How FLT3-ITDs and NPM1 Insertions Maneuver Into AML Using the Lymphoid Enzyme TdT to Initiate Replication Slippage

Recurrent genetic abnormalities in acute myeloid leukemia (AML) confer distinct clinicopathologic characteristics and are the foundation of modern AML subclassification systems. Among the most common recurrent genetic driver mutations in AML are activating internal tandem duplications (ITD) within the juxtamembrane domain of the type 3 receptor tyrosine kinase FLT3 (FLT3-ITD) and exon 12 frameshift mutations in nucleophosmin 1 (NPM1).¹  FLT3-ITDs are present in 25 percent of all AML cases and are also found in a subset of acute lymphocytic  leukemia (ALL), while NPM1 mutations are more  restricted to AML and occur in 35 percent of cases. NPM1-mutated AML is generally associated with a favorable prognosis, while the presence of a FLT3-ITD generally confers a poor prognosis, especially at a high (>0.5) FLT3-ITD allele fraction.

Although the molecular events underlying the formation of NPM1 mutations and FLT3-ITDs are poorly understood, DNA polymerase slippage near repetitive (microhomology) sequences has long been speculated as a culprit for FLT3-ITDs.²  A study published recently by Dr. Julian Borrow and colleagues analyzed 300 FLT3-ITDs from 271 patients for sequence-level evidence of polymerase slippage. The researchers hypothesized that if this mechanism were responsible for ITD formation, the sequences immediately flanking the 3′ and 5′ ends of the duplication would include short (1-6 base pairs), homologous sequences (microhomology), which is indeed seen in a minority of cases (Figure 1A compared to 1B).³  Unexpectedly, microhomology was not observed in a majority (62%) of ITDs. In fact, 38 percent of all ITDs contained an additional sequence of unknown origin (nontemplated “filler”) with the remaining 24 percent containing neither overt microhomology nor filler.

To explain the frequent presence of filler in FLT3-ITDs, the authors use a series of mathematical deductions and inferences to suggest an unusual mechanism: activity in the myeloid cells of terminal deoxynucleotidyl transferase (TdT), the DNA polymerase responsible for untemplated additions of nucleotides to V, D, and J exons during immunoglobulin and T-cell receptor rearrangement. Although canonically thought of as restricted in expression to lymphoid cells, TdT is also known to be expressed in a subset of AML at diagnosis and in myeloid progenitors. Supporting the involvement of TdT in FLT3-ITD formation, filler sequences were found to be enriched for G/C content at a level consistent with TdT’s known bias with filler sequences of a length consistent with TdT activity. Dipurines and dipyrimidines were observed at a higher than expected frequency in filler sequences, as would be expected during nontemplated polymerization by TdT. Finally, filler sequence correlated with frequency of TdT expression across AML subtypes and was significantly overrepresented in FLT3-ITDs derived from ALL compared to AML.

The authors went on to show how TdT activity could also provide a basis for generation of ITDs in the absence of germline microhomology. In short, TdT-added nucleotide runs that, by chance, have homology with nearby 5′ sequence (occult microhomology, Figure, part C) could allow for polymerase repositioning and subsequent duplication. These runs would not be apparent as filler sequence, but they should display the level of G/C bias associated with TdT activity. In a companion study, Dr. Borrow and colleagues extended their sequence analysis approach to 2,430 previously published NPM1 mutations in AML (114 unique mutations), showing that of 114 unique mutations, more than 90 percent of NPM1 mutations, contained filler sequences that were consistent with TdT activity in terms of G/C and dinucleotide bias as well as length distribution (Figure, part D).