Cell lines and patient-derived xenografts (PDX)

HW8 hiPSCs were generated from peripheral blood mononuclear cells (PBMCs) of a glioma patient using the CytoTune-iPS 2.0 Sendai Reprogramming kit (Thermo Fisher Scientific, #A16517) following written informed consent. Study approval was obtained by the internal review board at the National Institutes of Health (NIH, protocol number: 16CN 069). Cellartis human iPSC line 12 (ChiPSC12, #Y00280) was purchased from Takara Bio. BCP-ALL and 293T cell lines were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ). In-house leukemia patient samples for injection into NSG mice (The Jackson Laboratory) were retrieved from the Biobank of the University Hospital of Düsseldorf following informed consent in accordance with the Declaration of Helsinki. Study approval was obtained by the ethics committee of the Medical Faculty of the Heinrich Heine University (study number: 2019-566). All animal experiments adhered to regulatory guidelines set by the official committee at LANUV (Akt. 81-02.04.2017.A441) and were authorized by the animal research institute (ZETT) at Heinrich Heine University Düsseldorf.

Molecular cloning of a RUNX1 HDR template

A RUNX1 homology-directed repair (HDR) template targeting ETV6 intron 5 was constructed by combining RUNX1 exon sequences 2–8 with a puromycin resistance gene under control of the human EF-1α promoter. Homology arm sequences of ≈500 bp were polymerase chain reaction (PCR)-amplified from ChiPSC12 genomic DNA and ligated to both sides of the HDR template. The RUNX1 HDR template was subcloned into plasmid pUC19 (Addgene, #50005), linearized by PCR and concentrated by isopropanol precipitation to achieve a concentration ≥1 µg/µL. Single-guide RNA (sgRNA) sequences targeting the 5′ region of ETV6 intron 5 were designed using the online prediction tool CRISPOR (http://crispor.tefor.net/)²⁸ and subcloned into the pUC19-U6-BbsI-sgRNA plasmid. Target sequences used for CRISPR/Cas9 genome editing of hiPSCs were GGATGAGGCTAAATCCCTAA (hg38, chr12: 11,870,115–11,870,134, + strand) and GCCTAATTGGGAATGGTGCG (hg38, chr12: 11,870,054–11,870,073, − strand).

CRISPR/Cas9 editing of hiPSCs

Following incubation with 10 µM Y-27632 (STEMCELL Technologies, #72304) for 2 h, single-cell suspensions of HW8 or ChiPSC12 hiPSCs were prepared using StemPro Accutase (Thermo Fisher Scientific, #A1110501). 10 × 10⁶ hiPSCs were resuspended in 100 µL P3 solution with supplement (Lonza, #V4XP-3024) and transfected with 2.5 µg linearized RUNX1 HDR template and 4 µg each of pCW-Cas9 plasmid (Addgene, #50661) as well as two sgRNA plasmids using program CD-118 of the 4D Nucleofector system (Lonza). hiPSCs were plated onto 10 cm Geltrex-coated dishes in mTeSR Plus/Y-27632. Medium was exchanged to mTeSR Plus without Y-27632 after 24 h. Selection with 0.5 µg/mL puromycin (Thermo Fisher Scientific, #A1113803) was commenced 48 h after Nucleofection and single colonies were picked under microscopic guidance into a 96-well plate at days 7–10. Clones were expanded for subsequent confirmation of correct HDR template insertion.

In vitro differentiation of hiPSCs

Feeder-free differentiation of hiPSCs was performed using the STEMdiff Hematopoietic kit (Stemcell Technologies, #05310) according to the manufacturer's instructions. Briefly, hiPSCs were seeded onto Geltrex-coated (Thermo Fisher Scientific, #A1413301) 12-well plates as aggregates. The next day (D0), medium A (containing bFGF, BMP4, and VEGFA) was added to wells containing 16–40 hiPSC colonies to induce mesodermal differentiation, and a half-medium change was performed on D2. On D3, medium A was removed and medium B (containing bFGF, BMP4, VEGFA, SCF, Flt3L, and TPO) was added to induce hematopoietic differentiation. Half-medium changes were performed on D5, D7, and D10. After 12 days, differentiated hematopoietic cells float in suspension and were harvested from the supernatant for downstream analyses.

siRNA-mediated H1-0 knockdown

Specific siRNA sequences for knockdown of H1-0 in REH cells were designed using the Eurofins siRNA design tool (https://eurofinsgenomics.eu/en/ecom/tools/sirna-design/) and purchased from Eurofins Genomics. Sequences are listed in Table S1. The 1 × 10⁶ REH cells were transfected with 200 pmol of each siRNA pool (siCtrl, siH1-0_1, siH1-0_2) in the 100 µL nucleocuvette format using the 4D Nucleofector system (Lonza, SF solution, program DS-150).

Dual-luciferase reporter assay

Human H1-0 promoter sequence (nucleotides −351 to +161 from TSS) was PCR-amplified from REH cell genomic DNA and inserted into Firefly luciferase vector pGL4.22 (Promega, #E6771) at KpnI and HindIII restriction sites. 293T cells at 50%–70% confluency were transfected with 755 ng plasmid DNA using Xfect Transfection Reagent (Clontech Laboratories, #631317) in 24-well plates according to the manufacturer's instructions. To determine the effect of ETV6::RUNX1 on H1-0 promoter activation, each well was transfected with 500 ng pGL4.22 vector with or without H1-0 promoter expression as well as 5 ng Renilla luciferase control plasmid pGL4.73 (Promega, #E6911) and 250 ng of the respective pcDNA3.1 vectors (Thermo Fisher Scientific, #V79020) for expression of ETV6::RUNX1 or RUNX1 or empty vector in triplicates. To analyze the effect of Quisinostat on H1-0 promoter activation, transfection was performed with 500 ng pGL4.22 vector with H1-0 promoter expression and 5 ng Renilla luciferase control plasmid, and cells were treated with the indicated concentrations of Quisinostat or DMSO (1:10,000) for 24 h. Cells were lysed 48 h after transfection with Passive Lysis buffer and luciferase signal was measured on a Tecan SPARK 10 M reader using the Dual-Luciferase Reporter Assay System (Promega, #E1910). Firefly luciferase signal was normalized to Renilla luciferase activity. Adequate protein expression of ETV6::RUNX1 and RUNX1 was determined by Western blot.

In vitro inhibitor treatments

BCP-ALL cell lines were treated with 1 µM JNJ-26481585/Quisinostat at a concentration of 1 × 10⁶ cells per ml and RNA was extracted after 24 h for subsequent analysis of H1-0 expression by RT-qPCR and RNA-seq analysis. DMSO-dissolved compounds were purchased from Selleck Chemicals and MedChem Express. For drug synergy analysis, Quisinostat (concentration range: 0.2–20 nM), AR-42 (concentration range: 10–1000 nM), suberanilohydroxamic acid (SAHA)/Vorinostat (concentration range: 100–5000 nM), Vincristine (concentration range: 0.1–5 nM), Daunorubicin (concentration range: 1.5–50 nM), and Bortezomib (concentration range: 1–10 nM) were printed onto 384-well plates (Corning, #3570) in a randomized fashion in increasing concentrations of 8 × 8 matrices using a D300e digital dispenser (Tecan) and normalized with DMSO (Sigma-Aldrich, #2650). BCP-ALL cell lines were seeded at a concentration of 200,000 cells/mL (6000 cells/well). Due to limited amount of cells, PDX samples and siRNA-treated REH cells were seeded at 100,000 cells/mL (3000 cells/well). PDX samples were thawed and cultured for 24 h at 37°C and 5% CO₂ in StemSpan SFEM II (Stemcell Technologies, #09605) with added StemSpan CD34+ expansion supplement (Stemcell Technologies, #02691) before performing drug screens. Plates were incubated for 72 h and viability was determined by CellTiter-Glo Luminescent viability assay (Promega) using a Tecan SPARK 10 M reader. Most synergistic area scores (2 × 2 dose window) were determined using the zero interaction potency (ZIP) method using the SynergyFinder web application (version 3.0).²⁹ A most synergistic area score above 5 was considered synergistic.

Statistical analysis

Statistical analysis of data was performed using GraphPad Prism version 9.5.1. The number (n) of replicates and statistical tests are indicated in the figure descriptions. Statistical significance was considered for p values: *p < 0.05, **p < 0.01, and ***p < 0.001.

Additional methods can be found in the Supporting Information.

H1-0 is upregulated in a preleukemic hiPSC model and BCP-ALL expressing ETV6::RUNX1

To analyze specific gene expression patterns of ETV6::RUNX1-translocated preleukemia in a model without additional secondary alterations, we generated monoclonal hiPSC lines derived from two donors: HW8 and ChiPSC12. We used a CRISPR/Cas9-mediated knock-in approach to directly fuse RUNX1 exons 2–8 to ETV6 exon 5 and to place the resulting fusion gene under the physiological control of the endogenous ETV6 promoter (Figure 1A). We confirmed correct sequence of the RUNX1 insert by genotyping PCR of the ETV6 locus (Figure S1A,B) and Sanger sequencing (Figure S1C). ETV6::RUNX1 levels in the hiPSC lines as detected by reverse transcription quantitative PCR (RT-qPCR) and Western blot were lower compared to the ETV6::RUNX1+ BCP-ALL cell line REH (Figure 1B). All hiPSC lines maintained typical hiPSC microscopic morphology and expression of the pluripotency markers SSEA-4, DNMT3B, GDF3, POU5F1, and NANOG as determined by flow cytometric analyses and RT-qPCR; chromosomal integrity was confirmed by karyotype analysis (Figure S2). All ETV6::RUNX1+ hiPSC lines harbored a monoallelic insertion of the RUNX1 HDR template at the ETV6 locus as detected by PCR, RT-qPCR, and Sanger sequencing (Figure S1D–G). Since the RUNX1 HDR template disrupts one ETV6 allele, expression of full-length ETV6 was lower in the CRISPR/Cas9-edited hiPSCs compared to the wild-type controls (Figure 1C,D). This aligns with the genetic profile of ETV6::RUNX1+ ALL patients who commonly exhibit heterozygosity for the fusion gene. Since ETV6 exon 6 is not retained in the ETV6::RUNX1 fusion gene, full-length ETV6 was detected using an RT-qPCR spanning exons 5 and 6 (Figure S1F). REH cells served as negative control due to the deletion of the remaining copy of ETV6 (Figure 1C,D).

To identify significantly dysregulated genes in ETV6::RUNX1+ preleukemic cells, we performed bulk RNA-seq of ETV6::RUNX1+ and wild-type hiPSCs. Principal component analysis (PCA) clearly separated samples according to genotype (Figure 2A). Altogether, we found consistent differential expression of 20 genes with an absolute fold change > 2 and p < 0.05 in the three ETV6::RUNX1+ hiPSC lines compared to the respective wild-type hiPSC lines (Figures 2B and S3 and Tables S2–S5). These genes remained significantly dysregulated using a harsher cut-off of false discovery rate (FDR)-adjusted p value (or q value) < 0.1 (Table S5). Among these genes, H1-0 has previously been identified as the most significantly upregulated gene in dormant leukemia stem cell-like cells.³⁴ As a linker histone, H1-0 is involved in epigenetic regulation of chromatin and affects cellular differentiation states,²⁷ making it a compelling candidate for further investigation. Elevated levels of H1-0 identified by RNA-seq in the ETV6::RUNX1+ hiPSCs were confirmed both by RT-qPCR (2.4-fold increased mean expression; Figure 2C) and Western blot (Figure 2D). Moreover, upregulation of H1-0 in ETV6::RUNX1+ preleukemic cells is preserved during the differentiation of hiPSCs along the B lymphoid lineage in a published RNA-seq dataset¹⁰ (Figure 2E).

We previously found H1-0 expression to be restricted to ETV6::RUNX1+ bone marrow blasts compared to peripheral blood CD19+ cells,⁸ indicating that H1-0 upregulation is preserved upon leukemic transformation and highly specific for leukemic cells carrying the ETV6::RUNX1 fusion gene. To confirm this finding, we analyzed transcriptomic data derived from two patient cohorts encompassing a total of 1727 leukemia patient samples (PeCan St. Jude cohort^{30, 31} and GSE87070³²). Additionally, we determined H1-0 expression in nine BCP-ALL PDX samples by RT-qPCR (n = 2 ETV6::RUNX1+, n = 4 high-hyperdiploid, n = 2 BCR::ABL1+, n = 1 TCF3::PBX1+). Notably, ETV6::RUNX1+ BCP-ALL showed significantly elevated H1-0 levels compared to other leukemia entities (Figure 2F–H). Moreover, H1-0 was downregulated upon ETV6::RUNX1 knockdown in published RNA-seq data¹¹ of REH cells (p < 0.001, Figure 2I). In line with these observations, H1-0 expression closely anti-correlates with RUNX1 expression (Pearson r = −0.646, p < 0.0001, Figure 2J) in healthy bone marrow cells derived from the MILE study.³³ Altogether, these data support an association of the ETV6::RUNX1 fusion gene and linker histone H1-0 expression.

ETV6::RUNX1 skews lineage commitment during early hematopoiesis

To delineate effects of ETV6::RUNX1 in hematopoietic cells, we performed in-depth characterization of hematopoietic progenitor cells (HPCs) differentiated from ETV6::RUNX1+ and wild-type hiPSCs. As confirmed by RT-qPCR, ETV6::RUNX1 expression in HPCs increased to levels comparable with REH cells (Figure 3A) and expression of common hiPSC pluripotency marker genes decreased during differentiation (Figure S4). To infer lineage-commitment of hiPSC progenitors, we carried out single-cell RNA-seq (scRNA-seq) and annotated our in vitro-derived HPCs using published reference data (Figure S5A–C). Overlaying our data with published scRNA-seq atlas data³⁵ revealed that hiPSC-derived HPCs clustered closer to fetal liver compared to fetal bone marrow or cord blood-derived cells (Figure S5C). This finding is in keeping with other studies that showed similarity of hiPSC in vitro differentiation and early fetal hematopoiesis, for instance, in the fetal liver.^{10, 36} By mapping to fetal liver, fetal bone marrow and cord blood,³⁵ as well as to adult bone marrow³⁷ reference data, we identified three distinct trajectories of megakaryocyte, erythroid, and granulocyte/monocyte/dendritic cell (DC) progenitors in hiPSC-derived cells (Figure S5A,B). Comparable differentiation trajectories were observed by using a diffusion map representation of the data, which showed a central node of naïve cells and three branches of lineage-committed progenitors (Figure S5D). Of note, our data exhibited similar lineage commitment compared to a previous study that analyzed hiPSC-derived hematopoietic progenitors generated through embryoid body differentiation.³⁶ Expression of cell type-specific marker genes confirmed commitment to megakaryocytes in Leiden clusters 0 and 2 (TUBB1, expression of additional marker genes shown in Figure S6), and commitment to granulocytes/monocytes/DCs in cluster 3 (CEBPA), while naïve progenitors in cluster 4 expressed CD34 and the early megakaryocytic-erythroid marker FCER1A (Figure 3B–C). Naïve progenitors in cluster 5 exhibited an intermediate expression profile of megakaryocyte and granulocyte/monocyte/DC marker genes. In line with previous work showing that megakaryocytic-erythroid lineage specification is governed by cell cycle speed,³⁸ we observed that erythroid-committed cluster 1 was defined by differential expression of cell cycle regulators (Table S6 and Figure S5E), while megakaryocyte progenitors were annotated as non-cycling (Figure 3D).

Classification of HPCs by genotype revealed that expression of ETV6::RUNX1 skewed commitment towards granulocyte/monocyte/DC progenitors and resulted in increased abundance of naïve progenitors (Figure 3E,F; cell numbers per cluster are shown in Table S7). Comparison of ETV6::RUNX1+ and wild-type cells within the annotated clusters identified the highest number of differentially expressed genes within the megakaryocyte- and erythrocyte-committed clusters (0–2; Figure S5F and Table S8). In keeping with reduced commitment to erythrocyte progenitors, cell cycle scoring showed accumulation of non-cycling G0/G1 cells in ETV6::RUNX1+ HPCs (FDR = 0.018, Figure S5G). Additionally, transcriptional diversity and activity was reduced in ETV6::RUNX1+ HPCs as indicated by lower number of expressed genes per cell (n_genes) and unique transcripts detected per cell (n_UMIs), respectively (Figure S5H,I).

Next, we performed immunophenotyping of hiPSC-derived HPCs using an antibody panel designed to identify hematopoietic progenitors in human bone marrow³⁹ (gating strategy depicted in Figure S7). ETV6::RUNX1 expression led to increased numbers of phenotypic hematopoietic stem cells (HSC: CD34+LIN-CD38-CD90+CD45RA-, Figure 3G), in keeping with expansion of naïve progenitors observed by scRNA-seq (padj < 0.05; Table S9). Increased persistence of preleukemic ETV6::RUNX1+ HSCs has been reported previously.^{22, 40}

The majority of HPCs were characterized as multipotent progenitors (MPP: LIN-CD34+CD38-CD90-CD45RA-) by flow cytometry, while scRNA-seq analysis revealed a large proportion of megakaryocyte- and erythrocyte-committed HPCs. This discordance between transcriptionally and immunophenotypically defined cell types underlines the strength of scRNA-seq for in-depth characterization of cellular states and detection of differentiation trajectories. Correlating protein and mRNA levels would require further analyses that combine proteome and transcriptome profiling on single-cell level, such as cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq).

In colony forming assays, ETV6::RUNX1+ HPCs exclusively formed granulocyte-macrophage progenitor colonies (CFU-GM; Figure S8), while wild-type HPCs were also able to differentiate into common myeloid progenitor colonies (CFU-GEMM) and erythroid progenitors (BFU-E). This underlines the increased granulocyte/monocyte/DC lineage commitment that we observed in ETV6::RUNX1+ HPCs by scRNA-seq.

In summary, our preleukemic ETV6::RUNX1+ hiPSC model recapitulates the accumulation of phenotypic HSCs and exhibits increased commitment towards the granulocyte/monocyte/DC lineage during early hematopoiesis, as revealed by scRNA-seq analysis.

ETV6::RUNX1 induces H1-0 promoter activation

Given that our findings show strong association between ETV6::RUNX1 and H1-0 expression, we tested the potential of ETV6::RUNX1 to transactivate the H1-0 promoter. To this end, we cloned the H1-0 promoter region (−351 to +161 from TSS) into a luciferase reporter plasmid (Figure 4A), which was transfected into 293T cells along with either an empty vector or vectors containing FLAG-tagged ETV6::RUNX1 or RUNX1 sequences. Luciferase activity measurements confirmed that expression of ETV6::RUNX1 is sufficient to activate the H1-0 promoter (2.2-fold), while RUNX1 expression reduced luciferase activity (3.1-fold; Figure 4B). However, our previous analyses in murine cells^{8, 9} and analysis of the H1-0 promoter region using published chromatin immunoprecipitation sequencing (ChIP-seq) data of ETV6::RUNX1+ REH cells^{42, 43} did not show direct binding of either the fusion protein or RUNX1 to the H1-0 promoter region or distal enhancer regions upstream of H1-0 (Figure S9A), suggesting an indirect mechanism of H1-0 upregulation upon ETV6::RUNX1 expression.

Additionally to transcriptional control via binding of transcription factors, differential DNA methylation of the H1-0 CpG island (CGI) shore has been reported to regulate H1-0 expression in various solid tumor types, acting as an enhancer element.²⁷ Hence, we analyzed previously published 450K Infinium microarray DNA methylation data comprising patient samples of T-ALL and six B-ALL subtypes⁴¹ (n = 546). Indeed, the mean CGI shore methylation of H1-0, comprising probes cg07141002 and cg01883777, inversely correlated with H1-0 expression (Pearson r = −0.645, p < 0.0001; Figures 4C and S9B) and was lowest in ETV6::RUNX1+ BCP-ALL (Figure 4D), indicating that H1-0 expression is regulated via dynamic methylation of its CGI shore in leukemia. While the connection of ETV6::RUNX1 and H1-0 remains correlative, our data suggest that ETV6::RUNX1 induces upregulation of H1-0 in an indirect manner, possibly via the H1-0 promoter and CGI shore region.

H1-0 levels decrease during hematopoiesis

During hematopoiesis, H1-0 is expressed in undifferentiated, quiescent progenitor cells.⁴⁴ To characterize H1-0 expression during B lymphopoiesis, we analyzed published RNA-seq data from adult and pediatric bone marrow,⁴⁵ expression microarray data from umbilical cord blood and peripheral blood,⁴⁶ and scRNA-seq data from fetal liver.⁴⁷ Across these datasets, we observed a continuous decrease of H1-0 expression during B cell development (Figure 5A–C) and significant upregulation of H1-0 in ETV6::RUNX1+ ALL cells (n = 6) compared to HSCs and later B cell progenitor stages (Figure 5A).

To examine H1-0 expression in the context of cell cycle activity and hematopoietic differentiation, we employed scRNA-seq data of B precursor cells derived from adult bone marrow.⁴⁸ H1-0+ cell numbers decreased along the B lineage trajectory, clustering preferentially to G0/G1 cell cycle states, especially within the HSC, early lymphoid and pro-B populations (Figure 5D). Taken together, our data suggest that H1-0 is an indicator of differentiation state.

H1-0 is a key mediator of the ETV6::RUNX1-specific gene signature

To determine the contribution of H1-0 to ETV6::RUNX1+ BCP-ALL pathology, we knocked down H1-0 in the ETV6::RUNX1+ BCP-ALL cell line REH and performed gene set enrichment analysis (GSEA) on RNA-seq data. Knockdown reduced H1-0 RNA expression by ≈2.4-fold, translating to decreased protein levels compared to non-targeting siRNA (siCtrl) treatment (Figure 6A,B). Cell proliferation increased upon H1-0 knockdown (Figure 6C) along with a rise in apoptotic subG1 cells and reduced frequency of cells in G2/M (Figure 6D). GSEA using the canonical pathways collection (Human MSigDB Collections) revealed significant enrichment (cut-offs: p < 0.005, FDR q value of < 0.1) of gene signatures associated with DNA replication, histone modification, DNA repair and protein ubiquitination in siCtrl-treated REH cells (Figure 6E and Table S10), while no gene sets were identified as significantly enriched in REH cells treated with H1-0-targeting siRNA using the same cut-offs (Table S11). Notably, GSEA detected enrichment of gene sets linked to histone acetylation (Table S10 [in red]) in siCtrl-treated REH cells, consistent with previous reports highlighting strong correlation between H1-0 expression and chromatin acetylation.^{50, 51}

To ascertain common molecular drivers of gene expression changes observed upon H1-0 knockdown, we applied upstream regulator analysis using the Ingenuity Pathway Analysis (IPA) suite⁵² (Qiagen). Interestingly, the most significant potential driver detected by IPA upstream regulator analysis was ETV6::RUNX1, with p = 3.7 × 10⁻¹⁶ (for siH1-0_1 vs. siCtrl, Table S12) and p = 3.5 × 10⁻¹¹ (for siH1-0_2 vs. siCtrl, Table S13) respectively (Figure 6F). Negative activation z-scores indicated inhibition of the ETV6::RUNX1 transcription factor upon H1-0 knockdown. Given that ETV6::RUNX1 primarily functions as a repressor of RUNX1-regulated genes,⁵³ we employed a set of genes downregulated by ETV6::RUNX1 (cut-offs: log2 fold change > 0.9 and p < 0.05)⁴⁹ to validate our findings. Indeed, GSEA revealed significant upregulation of these ETV6::RUNX1 signature genes upon H1-0 knockdown (normalized enrichment score [NES] = 1.63, FDR q value = 0.001; Figure 6G and Figure S10A).

Furthermore, we detected significant activation of TP53 (encoding for the tumor suppressor p53) following H1-0 knockdown, as indicated by both upstream regulator analysis (Figure 6F) and GSEA (NES = 1.45, FDR q value = 0.009; Figure 6H). Indeed, previous studies have demonstrated that ETV6::RUNX1 suppresses p53 activity by upregulating MDM2.⁵⁴ Accordingly, we detected downregulated MDM2 in REH cells upon H1-0 knockdown (Figure S10B). Moreover, both EPOR and RAG1, two genes upregulated by ETV6::RUNX1 and key factors in ETV6::RUNX1+ BCP-ALL pathophysiology,^{6, 42, 55, 56} exhibited reduced levels upon H1-0 knockdown (Figure S10B) as well as significant correlation with H1-0 RNA expression in ETV6::RUNX1+ BCP-ALL patient samples derived from the PeCan St. Jude cohort^{30, 31} (n = 87, Figure S10C). Taken together, these data imply that linker histone H1-0 is a novel key regulator of ETV6::RUNX1-induced expression changes.

H1-0 inducer Quisinostat synergizes with frontline drugs in ETV6::RUNX1+ leukemic cells

Due to their cytostatic activity, HDACis are potent inducers of H1-0 expression.^{50, 51} Importantly, H1-0 has been identified as a mediator of the antitumor effect induced by the pan-HDACi Quisinostat in various solid cancers.⁵¹ Hence, basal expression levels of H1-0 could be a marker for HDACi activity in BCP-ALL. Indeed, we found a striking inverse correlation between H1-0 protein levels and sensitivity towards HDACis in a panel of 25 BCP-ALL cell lines (data derived from the Functional Omics Resource of Acute Lymphoblastic Leukemia [FORALL] platform, https://proteomics.se/forall/^{57, 58}), in particular with AR-42 and Vorinostat (p < 0.001), as well as 11 other HDACis, including Quisinostat (p < 0.01; Figure 7A). To examine the effect of H1-0 knockdown on Quisinostat sensitivity in ETV6::RUNX1+ BCP-ALL, REH cells were transduced with H1-0-targeting siRNA for 48 h and treated with Quisinostat (Figure S11A,B). While downregulation of H1-0 did not alter sensitivity towards single drug treatment with Quisinostat (Figure S11C), combination with the commonly used B-ALL chemotherapeutics Vincristine and Daunorubicin, or the proteasome inhibitor Bortezomib, increased drug synergism (Figure 7B).

To confirm the H1-0-inducing properties of Quisinostat, we again performed H1-0 promoter luciferase assays in 293 T cells and treated with increasing sub-lethal concentrations of Quisinostat for 24 h. Indeed, Quisinostat induced H1-0 promoter activation (Figure S11D) and an increase of endogenous H1-0 levels (Figure S11E) in a dose-dependent manner. Moreover, BCP-ALL cell lines with varying levels of basal H1-0 expression (high levels: REH and MHH-CALL-2, medium level: SUP-B15, low level: RS4;11) strongly upregulated H1-0 upon treatment with 1 µM Quisinostat for 24 h (Figure 7C,D). As anticipated, basal H1-0 levels reflected doubling times of BCP-ALL cell lines (Figure 7C) and REH cells were most resistant to Quisinostat treatment, as indicated by lowest fraction of subG1 cells after 24-hour Quisinostat treatment (Figure S11F). Transcriptome analysis revealed a drug-induced shift in all four BCP-ALL cell lines and dysregulation of similar signaling pathways (Figure S11G–I and Tables S14–S17). These include inhibition of MYC^{59, 60} and BRD4,⁶¹ as described previously, as well as the activation of TP53 signaling due to induction of apoptosis (Figure 7E). Interestingly, the ETV6::RUNX1 signature was activated upon Quisinostat treatment, indicating a connection of histone acetylation and ETV6::RUNX1 target gene transcription, that has also been proposed previously.²⁰

We further performed synergy drug screens with Quisinostat, as well as AR-42 and Vorinostat, the two HDACis that showed highest anti-correlation with H1-0 protein levels (Figure 7A). For this, we screened three ETV6::RUNX1- BCP-ALL cell lines (MHH-CALL-2, SUP-B15, and RS4;11) and the ETV6::RUNX1+ cell line REH as well as two ETV6::RUNX1+ PDX samples (Figures 7F and S12). Overall, simultaneous inhibition of HDACs and the proteasome showed high synergy. Clinical efficacy of combination treatment with Bortezomib and HDACis has been shown in previous studies targeting hematologic malignancies.⁶² Moreover, we found high synergy in ETV6::RUNX1+ samples using HDACis in combination with the topoisomerase II inhibitor Daunorubicin, while there was no or low synergy in ETV6::RUNX1− samples using this combination. Of note, Quisinostat induced effective killing in BCP-ALL cell lines at much lower concentration range (0.2–10 nM) than AR-42 (10–1000 nM) or Vorinostat (100–5000 nM; Tables S18S–20). Taken together, these analyses indicate that combinatorial drug treatment using Quisinostat in combination with Daunorubicin or Bortezomib might be beneficial for targeting ETV6::RUNX1+ leukemic cells.