Design, Construction, and Foxa2 Transcriptional Activity of the Foxa2ITA Allele
Transcriptional activity of a tetracycline responsive Cre-recombinase transgene (hereafter called “TRECre”; Perl et al.,2002) is dependent on binding of the tetracycline response element (TRE) by a complex containing the rtTA (Urlinger et al.,2000) and a tetracycline derivative (Gossen et al.,1995). We took advantage of the temporally inducible nature of this system and combined it with the tissue-specific properties of the Foxa2 locus to dually regulate Cre-recombinase activity. This system will be useful for controlling the activity of loxP-flanked (floxed) alleles during multiple stages of mouse embryogenesis.
With the goal of obtaining rtTA expression in the Foxa2 temporospatial domain without disrupting endogenous Foxa2 expression, we inserted an IRES and the rtTA coding sequence into the 3′-UTR of Foxa2, 5′ to the polyadenylation sequence (Fig. 1; the Foxa2ITA allele). An important part of this strategy was flanking the antibiotic selection gene (neor) with frt sites and its removal with Flp-recombinase after generating founder animals; this strategy avoids potential hypomorphic effects of neor on the function of Foxa2 or other loci (Olson et al.,1996).
Figure 1. Schematic of the Foxa2ITA;TRECre system. The Foxa2 targeted allele has an internal ribosomal entry site (IRES) followed by the reverse tetracycline transactivator (rtTA) coding sequence inserted between the stop codon and the polyadenylation sequence of the Foxa2 gene. The coding region of the Foxa2 gene remains intact, so as to minimally disrupt Foxa2 transcriptional function. Locations of polymerase chain reaction primers for genotyping are indicated. The TRECre transgene contains seven copies of the tet-operator, which requires binding of both doxycycline and the rtTA for Cre-recombinase expression (Perl et al.,2002).
Download figure to PowerPoint
To determine whether the targeted Foxa2ITA allele has preserved transcriptional function of Foxa2, we examined the phenotypes of Foxa2ITA/ITA mice. These homozygotes survive at expected Mendelian ratios, reproduce normally, and do not show any phenotypic features of Foxa2−/− embryos (early embryonic lethality with absence of node formation and neural and endodermal differentiation defects) or Foxa2+/− mice (jaw malocclusion with incisor overgrowth and an abnormal circling gait; Ang and Rossant,1994; Weinstein et al.,1994). The phenotypes of Foxa2ITA/ITA animals indicate adequate function of the allele, despite the presence of exogenous sequences in the 3′-UTR. In contrast, animals generated with the neor containing form of the allele (Foxa2ITAneo) have phenotypes suggestive of hypomorphic Foxa2 function (data not shown). We also compared Foxa2 mRNA expression in heterozygote and homozygote Foxa2ITA embryos and wild type controls using in situ hybridization. Foxa2 expression appears normal in the rostral floor plate/notochord and pharyngeal endoderm in 10 somite stage (ss) embryos (Fig. 2C–E), consistent with our phenotypic analyses.
Figure 2. The Foxa2ITA;TRECre system recapitulates endogenous Foxa2 expression. A–E: In situ hybridization to detect Foxa2 mRNA. A–E: Frontal view. A′–E′: Lateral view of corresponding embryo. A: At the 0 somite stage (ss). B: At the 5 ss. C–E: At the 10 ss. Genotypes are indicated. Note preservation of in situ pattern in Foxa2ITA/+ and Foxa2ITA/ITA embryos. F–J: Lineage analysis of Cre-recombination with the Foxa2ITA;TRECre system. LacZ staining reveals cells in which the Rosa26LacZ Cre reporter allele is recombined. The genotype of all embryos is Foxa2ITA/ITA;R26lacz;Cre. F–I: Frontal view. F′–I′: Lateral view of corresponding embryo. F: Late allantoic bud stage. G: At the 2 ss. H: At the 5 ss. I: At the 12 ss. J, J′: Right and left views of a 25 ss embryo, respectively. aip, anterior intestinal portal; ave, anterior visceral endoderm; en, endoderm; fg, foregut; fp, floor plate; hg, hindgut; ht, heart; lb, liver bud; lbs, late bud stage; mg, midgut; n, node; nc, notochord; pe, pharyngeal endoderm.
Download figure to PowerPoint
Cre Activity With the Foxa2ITA;TRECre System Recapitulates Endogenous Foxa2 Expression
To evaluate the fidelity and activity of the Foxa2ITA;TRECre system, we compared the temporospatial expression pattern of the wild-type Foxa2 locus with recombination resulting from Foxa2ITA-regulated Cre.
Whole-mount in situ hybridization of wild-type embryos from the 0 to 10 ss (Figs. 2A–C, 3A) confirmed the previously reported expression of Foxa2 in the node, notochord, floor plate, and definitive endoderm (Ang et al.,1993; Monaghan et al.,1993; Ruiz i Altaba et al.,1993; Sasaki and Hogan,1993; Ang and Rossant,1994). Foxa2 mRNA is present in the invaginating endoderm of the anterior intestinal portal of 0–10 ss embryos (Fig. 2A–C). At the 5 ss, many cells in the foregut endoderm of the first pharyngeal arch express Foxa2, but expression is patchy laterally as compared with more uniform staining in the midline (Fig. 3A). Note that the most intense Foxa2 staining detected at these stages is in the ventral neural tube/floor plate (Fig. 2A–C).
Figure 3. Sections of endogenous Foxa2 expression and lineage analysis of the Foxa2ITA;TRECre system. A: Transverse sections of a whole-mount in situ hybridized 5 somite stage (ss) wild-type embryo. B: Parasagittal section of a LacZ-stained late bud stage embryo, where blue staining indicates cells in which R26lacZ is recombined from prior or current expression of the Foxa2ITA and Cre. C: Transverse sections of a LacZ-stained 2 ss embryo. D: Transverse sections of a LacZ-stained 5 ss embryo. E: Transverse sections of a LacZ-stained 12 ss embryo. Genotypes of LacZ embryos are Foxa2ITA/ITA;R26lacZ;Cre. Sections progress rostral to caudal as denoted by primes. ave, anterior visceral endoderm; de, definitive endoderm; fg, foregut; fp, floor plate; ht, heart; lb, liver bud; mg, midgut; n, node; nc, notochord; pa1, pharyngeal arch 1; pp1, pharyngeal pouch 1; pp/nc, prechordal plate/notochord. Black arrowheads indicate areas of lateral gut where cells express lower levels of Foxa2, and there is resulting patchy Cre activity.
Download figure to PowerPoint
Lineage analysis of Cre activity in Foxa2ITA/ITA mice bearing TRECre and the Rosa26lacZ Cre reporter allele (Soriano,1999; genotype Foxa2ITA/ITA;R26lacZ;TRECre) was performed in embryos at multiple embryonic stages after exposure to doxycycline from e1.5 (Figs. 2F–J, 3B–E). Although Foxa2 expression has been demonstrated in visceral endoderm (Perea-Gomez et al.,1999; Kinder et al.,2001; Norris et al.,2002), recombination resulting from Foxa2ITA;TRECre activity is restricted to the definitive endoderm and not detected in visceral endoderm or its derivatives (Figs. 2F,G, 3B, and data not shown). Doxycycline is accessible to the early embryos, as indicated by bursts of doxycycline exposure as early as e0 to e2.5, resulting in Cre-mediated recombination when the TRECre is used with an rtTA expressing transgene (Perl et al.,2002). As will be discussed further below, several lines of evidence indicate that Cre activity within the Foxa2 domain is very sensitive to levels of rtTA which is, in turn, controlled by both the dosage of the Foxa2ITA allele, and by cell-specific transcriptional activity of the Foxa2 locus. Thus, we interpret the lack of recombination in visceral endoderm as a reflection of transcriptional activity of Foxa2ITA in these cells resulting in a level of rtTA below the threshold required for TRECre activation.
At the late allantoic bud stage, lacZ staining, indicative of previous Cre activity, is detected in the definitive endoderm and anterior primitive streak (Figs. 2F, 3B), and at e8.0 (2 ss), lacZ staining is detected in the definitive endoderm, node, and ventral neural tube (Figs. 2G, 3C), encompassing all other known expression domains of Foxa2. The mosaic appearance of Cre expression in the definitive endoderm at this stage may reflect different endodermal lineages arising within the primitive streak (Lawson et al.,1986; Lawson and Pedersen,1987; Tam et al.,2007), some of which may not express Foxa2, or it may be an indication that there is a subset of cells that express at levels insufficient to activate TRECre.
At e8.25–e8.75 (5–12 ss), lacZ-positive cells are evident in the invaginated gut tube and throughout the length of the notochord and floor plate (Figs. 2H,I, 3D,E). As observed in the in situ assay for Foxa2 (Fig. 3A), there is patchy lacZ staining in the lateral foregut endoderm (Figs. 2H, 3D) and in the midgut at the 5 ss stage (Fig. 3D′). By the 12 ss, lacZ staining is nearly confluent in the foregut (Figs. 2I, 3E).
At e9.5 (25 ss), lacZ-positive cells continue to reflect the Foxa2 lineage. Staining is confluent in the ventral foregut endoderm, while expression laterally in the foregut and in the midgut endoderm remains patchy (Figs. 2J, 4B–B″). Hindgut staining is relatively less robust (Fig. 2J). The lower number of lacZ-positive cells detected in both the mid- and hindgut (as compared with foregut) has several potential explanations. Again, population of this region of gut with cells of an endodermal lineage that has not previously expressed Foxa2, or loss of a Foxa2-positive lineage with growth, are both possibilities based on previous lineage tracer studies within the gut that reveal distinct endodermal origins of the mid- and hindgut from that of the foregut (Lawson et al.,1986; Tam et al.,2007). Because less Foxa2 expression is detectable within the nascent mid- and hindgut at e8.5–e9.0 than in the foregut (Ang et al.,1993; Monaghan et al.,1993; Sasaki and Hogan,1993; and data not shown), variability of rtTA production within these expression domains likely also contributes. Notably, at e9.5, we consistently detected more labeled cells in the left than in the right foregut of the second and third pharyngeal arches, suggesting higher levels of rtTA production on the left side (Fig. 4B′,B″,C′, arrowheads). This asymmetry was not seen in the first arch endoderm at earlier stages (see Fig. 3D,E), which suggests that ongoing Foxa2 activity in the developing pharynx\foregut endoderm influences Cre activity at this later stage and that this activity is asymmetric within the pharyngeal endoderm. Overall, these patterns of Cre-mediated recombination confirm the fidelity of the Foxa2ITA;TRECre system to endogenous Foxa2 regulation.
Figure 4. Timing of doxycycline administration affects Cre expression for the Foxa2ITA;TRECre system. Lineage analysis of Foxa2ITA;TRECre system by LacZ staining for Cre-recombinase activity using the Rosa26LacZ Cre reporter allele with varying bursts of doxycycline administration. All embryos are genotype Foxa2ITA/ITA;R26lacZ;Cre at embryonic day (e) 9.5. A: No doxycycline exposure. There is no LacZ staining, indicating the system requires doxycycline for Cre expression and is not leaky when the conditional reporter allele is passed maternally. B: Maternal doxycycline administration from e1.5 to e9.5 and at e5.5–e9.5 (C). Note that with these periods of administration, Cre activity is detected in all known expression domains of Foxa2, consistent with endogenous Foxa2 expression beginning at e5.5. D: Doxycycline pulse from e7.5–e9.5. Staining is restricted to the floor plate of the neural tube. Rostral(′) and caudal(″) sections with the location and plane of section indicated by black lines on the corresponding whole-mount embryo. Right and left are denoted on the first section. fg, foregut; fp, floor plate; ht, heart; lb, liver bud; nc, notochord; pa2, pharyngeal arch 2; pa3, pharyngeal arch 3; pe, pharyngeal endoderm. Black arrowheads indicate areas of left lateral gut where cells express higher levels of Foxa2ITA than right-sided cells.
Download figure to PowerPoint
Cre Production by Foxa2ITA;TRECre Requires Doxycycline and Is Responsive to rtTA Gene Dosage
For temporal regulation of Cre activity within the Foxa2 domain, it is crucial that the system is not leaky in embryos and is inducible with doxycycline. We mated Foxa2ITA/ITA;R26lacZ/lacZ females to Foxa2ITA/ITA;TRECre males and found no evidence of Cre activity in Foxa2ITA/ITA;R26lacZ;TRECre embryos in the absence of doxycycline administration (Fig. 4A). However, it is important to appreciate that, in the germline of TRECre males, Cre is sporadically produced. Approximately 50% of R26lacZ alleles transmitted from a TRECre male are recombined in the sperm, giving rise to embryos that bear a ubiquitously recombined R26lacZ allele and whose tissues are uniformly stained with lacZ (data not shown). This event is independent of the Foxa2 or TRECre genotype of the embryo, and occurs in the absence of doxycycline administration to the female. We observe this phenomenon commonly with Cre targeted to other loci, and avoid premature recombination of conditional alleles by using the mating strategy described above, in which conditional alleles are passed through the female.
We also found that the dosage of rtTA is crucial to obtaining reproducible Cre activity in the Foxa2 domain (Fig. 5). Foxa2ITA/ITA;R26lacZ;TRECre embryos (ITA homozygotes) demonstrate robust and reproducible Cre activity (Fig. 5A). In contrast, ITA heterozygotes (genotype Foxa2ITA/+;R26lacZ;TRECre) have patchy, sporadic recombination in cells within the Foxa2 domain (Fig. 5B).
Figure 5. Transactivator gene dosage affects Cre-recombinase expression. Cre-recombinase activity demonstrated by the Rosa26lacZ reporter allele in 5 somite stage (ss) littermates. A:Foxa2ITA homozygote with robust staining in all known Foxa2 expression domains. B:Foxa2ITA heterozygote with sporadic staining. C: Control without Cre transgene shows no staining. Frontal and left lateral (prime) views are shown. Embryo genotypes are noted.
Download figure to PowerPoint
To examine the effects of timing of induction, we varied the timing of doxycycline administration to the pregnant female and examined the Cre recombined lineage in e9.5 embryos (Fig. 4). Maternal doxycycline administration beginning on e1.5 or e5.5 and continuing until embryo harvest captures all of the previously demonstrated regions of Foxa2ITA;TRECre activity (Fig. 4B,C). If we provide only a brief pulse of doxycycline, from e7.5–e9.5, Cre activity is restricted to the floor plate (Fig. 4D). This finding is notable, because the only previously described Cre driver with floor plate expression begins at e8.5, and activity is not restricted to the floor plate (Li and Lufkin,2000). Note that the floor plate expresses Foxa2 at significantly higher levels than endoderm at these stages (Fig. 2A–C). This finding indicates that, with administration beginning at e7.5, peak levels of doxycycline in the embryo coincides with decreasing Foxa2 transcription in the definitive endoderm and notochord, resulting in insufficient rtTA production for activation of TRECre in these tissues at e7.5–e9.5. This timing is also consistent with a previous study that assessed the kinetics of rtTA activation of a TRE-linked reporter in which reporter expression was evident after 6 hr of maternal doxycycline exposure, but maximal induction was not evident until 24 hr (Ludwig et al.,2004). In our system, once TRECre expression is induced by doxycycline, additional time is required for Cre-mediated recombination and subsequent expression of β-galactosidase from the recombined R26lacZ reporter. In contrast to the lack of activity in definitive endoderm with a delayed doxycycline pulse (Fig. 4D), continued high level Foxa2 expression in floor plate from e7.5–e9.5 (Fig. 2B′,C′) permits robust Cre activity in this tissue (Fig. 4D).