Loss of ISWI ATPase SMARCA5 (SNF2H) in Acute Myeloid Leukemia Cells Inhibits Proliferation and Chromatid Cohesion

ISWI chromatin remodeling ATPase SMARCA5 (SNF2H) is a well-known factor for its role in regulation of DNA access via nucleosome sliding and assembly. SMARCA5 transcriptionally inhibits the myeloid master regulator PU.1. Upregulation of SMARCA5 was previously observed in CD34+ hematopoietic progenitors of acute myeloid leukemia (AML) patients. Since high levels of SMARCA5 are necessary for intensive cell proliferation and cell cycle progression of developing hematopoietic stem and progenitor cells in mice, we reasoned that removal of SMARCA5 enzymatic activity could affect the cycling or undifferentiated state of leukemic progenitor-like clones. Indeed, we observed that CRISPR/cas9-mediated SMARCA5 knockout in AML cell lines (S5KO) inhibited the cell cycle progression. We also observed that the SMARCA5 deletion induced karyorrhexis and nuclear budding as well as increased the ploidy, indicating its role in mitotic division of AML cells. The cytogenetic analysis of S5KO cells revealed the premature chromatid separation. We conclude that deleting SMARCA5 in AML blocks leukemic proliferation and chromatid cohesion.


Introduction
Acute myeloid leukemia (AML) is a malignant hematopoietic disease derived from myeloid-primed stem cells resulting in accumulation of myeloid blasts. AML patients have a poor prognosis and the only known efficient therapy is bone marrow transplantation combined with chemotherapy. Next-generation sequencing revealed that despite similar cytology and cellular features, the mutational profile of AML clones can be very heterogenic. Leukemogenesis involves multiple types of genomic alterations from single nucleotide variants to large chromosomal abnormalities (involving deletions, translocations, or chromosomal gains and losses). Targets of mutagenesis are often genes encoding regulators of gene transcription (e.g., RUNX1, CEBPA, GATA2), DNA methylation (e.g., DNMT3A, IDH1, IDH2), and genome organization (e.g., CTCF, RAD21, SMC3).

SMARCA5 Deletion Inhibits AML Cell Proliferation
To test requirement of SMARCA5 for AML cell growth, we produced a null allele using CRIPSR/Cas9 genome editing technology (Figure 2A). Targeted was exon5, which codes a portion of evolutionarily conserved ATPase domain and that was previously shown to be a targetable region using the Cre-loxP1 system. Deletion of exon5 results in a frame shift mutation disabling expression of Smarca5 protein in mouse [1]. For the experiments, human K562 cells (AML M6 subtype) were initially utilized as they were previously used for antisense oligonucleotide-mediated transient knockdown of SMARCA5 [2]. K562 cells were transfected by a pair of pX330-mVenus vectors containing sgRNAs complementary to a sequence in the SMARCA5 introns 4 & 5 and the the effect of CRISPR/Cas9-mediated deletion of exon5 was tested by PCR. Analysis of fragments amplified from genomic DNA of FACS-sorted mVenus-positive clonal populations identified 5 clones (#H10, D7, H4, E7, H7) with a single shortened PCR product (~632bp compared to 1175bp in controls) that were homozygously mutated ( Figure 2B). Sanger sequencing of PCR products confirmed that clones H10, D7, E7, and H5 contained the same deletion (543bp) and clone H4 an even larger deletion (582bp) within SMARCA5 exon5 ( Figure 2C). In addition, quantitative PCR and Western blot analyses of the cellular extracts confirmed that the Cas9-mediated deletion of the SMARCA5 gene resulted in loss of SMARCA5 expression (Figure 2 D,E). The resulting subclones had no expression of vector-coded & episomally expressed Cas9 nuclease. In addition, eight predicted off-target candidates (SRGAP2, RNF17, PRG4, GYPA, POLQ, CYB5R4, BCKDHB, NAV2) had no alteration of

SMARCA5 Deletion Inhibits AML Cell Proliferation
To test requirement of SMARCA5 for AML cell growth, we produced a null allele using CRIPSR/Cas9 genome editing technology ( Figure 2A). Targeted was exon5, which codes a portion of evolutionarily conserved ATPase domain and that was previously shown to be a targetable region using the Cre-loxP1 system. Deletion of exon5 results in a frame shift mutation disabling expression of Smarca5 protein in mouse [1]. For the experiments, human K562 cells (AML M6 subtype) were initially utilized as they were previously used for antisense oligonucleotide-mediated transient knockdown of SMARCA5 [2]. K562 cells were transfected by a pair of pX330-mVenus vectors containing sgRNAs complementary to a sequence in the SMARCA5 introns 4 & 5 and the the effect of CRISPR/Cas9-mediated deletion of exon5 was tested by PCR. Analysis of fragments amplified from genomic DNA of FACS-sorted mVenus-positive clonal populations identified 5 clones (#H10, D7, H4, E7, H7) with a single shortened PCR product (~632bp compared to 1175bp in controls) that were homozygously mutated ( Figure 2B). Sanger sequencing of PCR products confirmed that clones H10, D7, E7, and H5 contained the same deletion (543bp) and clone H4 an even larger deletion (582bp) within SMARCA5 exon5 ( Figure 2C). In addition, quantitative PCR and Western blot analyses of the cellular extracts confirmed that the Cas9-mediated deletion of the SMARCA5 gene resulted in loss of SMARCA5 expression ( Figure 2D,E). The resulting subclones had no expression of vector-coded & episomally expressed Cas9 nuclease. In addition, eight predicted off-target candidates (SRGAP2, RNF17, PRG4,

Smarca5 Deletion Inhibits Proliferation of Myeloblasts and Affects Function of Normal Stem Cells
To characterize the effect of SMARCA5 deletion in the AML-S5KO subclones, we monitored their growth in culture by the WST-1 assay correlating the number of metabolically active cells in the 72hr culture within a 96-well plate. We quantitated the data with a scanning multiwell spectrophotometer (ELISA reader) ( Figure 3A, upper panel) and also in parallel counted the viable cells with an automated cell counter ( Figure 3A, lower panel). We observed that starting day 1, the S5KO subclones produced less formazan product/s compared to AML 'control' cells, indicating that loss of SMARCA5 impaired proliferation of leukemic cells. We also attempted to create S5KO clones from additional AML cell lines. We repeatedly used OCI-M2, NB4, SKM1, MOLM-13, however, despite the fact that these AML cell lines grew normally in tissue culture conditions, the recombined cells by pX330-mVenus vectors followed by the single cell sorting could not produce clones with exon5 deletion. We therefore used the method of serial dilution of transfected cells. This approach, in contrast to the previous approach, produced populations of OCI-M2 and SKM1 cell lines with detectable Cas9-edited SMARCA5 loci. However, the signals of mutated alleles markedly decreased during long-term cultivation, suggesting that the S5KO cells were overgrown by cells containing at least one intact SMARCA5 allele. Thus, the deletion of the SMARCA5 gene completely impaired leukemic cell proliferation in most of the AML cell lines, while in K562 cells it was tolerated albeit under markedly lower proliferation activity, which allowed us to study it in more detail. (B) PCR verification of the exon5 deletion in the indicated S5KO clones. (C) Analysis of SMARCA5 gene region following the Cas9 nuclease deletion. PCR products (same as in B) were Sanger-sequenced and aligned with the wt sequence using the Kalign web tool. After sequencing, the precise length of the resultant PCR amplified region was determined (on the left in brackets). (D) Quantitative PCR analysis of SMARCA5 mRNA expression in the S5KO clones (n = 5) compared to controls (n = 10). Data normalized to the GAPDH mRNA. Student's t-test, p < 0.00001 ****. (E) Immunoblotting of SMARCA5 expression in CRISPR/Cas9-treated K562 or controls. β-actin controlled the load.

Smarca5 Deletion Inhibits Proliferation of Myeloblasts and Affects Function of Normal Stem Cells
To characterize the effect of SMARCA5 deletion in the AML-S5KO subclones, we monitored their growth in culture by the WST-1 assay correlating the number of metabolically active cells in the 72-hr culture within a 96-well plate. We quantitated the data with a scanning multiwell spectrophotometer (ELISA reader) ( Figure 3A, upper panel) and also in parallel counted the viable cells with an automated cell counter ( Figure 3A, lower panel). We observed that starting day 1, the S5KO subclones produced less formazan product/s compared to AML 'control' cells, indicating that loss of SMARCA5 impaired proliferation of leukemic cells. We also attempted to create S5KO clones from additional AML cell lines. We repeatedly used OCI-M2, NB4, SKM1, MOLM-13, however, despite the fact that these AML cell lines grew normally in tissue culture conditions, the recombined cells by pX330-mVenus vectors followed by the single cell sorting could not produce clones with exon5 deletion. We therefore used the method of serial dilution of transfected cells. This approach, in contrast to the previous approach, produced populations of OCI-M2 and SKM1 cell lines with detectable Cas9-edited SMARCA5 loci. However, the signals of mutated alleles markedly decreased during long-term cultivation, suggesting that the S5KO cells were overgrown by cells containing at least one intact SMARCA5 allele. Thus, the deletion of the SMARCA5 gene completely impaired leukemic cell proliferation in most of the AML cell lines, while in K562 cells it was tolerated albeit under markedly lower proliferation activity, which allowed us to study it in more detail.
whether Smarca5 deletion affects reconstitution of early blood progenitors after transplanting them into normal murine recipients, we utilized the hematopoietic reconstitution assay. We transferred E13.5 mouse fetal liver cells (C57Bl/6J Ly5.2 background) isolated either from control Smarca5 flox/+ Rosa26 eYFP/+ Vav1-iCRE or Smarca5-deficient (Smarca5 flox/-Rosa26 eYFP/+ Vav1-iCRE) embryos into lethally irradiated adult C57Bl/6J Ly5.1 recipients. Flow cytometric analyses of bone marrow and spleen at several weeks after transplantation revealed that repopulation was detected only in animals transplanted with cells in which the Smarca5 gene was preserved. Thus, homeostatic expression of Smarca5 is very important for hematopoietic reconstitution ( Figure 3B), implicating a possibility that the Smarca5 role in AML cells might also involve a very early leukemia-initiating compartment.

Inactivation of Smarca5 Causes Nuclear Abnormalities and Polyploidy
To gain insight into the subcellular structures of the AML S5KO cells, we utilized hematology staining using a standardized May-Grunwald and Giemsa-Romanowski stain procedure. As indicated within Figure 4A, the control AML cells were represented by a uniform layer of myeloblasts with large round nuclei, fine chromatin structure, and prominent nucleoli. Significantly more frequent nuclear abnormalities were observed in the S5KO cells compared to controls. These included nuclear budding, internuclear bridging, karyorrhexis, and multinuclearity seen in 10% to 65% of all analyzed cells ( Figure 4B). To study effect/s of S5 depletion in nonhematopoietic cells, we derived mouse embryonic fibroblast (MEF) with Tamoxifen-regulated Cre-recombinase activity (Cre-Esr1) from Smarca5 fl/fl Trp53 -/-animals. Trp53-mutated MEFs were chosen because of their lower propensity to enter proliferation senescence and because most AML cell lines including K562 have TP53 gene inactivation [22]. After 6 h incubation with 100 nM 4-hydroxy-tamoxifen (4OHT) and additional 90 h AML cell population resembles early hematopoietic progenitors. Thus, as controls to AML cells, we studied early murine blood progenitors. Previously it was shown that Smarca5 loss in mouse partially inhibits differentiation of early Lin − Sca-1 + c-Kit + hematopoietic progenitors [1]. To test whether Smarca5 deletion affects reconstitution of early blood progenitors after transplanting them into normal murine recipients, we utilized the hematopoietic reconstitution assay. We transferred E13.5 mouse fetal liver cells (C57Bl/6J Ly5.2 background) isolated either from control Smarca5 flox/+ Rosa26 eYFP/+ Vav1-iCRE or Smarca5-deficient (Smarca5 flox/− Rosa26 eYFP/+ Vav1-iCRE) embryos into lethally irradiated adult C57Bl/6J Ly5.1 recipients. Flow cytometric analyses of bone marrow and spleen at several weeks after transplantation revealed that repopulation was detected only in animals transplanted with cells in which the Smarca5 gene was preserved. Thus, homeostatic expression of Smarca5 is very important for hematopoietic reconstitution (Figure 3B), implicating a possibility that the Smarca5 role in AML cells might also involve a very early leukemia-initiating compartment.

Inactivation of Smarca5 Causes Nuclear Abnormalities and Polyploidy
To gain insight into the subcellular structures of the AML S5KO cells, we utilized hematology staining using a standardized May-Grunwald and Giemsa-Romanowski stain procedure. As indicated within Figure 4A, the control AML cells were represented by a uniform layer of myeloblasts with large round nuclei, fine chromatin structure, and prominent nucleoli. Significantly more frequent nuclear abnormalities were observed in the S5KO cells compared to controls. These included nuclear budding, internuclear bridging, karyorrhexis, and multinuclearity seen in 10% to 65% of all analyzed cells ( Figure 4B). To study effect/s of S5 depletion in nonhematopoietic cells, we derived mouse embryonic fibroblast (MEF) with Tamoxifen-regulated Cre-recombinase activity (Cre-Esr1) from Smarca5 fl/fl Trp53 −/− animals. Trp53-mutated MEFs were chosen because of their lower propensity to enter proliferation senescence and because most AML cell lines including K562 have TP53 gene inactivation [22]. After 6 h incubation with 100 nM 4-hydroxy-tamoxifen (4OHT) and additional 90 h of culture, the MEF cells were depleted from Smarca5 protein ( Figure 4C). Decrease of Smarca5 protein level negatively influenced the cell growth and the proliferation defect had already occurred within 40 h from the start of the 4OHT treatment while 4OHT untreated and control Cre-Esr1 lacking cells proliferated normally ( Figure 4D). This proliferative defect resembled one observed in AML S5KO clones. The flow cytometry analysis revealed that aberrant proliferation was accompanied by lower proportion of S-progressing and mitotic (pH3S10 + ) cells. In addition, we noted a higher number of cells with polyploid nuclei ( Figure 4E) that was concomitant to a decreased proportion of diploid cells upon S5 deficiency in MEFs. Taken together, inactivation of SMARCA5 triggers a cell proliferation blockade and results in nuclear abnormalities of exceedingly cycling leukemic as well as normal hematopoietic cells. of culture, the MEF cells were depleted from Smarca5 protein ( Figure 4C). Decrease of Smarca5 protein level negatively influenced the cell growth and the proliferation defect had already occurred within 40 h from the start of the 4OHT treatment while 4OHT untreated and control Cre-Esr1 lacking cells proliferated normally ( Figure 4D). This proliferative defect resembled one observed in AML S5KO clones. The flow cytometry analysis revealed that aberrant proliferation was accompanied by lower proportion of S-progressing and mitotic (pH3S10 + ) cells. In addition, we noted a higher number of cells with polyploid nuclei ( Figure 4E) that was concomitant to a decreased proportion of diploid cells upon S5 deficiency in MEFs. Taken together, inactivation of SMARCA5 triggers a cell proliferation blockade and results in nuclear abnormalities of exceedingly cycling leukemic as well as normal hematopoietic cells.

Cytogenetic Abnormalities and Gene Expression Dysregulation in the S5KO AML Cells
As pointed out in the Introduction section, SMARCA5 protein was previously shown to load cohesin complex onto human chromosomes [23]. As the canonical role of cohesin is the sister chromatid cohesion, we next analyzed the structures of mitotic chromosomes in the AML S5KO cells on metaphase spreads. The analysis of the S5KO subclone D7 consistently showed ( Figure 5A) that among other chromosomal abnormalities, the cohesion defects were by far the most frequent involving premature chromatid separation and loss of cohesion. Compared to the controls that contained only 12%, the S5KO mitotic cells displayed defects in chromatin cohesion in 70% of all cases. Similarly, the defects of chromatid cohesion were seen also in MEF cell-derived mitotic

Cytogenetic Abnormalities and Gene Expression Dysregulation in the S5KO AML Cells
As pointed out in the Introduction section, SMARCA5 protein was previously shown to load cohesin complex onto human chromosomes [23]. As the canonical role of cohesin is the sister chromatid cohesion, we next analyzed the structures of mitotic chromosomes in the AML S5KO cells on metaphase spreads. The analysis of the S5KO subclone D7 consistently showed ( Figure 5A) that among other chromosomal abnormalities, the cohesion defects were by far the most frequent involving premature chromatid separation and loss of cohesion. Compared to the controls that contained only 12%, the S5KO mitotic cells displayed defects in chromatin cohesion in 70% of all cases. Similarly, the defects of chromatid cohesion were seen also in MEF cell-derived mitotic chromosome spreads ( Figure 5B,C). These data suggest that SMARCA5 inhibition affects cohesin function in general.
1 transcripts with respect to SMARCA5. As expected, transcriptomic data from AML Cooperative Group München ( Figure 5D) showed an inverse correlation between SPI1/PU.1 and SMARCA5 expression in AML patient samples. To further assess the role of SMARCA5 in regulation of the hematopoietic transcription program, we determined the expression of a set of selected mRNAs upon the genetic ablation of the SMARCA5 gene in K562 cells. Compared with previously published data documenting an inverse relationship between SMARCA5 and hematopoietic transcription factors PU.1 or GATA-1, we observed that upon SMARCA5 deletion in K562 cells the level of SPI1/PU.1 and some of its targets (CSF1R) became downregulated while other transcription factors (GATA1, CBFB) were upregulated. The dysregulation of mRNA pattern of SMARCA5 targets upon SMARCA5 deletion can be attributed to the heterogeneity of the AML cell lines and also possibly to multiple genetic/cytogenetic abnormalities imposed by the SMARCA5 loss.  In order to better understand the cooperative nature of SMARCA5 and its interacting partners in AML, we correlated their expression using RNAseq data in AML patients. Hence, significant association exists between the expression pattern of SMARCA5 and BAZ proteins (BAZ1A, BAZ1B, BAZ2A, BAZ2B) as well as the members of the CTCF/cohesin complex across human AML samples. This implicates, albeit indirectly, a role of SMARCA5 in CTCF/cohesin function in AML that also coincides with karyotype abnormalities imposed by a SMARCA5 loss.
We recently showed that SMARCA5 (together with the CTCF/cohesin complex) represses PU.1-mediated myeloid differentiation [7] and similarly, we noted that SMARCA5 regulates GATA1-mediated erythropoiesis [1]. We therefore next decided to analyze the levels of SPI1/PU.1 and GATA-1 transcripts with respect to SMARCA5. As expected, transcriptomic data from AML Cooperative Group München ( Figure 5D) showed an inverse correlation between SPI1/PU.1 and SMARCA5 expression in AML patient samples. To further assess the role of SMARCA5 in regulation of the hematopoietic transcription program, we determined the expression of a set of selected mRNAs upon the genetic ablation of the SMARCA5 gene in K562 cells. Compared with previously published data documenting an inverse relationship between SMARCA5 and hematopoietic transcription factors PU.1 or GATA-1, we observed that upon SMARCA5 deletion in K562 cells the level of SPI1/PU.1 and some of its targets (CSF1R) became downregulated while other transcription factors (GATA1, CBFB) were upregulated. The dysregulation of mRNA pattern of SMARCA5 targets upon SMARCA5 deletion can be attributed to the heterogeneity of the AML cell lines and also possibly to multiple genetic/cytogenetic abnormalities imposed by the SMARCA5 loss.

Discussion
We herein studied how ISWI ATPase SMARCA5/SNF2H controls in AML the proliferation and gene expression of myeloblasts as SMARCA5 appeared to be an interesting target for anti-AML therapy. Our previous work demonstrated a pattern of SMARCA5 upregulation at AML diagnosis followed by its normalization upon achieving the hematologic remission. Importantly, additional work has not identified recurrent mutations of SMARCA5 in AML or any malignant disease (so far analyzed by next-generation sequencing-based techniques). For example, for the SMARCA5 gene, only 186 variants with an amino acid residue substitution exist in nearly~20 thousand oncologic patients (<1%). There also exist infrequently the variants in ISWI-interacting BAZ proteins detected in cancer, however, the significance of these variants remains also unknown. Importantly, among the AML-associated variants, only the SMARCA5-interacting proteins, CTCF and members of the cohesin complex, were shown consistently mutated in AML [24]. Based on this, we expected SMARCA5 indispensability for AML proliferation and its levels possibly reflecting the proliferative nature of AML cells. Indeed, the RNAseq analysis of a large set of AML patients confirmed that AML cells overexpressed SMARCA5 and its levels correlated with many ISWI-complex members including also cohesin complex, and finally, that the proliferative nature of AML cells marked by upregulation of SMARCA5 was supported by a trend in shorter OS albeit only in those AML patients that were marked by cytogenetic aberrations (see Figure 1).
Upon targeting of the SMARCA5 gene in AML cell lines with a CRISPR/Cas9-mediated deletion strategy, we could observe that AML cells lacking SMARCA5 markedly slowed the proliferation rate and became dysplastic with multiple karyotypic abnormalities. Inhibiting SMARCA5 to achieve suppression of AML growth may be thus a very efficient strategy as AML cells that are likely addicted to SMARCA5 in order to overcome various chromatin obstacles such as complex karyotype or also polyploidy often seen during progression of AML. Other data further implicated that SMARCA5 is very important also at the stem cell level to regulate their innate function: to repopulate the progeny. Indeed (as shown by Figure 3), repopulation activities were greatly reduced in normal hematopoietic stem cells in which the Smarca5 gene was genetically deleted. Our observation, however, does not rule out the possibility of SMARCA5 being an AML target as i) the AML cells are highly proliferating compared to their normal counterparts, and ii) SMARCA5 being expressed in stem cells implicates that antiSMARCA5 therapy would preferentially target the leukemia stem and progenitor cells.
While SMARCA5 expression represents a potential target for AML therapy, it may also serve as a factor of therapeutic resistance in AML. It is likely that additional factors will be involved in modulating therapy efficacy using SMARCA5 inhibitors in the future. As the Smarca5 loss was sensed in a mouse model by a) increased p53 levels and b) associated with DNA damage response (DDR), and c) activation of the p53 targets [1], very likely the tumor cells with DDR sensing defect would have a higher propensity to tolerate SMARCA5 level downregulation. This notion is supported by our other study demonstrating that proliferation defect imposed by Smarca5 deficiency can be partly restored with concomitant Trp53 deletion in murine thymocytes [3].
Our herein presented data indicate that AML growth is dependent on the expression of chromatin remodeling protein SMARCA5 that is a known partner of AML-associated targets: cohesin complex and CTCF [23]. Data presented in Figures 4 and 5 implicate that proliferation inhibition upon SMARCA5 targeting is at least in part caused by karyotype abnormalities, especially cohesion defects, and possibly also by a putative replication defect due to defective chromatin compaction as well as dysregulation of gene expression pattern of the key hematopoietic lineage restricted transcription factors. Interestingly, the nuclear changes after S5 deletion such as polyploidy were also described in other cell lines of hematologic origin [1,3] but not as a result of Smarca5 deletion of developing brain or eye lens [4,5]. Similar evidence was noted upon experimental manipulation with cohesin complex members; for example, the nonsense mutations in STAG2 (generated in the THP1 AML cell line) led to defects in sister chromatid cohesion and induced anaphase defects, which resulted in proliferation blockade [25]. Important connections between replication and cohesion have been established in the HeLa tumor cells, in which the interfering with replication affected chromatid cohesion and caused a defect in mitotic progression [26]. Others suggested that cohesion defects depend on a functional mitotic spindle checkpoint in regulating mitotic progression [27]. It seems that the strategy of inhibiting SMARCA5 in AML to block leukemogenesis becomes even more vital as shown recently using inhibitors of SMARCA5 (ED2-AD101) that target the HELICc-DExx domain to release the terminal AML cells into differentiation while sparing normal hematopoiesis in preclinical animal models [28]. Our work also suggests that upon inhibiting SMARCA5-mediated proliferation of AML cells, we also can face the problem of inhibiting proliferation of normal cells. Further work in this respect on experimental animals is under way. An additional strategy to inhibit AML cell growth specifically could be to target the SMARCA5 exon5 in AML cells by CRISPR/Cas9 as evidenced by the herein presented data. Data from global CRISPR/Cas9 screen identified that SMARCA5 targeting was very efficient and caused cell growth inhibition in several additional AML cell lines (OCI-AML2, OCI-AML3) and also in lymphoma and carcinoma cell lines [21]. Together, our as well as others' data demonstrate that SMARCA5 is a valuable epigenetic target suitable for inhibitor discovery projects and subsequent validation in MDS/AML and potentially also in other types of cancer.

CRISPR Vector Design
pX330-Venus (kindly provided by Dr. Bjoern Schuster) produces CRISPR/Cas9 enzyme that cleaves at a specific location based on sequence guide sgRNA defined target sequences in SMARCA5 intron4 (5 -TTCTTACGTTACCCATATACTGG-3 ) and SMARCA5 intron5 (5 -ATTTATCATATTTTCAGCGATGG-3 ). CRISPR/Cas9 enzyme is also fused with fluorescent protein mVenus, that enables selection of successfully transfected clones by FACS sorting. The DNA sequences for the sgRNA SMARCA5 intron4 and sgRNA SMARCA5 intron5 were synthesized by Sigma-Aldrich as four oligonucleotides with modifications at position 1 (to encode a Guanine due to the transcription initiation requirement of the human U6 promoter). These two pairs of complementary oligos were mixed together, boiled at 95 • C for 10 min, and allowed to cool down to RT to hybridize. Double-stranded oligos also designed with complementary BbsI overhangs on 3 and 5 ends were ligated into BbsI linearized pX330-Venus vector using T4 Ligase enzyme (Thermo Fisher Scientific, Waltham, MA, USA). Ligation mixtures were transformed into Subcloning Efficiency DH5α Competent Cells (Invitrogen, Carlsbad, CA, USA) following the manufacturer's protocol. pX330-Venus sgRNA hSMARCA5 intron4 and pX330-Venus sgRNA hSMARCA5 intron5 were isolated and purified by GenElute HP Plasmid Midiprep kit (Sigma-Aldrich, St. Louis, MO, USA) and correct oligo insertion verified by Sanger sequencing.

AML Patients and Statistics
RNA-Seq data sets from AML patient samples were previously described including the informed consent and ethical issues [29][30][31]. Reads were mapped with STAR aligner version 2.7.2d using GRCh37 reference and annotation version 32 from GENCODE (www.gencodegenes.org). Reads were counted using FeatureCounts version 1.6.5, normalized to transcripts per million (TPM) and log2 transformed. Log-rank test was performed in survival analysis, Wilcoxon test was used to assess differences in gene expression.

Cytogenetics
Standard cytogenetic methods published previously [10,11] were used for preparation of slides, with few modifications. Briefly, the K562 cells were synchronized with colcemid (10 µl/mL) at 37 • C and hypotonized in 0.075 M KCl for 20 min. The cells were then fixed in three changes of cold Carnoy's fixative (ethanol: glacial acetic acid, 3:1) and dropped onto a slide inclined at an angle of 45 degrees from a height. The chromosomal preparations were air-dried overnight and stained using 5% Giemsa blue solution (Sigma-Aldrich) prepared in standard Sorenson buffer. Preparations were inspected under a light microscope BX43 (Olympus, Sony, Shinjuku, Japan) with microscope camera Infinity 2-2 (Lumenera, Ottawa, ON, Canada). Selected plates were photographed under a 100x immersion oil objective using software QuickPHOTO CAMERA 3.1 (Olympus).

Analysis of S5KO MEF Cells
S5KO MEF cells (n = 3) were isolated from E14.5 embryos, in which the Smarca5 gene contained the LoxP1 sites upstream and downstream of exon5 and also expressed Cre Recombinase-Estrogen receptor fusion protein that translocated into the nucleus upon addition of 4OHT into the cultures for 6 h. Deletion of Smarca5-exon5 represents a null allele [2]. Production of stable MEF cells was enabled by concurrent deletion of Tp53 gene [32]. Gene targeting of the Smarca5 flox/flox Cre-Esr1 cells upon 4OHT addition was confirmed by previously published detection methods [2]. Analysis of cell proliferation of MEFs was determined by IncuCyte (Sartorius, Göttingen, Germany) that enables analysis in 96 wells under real-time continuous visualization and monitoring.
Author Contributions: CRISPR design and mouse transgenics and writing: T.Z., clone preparation and functional analysis: H.P., MEF cells: T.T., cytogenetics: S.T., IncuCyte: P.T., AML patient data and statistics: P.K. and P.A.G., hematopoietic reconstitution: J.K., supervision and writing: T.S. All authors have read and agree to the published version of the manuscript.