Eyeless/Pax6 initiates eye formation non-autonomously from the peripodial epithelium

Abstract

The transcription factor Pax6 is considered the master control gene for eye formation because (1) it is present within the genomes and retina/lens of all animals with a visual system; (2) severe retinal defects accompany its loss; (3) Pax6 genes have the ability to substitute for one another across the animal kingdom; and (4) Pax6 genes are capable of inducing ectopic eye/lens in flies and mammals. Many roles of Pax6 were first elucidated in Drosophila through studies of the gene eyeless (ey), which controls both growth of the entire eye-antennal imaginal disc and fate specification of the eye. We show that Ey also plays a surprising role within cells of the peripodial epithelium to control pattern formation. It regulates the expression of decapentaplegic (dpp), which is required for initiation of the morphogenetic furrow in the eye itself. Loss of Ey within the peripodial epithelium leads to the loss of dpp expression within the eye, failure of the furrow to initiate, and abrogation of retinal development. These findings reveal an unexpected mechanism for how Pax6 controls eye development in Drosophila.

Keywords: Dorsal-ventral patterning; Drosophila; Eye; Eyeless; Morphogenetic furrow; Pax6; Peripodial epithelium; Retina; Twin of Eyeless.

Fig. 1.

eyeless loss-of-function mutant flies have variable defects in eye development. (A) Wild-type adult head. (B) ey² mutant head lacking compound eyes. (C) ey² mutant with a small eye. (D) Percentage of flies that lack compound eyes. Error bars indicate s.d. An unpaired t-test between ey^LB and ey^LB/ey^J5.71 yields a P-value of 0.0998 (not statistically significant). Sample numbers: wild type=100, ey²=94, ey^LB=257, ey^LB/ey^J5.71=489. (E) Average number of ommatidia in wild-type and ey mutant flies. The right eyes of female flies were examined. Sample numbers: wild type=3, ey²=30, ey^LB=30. An unpaired t-test was used in all pairwise comparisons. (F) ey transcripts in wild-type and ey mutant larvae expressed relative to wild type. Error bars represent s.e.m. (G) Sequence of enhancer element that has been deleted in ey^LB. (H) Schematic of the ey² and ey^LB mutants. Blue boxes, coding exons. (I-K) Wild-type (I) and ey^LB (J,K) third larval instar eye-antennal discs. The green stain within the Bolwig's nerve in J and K is due to a 3PXP3-dsRED transgene that is associated with the CRISPR plasmid that was used to create the ey^LB allele. Anterior is to the right. wt, wild type. Scale bars: 50 µm.

Fig. 2.

Toy partially substitutes for Eyeless during eye development. (A,B) Third larval instar wild-type (A) and ey^LB (B) eye-antennal discs. (C) qRT-PCR analysis of toy transcript levels in wild-type and ey mutant retinas expressed relative to wild type. n=3. Error bars were determined by REST analysis (see Materials and Methods). (D-G) Examples of the medium-sized eye (E), small eye (F) and no-eyed (G) phenotypes observed in ey^LB mutants compared with w¹¹¹⁸ (D). (H) Eye sizes in the ey^LB mutant alone (n=168) and in ey^LB mutants in which toy has been overexpressed using ey-GAL4 (n=232). (I-N) Wing (I-K) and leg (L-N) discs in which either GFP (I,L) or toy (J,K,M,N) is being driven along the A/P axis by dpp-GAL4. wt, wild type. Scale bars: 50 µm.

Fig. 3.

The stoichiometry between core RD network members is essential for supporting eye development. (A,B) so (A) and eya (B) transcript levels in control and ey mutant eye-antennal discs expressed relative to wild type. n=3. Error bars were determined by REST analysis (see Materials and Methods). (C-F) Control and ey^LB eye-antennal imaginal discs. Anterior is to the right. Scale bars: 50 µm. (G) Simultaneous removal of one copy of ey and either so or eya results in adults with small, disorganized compound eyes. 30 animals were analyzed for each genotype. (H) Removal of one copy of so, eya or both together in an ey^LB mutant synergistically increases the percentage of flies that lack the compound eyes. Sample sizes: w¹¹¹⁸=30, ey²/ey²=94, ey^LB/ey^LB=257, eya²/+; ey^LB/ey^LB=146, so¹/+; ey^LB/ey^LB=51, so¹ eya²/+; ey^LB/ey^LB=36.

Fig. 4.

Mis-positioning of the D/V midline is correlated with a smaller eye field in ey^LB mutants. (A) Control third larval instar eye-antennal disc showing the expression pattern of the dorsal selector gene mirr. (B) Control adult fly that contains mirr-lacZ transgene. (C) Control adult eye. (D) Retinal section of a control adult eye. (E) ey^LB third larval instar eye-antennal disc. (F) ey^LB adult with a small eye. (G) ey^LB adult with a small eye. (H) ey^LB retinal section. The yellow arrows mark the two unit eyes that have ventral chirality. (I-N) Control (I-K) or ey^LB (L-N) third larval instar eye-antennal discs. Red arrows point to the optic stalk, green arrows show the position of midline markers in control discs and blue arrows indicate shifted or lost midline expression. Anterior is to the right. Scale bars: 50 µm.

Fig.5.

Loss of dpp expression is the underlying cause of the no-eyed phenotype of ey^LB mutants. (A-H) Control (A,B,E,F) and ey^LB mutant (C,D,G,H) eye-antennal imaginal discs. Anterior is to the right. Scale bars: 50 µm. (I) Distribution of eye sizes in ey^LB mutants (n=209) and ey^LB mutants in which dpp has been restored to the posterior margin (n=77). wt, wild type.

Fig. 6.

Removal of ey using RNAi recapitulates the ey^LB mutant phenotype. (A-C) ey-GAL4>UAS-ey RNAi third larval instar eye-antennal discs. (D,E) ey-GAL4>UAS-ey RNAi adult heads showing examples of a fly with small eyes (D) and a fly lacking eyes (E). Anterior is to the right. Scale bars: 50 µm. (F) Distribution of eye sizes when ey is removed using RNAi (n=279). Error bars indicate s.d.

Fig. 7.

Ey is required in the peripodial epithelium for proper development of the neural retina. (A) Third instar eye-antennal disc showing that Ey protein is present in cells of the peripodial epithelium (green arrow). (B,C) High magnification views of the peripodial epithelia of eye-antennal discs that carry either ey-GAL4 (B) or c311-GAL4 (C) drivers. Scale bars: 50 µm. (D,E) Orthogonal views of eye-antennal discs that carry either ey-GAL4 (D) or c311-GAL4 (E) drivers. Anterior is to the right. DP, disc proper; PE, peripodial epithelium. (F) ey transcripts in c311-GAL4>UAS-ey RNAi eye-antennal discs. Error bars were determined by REST analysis (see Materials and Methods). (G) Distribution of eye sizes when ey is removed from the peripodial epithelium (n=30).

Fig. 8.

Ey within the peripodial epithelium regulates dpp expression and pattern formation of the eye field. (A-C) Third instar eye-antennal discs in which ey expression has been removed from the peripodial epithelium while maintaining it within the eye disc proper. (D) Distribution of eye sizes in ey^LB mutants alone (n=111) and when dpp expression is added to the peripodial epithelium in an ey^LB mutant background (n=43). (E) ey^LB adult eye that has been completed rescued by the expression of dpp within the peripodial epithelium. Anterior is to the right. Scale bars: 50 μm.

Fig. 9.

A new model for how Ey/Pax6 regulates eye development. We propose that Ey (and other members of the PSED core unit of the RD network) function within the peripodial epithelium to activate Dpp signaling. In turn, as shown by many laboratories, Dpp signaling at the posterior margin (cuboidal cells) cooperates with other pathways such as Hh, Jak/STAT, Egfr and Notch to initiate the morphogenetic furrow. The RD network then, as currently envisioned, promotes the progression of the furrow by regulating Dpp signaling within the eye field itself. We further propose that in the absence of Ey, the other Pax6 protein, Toy, is unable to robustly activate the downstream PSED members and, as a result, the size of retina is reduced and variable. In animals that completely lack the compound eyes, the primary defect appears to be the loss of Dpp expression and a failure of the morphogenetic furrow to initiate from the posterior margin.