The AF4-MLL fusion transiently augments multilineage hematopoietic engraftment but is not sufficient to initiate leukemia in cord blood CD34+ cells

The translocation t(4;11)(q21;q23) is the hallmark genetic abnormality associated with infant pro-B acute lymphoblastic leukemia (B-ALL) and has the highest frequency of rearrangement in Mixed-lineage leukemia (MLL) leukemias. Unlike other MLL translocations, MLL-AF4-induced proB-ALL is exceptionally difficult to model in mice/humans. Previous work has investigated the relevance of the reciprocal translocation fusion protein AF4-MLL for t(4;11) leukemia, finding that AF4-MLL is capable of inducing proB-ALL without requirement for MLL-AF4 when expressed in murine hematopoietic stem/progenitor cells (HSPCs). Therefore, AF4-MLL might represent a key genetic lesion contributing to t(4;11)-driven leukemogenesis. Here, we aimed to establish a humanized mouse model by using AF4-MLL to analyze its transformation potential in human cord blood-derived CD34+ HSPCs. We show that AF4-MLL-expressing human CD34+ HSPCs provide enhanced long-term hematopoietic reconstitution in primary immunodeficient recipients but are not endowed with subsequent self-renewal ability upon serial transplantation. Importantly, expression of AF4-MLL in primary neonatal CD34+ HSPCs failed to render any phenotypic or hematological sign of disease, and was therefore not sufficient to initiate leukemia within a 36-week follow-up. Species-specific (epi)-genetic intrinsic determinants may underlie the different outcome observed when AF4-MLL is expressed in murine


INTRODUCTION
The translocation t(4;11)(q21;q23) encodes the chimeric proteins mixed-lineage leukemia (MLL)-AF4 and AF4-MLL and is the hallmark genetic abnormality associated with infant pro-B acute lymphoblastic leukemia (B-ALL), which has a dismal prognosis [1,2]. Our understanding of t(4;11)-mediated transformation is limited, and unlike other MLL fusions, MLL-AF4-induced leukemogenesis has been difficult to model. Current murine and humanized disease models do not faithfully recapitulate the pathogenesis/phenotype [3][4][5]. It has been claimed that the absence of a suitable t(4;11) disease model is the result of targeting a cell in a wrong developmental stage, or that the impact of secondary hits has not been properly addressed. Very recently, Lin et al [6] fused human MLL to murine Af4, resulting in a human-mouse chimeric fusion gene that produced high-titer retrovirus facilitating efficient transduction
Most chromosomal translocations that have been studied in cancer require only one fusion product for transformation, and in many human cancers the reciprocal fusion is not consistently expressed [7,8]. The relevance of the reciprocal product AF4-MLL in t(4;11) leukemia has been investigated in previous studies. AF4-MLL fusion protein was capable of inducing B-ALL in mice without requirement of MLL-AF4, indicating that it might represent a key genetic lesion contributing to t(4;11)-driven leukemogenesis [5]. Here, we have investigated in a human stem cell context whether ectopic expression of AF4-MLL contributes to transform cord blood (CB)-derived CD34 + HSPCs. Our data indicate that AF4-MLL transiently enhances long-term hematopoietic reconstitution in immunodeficient mice, but is not sufficient to initiate leukemia in primary neonatal CD34 + HSPCs.
IBMT provides the opportunity to assess migration of transplanted CD34 + cells in vivo. Accordingly, the migration ability of transplanted EV-, MLL-AF4-and AF4-MLL-CD34 + cells was assessed by analyzing the level of chimerism in injected tibiae (IT), non-injected tibia (contralateral, CL), spleen and peripheral blood (PB). CD34 + HSPCs were capable of migrating to and colonizing other hematopoietic sites in all the animals ( Figure 1H). However, the levels of chimerism in CL and PB were 2-fold higher in NSG mice transplanted with AF4-MLL-transduced CD34+ cells. Thus, enforced expression of AF4-MLL transiently augments multilineage hematopoietic engraftment and facilitates homing of CD34 + HSPCs.
To further confirm that AF4-MLL did not contribute to leukemia initiation, all the animals lacking disease signals after 36 weeks were sacrificed, and neither splenomegaly ( Figure 2E, left panel) nor hepatomegaly www.impactjournals.com/oncotarget Oncotarget 81939 www.impactjournals.com/oncotarget was particularly associated with the AF4-MLL genotype. Moreover, platelet counts, hemoglobin levels and leukocyte counts (WBC) were normal/similar between conditions ( Figure 2E, middle-right panels). Importantly, secondary recipients of primary mice engrafted with AF4-MLL-expressing CD34 + cells had no sign of splenomegaly, anemia or leukocytosis according to the limited engraftment observed upon serial transplantation ( Figure 2E). Collectively, the results show that enforced expression of AF4-MLL (or MLL-AF4) was not sufficient We acknowledge some limitations in our study. First, the AF4-MLL fusion is 8 kb in length, so the expressing lentivector may produce low viral titers. Our AF4-MLL-expressing lentivirus titer was consistently ~106 infective particles, which suffices to infect target cells at a reasonable MOI, rendering stable transcript expression (sequenced) in vitro ( Figure 1C, 1D) and in vivo ( Figure 1G) even 36 weeks after transplantation. The same transcript was successfully expressed in a "safe harbor" [12] using CRISPR/Cas9 genome-editing, suggesting that the inability of AF4-MLL to initiate leukemia may not be attributed to the absence of transcript expression. Second, it is plausible that co-expression of MLL-AF4 and AF4-MLL is required in the same target cell for leukemia initiation. Lentiviral expression of both cassettes in the same CD34 + HSPC is challenging, but a CRISPR/Cas9-mediated recreation of the t(4;11) by producing both allele-specific MLL fusions would undoubtedly open new avenues in future research [13]. Alternatively, the generation of an AF4-MLL knock-in mouse, where AF4-MLL is conditionally expressed in the context of the AF4 locus, could be a valuable model to test whether AF4-MLL is needed for leukemia onset. Arguments in favor for the need for AF4-MLL are: (i) the extreme rapid onset of disease after birth; the existing mouse data with AF4-MLL alone [5] and its proposed function [14]. Arguments against the need for AF4-MLL to be co-expressed with MLL-AF4 for B-ALL initiation are: (i) at least one third of t(4;11) + patients display complex genetic rearrangements and do not express particularly AF4-MLL [15,16]; (ii) ablation of AF4-MLL in leukemic cell lines seems dispensable for cell growth in short term siRNA experiments [14]. Therefore, AF4-MLL may well cooperate with MLL-AF4 in leukemia phenotype, aggressiveness or therapy-resistance rather than in leukemia onset/initiation.

MATERIALS AND METHODS
The study was IRB-approved by the Clinic Hospital of Barcelona and CB units from healthy newborns were accessed from the Barcelona Cord Blood Bank upon signed informed consent.

Author's contributions
C.P conceived the study, designed and performed experiments, analyzed and plotted the data and wrote the manuscript. R.M, A.K. and H.B. conceived the study, performed experiments and analyzed data. C.B and P.M conceived the study, designed experiments, analyzed data and wrote the manuscript.

CONFLICTS OF INTERESTS
The authors have nothing to disclose.

Editorial note
This paper has been accepted based in part on peerreview conducted by another journal and the authors'