Abstract
Significance
Disruption of PU.1-directed enhancer programs upon SWI/SNF inhibition causes differentiation of AML cells and induces leukopenia of PU.1-dependent B cells and monocytes, revealing the on- and off-tumor effects of SWI/SNF blockade.Free full text
SWI/SNF Blockade Disrupts PU.1-Directed Enhancer Programs in Normal Hematopoietic Cells and Acute Myeloid Leukemia
Abstract
In acute myeloid leukemia (AML), SWI/SNF chromatin remodeling complexes sustain leukemic identity by driving high levels of MYC. Previous studies have implicated the hematopoietic transcription factor PU.1 (SPI1) as an important target of SWI/SNF inhibition, but PU.1 is widely regarded to have pioneer-like activity. As a result, many questions have remained regarding the interplay between PU.1 and SWI/SNF in AML as well as normal hematopoiesis. Here we found that PU.1 binds to most of its targets in a SWI/SNF-independent manner and recruits SWI/SNF to promote accessibility for other AML core regulatory factors, including RUNX1, LMO2, and MEIS1. SWI/SNF inhibition in AML cells reduced DNA accessibility and binding of these factors at PU.1 sites and redistributed PU.1 to promoters. Analysis of nontumor hematopoietic cells revealed that similar effects also impair PU.1-dependent B-cell and monocyte populations. Nevertheless, SWI/SNF inhibition induced profound therapeutic response in an immunocompetent AML mouse model as well as in primary human AML samples. In vivo, SWI/SNF inhibition promoted leukemic differentiation and reduced the leukemic stem cell burden in bone marrow but also induced leukopenia. These results reveal a variable therapeutic window for SWI/SNF blockade in AML and highlight important off-tumor effects of such therapies in immunocompetent settings.
Significance:
Disruption of PU.1-directed enhancer programs upon SWI/SNF inhibition causes differentiation of AML cells and induces leukopenia of PU.1-dependent B cells and monocytes, revealing the on- and off-tumor effects of SWI/SNF blockade.
Introduction
Acute myeloid leukemia (AML) is an aggressive malignancy associated with poor outcomes (1–3). An important hallmark of AML cell identity is a deregulated hematopoietic transcriptional program that promotes malignant self-renewal and impairs differentiation (4). Altered gene expression in AML is supported by a core regulatory circuitry (CRC; ref. 5) composed of transcription factors (TF) and other regulators, including RUNX1 (6), MEIS1 (7), LMO2, and several others (8, 9), which form an autoregulatory circuit that maintains leukemic cell identity (10). Targeting the chromatin and transcriptional dependencies of the AML CRC therefore represents a promising approach to target the cell-intrinsic mechanisms sustaining this malignancy.
Excitingly, the recent development of selective small-molecule inhibitors of SWI/SNF chromatin remodeling complexes (11) has provided new opportunities to target AML based on its dependency on SWI/SNF activity (12). In AML cell lines, genetic knockdown of SWI/SNF ATPases (13) or treatment with BRM014 (14), a selective dual inhibitor of SWI/SNF ATPase subunits SMARCA4 (BRG1) and SMARCA2 (BRM), reduces MYC expression and induces leukemic differentiation. Loss of MYC is caused by disruption of the hematopoietic-specific blood enhancer cluster (BENC; ref. 15), which is regulated by SWI/SNF and occupied by the hematopoietic TF PU.1 and other members of the AML CRC.
However, SWI/SNF and PU.1 are both essential for many aspects of normal hematopoiesis. SWI/SNF complexes play several roles in hematopoietic stem cell (HSC) maintenance (16, 17), erythropoiesis (18, 19), myeloid (20, 21), and lymphoid (22–24) development, and interact with many hematopoietic lineage TFs, including PU.1 (25). PU.1 is a pioneer-like transcription factor (25–27), which is essential for myeloid and B-cell differentiation (28–31), and interacts with a large number of other TFs that regulate hematopoiesis (32, 33). Given these critical functions in normal hematopoiesis, important questions regarding the use of SWI/SNF inhibitors against AML have remained.
Here we evaluated the mechanisms and outcomes associated with SWI/SNF inhibition in clinically relevant models of AML and normal hematopoiesis. We establish that SWI/SNF blockade in AML cells disrupts PU.1-directed enhancer programs utilized by CRC members, and drives a genome-wide shift of PU.1 from enhancers to the promoters of many genes, including those involved in leukemic differentiation. We find that SWI/SNF inhibition is well tolerated by mice but has specific off-tumor effects on PU.1-dependent hematopoietic lineages. We also demonstrate potent sensitivity to SWI/SNF inhibition in an immunocompetent mouse model and in primary human AML samples. Our results identify altered PU.1-directed activity at enhancers as a critical on- and off-tumor effect of SWI/SNF inhibition, and establish SWI/SNF blockade as a promising antileukemic therapy.
Materials and Methods
Human cell lines
THP-1 (RRID: CVCL_0006), MV-4–11 (RRID: CVCL_0064), MOLM-13 (RRID: CVCL_2119), OCI-AML3 (RRID: CVCL_1844), and HL-60 cells (RRID: CVCL_0002) were acquired from ATCC or Leibniz Institute DSMZ and maintained in RPMI supplemented with 10% heat-inactivated FBS, 10 mmol/L HEPES, GlutaMAX, nonessential amino acids, 1 mmol/L sodium pyruvate, 2-mercaptoethanol, and pen/strep. Cells were cultured in a humidified incubator maintained at 37°C and 5% CO2. Common laboratory procedures for these cells are described in Supplementary Materials and Methods. After thawing, cells were passaged a minimum of three times before use in experiments and were regularly tested for Mycoplasma contamination using the MycoAlert Mycoplasma Detection Kit (Lonza).
Flow cytometry
Measurements were made on a Sony SH800 (RRID: SCR_018066), BD FACSCanto (RRID: SCR_018055), BD LSR II (RRID: SCR_002159), or BD FACSAria II (RRID: SCR_018934). Single-stained cells or compensation beads (Invitrogen, 01–3333–41) were used as compensation controls. Antibodies and filter sets for each instrument are listed in Supplementary Materials and Methods. All data were analyzed using FlowJo v10.4.1 (RRID: SCR_008520).
Colony forming assays
Cells were counted and resuspended in IMDM 10% heat-inactivated FBS 1% pen/strep (human) or DMEM 2% heat-inactivated FBS 1% pen/strep at a 1:50 dilution (murine). Cells were plated in triplicate at 10,000 (human AML), 500 (human CD34+), or 1,000 (murine MLL-AF9) viable cells/plate using human or mouse MethoCult methylcellulose media (StemCell Technologies, 04435 or 03434) with indicated concentrations of BRM014 or DMSO. After 10 to 14 days, plates were scored for colony counts and cells analyzed by flow cytometry where indicated.
Genome-wide analyses
ATAC library preparation and sequencing
Cells treated with 1 μmol/L BRM014, 3 μmol/L DB2313, or DMSO for indicated durations were prepared using the Omni-ATAC protocol (34) described in Supplementary Materials and Methods. Paired-end sequencing was performed on an Illumina NextSeq 500 (RRID: SCR_014983) high-output flow cell.
Chromatin immunoprecipitation sequencing library preparation and sequencing
Cells were treated with 1 μmol/L BRM014, 3 μmol/L DB2313, or DMSO for 72 hours. Chromatin immunoprecipitation (ChIP) libraries were prepared from a single-cell suspension fixed for 10 minutes in 1% formaldehyde (TF) or dual-fixed for 30 minutes in 2 mmol/L disuccinimidyl glutarate followed by 10 minutes in 1% formaldehyde (SMARCA4/BRG1). Excess formaldehyde was quenched by the addition of glycine to 125 mmol/L. Fixed cells were washed, pelleted, and snap-frozen using liquid nitrogen. ChIP was performed as described previously (35). The following antibodies were used for ChIP: PU.1 (Cell Signaling Technology, 2266s, RRID: AB_10692379), RUNX1 (Abcam, ab23980, RRID: AB_2184205), MEIS1 (Abcam, ab19867, RRID: AB_776272), LMO2 (R&D Systems, AF2726, RRID: AB_2249968), and SMARCA4/BRG1 (Proteintech, 21634–1-AP, RRID: AB_10858784).
Assay for transposase-accessible chromatin using sequencing and ChIP-seq analysis
As described previously (35, 36), reads were mapped to the hg38 human reference genome using Bowtie 2.4.1 (RRID: SCR_005476; ref. 37). Duplicate fragments and reads with mapping quality <10 were discarded. Peak calling was performed by MACS 2.1.1 (RRID: SCR_013291; ref. 38). DESeq2 (RRID: SCR_000154) was used for differential peak calling, with size factors determined using the top quartile of accessible sites. Differential peak calls were made by requiring fold changes of >1.5-fold in either direction and FDR-adjusted P values <0.10. TF motif enrichment was measured using HOMER (v4.11, RRID: SCR_010881; ref. 31) and ChromVAR (39). Overlap of peaks was assessed using bedtools (2.28.0; RRID: SCR_006646). Calculation of mean densities and preparation of genome-wide heat maps was performed using bwtool (RRID: SCR_003035; ref. 40).
Single-cell library preparation and sequencing
Bone marrow (BM) and peripheral blood were pooled by condition (N = 3 per condition). Peripheral red blood cells were lysed via osmotic shock and vortexing, and live nongranulocyte PBMCs and BM cells were isolated via FACS based on FSC/SSC profile and propidium iodide viability staining for multiome. Samples were processed on the 10× chromium platforms for single-cell 5′ RNA-seq (scRNA-seq; v2) or single-nuclei multiome (ATAC + Gene expression) kits according to the manufacturer's instructions. The resulting libraries were sequenced on an Illumina NovaSeq 6000 flow cell (RRID: SCR_016387).
5′ scRNA-seq and single-nuclei multiome analysis
scRNA-seq transcripts within each cell were counted using the 10× Cell Ranger 5.0.1 pipeline (RRID: SCR_017344). Multiome features within each cell were counted and linked using the Cell Ranger ARC 2.0.1 pipeline. All mapping was to the mm10 mouse genome build. Multiplets were removed using DoubletFinder 2.0.3 (RRID: SCR_018771; ref. 41) and AMULET 1.1 (42). To identify immune cell populations, gene expression data were visualized using UMAP embedding in Loupe Browser 6.1.0 (RRID: SCR_018555). For multiome, cell types incompatible with sn-multiome, including granulocytes and clusters with uniformly low ATAC reads were not considered for analysis. Cell type-specific transcripts and heat maps were generated using Seurat 3.2.3 (RRID: SCR_007322; ref. 43).
Human primary leukemias and CD34+ HSPCs
Culturing primary AML samples
Viably frozen de-identified primary AML samples obtained via leukapheresis were procured from the Texas Children's Cancer Center tissue repository under Institutional Review Board–approved protocols. Written informed consent and assent where appropriate were obtained from each participant and/or his or her guardian in accordance with the declaration of Helsinki prior to any study procedures. Cells were thawed at 37°C and transferred to fresh RPMI with 10% heat-inactivated FBS and 1% pen/strep at 1:50 dilution. After a 2-hour rest, cells were counted by trypan blue and resuspended in IMDM with 10% heat-inactivated FBS and 1% pen/strep. For liquid cultures, cells were seeded at 2e5 cells/mL and maintained in IMDM with 10% heat-inactivated FBS and 1 μmol/L BRM014 or DMSO for 48 hours.
Isolation of CD34+ cord blood stem/progenitor cells
Freshly isolated umbilical cord blood units from three anonymous healthy donors were obtained from MD Anderson Cancer Center on their Institutional Review Board–approved protocol and diluted 1:3 with PBS. Mononuclear cells were isolated using density centrifugation. CD34+ cells were isolated by MACS cell separation using the human CD34 MicroBead Kit UltraPure (Miltenyi Biotec, 130–100–453) and resuspended in fresh DMEM with 10% heat-inactivated FBS and 1% pen/strep.
Mice
Female 6-to-8-week-old C57BL/6J mice were purchased from the Baylor College of Medicine (BCM) Animal Core Facility. All animal experiments performed in this study were approved by the BCM Institutional Animal Care and Use Committee and BCM Veterinary Office.
Hematopoiesis mouse model
Dosing and sample collection
The treatment group (N = 5) received once daily oral gavage of 20 mg/kg of BRM014 in 10% DMSO and 1% methylcellulose (Sigma, M0430–100G), and the control group (N = 6) received vehicle alone for 14 days. Nearly 18 hours after the last treatment, peripheral blood was collected. A random subset of mice (N = 3 per condition) were euthanized, and BM was collected via crushing of bilateral femurs, tibias, and fibulas and homogenized using 40-μmol/L sterile cell strainers (Thermo Fisher Scientific, 22–363–547).
Competitive HSC transplant
Donor BM was harvested from the femurs, tibias and fibulas of CD45.2 C57BL/6 mice after treatment with daily BRM014 or vehicle control for 14 days and pooled from N = 3 mice per condition. Competitor BM cells were harvested from age-matched, untreated CD45.1 C57BL/6 mice. Recipient C57BL/6 CD45.1 mice were conditioned with a split dose of 10.5 Gy of irradiation then transplanted with donor and competitor whole BM by retro-orbital injection. Each recipient mouse was transplanted with 250 CD45.2+ donor HSCs (defined as Lin–, cKit+, Sca-1+, CD48–, CD150+) and 250 CD45.1 competitor HSCs, with N = 4 (vehicle) and N = 6 (BRM014). Fifteen weeks after transplantation, recipient mice were humanely euthanized, and lineage distributions of CD45.1 and CD45.2 peripheral blood cells were compared by flow cytometry.
Syngeneic KMT2A-MLLT3 mouse model
Tumor engraftment and dosing
Mice were sublethally irradiated and transplanted with 20,000 GFP+ murine KMT2A-MLLT3 (MLL-AF9) cells. Leukemia development was monitored by detection of GFP+ leukemic cells in peripheral blood. Two weeks after engraftment, mice were assigned to treatment or control groups (N = 11 vehicle, N = 10 BRM014) with essentially equivalent distributions of leukemic burden. For downstream analyses, all sample sizes in treatment and control conditions were chosen in advance but vary in some cases due to humane endpoints in control animals. The treatment group received once daily oral gavage of 20 mg/kg of BRM014 in 10% DMSO and 1% methylcellulose (Sigma, M0430–100G), and the control group received vehicle alone for 14 days. During treatment, mice were randomly assigned to two equivalent groups (N = 5 per condition), which were monitored at staggered intervals to minimize the frequency of bleeds.
Cell processing
For peripheral blood, RBCs were lysed via osmotic shock and vortexing. Nearly 18 hours after the last treatment, a random subset (N = 3 per condition) of mice was euthanized, and peripheral blood, BM, livers, and spleens were collected. Cell suspensions were obtained by manual trituration and homogenized using 40-μmol/L sterile cell strainers (Thermo Fisher Scientific, 22–363–547), and leukemic burden was assessed by flow cytometry.
Complete blood count analysis
For AML studies, peripheral blood was collected 3 days after the last treatment from MLL-AF9 mice and age-/sex-matched nonleukemic C57BL/6 controls. For all complete blood count (CBC) analyses, 15 μL of blood was analyzed using the Heska HT5 Hemocounter. One specimen for hematopoiesis studies was identified as a significant outlier by the Grubb test based on anomalously high total WBC count and was excluded from analysis.
Statistical analysis
Statistical tests were performed in R (3.6.1) as two-sided tests (RRID: SCR_001905). For multiple-comparison tests, P values were adjusted using the Benjamini–Hochberg FDR correction procedure. P values below the 64-bit double precision machine epsilon (2−52 = 2.22e−16) are reported as P < 2.2e−16. Independent biological replicates were chosen in advance and are indicated in figure legends. All N values indicate independent biological replicates, genome-wide studies used N = 2 per condition in agreement with recommendations of the ENCODE Consortium, flow cytometry data are presented as representative analyses from N ≥ 3 replicates, and single-cell experiments used N = 3 pooled animals per condition.
Data availability
High-throughput sequencing data generated for this project have been deposited in the Gene Expression Omnibus (GEO, RRID: SCR_005012) database with SuperSeries accession numbers GSE205572 and GSE190723. Public H3K4me3 ChIP-seq data were analyzed from GEO accession No. GSM3681460.
Results
PU.1-directed recruitment of SWI/SNF enables binding of the core tumor circuitry
Consistent with previous reports, treatment of a panel of AML cell lines with BRM014 resulted in leukemic differentiation (P < 2.2e−16 in all cases, Supplementary Figs. S1A–S1I, full data in Supplementary Dataset S1–S2). To dissect the underlying chromatin mechanisms, we performed assay for transposase-accessible chromatin using sequencing (ATAC-seq) on AML cells treated with BRM014 or DMSO, which revealed loss of accessibility at SWI/SNF sites (represented by SMARCA4 ChIP-seq peaks) across the genome (Supplementary Fig. S2A). Motif profiling identified the most outstanding losses to be at sites bearing the binding motif of PU.1 (Fig. 1A; Supplementary Fig. S2B, full data in Supplementary Dataset S3), a hematopoietic lineage-specific TF and ETS family member.
ChIP-seq revealed 136,513 PU.1 sites across the genome (Fig. 1B). Despite substantial overlap, the majority of PU.1 sites did not overlap with SWI/SNF peaks (55%) and many lacked DNA accessibility peaks (44%, Fig. 1C; Supplementary Fig. S2C), consistent with its pioneer-like activity (44–46). Following SWI/SNF inhibition with BRM014, PU.1 binding was largely preserved at 74,460 sites not bound by SMARCA4 (here referred to as “untargeted sites,” Fig. 1D), as well as at 41,150 sites that were bound by SMARCA4 but did not show reduced accessibility upon BRM014 treatment (“independent sites”). PU.1 was furthermore retained at ~60% of the 20,803 sites that showed reduced DNA accessibility following BRM014 treatment (“dependent sites”), reflecting a high capacity to bind its target sites in a SWI/SNF-independent manner. In contrast to PU.1, TFs and other factors that frequently co-bind with PU.1, including the AML CRC members RUNX1, LMO2, and MEIS1, were highly dependent on SWI/SNF activity and were uniformly reduced to near-background levels at dependent sites upon treatment with BRM014 (Fig. 1E). These losses occurred despite their continued expression (Supplementary Dataset S1) and the preservation of binding at sites not targeted by SWI/SNF (Supplementary Fig. S2D).
These findings led us to hypothesize that PU.1 recruits SWI/SNF to promote accessibility needed for other CRC TFs at PU.1 sites. To test this, we treated THP-1 cells with DB2313, an established small-molecule inhibitor that selectively blocks PU.1 binding to its target sites and reduces the expression of PU.1 target genes (Supplementary Figs. S2E–S2F; refs. 47–49). ChIP-seq of SMARCA4 following treatment with DB2313 but not DMSO selectively disrupted SMARCA4 recruitment to individual PU.1 sites (Fig. 1F), which reduced SMARCA4 binding (Fig. 1G) and DNA accessibility (Fig. 1H) at these sites genome-wide. PU.1 motifs were the highest enriched motifs at sites with decreased SMARCA4 binding or accessibility upon DB2313 treatment, demonstrating a high degree of selectivity (Supplementary Figs. S2G–S2H), and confirming PU.1-dependent recruitment of SWI/SNF.
Examination of altered PU.1 targeting following SWI/SNF inhibition revealed that the ~11% of genome-wide sites with decreased PU.1 occupancy were enriched at SWI/SNF-dependent sites, whereas the ~3% of sites with increased PU.1 occupancy were enriched at SWI/SNF-independent sites (all P < 2.2e−16, Fig. 1I). Although SWI/SNF-dependent sites were differentially enriched at intergenic regions, SWI/SNF-independent sites were enriched at promoters (Fig. 1J). In agreement, PU.1 increases were particularly associated with sites containing high levels of the promoter mark H3K4me3 (Fig. 1K). Enhanced PU.1 occupancy, including at myeloid differentiation-related genes, such CD14 and CD68 (Fig. 1L), corresponded with increased expression genome-wide (Supplementary Figs. S2I–S2K). Our data show that PU.1 recruits SWI/SNF to generate accessibility needed for CRC factor binding at distal sites (Fig. 1M). SWI/SNF inhibition reduces DNA accessibility required for CRC factors at these sites, and results in relocalization of PU.1 from a minority of its sites to promoters.
PU.1 and SWI/SNF activity converge on the BENC module
In hematopoietic cells, MYC expression is regulated by the BENC module (13, 15). In AML cells, BRM014 treatment induced loss of DNA accessibility and CRC factor binding at BENC compared with DMSO (Fig. 2A; Supplementary Figs. S3A–S3B). Although PU.1 was retained at some BENC sites, binding partners such as RUNX1 showed strong reductions at SWI/SNF-bound sites across the region (Fig. 2A). Inhibition of PU.1 resulted in reduced SMARCA4 recruitment to PU.1 sites within BENC (Supplementary Fig. S3C), consistent with the PU.1-directed SWI/SNF recruitment described above.
Treatment with either BRM014 or DB2313 caused loss of MYC RNA expression in AML cell lines (Fig. 2B). MYC protein levels were similarly reduced in MV-4–11 and MOLM-13 cells within 24 hours, but required 72 hours for complete reduction in THP-1 cells treated with DB2313 (Fig. 2C; Supplementary Fig. S3D). Despite modest reduction of PU.1 levels in some cell lines consistent with its autoregulatory function (50), PU.1 remained expressed and was retained at its chromatin binding sites. However, loss of accessibility at PU.1 motifs (Fig. 1A) arose even in cells lacking strong PU.1 expression changes (Fig. 2C).
Transcriptional changes induced by BRM014 and DB2313 treatment were strongly correlated genome wide (Spearman R = 0.38 P < 2.2e-16; Fig. 2D), and included downregulation of MYC targets, genes repressed during myeloid development, and HSC-related genes (Fig. 2E). Both inhibitors induced leukemic differentiation, as confirmed by increased expression of the cell surface markers CD14 and CD87 (Fig. 2F). Interestingly, SWI/SNF and PU.1 inhibition also induced loss of the stemness marker (51) CD44, which is not a feature of phorbol 12-myristate 13-acetate (PMA)-induced differentiation of AML cells (Fig. 2G and andHH and Supplementary Fig. S3E), reflecting a unique signature of differentiation via PU.1 deregulation. BRM014 and DB2313 also exhibited largely additive effects (Supplementary Fig. S3F), consistent with convergence on the same pathway. Together, our findings reveal that PU.1 and SWI/SNF collaboratively regulate MYC expression in AML and that inhibition of either factor converges on myeloid differentiation.
SWI/SNF inhibition has off-tumor effects on PU.1-dependent hematopoietic cell types
Because SWI/SNF and PU.1 are both critical regulators of hematopoiesis, we sought to examine the effects of SWI/SNF inhibition on nontumor hematopoietic cells in vivo. We treated healthy C57BL/6 mice with daily oral gavage of 20 mg/kg BRM014 or vehicle control for 14 days (Fig. 3A), which was well tolerated and did not cause significant weight loss (P = 0.49, Supplementary Fig. S4A). After the 14-day treatment period, we observed decreased white blood cell (WBC) counts in the peripheral blood of BRM014-treated mice but no change in hemoglobin levels (Fig. 3B).
Single-nuclei multiome (sn-multiome; integrated snATAC and snRNA-seq) analysis of peripheral blood collected from BRM014- and vehicle-treated mice (N = 3 per condition) revealed cells in distinct clusters, which we identified as B cells, T cells, NK cells, and monocytes using established markers (Fig. 3C; Supplementary Fig. S4B). SWI/SNF inhibition caused a notable reduction in the proportion of peripheral B cells and monocytes (Fig. 3D), but gene expression changes were observed in all cell types (Supplementary Fig. S4C). PU.1 motifs were among the most significantly down-regulated TF motifs in B cells and monocytes, but not in NK or T cells (Fig. 3E). Moreover, reduced cell numbers corresponded to PU.1 motif utilization, with B cells and monocytes exhibiting the highest PU.1 motif accessibility, which was furthermore reduced by SWI/SNF inhibition (Fig. 3F and andG).G). Remarkably, more detailed analysis of sites with PU.1 motifs revealed DNA accessibility losses at distal sites but gains at promoters (Fig. 3H), similar to the effects observed in AML cells.
Flow cytometry (Fig. 3I and andJ)J) confirmed that compared with vehicle treatment, BRM014 induced no significant effect on neutrophils (P = 0.18) but reduced total monocytes 1.8-fold (P = 0.05), with specific loss of the Ly6Chigh population, a marker for monocytes recently derived from BM (P = 3.7e−3, Fig. 3I; Supplementary Fig. S4D; refs. 52, 53). Within the lymphoid compartment, BRM014 treatment had no significant effect on peripheral T cells (P = 0.38) but caused a 5-fold depletion of B cells (P = 0.01, Fig. 3J; Supplementary Fig. S4E). Our findings reveal important off-tumor effects of SWI/SNF inhibition that arise through the same mechanisms responsible for therapeutic effects against AML.
To evaluate the reversibility of SWI/SNF inhibition's effects on hematopoiesis, we assessed CBCs from mice on the last day of treatment and 3 weeks after cessation of treatment. This comparison revealed that BRM014-induced effects on peripheral blood cells recover within 3 weeks of drug withdrawal (Fig. 3K), demonstrating that hematological and immune-related effects of transient SWI/SNF inhibition are reversible.
Effects of SWI/SNF inhibition on hematopoiesis arise within BM
To determine at which stage of hematopoietic development SWI/SNF sensitivity arises, we evaluated the effect of BRM014 on BM harvested from BRM014- and vehicle-treated mice at day 14 (N = 3 per condition). BM from BRM014-treated mice contained fewer total cells than vehicle control (P = 0.017, Fig. 4A). To assess the chromatin effects of BRM014 treatment, we performed sn-multiome on BM harvested at day 14, which revealed distinct clusters that we classified as B cells, erythroid cells, monocytes, T cells, plasmacytoid dendritic cells (pDC), and Kit+ hematopoietic stem and progenitor cells (HSPC) using standard cell-type-specific markers (Supplementary Fig. S5A). This study revealed the same loss of PU.1 motifs in B cells (P = 5.9e−17) and monocytes (P = 7.4e−3) in BM as observed in the peripheral blood, with no other cell types showing significant reductions (Fig. 4B; Supplementary Figs. S5A–S5C).
To provide increased resolution of expression changes within BM populations, we performed single-cell RNA-seq (scRNA-seq), which revealed similar clustering of cell identities but with increased granularity (Fig. 4C). Interestingly, scRNA-seq revealed reduced BM cell counts to be caused primarily by a loss of lineage-committed (Kit–) cells (Fig. 4D). This finding was orthogonally confirmed by flow cytometry, which showed that SWI/SNF inhibition differentially affects BM cells that have become committed and lineage-restricted (P = 7.0e−3, Fig. 4E; Supplementary Figs. S5D–S5E). Further, reductions of B cells and monocytes varied between early proliferative cells and late nonproliferative cells across cell types (Fig. 4F), demonstrating that SWI/SNF inhibition has lineage-specific effects and is required at distinct developmental stages for each lineage (Supplementary Figs. S5F–S5G).
Importantly, BRM014 did not alter the number of LSK (Lin–Sca-1+Kit+) HSPCs in BM (P = 0.57, Fig. 4G; Supplementary Fig. S5H). To assess the effects of SWI/SNF inhibition on HSC fitness and function, we performed a competitive HSC transplant of BM harvested from BRM014- and vehicle-treated mice (Fig. 4H). Excitingly, HSCs isolated from BRM014-treated mice retained the ability to engraft and to reconstitute all major hematopoietic lineages (Fig. 4I). We conclude that transient SWI/SNF inhibition does not permanently suppress HSC function.
SWI/SNF inhibition induces regression of AML in an immunocompetent in vivo model
To assess the efficacy of SWI/SNF inhibition against AML in vivo, we employed a syngeneic KMT2A-MLLT3 (“MLL-AF9”) leukemic model in C57BL/6 mice, which is an aggressive and immunocompetent AML model (54) with predictable latency. In vitro, leukemic cells displayed a dose-dependent inhibition of colony-forming capacity upon treatment with BRM014 (Supplementary Fig. S6A). We treated mice engrafted with GFP+ MLL-AF9 cells with BRM014 or vehicle control for 14 days (Fig. 5A). Compared with vehicle control, BRM014 strongly prevented leukemic expansion (P < 2.2e−16, two-way ANOVA, Fig. 5B and andC).C). Notably, we observed significant regression of disease burden (P = 8.3e−3), including one mouse in which the peripheral leukemic burden was reduced from 32% to 7% after 6 days of treatment (Supplementary Fig. S6B).
Treatment with BRM014 prevented leukemia-induced splenomegaly (P = 2.4e−4) and hepatomegaly (P = 5.5e−3, Fig. 5D; Supplementary Figs. S6C–S6D), with leukemic burden reduced 14.8-fold (P = 4.8e−6) in the spleen and 18.8-fold (P = 3.3e−5) in the liver (Fig. 5E and andF).F). Importantly, BRM014 also reduced leukemic burden in BM 7-fold compared with vehicle control (P = 1.2e−4, Fig. 5G). To understand the leukemic subpopulations affected by SWI/SNF inhibition, we assessed the effects on committed lineage-marker positive (Lin+) and immature (Lin–) leukemic blasts (Fig. 5H and andI).I). BRM014 effectively reduced all leukemic subpopulations, resulting in a 5.3-fold reduction of Lin+ (P = 3.9e−4) and a 12-fold reduction of Lin– leukemic BM cells (P = 1.9e−5). Notably, Lin–Sca-1–Kit+ leukemic cells, a population enriched for leukemic stem cells (55, 56), were reduced 20-fold (P = 1.8e−3, Fig. 5J).
To detect leukemic differentiation in vivo, we assessed the expression of CD44 and myeloid differentiation markers CD11B and CD14. Leukemic BM cells from mice treated with BRM014 for 14 days displayed decreased CD44 (P = 2e−3) and increased expression of CD11B (P = 0.05) and CD14 (P = 4.7e−5, Fig. 5K–M). Similar results were observed in the peripheral blood as early as day 9 (Supplementary Figs. S6E–S6G). Hence, our results confirm that leukemic differentiation occurs in both peripheral blood and BM in an immunocompetent context.
BRM014 was well tolerated and caused no significant weight loss (P > 0.05, two-way ANOVA; Supplementary Fig. S6H). CBCs 3 days after completion of therapy revealed vehicle-treated mice had high WBC counts, anemia, and thrombocytopenia, reflecting their leukemic state (Supplementary Figs. S6I–S6K). BRM014-treated mice had reduced WBC (P = 2.3e−4) and lymphocyte (P = 1.1e−5) counts compared with nonleukemic control mice (Fig. 5N, Supplementary Figs. S6I–S6K; Supplementary Dataset S4). Platelets, neutrophils, and monocytes were not significantly different (P > 0.05 in all cases). While hemoglobin was mildly reduced in BRM014-treated mice compared with normal controls (P = 6.1e−6), BRM014-treated mice were not anemic, with hemoglobin levels within a normal range. The modest inflation of monocytes in BRM014-treated mice was nonsignificant and caused by the identification of differentiated leukemic blasts as monocytes by CBC. However, flow cytometry of GFP-negative cells revealed significant reduction of nontumor myeloid cells in BRM014-treated compared with nonleukemic reference mice, confirming similar off-tumor effects in a leukemic setting (Supplementary Figs. S6L–S6M).
Treatment with BRM014 significantly prolonged survival (P = 1.9e−4, log-rank test), with a median survival of 44 days for BRM014-treated mice compared with 16 days for vehicle-treated mice (Fig. 5O). Overall, our results demonstrate that small-molecule inhibition of SWI/SNF ATPase activity significantly reduces disease burden and prolongs survival of AML.
Primary human samples mirror the on- and off-tumor effects of SWI/SNF inhibition in mice
Upon treatment with BRM014, primary human CD34+ HSPCs exhibited dose-dependent reduction of colony-forming capacity (Fig. 6A), with an IC50 value of 58 ± 33 nmol/L (mean ± SEM). Notably, we observed selective reduction in the development of myeloid lineage cells, consistent with the PU.1-related effects observed in mice (P = 1.5e−3; Supplementary Fig. S7A–S7C). These results reveal that similar off-tumor effects arise in both mice and in primary human samples.
To assess the effect of SWI/SNF inhibition on human AML samples, we obtained leukemic blasts from patients with AML and evaluated their sensitivity to BRM014. BRM014 caused dose-dependent reduction of colony formation potential (Fig. 6B; Supplementary Fig. S7D) and altered cell morphology (Fig. 6C). Although the majority of AML specimens had a lower IC50 compared with nontumor CD34+ HSPCs, three of the seven had higher IC50 values. This variation in therapeutic response indicates that the balance between on- and off-tumor effects gives rise to a unique therapeutic window for each patient. Effective concentrations of BRM014 varied between donors, but 1 μmol/L of BRM014 was sufficient to ablate colony formation potential of all leukemias. Remarkably, one patient sample (358) that was highly responsive to BRM014 treatment bore the KMT2A-MLLT4 (MLL-AF6) fusion [t(6;11)], which is the most therapy-refractory KMT2A fusion, associated with event-free survival odds of only 11% to 17% (57, 58). No systematic variation was observed between KMT2Ar and non-KMT2Ar samples, indicating that SWI/SNF inhibition targets mechanisms that contribute to AML maintenance independently of KMT2A fusion status.
Treatment with BRM014 did not result in significant loss of viability of primary leukemia cells within 48 hours, even at concentrations up to 1 μmol/L (Supplementary Fig. S7E). Flow cytometry of human samples confirmed accumulation of myeloid differentiation markers (CD11B, CD68) and loss of the stemness marker CD44, a signature of PU.1 deregulation in AML (Fig. 6D and andE).E). Markers activated upon SWI/SNF inhibition varied by donor, but indicated a general myeloid differentiation phenotype.
Similar to our cell line observations, ATAC-seq revealed PU.1 to be the most decreased motif following BRM014 treatment (Fig. 6F). Moreover, BRM014 caused loss of accessibility at BENC in all specimens (Fig. 6G), with concomitant loss of MYC expression (Fig. 6H). Notably, we observed significant heterogeneity in the specific peaks lost at BENC, and hence variation in chromatin state may underlie crucial aspects of the heterogeneous patient-specific responses to SWI/SNF inhibition. Excitingly, BRM014 caused accessibility losses at intergenic regions with PU.1 motifs (P = 2.5e−7) and gains at promoter regions with PU.1 motifs (P = 1.4e−6, Fig. 6I), consistent with the mechanisms observed in our cell line studies. Overall, our findings reveal that primary human AMLs are highly sensitive to pharmacological targeting of SWI/SNF and indicate that the balance between on- and off-tumor effects on hematopoietic cells will be an important clinical consideration in humans.
Discussion
Subunits of SWI/SNF complexes act as oncogenic dependencies in multiple cancer types (59–62). In these cancers, pharmacologic inhibition of SWI/SNF represents a promising and novel strategy to target tumor-specific chromatin dependencies. Targeting tumor cell-intrinsic dependencies may be particularly valuable for cancers with low mutational burdens, which often have few targetable mutations (63, 64) and respond poorly to immunotherapies due to low immunogenicity (65). Hematopoietic malignancies such as leukemia are ideal candidates for this strategy, due to their low mutation burden and demonstrated dependence on SWI/SNF activity.
Acute leukemias rely on SWI/SNF activity to maintain the stemness of leukemic cells and to prevent terminal differentiation. In this setting, SWI/SNF complexes as well as PU.1 and CRC member RUNX1 have been linked to regulation of MYC expression (13, 66–68). Our work reveals that PU.1 requires SWI/SNF activity to recruit CRC factors to distal enhancers, including the BENC module that drives MYC. When SWI/SNF is inhibited, PU.1 is retained at many sites, but is also relocalized to H3K4me3-bound promoter sites associated with myeloid differentiation genes.
Variable outcomes regarding the requirement for SWI/SNF subunits for hematopoiesis (59, 69, 70) have created uncertainty about whether SWI/SNF inhibition would cause adverse hematologic effects. Here, we demonstrate that transient SWI/SNF inhibition dysregulates hematopoietic development within BM, with lineage-specific effects similar to PU.1 inhibition, which also impacts committed PU.1-dependent lineages but spares HSCs (47). Nevertheless, long-term PU.1 impairment has been reported to impair HSC function (71), therefore additional studies on the long-term use of SWI/SNF inhibition are warranted. Although our work identifies impaired development in B cells and monocytes, lineage tracing studies can be expected to provide deeper insight into the altered mechanisms within each lineage. Importantly, the immune-related toxicities observed during short-term treatment compare favorably to the dose-limiting and persistent immunosuppression associated with standard-of-care chemotherapy (72, 73).
Recent studies showed that ex vivo inhibition of SWI/SNF complexes in T cells enhances memory T-cell formation (74) and limits T-cell exhaustion (75), which is not assessed in our studies due to the absence of an immune challenge. Importantly, unlike ex vivo treatment, in vivo SWI/SNF inhibition also impacts antigen-presenting cells such as macrophages. Our work shows that full consideration of the distinct effects on diverse cell types is required to understand the impact of SWI/SNF inhibition on immune response in vivo.
The strong leukemic regression achieved by SWI/SNF inhibitors in vivo demonstrates their remarkable potential for use in AML. Still, dosing regimens and treatment duration will require further optimization to achieve optimal outcomes in vivo, for example, by accounting for external factors including drug metabolism and small-molecule pharmacokinetics/pharmacodynamics that limit drug efficacy. Our work highlights that normal hematopoiesis can be affected at doses similar to those needed for therapeutic response in AML specimens, which will be an important consideration for clinical use of SWI/SNF inhibitors against AML and other tumors. Nevertheless, the robust regression of leukemic burden seen over a short 2-week treatment period suggests a considerable therapeutic window in immunocompetent settings for some patients, and provides a compelling justification for continued study of SWI/SNF inhibitors in the treatment of AML.
Acknowledgments
The authors thank Zainab Jagani and Florencia Rago (Novartis) for helpful comments and feedback. This work was supported by grants to H.C. Hodges from Gabrielle's Angel Foundation, the V Foundation (V2018–003), the Mark Foundation for Cancer Research (20–024-ASP), and the Adrienne Helis Malvin Medical Research Foundation. H.C. Hodges is a CPRIT Scholar supported by grants from the NIH (R35GM137996 and R01CA272769) and the Cancer Prevention and Research Institute of Texas (CPRIT RR170036). C. Chambers was supported by NIH grant F31AI161906. H.D. Lacorazza was supported by NIH grant R01CA207086 and the ASH Bridge Award. This project was also supported by the UTMB Next-Generation Sequencing Core, the BCM Cytometry and Cell Sorting Core (CPRIT RP180672, NIH P30CA125123, and S10RR024574), the BCM Single Cell Genomics Core (NIH S10OD018033, S10OD023469, S10OD025240, and P30EY002520), the BCM Genomic and RNA Profiling Core (NIH S10OD023469), the Texas Children's Cancer Center tissue repository, and the MD Anderson Next-Generation Sequencing Core (CPRIT RP120348).
The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.
Footnotes
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Authors' Disclosures
H.C. Hodges reports grants from Gabrielle's Angel Foundation, Adrienne Helis Malvin Medical Research Foundation, Mark Foundation for Cancer Research, V Foundation, NIH/NIGMS, NIH/NCI, and Cancer Research and Prevention Institute of Texas during the conduct of the study. No disclosures were reported by the other authors.
Authors' Contributions
C. Chambers: Formal analysis, visualization, methodology, writing–original draft. K. Cermakova: Investigation, writing–review and editing. Y.S. Chan: Formal analysis, writing–review and editing. K. Kurtz: Investigation. K. Wohlan: Investigation. A.H. Lewis: Investigation. C. Wang: Investigation. A. Pham: Formal analysis. M. Dejmek: Resources. M. Sala: Resources. M. Loeza Cabrera: Investigation. R. Aguilar: Investigation. R. Nencka: Resources. H.D. Lacorazza: Resources, methodology. R.E. Rau: Resources, investigation, methodology, project administration, writing–review and editing. H.C. Hodges: Conceptualization, resources, formal analysis, supervision, funding acquisition, visualization, methodology, writing–original draft, project administration, writing–review and editing.
References
Full text links
Read article at publisher's site: https://doi.org/10.1158/0008-5472.can-22-2129
Read article for free, from open access legal sources, via Unpaywall: https://aacrjournals.org/cancerres/article-pdf/83/7/983/3299969/983.pdf
Citations & impact
Impact metrics
Citations of article over time
Alternative metrics
Discover the attention surrounding your research
https://www.altmetric.com/details/141608455
Smart citations by scite.ai
Explore citation contexts and check if this article has been
supported or disputed.
https://scite.ai/reports/10.1158/0008-5472.can-22-2129
Article citations
Development of an orally bioavailable mSWI/SNF ATPase degrader and acquired mechanisms of resistance in prostate cancer.
Proc Natl Acad Sci U S A, 121(15):e2322563121, 01 Apr 2024
Cited by: 0 articles | PMID: 38557192 | PMCID: PMC11009648
The IL6/JAK/STAT3 signaling axis is a therapeutic vulnerability in SMARCB1-deficient bladder cancer.
Nat Commun, 15(1):1373, 14 Feb 2024
Cited by: 0 articles | PMID: 38355560 | PMCID: PMC10867091
MYC function and regulation in physiological perspective.
Front Cell Dev Biol, 11:1268275, 24 Oct 2023
Cited by: 3 articles | PMID: 37941901 | PMCID: PMC10627926
Review Free full text in Europe PMC
The ncBAF Complex Regulates Transcription in AML Through H3K27ac Sensing by BRD9.
Cancer Res Commun, 4(1):237-252, 01 Jan 2024
Cited by: 1 article | PMID: 38126767 | PMCID: PMC10831031
Reactivation of the G1 enhancer landscape underlies core circuitry addiction to SWI/SNF.
Nucleic Acids Res, 52(1):4-21, 01 Jan 2024
Cited by: 0 articles | PMID: 37993417 | PMCID: PMC10783513
Go to all (8) article citations
Data
Data behind the article
This data has been text mined from the article, or deposited into data resources.
GEO - Gene Expression Omnibus (3)
- (1 citation) GEO - GSE190723
- (1 citation) GEO - GSM3681460
- (1 citation) GEO - GSE205572
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
Exquisite Sensitivity to Dual BRG1/BRM ATPase Inhibitors Reveals Broad SWI/SNF Dependencies in Acute Myeloid Leukemia.
Mol Cancer Res, 20(3):361-372, 01 Mar 2022
Cited by: 17 articles | PMID: 34799403
SMARCB1 Deficiency Integrates Epigenetic Signals to Oncogenic Gene Expression Program Maintenance in Human Acute Myeloid Leukemia.
Mol Cancer Res, 16(5):791-804, 26 Feb 2018
Cited by: 13 articles | PMID: 29483235
Helicase-like transcription factor is a RUNX1 target whose downregulation promotes genomic instability and correlates with complex cytogenetic features in acute myeloid leukemia.
Haematologica, 101(4):448-457, 22 Jan 2016
Cited by: 10 articles | PMID: 26802049 | PMCID: PMC5004395
Emerging Themes in Mechanisms of Tumorigenesis by SWI/SNF Subunit Mutation.
Epigenet Insights, 15:25168657221115656, 26 Jul 2022
Cited by: 10 articles | PMID: 35911061 | PMCID: PMC9329810
Review Free full text in Europe PMC
[Research Progress on the Role of Chromatin Remodeling Factor BRG1 in Acute Myeloid Leukemia].
Zhongguo Shi Yan Xue Ye Xue Za Zhi, 24(3):930-933, 01 Jun 2016
Cited by: 0 articles | PMID: 27342536
Review
Funding
Funders who supported this work.
Adrienne Helis Malvin Medical Research Foundation
American Society of Hematology
Cancer Prevention and Research Institute of Texas (3)
Grant ID: RP120348
Grant ID: RR170036
Grant ID: RP180672
Gabrielle's Angel Foundation for Cancer Research
Mark Foundation For Cancer Research (1)
Grant ID: 20-024-ASP
NCI NIH HHS (3)
Grant ID: P30 CA125123
Grant ID: R01 CA207086
Grant ID: R01 CA272769
NCRR NIH HHS (1)
Grant ID: S10 RR024574
NEI NIH HHS (1)
Grant ID: P30 EY002520
NIAID NIH HHS (1)
Grant ID: F31 AI161906
NIGMS NIH HHS (1)
Grant ID: R35 GM137996
NIH HHS (3)
Grant ID: S10 OD025240
Grant ID: S10 OD018033
Grant ID: S10 OD023469
NIH Office of the Director (3)
Grant ID: S10OD023469
Grant ID: S10OD018033
Grant ID: S10OD025240
National Cancer Institute (3)
Grant ID: P30CA125123
Grant ID: R01CA272769
Grant ID: R01CA207086
National Eye Institute (1)
Grant ID: P30EY002520
National Institute of Allergy and Infectious Diseases (1)
Grant ID: F31AI161906
National Institute of General Medical Sciences (1)
Grant ID: R35GM137996
National Institutes of Health (1)
Grant ID: S10RR024574
V Foundation for Cancer Research (1)
Grant ID: V2018-003