Cell experiments were performed using K-562 cells (RRID: CVCL_0004), a cell line derived from a 53-year-old female CML patient in blast crisis (Lozzio and Lozzio, 1975) provided by the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). Cells were maintained as previously described (Turrini et al., 2012; Kaehler et al., 2022). Imatinib-resistant replicates were obtained by exposing treatment-naïve K-562 cells to increasing concentrations of imatinib, resulting in cells resistant to 0.5 µM and 2 µM imatinib.

RNA extraction was performed using E. Z.N.A total RNA Kit I (Omega bio-tek, Norcross, Georgia, United States) following the manufacturer’s instructions with the added step of centrifuging the cell lysate within QIAshredder homogenizers (Qiagen, Hilden, Germany) for 1 min at 10,000 x g after exposure to the lysis buffer to enhance RNA extraction. DNA extraction was performed using the Gentra Puregene Kit (Qiagen).

Reverse transcription of 1 µg RNA was conducted with the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Darmstadt, Germany) according to the manufacturer’s protocol. RT-qPCR was performed with the QuantStudio 7 Flex (Thermo Fisher Scientific) applying default cycling conditions. Samples were examined in triplicates using the TaqMan Universal Master Mix without UNG (Thermo Fisher Scientific) and the following TaqMan assays obtained from Thermo Fisher Scientific: KMT2D (Hs00912419_m1), CDK4 (Hs00364847_m1), CCND3 (Hs00236949_m1), CCNE2 (Hs00180319_m1), TBP (Hs00427620_m1), GAPDH (Hs02786624_g1), 18S (Hs99999901_s1). The cycle threshold (CT) value of the target genes were normalized to the housekeeping genes TBP, GAPDH and 18S with relative mRNA expression being calculated as 2^−ΔΔCT (Livak and Schmittgen, 2001).

Amplicons of the KMT2D gene were generated using the AmpliTaq Gold 360 Mastermix (Thermo Fisher Scientific) and the primers 5′-GAT​GTC​CAT​GGC​TTT​ACC​ACT​TCC​CCT​GC-3′ and 5′-AAA​GCC​ATG​GAC​ATC​CAG​GTG​AGC​GG-3’ (obtained from Merck, Darmstadt, Germany) with an annealing temperature of 58 °C and an elongation time of 7 min. The PCR products were purified using the GeneJET Gel extraction Kit (Thermo Fisher Scientific). Next-Generation Sequencing was performed using the Nextera XT Sequencing kit (Illumina, San Diego, California, United States) adhering to the manufacturer’s protocol as previously described (Kaehler et al., 2021; Kaehler et al., 2023).

The KMT2D-encoding plasmid was provided by Promega (Cat# FHC12732, Madison, Wisconsin, United States) and the plasmid harboring the KMT2D p. (Trp191Arg) variant was obtained by mutagenesis at GenScript (Rijswijk, Netherlands). The empty pFN21A vector was obtained through restriction enzyme cloning with AsiSI and Pme1 (both New England Biolabs). Plasmid DNA was isolated using PureYield Plasmid Multiprep System (Promega) or NucleoBond Xtra Midi Kit (Macherey Nagel GmbH).

Transient transfection was performed with 4 x 10⁶ cells using the Amaxa Cell Line Nucleofector Kit V (Lonza, Basel, Switzerland) with the Nucleofector I device (Lonza) following the manufacturer’s recommendation for K-562 cells. After respective incubation periods, cell seeding was carried out to investigate cell viability under exposure to 2 µM imatinib in cellular fitness assays as described below. For the KMT2D knockdown, treatment-naïve cells were transfected with 200 nM Ambion MLL2 Silencer siRNA (Cat# AM51331) or negative control #1 siRNA (Cat# AM4611) with subsequent cell seeding after an incubation period of 6 h. Imatinib-resistant cells were transfected with 10 µg of a KMT2D-encoding plasmid or empty pFN21A vector followed by cell seeding after 1 h incubation. Furthermore, treatment-naïve K-562 cells were transfected with 10 µg of a plasmid harboring the KMT2D p. (Trp191Arg) variant, KMT2D wild-type or the empty pFN21A vector as negative control. Cells were seeded 24 h after transfection. The cells were additionally exposed to 100 µM LSD1 inhibitor or DMSO as a solvent control.

Cells were seeded into 12-well plates with 1 x 10⁶ cells/mL for the Ki-67 assay as well as immunoblotting, whereas 96-well plates were used to determine cell numbers with 2 x 10⁵ cells/200 µL medium and metabolic activity with 5 x 10⁴ cells/100 µL medium. Cells were exposed to either 2 µM imatinib or medium and incubated at 37 °C. To determine cell numbers, the cell suspension was mixed with trypan blue (Sigma Aldrich) to mark viable, unstained cells, which were then quantified with a Fuchs-Rosenthal cell counting chamber after 24 and 48 h. Metabolic activity was measured using the WST assay (Merck) as previously described (Kaehler et al., 2017).

Cell proliferation was determined 24 h after transient transfection using the Human Antigen Ki-67 ELISA Kit (Cat# MBS764543, MyBioSource, San Diego, California, United States) with 10 µg of protein according to the manufacturer’s protocol. To determine the influence of imatinib on cell viability, the results of cells treated with imatinib were normalized to treatment-naïve cells.

Cell lysis and immunoblots were performed as described elsewhere (Kaehler et al., 2017; Waetzig et al., 2019; Bruhn et al., 2020). Using 15% v/v polyacrylamide gels, 20 µg of protein were transferred onto nitrocellulose membranes and membranes were probed with the following antibodies: Histon H3: Cat# sc-517576 (Santa Cruz, Dallas, Texas, United States), RRID: AB_2848194, 1:250; H3K4me1: Cat# 710795-20UG (Thermo Fisher Scientific), RRID: AB_2848515, 1:1,000; HSP90: Cat# 4877 (Cell Signaling Technology, Danvers, Massachusetts, United States), RRID: AB_2233307, 1:1,000; anti-mouse: Cat# 926-68070, RRID: AB_10956588, Cat# 926-32210, RRID: AB_621842; anti-rabbit: Cat# 926-68071, RRID: AB_10956166, Cat# 926-32211, RRID: AB_621843; all 1:10,000, LiCOR (Bad Homburg, Germany). Primary antibodies were diluted with the Intercept TBS Blocking Buffer supplemented with 0.2% v/v Tween20, whereas secondary antibodies were diluted in TBS with 0.1% v/v Tween20.

Inhibition experiments were conducted in 96-well plates with triplicates of 5 x 10⁴ cells/100 µL medium supplemented with 2 µM imatinib with DMSO as solvent control. LSD1 was inhibited using 1–200 µM LSD1 Inhibitor II (S2101, Merck Millipore, United States). Furthermore, 1–100 µM of the histone-deacetylase (HDAC) inhibitor vorinostat (Cat# SML0061, Merck) and 0.1–250 µM of the DNA-methyltransferase (DNMT) inhibitor 5′-azacytidine (Hölzel Diagnostika, Köln, Germany) were used. After an incubation of 48 h at 37 °C, metabolic activity was measured as described above. IC50 values were calculated by non-linear regression with variable slope (four parameters) for N = 3 including at least six concentrations.

Exome sequencing data of TKI-resistant biological replicate cell lines was obtained from the European Nucleotide Archive (ENA), accession number PRJEB60565. KMT2D variants were identified as previously described (Kaehler et al., 2023). In silico prediction of the variant effect was performed using gnomAD (gnomad.broadinstitute.org). Genome-wide gene expression data was derived from the GEO datasets GSE227347 and GSE203342 as previously published (Kaehler et al., 2022; Kaehler et al., 2023). Comparing treatment-naïve and imatinib-resistant cell lines, genes with a fold change ±2 and a false discovery rate (FDR)-corrected p-value p < 0.05 were considered to be differentially expressed. Venn diagrams for the comparison of these differentially expressed genes with the KMT2D essentiality network (Takemon et al., 2024) were obtained using the PNNL software (omics.pnl.gov (Oliveros, 2007)). KEGG pathway prediction was performed using DAVID Functional Annotation Tool (DAVID Bioinformatics Resources (Huang da et al., 2009; Sherman et al., 2022)) and interaction networks using the STRING database (string-db.org. Version 12.0 with medium confidence).

Primers were designed with the NCBI primer design tool (National Center for Biotechnology Information, Bethesda, Maryland, United States). Unless indicated otherwise, statistical analyses were performed using student’s t-tests or One-way ANOVA followed by Dunnett’s tests to examine multiple comparisons with the GraphPad Prism software (Version 10.2.3 for Windows, San Diego California, United States). For all experiments, data from at least three replicates was analyzed. Results were considered as statistically significant with a p-value <0.05.

First, genetic variants contributing to imatinib resistance in an in vitro-K-562 CML cell line model were analyzed by exome sequencing. In two out of seven imatinib-resistant biological replicate cell lines that did not harbor BCR::ABL1 mutations, the KMT2D variant p. (Arg191Trp) (NM_003482) was recurrently detected with allele frequencies of 27% and 37%, respectively ((Kaehler et al., 2023), unpublished data, Figure 1A), while this variant was not detected in treatment-naïve K-562 cells. The presence of this variant was confirmed in these two cell lines by in-depth sequencing, as it was present in 43% and 52% of cells resistant to low dose imatinib (0.5 µM) and 43% and 56% cells resistant against high dose imatinib (2 μM, Figure 1A). This raised the question on the role of this gene and this particular KMT2D variant in imatinib resistance. In silico prediction revealed a CADD score of 29.6 and a PolyPhen score of 0.999 indicating a detrimental effect on KMT2D protein function. Thus, the KMT2D mRNA expression was analyzed and found to be significantly downregulated in both imatinib-resistant cell lines harboring the KMT2D p. (Arg191Trp) variant compared to treatment-naïve cells (R1: 2 µM IM: −33.3%, p = 0.03; R2: 0.5 µM IM: −15.1%, p = 0.03; 2 µM IM: −33.9%, p < 0.001, Figure 1B). In imatinib-resistant cell line replicates not carrying KMT2D variants, KMT2D mRNA was not differentially reduced, but even significantly upregulated in one replicate (Supplementary Figure S1). As KMT2D regulates methylation of H3K4, protein levels of histon 3 (H3) and its methylated form H4K4me1 were investigated, but did not reveal significant changes in the methylation between treatment-naïve and KMT2D-variant imatinib-resistant cells (Figure 1C).

In a next step, the effect of KMT2D downregulation on imatinib susceptibility was analyzed by a siRNA-mediated knockdown. After successful knockdown of KMT2D (p = 0.02, Figure 2A), the cells were exposed to imatinib and cellular fitness was investigated. A significant increase in the cell number (90.6%, p < 0.001), metabolic activity (23.9%, p < 0.001) and proliferation rates (Ki-67-expression: 2.0-fold, p = 0.01) in the KMT2D knockdown was observed compared to negative control-transfected cells (Figures 2B–D). In addition, methylation of H3K4 was analyzed after silencing of KMT2D, revealing a slight decrease in H3K4me1 compared to Histone 3 and HSP90 levels (Figure 2E). This indicates that CML cells benefit from the loss of KMT2D expression under imatinib exposure, while histone methylation is reduced.

As KMT2D expression was significantly downregulated in imatinib-resistant cell lines harboring the p. (Arg191Trp) variant, we were interested whether restoration of its expression by transfection of a KMT2D-encoding plasmid in these cell lines would increase imatinib-sensitivity (both: p < 0.001, Figure 3A). After KMT2D rescue, exposure to imatinib led to a reduction of cell numbers (IM-R1: −46.6%, p < 0.001, IM-R2: −41.2%, p < 0.001) and metabolic activities (IM-R1: −12.9%, p = 0.02, IM-R2: −15.7%, p = 0.02) compared to the empty vector control transfection. These findings indicate a restored susceptibility towards imatinib in both resistant cell lines (Figure 3B,C). However, proliferation rates were only significantly reduced in IM-R1 (−29.7%, p = 0.03, Figure 3D). In addition, analysis of H3K4 methylation did not reveal any changes (Figure 3E).

Our previous findings suggested that the absence of KMT2D would be favorable for the development of imatinib resistance. In addition, it also pointed to a detrimental effect of the p. (Arg191Trp) variant on the KMT2D protein function. However, as the KTM2D variant’s effect on the protein function was still unclear, transfection experiments were performed to compare H3K4-methylation, cell numbers, proliferation and metabolic activity of either KMT2D wild-type or p. (Arg191Trp) in treatment-naïve K-562 cells (WT: p < 0.001; p. (R191W): p = 0.003, Figure 4A). In the presence of the KMT2D variant, methylation of H3K4 was slightly decreased compared to the KTM2D wild-type (Figure 4B). Under imatinib treatment, the presence of KMT2D p. (Arg191Trp) led to a significant increase in cell number compared to wild-type KMT2D (93%, p < 0.001, Figure 4C). In addition, metabolic activities (23%, p < 0.001) and proliferation rates (19.7%, p = 0.04), were also significantly increased under imatinib treatment (Figures 4D,E). These data confirm that the KMT2D p. (Arg191Trp) variant augments the development of imatinib resistance in CML.

To restore the reduced methylation of H3K4 caused by the potential loss-of-function KMT2D p. (Arg191Trp) variant, the lysine-specific histone demethylase (LSD1) counteracting KMT2D function was inhibited. As expected, methylation of H3K4 was slightly increased under treatment with the LSD1 inhibitor in cells carrying the KMT2D variant (Figure 5A). Subsequently, the effect on imatinib susceptibility in these cells was analyzed. LSD1 inhibition led to a significant decrease in cell number (−26.7%, p < 0.001), metabolic activity (−58.4%, p = 0.007) and proliferation (−52.2%, p < 0.001, Figure 5B). Moreover, to assess whether LSD1 inhibition alters imatinib sensitivity in the context of the KMT2D variant, we compared the IC50 values of imatinib in the presence of the LSD1 inhibitor. Cells harboring the KMT2D variant exhibited a significantly higher IC50 compared to wild-type (imatinib IC50: 60.7 µM vs. 72.4 µM, 19%, p = 0.004, Figure 5C) indicating a reduced imatinib susceptibility in the presence of the KMT2D variant.

To get a deeper insight into mechanisms underlying the loss of imatinib-susceptibility in presence of KMT2D p. (Arg191Trp), genome-wide expression data from imatinib-resistant cells harboring KMT2D wild-type or p. (Arg191Trp) were obtained and compared to imatinib-sensitive K-562 cells ((Kaehler et al., 2022; Kaehler et al., 2023); GSE227347, GSE203342). These expression profiles were compared with the KMT2D essentiality network, a list of 1954 genes from Takemon et al. (Takemon et al., 2024), to identify potential genes targeted by KMT2D. In KMT2D wild-type imatinib-resistant cells, 51 genes were detected, while in KMT2D variant cells 197 genes were found (Figure 6A). By subsequent KEGG pathway cluster analysis, an enrichment with genes involved in the p53 signaling pathway was detected (Figure 6B), also showing an interaction of the respective genes in the STRING annotation (Figure 6C). These upregulated genes were the cyclins CCND3 (3.2-fold) and CCNE2 (4.3-fold) and the cyclin-dependent kinase CDK4 (2.2-fold enriched in the imatinib-resistant KMT2D variant cells). As these genes are putative indirect interaction partners of KMT2D via the chromatin-remodeling complex protein ARID1A or the co-activator of transcription CREBBP, the question arose if their expression is influenced by the KMT2D variant. In subsequent analysis of mRNA expression levels, upregulation of CCNE2 (2.2-fold, p < 0.001) in imatinib-resistant cells harboring the KMT2D variant compared to treatment-naïve cells was confirmed, while expression of CCND3 and CDK4 was not altered (Figure 6D). These findings indicate that proliferation of cells harboring the KMT2D variant could be mediated by upregulation of CCNE2.

Besides KMT2D, deregulation of other epigenetic factors in imatinib-resistant CML was analyzed. Thus, genome-wide gene expression from imatinib-resistant and treatment-naïve K-562 biological replicate cell lines harboring KMT2D wild-type or p. (Arg191Trp) derived from the GSE203342 and GSE227347 datasets were compared and filtered for significant deregulation of genes encoding epigenetic modulators and histones. The number of differentially expressed genes varied between four and 54 between KMT2D wild-type and variant (Figure 7A). As DNMT or HDAC genes were differentially deregulated between KMT2D wild-type or variant imatinib-resistant cells, this raised the question on the efficiency of epigenetic modulation in the presence and absence of the KMT2D variant. Thus, the cell lines were exposed to the DNMT inhibitor 5′-azacytidine or the HDAC inhibitor vorinostat. Under treatment with 5′-azacytidine, the imatinib IC50 was significantly reduced in KMT2D p. (Arg191Trp) cells compared to wild-type cells (−60%, p = 0.04, Figure 7B), while the HDAC inhibitor vorinostat did not alter the response to imatinib (Figure 7C).