Genetic targeting of the endoderm with claudin-6CreER

A full description of the ontogeny of the β cell would guide efforts to generate β cells from embryonic stem cells (ESCs). The first step requires an understanding of definitive endoderm: the genes and signals responsible for its specification, proliferation, and patterning. This report describes a global marker of definitive endoderm, Claudin-6 (Cldn6). We report its expression in early development with particular attention to definitive endoderm derivatives. To create a genetic system to drive gene expression throughout the definitive endoderm with both spatial and temporal control, we target the endogenous locus with an inducible Cre recombinase (Cre-ERT2) cassette. Cldn6 null mice are viable and fertile with no obvious phenotypic abnormalities. We also report a lineage analysis of the fate of Cldn6-expressing embryonic cells, which is relevant to the development of the pancreas, lung, and liver.


INTRODUCTION
During gastrulation, pluripotent cells of the epiblast become specified to form one of the three embryonic germ layers: ectoderm, mesoderm, or definitive endoderm. The definitive endoderm gives rise to the epithelia of numerous organs including the thyroid, lung, liver, stomach, pancreas, and intestines. In the past decade, the potential use of embryonic stem cells (ESCs) as cell based therapeutic vehicles to treat diseases has been amply discussed. Current investigations are underway in numerous laboratories to elucidate paracrine factors that can direct differentiation of ESCs to definitive endoderm, which can then be further instructed to differentiate into an organ progenitor cell, and ultimately to a fully differentiated cell type of interest (such as hepatocytes of the liver, epithelial cells of the lung, or ␤ cells of the pancreas).
This presents the challenge of understanding how definitive endoderm is formed and patterned. Specification of definitive endoderm is a Nodal-dependent process (Schier, 2003) but little is known about endoderm progenitor cells; the cues responsible for their expansion and the signals that pattern the gut tube along the anterior-posterior axis are yet unknown. Tools that would enable the genetic manipulation of the definitive endoderm will help to better understand these processes. To that end, we have sought a marker that would target definitive endoderm during development. A promising candidate arose from an in situ hybridization screen performed by Rosa Beddington's group (Sousa-Nunes et al., 2003). They examined expression of genes from an E7.5 endoderm cDNA library, which was derived from visceral and definitive endoderm, and the node-tissues responsible for specification of the axial mesendoderm. One clone, r8707a53, was expressed in the forebrain, visceral yolk sac, otic vesicle, and throughout the entire definitive endoderm at E9.0. This clone encoded a member of the Claudin superfamily of proteins, Claudin-6.
Claudins are the major structural proteins involved in tight junction formation. Currently in the mouse, there are 24 genes in the Claudin superfamily. Based on expression patterns, it is possible that Claudins come in two varieties: "housekeeping" Claudins and tissue-specific Claudins (Van Itallie and Anderson, 2006).
Cldn6 was first identified by searching for sequences in genomic databases that were similar to Cldn1 and Cldn2. It was isolated by PCR from mouse kidney and found to encode a protein of 219 amino acids in length with a molecular mass of 23.4 kDa (Morita et al., 1999). Cldn6 has been subsequently studied both in vitro and in vivo (Turksen and Troy, 2001;Turksen and Troy, 2002;Troy et al., 2005;Arabzadeh et al., 2006). While its functional role in the epidermal permeability barrier is of importance, our interest in Cldn6 is as a marker of endoderm and a means to manipulate gene expression therein.

Expression of Claudin-6, a Global Marker of Definitive Endoderm
We examined expression of Cldn6 by in situ hybridization, in sections and whole mounts, from gastrulation to the initial stages of organogenesis ( Fig. 1). At E6.5, Cldn6 is broadly expressed throughout the epiblast as well as the hypoblast. It is absent in the primitive streak (Fig. 1A) and in some visceral endoderm surrounding the epiblast (Fig. 1A,D). This lack of expression in the primitive streak is not surprising, as cells undergo an epithelial-to-mesenchymal transition (EMT) as they migrate to form mesoderm and definitive endoderm, thus dissolving adherens and tight junctions (Solursh and Revel, 1978;Tam et al., 1993;Cano et al., 2000;Batlle et al., 2000;Zohn et al., 2006;Ikenouchi et al., 2003;Ohkubo and Ozawa, 2004). At E7.5, expression is still evident throughout most of the epiblast (Fig. 1B,E). The primitive streak and nascent mesoderm are devoid of Cldn6 expression (Fig. 1E). By E8.5, expression begins to become restricted to the definitive endoderm (Fig. 1C,F). This is the stage when initial patterning of the gut tube along the anteriorposterior axis is thought to occur. At E9.5, intense expression is seen in the entire gut, otic vesicles, and a small region of the forebrain (Fig. 1G). This expression pattern at E9.5 is comparable to what is reported by others (Gitton et al., 2002;Sousa-Nunes et al., 2003). At E10.5, expression is similar to E9.5, but staining in the mesonephros and forebrain is now more apparent (Fig. 1H).
To examine whether Cldn6 maintains epithelial expression in endoderm-derived organs, we performed whole mount and section in situ hybridization on embryonic lung and pancreas. Cldn6 is strongly expressed in the epithelia of both organs ( Fig. 2A-F, 2G-H, pancreas). On sections, no staining in the mesenchyme is evident (Fig. 2E,F,H).
To summarize, Cldn6 is expressed broadly at early stages of development. However, by E9.5, expression of Cldn6 is largely restricted to definitive endoderm along its entire length, mak- Fig. 1. Expression of Cldn6 from commencement of gastrulation to the early stages of gut tube organogenesis. A,D: Expression at E6.5 is quite broad throughout the epiblast, with the exception of the primitive streak and visceral endoderm. Similar expression is seen at E7.5 (B,E). C,F: By E8.5, expression begins to be restricted to the endoderm and by E9.5 (G) and E10.5 (H), expression is restricted to the endoderm, otic vesicle (arrowhead), and mesonephros (arrow). A-C are representative whole mount stainings for the sections in D-F), respectively. Embryos in D-F were sectioned in their deciduas. Plane of section is indicated by a line in the corresponding whole mount embryo. For E, posterior is facing up. dec, decidua; ect, ectoderm; mes, mesoderm; end, endoderm; rm, Reichert's membrane; nf, neural fold; he, heart.
ing it a good candidate as a global endoderm marker. Other markers that have been used to mark endoderm include Shh, FoxA genes, and Sox17. Shh, while initially expressed throughout the definitive endoderm, is downregulated in the prospective pancreatic region (Apelqvist et al., 1997). Of the FoxA genes, FoxA3 is not expressed in the most anterior endoderm derivatives, while FoxA1 and FoxA2 are expressed throughout the definitive endoderm as well as other tissues such as the notochord and neural tube (Monaghan et al., 1993). Sox17, while expressed in nascent endoderm, is downregulated in the endoderm at stages post E7.5 and is expressed in other tissues such as vasculature (Kanai-Azuma et al., 2002;Matsui et al., 2006;unpublished observations). This makes Cldn6 one of the few genes that allows for targeting of the definitive endoderm throughout the entire anterior-posterior axis before and after gut tube patterning (E8.0 onward).

Expression of Cldn6 in the Adult Mouse
We investigated whether Cldn6 was expressed in epithelial structures in the adult. Adult mice of both sexes were sacrificed and organs were harvested and transcriptional analysis was performed by RT-PCR. Out of the 23 tissues examined, Cldn6 was present in 11 tissues (see Supplemental Fig. 1, which can be viewed at www.interscience.wiley.com/jpages/1058-8388/suppmat). Using ␤-actin as a loading control, we can make a semiquantitative assessment of relative expression levels. Cldn6 is most highly expressed in the olfactory epithelium, followed by the coagulating gland, kidney, eye, trachea, prostate, epididymis, lung, liver, eye gland, and uterine horns. Immunohistochemistry with goat anti-Cldn6 polyclonal antibodies, however, failed to detect Cldn6 in lung, trachea, kidney, or testis (data not shown). These results are in broad agreement with previous reports that Cldn6 expression in the adult is very weak or absent (Morita et al., 1999;Turksen and Troy, 2004).

Generation of the Cldn6 CIHV Allele
Based on its expression during early development, we hypothesized that the Cldn6 gene would be useful to drive expression throughout the entire endoderm. One genetic method that has been used extensively for gain of function or loss of function studies is the Cre-ER system (Branda and Dymecki, 2004). Applications of this system can include activation or removal of target genes in a spatially and temporally controlled manner.
A multifunctional cassette was generated consisting of the Cre-ER T2 gene for genetic manipulation, followed by an internal ribosome entry site (IRES), the gene encoding a nuclear localized yellow fluorescent protein variant (the histone 2B/Venus fusion) as a reporter for gene expression, a strong transcriptional stop sequence using the SV40 polyadenylation signal to prevent transcriptional readthrough, and a floxed selection cassette that self-excises in the male germline. We termed this cassette "CIHV" for Cre-ER T2 -IRES-H2B/Venus (Zhou et al., 2007).
We first tested this construct by transfection into COS cells. A plasmid was made with the CIHV cassette cloned downstream from a CAGGS promoter. As shown in Supplemental Figure 2, all transfected cells express a nuclear localized YFP. To test whether Cre recombinase activity is tamoxifen-dependent, we co-transfected a plasmid with a CAGGS promoter upstream of a floxed transcriptional stop and cytoplasmic RFP. RFP expression was dependent upon addition of 4-hydroxytamoxifen to the culture media, verifying that the Cre-ER T2 system is tightly regulated (see Suppl. Fig. 2).
AV3 mouse ES cells (a gift from Jill McMahon) were targeted with the CIHV cassette (Suppl. Fig. 3). Out of a total of 1,474 colonies screened, one clone (E40) successfully underwent homologous recombination at the Cldn6 locus (Fig. 3A,B). Chimeras were generated by blastocyst injection and transmitted the allele through the germline (Fig. 3B). Heterozygotes from this cross were renamed Cldn6 CIHV/ϩ .

A Genetic Lineage Analysis of Cldn6 ؉ Cells at Different Time Points in Development
To test Cre activity, we crossed Cldn6 CIHV/ϩ males to homozygous ROSA26 Reporter (R26R) females (Suppl. Fig. 4A). The R26R line has a floxed transcriptional stop sequence followed by the neomycin resistance/ lacZ fusion gene ␤geo inserted into the ROSA26 locus, a non-coding open reading frame in the mouse that is ubiquitously expressed in all tissues (Soriano, 1999). This cross permits indelible labeling of cells that express Cldn6 at the time of tamoxifen injection and all of their progeny.
We experimented with tamoxifen dosage as was done previously (Hayashi and McMahon, 2002) and found that a 4-mg injection of tamoxifen to pregnant females provided maximum labeling of embryos without significant embryonic loss (between 0 -2 embryos resorbed per litter). We then performed a time course of tamoxifen injection during embryonic development, which is schematized in Supplemental Figure 4B. Four milligram tamoxifen injections were administered to pregnant females between E5.5 and E12.5, with embryos whose mothers were injected between E5.5 and E10.5, harvested on E12.5, and those whose mothers were injected on E11.5 and E12.5, harvested on E15.5. Selected whole mount images are shown in Figure 4. At all stages examined, neither Cre-independent ␤-galactosidase expression nor tamoxifen-independent ␤-galactosidase expression was detected (data not shown). Embryos at each stage of injection were sectioned post-staining to better visualize localization of the signal to the epithelial or mesenchymal compartments (data not shown).
Tamoxifen injection to pregnant females at E5.5 labels cells throughout the embryo, both in the epithelia and mesenchyme (Figs. 4A-C, 5A). This is consistent with the ubiquitous expression of Cldn6 at the time Cre recombinase would be most active (roughly E6.0 -7.0). Based on the composite data, it appears E7.5 is the best time point for tamoxifen administration (Fig. 4D-F) if one aims to mostly label definitive endoderm and its derivatives without ubiquitous labeling of the embryo. From this stage forward, Cre-mediated recombination is detected in epithelia including all endoderm-derived organs (Figs. 4D,E,G-J, 5C,F), the kidney (Figs. 4G,K, 5G,H), olfactory epithelia, lens epithelia, and semicircular canal (data not shown). Cells that appear to be melanocytes also label (Fig.  4C,F,L), although Cldn6 expression in these cells is too low to visualize by in situ hybridization. Based on the kinetics of tamoxifen, E7.5 would be the optimal stage for misexpression in definitive endoderm as Cre recombinase would be active around the time the endoderm is patterned and before or-  Upon successful recombination, the entire coding sequence for Cldn6 is removed, generating a null allele. After NcoI digest, the wild type locus generates a band of approximately 12.5 kb, while the targeted allele has a smaller band close to 8 kb. For reference, the first two bands on the 1-kb ladder are 10 and 8 kb, respectively. As seen in the parental ES cell lane (AV3), only one band is present higher than 10 kb, while in the targeted ES cell clone lane (E40), there is a band at the same height as in the wild type lane, but a second band present at 8 kb. B: Genotyping by PCR confirms germline transmission of the CIHV allele. AV3 and E40 refer to the parental ES cell line and targeted ES cell line used for blastocyst injection, respectively. The upper gel distinguishes between the wild type and targeted locus. The lower gel is a PCR to demonstrate ACN cassette removal during germline transmission. C,D: Cldn6 CIHV/CIHV mice are null mutants for Cldn6. In situ hybridization for Cldn6 in embryos from a Cldn6 CIHV/ϩ intercross. C: Cldn6 CIHV/ϩ ; D: Cldn6 CIHV/CIHV . As shown in D, Cldn6 mRNA is absent in the Cldn6 CIHV/CIHV embryo. ganogenesis begins. Of note, given that the degree of Cre-mediated recombination is dependent upon the dose of tamoxifen administered, not all Cldn6-expressing cells undergo recombination at the dose used here.
In summary, Cldn6 is expressed in various epithelia in the developing em-bryo. It is a more useful definitive endoderm marker than others that have been described to date including the widely used marker Sox17. Using a Fig. 4. Lineage analysis of Cldn6 ϩ cells at different times in embryonic development. Tamoxifen injection at E5.5 shows widespread labeling throughout the embryo (A-C). Tamoxifen injection at E7.5 and later shows epithelial-restricted expression of ␤-galactosidase in definitive endoderm derivatives (D,E, G-J). Epithelia of the kidney is also labeled (G, K). Melanocytes also label (C, F, scattered cells in L), with a decrease in labeling correlating to later times of tamoxifen injection. Embryos administered tamoxifen at E5.5 (A-C) and E7.5 (D-F) were harvested at E12.5 for analysis. Embryos administered tamoxifen at E11.5 (G-L) were harvested at E15.5 for analysis. panc, pancreas; lu, lung; liv, liver; st, stomach; kid, kidney; int, intestine.
tamoxifen-inducible form of Cre recombinase, we are able to target the entire endoderm using the Cldn6 CIHV line. Based on the timing of tamoxifen administration, one can target various epithelia in the developing embryo for gain of function or loss of function studies. To our knowledge, this is the first and only Cre-ER line available to globally target the endoderm at stages prior to patterning and during organogenesis without affecting most other tissues.
The Cldn6 CIHV line has distinct advantages compared to other genetically modified mice. While Shh Cre-ER mice (Harfe et al., 2004) can target definitive endoderm, they cannot target pancreatic endoderm during initial stages of organogenesis. As mentioned earlier, FoxA3 Cre transgenic mice do not target the most anterior definitive endoderm (Monaghan et al., 1993;Lee et al., 2005). A FoxA2 rtTA line was recently reported (Frank et al., 2007). This will prove to be quite useful for crosses with tetracycline operator responder lines. Of note, the authors preserved the FoxA2 coding sequence, as FoxA2 is haploinsufficient (Ang and Rossant, 1994;Weinstein et al., 1994). Crossing to Cre responder lines, however, will be somewhat laborious as it involves generation of a triple transgenic. Sox17 Cre-ER mice have not been reported, though they would not be useful for endoderm targeting post-E7.5 due to predominant vascular expression (Matsui et al., 2006;unpublished observations).

Cldn6 CIHV/CIHV Mice Are Viable and Exhibit No Overt Phenotype
To examine whether Cldn6 plays a role in gastrulation or gut tube morphogenesis, mice heterozygous for the Cldn6 allele (Cldn6 CIHV/ϩ ) were mated. Female mice were allowed to give birth and progeny were genotyped by tail biopsy upon weaning. Homozygous null mice (Cldn6 CIHV/CIHV ) were retrieved in the expected Mendelian ratio by PCR genotyping. These mice show no obvious anatomical phenotypic abnormalities and responded to sound (data not shown).
To ensure that targeting the Cldn6 locus with the CIHV cassette generated a complete null allele, mice het- pancreas, injection at E6.5 (E) leads to strong staining in the epithelia as well as in some scattered mesenchyme. Injection at E7.5 (F) is restricted to the epithelia. No mesenchymal staining was seen in any section examined. G,H: In the E15.5 kidney, injection at E11.5 (G) and E12.5 (H) are shown. All magnification at 10ϫ except F at 20ϫ. inj., injection. erozygous for Cldn6 were again mated and embryos were harvested at E9.5 for in situ hybridization. As seen in Figure 3D, no Cldn6 mRNA was detected in Cldn6 CIHV/CIHV embryos.
To investigate whether Cldn6 null mice are able to mate normally, Cldn6 CIHV/CIHV mice were mated and allowed to give birth. Females gave birth to litters of normal size. Newborns looked healthy with no grossly abnormal phenotype (data not shown). Mothers showed no rejection of the pups. Mothers nursed normally and successfully exhibited pup retrieval behavior (data not shown). This would suggest that pheromone detection and olfactory cues, at least with regards to mating and nursing behaviors, are normal in these mice.
Concurrently, E8.5 embryos were dissociated and embryonic germ layers were analyzed by microarray analysis (Sherwood et al., 2007). Upon examining the list for members of the Claudin gene family, there are eight different Claudins expressed in the endoderm, with Cldn6 being the highest relative value (Table 1). The other seven, in order of relative expression, are Cldn7, Cldn12, Cldn8, Cldn4, Cldn10, Cldn3, and Cldn11. While this type of analysis does not provide cellular resolution, it suggests that other Claudin family members may compensate for loss of Cldn6 in the endoderm. Further analysis by in situ hybridization and double loss of function studies may shed insight onto this mechanism. In addition, Cldn6 CIHV/CIHV mice may have subtle phenotypic abnormalities that will be more difficult to reveal or that may appear more noticeable with age.

Role of Cldn6 in Gastrulation and Endoderm Development
Cldn6 null mice appear phenotypically normal based on anatomical criteria and overt behavior. From microarray data, this is most likely due to compensation by another Claudin family member, with Cldn7 as a likely candidate. This does not rule out a possible role for Cldn6 in endoderm development. Functional redundancy has been shown in processes important for gastrulation before. For example, in the frog, Cer1 and Lefty1 both play important roles in blocking BMP and Nodal signaling. In mouse, Cer1 Ϫ/Ϫ mutants (Simpson et al., 1999;Shawlot et al., 2000;Stanley et al., 2000;Belo et al., 2000) and Lefty1 Ϫ/Ϫ (Meno et al., 1998) mutants have no overt phenotype at gastrulation. However, Cer1 Ϫ/Ϫ ; Lefty1 Ϫ/Ϫ double mutants develop multiple primitive streaks due to Nodal expression expanding outside of the posterior region of the embryo (Perea- Gomez et al., 2002). Therefore, a Cldn6 Ϫ/Ϫ ; Cldn7 Ϫ/Ϫ mutant might show a severe phenotype, and manifest at gastrulation or perhaps even earlier upon egg cylinder formation.
An application of this genetic system is to target the endoderm for misexpression studies. One means is to use a gain-of-function approach, crossing the Cldn6 CIHV line with different Cre responder lines that constitutively activate different signaling pathways upon removal of a floxed stop cassette. The role of the Notch, Hedgehog, and Wnt signaling pathways in Cldn6-positive cells would be interesting to examine given that many responder lines currently exist (Murtaugh et al., 2003;Jeong et al., 2004;Harada et al., 1999). Further a Expression of different members of the Claudin family at E8.5. E8.5 embryos were dissociated and separated by embryonic lineage. cDNA was isolated for each sample and hybridized to Affymetrix microarrays. Numbers reflect relative expression levels. Only members of the Claudin family of genes are shown here. In the definitive endoderm, Cldn6 has the highest expression, followed by Cldn 7,12,8,4,10,3,and 11. Accession, accession number for gene in public NCBI database; endoderm, definitive endoderm; EE endoderm, extraembryonic endoderm; EE mesoderm, extraembryonic mesoderm.
analysis will be necessary to determine what role these different signaling pathways play in the definitive endoderm with regards to both organ specification and regulation of endoderm progenitor cells. The Cldn6 CIHV line, combined with the recently reported FoxA2 rtTA line (Frank et al., 2007), will prove to be useful reagents in this regard. This understanding will lead to more efficient strategies to generate and expand definitive endoderm from ESCs, as well as subsequent differentiation to various organ progenitor cells.

In Situ Hybridization
In situ hybridization was performed essentially as described using digoxigenin-labeled probes (Wilkinson, 1992). Tammy-Claire Troy (Ottawa Health Research Institute) kindly provided the Cldn6 in situ plasmid.

RT-PCR
Organs were harvested from adult female and male ICR mice (Taconic) and stored in RNAlater (Ambion) until use. Total RNA was extracted with a RNeasy Kit (Qiagen).

Generation of the CIHV Cassette
We generated a construct consisting of Cre-ER T2 , followed by an IRES element, H2B-Venus, a strong transcriptional stop, and a self-excising drug selection cassette (ACN) (construct referred to as CIHV). Pfx Taq polymerase was used to generate 5Ј and 3Ј homology arms for Cldn6. PCR conditions were as follows: 1 l AV3 (strain 129Sv) genomic DNA, 10 l 10ϫ PCR buffer, 5 l Pfx enhancer, 1.5 l 10 mM dNTPs, 22 l 3M betaine, 1 l 50 mM MgSO 4 , 0.65 l DMSO, 1.5 l each of 10 M forward and reverse primers, 5.35 l dH 2 O, and 0.5 l Platinum Pfx polymerase (Invitrogen). Thermocycler conditions were as follows: initial denaturation at 94°C for 2 min, followed by 35 cycles of denaturation at 94°C for 10 sec, annealing at 55°C for 30 sec, and extension at 68°C for 2 min plus 10 sec per cycle, with a final extension at 68°C for 7 min. Fragments were TOPO cloned into the pCRII vector (Invitrogen) and confirmed by DNA sequencing (MCB Sequencing Facility). Both arms were subsequently ligated with the CIHV cassette.

Targeting the Cldn6 Locus With the CIHV Cassette
C6-CIHV vector was linearized and phenol:chloroform extracted. AV3 mouse ES cells (7 ϫ 10 6 cells/ml; a kind gift from Jill McMahon) in Em-bryoMax qualified electroporation buffer (Specialty Media) were electroporated with 40 g of DNA and selected with 300 g/ml Gentacin (Invitrogen) for 8 -9 days. Roughly 1,000 colonies were picked and subjected to Southern blot with a 995-bp 5Ј probe (Sambrook et al., 1989). Correctly targeted colonies were identified (E40). The E40 ES cells were used for blastocyst injection (Genome Manipulation Facility). Eighteen chimeras were born, 11 of which were high-grade chimeras. Three (males D1, D4, and E3) transmitted through the germline from crosses with C57Bl/6J females (Jackson Laboratories).
All animal experiments described in this report have been approved by the Harvard University Institutional Animal Care and Use Committee.

PCR Genotyping
Genomic DNA isolation from tail biopsy was performed following standard protocols (Laird et al., 1991). Offspring were genotyped by PCR using the following strategy. The allele was genotyped using a forward primer located at the end of the 5Ј homology arm (gtCldn6F) with two reverse primers, one located in the Cre coding sequence (Cre-R) and the other located in the middle of the endogenous Cldn6 coding sequence (gtCldn6R1). If the locus was undisrupted, gtCldn6F and gtCldn6R1 would amplify a 294-bp product. If the locus was targeted with the CIHV cassette, the product would be slightly less than 200 bp. To ensure self-excision of the selection cassette, a PCR strategy was designed to check for the presence of the ACN cassette. Two forward primers, one in the Venus coding sequence and the other in the ACN cassette itself were combined with a reverse primer, gtCldn6R2, located in the 3Ј homology arm. If the ACN cassette was still present, ACN-F and gtCldn6R2 would produce ϳ200-bp fragment, while if the ACN cassette was excised, the Venus-F and gtCldn6R2 would produce ϳ600-bp fragment. In all cases, there was successful ACN excision through the male germline. Sequences for the primers used are: ACN-F: 5Ј-ATC GCC TTC TAT CGC CTT CTT GAC GAG TTC-3Ј; Cre-R: 5Ј-CGG ACA GAA GCA TTT TCC AGG TAT GCT CAG-3Ј; gtCldn6F: 5Ј-CAC TAC ACA GCC CCC TCA AC-3Ј; gtCldn6R1: 5Ј-TGA GGA GGG TGA CAA CAC AG-3Ј; gtCldn6R2: 5Ј-GAG GTG GAG CTT GGA CTC AG-3Ј; Venus-F: 5Ј-ATC ACT CTC GGC ATG GAC GAG CTG TAC AAG-3Ј.

Tamoxifen Administration
Tamoxifen (Sigma) was dissolved in corn oil (Sigma) at a concentration of either 10 or 20 mg/ml and passed through a 0.2-m filter. Injections were administered to pregnant females intraperitoneally.

X-gal Staining of Cldn6 CIHV/؉ ; ROSA26 Reporter/؉ Embryos
Timed pregnancies were collected, using E0.5 as the date of plug detection. Embryos were dissected out of the decidua, saving extraembryonic membranes. To ensure penetration, heads were removed and the body was eviscerated. The tissues were fixed in 4% paraformaldehyde in PBS for 1 hr at 4°C, followed by two PBS washes for 15 min each. ␤-galactosidase staining was performed using the HistoMark X-gal Substrate Set (Kirkegaard & Perry Laboratories). Incubations were continued overnight at room temperature for maximal sensitivity.

ACKNOWLEDGMENTS
We thank Amy Greenwood for helpful discussions and Andy McMahon for excellent advice and counsel. We thank Tammy-Claire Troy for sharing her Cldn6 in situ plasmid with us. We also thank Jill McMahon for providing us with AV3 mouse ESCs. W.J.A. was supported by the NIH and HHMI. D.A.M. is an Investigator of the Howard Hughes Medical Institute.