Dosage requirement of Pitx2 for development of multiple organs

The bicoid-like homeobox gene Pitx2 is a member of a multigene family with overlapping and distinctive expression patterns. Multiple members of this gene family have been identified in vertebrates including amphibians, fish, birds and mammals (Gage et al., 1999). Identification of Pitx in Drosophila implied an important evolutionary role for the Pitx gene family, yet Drosophila mutants have no obvious anomalies (Vorbruggen et al., 1997). Recent experiments in chick and frog have indicated a role for Pitx2 downstream of sonic hedgehog and nodal in a genetic pathway regulating laterality of heart, gut and other asymmetric organs (Logan et al., 1998; Piedra et al., 1998; Ryan et al., 1998; Yoshioka et al., 1998). Pitx2 expression is altered in mouse mutants with laterality defects, including iv, inv and a targeted mutant in Lefty-2, suggesting that Pitx2 participates in left-right determination in mammals as well (Meno et al., 1998; Piedra et al., 1998; Ryan et al., 1998).

The importance of each member of the Pitx gene family is emerging through analysis of human patients and targeted mutations in mice. PITX2 mutations cause Rieger syndrome, type I (RGS1, MIM 180500) (Semina et al., 1996), an autosomal dominant haploinsufficiency syndrome with defects in the anterior segment of the eye (cornea, trabecular meshwork and iris stroma), and other features of varying severity including cataracts, glaucoma, missing or misplaced teeth (hypodontia), umbilical abnormalities and cardiac defects (Jorgenson et al., 1978; Semina et al., 1996). PITX3 mutations result in autosomal dominant cataracts and anterior segment mesenchymal dysgenesis (MIM 107250) (Semina et al., 1998). In contrast to PITX2 and PITX3, Pitx1 mutations in mice are recessive, affecting development of the palate, mandible and limbs (Lanctot et al., 1999; Szeto et al., 1999).

Pitx2 is expressed in many tissues during development, including the left lateral plate mesoderm, derivatives of the first branchial arch, the eye, brain, pituitary gland, mandible, heart and limbs (Arakawa et al., 1998; Gage and Camper, 1997; Kitamura et al., 1997; Muccielli et al., 1996; Semina et al., 1996). The phenotypes of RGS patients with null alleles establish an essential role for this gene in eye, tooth and umbilical development, and glaucoma (Amendt et al., 1998). The occasional occurrence of isolated growth insufficiency and aortic arch defects in RGS patients suggests that PITX2 may also play a role in pituitary gland and heart development (Mammi et al., 1998; Sadeghi-Nedjad and Senior, 1974). However, the importance of PITX2 is still enigmatic because of the one remaining functional allele in these heterozygous patients.

We used gene targeting in embryonic stem (ES) cells, combined with cre-loxP technology to generate an allelic series of Pitx2 mutations of varying severity. Heterozygotes for hypomorphic (neo) or null (−) alleles for Pitx2 mimic some aspects of RGS, including eye and tooth defects. The phenotypes of homozygotes for both alleles establish that Pitx2 is essential for positioning and development of the heart and for lung asymmetry. While pituitary and eye development are grossly normal in neo/neo fetuses, −/− mice eye exhibit an early arrest in pituitary development and multiple eye defects. The differing sensitivity of various organs to Pitx2 deficiency demonstrates a tissue-specific dosage-dependence on Pitx2 for both laterality and organogenesis.

A P1 clone containing Pitx2 was obtained from a 129/OLA genomic library by screening with Pitx2 exon 5 PCR primers (Sternberg, 1992; Genomic Systems, Inc.; St. Louis, MO). Since herpes simplex virus thymidine kinase (HSV-TK) interferes with spermatogenesis (Al-Shawi et al., 1991), pFLOXΔRI was generated by deleting the HSV-TK sequences from pFLOX (Orban et al., 1992) by EcoRI digestion and religation. Three contiguous Pitx2 genomic fragments were inserted sequentially into pFLOXΔRI: a 631-bp fragment spanning the homeodomain-encoding exon 4 was placed between loxP2 and loxP3, a 1.7 kb fragment was placed upstream of loxP1 and the phosphoglycerate kinase-neo^R cassette, and a 3.2 kb fragment was placed downstream of loxP3 (Fig. 1).

The targeting vector was linearized with HindIII and electroporated into R1 ES cells (Nagy et al., 1993). G418-resistant colonies were screened by PCR to identify homologous recombinants. The 29 positive clones (3.3%) were rescreened by Southern blot to confirm homologous recombination 5′ and 3′ of the neo^R cassette, and by PCR to demonstrate the incorporation of loxP3. Two cell lines heterozygous for Pitx2^neo (6A.C10 and 8A.A9) were injected into C57BL/6J blastocysts to generate chimeric mice, which were mated to C57BL/6J females to transmit the neo allele.

The two neo/+ cell lines were also transiently transfected with a cre expression vector (pMC-Cre, (Gu et al., 1993)) to derive the null (−) and floxed alleles by cre-mediated site-specific recombination. The +/− and +/flox cell lines were identified by PCR (Fig 1). Two +/− cell lines were injected into C57BL/6J blastocysts; resulting chimeric founder males were bred to C57BL/6J females to transmit the null allele. Heterozygotes of the N1-N3 generations were mated to generate timed pregnancies. The morning after mating was designated e0.5.

Experiments with mice and recombinant DNA were approved by university committees.

Total RNA was isolated from embryos or cell line pellets using TRIzol reagent (Gibco; Gaithersburg, MD) (Gage and Camper, 1997). PCR was performed on first-strand synthesis product using Pitx2-specific primers located in exons 1C and 4.

Embryos were flash frozen or fixed overnight in 4% phosphate-buffered saline (pH 7.2). Fixed embryos were embedded in ParaPlast (Oxford Labware; St Louis, MO). Sections were stained in Hematoxylin (Fisher; Fair Lawn, NJ) and Eosin (Sigma; St. Louis, MO). In situ histochemistry was previously described (Schaeren-Weimers and Gerfin-Moser, 1993). Plasmid templates for Lhx3 (from H. Westphal) and Rpx (officially designated as Hesx1, from K. Mahon) were used as described (Hermesz et al., 1996; Zhadanov et al., 1995).

Analysis of Pitx2 mRNA transcripts reveals the potential for three different protein isoforms of wild-type Pitx2 that arise through the use of alternative promoters and splicing (Fig. 1A) (Gage et al., 1999). Pitx2a and Pitx2b use an upstream promoter and differ from each other by alternative splicing of exon 3. Pitx2c utilizes a downstream promoter located in intron 3. Each Pitx2 isoform would contain the homeodomain, encoded primarily in exon 4, and have identical C termini but different N termini (Gage et al., 1999).

We used gene targeting in ES cells and cre-lox technology (Gu et al., 1994) to construct two Pitx2 alleles that differ in severity. Gene targeting was performed with a vector engineered to contain loxP sites flanking the critical exon 4 and a neomycin resistance selectable marker cassette (neo^R) flanked by loxP sites and inserted into intron 4 (Fig. 1A). The Pitx2^neo allele was generated by homologous recombination after electroporation of the targeting vector into ES cells. Transfection of these clones with a cre expression vector produced the Pitx2^null and Pitx2^flox alleles. Clones heterozygous for Pix2^null (+/−) and Pitx2^flox (flox/+) were identified by PCR (Fig. 1C). Deletion of both the neo^Rcassette and exon 4 via recombination at sites P1 and P3 produced a presumptive null allele due to loss of the essential homeodomain and a frameshift leading to a series of missense codons and a premature stop codon. Deletion of only the neo^R cassette, by recombination between lox P1 and P2, produced Pitx2^flox, an allele that should be fully functional and conditionally converted to Pitx2^null in the presence of tissue-specific or temporally regulated cre transgenes.

Two neo/+ clones and two +/− clones were injected into C57BL/6J blastocysts to generate chimeric mice. Chimeric males were bred to C57BL/6J females to transmit the modified Pitx2 alleles, and independent mouse colonies derived from each injected cell line were established. Progeny were genotyped by PCR (Fig 1C). Heterozygotes for both alleles exhibited normal fertility. The phenotypes associated with each allele were replicated in two independent lines.

Insertion of a neo^R cassette into an intron by gene targeting can alter or disrupt gene expression by virtue of the cryptic splice sites and polyadenylation signals in the cassette, leading to hypomorphic or null alleles (Meyers et al., 1998). neo/+ mice were intercrossed to test for viable progeny. No neo/neo mice were identified in 334 two-week old progeny (Table 1), demonstrating that the neo allele is not functioning normally. Genetic analysis of timed pregnant litters from e9.5 to p1 revealed that individual homozygotes die throughout gestation beginning at e10.5. The loss of viable embryos is significant at e18.5 when only 2.4% (1/41) of the fetuses were neo homozygotes instead of the expected 25% (P<0.01). Death this late in gestation is usually attributable to structural cardiovascular defects or hematopoietic problems (Copp, 1995).

Normal Pitx2 transcripts, produced by splicing from exon 1A to exon 4, were detected in mRNA from neo/neo embryos by RT-PCR (Fig. 2A). This observation and the more severe effects of the null allele (see below) demonstrate that the neo allele is a hypomorph, rather than a null (Table 2).

The RGS phenotype can be highly variable, even within families. However, all patients exhibit congenital defects of the anterior segment of the eye. Abnormalities consistent with RGS1 were evident. These included corectopia (asymmetrically placed pupils), anisocoria (uneven pupil size), multiple pupillary openings and clouded lenses (Fig. 3). Like RGS in humans, these features were not present in all heterozygotes for the neo or null (−) allele (∼10%). The majority of the +/− mice (n>30) had grossly normal teeth and body size, however two +/− males were approximately 1/3 normal body size, had mal-occluded incisors and obviously small eyes.

Examination of −/− embryos revealed a broad requirement for Pitx2 during eye development (Fig. 4A,B,D,G). Pitx2 is expressed in the perioptic mesenchyme by e9.5, but it is not expressed in the optic stalk or eminence. At e12.5, +/+ and neo/neo eyes are located immediately beneath the surface with the rim of the optic cup in contact with the developing cornea. In contrast, −/− eyes are smaller than normal and lie below the surface epithelium (Fig. 4C,E,H). They are rotated toward the nose so that the lens does not directly face the surface ectoderm. The developing corneal epithelium is thicker in −/− mice, and the optic cup is separated from the surface ectoderm by an abnormally thick mesenchymal layer (Fig. 4E,H). We confirmed the increased thickness of corneal epithelium by staining with a Pax6 antibody (data not shown). Vestigial lens pits are present in some −/− eyes at e12.5 (Fig. 4H), while lens pit closure is complete in +/+ eyes by e11.

Mesenchymal condensations leading to extrinsic eye muscle formation are clearly evident in wild-type eyes but absent in −/− embryos at e12.5 (Fig. 4D-I). Muscles form in neo/neo eyes but they are reduced in size (data not shown). Closure of the optic fissure is also defective or significantly delayed in the −/− embryos. No other gross defects were noted in the development of the neo/neo eyes or in the neural retina and retinal pigmented epithelium of −/− eyes.

The anterior and intermediate lobes of the pituitary gland derive from Rathke’s pouch, an invagination of oral ectoderm within the roof of the mouth that begins at e8.5. Sequentially, the early steps are formation of the pouch rudiment, development of the committed pouch and pouch expansion (Sheng et al., 1996). Pitx2 is broadly expressed early in Rathke’s pouch, including in the rudiment (Gage and Camper, 1997; Muccielli et al., 1996). The Pitx2a and Pitx2b mRNA isoforms are expressed in the anterior pituitary cells of the thyrotrope, somatotrope, lactotrope and gonadotrope lineages, but not corticotropes (Gage and Camper, 1997). The Pitx2c isoform is expressed in all five anterior pituitary cell lineages, including corticotropes (Fig. 2B). Examination of −/− embryos revealed a committed pouch at e10.5, indicating that Pitx2 is not required for the initial stages of pituitary development (Fig. 5A,B). However, the mesenchymal layer surrounding the pouch was clearly hypercellular at this time and comparison of serial sagittal sections spanning the midline from +/+ and −/− littermates at e10.5 demonstrated that the pouch was smaller. Arrest of growth and differentiation is clearly evident in −/− pouches by e12.5. The wild-type pouch wall is thicker and cell proliferation on the ventral side of the pouch has formed the nascent anterior lobe. In contrast, the ventral and dorsal aspects of the −/− pouch wall remain relatively thin (Fig. 5D). This indicates that Pitx2 is required for pituitary gland development shortly after formation of the committed pouch. No changes were noted in the nascent posterior pituitary gland. Pituitary glands from neo/neo embryos were morphologically normal at e16.5 and e18.5.

Hesx1 (Rpx) and Lhx3 are homeobox genes that are required early during early pituitary development (Hermesz et al., 1996; Sheng et al., 1996). Pitx2 is expressed before Hesx1 and Lhx3 during pouch development, making them candidates for downstream targets of Pitx2. Hesx1 expression was detectable throughout the pouch in +/+ embryos at e12.5, but no transcripts were evident in −/− littermates (Fig. 5G,H). This indicated that Pitx2 is a regulator of Hesx1 expression during pituitary development. In contrast, similar levels of Lhx3 expression were detected in both +/+ and −/− pouches, demonstrating that Lhx3 expression is not dependent on Pitx2 (Fig. 5I,J). Essential inductive interactions occur between Rathke’s pouch and the base of the diencephalon early during pituitary development, including induction of pro-opiomelanocortin (Pomc) in the diencephalon (Diakoku et al., 1982; Sheng et al., 1996). We demonstrated that Pomc expression in the diencephalon of +/+ versus −/− embryos is indistinguishable, indicating that some inductive interactions are intact in −/− mice (Fig. 5K,L).

Neo/neo embryos are indistinguishable externally from wild type (Fig. 6A,B). The −/− embryos are moderately undersized relative to their heterozygous or wild-type littermates, although there were no signs of edema (Fig. 6C,D). All the mutant embryos bend severely rightward rostral to the developing hindlimb. In some embryos (∼5%), the crown of the head is open over the midbrain. Other brain defects were also varied but included excess mesenchyme and abnormalities of the expansion of the pontine and midbrain flexures. Rarely, embryos lacked the left forelimb (∼5%, Fig. 6J), mandibular arch, tongue and/or aspects of the caudal region.

Individuals with RGS1 have phenotypes consistent with variable defects in development of the ventral body wall. Failure of umbilical cord involution ranges in severity from a protruding navel to omphalocele (umbilical hernia) requiring surgical repair (Jorgenson et al., 1978). In wild-type mice, the thoracic and abdominal organs are normally internalized by e12.5 (Fig. 6A). Remarkably, the heart, liver and other abdominal organs are almost universally external and displaced profoundly leftward and outward in −/− embryos at e12.5 due to a large fissure in the ventral body wall (Fig. 6C,D). This fissure results from incomplete ventral closure and is generally displaced towards the left of the midline. The open ventral walls thicken as they extend outwardly and dorsally (Fig. 6L,P). They remain contiguous with the amnion, similar to their relationship during and shortly after turning of the embryo. The heart and liver are usually within thin membranes, presumably the remnants of the pericardial sac and liver capsule, respectively (Fig. 6K). Thus, the phenotype of −/− mice appears to be a more severe expression of the omphalocele characteristic of some RGS patients.

Hearts of −/− embryos are dramatically enlarged by e12.5 and continue to increase in size (Fig. 6 G,H). A septal ridge is clearly visible extending across the ventricle, indicating that septation has at least initiated. However, even as late as e12.5-13.5 only a single, hugely distended atrium can be identified. Contraction of this single atrium injects blood into both ventricles. Blood flows from the atrium into the ventricles between contractions and atrial contraction can inject blood backward into the vena cava.

Several organs of the body exhibit left-right asymmetry, including the lungs, heart and stomach. The heart begins as a symmetrical, straight tube and initates asymmetry by looping to the right in an ‘S-shaped’ structure. The atria and inflow tracts at the caudal end of the tube move dorsally and anteriorly behind the ventricular regions. Right looping is initated properly in both neo/neo and −/− embryos and the asymmetry appears grossly normal at later stages (Fig. 7A-H). In rodents, the left lung is a single lobe but the right lung consists of four lobes. Both neo/neo and −/− embryos exhibit right-isomerization of the lungs (Fig. 7I-L). Both the right and left lungs were mirror images with multiple lobes normally characteristic of the right lung. The stomach was positioned on the left, as expected, in both neo/neo and −/− mutants (data not shown). Positioning of the tail and umbilicus over the right flank indicated that mutant embryos turn in the normal direction (data not shown). However, the orientation of the hindlimbs is sometimes in the opposite direction to the forelimbs, suggesting that the process of turning is not always complete.

The extremely abnormal heart morphology indicated that −/− embryos probably died of heart defects. A massive proliferation of mesoderm cells has displaced the mutant heart ventrally and leftward at e12.5 which, combined with the deficit in ventral body wall closure, results in its grossly external position (Fig. 8A,D). There is a single, distended atrium located to the left of the ventricles (Fig. 8E). Ventricular septation in −/− hearts is dysmorphic and there is little evidence of atrial septation. In contrast, +/+ hearts have distinct left and right atria developing on either side of the ventricles and ventricular septation is complete (Fig. 8B). Both the leaflets of the vena cava and atrial-ventricular (A-V) valve formation are defective in −/− embryos (Fig. 8E,F). In the case of the A-V valve, this appears to result from a reduction in the membranous component of the valve. Although the hearts of neo/neo animals were located within the thoracic cavity (Fig. 8I), they exhibited ventricular septation defects possibly resulting from a reduction in the membranous component (Fig. 8J). The structural defects of the heart are likely to account for the death of both the neo/neo and −/− embyros.

Malpositioning of the heart is evident in homozygotes for both alleles. The position of the hearts in neo/neo embryos was biased toward the right and the ventricular septum appeared reversed relative to wild-type hearts (compare Fig. 8I,J with 8G,H). The isomerization of the lungs results in a larger than normal left lung. This may contribute to the apparent shift in position of the heart, as heart looping initiated properly. While the heart was placed on the left in −/− embryos, its orientation was off by approximately 90°. The size of the left lung and the mass of mesoderm cells on the left could contribute to this rotation.

An allelic series ranging from partial to complete loss of function can be especially valuable in assessing gene function (Greco et al., 1996). The embryonic lethality frequently resulting from complete loss-of-function mutations in homeobox genes makes the availability of milder alleles highly desirable and relevant as models for human birth defects. We have generated an allelic series for the murine Pitx2 gene (Table 2). Two alleles differ in severity and the floxed allele has the potential for tissue-specific and temporally induced disruption of Pitx2 function. The constellation of phenotypes produced in heterozygotes indicate that these mice are a model for Rieger syndrome, type I. Pitx2 is essential for many aspects of normal development, with striking roles in eye, pituitary gland, heart and lung ontogeny, including developing functional heart valves and establishing the asymmetry of the lungs.

The congenital eye defects of RGS patients indicate an essential role for PITX2 in development of the anterior segment of the eye. In −/− embryos, the anterior segment is abnormally thick, due to an increase in both the epithelial and mesenchymal components of the cornea. It is striking that both layers are affected since Pitx2 is not expressed in the corneal epithelium (Semina et al., 1996). RGS patients exhibit corneal opacity. Therefore, the changes observed in −/− mice are consistent with the defects seen in RGS patients. The two posterior eye defects, lack of ocular muscles and optic nerve coloboma, were unexpected because RGS involves anterior segment abnormalities. However, the majority of RGS patients develop early onset glaucoma, a progressive loss of retinal ganglion cells and degeneration of the optic nerve head (Chang et al., 1999). The optic fissure defects in −/− embryos suggest that subtle optic nerve defects in heterozygotes might result in glaucoma as the animals age, providing another parallel with RGS patients.

Only a small fraction of adult +/neo or +/− mice exhibited anterior chamber defects of the eye, consistent with the remarkable variability in RGS and septo-optic dysplasia (Hesx1 MIM #182230) (Dattani et al., 1998). In contrast to the variability in heterozygotes, profound disturbances in eye development were observed consistently in the absence of Pitx2. The dosage sensitivity of Pitx2 may contribute to the variability of the phenotype. Inbred strains of mice and strains with sensitizing mutations in Pax6, Chx10 and Trp1 have been used to demonstrate the effects of genetic background on eye development and glaucoma (Burmeister et al., 1996; Chang et al., 1999; Glaser et al., 1999). Modifiers of PITX2 may be responsible for unlinked cases of RGS in addition to other

eye abnormalities and diseases such as cataracts and glaucoma. Thus, the Pitx2 mutants we have generated could be valuable for identifying other genes important for eye development and disease.

A genetic hierarchy of homeobox genes, including Hesx1, Lhx3, Prop1 and Pit1, is essential for pituitary gland development. (Dattani et al., 1998; Gage et al., 1996; Sheng et al., 1996; Sornson et al., 1996). Loss of Hesx1, Lhx3 and Prop1 results in dysmorphology of the pituitary primordium. Pit1 is required later during pituitary development for specification of three anterior pituitary cell types (Camper et al., 1990; Li et al., 1990). Molecular genetic analyses suggest that some of the genes act sequentially in pituitary development: Hesx1→Prop1→Pit1 (Anderson et al., 1995; Gage et al., 1996; Sornson et al., 1996). Our demonstration of altered Hesx1 expression in −/− mice places Pitx2 early in this pathway. Lhx3 and Pitx2 affect pouch development similarly and both genes are required for Hesx1 expression. Lhx3 expression does not require Pitx2. Thus, Pitx2 is an early regulator of pituitary ontogeny that acts with Hesx1 and Lhx3, prior to the action of Prop1 and Pit1 to advance development of the committed Rathke pouch.

The expression patterns of Pitx1 and Pitx2 overlap in the developing pituitary and they have nearly identical homeodomains, suggesting functional redundancy (Lamonerie et al., 1996; Lanctot et al., 1997; Muccielli et al., 1996; Szeto et al., 1996; Tremblay et al., 1998). Pitx1^−/− fetuses have normal pituitary morphology and only modest quantitative changes in three of five pituitary cell types (Szeto et al., 1999). In contrast, development is arrested at the committed pouch stage in Pitx2^−/− fetuses. The relative roles of Pitx1 and Pitx2 may be similar to the relative roles of Lhx4 and Lhx3. Loss of Lhx4 alone results in a relatively mild pituitary phenotype, similar to Pitx1 null mice (Sheng et al., 1997), but loss of Lhx3 has an early and dramatic effect on pouch development, like Pitx2. Loss of both Lhx3 and Lhx4 affected pouch development more profoundly, indicating a partial ability of these two genes to compensate for each other (Sheng et al., 1997). Analysis of Pitx1, Pitx2 double mutants will be necessary to reveal any functional overlap between these two genes.

Prior reports of asymmetric expression of Pitx2 in early mouse development, before the onset of visceral organogenesis, suggested it might have a role in development of left-right asymmetry (Logan et al., 1998; Piedra et al., 1998; Ryan et al., 1998; Yoshioka et al., 1998). Its expression is perturbed in mouse mutants with reversals of the heart and visceral organs (Meno et al., 1998; Piedra et al., 1998; Ryan et al., 1998). Manipulation of Pitx2 expression was sufficient to reprogram the left/right axis in chick and frog, providing further circumstantial evidence for an important role. In these experiments, Pitx2 was expressed ectopically on the right side and an early player in the pathway was blocked (sonic hedgehog, Shh), preventing activation of Pitx2 on the left. Isomerized or reversed development of the heart and gastrointestinal tract resulted. The defects in the ventral body wall that caused externalization of the viscera in −/− embryos revealed the importance of Pitx2 at the time the embryo turns. However, the right-ward looping of the heart initiated properly and the stomach was located on the left in both neo/neo and −/− fetuses, indicating that Pitx2 is not required to initiate the left-right asymmetry of the heart or the gastrointestinal tract. However, the lungs were right-isomerized, suggesting that Pitx2 is required for establishing the lobular pattern characteristic of the left lung. Thus, there is a dramatic difference between ectopic expression of Pitx2, and absent or low levels of Pitx2 in establishing the left-right asymmetry of the body axis. Ectopic expression of Pitx2 is sufficient to reprogram the left-right body axis, but it is only required for developing the asymmetry of the lungs. This suggests that there are other genes that partially compensate for the lack of Pitx2.

The defects that we observed in homozygous mutant mice clarify some questions about the Rieger syndrome phenotype. Retrospective studies have demonstrated that defects of the heart and umbilicus are more common than previously appreciated (Jorgenson et al., 1978; Mammi et al., 1998). However, the possibility existed that these features resulted from a contiguous gene syndrome. This is unlikely because reduced Pitx2 levels cause heart defects similar to those found in Rieger patients and failure of the ventral body wall to close that is consistent with omphalocele requiring surgical repair.

The concentration of Pitx2 is important for normal development. This is evident by comparing the mild eye phenotypes of adult +/− and +/neo mice with the severe eye defects in −/− embryos and by comparing the heart development of neo/neo and −/− fetuses. The sensitivity of each organ to Pitx2 levels is variable. The eye and heart appear to be more sensitive than the pituitary, for example. Pitx1 is expressed in the pituitary but not the eye or heart, thus functional redundancy may account for some of the differential sensitivity of each organ. In conjunction with tissue-specific, developmentally regulated, and/or inducible cre transgenes, the Pitx2^flox allele will allow spatial and temporal requirements of Pitx2 function to be dissected.

We thank the following: S. Lindert, P. Gillespie, T. Saunders and the University of Michigan Transgenic Animal Core for assistance with generation and propogation of the mice; T. Glaser, N. Brown and J. Lauderdale for discussions of eye development and Pax6 antibodies, G. Dressler for Pax2, A. Seasholtz for pituitary cell line RNA, and D. M. Martin and S. O’Shea for helpful discussions. This work was supported by The Glaucoma Research Foundation (PJG) and NICHD (34283 and 30428, SAC).