Rod differentiation factor NRL activates the expression of nuclear receptor NR2E3 to suppress the development of cone photoreceptors

Abstract

Neural developmental programs require a high level of coordination between the decision to exit cell cycle and acquisition of cell fate. The Maf-family transcription factor NRL is essential for rod photoreceptor specification in the mammalian retina as its loss of function converts rod precursors to functional cones. Ectopic expression of NRL or a photoreceptor-specific orphan nuclear receptor NR2E3 completely suppresses cone development while concurrently directing the post-mitotic photoreceptor precursors towards rod cell fate. Given that NRL and NR2E3 have overlapping functions and NR2E3 expression is abolished in the Nrl^−/− retina, we wanted to clarify the distinct roles of NRL and NR2E3 during retinal differentiation. Here, we demonstrate that NRL binds to a sequence element in the Nr2e3 promoter and enhances its activity synergistically with the homeodomain protein CRX. Using transgenic mice, we show that NRL can only partially suppress cone development in the absence of NR2E3. Gene profiling of retinas from transgenic mice that ectopically express NR2E3 or NRL in cone precursors reveals overlapping and unique targets of these two transcription factors. Together with previous reports, our findings establish the hierarchy of transcriptional regulators in determining rod versus cone cell fate in photoreceptor precursors during the development of mammalian retina.

INTRODUCTION

The central nervous system is assembled from thousands of distinct cell types that must be produced in correct numbers and within appropriate spatial and temporal constraints [19,20,48]. Unraveling the molecular mechanisms that control specification of diverse cell types during neurogenesis has been the subject of intense scrutiny. The vertebrate retina offers a convenient paradigm to examine cell fate decisions, as it is relatively less complex and more accessible to experimental manipulations. Birth-dating studies reveal that six neuronal cell types and one type of glia are produced from common pools of neuroepithelial progenitors in a conserved order of birth [1,7,34,35,53,54,57]. Retinal ganglion cells are generated first, followed by cone, amacrine and horizontal cells, whereas rods, bipolar cells and Muller glia are generated later [7,45,60]. A competence model has been proposed to explain the sequential birth order [10,34]; according to this, multipotent pools of progenitor cells go through discrete competence states during which they can only give rise to specific subsets of neurons. Restriction(s) in developmental potential can occur prior to or after exit from the final cell cycle through combinatorial regulatory mechanism(s) involving intrinsic factors and/or extrinsic signaling molecules [8,21,31]. As considerable overlap exists in the timings of cell birth, specific instructions and stringent controls are required to produce particular subtypes of neurons from a set of progenitor cells.

Rod and cone photoreceptors are specialized light-sensing neurons with highly structured membrane disks (called outer segments), which contain visual pigments and other phototransduction components. In most mammals (including humans and mice), rod photoreceptors greatly outnumber cones and generally account for over 70% of all cells in the retina [7]. Although the morphology and physiology of photoreceptors are well documented, the developmental pathways from a multipotent retinal progenitor to a committed precursor and a terminally-differentiated photoreceptor are only beginning to be elucidated.

NRL belongs to the basic motif-leucine zipper family of transcription factors and is an essential regulator of early events leading to the birth and development of rod photoreceptors [4,37,40,47,51]. It is preferentially expressed in rods (and not other retinal cells) and pineal gland [4,50]. In mice lacking NRL, rods are transformed to cones that share morphological, molecular, and electrophysiological characteristics similar to wild type cones [18,37,39]. More recently, gain of function studies reveal that NRL can influence all photoreceptor precursors to initiate a rod differentiation program at the expense of cones [40]. Despite the absence of cones, cone bipolar and horizontal interneurons are present in the adult retina but do not attain appropriate neuronal morphology during synaptogenesis [40,46]. CRX is a photoreceptor-specific homeodomain protein that plays a critical role in the maturation of photoreceptors [14,23], but it does not appear to be essential for initial specification events [24]. Mis-expression studies in adult iris cells [2] suggest that CRX acts as a photoreceptor competence factor before NRL defines rod identity. Control of photoreceptor cell fate also involves the participation of NR2E3, a photoreceptor-specific orphan nuclear receptor [3,12,16,26,42]. The rd7 mice carrying an antisense L1 insertion into exon 5 of the Nr2e3 gene exhibit a progressive photoreceptor degeneration accompanied by 1.5–2 fold increase in the number of S-cones [3,13,27,55]. Ectopic expression of NR2E3 or NRL [15,40] in the photoreceptor precursors of Nrl^−/− mice results in the complete inhibition of cone developmental program [15]; however, in contrast to NRL [40], functional rods are not generated by NR2E3 expression alone [15].

Given that NRL and NR2E3 functions are overlapping and NR2E3 expression is undetectable in the Nrl^−/− mice [15,36,37,40] it has been suggested that NR2E3 is downstream of NRL in transcriptional hierarchy controlling retinal development [37]. In this report, we have examined whether NR2E3 is a direct target of NRL and evaluated the precise role NRL in cone specification in the absence of NR2E3. We also present expression profiles of retinas from transgenic mice that ectopically express either NRL and NR2E3 or NR2E3 alone in cone precursors, with a goal to identify cone-enriched genes in mature photoreceptors.

RESULTS

NRL directly binds to the Nr2e3 promoter

To determine whether NRL can modulate NR2E3 expression, we first analyzed the promoter of the Nr2e3 gene and identified four sequence regions that are conserved in mammals (Figure 1 A). In silico analysis revealed a putative NRL response element (NRE) in one of the conserved regions (see Figure 1 A, grey box). Addition of nuclear extracts from COS-1 cells expressing the NRL protein, but not from mock-transfected cells, to ³²P-labeled NRE oligonucleotide resulted in band-shift in electrophoretic mobility shift assays (EMSA) (Figure 1 B; lanes 1–3), suggesting the binding of NRL to NRE sequence in the Nr2e3 promoter region. The specificity of binding was substantiated by competition with an excess of unlabeled oligonucleotide spanning the NRE but not with a mutant sequence (lanes 4–6). The major shifted band (shown by the arrowhead) was clearly detectable upon the addition of rabbit IgG but not anti-NRL antibody (lanes 7, 8), providing further evidence in support of NRL’s binding to Nr2e3-NRE. To determine whether NRL could bind the Nr2e3 promoter in vivo, we performed chromatin immunoprecipitation (ChIP) experiments. Cross-linked protein-DNA complexes from wild-type adult retinas were immunoprecipitated with anti-NRL antibody, and purified ChIP DNA was used for PCR with primers flanking the Nr2e3-NRE site. A strong enrichment of the Nr2e3-NRE promoter fragment was observed with anti-NRL antibody compared to rabbit IgG (Figure 1 C). Additionally, no significant enrichment was detected for another randomly-selected sequence in the Nr2e3 gene (negative control) (Figure 1 C).

Figure 1. Binding to and activation of the Nr2e3 promoter by NRL.

(A) Schematic of approximately 4.5 kb genomic DNA upstream of the Nr2e3 transcription start site (denoted as +1). The four boxes indicate sequence regions conserved in mammals. A comparison of sequences in the second conserved region including a putative NRE (highlighted in grey) is shown below the schema. (B) EMSA showing the binding of NRL to NRE site in the Nr2e3 promoter. Lanes are as indicated above the autoradiograph. Nr2e3 oligo* indicates ³²P-labeled NRE oligonucleotide (−2820 nt to −2786 nt) (in all lanes). Nrl – NE shows 10 µg nuclear extract from untransfected COS-1 cells (lane 2), whereas Nrl + NE means 10 µg nuclear extract from COS-1 cells transfected with Nrl cDNA expression plasmid (lanes 3–8). Lane 4 and 5 included 50- or 100 fold molar excess of unlabeled NRE oligonucleotide. Lane 6 included 100-fold molar excess of unlabeled mutant NRE oligonucleotide. Lane 7 contains 2 µg anti-NRL antibody, whereas lane 8 has 2 µg normal rabbit IgG. Arrowhead represents the specific shifted band, which is undetectable when anti-NRL antibody is included. Asterisk indicates a shifted band (of low molecular mass) that does not seem to be altered by the addition of anti-NRL antibody. These experiments were performed three times, and similar results were obtained. (C) PCR assays using immunoprecipitated chromatin from adult C57BL/6J retinas. Lane1, NRL antibody used for IP; lane 2, normal rabbit IgG used for IP (negative control); lane 3, input DNA used as template. Top panel: primers amplifying the NRE containing region (−2989 nt to −2742 nt) in the Nr2e3 promoter region were used for PCR. Bottom panel: primers amplifying an irrelevant region (1230 nt to 1438 nt) in the Nr2e3 gene were used for PCR. (D) Luciferase reporter assays showing the activation of Nr2e3 promoter by NRL and CRX.

NRL induces the Nr2e3 promoter activity in transfected cells

We then examined the activity of a 4.5 kb Nr2e3 promoter fragment (encompassing the conserved NRE sequence; see Figure 1 A) in the presence of NRL. Transfection of HEK-293 cells with NRL, but not CRX, expression plasmid induced the luciferase reporter activity that was driven by the Nr2e3 promoter (Figure 1 D). Co-transfection of HEK-293 cells with both NRL and CRX plasmids resulted in further increase of the Nr2e3 promoter activity (Figure 1 D). This is consistent with previously-reported synergistic activation of several rod-specific genes by NRL and CRX [14,16,38,44].

Overlapping yet distinct gene profiles are generated by NRL and NR2E3

Recent investigations into the role of NRL and NR2E3 [12,15,29,40,42] and our findings reported here (Figure 1) suggest that NRL suppresses cone differentiation by directly signaling through NR2E3. This level of regulation also implies that many molecular defects observed in mice lacking functional NR2E3 (e.g., the rd7 mouse) are also present in the Nrl^−/− mice [17,37]. To dissect the transcriptional activity of NRL versus NR2E3 in mature photoreceptors, we took advantage of two recently-generated transgenic mouse models – Crxp-Nrl/WT [40] and Crxp-Nr2e3/WT [15]. In these mice, a 2 kb Crx proximal promoter [22] leads to the expression of NRL or NR2E3 in photoreceptor precursors and transformation of cones to rod photoreceptors, without any obvious perturbation in retinal lamination or development of other cell types [15,40].

In the Crxp-Nrl/WT retinas, NRL and consequently NR2E3 ([40], see Fig. 1) are ectopically expressed in cone precursors; while only NR2E3 (and not NRL) is ectopically expressed in cone precursors of the Crxp-Nr2e3/WT retina. NRL and NR2E3 are also expressed in the developing rod precursors of both transgenic lines. Therefore, gene profiling of retinas from Crxp-Nrl/WT and Crxp-Nr2e3/WT mice can reveal expression changes induced by NRL+NR2E3 or NR2E3 alone in cone precursors, respectively. Retinal RNA from P28 adult mice was hybridized to Affymetrix MOE430.2.0 GeneChips, which contain 45,101 probesets for mouse transcripts. A comparative analysis of gene clusters from Crxp-Nrl/WT and Crxp-Nr2e3/WT retinas to WT samples revealed a number of genes involved in diverse signaling pathways and transcriptional regulation; Table 1 shows the genes with FDRCI P-value of <0.1 and a fold change >4. In addition to established cone-specific genes, we also discovered several new genes down-regulated in the Crxp-Nrl/WT and Crxp-Nr2e3/WT coneless groups which are potential cone-enriched target genes. We then compared Crxp-Nrl/WT and Crxp-Nr2e3/WT gene profiles to Nrl^−/− (cone-only) and rd7 (1.5–2 fold more S-cones) profiles. Many cone phototransduction genes that are up-regulated in the Nrl^−/− (cone-only, Table 2) and rd7 (1.5–2 fold more S-cones, Table 3) retinas are also significantly repressed in the Crxp-Nrl/WT and Crxp-Nr2e3/WT coneless samples. Gene expression changes showing FDRCI P-value < 0.1 and a fold change > 10 are listed in Table 2 and Table 3.

Table 1.

Non-redundant differentially expressed genes in Crxp-Nrl/WT or Crxp-Nr2e3/WT samples compared to WT retinas. Gene profiles of P28 retinal samples from Crxp-Nrl/WT or Crxp-Nr2e3/WT mice were compared to those from the WT retina. Common genes in Crxp-Nrl/WT and Crxp-Nr2e3/WT, or unique genes from Crxp-Nrl/WT or Crxp-Nr2e3/WT with a minimum fold change of 4 and FDRCI P-value of < 0.1 are shown. AFC, average fold change.

Gene Symbol	Gene Title	AFC Crxp-Nrl/WT versus WT	AFC Crxp-Nr2e3/WT versus WT	GO biological process description
Overlapping genes in Crxp-Nrl/WT and Crxp-Nr2e3/WT versus WT group
Gdpd3	Glycerophosphodiester phosphodiesterase domain containing 3	21.4	4.6	Alcohol metabolic process
Sgcg	Sarcoglycan, gamma (dystrophin-associated glycoprotein)	14.8	12.1	Cytoskeleton organization and biogenesis
Cap1	CAP, adenylate cyclase-associated protein 1 (yeast)	10.2	2.1	Actin cytoskeleton organization and biogenesis
Plekhk1	Pleckstrin homology domain containing, family K member 1	6.3	4.7	Regulation of cell proliferation
Tsc22d1	TSC22 domain family, member 1	6.0	6.5	Regulation of transcription
Paip1	Polyadenylate binding protein-interacting protein 1	4.4	2.6	Regulation of translation
Huwe1	HECT, UBA and WWE domain containing 1	4.1	3.0	DNA packaging
Abca13	ATP-binding cassette, sub-family A (ABC1), member 13	−4.0	−6.5	Intracellular protein transport
Clca3	Chloride channel calcium activated 3	−4.5	−4.9	Chloride transport
Camk2b	Calcium/calmodulin-dependent protein kinase II, beta	−5.1	−20.0	Regulation of phosphorylation
Gnat2	Guanine nucleotide binding protein, alpha transducing 2	−8.3	−10.1	Phototransduction
Pttg1	Pituitary tumor-transforming 1	−10.3	−4.2	Meiotic chromosome segregation
Pde6c	Phosphodiesterase 6C, cGMP specific, cone, alpha prime	−15.9	−13.8	Activation of MAPK activity/visual perception
Opn1mw	Opsin 1 (cone pigments), medium-wave- sensitive (color blindness, deutan)	−38.7	−37.6	Phototransduction
Arr3	Arrestin 3, retinal	−56.3	−31.1	Regulation of phosphorylation
Pde6h	Phosphodiesterase 6H, cGMP-specific, cone, gamma	−75.6	−70.9	Activation of MAPK activity/visual perception
Opn1sw	Opsin 1 (cone pigments), short-wave- sensitive (color blindness, tritan)	−124.0	−98.4	Phototransduction
Unique genes in Crxp-Nrl/WT versus WT group
Rds	Retinal degeneration, slow (retinitis pigmentosa 7)	56.5	1.5	Visual perception
Nudt21	Nudix (nucleoside diphosphate linked moiety X)-type motif 21	10.3	1.1
Deadc1	Deaminase domain containing 1	6.6	1.9	tRNA processing
Atp2b2	ATPase, Ca++ transporting, plasma membrane 2	5.6	−1.1	Inner ear development
Scd2	Stearoyl-Coenzyme A desaturase 2	5.5	−1.2	Fatty acid biosynthetic process
Atp6v0a1	ATPase, H+ transporting, lysosomal V0 subunit A1	5.2	−1.1	Transcription initiation
Stk35	Serine/threonine kinase 35	5.2	1.2	Protein amino acid phosphorylation
Uhmk1	U2AF homology motif (UHM) kinase 1	4.3	−1.3	Protein amino acid phosphorylation
Gucy2e	Guanylate cyclase 2e	4.3	1.4	Visual perception
Thbs1	Thrombospondin 1	4.2	1.1	Blood vessel morphogenesis
Fabp4	Fatty acid binding protein 4, adipocyte	−4.0	1.2	Regulation of protein kinase activity
5730410E15Rik	RIKEN cDNA 5730410E15 gene	−4.1	−1.2
5330426P16Rik	RIKEN cDNA 5330426P16 gene	−4.1	−1.4
Ssbp2	Single-stranded DNA binding protein 2	−4.3	−1.0	Regulation of transcription
Ing3	Inhibitor of growth family, member 3	−4.4	1.0	Regulation of transcription
Dmd	Dystrophin, muscular dystrophy	−4.4	1.1	Peptide biosynthetic process
A930033H14Rik	RIKEN cDNA A930033H14 gene	−4.7	1.4
Aff1	AF4/FMR2 family, member 1	−5.0	−1.3	Regulation of transcription
Vapb	Vesicle-associated membrane protein, associated protein B and C	−5.1	−1.2
Scn2b	Sodium channel, voltage-gated, type II, beta	−8.2	−1.1	Sodium ion transport
Unique genes in Crxp-Nr2e3/WT versus WT group
Sox30	SRY-box containing gene 30	1.8	6.8	Regulation of transcription
2900056M20Rik	RIKEN cDNA 2900056M20 gene	−1.9	−6.6
LOC552908	Hypothetical LOC552908	−1.7	−10.1

Table 2.

Non-redundant differentially expressed genes in Crxp-Nrl/WT or Crxp-Nr2e3/WT samples compared to Nrl^−/− retinas. Gene profiles of P28 retinal samples from Crxp-Nrl/WT or Crxp-Nr2e3/WT were compared to the profiles from the Nrl^−/− retina. We show common differentially expressed genes in Crxp-Nrl/WT and Crxp-Nr2e3/WT retina, or unique genes from Crxp-Nrl/WT or Crxp-Nr2e3/WT, with a minimum fold change of 10 and FDRCI P-value of < 0.1. AFC, average fold change.

Gene Symbol	Gene Title	AFC Crxp-Nrl/WT versus Nrl−/−	AFC Crxp- Nr2e3/WT versus Nrl−/−	GO biological process description
Overlapping genes in Crxp-Nrl/WT and Crxp-Nr2e3/WT versus Nrl−/− group
Nrl	Neural retina leucine zipper gene	386.9	381.8	Photoreceptor cell development
Rho	Rhodopsin	347.1	332.5	Phototransduction
Nr2e3	Nuclear receptor subfamily 2, group E, member 3	115.8	83.9	Photoreceptor cell development
Gnb1	Guanine nucleotide binding protein, beta 1	69.7	46.1	Phototransduction
Slc24a1	Solute carrier family 24 (sodium/potassium/calcium exchanger), member 1	50.9	43.3	Calcium ion transport/visual perception
A930036K24Rik	RIKEN cDNA A930036K24 gene	50.1	88.7
BC016201	cDNA sequence BC016201	40.8	33.8
Esrrb	Estrogen related receptor, beta	32.4	26.9	Intracellular protein transport
Susd3	Sushi domain containing 3	23.4	21.5
Aqp1	Aquaporin 1	21.3	31.4	Water transport
BC038479	cDNA sequence BC038479	20.4	17.5
Reep6	Receptor accessory protein 6	17.9	20.9	Protein binding
Mef2c	Myocyte enhancer factor 2C	16.7	10.8	Regulation of transcription
Pde6b	Phosphodiesterase 6B, cGMP, rod receptor, beta polypeptide	16.7	19.4	Activation of MAPK activity/Visual perception
Wisp1	WNT1 inducible signaling pathway protein 1	16.1	20.3	Regulation of cell growth
Sh2d1a	SH2 domain protein 1A	15.8	18.7	Intracellular signaling cascade
Sgcg	Sarcoglycan, gamma (dystrophin-associated glycoprotein)	14.6	10.6	Cytoskeleton organization and biogenesis
Samd11	Sterile alpha motif domain containing 11	13.9	16.5	Negative regulation of transcription
Plekha2	Pleckstrin homology domain-containing, family A (phosphoinositide binding specific) member 2	11.4	14.7	Phosphatidylinositol binding
Vax2os1	Vax2 opposite strand transcript 1	10.3	15.6
Guca1b	Guanylate cyclase activator 1B	10.2	10.6	Phototransduction
Gulo	Gulonolactone (L-) oxidase	−10.0	−31.3	L-ascorbic acid biosynthetic process
Pik3ap1	Phosphoinositide-3-kinase adaptor protein 1	−11.1	−11.3	Transmembrane receptor protein tyrosine kinase signaling protein activity
Gngt2	Guanine nucleotide binding protein (G protein), gamma transducing activity polypeptide 2	−11.7	−11.9	G-protein coupled receptor protein signaling pathway
En2	Engrailed 2	−12.0	−10.4	Regulation of transcription
Myocd	Myocardin	−12.5	−12.7	Regulation of cell growth
Kcne2	Potassium voltage-gated channel, Isk-related subfamily, gene 2	−12.7	−12.4	Potassium ion transport
Arhgdib	Rho, GDP dissociation inhibitor (GDI) beta	−13.8	−12.8	Rho GDP-dissociation inhibitor activity
Parvb	Parvin, beta	−13.9	−14.0	Actin binding
Cckbr	Cholecystokinin B receptor	−17.4	−33.3	G-protein coupled receptor protein signaling pathway
Klhl4	Kelch-like 4 (Drosophila)	−19.8	−17.4
A930009A15Rik	RIKEN cDNA A930009A15 gene	−24.4	−28.9
Otop3	Otopetrin 3	−39.2	−39.3
Cngb3	Cyclic nucleotide gated channel beta 3	−40.8	−49.4	Potassium ion transport
Gnat2	Guanine nucleotide binding protein, alpha transducing 2	−46.0	−56.2	Phototransduction
Fabp7	Fatty acid binding protein 7, brain	−60.1	−48.8	Lipid binding
Opn1mw	Opsin 1 (cone pigments), medium-wave- sensitive (color blindness, deutan)	−77.9	−76.2	Phototransduction
Clca3	Chloride channel calcium activated 3	−91.6	−93.5	Chloride transport
Pde6c	Phosphodiesterase 6C, cGMP specific, cone, alpha prime	−158.3	−136.8	Activation of MAPK activity/visual perception
Arr3	Arrestin 3, retinal	−271.2	−166.0	Regulation of phosphorylation
Pde6h	Phosphodiesterase 6H, cGMP-specific, cone, gamma	−429.1	−409.3	Activation of MAPK activity/visual perception
Opn1sw	Opsin 1 (cone pigments), short-wave- sensitive (color blindness, tritan)	−559.2	−546.8	Phototransduction
Unique genes in Crxp-Nrl/WT versus Nrl−/− group
Rds	Retinal degeneration, slow (retinitis pigmentosa 7)	149.7	3.3	Sensory perception of light stimulus
Stk35	Serine/threonine kinase 35	11.7	2.3	Protein amino acid phosphorylation
Mtmr7	Myotubularin related protein 7	−11.8	1.8	Phospholipid dephosphorylation
Pcdh15	Protocadherin 15	−12.6	−3.3	Sensory perception of light stimulus
Pip5k2b	Phosphatidylinositol-4-phosphate 5-kinase, type II, beta	−12.6	1.3	Glycerophospholipid metabolic process
Unique genes in Crxp-Nr2e3/WT versus Nrl−/− group
A930003C13Rik	RIKEN cDNA A930003C13 gene	3.5	11.5
Skiv2l2	Superkiller viralicidic activity 2-like 2 (S. cerevisiae)	−4.4	−17.0	RNA splicing

Table 3.

Non-redundant differentially expressed genes in Crxp-Nrl/WT or Crxp-Nr2e3/WT samples compared to rd7 retinas. Gene profiles of P28 retinal samples from Crxp-Nrl/WT or Crxp-Nr2e3/WT were compared to those of rd7 retina. Common genes in Crxp-Nrl/WT and Crxp-Nr2e3/WT, or unique genes from Crxp-Nrl/WT or Crxp-Nr2e3/WT with a minimum fold change of 10 and FDRCI P-value of < 0.1 are shown. AFC, average fold change.

Gene Symbol	Gene Title	AFC Crxp-Nrl/WT versus rd7	AFC Crxp-Nr2e3/WT versus rd7	GO biological process description
Overlapping genes in Crxp-Nrl/WT and Crxp-Nr2e3/WT versus rd7 group
Eif2s3y	Eukaryotic translation initiation factor 2, subunit 3, structural gene Y-linked	65.3	79.1	Macromolecule biosynthetic process
Ddx3y	DEAD (Asp-Glu-Ala-Asp) box polypeptide 3, Y-linked	62.1	74.0
Sgcg	Sarcoglycan, gamma (dystrophin-associated glycoprotein)	14.8	10.7	Cytoskeleton organization and biogenesis
Jarid1d	Jumonji, AT rich interactive domain 1D (Rbp2 like)	11.8	11.2	Regulation of transcription
LOC640072 ///	Hypothetical protein LOC640072 ///	−11.3	−12.6
LOC677194	Hypothetical protein LOC677194
A230097K15Rik	RIKEN cDNA A230097K15 gene	−12.7	−11.9
Arhgdib	Rho, GDP dissociation inhibitor (GDI) beta	−12.8	−11.9	Rho GDP-dissociation inhibitor activity
Gulo	Gulonolactone (L-) oxidase	−13.6	−42.6	L-ascorbic acid biosynthetic process
Socs3	Suppressor of cytokine signaling 3	−15.8c	−11.6	Regulation of phosphorylation
Bub1b	Budding uninhibited by benzimidazoles 1 homolog, beta (S. cerevisiae)	−20.1	−14.7	Regulation of phosphorylation
Edn2	Endothelin 2	−20.1	−11.1	Regulation of vasoconstriction
Otop3	Otopetrin 3	−28.1	−28.1
Fabp7	Fatty acid binding protein 7, brain	−37.7	−30.6	Lipid binding
A930009A15Rik	RIKEN cDNA A930009A15 gene	−38.0	−45.1
Gnat2	Guanine nucleotide binding protein, alpha transducing 2	−39.7	−48.5	Phototransduction
Opn1mw	Opsin 1 (cone pigments), medium-wave- sensitive (color blindness, deutan)	−40.5	−39.6	Phototransduction
Arr3	Arrestin 3, retinal	−50.1	−25.3	Regulation of phosphorylation
Clca3	Chloride channel calcium activated 3	−78.5	−80.1	Chloride transport
Pde6c	Phosphodiesterase 6C, cGMP specific, cone, alpha prime	−127.0	−109.7	Activation of MAPK activity/visual perception
Opn1sw	Opsin 1 (cone pigments), short-wave-sensitive (color blindness, tritan)	−223.2	−218.3	Phototransduction
Pde6h	Phosphodiesterase 6H, cGMP-specific, cone, gamma	−365.1	−348.2	Activation of MAPK activity/visual perception
Unique genes in Crxp-Nrl/WT versus rd7 group
Rds	Retinal degeneration, slow (retinitis pigmentosa 7)	74.7	1.6	Sensory perception of light stimulus
Cap1	CAP, adenylate cyclase-associated protein 1 (yeast)	11.3	2.3	Actin cytoskeleton organization and biogenesis
Scn2b	Sodium channel, voltage-gated, type II, beta	−10.2	−1.3	Sodium ion transport
Fabp4	Fatty acid binding protein 4, adipocyte	−11.3	−2.4	Regulation of protein kinase activity
Mtmr7	Myotubularin related protein 7	−27.7	−1.3	Phospholipid dephosphorylation
Unique genes in Crxp-Nr2e3/WT versus rd7 group
Camk2b	Calcium/calmodulin-dependent protein kinase II, beta	−3.1	−12.2	Regulation of phosphorylation
LOC552908	Hypothetical LOC552908	−2.4	−14.0

Expression of NRL can only suppress a subset of S-cones in the absence of NR2E3

Similarities in gene profiles of Crxp-Nrl/WT and Crxp-Nr2e3/WT retinas raise the question whether NRL can suppress cone gene expression and differentiation even in the absence of NR2E3. To evaluate this, we mated Crxp-Nrl/WT mice to rd7 mice to generate a transgenic mouse line (Crxp-Nrl/rd7) that expresses NRL, but not NR2E3, in both cone and rod precursors. We first analyzed cone markers, such as S- and M-opsin, in retinal whole mounts. As previously demonstrated [5], we observed an inferior to superior gradient of S-opsin expression (Figure 2 A–C) and a superior to inferior gradient of M-opsin in the WT mice (data not shown). As predicted, S-opsin was detected throughout in the Nrl^−/− retinal whole mounts (Figure 2 D–F) and increased S-opsin staining was observed in the rd7 retinas (Figure 2 J–L); however, both S-opsin and M-opsin could not be detected in Crxp-Nrl/WT retinas (Figure 2 G–I, and data not shown). In both Nrl^−/− and rd7 mice, whorls are detected in the whole mount preparations (Figure 2 D–F and J–K). In Crxp-Nrl/rd7 retinal whole mounts, we observed a large absence of S-opsin staining in the superior domain (Figure 2 M, O) yet detected a small population of S-opsin positive cells in the inferior retina (Figure 2 M, N). The expression of M-opsin was unaltered (data not shown), and whorls could be detected throughout the retinas (Figure 2 M–O).

Figure 2. Incomplete suppression of S-opsin expression by NRL in the absence of NR2E3.

(A–C) WT retina, showing superior to inferior gradient of S-opsin expression. (D–F) Nrl^−/− retina. In the absence of NRL and NR2E3, whorls (arrows) and S-opsin can be detected throughout the retina. (G–I) Crxp-Nrl/WT retina. Ectopic expression of NRL in early cone precursors results in the complete absence of S-opsin. (J–L) rd7 retina. In the absence of functional NR2E3, enhanced S-opsin expression and whorls (arrows) are observed in both superior and inferior domain. (M–O) Crxp-Nrl/rd7 retina. In the presence of NRL but absence of NR2E3, expression of S-opsin is reduced but detectable in the inferior domain. Asterisks are positioned at 3’o clock relative to the whorls (M). Arrows indicate the irregular S-opsin staining of whorls (O).

Scale bar: 200 µm (A, D, G, J, M), and 50 µm (B, C, E, F, H, I, K, L, N, O).

As shown previously [27,37,40], the number of cone arrestin (mCAR) and S-opsin positive cells in retinal cross-sections from Nrl^−/− and rd7 retinas were increased compared to WT, and there is an absence of cone-specific markers in Crxp-Nrl/WT mice (Figure 3 A: a–o). In Crxp-Nrl/rd7 sections, we observed normal cone arrestin and M-opsin staining but an absence of S-opsin in the superior domain (Figure 3 A: m–o). In the inferior domain, we identified a few S-opsin positive cones and many S-opsin positive cell bodies at the inner portion of the ONL (Figure 3 B: i, j). These findings were in contrast to S-opsin positive cell bodies distributed throughout the ONL and INL in Nrl^−/− and rd7 retinas (Figure 3 B: c–d and g–h). The expression of cone arrestin and M-opsin in the Crxp-Nrl/rd7 mice (harboring the Crxp-Nrl transgene in rd7 background with no NR2E3 function) but not in the Crxp-Nrl/WT mice (harboring the Crxp-Nrl transgene in wild-type background) demonstrates that NR2E3 is the primary suppressor of cone gene expression and cone differentiation.

Figure 3. Expression of cone-specific markers and targeting of photoreceptors to the ONL.

(A) Superior, and (B) inferior regions of the retina showing staining for cone arrestin (mCAR), S-opsin, and M-opsin antibodies (as shown on the left). Mouse strains are indicated. Compared to WT (B: a–b) and Crxp-Nrl/WT (B: e–f), targeting of S-cones (arrows) to the ONL is perturbed in Nrl^−/− (B: c–d) and rd7 retinas (B: g–h), and S-opsin positive nuclei are present in the INL. S-cone staining (arrowheads) in the Crxp-Nrl/rd7 retinas (B: i–j) is observed in cells closest to the outer plexiform layer. OS, outer segments; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; BBZ, bisbenzamide. Scale bar: 25 µm.

Cone function is detected but reduced in the Crxp-Nrl/rd7 mice

We performed electroretinography (ERG) recordings to measure the massed-field potential across the retina in the different transgenic lines. As reported previously [40], the ectopic expression of NRL in cone precursors (Crxp-Nrl/WT) resulted in an absence of cone-driven responses, whereas rod-driven components were preserved (Figure 4 A–E). To characterize the functionality of cone-driven neurons in the absence of NR2E3, we analyzed the photopic response from Crxp-Nrl/rd7 mice (Figure 4 C, D). In response to brief flashes of white light, we first detected a cone-driven b-wave at 0.09 log cd-s/m². At the higher flash intensity of 1.09 log cd-s/m² the maximum b-wave amplitude was about 40% of the WT and rd7 response amplitude (Figure 4 C, D).

Figure 4. Absence of normal cone function in cone photoreceptors expressing NRL but not NR2E3.

(A) Representative dark-adapted ERGs for increasing stimulus intensities are shown for WT, Nrl^−/−, Crxp-Nrl/WT, rd7 and Crxp-Nrl/rd7 mice at two months age. Intensity-response functions for the (B) a-wave and (C) b-wave amplitude were plotted on log-linear coordinates. (D) Representative light-adapted ERGs waveforms with increasing stimulus intensity for WT, Nrl^−/−, Crxp-Nrl/WT, rd7 and Crxp-Nrl/rd7 mice, as indicated. (E) Plots of the b-wave amplitude as a function of stimulus intensity for light-adapted conditions. At 2 months of age, there was no significant difference in the photopic response between WT and rd7 mice. (D–E) B-wave amplitude at the maximum intensity for the Crxp-Nrl/rd7 mice. A reduction of about 40% is observed from the WT and rd7 mice. (F) Representative light-adapted S- (360 nm) and M- (510 nm) cone ERGs showing a smaller M-cone response and undetectable S-cone response in Crxp-Nrl/rd7 mice compared to WT mice. Bars indicate ± standard error.

We further examined the photopic ERG in the Crxp-Nrl/rd7 transgenic mice by recording light-adapted cone-mediated responses at 360 nm and 510 nm to isolate S-cone and M-cone function, respectively (Figure 4 F). As predicted, the Nrl^−/− mice showed an enhanced S-cone response when compared with WT mice [18,37]. There was no significant difference in the M-cone response amplitude between Nrl^−/− and WT mice. We then recorded from Crxp-Nrl/rd7 mice and found that while the M-cone response was reduced by 40% from WT mice, S-cone responses were undetectable (Figure 4 F).

DISCUSSION

Regulatory networks defining rod versus cone identity are under the direct control of bZIP transcription factor NRL [37,40]. In this report, we demonstrate that NR2E3 is a direct transcriptional target of NRL and that specification of rod cell fate over cone differentiation is dictated by the activation of NR2E3 in response to NRL. Restricted expression of these two key transcriptional regulators in photoreceptor precursors is essential for proper development of rods. Ectopic expression of either protein in cone precursors can reprogram the cone development pathway to generate rod photoreceptors [15,40]. We had shown previously that ectopic NR2E3 expression can inhibit the development of functional S and M-cones in the Nrl^−/− retina [15]. The current data suggest that NR2E3 is necessary to completely repress the development of M and some S-cones, and NRL alone can only repress a subset of S-cones. These genetic models therefore raise the possibility of heterogeneity within S-cones.

Several studies have indicated the association of NRL and NR2E3 with promoter elements of cone-specific genes [40,42,43]. In this report, we analyzed the relationship of NRL and NR2E3 in modulating the cone developmental program. Data from immunohistochemical and physiological studies presented herein suggest that NRL modulates the development of S-cones, and its gain or loss of function primarily results in alterations of the S-cone pathway. One possibility is that S-cones represent the “default fate” for photoreceptors in mice [9,52] and that the expression of NRL controls an important node for this process.

The presence of ectopic S-opsin cells in the INL of rd7 and Nrl^−/− retinas is reminiscent of previous findings showing opsin-like immunoreactive cells in the developing retina [25]. Our study reveals the presence of ectopic S-opsin positive cells that persist and survive in the adult retinas from Nrl^−/− and rd7 mice. What can account for the existence and survival of these neurons outside of their normal retinal photoreceptor layer? It is possible that NRL and NR2E3 dictate the expression of specific guidance cues that facilitate photoreceptor pathfinding to the vicinity of their appropriate target regions in a highly stereotyped and directed manner. Several candidate proteins that show an altered expression profile in the Nrl^−/− retina appear to match the role of an axonal guidance cue [59,61]. These include members of families of secreted signaling molecules, such as Wingless/Wnt and Decapentaplegic/Bone Morphogenic Protein/Transforming Growth Factor B (Dpp/BMP/TGFb) [11], which appear to have important functions during retinal development [6,32,33,41,56,58,61]. We hypothesize that in the absence of NRL, and consequently NR2E3, changes in Wnt and BMP pathway may create noise in a homing signal that is required to (i) bring all photoreceptors to the ONL, and/or (ii) promote the appropriate wiring of rods and cones to bipolar and horizontal neurons. Although our current microarray experiments of P28 retinas did not reveal significant changes in classical pathfinding genes, future efforts will focus on profiling early postnatal stages of retinal development.

The absence of cones in the Crxp-Nrl/WT and Crxp-Nr2e3/WT retinas resulted in normal architecture and lamination features, similar to the WT. A lack of structural abnormalities has allowed us to profile expression changes that may be specifically due to the absence of one class of neurons (i.e., cones). While the retinal profiles of Crxp-Nrl/WT and Crxp-Nr2e3/WT mice had many common genes, the Crxp-Nrl/WT profile contained more unique changes in gene expression, consistent with NRL being upstream of NR2E3 in transcriptional hierarchy. One interesting novel gene revealed from the gene profiling experiments is PTTG1, which is down-regulated in the coneless retinas. The inhibitory chaperone PTTG1 has been implicated as a mitotic checkpoint gene involved at the metaphase-anaphase interface [63]. Elucidation of specific roles of PTTG1 and other cone-enriched genes will require further investigation.

In conclusion, our work refines the roles of NRL and NR2E3 during photoreceptor differentiation. We show, for the first time, that NR2E3 is a direct downstream target of NRL and that the correct sequential expression of these transcriptional regulators may be required for appropriate expression of rod-specific opsin and suppression of cone phototransduction genes during normal retinal development. Additional studies are needed to precisely define how specific down-stream targets of NRL and NR2E3 fine-tune the differentiation of functional photoreceptors from post-mitotic committed precursors.

EXPERIMENTAL PROCEDURES

Transgenic mice

The Crxp-Nrl/WT and Crxp-Nr2e3/WT mice were generated previously [15,40]. We mated Crxp-Nrl/WT mice with the rd7 mice (procured from Jackson Laboratory) to generate Crxp-Nrl/rd7 mice. The mice, used for analysis reported here, were in a mixed background of 129X1/SvJ and C57BL/6J. PCR primers for genotyping the Crxp-Nrl/WT allele are: F: 5’-AGCCAATGTCACCTCCTGTT-3’ and R: 5’-GGGCTCCCTGAATAGTAGCC-3’. PCR primers for genotyping the rd7 allele are as reported [27]. All studies involving mice were performed in accordance with institutional and federal guidelines and approved by the University Committee on Use and Care of Animals at the University of Michigan.

Gene profiling and analysis

The details of microarray analysis have been described earlier [59,61,62]. Briefly, total retinal RNA was isolated (Trizol, Invitrogen) from one P28 mice (two eyes) and used to generate double-stranded cDNA for hybridization to mouse GeneChips MOE430.2.0, per guidelines (Affymetrix). Retinal RNA from four independent WT, Nrl^−/−, rd7, Crxp-Nrl/WT, and Crxp-Nr2e3/WT mice was used for each evaluation. RMA (Robust Multichip Average) was used to normalize and obtain gene expression scores [30]. Normalized data were subjected to two-stage analysis based on false discovery rate with confidence interval (FDRCI) which controls for both statistical and biological significance when identifying differentially expressed genes [62]. The differentially expressed genes were further classified into distinct functional categories using NCBI and FatiGO (http://fatigo.bioinfo.cipf.es).

Immunohistochemistry

Retinal whole mounts and 10 µm sections were probed with the following antibodies [16,49]: rabbit S-opsin, rabbit M-opsin, and rabbit cone-arrestin (mCAR; generous gift from C. Craft, University of Southern California, Los Angeles, CA, and Chemicon), mouse anti-rhodopsin (1D4 and 4D2; generous gift from R. Molday, University of British Columbia, Vancouver, Canada). The secondary antibodies for fluorescent detection were AlexaFluor 488 and 546 (Molecular probes, Invitrogen). Sections were visualized using an Olympus FluoView 500 laser scanning confocal microscope. Images were subsequently digitized using FluoView software version 5.0.

EMSA

The electrophoretic mobility shift assays were performed using established methods [28], with minor modifications. Nuclear protein extracts from transfected COS-1 cells were prepared using a commercial kit (Active motif, Carlsbad, CA), and expression of NRL protein was confirmed by SDS-PAGE followed by immunoblotting. Nuclear extracts were incubated with 1 µg poly (dI-dC) at 4°C for 15 min in the binding buffer (12 mM HEPES [N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid], pH 7.9; 60 mM KCl; 4 mM MgCl₂; 1 mM EDTA [ethylenediaminetetra acetic acid]; 12% glycerol; 1 mM dithiothreitol; and 0.5 mM phenylmethylsulfonyl fluoride [PMSF]). Then, ³²P-labeled double-stranded oligonucleotide (40,000 cpm) was added and the reaction was incubated at 4°C for 20 min. The DNA oligonucleotide (−2820 nt to −2786 nt: NRE F5’-TGGCCTCTGGTGGCTTTGTCAGCAGTTCCAAGGCT-3’, NRE R 5’-AGCCTTGGAACTGCTGACAAAGCCACCAGAGGCCA-3’) contains a putative NRL-response element (NRE) (underlined) that is predicted by Genomatix. In competition studies, nuclear extracts were pre-incubated with 50 or 100X unlabeled oligonucleotide for 30 min at room temperature and incubated with labeled oligonucleotide at room temperature for 20 min. A mutant oligonucleotide (F: 5’-TGGCCTCTGGTGGCTTTATTTGCAGTTCCAAGGCT-3’, R: 5’-AGCCTTGGAACTGCAAATAAAGCCACCAGAGGCCA-3’) with three nucleotides changed in the NRE site was also used to compete for the protein binding to the oligonucleotide. In order to immunologically identify the components in protein-DNA complexes, nuclear extracts were incubated with 2 µg of the anti-NRL antibody or normal rabbit IgG for 30 min at room temperature, followed by the addition of labeled oligonucleotide and a further incubation for 20 min at room temperature. The reaction mixtures were electrophoresed on 6% polyacrylamide gels at 175 volts for 2.5 hr and subjected to autoradiography.

ChIP

The chromatin immunoprecipitation assays were performed using a commercial kit (Active motif, Carlsbad, CA). Briefly, four snap-frozen retinas from wild type C57BL/6J mice were cross-linked for 15 min at room temperature with 1% formaldehyde in PBS containing protease inhibitors [40]. The reaction was stopped by adding glycine (125 mM), followed by 5 min incubation at room temperature. The remaining steps were performed according to the manufacturer’s instructions, using anti-NRL polyclonal antibody or normal rabbit IgG. ChIP DNAs were used for PCR amplification of a 248-bp fragment (−2989nt to −2742nt), containing a putative NRE (as determined by Genomatix), with primers 5’-GCATGCACTGTTCAAACACC-3’ and 5’-GATAGGCTGTGCAGGGGTTA-3’. PCR with another pair of primers (5’-TGTCCTGAGTCTCC CTGCTT -3’ and 5’- TAAGGCTGGCCAT AAAGTGG -3’) that amplifies a 209-bp fragment (1230 nt to 1438 nt) located about 4 kb downstream from the NRE site, served as a negative control.

ERG

Electroretinograms (ERGs) were recorded from 2–3 month old adult mice. Animals were dark-adapted for at least 12 hours before intraperitonial administration of ketamine (93 mg/kg) and xylazine (8 mg/kg). After pupil dilation with topical 1% atropine and 0.5% tropicamide, corneal ERGs were recorded from both eyes using gold wire loops with 0.5% tetracaine topical anesthesia and a drop of 2% methylcellulose for corneal hydration. A gold wire loop placed in the mouth was used as reference, and the ground electrode was attached to the tail. Body temperature was maintained at 37°C with a heating pad. ERGs were recorded to single xenon white flashes (PS22 Photic Stimulator, Grass Telefactor, West Warwick, RI) presented in a Ganzfeld bowl. Responses were amplified at 10,000 gain at 1 to 1000 Hz (CP511 AC amplifier, Grass Telefactor), and digitized at a rate of 32 KHz. A notch filter was used to remove 60 Hz line noise. Stimulus intensity was attenuated with neutral density filters and ERGs were recorded to increasing intensity (−6.0 to 1.09 log cd-s/m²). Scotopic ERGs were recorded at 3 to 60 second interstimulus intervals depending on the stimulus intensity and responses were computer averaged with at least 20 averages at the lower intensities. Animals were then light adapted for 10 min by exposure to a white 32 cd/m² rod saturating background, and photopic ERGs were recorded for single flash white stimuli over a 2 log unit range. The a-wave was measured from the pre-stimulus baseline to the initial trough. B-waves were measured from the trough of the a-wave when present or from the baseline to the b-wave maximum. A second recording system was used to record S- and M-cone ERGs (Espion e2, Diagnosys LLC, Lowell, MA). Light-adapted ERGs were recorded on a 40 cd/m² background to a xenon flash and a UV filter (360 nm peak; Hoya U-360 filter, Edmund Optics, Barrington, NJ). M-cone ERGs were isolated using a green light-emitting diode (510 nm peak) on the Espion e2. The flash energy was adjusted to elicit responses of approximately equal amplitude for the two wavelengths in WT mice. These stimuli were then used to record S- and M-cone ERGs in Nrl^−/− and in Crxp-Nrl/rd7 mice.

ABBREVIATIONS

ChIP

chromatin immunoprecipitation

EMSA

electrophoretic mobility shift assay

ERG

electroretinography

NRE

NRL-response element

rd

retinal degeneration

nt

nucleotide