The hepatocyte nuclear factor 4 (hnf4α) is a member of the steroid hormone receptor family that plays an important role in regulation of hepatic gene expression (Li et al., 2000; Sladek et al., 1990). It has been shown that a complete loss of hnf4α results in embryonic lethality caused by defects in visceral endoderm function (Chen et al., 1994; Duncan et al., 1997). This early embryonic lethality can be overcome by complementing the hnf4α null embryos with an hnf4α+/+ visceral endoderm by tetraploid aggregation (Duncan et al., 1997; Li et al., 2000). Unfortunately, we were unable to generate latestage embryos using this procedure that were suitable for the analysis of hnf4α’s role in liver morphogenesis. Here we describe the generation of a conditionally null allele of hnf4α that will be suitable for this purpose. Figure 1a shows that we employed a two-step Cre/loxP recombination strategy to delete exon 2 of the hnf4α gene that encodes a portion of the hnf4α DNA binding domain (Hadzopoulou et al., 1997). A targeting vector containing three loxP elements was constructed. The first loxP site was introduced into a SmaI restriction endonuclease cut site in intron 1 and another two loxP elements that flanked a thymidine kinase/neomycin phosphotransferase (tk/neo) expression cassette were introduced into an Asp718 site in intron 2 (Fig. 1a). The targeting vector also contained a diphtheria toxin (DT) gene that selected against random integration events. Following electroporation of the targeting construct into R1 ES cells, G418-resistant ES cell clones were tested for homologous recombination at the hnf4α locus by genomic Southern blot analysis (Fig. 1b)(Nagy et al., 1993). After digestion with EcoRV, a 700-bp probe outside the 5! arm of the targeting vector, probe A, detects an endogenous 10.2-kb fragment while a correctly targeted allele should contain an additional 6.8-kb fragment due to the introduction of a novel EcoRV site from tk/neo. Similarly, a 600-bp probe from outside the 3! end of the targeting vector, probe B, should detect the same wild-type 10.2-kb and a targeted 6.4-kb fragment. Twenty percent of G418-resistant clones collected were found to contain a correctly targeted allele, an example of which is shown in Figure 1b. Next, the tk/neo expression cassette was removed by transient expression of Cre recombinase, and successfully recombined clones were identified by selection for growth in gancyclovir, as described previously (Sund et al., 2000). As shown in Figure 1a, two possible outcomes could arise. Recombination between loxP elements b and c would generate an allele containing hnf4α exon 2 flanked by loxP elements. In contrast, recombination between sites a and c would replace exon 2 with a single loxP site. These two alleles are referred to as hnf4αloxP/+ and hnf4αloxPΔ/+ to indicate these events. As shown in Figures 1a and b, these recombination events can be distinguished by Southern blot analysis of genomic DNA that has been digested with EcoRV and probed with probe A. We have shown previously that loss of hnf4α function can be determined in ES cell embryoid bodies by measuring changes in hnf4α-mediated gene expression (Duncan et al., 1997). We, therefore, generated ES cell lines in which both copies of hnf4α exon2 had been deleted through recombination between flanking loxP elements by repeating the targeting procedure discussed above on the remaining wild-type allele. Figure 1b shows Southern blot analysis confirming the genotype of hnf4αloxPΔ/loxPΔ ES cell lines in which both alleles of hnf4α exon 2 are replaced by single loxP elements. This is the genotype expected of hepatocytes that have …