The canonical Wnt signaling antagonist DKK2 is an essential effector of PITX2 function during normal eye development

Abstract

Local control of cell signaling activity and integration of inputs from multiple signaling pathways are central for normal development but the underlying mechanisms remain poorly understood. Here we show that Dkk2, encoding an antagonist of canonical Wnt signaling, is an essential downstream target of the PITX2 homeodomain transcription factor in neural crest during eye development. Canonical Wnt signaling is ectopically activated in central ocular surface ectoderm and underlying mesenchyme in Pitx2- and Dkk2-deficient mice. General ocular surface ectoderm identity is maintained during development in Dkk2-deficient mice but peripheral fates, including conjunctival goblet cells and eyelash follicles, are ectopically permitted within more central structures and eyelids are hypomorphic. Loss of DKK2 results in ectopic blood vessels within the periocular mesenchyme and PITX2 expression remains persistently high, providing evidence for a negative feedback loop. Collectively, these data suggest that activation of Dkk2 by PITX2 provides a mechanism to locally suppress canonical Wnt signaling activity during eye development, a paradigm that may be a model for achieving local or transient inhibition of pathway activity elsewhere during embryogenesis. We further propose a model placing PITX2 as an essential integration node between retinoic acid and canonical Wnt signaling during eye development.

Figure 1. Dkk2 is regulated by Pitx2 gene dose and suppresses canonical Wnt signaling in the developing eye

(A) Dkk2 expression levels correlate strongly with Pitx2 gene dose in TaqMan Gene expression assays performed on total RNA isolated from microdissected eye primordia of e12.5 Pitx2 embryos of the indicated genotypes. Values are derived from N=4 samples/genotype. (B) RNA in situ hybridization to detect Pitx2 or Dkk2 expression in e12.5 wild type and Pitx2 mutant eyes. Dkk2 (i) and Pitx2 (ii) are both expressed in periocular mesenchyme of wild type eyes but Dkk2 is completely undetectable in eyes of global (iii) or neural crest-specific (iv) Pitx2 knockout embryos. (C) Axin2 expression by RNA in situ hybridization in eyes from e15.5 wild type and Dkk2^−/− littermates. Axin2 is highly upregulated in ocular surface ectoderm (open arrowheads) and underlying mesenchyme (m) within the iridocorneal angle (ii) and central cornea (iv) of mutant eyes. (D) Axin2 expression by RNA in situ hybridization in eyes from e13.5 wild type and Pitx2^−/− littermates. Axin2 expression ceases at the border between conjunctival and corneal ectoderm in wild type eyes (insert i, closed arrowhead) but extends throughout the ocular surface ectoderm in Pitx2−/− eyes (ii, open arrowheads). Key: L, lens; R, retina; POM, periocular mesenchyme; RPE, retinal pigmented epithelium; *, presumptive cornea.

Figure 2. PITX2 physically associates with Dkk2 promoter sequences and transactivates a linked reporter

(A) Schematic of Dkk2 genomic sequences spanning from −3,500 to +500 base pairs from the transcription start site illustrating position of potential PITX2 binding sites and regions amplified by Dkk2 primer sets A and B. (B) ChIP analysis on primary cultures of mesenchyme grown from wild type e12.5 eye primordia with an anti-PITX2 antibody showed that PITX2 specifically interacts with elements of the Dkk2 promoter but not unrelated sequences. (C) Reporter assay using a 4-kb Dkk2 fragment including 3.5-kb of promoter sequences and 0.5-kb of untranslated 1^st exon sequences linked to luciferase. CHO cells were transfected with the reporter plasmid and an empty expression vector or vector encoding one of the three PITX2 isoforms. Error bars represent s.e.m.

Figure 3. Prenatal Dkk2 mutant eyes are defective in ocular anterior segment development

Coronal JB-4 plastic (A-F) and paraffin (G,H) sections taken from e18.5 littermates with genotypes as marked. (A,B) Eyelids of Dkk2^null/null mutant embryos are hypomorphic compared to wild type littermate controls, with the dorsal lid (D) always more severely affected than the ventral lid (V). (C,D) Dkk2 mutant embryos have frequent ectopic eyelash follicles within the presumptive conjunctiva and limbus (arrowheads). (E,F) High magnification central cornea images taken from boxed regions in panels A&B, respectively. The distinctive wavy morphology and highly ordered lamellar organization of emerging keratinocytes in the wild type corneal stroma is replaced by more stellate-shaped cells and loss of regular organization in presumptive stroma cells of Dkk2 mutant eyes, and the mutant corneal epithelium is moderately thickened. (G,H) VEGFR2 (red) and VEFGR3 (green) immunofluorescence highlights the avascular corneal stroma of the wild type littermate compared to the presence of ectopic blood vessels penetrating the anterior corneal stroma of Dkk2 mutant eyes. Note: for examples from each genotype, the dashed lines outline the inner and outer curvature of the cornea. Key: L, lens; R, retina; box, central cornea; E, eyelids; Ep, corneal epithelium; S, corneal stroma; En, corneal endothelium. All images are oriented with dorsal to the left.

Figure 4. Postnatal Dkk2 mutant eyes have additional defects

All samples are of 3-week-old eyes, except where noted. Position of normal eyelashes in wild type littermate (A) compared to ectopic eyelashes lying across cornea in Dkk2^null/null eye (B). Note also periorbital edema and ptosis in mutant eye. (C) Dkk2^null/null eyes have iridocorneal attachments (box and inset). (D) Histologic section of Dkk2^null/null eye showing iridocorneal attachments consisting of ectopic small blood vessels spanning anterior chamber from iris collarette (i) to central corneal endothelium (e). Keratinocytes in a mature wild type corneal stroma (E) have a characteristic appearance and are highly organized while in corneas of Dkk2^null/null mice (F) the presumptive stroma cells (S) are disorganized and numerous small blood vessels (<) populate the anterior corneal stroma. In addition, the morphology of mutant corneal epithelium cells (Ep) is altered relative to wild type and apparent goblet cells are present. (G) Periodic acid schiff staining confirms the presence of mucin-containing goblet cells in Dkk2^null/null corneas. (H) Fully developed ectopic eyelash cilia are present in the limbus and conjunctiva of Dkk2^null/null eyes. (I) Ectopic pigmented sheath extending onto optic nerve (open arrowhead) of a Dkk2^null/null eye at 12 weeks of age. Wild type control littermate eye is on left. (J) Plastic section of ectopic sheath ([]) and optic nerve ((Evans and Gage, 2005)) showing small capillaries and associated pigment cells, and continuity with the choroidal capillaris.

Figure 5. Expression of lineage-specific markers shows prenatal bias of corneal surface ectoderm towards a conjunctival fate but maintenance of an overall OSE identity in Dkk2 mutant eyes

(A) Expression of the lineage-specific markers cytokeratin 12 (corneal epithelium) and cytokeratin 4 (conjunctival epithelium) in eyes from e18.5 wild type control and Dkk2^−/− littermates by immunofluorescence. (B) Expression of the general OSE marker PAX6 in e16.5 wild type control and Dkk2^−/−mutant littermate eyes by immunofluorescence. Magnified central corneas from wild type (iii) and Dkk2 mutant (iv) eyes (boxed regions in panels i & ii, respectively). Nuclear PAX6 expression, a signature of all ocular surface ectoderm, is maintained in all components of the ocular surface ectoderm in Dkk2 mutant eyes (e.g. inset in panel ii). Note: non-nuclear epidermis label (*) is non-specific background. Key: Ep, corneal epithelium; S, corneal stroma; En, corneal endothelium; Le, anterior lens epithelium (not present in panel ii). (C) Expression of mature epidermis markers Filaggrin and Loricin in expression in eyes from e18.5 wild type control and Dkk2^−/− littermates by immunofluorescence.

Figure 6. Altered FOXC protein expression in wild type versus Dkk2 mutant eyes

Detection of FOXC1 (A) and FOXC2 (B) in e15.5 wild type control (i) and Dkk2 mutant (ii) littermate eyes by immunofluorescent histochemistry. For each, boxed areas in panels i and iii are enlarged in panels ii and iv, respectively. Iridocorneal angles (*) are marked in each low power view while the curvature of dorsal conjunctival ectoderm (arrowheads) is marked in each high power view. Note: weak signal from e.g. eyelid epidermis and lens is a background characteristic of this FOXC2 antibody that does not correspond with the RNA in situ hybridization pattern (data not shown).

Figure 7. Persistent PITX2 expression in Dkk2 mutant eyes

(A,B) PITX2 immunofluorescence intensity is indistinguishable in e12.5 wild type (control) vs. Dkk2^−/= eyes. (C,D) PITX2 immunofluorescence in e16.5 wild type (control) vs. Dkk2^−/− (mutant) eyes. Staining intensity in the presumptive iridocorneal angles (*) is indistinguishable between the two genotypes. (E,F) Enlargements of boxed areas in C and D, respectively. High-level PITX2 staining intensity persists in the differentiating corneal stroma (S) of mutant but not wild type eyes. Staining intensity in the corneal endothelium (><) is not noticeably different between the two genotypes. Note: Sections were photographed with invariant camera settings and exposure times to facilitate comparison of relative PITX2 expression levels.

Figure 8. Model for PITX2-dependent integration of retinoic acid and canonical Wnt signaling during ocular anterior segment development

Pitx2 expression requires retinoic acid produced by the surface ectoderm, optic cup, and lens (Matt et al., 2005). Dkk2 expression in neural crest is activated downstream of Pitx2 and DKK2 acts to locally suppress canonical Wnt signaling in the neural crest and overlying ocular surface ectoderm. The persistently elevated PITX2 protein levels in Dkk2-deficient mice are consistent with a role for canonical Wnt signaling in contributing to net Pitx2 expression levels, either through stimulation of Pitx2 transcription (Kioussi et al., 2002) or enhancement of mRNA stability (Briata et al., 2003). Several Wnt genes are highly expressed in the overlying ocular surface ectoderm (Liu et al., 2003), suggesting this as the likely source of the Wnt ligands.