The Rx-like homeobox gene (Rx-L) is necessary for normal photoreceptor development

Abstract

Purpose: The retinal homeobox (Rx) gene plays an essential role in retinal development. An Rx-like (Rx-L) gene from Xenopus laevis has been identified. The purpose of this study was to analyze the function of Rx-L in the developing retina.

Methods: DNA-binding properties of Rx-L were analyzed by electrophoretic mobility shift assay (EMSA), with in vitro-translated proteins and radiolabeled oligonucleotide probe. The Rx-L expression pattern was analyzed by in situ hybridization using whole or sectioned embryos and digoxigenin-labeled antisense riboprobes. Rx-L loss of function was studied by using antisense morpholino oligonucleotides targeted to the Rx-L translation initiation site. Embryos injected with control or Rx-L morpholinos were analyzed at stage 41 or 45.

Results: Rx-L shares homology with Rx at the homeo-, OAR, and Rx domains, but lacks an octapeptide motif. Rx-L is expressed in the developing retina beginning in the early tailbud stage. In the maturing retina, Rx-L expression is restricted primarily to the developing photoreceptor layer and the ciliary marginal zone. Rx-L can bind a photoreceptor conserved element-1 (PCE-1) oligonucleotide, an element conserved among all known photoreceptor gene promoters. In a promoter activity assay, Rx-L functions as a stronger transcriptional activator than Rx. Antisense morpholino-mediated knockdown of Rx-L expression resulted in a decrease in rhodopsin and red cone opsin expression levels in Xenopus retinas. Injection of the Rx-L antisense morpholino oligonucleotide also resulted in a decrease in the length of both rod and cone outer segments.

Conclusions: The results suggest that Rx-L functions to regulate rod and cone development by activating photoreceptor-specific gene expression.

Figure 1

Rx-L is related to Rx and similar to other Rx-like proteins. (A) Comparison of the predicted protein sequence of Xenopus laevis Rx-L and Rx1A. Rx1A and Rx-L sequences were aligned by using ClustalW (http://www.ebi.ac.uk/clustalw; European Bioinformatics Institute, European Molecular Biology Laboratory, Heidelberg, Germany). The percentage of amino acid identity and similarity (with respect to Rx-L) are summarized in the figure. Canonical domains are indicated. (B) Alignment of Xenopus laevis (Rx-L), chicken (RaxL), and human (QRX) Rx-like gene product sequences by ClustalW. Canonical domains are denoted with brackets above the alignment. OP, octapeptide; HD, homeodomain; RX, Rx domain; OAR, orthopedia-aristaless-Rx domain.

Figure 2

Rx-L was expressed in the developing retinas of tailbud embryos and tadpoles. (A, L) Whole-mount in situ hybridization (ISH) with an Rx-L (A–E) or an Rx(G–K) antisense riboprobe. (F, L) ISH with Rx-L (F) or Rx (L) sense control probes. (M–U) ISH on sections of paraffin-embedded tadpoles with antisense riboprobes for Rx-L (M–O), Rx (P–R), or rhodopsin (S) at st 36 (M, P), st 38 (N, Q, S), or st 41 (O, R). (T, U) ISH with sense control riboprobes for Rx-L (T) or Rx (U) at st 36. All wholemounts and sections are oriented with dorsal side to the top of the images. Embryo staging is indicated in each panel. (V) Analysis of expression of Rx and Rx-L in early development by RT-PCR with RNA purified from whole embryos (at indicated stages). EF-1α was used as a control for RT-PCR.

Figure 3

Function of Rx-L as a transcription factor. (A) Rx-L binds the same oligonucleotide as RLamx. Synthetic Rx1A (lanes 2, 6) and Rx-L (lanes 7, 11) were used in an EMSA with a radiolabeled PCE-1 probe and various oligonucleotide competitors. W, wt PCE-1; M, mutated PCE-1; B, BAT-1; R, Ret4. The major specific Rx and Rx-L complexes are indicated (solid and open arrowheads, respectively). (B) Rx-L is a stronger transcriptional activator than Rx. Lucif-erase assay performed with lysates from embryos coinjected with XOP-Luc reporter plasmid and Rx or Rx-L RNAs, as shown.

Figure 4

Injection of the Rx-L anti-sense MO oligonucleotide specifically inhibited translation of Rx-L but did not result in gross changes in eye morphology. (A) The Rx-L antisense MO oligonucleotide (Rx-L MO), but not the control MO oligonucleotide (ctl MO), can inhibit Rx-L translation in vitro. (B–G) The Rx-L MO, but not a control MO, can inhibit expression of a Rx-L-GFP fusion protein containing the Rx-L MO target in Xenopus embryos. (B, C) bright-field (B) and fluorescent (C) views of embryos injected with Rx-L-GFP plasmid. (D, E) GFP fluorescence in embryos injected with Rx-L-GFP plasmid and 0.1 (D) or 0.25 (E) mM ctl MO. (F, G) GFP fluorescence in embryos injected with Rx-L-GFP plasmid and 0.1 (F) or 0.25 (G) mM Rx-L MO. Insets: visualization of lissamine tag of MO (red fluorescence). (H–N) Microinjection of 0.25 mM Rx-L MO did not affect external morphology of the eye. Embryos were unilaterally co-injected with lissamine-conjugated Rx-L MO and GFP RNA at the four-cell stage. (H–J) Dorsal views of injected embryos with bright-field illumination (H), green fluorescence to detect GFP tracer (I) or red fluorescence to detect lissamine-conjugated MO (J). (K, L) lateral view of the injected side of the same embryo viewed with bright-field illumination (K) or red fluorescence (L). (M, N) lateral view of the uninjected side of the same embryo viewed with bright-field illumination (M) or red fluorescence (N). (O–V) Retinas from embryos injected with Rx-L MO appeared histologically similar to those from embryos injected with ctl MO at st 41 and 45 (as indicated).

Figure 5

The Rx-L antisense MO oligonucleotide resulted in perturbations in photoreceptor development. (A–D) Analysis of rod photoreceptor morphology visualized by immunohistochemistry with an antibody to rhodopsin (Ret-P1). (E) Diagram of measurements used to analyze changes in photoreceptor layer thickness. (F) Quantification of the effect of micro-injection of control or Rx-L antisense MO oligonucleotide on rod photoreceptor layer thickness. (G–J) Analysis of cone morphology visualized by binding to PNA. (A’–D’) and (G’–J’): high-magnification (100×) views of boxed area indicated in (A–D) and (G–J), respectively. Green bars indicate the width of the photoreceptor layer. All images were captured at 40× magnification. *P < 0.00013.

Figure 6

Rx-L knockdown decreased rod and cone opsin expression. (A–D) Analysis of rhodopsin gene expression by in situ hybridization performed on sectioned retinal tissue. (E–H) Analysis of red cone opsin expression by in situ hybridization. (I) Quantatative analysis of red cone opsin expression levels in control and Rx-L antisense MO oligonucleotide–injected embryos by real-time RT-PCR. *P < 0.002; **P < 0.001.

Figure 7

Microinjection of the Rx-L MO oligonucleotide was not broadly deleterious to retinal development. (A–D) The expression of Pax6 in the ganglion cell and inner nuclear layers of the neural retina are unaffected by injection of the Rx-L MO. (E–H) Injection of the Rx-L MO did not affect the development of nonphotoreceptor cell types such as ganglion, amacrine, and horizontal cells, visualized by immunohistochemistry with an antibody to Isl-1. (A), (E) and (C), (G): retinas from uninjected sides of embryos injected with control or Rx-L MO, respectively. (B), (F) and (D), (H): retinas from injected sides of embryos injected with control or Rx-L MO, respectively. A, amacrine cells; G, ganglion cell layer; H, horizontal cells; I, inner nuclear layer; L, lens; P, photoreceptor layer.