The role of CBFβ in AML1-ETO's activity

To the editor:

AML1-ETO is the chimeric protein generated as a result of the t(8;21) in acute myeloid leukemia. Understanding which of the proteins that interact with AML1-ETO are essential for its activity is vital for developing targeted small-molecule AML1-ETO inhibitors. Two papers recently assessed the importance of CBFβ, the non–DNA-binding subunit of the core binding factors, for AML1-ETO's activity.^1,2  In both studies, the authors introduced amino acid substitutions at the CBFβ-binding interface of the Runt domain of AML1-ETO, and assessed the effects of these mutations on AML1-ETO's ability to confer serial replating ability to primary mouse bone marrow cells. Our group (Roudaia et al)¹  and that of Kwok et al²  reached opposite conclusions. Roudaia et al¹  reported that CBFβ is essential for AML1-ETO's activity, whereas Kwok et al²  claimed it is dispensable.

Roudaia et al¹  compared the effect of a single T161A mutation, which decreases the affinity of CBFβ for the Runt domain of AML1-ETO by 63-fold, with a combined Y113A/T161A mutation that decreases binding by 430 fold, on AML1-ETO's activity (Figure 1A). The side chains of both T161 and Y113 make direct contacts to CBFβ.^3-5  The extent to which binding was impaired was determined by isothermal titration calorimetry using purified proteins.¹  We showed that the T161A mutation alone only slightly impaired AML1-ETO's activity in serial replating assays, whereas the combined Y113A/T161A mutations severely compromised AML1-ETO's serial replating activity, as well as its ability to cooperate with TEL-PDGFβR to cause acute myeloid leukemia in mice.¹ 

Kwok et al²  introduced M106V and A107T mutations into AML1-ETO. The M106 side chain directly contacts CBFβ, although we previously showed that a very similar M106A mutation seriously perturbed the structure of the Runt domain.³  A107 is buried in the hydrophobic core of the Runt domain,^3-5  making the rationale for this mutation unclear. Kwok et al reported that neither mutation impaired AML1-ETO's serial replating activity, and concluded that the interaction of AML1-ETO with CBFβ was not essential.

We purified AML1-ETO Runt domains containing the M106V and A107T mutations used by Kwok et al²  and measured their affinity for CBFβ by isothermal titration calorimetry (Figure 1B). We found that the M106V and A107T mutations impaired binding to CBFβ by 20- and 43-fold, respectively. Neither mutation impaired CBFβ binding as much as the T161A mutation, which by itself had only very modest effects on AML1-ETO's activity. Therefore, the contradictory conclusions reached by the 2 groups result from differences in the extent to which CBFβ binding was disrupted by the amino acid substitutions used in each study. Twenty- to 60-fold decreases in CBFβ-binding affinity in both studies were insufficient to cripple AML1-ETO, whereas a 400-fold decrease severely impaired its activity. This is likely a result of the relatively high concentration of CBFβ in cells,⁶  necessitating a more robust disruption of the interaction to shift the ratio of free to bound protein. We assert that CBFβ is, indeed, essential for AML1-ETO's activity.