Homologous recombination in embryonal stem cells has been used to produce a fusion oncogene, thereby mimicking chromosomal translocations that frequently result in formation of tumor-specific fusion oncogenes in human malignancies. AF9 sequences were fused into the mouse Mll gene so that expression of the Mll-AF9 fusion gene occurred from endogenous Mll transcription control elements, as in t(9;11) found in human leukemias. Chimeric mice carrying the fusion gene developed tumors, which were restricted to acute myeloid leukemias despite the widespread activity of the Mll promoter. Onset of perceptible disease was preceded by expansion of ES cell derivatives in peripheral blood. This novel use of homologous recombination formally proves that chromosomal translocations contribute to malignancy and provides a general strategy to create fusion oncogenes for studying their role in tumorigenesis.

SummaryMaternal metabolism provides essential nutrients to enable embryonic development. However, both mother and embryo produce reactive metabolites that can damage DNA. Here we discover how the embryo is protected from these genotoxins. Pregnant mice lacking Aldh2, a key enzyme that detoxifies reactive aldehydes, cannot support the development of embryos lacking the Fanconi anemia DNA repair pathway gene Fanca. Remarkably, transferring Aldh2−/−Fanca−/− embryos into wild-type mothers suppresses developmental defects and rescues embryonic lethality. These rescued neonates have severely depleted hematopoietic stem and progenitor cells, indicating that despite intact maternal aldehyde catabolism, fetal Aldh2 is essential for hematopoiesis. Hence, maternal and fetal aldehyde detoxification protects the developing embryo from DNA damage. Failure of this genome preservation mechanism might explain why birth defects and bone marrow failure occur in Fanconi anemia, and may have implications for fetal well-being in the many women in Southeast Asia that are genetically deficient in ALDH2.

One subunit of both type I and type III restriction and modification enzymes contains motifs characteristic of DEAD box proteins, which implies that these enzymes may be DNA helicases. This subunit is essential for restriction, but not modification. The current model for restriction by both types of enzyme postulates that DNA cutting is stimulated when two enzyme complexes bound to neighbouring target sequences meet as the consequence of ATP‐dependent DNA translocation. For type I enzymes, this model is supported by in vitro experiments, but the predicted co‐operative interactions between targets have not been detected by assays that monitor restriction in vivo. The experiments reported here clearly establish the required synergistic effect but, in contrast to earlier experiments, they use Escherichia coli K‐12 strains deficient in the restriction alleviation function associated with the Rac prophage. In bacteria with elevated levels of EcoKI the co‐operative interactions are obscured, consistent with co‐operation between free enzyme and that bound at target sites. We have made changes in three of the motifs characteristic of DEAD box proteins, including motif III, which in RecG is implicated in the migration of Holliday junctions. Conservative changes in each of the three motifs impair restriction.

Bacteriophage lambda encodes an antirestriction function, RaI, which is able to modulate the activity of the Escherichia coli K-12 restriction and modification system, EcoKI. Here we report the characterization of an analogous function, Lar, expressed by E. coli sbcA mutants and the hybrid phage lambda reverse. E. coli sbcA mutants and lambda reverse both express genes of the Rac prophage, and we have located the lar gene immediately downstream of recT in this element. The lar gene has been cloned in an expression plasmid, and a combination of site-directed mutagenesis and labelling of plasmid-encoded proteins has enabled us to identify a number of translational products of lar, the smallest of which is sufficient for restriction alleviation. Lar, like RaI, is able both to alleviate restriction and to enhance modification by EcoKI. Lar, therefore, is functionally similar to RaI and the nucleotide sequences of their genes share 47% identity, indicating a common origin. A comparison of the predicted amino acid sequences of Lar and RaI shows only a 25% identity, but a few short regions do align and may indicate residues important for structure and/or function.