group). I) Relative mRNA levels of Kmt2d in bone marrow cells of sacrificed control TN (Trp53 −/− ; shNf1; shRen) and TNK mice were quantified by qRT-PCR (normalized to Actin; n = 3 per group). J) Western blotting analyses showing the H3K4me1 and H3K4me2 levels in bone marrow cell lysates from sacrificed TN and TNK mice. E,I) Graph represents the mean ± SD; *p < 0.05, **p < 0.01, ***p < 0.001, ns, not significant (unpaired two-tailed t-test).
knock-in transgenic mice, followed by transplantation into 5.5 Gy irradiated syngeneic wild-type mice (Figure 2A). A sgRNA targeting Cas9 (sgCas9) was also used to prevent potential adverse effects associated with constitutive Cas9 expression.
Recipients carrying Trp53 −/− ; sgNf1; sgKmt2d; sgCas9 HSPCs (hereafter referred to as TNKC) died of AML at a median survival of 59 days, while none of the control sgScramble recipients developed diseases during the observation period ( Figure 2B). Consistent with shKmt2d AML developed above, Kmt2d mutants accelerated AML development evidenced by leukocytosis, anemia, and thrombocytopenia in recipients of sgKmt2d HSPCs compared to controls 2 months post-transplant ( Figure 2C). Harvested bone marrow cells from moribund sgKmt2d mice were CD11b/Gr-1 + and c-Kit + ( Figure 2D). Blood smear and H&E staining of the liver, spleen, and bone marrow showed accumulated blasts and aggressive leukemia cell infiltration ( Figure 2E). In the harvested leukemia cells, Kmt2d mutations were confirmed by T7 endonuclease I mismatch detection assay and sanger sequences ( Figure 2F,G). Of note, all Kmt2d detected mutations were truncating, supporting a loss-of-function mechanism of Kmt2d in AML.

Kmt2d Deficiency Upregulates Ribosome Biogenesis
To investigate the role of Kmt2d in AML, we generated a mouse inducible shKmt2d-driven AML with the Tet-ON system. In this system, shKmt2d with the fluorescence gene dsRed is regulated by the tetracycline-inducible promoter TRE. Treatment with doxycycline activates the TRE promoter resulting in the transcription of shKmt2d (Kmt2d knockdown, KD), while a withdraw time of 4 days shuts down the TRE promoter and shKmt2d transcription (Kmt2d restored, RS, Figure S3A, Supporting Information). shKmt2d AML cells presented with venus and dsRed double-positive population by flow cytometry and the expression of Kmt2d was reduced, assessed by qRT-PCR ( Figure 3A and Figure S3B, Supporting Information). Remarkably, repressing Kmt2d expression resulted in a significant growth increment of AML cells ( Figure 3B). Moreover, both cell size and nuclear size were enlarged significantly in Kmt2d-deficient AML cells (Figure 3C).
Since cell proliferation, cell size, and carcinogenesis have a close relationship with the upregulated ribosome biogenesis, we detected cellular ribosomal function in Kmt2d-deficient AML cells. [31][32][33][34] Nucleoli are the sites to produce and assemble ribosomes. Studies have shown the connection between structural and functional alterations of nucleoli and tumorigenesis, where increased ribosome biogenesis is associated with larger nucleoli and malignant tumors generally have larger ones. [35,36] From the analysis of transmission electron microscopic (TEM) images and immunofluorescence staining using an antibody against nucleolar protein fibrillarin, shKmt2d AML cells contained significantly enlarged nucleoli and brighter ribosome biogenesis key molecule fibrillarin staining, compared to control cells ( Figure 3D,E). Ribosomes are comprised of ribosomal proteins and ribosomal RNAs (rRNAs). We evaluated the rRNA synthesis by the immunofluorescence detection of 5-fluorouridine (5-FUrd) incorporation into nascent rRNAs. Cells were pulse-labeled for 30 min with 5-FUrd and immunostained with an antibody against BrdU. Results showed that Kmt2d deficiency greatly enhanced rRNA transcription ( Figure 3F). Further, we estimated the total rRNA concentration by nondenaturing agarose gel electrophoresis and observed that rRNA levels were elevated when Kmt2d knockdown ( Figure S3C, Supporting Information). Quantitatively, both 18S and 28S rRNA transcription levels were increased in Kmt2d knockdown AML cells, compared to Kmt2d restored cells ( Figure 3G). To investigate the function of increased ribosomes, we measured the newly synthesized peptides with O-propargyl-puromycin (OPP) protein synthesis assays. Briefly, OPP, a membrane-permeable puromycin analog, was added to leukemia cells and then measured by flow cytometry. OPP exerts its inhibition by incorporating into nascent peptides, disrupting peptides transfer on ribosomes, and causing premature chain termination during translation. As a result, Kmt2d-deficient AML cells contained stronger OPP signals indicating higher translational activity, compared to Kmt2d restored AML cells ( Figure 3H). Thus, our results demonstrated that Kmt2d deficiency upregulates ribosome biogenesis and increases protein synthesis.
To further get insight into the alterations in ribosome biogenesis, we performed RNA sequencing (RNA-seq) to profile the transcriptomes of Kmt2d knockdown versus restored AML cells. As expected, Kmt2d deficiency induced wide gene expression changes, with more significantly downregulated genes (945 down vs 499 up genes, absolute log2-fold change > 0.5, p < 0.05), in line with the impact of H3K4 methylation on activating gene expressions ( Figure S3D and Table S2, Supporting Information). Gene ontology (GO) analyses showed that downregulated genes in shKmt2d AML cells were highly enriched in pathways including hematopoietic cell differentiation, while upregulated genes were highly enriched in pathways including ribosome biogenesis, ribonucleoprotein complex biogenesis, and rRNA metabolic process ( Figure 3I and Figure S3E, Supporting Information). Gene set enrichment analyses (GSEA) further confirmed the enrichment of genes implicated in the aforementioned biological processes, with the gene signature of KEGG_RIBOSOME, GO_NUCLEOLAR_PART, GO_RRNA_METABOLIC_PROCESS, and REAC-TOME_TRANSLATION being significantly positively enriched in Kmt2d knockdown AML cells compared to Kmt2d restored cells ( Figure 3J). The upregulated expressions of several ribosome biogenesis-related genes like Rpp40, Tbl3, and Fbl in shKmt2d AML cells have been validated by qRT-PCR ( Figure S3F, Supporting Information). A similar correlation between Kmt2d deficiency and upregulated ribosome biogenesis was observed in HSPCs (Figure S3G-P and Table S3, Supporting Information). Interestingly, the higher ribosome biogenesis correlated with Kmt2d deficiency was not observed in Kmt2c, another member of the COMPASS-like complex, -deficient HSPCs ( Figure S3Q-S, Supporting Information).

Kmt2d Directly Regulates Ddit4, Encoding a Negative Regulator of the mTOR Pathway
To explore the molecular mechanism under which Kmt2d regulates ribosome biogenesis, we sought to conduct multiomics analyses to identify downstream pathways and targets of Kmt2d. First, digging deeper into RNA-seq data using GSEA, we found that genes in the HALLMARK_MTORC1_SIGNALING pathway were significantly positively enriched in shKmt2d leukemia cells and HSPCs ( Figure 4A and Figure S4A, Supporting Information). Given the role of the mammalian target of rapamycin complex 1 (mTORC1) in promoting ribosome biogenesis, [37,38] we hypothesized that mTOR activity may underlie Kmt2d deficiencyinduced ribosome upregulation. The activation of the mTOR pathway was confirmed in Kmt2d-deficient leukemia cells, as shown by Western blots with antiphosphorylated ribosomal protein S6, a substrate of direct mTORC1 target S6K1 ( Figure 4B). The inhibition of mTOR by rapamycin reduced the fluorescence intensity of fibrillarin staining in shKmt2d AML cells, indicating the repression of ribosome biogenesis ( Figure 4C). Further, rapamycin-treated leukemia cells were presented with smaller cell and nuclear sizes, pharmacologically reversing the effect of Kmt2d knockdown, further providing experimental evidence for the role of the mTOR signaling pathway played in the tumorigenesis ( Figure 4D).
To assess how the mTOR signaling pathway is highly activated by Kmt2d deficiency and given that KMT2D, as a histone lysine N-methyltransferase, is required for H3K4 mono-and dimethylation, we performed the cleavage under targets and tagmentation (CUT&Tag) technique to investigate the corresponding H3K4me1, H3K4me2, or H3K27ac levels in Kmt2d restored and knockdown AML cells. Consistent with the KMT2D enzymatic activity, Kmt2d-deficient AML cells had a decrease in differentially modified peak numbers of H3K4me1, H3K4me2, and H3K27ac. Specifically, in comparison with the control group, Kmt2d-deficient AML cells exhibited 6680, 2592, and 1486 significantly lower H3K4me1, H3K4me2, and H3K27ac modified peaks, respectively (p < 0.05, log2-fold change < −1; Figure S4B and Table S4, Supporting Information). ATAC-seq also demonstrated a great genome accessibility change, with 1064 decreased chromatin accessibility sites in shKmt2d AML cells compared to the control group (p < 0.05, log2-fold change < −0.5; Figure S4C and Table S4, Supporting Information). Remarkably, despite Kmt2d deficiency reprogrammed epigenetic landscape in AML cells with decreases in differentially modified peak numbers and average signal levels of gene-activating mark H3K27ac as well as chromatin accessibility globally, RNA polymerase I (Pol I) and RNA polymerase III (Pol III) bound regions, which specialize in the transcription of rRNA, were marked with elevated H3K27ac modifications and possessed increased genome accessibility levels in Kmt2d-deficient AML cells, concordant with increased gene expression and enhanced rRNA concentration (Figure S4D, Supporting Information). Meanwhile, we analyzed the KMT2D binding, H3K4me1, or H3K4me2 modification levels in Pol I & III binding sites and found that the KMT2D binding levels were dramatically reduced in rDNA regions compared to those in the whole genome. Similar results were obtained for the H3K4me1 or H3K4me2 modification levels ( Figure S4E, Supporting Information). Further, we found that there was little difference in the KMT2D-binding, H3K4me1, or H3K4me2 levels between Kmt2d knockdown and restored AML cells in Pol I or III binding sites ( Figure S4F, Supporting Information). These data suggest that the role of KMT2D in rDNA expression is not likely through a direct effect on rDNA-associated chromatin. By comparing the KMT2D binding and histone modification levels between Kmt2d restored and knockdown AML cells, we found that 2516 peaks had a significant decrease of KMT2D binding levels and contained reduced H3K4me1, H3K4me2, and H3K27ac modifications in Kmt2d-deficient AML cells, which indicated these regions were specifically bound by KMT2D (p < 0.05, log2-fold change < −1; Figure S4G and Table S4, Supporting Information). These KMT2D-specific binding sites were preferentially distributed in distal intergenic (38.92%), intron (38.31%), and promoter regions (13.20%), similar to previously reported [15,16,21,39,40] (Figure S4H, Supporting Information). KMT2D-targeted genes significantly overlapped with genes that had decreased H3K4me1 and H3K4me2 modification levels in shKmt2d AML cells. Specifically, 730 genes were identified with significant KMT2D, H3K4me1, and H3K4me2 downregulation ( Figure S4I, Supporting Information). These genes also possessed significant decreases in H3K27ac modification, chromatin accessibility, and gene expression levels ( Figure S4J, Supporting Information). Among them, 178 genes had significantly reduced expression levels and were considered as KMT2D directly regulated genes, in which we found one encoding a negative regulator of mTORC1, Ddit4 [41] (Figure S4I, Supporting Information). KMT2D bound the near transcriptional start site (TSS) and the region around 20 kb upstream of the TSS of Ddit4, and caused a significant reduction in H3K4me1 and H3K4me2 occupations as well as gene expression ( Figure 4E). The reduction of Ddit4 expression was confirmed by qRT-PCR ( Figure 4F). To prove the role of Ddit4 participated in the Kmt2d-deficiencymediated hyperactivation of the mTOR pathway, we conduct a rescue experiment by overexpressing Ddit4 in Kmt2d-deficient AML cells ( Figure S4K, Supporting Information). As a consequence, overexpressed Ddit4 rescued almost all of the phenotypes associated with Kmt2d deficiency. Specifically, mTOR activation was inhibited evidenced by the impediment to the increase of phosphorylated ribosomal protein S6 ( Figure 4G). Ddit4 over-expression reduced the fluorescence intensity of nucleolar protein fibrillarin, rRNA transcriptions, and protein synthesis rate ( Figure 4H-J). Eventually, we observed a significant inhibition of cell proliferation and a reduction of cell and nuclear sizes by Ddit4 overexpression ( Figure 4K and Figure S4L, Supporting Information). Hence, Kmt2d might directly control the expression of mTOR negative regulator Ddit4, whose deficiency activates the mTOR pathway and thus induces ribosome biogenesis.

Kmt2d-Deficient AML Cells Are Sensitive to the Inhibitor of Ribosome Biogenesis
To investigate the impact of ribosome biogenesis on Kmt2ddeficient leukemia cell growth, we treated mice bearing shKmt2d AML with CX-5461, an inhibitor of rRNA synthesis (Figure 5A). CX-5461 could directly and selectively target Pol I-mediated transcription by disrupting the binding of the SL1 transcription factor to the promoter of ribosomal RNA genes. [42] CX-5461 treatment significantly prolonged the survival of recipient mice transplanted with shKmt2d AML cells compared to the vehicle-treated group ( Figure 5B). We repeated this experiment and harvested recipient mice simultaneously 4 weeks after the transplant. We www.advancedsciencenews.com www.advancedscience.com found that shKmt2d leukemia cells in the peripheral blood of CX-5461-treated mice were significantly reduced compared to those in control vehicle-treated mice, as shown by decreased WBC counts and the percentages of GFP and mCherry doublepositive population ( Figure 5C,D). Blast cells were also dramatically reduced as shown in the blood smears and bone marrow cytospins ( Figure 5E). The harvested spleens and livers of CX-5461treated mice had reduced size, weight, and a diminished infiltration of leukemia cells as examined by H&E staining, compared to splenomegaly and hepatomegaly correlated with severe infiltration of leukemia cells in the vehicle-treated group ( Figure 5F,G). We also compared the response to CX-5461 in Kmt2d-deficient leukemia cells (TNK) and normal Kmt2d control leukemia cells (TN) side by side. Results showed that Kmt2d-deficient AML cells were more sensitive to the CX-5461 treatment ( Figure S5A, Supporting Information). Taken together, our data showed that CX-5461 is an effective drug for treating Kmt2d-deficient AML, indicating a ribosome biogenesis vulnerability.

KMT2D Deficiency Upregulates Ribosome Biogenesis in Human AML
To translate our findings into human settings, we analyzed the transcriptomic profiles of 142 AML patients in the TCGA-LAML cohort. [4] GSEA results showed that upregulated genes in KMT2D low expression AML patients were significantly positively enriched in GO_ribosome_biogenesis, GO_ribosome, GO_nucleolar_part, GO_rRNA_metabolic_process, HALL-MARK_mTORC1_signaling, GO_translational_initiation, and GO_translational_elongation pathways, compared to KMT2D high expression ones ( Figure 6A and Figure S6A, Supporting Information). Consistently, KMT2D expression was negatively correlated with the expressions of ribosomal biogenesis-related genes ( Figure 6B).
Additionally, we also investigated the correlation between KMT2D and ribosome biogenesis in human AML cell line MOLM-13. We first constructed KMT2D-deficient MOLM-13 cells with CRISPR/Cas9 genome editing technology (Figure S6B,C, Supporting Information). Consistent with our previous findings, upregulated ribosome biogenesis was observed in KMT2D-mutated MOLM-13 cells. Compared to control cells (sgScramble), we found that in KMT2D-mutated cells, cell and nuclear sizes were enlarged ( Figure 6C); fibrillarin staining was significantly increased ( Figure 6D); rRNA concentration was enhanced ( Figure S6D, Supporting Information). More importantly, translation activity in KMT2D-mutated MOLM-13 cells was enhanced as measured by OPP incorporation ( Figure 6E). Furthermore, we observed that the mTOR pathway was activated in sgKMT2D cells, as shown by increased phosphorylated S6 levels ( Figure 6F), and rapamycin treatment reduced the cell and nuclear size ( Figure S6E, Supporting Information). Altogether, these results supported that our findings of KMT2D-deficiencyinducing ribosome biogenesis are conserved in human AML samples.
In summary, our study identified KMT2D as a tumor suppressor gene, whose deficiency was critical for AML tumorigenesis. KMT2D deficiency reduced H3K4 methylation levels and suppressed the expression of DDIT4, the negative regulator in the mTOR signaling pathway, which led to the activation of the mTOR pathway and enhancement of ribosome biogenesis, thus contributing to AML development ( Figure 6G).

Discussion
KMT2D is one of the most frequently mutated genes in human cancers. Its loss-of-function mutations or loss have been demonstrated as driving forces for the initiation and development of many cancers like B-cell lymphoma, melanoma, medulloblastoma, lung, and pancreas cancers. [15][16][17][18][19][20][21] In spite of the wellestablished tumor suppressive effect of KMT2D in lymphoma and solid cancers as well as KMT2C and KDM6A encoding components in the COMPASS-like complex were identified as tumor suppressor genes in AML, the role of KMT2D in myeloid malignancies seems the opposite. Nussenzweig and Hess's groups found that Kmt2d deficiency prevents the MLL-AF9-or HOXA9driven AML formation in mice. [26,27] They further illustrated that Kmt2d deficiency induced myeloid differentiation in MLL-AF9 leukemic blasts, suggesting a tumor-promoting role of Kmt2d. [26] However, we found that KMT2D expressions were generally lower in AML cells compared to normal peripheral blood cells, suggesting that KMT2D may have a tumor-suppressive function in some AMLs. Functional studies demonstrated that Kmt2d deficiency together with Trp53 and Nf1 loss in hematopoietic stem and progenitor cells could promote AML in recipient mice, supporting that KMT2D is a tumor suppressor in TP53 −/− AML. One gene, having a distinct or even opposite impact on AML with different genetic alterations or in different phases, has been reported before. SETD2 loss drove leukemogenesis when cooperated with NUP93-HOXD13, [43] while inhibiting MLL-AF9 AML cell growth both in vitro and in vivo. [44] Ezh2 acts as a tumor suppressor during AML induction, while it exerts an oncogenic function during disease maintenance. [45] As an important tumor suppressor gene, KMT2D regulates the expressions of genes involved in multiple cellular functions like differentiation and metabolism reprogramming in cancer cells. [17,19,26] Here, we revealed a new role of KMT2D, enhancing ribosome biogenesis when repressed. We found that when Kmt2d was suppressed, the synthesis of ribosome RNA, the expressions of ribosome biogenesis-related genes and genes encoding RNA polymerase I or III complexes were increased. Consequently, the nucleolus was enlarged and more newly synthesized peptides were produced in Kmt2d-deficient cells. This correlation between KMT2D defect and enhanced ribosome biogenesis is consistent in human AML cell lines and AML patients. From the multiomics analysis, we observed that the hyperactivated mTOR pathway was associated with KMT2D deficiency, and mTORC1 inhibitor rapamycin could reduce ribosome biogenesis. The mTORC1 signaling pathway has been reported to control ribosome biogenesis in multiple steps, including ribosome protein translation and ribosome RNA transcription. [38,46] We identified that KMT2D could bind to and regulate the expression of Ddit4, encoding a negative regulator of mTORC1. The overexpression of Ddit4 could repress Kmt2d deficiencyinduced upregulation of ribosome biogenesis, suggesting that KMT2D has an impact on translation likely through DDIT4-mTOR. Meanwhile, we would not exclude other possible factors and pathways that may be involved. We noticed that genes in MYC_TARGETS were significantly positively enriched in Kmt2ddeficient HSPCs and AML cells (unpublished data). Therefore, KMT2D may also regulate ribosome biogenesis and leukemogenesis through mTOR-independent mechanisms. Altogether, we illustrated a novel function of KMT2D that could regulate protein translation in addition to its critical role in controlling gene transcription.
Furthermore, our work revealed that CX-5461, a specific RNA polymerase I inhibitor, could effectively reduce tumor burden and significantly prolong the survival of mice bearing Kmt2ddeficient AML. The drug experiment not only confirmed the important role of ribosome biogenesis in Kmt2d-deficiency-induced AML, but also suggested a potential therapeutic target for KMT2D low-expression or mutated AML patients. Ribosome biogenesis has emerged as an effective pathway for cancer therapeutics. Apart from chemotherapeutic drugs which exert their cytotoxic effects by perturbation of ribosome biogenesis at various levels, several novel compounds that selectively target ribosome production or function, mainly inhibitors of Pol I transcription, such as CX-5461, have recently entered clinical trials. [47][48][49] CX-5461 has shown therapeutic potential in hematological malignancies. CX-5461 treatment could effectively inhibit mouse MLL-fusion AML progression and human AML cell lines in both Trp53-dependent and -independent mechanisms. [50,51] www.advancedsciencenews.com www.advancedscience.com Also, combination therapy targeting ribosome biogenesis and mTORC1-dependent translation synergistically extends survival in MYC-driven lymphoma. [52] Thus, several phase I clinical trials of CX-5461 in B-cell lymphomas and solid cancers have been done [53] or are ongoing (NCT02719977, NCT04890613, and NCT05425862). According to these studies, CX-5461 treatment in patients is safe and holds promise. [53] Further validation of CX-5461 and other inhibitors with similar functions in human cancer cells with KMT2D mutations or deficiency would pave the way for their potential applications.
In summary, we demonstrate that KMT2D could be a tumor suppressor gene in AML. KMT2D regulates the expression of a negative regulator DDIT4 in the mTOR pathway via histone methyltransferase activity, suppresses ribosome biogenesis, and eventually prevents leukemogenesis. CX-5461, the selective inhibitor of RNA Pol I transcription, has an antileukemia effect on Kmt2d-deficient AML. Besides, KMT2D mutations and abnormalities are also common in other leukemias. [54,55] As such, our observations may have implications beyond AML.
cDNA Cloning: Ddit4 cDNA sequences were obtained from the cDNA library of mouse AML cells and cloned into retroviral vector MSCV-cDNA-IRES-GFP.
Flow Cytometry: Flow cytometry analyses were performed on the BD LSRFortessa Flow Cytometer (BD Biosciences, San Jose, CA). Data were analyzed using FlowJo software (RRID:SCR_008520). [58] Antibodies used in flow cytometry are displayed in Table S8 in the Supporting Information.
OPP Protein Synthesis Assay: Nascent protein synthesis was detected using Click-iT Plus OPP Protein Synthesis Assays kit (Cat# C10458, Thermo Fisher Scientific) following the manufacturer's instructions. The intensity of OPP signals was detected by flow cytometry analyses and presented as the mean fluorescence index.
In Vivo Treatment: 10E+6 bone marrow leukemia cells were injected into 5.5 Gy irradiated C57BL/6 recipient mice. Drug treatments were initiated 10 days after transplantation. Mice were treated with either vehicle (0.5% CMC-Na) or 40 mg kg −1 CX-5461 (Cat# S2684, Selleck, Shanghai, China) every 3 days via oral gavage. Administration timing can be adjusted according to the physical condition of the mice. The leukemia progression in recipient mice was monitored by CBC, flow cytometry, and blood smears.
ATAC-seq Analysis: Library preparation was performed as previously described, [67] and the transposase was from TruePrep DNA Library Prep Kit V2 for Illumina (Cat# TD501, Vazyme, Nanjing, China). The library was sequenced on the Illumina NovaSeq 6000 platform. NGmerge was performed to remove adapters from 150 bp paired-end raw data with "NGmerge -a -v -n 20" option. [68] The mouse genome (mm10) index construction and follow-up alignments were performed using Bowtie2 (RRID:SCR_016368) with "-very-sensitive -X2000 -x mm10" option. Duplicates were removed using MarkDuplicates tools from Picard (RRID:SCR_006525). The mitochondrial genome contamination was removed and bam files were obtained using SAMtools (RRID:SCR_002105) and awk commands. Normalized bw files were generated using Deep-Tools (RRID:SCR_016366) with "bamCoverage -bs = 1 -normalizeUsing BPM" option for further IGV (RRID:SCR_011793) visualization. The standard workflow of HMMRATAC [69] was utilized for the ATAC-seq peak calling step. The genomic distribution of accessibility sites was identified by ChIPseeker (RRID:SCR_021322) with "annotatePeak" option and the TSS region "tssRegion" was set as (−3000, 3000). "gappedPeak" files were transformed into GrangesList forms and peaks were converted into consensus counts. Data normalization and difference comparison were performed using DESeq2 (RRID:SCR_015687). Pol I and Pol III binding regions were cited from GSE145874. [70] Peaks with a p-value cutoff of 0.05 (Wald test) were identified as differential accessibility regions for downstream analyses.
Referred AML Patient Transcriptome Data Analysis: Genetic alteration data were analyzed from The Cancer Genome Atlas AML project (TCGA-LAML) [4] and OHSU AML cohorts. [30] The co-occurrence of KMT2D, TP53, and NF1 mutations in leukemia patients was analyzed by GENIE Cohort v11.0-public datasets (https://genie.cbioportal.org/, n = 4670). Tran-scriptome data of normal and AML patients were acquired from highthroughput sequencing data GSE48173 (AML, n = 43; normal, n = 17) [28] and microarray data GSE1159 (AML, n = 285; normal, n = 8). [29] RPKM of GSE48173 was performed with log2 transformation to obtain the standardized expression value. p value was determined by the Wilcoxon signedrank test. To estimate the impact of KMT2D expression on the prognosis of AML patients, patients were divided into KMT2D-high and KMT2Dlow groups by FPKM in the TCGA-LAML [4] and Beat AML cohorts. [30] The optimal cut-point for numerical variables of KMT2D expression was calculated with the maximally selected rank statistics by the survminer (RRID:SCR_021094) package to stratify patients for the survival analysis, set as 20.78 and 7.12 in TCGA-LAML and Beat AML cohort, respectively. According to the algorithm, 122 and 20 patients in the TCGA-LAML cohort were divided into KMT2D-high and KMT2D-low groups. GSEA (RRID:SCR_003199) was performed to identify significantly enriched pathways between the two groups. The ribosome biogenesis gene signature was determined by GOBP_RIBOSOME_BIOGENESIS. [71] Correlations between KMT2D expression and mean expression levels of genes involved in ribosome biogenesis were visualized by ggpubr (RRID:SCR_021139) and correlation values were determined by Spearman's rank correlation coefficient.
Data Visualization: R packages ggpubr and ggplot2 were performed to show the expression and accessibility level differences by box plots. Heatmaps were generated by the R package pheatmap to present differentially expressed genes. Bar plots and dot plots of GO and GSEA enrichment results were drawn by ggplot2. The program DeepTools was utilized to visualize the histone modification levels. Matrix computations of differential modification and accessibility peaks were performed using "ComputeMatrix reference-point -referencePoint center -b 3000 -a 3000 -skipZeros" option.
Statistical Analysis: Statistical test methods, sample sizes, and p values are indicated in the corresponding figure legends. Statistical significance was determined using GraphPad Prism (v5.01, RRID:SCR_002798) and PASW Statistics18 software by unpaired two-tailed t-test, Wilcoxon signed-rank test, Wald test, hypergeometric distribution, or log-rank test. Statistical significance was defined as p < 0.05. Error bars were shown as SD.
Study Approval: All animal study procedures and experiments were reviewed and approved by the Experimental Animal Ethics Committee of the State Key Laboratory of Biotherapy and Cancer Center, Sichuan University (approval number: 20170209) and were in accordance with the 8th edition of the Guide for the Care and Use of Laboratory Animals.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.