TECHNOLOGY REPORT

Tissue-Specific Expression of Cre Recombinase from thePax8 LocusMaxime Bouchard, Abdallah Souabni, and Meinrad Busslinger*Research Institute of Molecular Pathology, Vienna Biocenter, Vienna, Austria

Received 28 July 2003; Accepted 14 December 2003

Summary: The transcription factor Pax8 is expressed inthe developing thyroid gland, inner ear, kidney, and mid-hindbrain region. Pax8 mutant mice die only postnatallydue to a thyroid gland defect. Here we report the gen-eration and expression analysis of a Pax8cre allele. Crerecombinase activity was detected in known Pax8 ex-pression domains of Pax8cre/� and Pax8cre/cre embryoscarrying the Z/AP transgene, which expresses alkalinephosphatase only after Cre-mediated excision of lacZsequences. Alkaline phosphatase expression was addi-tionally detected in the adrenal gland and in the facial,vestibulocochlear, and cuneate nerves, which have sofar not been associated with Pax8 expression. Our dataindicate that the Pax8cre allele provides a novel impor-tant tool for conditional gene manipulation and lineagetracing in different Pax8 expression domains. genesis38:105–109, 2004. © 2004 Wiley-Liss, Inc.

Key words: Pax8; Cre recombinase; knockin mouse; kid-ney; thyroid gland; inner ear; mid-hindbrain boundary

Gene inactivation by homologous recombination in em-bryonic stem (ES) cells has allowed great progress inmouse genetics by providing important insight into genefunction during development. Mutation of many lociresults, however, in embryonic lethality, which reflectsthe earliest nonredundant role of the mutated gene, andthus precludes analysis of its function in later develop-ment. To circumvent this limitation, a number of strate-gies have been developed in order to inactivate a givengene in a stage- or tissue-specific manner (reviewed byLewandoski, 2001). The most commonly used strategyrelies on the binary Cre/loxP system, which facilitatesexcision of DNA sequences located between two loxPsites (floxed) by the Cre recombinase of bacteriophageP1. Optimal use of the Cre/loxP system critically de-pends on tight regulation of the tissue-specific cre trans-gene, which is used for conditional gene inactivation(Lewandoski, 2001). This aspect has often been prob-lematic, as the expression of conventional cre transgenesunder the control of isolated enhancers and promoterscan be strongly influenced by sequences at the chromo-somal integration site, resulting in ectopic Cre recombi-nase activity. A tighter control of cre expression can be

achieved by insertion of the cre gene into an endoge-nous locus. We have used such a knockin approach togenerate a mouse line expressing a cre gene under thecontrol of the endogenous Pax8 locus (Fig. 1A).

The paired domain-containing transcription factorPax8 is expressed during embryogenesis in various or-gans, including the thyroid gland, inner ear, kidney (pro-,meso-, and metanephros), and mid-hindbrain boundaryregion (Bouchard et al., 2002; Pfeffer et al., 1998; Pla-chov et al., 1990). Despite this complex expressionpattern, Pax8 mutant mice are born at mendelian fre-quency and die only at weaning age due to the absenceof follicular cells in the thyroid gland (Mansouri et al.,1998). The lack of a mutant phenotype in other Pax8-expressing tissues reflects functional compensation ofthe Pax8 loss by the related Pax2 and Pax5 proteins(Bouchard et al., 2002; unpubl. data).

We have recently reported the targeted inactivation ofPax8 by inserting a cre gene in-frame into Pax8 exon 3(Bouchard et al., 2002). However, the Pax8neo allelefailed to express Cre recombinase activity in vivo, pos-sibly due to the presence of the neomycin resistancegene (flanked by two frt sites; Fig. 1A). To investigatethis hypothesis, we eliminated the neo gene in thePax8cre allele (Fig. 1A) by crossing Pax8neo/� mice withan FLPe recombinase-expressing deleter strain (Rodri-guez et al., 2000). Correct generation of the Pax8cre

allele was verified by Southern blot analysis of tail DNA(Fig. 1B).

The Pax8cre allele was tested for recombinase activityby mating Pax8cre/� mice with the Z/AP reporter line(Lobe et al., 1999). This transgenic line ubiquitously

Present address for Maxime Bouchard: McGill Cancer Centre, McGillUniversity, 3655 Promenade Sir-William-Osler, Montreal, Quebec H3G 1Y6,Canada.

* Correspondence to: Meinrad Busslinger, Research Institute of MolecularPathology, Vienna Biocenter, Dr. Bohr-Gasse 7, A-1030 Vienna, Austria.E-mail: busslinger@imp.univie.ac.at

Contract grant sponsors: Boehringer Ingelheim, the Austrian IndustrialResearch Promotion Fund.

DOI: 10.1002/gene.20008

© 2004 Wiley-Liss, Inc. genesis 38:105–109 (2004)

expresses �-galactosidase from a floxed �geo (lacZ)gene and switches to the expression of human placentalalkaline phosphatase (hPLAP) only upon Cre-mediateddeletion of the upstream �geo sequences. Alkaline phos-phatase staining thus provides a simple readout for Crerecombinase activity (Lobe et al., 1999). It is important,however, to note that hPLAP expression is maintained inall descents of Cre-expressing cells, thus defining devel-opmental ancestry by lineage tracing.

hPLAP activity was first detected at embryonic day 8.5(E8.5) in discrete cell clusters within the otic placoderegion of Pax8cre/� Z/AP embryos (Fig. 2A). Followinginvagination of the otic placode at E9.5, hPLAP stainingwas predominantly observed within the otic vesicle,although the ectoderm adjacent to the vesicle containedsome hPLAP� cell clusters (Fig. 2B). By E10.5, most ofthe otic vesicle was hPLAP-positive (Fig. 2D,E). Thisbroad expression resulted in staining of most epithelialcomponents of the inner ear by E16.5, with the excep-tion of the semicircular canals, which showed patchyhPLAP expression in Pax8cre/� Z/AP embryos (Fig. 4C).In contrast, homozygous Pax8cre/cre Z/AP embryos re-vealed uniform hPLAP staining in the entire inner ear(Fig. 4D), indicating that a 2-fold increase in cre expres-sion efficiently activated the Z/AP transgene also in thedorsal part of the inner ear. No hPLAP staining, however,was detected in the otic capsule surrounding the innerear (Fig. 4C,D). Sections of E10.5 embryos at the level ofthe otic vesicle revealed strong hPLAP activity in thevestibulocochlear and geniculate ganglia (Fig. 2E). Thevestibulocochlear ganglion gives rise to the 8th (VIII)cranial nerve that connects the cochlea and vestibularsystems to specific nuclei of the brain stem. The genic-ulate ganglion is the origin of the facial (VII) nerveinnervating the facial musculature, the anterior two-thirds of the tongue, and part of the external ear epithe-lium. hPLAP staining visualized the corresponding neu-ron projections in the tongue and external ear of

Pax8cre/� Z/AP embryos at E15.5 (Fig. 3C; data notshown).

The brain of Pax8cre/� Z/AP embryos expressed Crerecombinase activity at E9.5 only in a few cells at themidbrain–hindbrain boundary (mhb; Fig. 2B,C), whichevolved into a patchy expression pattern in the anteriorhindbrain and tegmentum of the midbrain at E10.5 (Fig.2D) and E15.5 (Fig. 3A). At the later time point, hPLAPactivity was also detected in specific axon tracts of thedorsal hindbrain (Fig. 3B), which most likely correspondto the cuneate fascicle (Paxinos et al., 1991).

The mesonephros of Pax8cre/� Z/AP embryos alsoexpresses hPLAP, which was first detected in a few cellsat E9.0 (data not shown) and then became more evidentby E9.5 (Fig. 2C). The expression in the nephric ductand in some discrete regions of the nephrogenic cordwas irregular, thus reflecting inefficient Z/AP gene acti-vation by Pax8cre in the mesonephros (Fig. 2C,D,F). AtE10.5, hPLAP expression was also visible in the region ofthe cloaca (Fig. 2D,F). The ureter, which induces thedevelopment of the metanephros (adult kidney) byevaginating from the posterior nephric duct (Saxen,1987), revealed a discontinuous pattern of hPLAP ex-pression at E15.5 (Fig. 3B). Sections through the meta-nephric kidney identified hPLAP activity in the nephro-genic tubules and the nonvascular components of theglomeruli (Fig. 4A,B). The Z/AP transgene was moreefficiently activated in glomeruli than in tubules. More-over, S-shaped bodies expressed hPLAP activity only inhomozygous Pax8cre/cre Z/AP embryos (Fig. 4B), sug-gesting that a 2-fold increase in Pax8cre expression leadsto efficient Z/AP activation at this early nephron precur-sor stage. hPLAP staining was also detected in the adre-nal gland, Mullerian duct, and occasionally at the tip ofthe ureter in Pax8cre/� and Pax8cre/cre Z/AP embryos(Fig. 3B; data not shown).

Strong expression of hPLAP activity was observed inthe thyroid gland of Pax8cre/� Z/AP embryos at E15.5

FIG. 1. Generation of the Pax8cre allele. A: Structure of wildtype and mutant Pax8 alleles. The ACTB:FLPe transgene (Rodriguez et al., 2000)was used to generate the Pax8cre allele from the Pax8neo allele (Bouchard et al., 2002) by FLPe-mediated germline deletion of the neoexpression cassette, which was flanked by frt sites (red arrowheads). The different alleles were verified by Southern blot analysis ofEcoRI-digested DNA with the indicated probe. The length of the characteristic DNA fragments is indicated in kilobases. E, EcoRI; pA, SV40polyadenylation site; Sa, SacI; Ss, SspI. B: Identification of the different Pax8 alleles by Southern blot analysis of EcoRI-digested tail DNA.

106 BOUCHARD ET AL.

(Fig. 3D). Surprisingly, however, homozygous Pax8cre/cre

Z/AP embryos revealed hPLAP staining even in the rem-nant of the thyroid gland (data not shown) that lacksfollicular cells in the absence of Pax8 (Mansouri et al.,1998). Hence, Pax8 appears to be expressed also inparafollicular C-cells or their precursors, contrary to pub-lished data (Mansouri et al., 1998). This unexpectedfinding should prompt a reevaluation of the developmen-tal origin of follicular and parafollicular cells, which arethought to differentiate from endoderm and neural crestcells, respectively (Mansouri et al., 1998).

In conclusion, we have shown that cre expression ofthe Pax8cre allele accurately reflects the developmentalexpression pattern of the endogenous Pax8 gene (Bou-chard et al., 2002; Pfeffer et al., 1998; Plachov et al.,1990). Based on the expression in the otic placode andnephric duct, we estimate a delay of about 12 h betweenthe initiation of Pax8 transcription and the detection ofhPLAP activity expressed from the Pax8cre allele. Hence,

the Pax8cre/� mouse cannot be used for Cre-mediatedgene inactivation during the earliest phase of ear andkidney development. However, the Pax8cre line is apowerful tool for Cre-mediated gene manipulation atlater developmental stages, particularly in the inner earand thyroid gland, where the efficiency of Cre recombi-nation is uniformly high. The efficiency of Cre-mediateddeletion is known to vary between different floxed trans-genes and endogenous loci (Lewandoski, 2001). For in-stance, the activation of a Z/EG reporter transgene,which differs from the Z/AP gene only by the replace-ment of the hPLAP gene by GFP (Novak et al., 2000), is4-fold less efficient than deletion of a floxed Pax5 allelein the same CD19cre-expressing B cells (Horcher et al.,2001; unpubl. data). By analogy, the Pax8cre allele mayalso delete certain floxed genes more efficiently com-pared to the Z/AP transgene, thus reducing the mosa-icism of gene deletion in the developing kidney andmidbrain. Finally, the Pax8cre/� Z/AP system identified

FIG. 2. Cre activity in heterozy-gous Pax8cre/� Z/AP embryos.Embryos were analyzed forhPLAP activity at E8.5 (A), E9.5(B,C), and E10.5 (D–F). Duringthis time interval the presence ofCre activity was successively de-tected in the region of the oticplacode (op), in the otic vesicle(ov), at the mid-hindbrain bound-ary (mhb), and in the mesone-phros (ms) (A–D). A sectionthrough the otic region (E) re-vealed hPLAP activity in the oticvesicle, vestibulocochlear gan-glion (vcg), and geniculate gan-glion (gg). The plane of sectioningis indicated by a dashed line in D.A dorsal view of the tail region (F)indicates that the hPLAP activityin the mesonephros was mostlylocalized in a discontinuous pat-tern in the nephric duct (nd), whilesome clusters of hPLAP� cellwere also detected in the nephro-genic cord (nc). Staining in the ec-toderm close to the otic vesicle isindicated by an asterisk (B). cl,cloaca; teg, tegmentum.

107PAX8CRE KNOCKIN MOUSE

FIG. 3. Cre-mediated hPLAP ex-pression in Pax8cre/� Z/AP em-bryos at E15.5. A: hPLAP activitywas detected in the tegmentum(teg) of the midbrain as well as inthe vestibulocochlear nerve (vcn)and cuneate fascicules (cf) of thehindbrain. The observed hPLAPactivity in the urogenital system(B) reflects Pax8 expression in thekidney (k) and epididymis (ep), aswell as development of the adre-nal gland (a) and ureter (u) fromearlier Pax8 expression domains.The hPLAP expression in the fa-cial nerve (fn) of the tongue (C)results from Pax8 expression inthe geniculate ganglions, whichwas observed at E10.5 (Fig. 2E).D: hPLAP expression in the thy-roid gland (tg). b, bladder; t, tes-tis; tb, taste bud.

FIG. 4. Efficient Cre recombina-tion in homozygous Pax8cre/cre

Z/AP embryos at E16.5. A,B: Kid-ney sections revealed hPLAPstaining already in S-shaped bod-ies (s) of Pax8cre/cre embryos (B),whereas this structure appearedunstained in Pax8cre/� embryos (A).hPLAP activity was readily visible inthe glomeruli (g) and some nephrictubules (tu) of both genotypes.C,D: hPLAP staining detected Creactivity in the entire inner ear of ho-mozygous Pax8cre/cre embryos (D),in contrast to the lower efficiencyof Z/AP activation observed in thesemicircular canals (sc) ofPax8cre/� embryos (C). The Creactivity was similar for both geno-types in the cochlea (co) andother epithelial components ofthe inner ear. es, endolymphaticsac; oc, otic capsule.

108 BOUCHARD ET AL.

hPLAP expression in the cuneate, facial, and vestibulo-cochlear nerves, as well as in the adrenal gland, whichmay correspond to novel Pax8 expression domains or,more likely, are derived from Pax8-expressing cells dur-ing development. Consequently, a potential role of Pax8in the differentiation of these nerves and in adrenal glanddevelopment should be investigated, as the correspond-ing mutant phenotypes may have been missed in previ-ous analyses (Mansouri et al., 1998).

MATERIALS AND METHODS

MiceThe generation of Pax8neo/� mice was previously de-

scribed (Bouchard et al., 2002). The neo selection cas-sette, which was flanked by frt sites in the Pax8neo allele,was deleted in vivo by mating Pax8neo/� mice with theACTB:FLPe transgenic line (Rodriguez et al., 2000), re-sulting in efficient generation of Pax8cre/� mice in the F1generation. The different Pax8 alleles were genotypedby PCR with the following primers: 5�-TCTCCACTCCAA-CATGTCTGC-3� (Pax8 intron 2), 5�-CCCTCCTAGTT-GATTCAGCCC-3� (Pax8 exon 3), and 5�-AGCTGGC-CCAAATGTTGCTGG-3� (cre gene). The wildtype Pax8alleles gave rise to a PCR product of 389 bp, while boththe Pax8neo and Pax8cre alleles generated a 673-bp frag-ment. Z/AP transgenic mice were genotyped by �-galac-tosidase staining of ear punches (Lobe et al., 1999).

Alkaline Phosphatase Staining and HistologicalAnalysis

Alkaline phosphatase staining was performed as de-scribed previously (Lobe et al., 1999). Embryos werestained for 24 h except for E15.5 and E16.5 embryos,whose Pax8-expressing organs were dissected after 24 hand reincubated for another 24 h in BM purple stainingsolution (Roche, Nutley, NJ). Stained embryos or organswere postfixed in cold 4% paraformaldehyde for 1 h,washed in PBS, and equilibrated in 50% and then 80%glycerol prior to photography. Alternatively, fixed em-

bryos or organs were dehydrated in methanol, isopropa-nol, cleared in tetrahydronaphtalene, and photographed.Tissues processed for sectioning were equilibrated in15% and 30% sucrose and frozen in OCT medium (SakuraFine Tek, Torrance, CA). Sections of 10 �m thicknesswere obtained by cryostat sectioning, counterstainedwith eosine in 80% ethanol, dehydrated, and mounted inEntellan medium (Merck, Darmstadt, Germany).

ACKNOWLEDGMENT

We thank C. Lobe and S. Dymecki for providing Z/APand ACTB:FLPe transgenic mice.

LITERATURE CITED

Bouchard M, Souabni A, Mandler M, Neubuser A, Busslinger M. 2002.Nephric lineage specification by Pax2 and Pax8. Genes Dev 16:2958–2970.

Horcher M, Souabni A, Busslinger M. 2001. Pax5/BSAP maintains theidentity of B cells in late B lymphopoiesis. Immunity 14:779–790.

Lewandoski M. 2001. Conditional control of gene expression in themouse. Nat Rev Genet 2:743–755.

Lobe CG, Koop KE, Kreppner W, Lomeli H, Gertsenstein M, Nagy A.1999. Z/AP, a double reporter for Cre-mediated recombination.Dev Biol 208:281–292.

Mansouri A, Chowdhury K, Gruss P. 1998. Follicular cells of thethyroid gland require Pax8 gene function. Nat Genet 19:87–90.

Novak A, Guo C, Yang W, Nagy A, Lobe CG. 2000. Z/EG, a doublereporter mouse line that expresses enhanced green fluorescentprotein upon Cre-mediated excision. genesis 28:147–155.

Paxinos G, Tork I, Tecott LH, Valentino KL. 1991. Atlas of the rat brain.San Diego: Academic Press.

Pfeffer PL, Gerster T, Lun K, Brand M, Busslinger M. 1998. Character-ization of three novel members of the zebrafish Pax2/5/8 family:dependency of Pax5 and Pax8 expression on the Pax2.1 (noi)function. Development 125:3063–3074.

Plachov D, Chowdhury K, Walther C, Simon D, Guenet JL, Gruss P.1990. Pax8, a murine paired box gene expressed in the develop-ing excretory system and thyroid gland. Development 110:643–651.

Rodriguez CI, Buchholz F, Galloway J, Sequerra R, Kasper J, Ayala R,Stewart AF, Dymecki SM. 2000. High-efficiency deleter mice showthat FLPe is an alternative to Cre-loxP. Nat Genet 25:139–140.

Saxen L. 1987. Organogenesis of the kidney. Cambridge, UK: Cam-bridge University Press.

109PAX8CRE KNOCKIN MOUSE