Systematically Interrogating Downstream Regulation Mechanisms of HOXA9 in MLL-r Leukemia Via Functional Noncoding CRISPR Screens

HOXA9 overexpression is observed in 50-70% of human acute myeloid leukemia (AML) and a subset of acute lymphoblastic leukemia (ALL) and positively correlates with poor patient outcomes. Leukemia subtypes with hallmark overexpression of HOXA9 include those carrying MLL gene rearrangements (MLL-r), NPM1c mutations, and other genetic alterations. Frequent genetic translation of NUP98-HOXA9 retained the DNA binding domain of HOXA9, suggesting the transcription factor function is essential for leukemia initiation and maintenance. Accumulating evidence indicates that HOXA9 dysregulation is sufficient and necessary for leukemic transformation. However, HOXA9 protein itself is a poor therapeutic target as it lacks targetable binding domains. Therefore, understanding HOXA9's functional downstream targets will provide alternative therapeutic targets. However, it remains largely unknown how HOXA9, as a homeobox transcriptional factor, binds to noncoding regulatory sequences and controls the downstream genes in MLL-r leukemia and across leukemias of other genetic subtypes.

In this study, we have successfully established a HOXA9-miniAID expressing MLL-r B-ALL cell line to execute acute HOXA9 degradation in the same isogenic cell line background upon auxin treatment. Compared with conventional loss-of-function strategies, our AID cellular model allows the investigation of immediate downstream targets controlled by HOXA9. We have therefore identified about 1,800 reproducible peaks bound by HOXA9 at a genome-wide scale. About 80% of HOXA9 binding peaks were co-bound by HOXA9's binding partner MEIS1, and motif enrichment confirmed that more than 75% of peaks contained a typical HOXA9 consensus motif. Interestingly, >80% binding peaks are located in non-promoter cis-regulator elements.

Given the essential role of HOXA9 in regulating MLL-r leukemia cell survival and differentiation, we reasoned that targeting specific HOXA9-bound peaks will affect downstream gene expression leading to a similar survival crisis seen in HOXA9 knockout cells. Based on this rationale, combinatorial CRISPR screens were performed in three biological replicates on the MLL-r SEM cell line stably expressing Cas9 and base editor (ABE8.0), followed by time-course dropout selection on days 7, 14, and 21. Integrative CRISPR screen analysis identified six reproducible noncoding hits, including a positive control located in the distal enhancer of FLT3. Functional validation by targeting these six noncoding segments with Cas9, dCas9-KRAB in additional MLL-r leukemia cell lines MOLM13, MV4,11 confirmed the survival essential of these HOXA9-bound regions. No cellular phenotype was observed in non-MLL-r leukemia cell lines, including NALM6, JURKAT, or OCI-AML2. Transplanting SEM cells targeted with sgHOXA9-in-FLT3 into NSG mice, significant growth retardation was monitored by flow analysis of peripheral blood. In addition, RNA-seq and Q-PCR analysis further identified the functional relevant downstream genes upon CRISPR editing of these HOXA9-bound noncoding segments, including FLT3, XBP1, JUN, BAHCC1, and CCDC200. Chromatin conformation characterization by Capture-C further confirmed the long-range interaction between HOXA9 bound enhancers and corresponding gene promoters in MLL-r SEM cells.

In summary, using state-of-art research tools and newly established cell models, we have conducted a systematic functional screen and mechanism study of an undocumented downstream transcriptional network of HOXA9 in MLL-r leukemia cells. The work is significant as it will advance our understanding of how HOXA9-associated transcription programs reconstruct the regulatory network specifying MLL-r dependency. Moreover, our study will promote the future development of alternative therapeutic targets in HOXA9-driven (including HOXA9 and possibly NUP98-HOXA9 subtype) leukemia in patients.

No relevant conflicts of interest to declare.