Systematic characterization of the HOXA9 downstream targets in MLL-r leukemia by noncoding CRISPR screens

Accumulating evidence indicates that HOXA9 dysregulation is necessary and sufficient for leukemic transformation and maintenance. However, it remains largely unknown how HOXA9, as a homeobox transcriptional factor, binds to noncoding regulatory sequences and controls the downstream genes. Here, we conduct dropout CRISPR screens against 229 HOXA9-bound peaks identified by ChIP-seq. Integrative data analysis identifies reproducible noncoding hits, including those located in the distal enhancer of FLT3 and intron of CDK6. The Cas9-editing and dCas9-KRAB silencing of the HOXA9-bound sites significantly reduce corresponding gene transcription and impair cell proliferation in vitro, and in vivo by transplantation into NSG female mice. In addition, RNA-seq, Q-PCR analysis, chromatin accessibility change, and chromatin conformation evaluation uncover the noncoding regulation mechanism of HOXA9 and its functional downstream genes. In summary, our work improves our understanding of how HOXA9-associated transcription programs reconstruct the regulatory network specifying MLL-r dependency.

by regulating cell proliferation and differentiation.Ablation of HOXA9 led to the arrested proliferation and enhanced apoptosis, especially in MLL-r cells 13 .Hence, HOXA9 represents a promising therapeutic target.However, targeting HOXA9 directly as a transcription factor is quite challenging.As an alternative, an epigenetic silencing upstream activator of HOXA9 has been investigated [14][15][16][17] .For instance, a large epigenetic complex consisting of the MLL-fusion protein and its cofactors, DOT1L 18 , WDR5 19 , MENIN 20 , and LEDGF 21 maintains constitutive expression of HOXA9 by a direct binding at the HOXA9 locus; therefore, inhibitors that disrupt the interaction between the MLL-fusion protein and its co-factors represent a promising way of abolishing the HOXA9 expression.So far, inhibitors targeting DOT1L have been developed and evaluated in clinical trials even though they have demonstrated limited therapeutic efficacy when used as single agents 22 .The MENIN inhibitors targeting the interaction between MLL1 and MENIN have also shown promising therapeutic potential in preclinical evaluations 23,24 .We previously established an endogenous HOXA9-knockin reporter in an MLL-r B-ALL cell line, SEM 25 .We have identified positive and negative regulators responsible for HOXA9 transcription through a loss-of-function CRISPR screen 26 .In contrast, few studies have explored the direct and functional targets of HOXA9 to reveal the molecular mechanism of its downstream signaling as an avenue to pursue potential therapeutic targets.
In this work, we use a CRISPR-mediated loss-of-function screen against HOXA9-bound peaks identified by ChIP-seq to systematically evaluate HOXA9-bound cis-regulatory elements in MLL-r leukemia cell models.We also conduct integrative analyses to reveal the long-range chromatin interaction regulation mechanism of genes essential for survival through HOXA9/MEIS1 binding occupancy.We demonstrate the downstream functional effectors of HOXA9 and provide additional clues to identify potential therapeutic targets in MLL-r leukemia.

Results
Genome-wide CRISPR dropout screening reveals essential HOXA9-bound cis-regulatory elements critical for leukemia maintenance Considering the transcription factor HOXA9 is an oncogene and critical to the survival of MLL-r leukemia cells, exploring its direct downstream targets promises to provide insights into the HOXA9 regulome and to identify additional therapeutic targets.We hypothesized that genome editing-mediated disruption of HOXA9-bound sites should mimic the loss-of-function effect of HOXA9.This approach would affect functional target gene expression, impair cell fitness, and prevent functional compensation by other HOXA genes.To this end, we used Cas9-mediated CRISPR screening to perturb HOXA9-bound sites genome-wide.Remarkably, no high-quality published ChIP-seq data is available to locate reproducible HOXA9-bound sites.Based on these observations, we established a doxycycline-modulated Tet-On system to ectopically express HA-tagged HOXA9 cDNA in MLL-r SEM cells (Fig. 1A and Supplementary Fig. 1A).Then, ChIP-seq was performed using specific antibodies against the HA tag in both nodoxycycline and doxycycline-treated cells.In total, we identified 229 reproducible HOXA9-bound peaks in SEM cells.Among these peaks, 50% bind to introns, 12.5% bind to promoters, and 29.1% bind to distal regions (Fig. 1B).Motif analysis confirmed that the top consensus motif is HOXA9 (Fig. 1C).Next, we analyzed the co-factor profiling of the 229 peaks based on collected ChIP-seq data of transcription factors (TFs) from human B-ALL SEM.These data suggest MEIS1 (a known co-factor of HOXA9) and other leukemia-specific TFs co-bind with HOXA9 (Fig. 1D and Supplementary Fig. 1B).
Next, we designed 5718 sgRNAs to target the 229 HOXA9-binding peaks, including 100 non-target sgRNAs and 120 positive-control sgRNAs against ribosomal genes.The pooled lentiviral sgRNA library was used to infect Cas9-expressing or dCas9-KRAB-expressing MLL-r SEM cells, followed by a survival-based dropout screen.The loss of representation of sgRNAs at day 7 and day 14, as compared with their presence on day 0, implies that CRISPR targeting of the candidate targeted regions impairs cell growth and survival, as exemplified by sgRNAs targeting RPS19, an essential gene that served as a positive control (Fig. 1E-I).The MAGeCK 27 analysis algorithm was used to identify sgRNA depleted during the two-week culture.Notably, sgRNAs (FLT3 Enh -left and right) targeting a known distal enhancer of the FLT3 locus 28 were among the top drop-out candidate sgRNAs (FDR < 0.05) in both Cas9 and dCas9-KRAB screens (Fig. 1F-I).Our dropout screen system against HOXA9-bound noncoding regions reliably identified putative target genes of the HOXA9 protein.
Epigenetic regulation of FLT3 transcription and the biological significance of HOXA9 and FLT3 regulation through the distal enhancer regulation The FLT3 gene is well known to be essential for survival in MLL-r subtypes based on genomic and genetic evidence [29][30][31] .Although several studies reported the HOXA9 protein could bind and regulate FLT3 at the promoter in mouse settings 32,33 , the detailed transcriptional regulation mechanism has not been addressed.Moreover, we identified a different HOXA9-bound site distant from FLT3 in MLL-r leukemia cells.To further explore the regulatory mechanism of the HOXA9-bound cisregulatory element regulating the FLT3 gene, we used integrative analysis of epigenetic profiling and genome editing approaches.First, we analyzed publicly available ATAC-seq datasets derived from a panel of human acute lymphoblastic leukemia (ALL) cell lines with and without MLL rearrangement and 13 normal human hematopoietic lineages 34 .A distal human cis-regulatory element (CRE) (h-FLT3) approximately 170 kb upstream of FLT3 showed significantly higher chromatin accessibility in MLL-r ALL cells (SEM and RS4;11 cells), as compared with non-MLL-r ALL cells (697, Nalm6, REH, SUPB15, and UOCB-1 cells) (Fig. 2A).Interestingly, a solid ATAC-seq peak was also observed in HSPCs (CD34 + , HSC, MPP, CMP, GMP, MEP, and CLP cells) but not in mature lineages (monocytes, erythrocytes, CD4 + , and CD8 + T cells, NK cells, and B cells)(Fig.2A) 34 .Similarly, the enhancer activity of h-FLT3 was supported by more robust H3K27ac ChIP-seq signals in MLL-r acute myeloid leukemia (AML) cells (MOLM13, MOLM14, and MV4;11 cells) as compared with non-MLL-r AML cells (IMSM2, HEL, K562, SET2, U937, and HL60 cells) and normal CD34+ cells (Fig. 2B) [35][36][37] .Notably, within the FLT3 distal enhancer, we confirmed a specific binding peak of both HOXA9 and MEIS1 using our HOXA9-HA ChIP-seq and publicly available HOXA9 and MEIS1 ChIP-seq datasets.Notably, the HOXA9-bound peaks did not show CTCF binding occupancy within the ± 2Kb, suggesting that transcription regulation through HOXA9 likely occurs independently of co-regulation from the looping factor CTCF.
By exploring HiC data from SEM cells 25 , we found evidence of a long-range DNA interaction between the FLT3 promoter and the HOXA9/MEIS1-bound h-FLT3 enhancer (Fig. 2C).Promoter Capture-C data further confirmed this looping event (Fig. 2D).We have previously established an auxin-inducible degron model of looping factor CTCF in SEM cells 25 , allowing acute depletion of CTCF protein (Fig. 2E).Using this cellular model, we demonstrated that upon CTCF loss, FLT3 expression was indeed not altered at both protein and total mRNA levels (Fig. 2F-H).In addition, we also confirmed that nascent RNA of FLT3 and neighbor gene PAN3 were not changed as characterized by SLAM-seq (Supplementary Fig. 2A, B).Interestingly, neither the chromatin accessibility at the h-FLT3 enhancer nor the chromatin loop between the FLT3 promoter and enhancer was altered (Fig. 2I), suggesting that additional chromatin looping factors other than CTCF may be involved in controlling FLT3 transcription.
Consistent with previous reports [29][30][31][38][39][40] , FLT3 expression was significantly higher in MLL-r leukemia cell lines (n = 11) than in non-MLL-r leukemia cell lines (n = 89) (p < 0.0001). Also,MLL-r leukemia cell lines (n = 8) showed more survival dependency on FLT3 than non-MLL-r leukemia cell lines (n = 39) (p < 0.0001), according to the CRISPR essential score from the DepMap dataset 41 .Furthermore, there was a significant negative correlation (r = −0.684,p = 8.28e−8) between the CRISPR essential score and FLT3 expression (Supplementary Fig. 3A-C).Of note, according to the TCGA database, high FLT3 expression was associated with poor overall survival in a cohort of 106 patients with AML (Supplementary Fig. 3D, E).A similar trend was also observed in the TARGET and BEAT cohorts (Supplementary Fig. 3F-I).Collectively, the results of the epigenetic profiling, the chromatin interaction, and the expression correlation suggest that there is h-FLT3-mediated transcriptional activation of FLT3 in MLL-r leukemia cells.Inspired by the profiling and correlation analysis results, we conducted functional validation in both MLL-r and non-MLL-r cell lines.We found that the HOXA9-bound site is specific at the distal enhancer of FLT3 but not the promoter (Fig. 3A, B).Therefore, two sgRNAs (sgFLT3-DE-1 and sgFLT3-DE-2) from the CRISPR sgRNA library were selected to disrupt the HOXA9 occupancy motif at the h-FLT3 enhancer (Fig. 3C).TIDE-seq analysis confirmed that more than 90% indel frequency was observed (Supplementary Fig. 4A, B).As a result, Cas9-mediated genome editing by these two sgRNAs resulted in significant downregulation of FLT3 expression and the adjacent PAN3, compared with that of the non-target control (Fig. 3D).Subsequently, we conducted a time-course competitive proliferation assay (CPA) to characterize the impact on the proliferation of decreased FLT3 expression.The CFP fluorescence in the sgRNAexpressing lentiviral vectors was used to monitor the compromised cell survival observed in MLL-r SEM cells (Fig. 3E), as targeted by sgFLT3-DE-1 and sgFLT3-DE-2.Similar results were observed in two other MLL-r AML cell lines (MOLM13 and MV4;11) (Fig. 3F-G) but not in non-MLL-r cell lines (Nalm6 and GM12878) (Fig. 3H, I).Of note, the MLL-r AML OCI-AML2 cells expressed high FLT3 expression but displayed weak h-FLT3 enhancer activity as indicated by H3K27ac ChIPseq (Fig. 3C).Therefore, as expected, there was no change in cellular phenotype in OCI-AML2 cells infected with sgFLT3-DE-1 and sgFLT3-DE-2 (Fig. 3J).In summary, genome editing of the HOXA9 binding site in the h-FLT3 enhancer functionally impaired cell proliferation in the context of MLL-r.It has been reported that double-stranded breaks may induce cell death and may cause the HOXA9-binding motif-independent phenotype 42,43 .To mitigate this potential impact, we designed additional complementary experiments to confirm the result obtained from Cas9-mediated genome editing.Firstly, a full-length wildtype FLT3 cDNA was overexpressed via the lentiviral system in the h-FLT3 enhancer-targeted SEM cells.This resulted in a significant rescue at the protein level and consistent recovery in CPA (Supplementary Fig. 4C, D).Second, we used the CRISPR interference (CRISPRi) assay by blocking the chromatin accessibility of the HOXA9-bound site at the h-FLT3 enhancer with the same sgRNA DE-1 in SEM cells stably expressing dCas9-KRAB (Fig. 4A).As expected, on a genome-wide scale, the ATAC-seq peak that was most significantly decreased upon CRISPRi targeting by sgFLT3-DE-1, as compared with the non-target control sgRNA, corresponded to our target locus (FDR < 0.05) (Fig. 4B, C).Subsequently, expressions of both FLT3 and PAN3 were significantly downregulated compared with the non-target control sgRNA (Fig. 4D, E).Moreover, consistent with our observations with Cas9-edited cells, cell growth was significantly retarded, as indicated by the time-dependent decrease in the CFP + percentage of dCas9-KRAB SEM cells that were positive for sgFLT3-DE-1 (Fig. 4F). Hoever, the interaction between the FLT3 promoter and the h-FLT3 enhancer was not altered by blocking the binding of HOXA9 at the h-FLT3 enhancer with sgFLT3-DE1 sgRNA, as indicated by Capture-C using two baits (bait 1 and 2) against the FLT3 promoter (Fig. 4G).Our previous ChIP-seq data showed that the binding peaks of CTCF and HOXA9 at this locus are ~5Kb away.Therefore, targeting the HOXA9bound site does not affect CTCF or CTCF-associated chromatin interaction.Also, it is possible that other chromatin looping factors are required for FLT3 enhancer/promoter regulation.Finally, FLT3 was an essential gene in MLL-r according to the DepMap dataset (Fig. 4H).As expected, MLL-r SEM cells exhibited superior sensitivity to pharmacological inhibition of FLT3 by Gilteritinib (LC50: 54 nM) when compared to Nalm6 (LC50: 1852 nM), REH (LC50: 1523 nM), and 697 (LC50: 1729) cells (Fig. 4I).Moreover, sgFLT3-DE-1-mediated CRISPRi blunted the drug response of SEM cells to Gilteritinib (Fig. 4J).
To test further the effect of disrupting the h-FLT3 enhancer on the in vivo growth of SEM cells, we performed an in vivo competitive assay by transplanting a mix of sgFLT3-DE1 SEM cells (CFP + cells) expressing either dCas9-KRAB or Cas9 and parental SEM cells (CFP -cells) into NSG mice at a ratio of 4:1 (Fig. 5A).The percentage of CFP + SEM cells out of the total SEM cells (hCD45 + cells) in the peripheral blood was monitored by flow cytometry.As shown in Fig. 5B, C, the in vivo cell growth of SEM cells positive for sgFLT3-DE-1 was outcompeted by parental SEM cells 2 and 3 weeks after injection.Our use of integrative genetic tools revealed the HOXA9-binding and cis-regulatory regulation mechanism of FLT3 in MLL-r leukemia models.

Functional validation of additional HOXA9 targets
In addition to sgRNAs (sgFLT3-DE-1 and sgFLT3-DE-2) targeting h-FLT3 enhancer, several other sgRNAs were also enriched in the dropout CRISPR screen.To test the functional relevance with MLL-r survival, we conducted a CPA assay in Cas9-expressing SEM and MOLM13 cells targeted with sgRNAs against the HOXA9-bound sites in the genes of CDK6, RUNX1, DCAF11, PEBP4, AHI1, NDUFS8, ZCCHC7 and MAN1C1.TIDE-seq analysis confirmed that indels were observed in the coding regions of each gene by CRISPR-Cas9 targeting (Supplementary Fig. 5).Consistent with our dropout screen results, all the targeted cells showed impaired proliferation (Fig. 6A, B).To independently validate the targeting effect, the sgRNAs against CDK6, RUNX1, and DCAF11 were delivered into dCas9-KRAB expressing SEM and MOLM13 cells.CRISPRi system reproducibly confirmed the decreased proliferation phenotype (Fig. 6C, D).To further reveal the molecular mechanism underlying the HOXA9-target gene regulation, we focused on the CDK6 locus.The HOXA9-bound site is located in the intronic region and colocalizes with enhancer markers H3K27ac and BRD4 (Fig. 1D and Supplementary Fig. 6).HiC data also reveal strong loops between this region and the CDK6 promoter (Fig. 6E).We also conducted ATAC-seq on SEM cells targeted with sgCDK6 (HOXA9-bound site in CDK6 intron) and sgNT.These data demonstrated that CRISPRi against the HOXA9bound site in CDK6 intron significantly decreased the chromatin accessibility and the CDK6 expression (Fig. 6E-G, and Supplementary Fig. 7A, B).In contrast, chromatin accessibility of DCAF11 and RUNX1 loci was not obviously altered, pointing to the presence of alternative mechanisms (Supplementary Fig. 7C, D).Total RNA-seq and differential gene expression analysis confirmed that the CDK6 transcription is the most significantly decreased gene upon CRISPRi (FDR < 0.01), further supporting distal noncoding regulation is specific to CDK6 (Fig. 6H).
Fig. 1 | A CRISPR editing screen discovered HOXA9 binding sites essential for the growth of MLL-r leukemia cells.A Diagram of the all-in-one inducible Tet-On system to ectopically express HOXA9-HA cDNA.B Genomic distribution of the 229 HOXA9 ChIP-seq peaks.C The top three consensus motifs and P value were shown using the HOMER motif analysis algorithm against 229 peaks compared with control peaks.D Heat maps of HOXA9-HA ChIP-seq peaks compared with publicly available HOXA9 and MEIS1 ChIP-seq datasets and additional transcription factor ChIP-seq datasets in SEM cells (GEO: GSE117864) 70 .E The 229 reproducible HOXA9bound peaks were identified from our ChIP-seq data and targeted by a library of 5718 sgRNAs, including 100 non-target (NT) sgRNAs used as negative controls and In addition to HOXA9-FLT3 and HOXA9-CDK6 regulation axis, it will be interesting to investigate the noncoding regulation of other downstream target genes by HOXA9 in a follow-up study in the future.Taken together, these extensive efforts will undoubtedly improve our understanding of how HOXA9 works in MLL-r leukemia maintenance and provide insights for alternative targeting and therapeutic innovation.Fig. 7 Discussion Numerous studies have suggested that dysregulation of HOXA9 is a dominant driver in the pathogenesis of MLL-r leukemia.Aberrant high expression of HOXA9 is a hallmark of MLL-r leukemia, including AML and ALL, and is associated with a poor prognosis for this disease.Nonetheless, the exact downstream targets of HOXA9 accounting for its oncogenic role remain to be fully elucidated due to the lack of commercial ChIP-seq grade antibody available 44 .Moreover, HOXA9 is undruggable, adding another layer of difficulty to leverage the therapeutic potential of targeting HOXA9.CRISPR-based (either Cas9 or dCas9) screenings have recently been reported to efficiently manipulate the transcription of target genes by perturbating noncoding cisregulatory elements [45][46][47] .Inspired by these studies, we established the HA-tag-based HOXA9 ChIP-seq using an inducible expression system and identified 229 reproducible HOXA9-bound peaks in MLL-r leukemia cell lines.Subsequently, we designed a library of sgRNAs targeting binding sites of HOXA9 and performed Cas9-based genome editing, aiming to determine functional cis-regulatory elements: 1) directly targeted by HOXA9; 2) responsible for cell fitness.Notably, several distal cis-regulatory elements influencing cell fitness associated with poorly explored target genes were identified.Targeting the HOXA9bound sites by CRISPR/Cas9 or CRISPRi reduced the chromatin accessibility of targeted regions (Fig. 4B and Supplementary Fig. 7A, B), substantially compromised the enhancer activity of HOXA9-bound sites.However, targeting other HOXA9-bound sites at the promoters of RUNX1 and DCAF11 did not impact chromatin accessibility, indicating other mechanisms may exist (Supplementary Fig. 7C, D).
In this study, we explored the functional binding sites of HOXA9 using CRISPR/Cas9-based genome editing, and a subsequent dropout CRISPR screen in MLL-r positive SEM cells.We identified a series of HOXA9-driven cis-regulatory elements that are critical for the survival of MLL-r leukemia cells, including h-FLT3, a distal enhancer critical for the high expression of FLT3 in MLL-r leukemia and its survival 33,48 .Another example is HOXA9-binding and regulation at the intronic region of CDK6.Both genes were essential for MLL-r leukemia cell survival.Our integrative mechanism studies revealed that chromatin-looping mediated enhancer represents another layer of regulation in the HOXA9/ downstream gene axis.Our data also suggest that additional looping factors other than CTCF may be required for the HOXA9 regulation.
Notably, most of the functionally validated HOXA9-bound sites co-localized with enhancer markers as H3K27ac and BRD4 regardless of their genomic location (Supplementary Fig. 6), suggesting HOXA9 may be functional through controlling enhancer activity.These observations are consistent with the previous report that HOXA9 may reprogram the enhancer landscape to promote leukemogenesis 49 .However, more stringent functional assays and research tools are required to explore the detailed mechanism.
High expression of FLT3 has been recognized as a critical component of the unique identity of MLL-r ALL, compared to non-MLL-r ALL and AML 50 .However, the underlying mechanism for the dysregulation of FLT3 expression is largely unknown.In this study, we have identified a distal enhancer (h-FLT3) highly responsible for FLT3 transcription under the direct control of HOXA9.Disrupting the binding of HOXA9 at this enhancer via CRISPR editing resulted in significant downregulation of FLT3.Collectively, our results determined FLT3 as a downstream effector of HOXA9 and established the axis of MLL-r fusion, HOXA9, and FLT3 in MLL-r leukemia.Moreover, high FLT3 expression also conferred preferential sensitivity to FLT3 inhibition, especially in MLL-r leukemia cells 38,39 .Consistent with these observations, down-regulation of FLT3 by disrupting HOXA9 binding at the h-FLT3 enhancer led to decreased growth of MLL-r leukemia cells in vitro and in vivo and decreased their sensitivity to the FLT3 inhibitor, Gilteritinib.
HOXA9 overexpression was also identified in many different leukemia subtypes, including onco-fusion proteins, NPM1c+, etc.; experimental validation of the target binding site and cellular function is interesting for understanding HOXA9's role in other HOXA9-driven leukemia cells.Inspired by the NUP98-HOXA9 fusion containing the homeodomain of HOXA9 protein, we think it is also worth using the NUP98-HOXA9 AML model to further evaluate the HOXA9 binding peak function.
In summary, we have conducted a genome-wide CRISPR screen for HOXA9-binding sites, leading to a refined understanding of the role of HOXA9 in noncoding segment regulation.The system itself was confirmed to be a powerful tool for identifying downstream targets of HOXA9 that represent potential therapeutic targets for MLL-r and potentially for other HOXA9-driven leukemia (Fig. 7).

Methods
Our research complies with all relevant ethical regulations.Protocols for mouse studies were approved by the St. Jude Children's Research Hospital Institutional Animal Care and Use Committee.

Cell culture
The REH, Nalm6, and MV4;11 cell lines were originally purchased from ATCC, and the SEM, OCI-AML-2, and 697 cell lines were originally purchased from DSMZ.MOLM13 was originally purchased from ACCEGEN.GM12878 was from Coriell Institute.All cell lines were maintained in RPMI 1640 medium (GIBCO, Life Technologies) supplemented with 10% heat-inactivated fetal bovine serum and 2 mM Lglutamine, at 37 °C in 5% CO 2 .A high-titer virus was generated using 293 T cells in this study, which were maintained in DMEM (GIBCO, Life Technologies) supplemented with 10% heat-inactivated fetal bovine serum and 2 mM L-glutamine, at 37 °C in 5% CO 2 .All cell lines were validated by STR before use in this study.All cell lines used in this study were Mycoplasma-free examined by Lookout Mycoplasma PCR Detection Kit (Sigma, #MP0035) following the instruction.

Vectors
The Lenti-Cas9-Blast (Addgene #83480), Lenti-dCas9-KRAB-blast (Addgene #89567), and the Lenti-Guide-Puro (Addgene #52963) plasmids were purchased from Addgene.The IRES-CFP cassette was Fig. 2 | Characterizing the epigenetic regulation of FLT3.A Functional annotation of the FLT3 locus, using ATAC-seq data from seven B-ALL leukemia cell lines either with the MLL gene rearrangement (MLL-r cells) (SEM and RS4;11) or without MLL rearrangement (non-MLL-r cells) (697, Nalm6, REH, SUP-B15, and UOC-B1) (GEO: GSE129066) 68 and from 13 different normal human hematopoietic lineages (GEO: GSE74912) 34 .A potential FLT3 enhancer (FLT3 Enh ) at ~170 kb upstream of FLT3 was observed in all leukemia cell lines, normal human stem cells, progenitor cells, and B cells.The signal of the FLT3 enhancer was significantly more robust in cells with the MLL gene rearrangement than in cells without the MLL rearrangement.B A strong H3K27ac ChIP-seq signal was observed at the FLT3 enhancer in AML cells with the MLL rearrangement but not in AML cells without MLL rearrangement (upper panel)(GSE17312 71 , GSE65138 36 , GSE80779 35 , GSE79899 72 , GSE109492 73 , GSE137652 74 , GSE111179 76 , GSE111293 75 ).Strong CTCF, HOXA9, and MEIS1 ChIP-seq signals were also observed at the h-FLT3 enhancer locus in SEM cells (lower panel).C HiC data from parental SEM cells demonstrated long-range chromatin interaction between the FLT3 promoter and the h-FLT3 enhancer (upper panel)(GEO: GSE138862).ChIP-seq data (HOXA9, MEIS1, and CTCF) served as a reference to locate the HOXA9-bound site (lower panel).D Capture-C was conducted to characterize the chromatin looping between the FLT3 promoter and the h-FLT3 enhancer using biotin-labeled DNA oligos against the FLT3 promoter.E Schematic diagram of auxin-inducible degron system to target and acutely deplete CTCF protein.F Immunoblotting was conducted to confirm the acute protein degradation of CTCF in the presence of auxin for 24 hours.The results were confirmed by consistent two replicates and one representative result was shown here.G Immunoblotting was performed to detect the expression of FLT3 protein upon acute degradation of CTCF.GAPDH was included as an internal control.H Q-PCR was used to examine the mRNA expression of FLT3 upon CTCF protein degradation.Data are shown as mean values ± SEM of three biological replicates (center of the error bar).P values were estimated using a two-tail un-paired t-test.I Capture-C and ATAC-seq were conducted to characterize the chromatin accessibility change and chromatin conformation change following CTCF degradation, respectively.The biotin-labeled DNA oligos against the FLT3 promoter were used as baits to quantify chromatin looping between the FLT3 promoter and the h-FLT3 enhancer.Source data are provided as a Source Data File.Cas9/sgFLT3-DE-1/2-mediated disruption of the h-FLT3 enhancer resulted in retarded cell growth of MLL-r leukemia cells (E-G) but not non-MLL-r leukemia cells or OCI-AML-2 cells with low enhancer activity of the h-FLT3 (H-J).The sgRPS19-11 sgRNA served as the positive control.The percentage of cell numbers was normalized to CFP + control cells (sgNT).Data are shown as mean values ± SEM of three biological replicates (center of the error bar).P values were estimated using a twotail un-paired t-test.Source data are provided as a Source Data File.cloned into the Lenti-Guide-Puro plasmid to enable the flow tracing of sgRNA-targeted cells in a competitive proliferation assay.A pair of oligonucleotides containing a 20-bp sgRNA sequence targeting the candidate region was synthesized (Thermo Fisher Scientific) and cloned into the Lenti-Guide-Puro-IRES-CFP construct between two BsmBI sites.Correct clones were selected by screening and confirmed by Sanger sequencing with the U6-Forward sequencing primer (5'-GAGGGCCTATTTCCCATGAT-3').For lentiviral overexpression of FLT3, the FLT3 cDNA containing the coding region was cloned with a pair of primers (Supplementary Data 1) and inserted into the lentiviral expression vector by Gibson Assembly.SnapGene software was used to design all primers used for cloning.The PCR amplifications of products for cloning were performed using CloneAmp polymerase (Clontech), and the cycling parameters were as follows: 98 °C for 5 min, followed by 40 cycles of 98 °C for 15 s, 55 °C for 20 s, and 72 °C for 20 s.A Gibson Assembly Cloning Kit (NEB #E5510S) was used in accordance with the manufacturer's instruction, and all reactions were carried out at 50 °C for 20 min.All primer information was included in Supplementary Data 1.

CRISPR library construction and screening
A set of approximately 5718 sgRNA oligonucleotides that targeted 229 HOXA9 binding peaks carrying consensus motifs was designed for array-based oligonucleotide synthesis (CustomArray).The unique binding of each sgRNA was verified by sequence blast against the A A dCas9-KRAB-mediated CRISPR interfering system was applied to disrupt the h-FLT3 enhancer.B, C ATAC-seq results showed that Cas9-KRAB/sgFLT3-DE-1 only influenced the chromatin accessibility of the h-FLT3 enhancer locus in a genomewide analysis.The non-relevant RBM45 locus was one of the negative loci.D-F dCas9-KRAB/sgFLT3-DE-1-mediated disruption of the h-FLT3 enhancer led to down-regulation of FLT3 and PAN3 expressions and retarded cell growth of SEM cells.Expressions of FLT3 and PAN3 were normalized to SEM (dCas9-KRAB) cells infected with non-target sgRNA (sgNT).The percentage of cell numbers was normalized to CFP + control cells.Data are shown as mean values ± SEM of three biological replicates (center of the error bar), and results represent three independent experiments.P values were estimated using a two-tail un-paired t-test.G Capture-C results showed a strong interaction between the h-FLT3 enhancer and the FLT3 promoter, and this interaction was not influenced by dCas9-KRAB/sgFLT3-DE-1mediated disruption of the h-FLT3 enhancer.Two baits (Bait 1 and Bait 2) were designed against the FLT3 promoter.FLT3 was shown as an essential gene in SEM (MLL-r) cells, indicated by gene effect (<-0.5 is recognized as essential)(DEPMAP) (H).Its high expression conferred SEM superior sensitivity to the FLT3 inhibitor Gilteritinib (I), whereas dCas9-KRAB/sgFLT3-DE-1-mediated disruption of the h-FLT3 enhancer led to the down-regulated expression of FLT3 and decreased sensitivity to Gilteritinib (J).Data are shown as mean % viability relative to vehicle ± SEM of three biological replicates (center of the error bar), and results represent three independent experiments.Source data are provided as a Source Data File.After the cells had recovered from the selection process, the pooled population was collected on day 0, and portions were maintained in liquid culture for 7 or 14 days.The sgRNA sequences were recovered by genomic PCR analysis and deep sequencing using a HiSeq 4000 or Novasesq system (Illumina) for single-end 150-bp reads.The primer sequences used for cloning and sequencing are listed in Supplementary Data 1.The sgRNA sequences and counts are described in Supplementary Data 2.
Reverse transcription was performed using a High-Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems #4374966).Real-time PCR was performed using FAST SYBR Green Master Mix (Applied Biosystems #4385612) in accordance with the manufacturer's instructions.Relative gene expression was determined by using the ΔΔ-CT method 51 .All Q-PCR primers used in this study are listed in Supplementary Data 1.

RNA-seq
Total RNA was extracted by Trizol (Thermo Fisher Scientific #15596026) from replicate samples.About 200 ng total RNA was treated using Kapa rRNA depletion reagents to remove ribosomal RNA, then converted into cDNA libraries using Kapa RNA HyperPrep Kit with RiboErase (HMR).After the end repair procedure, dA-tailing, and adapter ligation, each cDNA library was purified and enriched by 11 cycles of PCR amplification.

RNA-seq data analysis
Paired-end 11-cycle sequencing was performed on the NovaSeq 6000 sequencer following the manufacturer's instructions (Illumina).Raw reads were first trimmed using TrimGalore (v0.6.3) with parameters '--paired --retain_unpaired.' Filtered reads were then mapped to the Homo sapiens reference genome GRCh37.p13using STAR (v2.7.9a) 52 .Gene-level read quantification was done using RSEM (v1.3.1) 53on the Gencode annotation v19 54 .To identify the differentially expressed genes, normalization factors were first estimated using the TMM, and genes with CPM < 1 in all samples were removed.Next, the TMM normalization factors and raw counts were then used for the Limma-voom analysis using the "voom," "lmFit" and "eBayes" functions from the limma R package 55 .Gene set enrichment analysis (GSEA) was performed using the MsigDB database (v7.1) 56,57 .Differentially expressed genes were ranked based on their log 2 (FC).RNA expression value was provided in Supplementary Data 3.

Competitive proliferation assay
To evaluate candidate sgRNAs' impact on leukemia expansion, cell cultures were transduced with individual sgRNAs in CFP expressing lentiviral vectors, followed by measurement of the percentage of CFPpositive cells at various days post-infection, using flow cytometry 26,58 .
The CFP-positive percentage was normalized to that at the starting time point and declined over time.The decline was used to infer a defect in cell accumulation conferred by a given sgRNA associated with survival, relative to the uninfected cells in the same culture.An infected pool population could also be injected into host NSG mice via the tail vein.Blood samples collected from the eyeballs were used for flow cytometric analysis of the CFP-positive percentage.

ChIP-seq analysis
Twenty million HOXA9-HA cells were treated with 1ug/mL doxycycline for 48 hours to induce the HOXA9-HA protein expression.DMSO treatment was used as a negative control.Cells were fixed with 1% formaldehyde for 5 mins at room temperature (Covaris TruChIP Chromatin Shearing Kit).Nuclei were prepared according to the Tru-ChIP protocol, and chromatin was sheared in a Covaris milli tube using the Covaris M220 ultrasonicator set at a duty factor of 10 and 200 cycles/burst for 10 min at set point 6 °C.Sheared chromatin was centrifuged for 10 mins at 8000 × g, and clarified chromatin was moved to a new 1.To ensure replicability, reproducible peaks for each group were finalized as only peaks retained if called with a stringent cutoff (-q 0.05) in one sample and at least called with a lower cutoff (-q 0.5) in the other samples.Homer was used for motif analysis 61 .HOXA9-bound peak files were provided in Supplementary Data 4. Raw ChIP-seq data were deposited in the dataset GSE215928 in GEO.

ATAC-seq
Briefly, 10,000 SEM cells were treated with a transposition mix consisting of 22 μL nuclease-free water; 25 μL 2 × TD Buffer; 2.5 μL TDE1 transposase; and 0.5 μL of 1% digitonin (added immediately before removing the supernatant from cell pellets), followed by incubation at 37 °C on an Eppendorf ThmorMixer at 300 rpm for 30 minutes.DNA was purified using the MinElute Reaction Cleanup Kit (Qiagen, 28204) and amplified for 5 cycles using the PCR mix composed of 10 μL transposed DNA, 2.2 μL nuclease-free water, 6.25 μL 10 μM barcoded Nextera primer-F, 6.25 μL 10 μM barcoded Nextera primer-R, 0.3 μL 100 × SYBR Green I (Invitrogen, S-7563), and 25 μL NEBNext® High-Fidelity 2 × PCR Master Mix (NEB, M0541).PCR products were subsequently re-amplified for another 6 PCR cycles, followed by purification using SPRIselect beads (Beckman Coulter, B23317).Paired-end sequencing (2 × 100 bp) was performed using the Illumina HISeq 4000 platform.Paired-end reads were analyzed using cutadapt (version 1.18) 62 for adaptor trimming and then mapped to the human hg19 genome by Bowtie2 (version 2.2.9) 63 .Peak calling was performed on each sample by using MACS2 64 with default parameters, and peaks were then merged by BEDtools (version 2.25.0) 65to retain nonoverlapped regions for identifying differentially enriched peaks, using ABSSeq under the aFold model 66 with read count from HTSeq 67 .A cutoff of the adjusted P-value (FDR) < 0.05 and Ilog 2 fold changeI ≥1 was used to define high-confidence ATAC-seq peaks.Enriched regions were mapped to the nearest gene in human hg19 by Homer 61 .ATACseq peak files were provided in Supplementary Data 5.

Flow cytometry
To determine the percentage of cells targeted with a sgRNA coupled with CFP, cells grown in suspension culture were collected and filtered through a 70-micron cell strainer before sorting with flow cytometry.DAPI was added to the cell suspension to exclude dead cells.Fluorescence from CFP was detected using a blue-violet laser at a wavelength of 445 nm.

Mouse studies
NOD.CgPrkdc scid Il 2rg tm1Wjl /SzJ (NSG) mice were purchased from the Animal Resource Center at St. Jude Children's Research Hospital.NSG male mice are showing aggressive behavior than female settings particularly implanted with leukemia and developed disease progression.
To avoid potential physical fighting and injury followed with infection, only female mice were included in this type of studies as a default setting.All mice were maintained in sterilized conditions with a temperature between 20 °C and 23 °C and humidity between 40% and 60%.).The mice were euthanized by CO 2 treatment and cervical dislocation following the approved animal protocol, when the mice reached a threshold of engraftment (generally above 80% human leukemia cells in the peripheral blood).Most mice show no overt symptoms at this point, which we consider a humane endpoint of the study.We also euthanized mice immediately if they had any of the following severe clinical signs: dragging hind limbs, inability to eat or drink, failure or delay to the right when placed on the back, dyspnea, bleeding from GI or respiratory tract, or any neurological signs such as circling head tilt, or seizures.

Statistical analysis
Statistical analyses were performed using GraphPad Prism 9.0.For Q-PCR experiments, p values were calculated using a two-tailed unpaired t-test from 3 independent biological replicates.For animal experiment, three or four female NSG mice were used for transplantation of control and CRISPR targeted cells.The p values were calculated using a two-tailed un-paired t-test.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Characterization of the HOXA9/FLT3 regulation phenotype and mechanism in MLL-r leukemia

Fig. 3 |
Fig.3| Cas9-mediated disruption of the h-FLT3 enhancer blunted MLL-r leukemia cell growth.ChIP-seq tracks of transcription factors and H3K27ac (GEO: GSE117864)70 were used to demonstrate the epigenetic status of the FLT3 promoter (A) and the HOXA9-bound site in the h-FLT3 enhancer (B).C Two sgRNAs (sgFLT3-DE-1 and -DE-2) targeting the h-FLT3 enhancer were used to disrupt the h-FLT3 enhancer activity.D Q-PCR detection of gene expression demonstrated notable downregulation of FLT3 and PAN3 expressions in SEM cells upon CRISPR targeting.

Fig. 4 |
Fig.4| dCas9-KRAB-mediated of the h-FLT3 enhancer blunted growth of MLL-r leukemia cells and decreased sensitivity to an FLT3 inhibitor.A A dCas9-KRAB-mediated CRISPR interfering system was applied to disrupt the h-FLT3 enhancer.B, C ATAC-seq results showed that Cas9-KRAB/sgFLT3-DE-1 only influenced the chromatin accessibility of the h-FLT3 enhancer locus in a genomewide analysis.The non-relevant RBM45 locus was one of the negative loci.D-F dCas9-KRAB/sgFLT3-DE-1-mediated disruption of the h-FLT3 enhancer led to down-regulation of FLT3 and PAN3 expressions and retarded cell growth of SEM cells.Expressions of FLT3 and PAN3 were normalized to SEM (dCas9-KRAB) cells infected with non-target sgRNA (sgNT).The percentage of cell numbers was normalized to CFP + control cells.Data are shown as mean values ± SEM of three biological replicates (center of the error bar), and results represent three independent

Fig. 5 |
Fig. 5 | CRISPR-based disruption of the h-FLT3 enhancer blunted in vivo MLL-r leukemia cell growth.A A diagram of in vivo competitive transplantation in female NSG mice to evaluate the effect of CRISPR-based disruption of the h-FLT3 enhancer on cell growth of SEM cells.The percentage of SEM cells in the peripheral blood was monitored weekly by flow cytometry.CRISPR-based (dCas9-KRAB and Cas9) disruption of h-FLT3 enhancer compromised in vivo growth of SEM cells at 2 weeks (B) and 3 weeks (C) after transplantation.Four animals in sgNT group and three animals in targeted experiment groups were used.Data are shown as mean values ± SEM of three biological replicates (center of the error bar).P values were estimated using a two-tail un-paired t-test.Source data are provided as a Source Data File.

Fig. 6 |Fig. 7 |
Fig. 6 | Functional interrogation of HOXA9-bound targets by genome editing.A Competitive proliferation assay was conducted in Cas9-expressing SEM cells targeted with sgRNAs against HOXA9-bound sites close to the genes as PEBP4, ZCCNC7, AHI1, RUNX1, DCAF11, CDK6, NDUFS8, and MAN1C1.The sgRPS19 and sgFLT3-DE-1 sgRNA served as positive controls.Disruption of the loci targeted by these selected sgRNAs led to retarded cell growth of SEM cells in a time-dependent manner.The percentage of cell numbers was normalized to CFP + control cells infected with non-target sgRNA (sgNT).B Competitive proliferation assay was conducted in Cas9-expressing MOLM13 cells targeted with sgRNAs against HOXA9bound sites close to the genes as PEBP4, ZCCHC7, AHI1, RUNX1, DCAF11, CDK6, NDUFS8 and MAN1C1.The sgRPS19 and sgFLT3-DE-1 sgRNA served as positive controls.Disruption of the loci targeted by these sgRNAs led to retarded cell growth of MOLM13 cells in a time-dependent manner.The percentage of cell numbers was normalized to CFP + control cells infected with non-target sgRNA (sgNT).C Competitive proliferation assay was conducted in dCas9-KRABexpressing SEM cells targeted with sgRNAs against HOXA9-bound sites close to the genes RUNX1, DCAF11, and CDK6.The sgFLT3-DE-1 sgRNA served as positive control.D Competitive proliferation assay was conducted in dCas9-KRABexpressing MOLM13 cells targeted with sgRNAs against HOXA9-bound sites close to the genes RUNX1, DCAF11, and CDK6.The sgFLT3-DE-1 sgRNA served as positive control.E Characterization of the chromatin conformation change upon dCas9-KRAB targeting against the HOXA9-bound site in the intron of CDK6.HiC (GEO: GSE138862), HOXA9 ChIP-seq, H3K27ac ChIP-seq and BRD4 ChIP-seq (GEO: GSE117864)70 tracks were shown to characterize the epigenetic status of the HOXA9-bound site.F Q-PCR was conducted to quantify the transcription decrease of CDK6 when CRISPRi targeted the HOXA9-bound site in the intron of CDK6 in SEM cells.G Q-PCR was conducted to quantify the transcription decrease of CDK6 when CRISPRi targeted the HOXA9-bound site in the intron of CDK6 in MOLM13 cells.H Total RNA-seq was performed using SEM cells targeted with sgCDK6 against the HOXA9-bound site in the CDK6 intron.Differential gene expression was defined by FDR < 0.01.The CDK6 expression is the top hit.Data are shown as mean values ± SEM of three biological replicates.P values were estimated using a two-tail unpaired t-test.Source data are provided as a Source Data File.