The Mll partial tandem duplication: differential, tissue-specific activity in the presence or absence of the wild-type allele

Approximately 5% to 10% of patients with acute myeloid leukemia (AML) and normal cytogenetics present with rearrangement of the mixed-lineage leukemia (MLL, also known as ALL1 or HRX) gene as the result of a partial tandem duplication (PTD) within a single MLL allele.^1,2  In AML blasts harboring the somatic MLL PTD mutation, the MLL wild-type (WT) allele is not expressed and, when reexpressed, leukemic cell death was observed.³  We previously reported on the Mll^PTD/WT knockin mice that are fully viable with modest developmental defects, have aberrant gene expression and altered hematopoiesis, but do not develop leukemia.⁴  These mice express both the Mll PTD and WT Mll alleles. Thus, to partially recapitulate what is observed in human primary AML blasts regarding the MLL-PTD and absence of MLL-WT expression, we generated mice that harbor a Mll-PTD but lack Mll-WT (Mll^PTD/−) in the germ line. We then asked how loss of function of Mll-WT in the context of Mll-PTD would affect HoxA gene expression and hematologic abnormalities previously observed in Mll^PTD/WT mice and, eventually, occurrence of leukemia.

Mll^WT/− mice⁵  were generously provided by the late Dr Stanley Korsmeyer (Dana-Farber Cancer Institute, Boston, MA). These mice were maintained on a B6C3F1 background. To obtain Mll^PTD/− mice, F1 offspring were obtained by crossing the Mll^WT/− mice with the Mll^PTD/WT mice⁴  (maintained on a pure C57Bl/6J background). This work was performed with approval of The Ohio State University institution review board and under an Institutional Animal Care and Use Committee–approved proposal.

Total RNA was extracted from E17.5 fetal liver cells (FLC) from Mll^PTD/WT, Mll^PTD/−, Mll^WT/−, and Mll^WT/WT embryos. Comparative real- time reverse transcription–polymerase chain reaction (RT-PCR) assays on whole fetal liver and c-kit⁺- and CD11b⁺-sorted populations were performed as previously described.^4,6 

H3 (Lys4) dimethylation has been shown to occur as a direct result of MLL's SET domain methyltransferase activity.^4,7  Therefore, chromatin immunoprecipitation (ChIP) assays were performed on 2 × 10⁶ FLC using the EZ ChIP Assay Kit with the anti-dimethyl Histone H3 (Lys4) antibody (Millipore, Billerica, MA) according to the manufacturer's standard protocol. DNA was quantified using PCR and nested real-time quantitative PCR with SYBR green incorporation (Applied Biosystems, Foster City, CA) using previously described methods.⁴ 

Single cell suspensions were plated at a density of 50 000 cells/dish in M3434 methylcellulose (StemCell Technologies, Vancouver, BC), and were performed according to the manufacturer's protocol (StemCell Technologies) and methods as previously described.⁴ 

To evaluate whether significant differences in pluripotent hemopoietic progenitors (colony forming unit [CFU]–GEMM), CFU-GM, or burst forming unit-erythroid (BFU-E) existed between mouse genotypes as indicated in the Figure 1 legend, paired t tests were carried out using siblings.

Results of the comparative analysis among the Mll^WT/WT, Mll^WT/−, Mll^PTD/WT and Mll^PTD/− mice are summarized in Table 1. In terms of survival, the first 3 genotypes were all viable and born at expected Mendelian ratios, however, although present at normal Mendelian ratios, 100% of the pups having the Mll^PTD/− genotype died by postpartum day 1 (P1). Interestingly, Mll^−/− mice are also nonviable, but die around embryonic day 10.5 (E10.5).⁵  These results indicate the Mll PTD itself provides some, albeit insufficient, compensation for embryonic development in the absence of both Mll WT alleles.

With regards to HoxA gene expression, we found that the Mll PTD is required for aberrant overexpression of HoxA genes in E17.5 FLC. Mll^PTD/WT and Mll^PTD/− mice showed nearly equivalent levels of HoxA over-expression in unsorted FLC, while Mll^WT/− cells expressed HoxA levels that were consistently lower but not significantly different from the expression levels found in Mll^WT/WT FLC (Figure 1A). To determine whether the overexpression of Hoxa9 was on a per cell basis rather than an increase in a Hox-expressing subpopulation within the unsorted FLC, we sorted out equivalent numbers of c-kit⁺ (Figure 1B) and CD11b⁺ (Figure 1C) cells from Mll^PTD/WT, Mll^PTD/−, Mll^WT/WT and Mll^WT/− E17.5 fetal livers. Significantly increased levels of Hoxa9 were found in the Mll^PTD/WT- and Mll^PTD/−-sorted FLC but not in Mll^WT/WT- or Mll^WT/−-sorted FLC. In fact, Mll^WT/− CD11b⁺ cells had no detectable levels of Hoxa9 transcript. These results support the notion that the overexpression of HoxA is occurring on a per cell basis within the fetal liver. Furthermore, ChIP assays demonstrate that the presence of the Mll PTD is associated with increased H3 (Lys4) methylation at the Hoxa9 promoter in the presence or absence of the Mll-WT allele (Figure 1D). This increased H3 (Lys4) methylation likely accounts for the observed HoxA gene overexpression in both Mll^PTD/WT and Mll^PTD/− FLC.

Although premature death of the Mll^PTD/− mice precluded assessment of leukemia development, we did examine fetal livers for any alterations in normal hematopoiesis. We performed CFU assays to assess fetal hematopoietic liver function in vitro using E17.5 FLC obtained from each of the 4 genotypes. Significant increases in the CFU-GM, BFU-E, and the more immature CFU-GEMM were seen in cells obtained from the Mll^PTD/WT mice compared with Mll^WT/WT mice (P < .01, P < .01, and P < .05, respectively), suggesting that the Mll PTD may cooperate with the Mll WT allele at an early stage of hematopoiesis such as the common myeloid progenitor cell. In contrast, FLC obtained from Mll^PTD/− mice had increases in the CFU-GM and BFU-E populations compared with Mll^WT/WT mice (P < .05 and P < .05, respectively), but not in the CFU-GEMM population (Figure 1E). Finally, FLC obtained from Mll^PTD/− mice had a significantly lower number of BFU-E progenitors compared with Mll^PTD/WT FLC. Together these results are consistent with the notion that the Mll PTD itself is required and sufficient for abnormal expansion at some stages of progenitor cell differentiation but may not be at other stages. These results also support 2 recent reports that showed the role of Mll WT in hematopoietic stem cells is distinct from its role in hematopoietic progenitor cells.^8,9 

Similar phenotypic abnormalities observed in Mll^PTD/− and Mll^PTD/WT genotypes support the notion that the Mll PTD by itself is capable of dysregulating downstream targets and can therefore behave as a dominant gain-of-function mutation in the absence of the Mll WT. Although these studies suggest a direct role for the Mll PTD, it is important to note that a more stable (lacZ fused) Mll protein containing several important N-terminal Mll functional motifs exists in the knockout model. Because our heterozygous animals exhibit a more severe phenotype than other (non-lacZ fused) Mll heterozygous knockout mice,^8,9  we cannot exclude the possibility that the knockout allele acts in an interfering manner.

Mll has now been shown to have very different functions in different subpopulations in the hematopoietic compartment.¹⁰  Our results suggest that in some cases Mll function may have been lost and cannot be replaced by the Mll PTD allele, as in the case of Mll^PTD/− early lethality at P1. However, in some cases such as HoxA gene overexpression and CFU-GM expansion, the Mll PTD appears to behave more as a dominant gain-of-function mutation because the quantifications performed for these experiments were similar between the Mll^PTD/WT and Mll^PTD/− genotypes, ie, in the presence and absence of Mll WT, respectively. In contrast, differences in the number of BFU-E and CFU-GEMM progenitors seen between the Mll^PTD/WT and Mll^PTD/− genotypes reveal the lack of dominant activity by the Mll PTD allele. In this tissue-specific context, the Mll PTD may require interaction with the Mll WT in order to achieve the maximum manifestation of the abnormality.

This work was funded in part by the Lady Tata Memorial Trust (London, United Kingdom) Award (to A.M.D.) and National Cancer Institute (Bethesda, MD) grant funding (R01 CA89341 to M.A.C.).

National Institutes of Health

Contribution: A.M.D. and M.A.C. designed experiments, analyzed and interpreted data, and cowrote the manuscript; A.M.D., S.L., A.C., B.R.P., D.N., M.G., W.Y., and D.C. performed experiments; and S.P.W. and G.M. provided intellectual expertise and careful review and editing of the manuscript. All authors agreed on the final text version.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Michael A. Caligiuri, MD, 458A Starling-Loving Hall, West 10th Avenue, Columbus, OH, 43210; e-mail: Michael.caligiuri@osumc.edu.