A human genome editing–based MLL::AF4 ALL model recapitulates key cellular and molecular leukemogenic features

TO THE EDITOR:

Chromosomal translocations involving the mixed-lineage leukemia (MLL) gene drive leukemia development; key determinants of disease outcome are patient age, MLL fusion partner, and cell-of-origin.¹ The MLL::AF4 (also known as KMT2A::AFF1) fusion protein (MA4) is present in 80% of infants with B-cell acute lymphoblastic leukemia (iB-ALLs) and is associated with prenatal origin and poor prognosis.^1,2 In fact, the MA4 fusion occurs in early hematopoietic stem/progenitor cells (HSPCs) and is a causal oncogenic driver in iB-ALL.^3,4 The prevalence of the MA4 fusion is higher in infant than in noninfant patients with B-ALL and is associated to higher frequency of refractoriness, relapse, and poor prognosis transcriptomic signatures.⁵ Molecularly, the distribution of genomic breakpoints within the MLL breakpoint cluster region (BCR) varies significantly, with a notable telomeric hot spot in infants and a preference for centromeric breakpoints in noninfants.^6,7 Overall, these findings support the notion that age-related mechanisms and inherent biological differences contribute to MA4⁺ B-ALL. Studying these mechanisms is, however, challenging because of the dearth of human cellular models.^8-12 CRISPR-Cas9 gene editing was recently used to recreate the t(4;11)/MLL::AF4 translocation in human HSPCs, mimicking the characteristics of the disease,^13,14 but the impact of the cell-of-origin and its functional consequences were not compared nor was the potential to induce pro–B-ALL in vivo determined. Here, we compared the molecular impact and leukemogenic potential of both centromeric (intron 10 [MLLⁱ¹⁰]) and telomeric (intron 12 [MLLⁱ¹²]) MLL breakpoints in human CD34⁺ HSPCs from samples with different developmental stages, such as second trimester prenatal fetal liver (FL) and postnatal cord blood (CB).

We designed and tested several single-guide RNAs to target intron 10 (centromeric) or intron 12 (telomeric) of MLL, along with AF4 intron 3 (Figure 1A; supplemental Table 1, available on the Blood website). Nucleofected FL and CB cells were cultured with cytokines to promote MLL-edited cell expansion¹³ (Figure 1B). In contrast to mock cells, edited cells showed exponential growth starting on day 20 to 25 after nucleofection, resulting in billions of MA4-immortalized cells irrespective of their origin (Figure 1C; supplemental Figure 1A-B). Most cells in both cultures harbored MLL rearrangements (Figure 1D; supplemental Figure 1C), although the emergence of MLLⁱ¹²-rearranged cultures was consistently slower than that of MLLⁱ¹⁰-rearranged cultures (Figure 1E; supplemental Figure 1D). We confirmed the presence of the chromosomal translocation, the fusion gene RNA transcript (supplemental Figure 1D-E), and the chimeric protein (supplemental Figure 1F). Furthermore, we found that targeting either intronic regions equally immortalized in vitro both prenatal and neonatal CD34⁺ cells.

We confirmed similar MA4 expression levels between CRISPR-edited cells and patient-derived cells (supplemental Figure 2A). Transcriptome analysis revealed distinct clustering patterns of MLL-edited HSPCs according to the location of the MLL breakpoint (centromeric-Mⁱ¹⁰A4 vs telomeric-Mⁱ¹²A4) irrespective of the cellular ontogeny (FL or CB; supplemental Figure 2B). Specifically, the expression of HOX cluster genes, known to be involved in MA4-mediated pathogenesis,^15,16 exhibited variation dependent on the MLL breakpoint location rather than the tissue of origin (Figure 2A). Notably, crucial HOX/MEIS genes, such as HOXA9, HOXA10, and MEIS1 showed significant upregulation in FL and CB Mⁱ¹⁰A4-edited cells but not in Mⁱ¹²A4-edited cells (supplemental Figure 2C). We performed chromatin immunoprecipitation sequencing analysis to understand how HOXA9, HOXA10, and MEIS1 expression is regulated in gene edited–derived cells, which revealed specific MA4 binding at the promoter and spreading into the gene bodies of HOXA9, HOXA10, and MEIS1 in Mⁱ¹⁰A4-edited FL and CB CD34⁺ cells but not in Mⁱ¹²A4-edited cells (Figure 2B). Notably, no significant differences were observed in other epigenetic marks across all edited cells. Collectively, these findings confirm the transcriptional regulatory capacity of Mⁱ¹⁰A4, but not Mⁱ¹²A4, in regulating the expression of HOXA9, HOXA10, and MEIS1 through direct and specific binding to their regulatory regions. This recapitulates the HOXA^low transcriptomic signature reported in infant patients.^3,15,17 

The MA4⁺-edited cells exhibited in vitro expansion with a CD45⁺CD33⁺CD34^–CD19^– myeloid phenotype (Figure 1). Therefore, we next examined the myeloid clonogenic potential of MLL-edited FL and CB CD34⁺ HSPCs. Both mock-targeted and edited cells showed similar primary colony-forming unit potential and sustained expression of MA4 (supplemental Figure 2D-E); however, only MLLⁱ¹⁰-edited cells displayed clonogenic capacity upon serial replating, regardless of their cellular ontogeny (supplemental Figure 2D-E). Notably, these results align with clinical data,¹⁸ indicating a higher prevalence of acute myeloid leukemia in patients with a centromeric (42%) rather than a telomeric (30%) MLL breakpoint (supplemental Figure 2F).

We next investigated the impact of cellular ontogeny and the location of the MLL breakpoint on leukemia initiation in vivo by transplanting t(4;11)-CRISPR/Cas9–edited cells into NSG mice (supplemental Figure 2G). No significant bias in the overall engraftment potential was observed across the samples; however, marked differences were noted in the rates of engraftment driven by MA4-rearranged CD34⁺ cells (from 11% to 70%; Figure 2C). The frequency of Mⁱ¹⁰A4⁺ and Mⁱ¹²A4⁺ engraftment was higher when the translocation was generated in CB-CD34⁺ cells (16 of 28, 57%) than in FL-CD34⁺ cells (5 of 29; 18%). Furthermore, transplantation of these cells into secondary recipients resulted in human leukemic MA4⁺, NG2⁺, lymphoid-based (pro–B and CD19⁺CD10^-–), and multiorgan (bone marrow, peripheral blood, spleen, and liver) grafts, except for Mⁱ¹⁰A4 FL–derived cells, which failed to engraft after serial transplantation and consistently exhibited a nonleukemic phenotype (MA⁻, NG2⁻, and multilineage; Figure 2D). Of note, clinicobiological data from patients with MLL-rearranged (MLLr) B-ALL studied within the MLL recombinome consortium revealed a lower incidence (30% vs 49%) of centromeric breakpoints (i9-i10) in the MLL BCR among infant patients (<6 months) than among pediatric patients (>6 months)¹⁸ (Figure 2E). Collectively, our results show that genome-edited t(4;11)/MA4⁺ in human FL and CB CD34⁺ cells are sufficient to induce a transplantable MA4⁺ pro-B-ALL that recapitulates key phenotypic features of MLLr iB-ALL, but the leukemogenic potential of MA4⁺ CD34⁺ cells appear to be influenced by the cellular ontogeny and the MLL breakpoint location within the MLL BCR.

We finally characterized the pro–B-ALL generated from Mⁱ¹⁰A4⁺, Mⁱ¹²A4⁺ CB–derived cells, and Mⁱ¹²A4⁺ FL–derived cells in secondary mice. Recurrent mutations in K- and N-RAS were found, phenocopying those found in patients with primary iB-ALL (n = 4 of 9, 44%, Figure 2F),^3,4 and confirming the leukemogenic potential of t(4;11)/MA4⁺ (Figure 2F). Differential methylation analysis revealed thousands of hyper and hypomethylated sites (supplemental Figure 3A), and enrichment analyses of transcription factor–binding sites and chromatin states indicated similarities in DNA methylation and chromatin patterns between leukemias from edited cells and primary MLLr iB-ALL (supplemental Figure 3B-C). Transcriptome and methylome analyses revealed that the pro–B-ALL from edited cells clustered with iB-ALL primary MLLr from patients (supplemental Figure 3D-E). Overall, our study provides evidence that pro–B-ALL generated in vivo from MA4-edited CB- or FL-derived CD34⁺ cells recapitulate key molecular features of iB-ALL MLLr.

Our findings demonstrate that the origin of CD34⁺ cells does not affect the efficiency of t(4;11)/MA generation. Our human genome editing t(4;11)/MA4⁺ B-ALL model recapitulates the HOXA^low and HOXA^hi transcriptomic signatures reported in patients with t(4;11) B-ALL with telomeric-biased (Mⁱ¹²A4⁺) and centromeric-biased (Mⁱ¹⁰A4⁺) MLL breakpoint hotspots, respectively, regardless of the source (prenatal or neonatal) of the HSPCs. We confirmed that Mⁱ¹⁰A4, but not Mⁱ¹²A4, directly and specifically regulates the expression of HOXA9, HOXA10, and MEIS1 by binding to their regulatory regions. Although our ex vivo model failed to replicate the upregulation of other previously reported genes,¹⁹ it establishes a link between HOXA cluster gene expression and intronic distribution of MLL breakpoints in patients with t(4;11)/MA4⁺.

Our model also recapitulates the presence of recurrent mutations exclusively in K- and N-RAS, phenocopying those of patients with primary iB-ALL,^3,4 and confirming that t(4;11)/MA4⁺ is a leukemogenic driver. It is worth noting that targeting the MLL BCR of the centromeric region enhances in vitro myeloid clonogenic replating in MLL-edited CD34+ HSPCs regardless of their cellular origin. These findings align with clinical and biological data,¹⁸ and reveal an increased prevalence of acute myeloid leukemia among patients with MLL-r with a centromeric MLL breakpoint. However, all leukemias initiated in mice that received transplantation with either Mⁱ¹⁰A4-edited or Mⁱ¹²A4-edited CD34⁺ cells were pro–B-ALL, suggesting that the genomic localization of MLL breakpoint influences the resulting leukemia phenotype in cooperation with in vivo–occurring interactions in specific bone marrow niches,²⁰ secondary oncogenic hits,²¹ and the nature of the cell-of-origin.¹ There remain some differences between human leukemias and engineered cells that warrant further exploration.

Human samples were accessed and processed following the institutional guidelines approved by our local institutional review board (HCB/2013/8648).

The authors thank the Medical Research Council/Wellcome Trust Human Developmental Biology Resource for providing institutional review board–approved fetal material. The authors thank Centres de Recerca de Catalunya/Generalitat de Catalunya and Fundació Josep Carreras-Obra Social la Caixa for core support.

Financial support for this work was obtained from the Spanish Ministry of Economy and Competitiveness (PID2022-142966OB-100ID2019), Heroes hasta la médula initiative and Instituto de Salud Carlos III (ISCIII)–Redes de Investigación Cooperativa Orientadas a Resultados en Salud within the Next Generation EU program (plan de recuperación, transformación y resiliencia) (P.M.) and the Health Institute Carlos III (ISCIII/Federacion Española de Enfermedades Raras) PI20/00822 (C.B.), PI20/01837 (S.R.-P.) and PI21/01641 (R.T.-R.). Asociación Española Contra el Cancer (PRYGN211192BUEN) (C.B.), AECC-LABAE20049RODR (S.R.-P.), and the Fundación Uno entre Cienmil (C.B.). R.T.-R. and O.M. were supported by investigator fellowships from the Spanish Association of Cancer Research (INVES211226MOLI). R.M. is supported by grants from the Deutsche Forschungsgemeinschaft (Ma 1876/12-1) and Wilhelm Sander foundation (2018.070.2).

Contribution: C.B., R.T.-R., T.V.-H., O.M., P.P., A.M., V.R., M.V., S.C., N.V.-G., S.R.-P., J.C.S., O.Q.-B., A.L.S., and J.R.T. performed experiments and interpreted data; O.W., M.F.F., A.R., C.M., R.M., T.A.M., and P.M. supervised research and contributed key knowledge, techniques, and reagents; C.B., R.T.-R., and P.M. conceived the study and funded the research; and all authors have read and agreed to publish the manuscript.

Conflict-of-interest disclosure: P.M. is the founder of the spin-off OneChain Immunotherapeutics, which has no connection with the present research. T.A.M. is a paid consultant for and shareholder in Dark Blue Therapeutics Ltd, a company that has no connection to the results in this paper. The remaining authors declare no competing financial interests.

Correspondence: Clara Bueno, Department of Hematology, School of Medicine Barcelona University, Carrer Casanova 143, 08036 Barcelona, Spain; e-mail: cbueno@carrerasresearch.org; and Pablo Menéndez, Department of Hematology, School of Medicine Barcelona University, Carrer Casanova 143, 08036 Barcelona, Spain; e-mail: pmenendez@carrerasresearch.org.