Skip Navigation


Human Molecular Genetics Advance Access originally published online on May 26, 2004
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Material
Right arrow All Versions of this Article:
13/14/1487    most recent
ddh160v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (41)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Yoshida, S.
Right arrow Articles by Swaroop, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Yoshida, S.
Right arrow Articles by Swaroop, A.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Human Molecular Genetics, 2004, Vol. 13, No. 14 1487-1503
DOI: 10.1093/hmg/ddh160
Human Molecular Genetics, Vol. 13, No. 14 © Oxford University Press 2004; all rights reserved

Expression profiling of the developing and mature Nrl–/– mouse retina: identification of retinal disease candidates and transcriptional regulatory targets of Nrl

Shigeo Yoshida1,{dagger},{ddagger}, Alan J. Mears1,{dagger},, James S. Friedman1, Todd Carter4, Shirley He1, Edwin Oh1, Yuezhou Jing2, Rafal Farjo1,§, Gilles Fleury5, Carrolee Barlow4,||, Alfred O. Hero2 and Anand Swaroop1,3,*

1Department of Ophthalmology and Visual Sciences, 2Departments of EECS, Biomedical Engineering and Statistics and 3Department of Human Genetics, University of Michigan, Ann Arbor, MI, USA, 4The Salk Institute for Biological Studies, San Diego, California, USA and 5Service des Mesures Ecole Supérieure d'Electricité, Gif-sur-Yvette, France

Received March 20, 2004; Accepted May 14, 2004


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
The rod photoreceptor-specific neural retina leucine zipper protein Nrl is essential for rod differentiation and plays a critical role in regulating gene expression. In the mouse retina, rods account for 97% of the photoreceptors; however, in the absence of Nrl (Nrl–/–), no rods are present and a concomitant increase in cones is observed. A functional all-cone mouse retina represents a unique opportunity to investigate, at the molecular level, differences between the two photoreceptor subtypes. Using mouse GeneChips (Affymetrix), we have generated expression profiles of the wild-type and Nrl–/– retina at three time-points representing distinct stages of photoreceptor differentiation. Comparative data analysis revealed 161 differentially expressed genes; of which, 78 exhibited significantly lower and 83 higher expression in the Nrl–/– retina. Hierarchical clustering was utilized to predict the function of these genes in a temporal context. The differentially expressed genes primarily encode proteins associated with signal transduction, transcriptional regulation, intracellular transport and other processes, which likely correspond to differences between rods and cones and/or retinal remodeling in the absence of rods. A significant number of these genes may serve as candidates for diseases involving rod or cone dysfunction. Chromatin immunoprecipitation assay showed that in addition to the rod phototransduction genes, Nrl might modulate the promoters of many functionally diverse genes in vivo. Our studies provide molecular insights into differences between rod and cone function, yield interesting candidates for retinal diseases and assist in identifying transcriptional regulatory targets of Nrl.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
The mammalian retina contains a diverse array of anatomically and functionally distinct neurons (1). Rod and cone photoreceptors account for >70% of all cells in the retina. In most mammals, rods are almost 20-fold more in number compared with cones though their distribution may vary greatly in different regions (2). Photoreceptors are highly metabolically active post-mitotic neurons; it is estimated that almost 10 billion opsin molecules are synthesized per second in each human retina (3). Hence, it is not surprising that altered expression or function of opsin and other phototransduction proteins results in photoreceptor degeneration (4–6). The transcriptional regulatory networks underlying photoreceptor differentiation and function are understood poorly.

The neural retina leucine zipper (Nrl) protein, a transcription factor of the Maf-subfamily, is expressed specifically in the rod photoreceptors of the retina (7,8) and the pineal gland (A.J. Mears and A. Swaroop, unpublished data). Nrl has been shown to interact with the retina-specific homeodomain protein Crx (9) and regulate the expression of rhodopsin (10) and rod cGMP-phosphodiesterase {alpha}- (11) and ß-subunits (12). In humans, missense mutations of NRL are associated with autosomal dominant retinitis pigmentosa (13–17), and in at least one instance (Ser50Thr mutation), the disease may be a result of increased activity of the NRL protein. Targeted deletion of Nrl in mice results in a complete loss of rods and a supernormal S-cone function, as demonstrated by histology, immunocytochemistry, ERG and expression analysis (18). These observations led to the hypothesis that Nrl plays a critical role in the differentiation of rod photoreceptors, and in its absence, the immature photoreceptors adopt an S-cone phenotype (18). The retina of the Nrl–/– mouse exhibits similarities to the Nr2e3rd7 mouse (19,20) and its corresponding human disease enhanced S-cone syndrome (21). One plausible explanation of the phenotypic overlap is that Nrl directly or indirectly regulates Nr2e3 expression, which is undetectable in the Nrl–/– mouse retina (18).

Although several transcription factors have been implicated in photoreceptor differentiation or gene regulation (22–25), their direct impact on the photoreceptor transcriptome has not been elucidated. Microarray-based global expression profiling of tissues from mice deficient in a transcription factor gene can point to downstream regulatory targets and provide candidate genes for functional studies and cloning of disease loci (26). This approach has been utilized successfully in studies of the mouse retina (27,28). The Nrl–/– retina is particularly amenable to this analysis because of its dramatic phenotype of no rods and enhanced cones. In the retina of Nrl–/– mice, rod bipolar cells have normal morphology, pattern of staining and lamination, and form functional connections with the cones, and the axonal arbors of horizontal cells and AII amacrine cells maintain a normal morphology and stratification pattern (E. Strettoi, A.J. Mears and A. Swaroop, unpublished data). We, therefore, hypothesize that the comparative analysis of gene profiles from the wild-type and Nrl–/– retina will, to a large extent, reveal expression differences between rod and cone transcriptomes. Based on our initial analysis of phototransduction genes (18), we predict that transcripts encoding rod photoreceptor proteins would be expressed at lower levels (or undetectable) in the Nrl–/– retina. Conversely, the transcripts specific to the normally sparse population of cones are expected to be enriched in the Nrl–/– retina.

Here, we report the gene expression profiles, obtained by using Affymetrix GeneChips (MGU74Av2), of the wild-type and Nrl–/– retina at three time-points (post-natal days 2 and 10, and 2 months). After data normalization by robust multichip average algorithm (RMA) (29) and ranking the statistically validated genes with a minimum 1.5 average fold-change (AFC) in expression, we have identified 161 differentially expressed genes, which include the known rod- or cone-specific genes represented on these chips. Functional annotation suggests a wide spectrum of physiological changes that likely correspond to differences between rods and cones and/or remodeling of retina in the absence of rods. Our analysis suggests that ~25% of all differentially expressed genes identified in this study are either associated with (15) or are candidates (26) for retinal diseases. Using chromatin immunoprecipitation (ChIP) analysis, a significant proportion of the top ranked genes showing reduced expression in the Nrl–/– retina are demonstrated to be putative direct targets of Nrl, indicating the breadth of its influence on the rod transcriptome.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
Identification of differentially expressed genes in the Nrl–/– retina
The three time-points, P2, P10 and 2 months, were chosen to cover distinct critical stages of photoreceptor development in mouse. In the wild-type retina, at P2 many retinal progenitor cells are still exiting the cell cycle and a majority of these will become rods (30). The photopigment of the rod photoreceptors, rhodopsin, is first detected at P4. At P10, retinogenesis is complete, the cells are undergoing terminal differentiation and photoreceptor outer segments are beginning to form. We chose 2 months of age as another suitable time-point as the retina is structurally and functionally matured and yet old enough to avoid any potential delayed differentiation effects which may occur due to the re-specification of the photoreceptor cell types in the Nrl–/– retina.

To facilitate statistical analysis and identification of ‘true’ positives, four replicate MGU74Av2 GeneChips were utilized for each time-point and strain. Based on absent/present calls (MAS5 analysis), ~60% of the probesets (out of ~12 400) were reported as present or marginally detectable in at least one of the 12 wild-type GeneChips, consistent with other studies analyzing single tissue types. Signal quantification and normalization were performed using RMA, a reliable and effective algorithm in control studies (31,32). The normalized data were then analyzed with a robust two-step procedure to identify statistically significant differentially expressed genes. Due to the tendency of microarrays to quantitatively underestimate fold-change in expression and since RMA normalization compresses the signals (and resulting ratios), an empirical 1.5 AFC cut-off was selected as the minimum fold-change (minfc) for statistical analysis. Using these criteria and after removal of those scored as absent on all 24 chips, a total of 173 probesets were reported as differentially expressed for at least one of the three time-points (i.e. P-value <1). Of these, 86 show decreased (Table 1, down-regulated genes) and 87 increased (Table 1, up-regulated genes) expression in the Nrl–/– retina. The differentially expressed genes are ranked based on increasing false discovery rate confidence interval (FDRCI) P-values, which are similar to FDR P-values except that they account for a specified minfc level in addition to a level of statistical significance. Although the highly differentially expressed genes are near the top of the lists as expected, the order is based on both the AFC and the variability of the signal data across the GeneChips. For this reason, probesets displaying a relatively high AFC for a given time-point may still be reported as non-significant [e.g. Nt5e (down-regulated) and Fin15 (up-regulated) at P10 in Table 1]. After removing probesets that belong to the same gene and show similar gene expression profile, a non-redundant set comprising 78 down-regulated and 83 up-regulated genes is obtained. Almost 90% of these genes are categorized as ‘known’, whereas 18 are novel sequences that are represented currently only in expressed sequence tag (EST) or genomic sequence databases.


View this table:
[in this window]
[in a new window]
  		Table 1. Summary of microarray-based expression
 
Validation by quantitative real-time PCR
Fifty-four different gene/time-point values spanning a broad spectrum of AFC and FDRCI rankings were examined by quantitative real-time PCR (Q-PCR) (Table 2). There is a good correlation (R2=0.91, data not shown) between AFC reported by microarray and by Q-PCR. Underestimation of the relative degree of fold-change in microarray data is likely due to background noise and limited sensitivity that restricts the dynamic range of this hybridization-based technique. Only three genes (Gas5, Sox11 and 1110002B05Rik) showed disagreement between the two methods (94% validation rate). The discrepancy could be due to the existence of multiple isoforms, which have been identified for these genes. The importance of validation is evident, not only for identifying possible false positives but also for determining the relative fold-change in transcripts (i.e. biological change) compared with the AFC reported by microarray. For example, Guca1a and Kibra are both predicted to be moderately up-regulated (5.6 and 6.0, respectively) in the Nrl–/– retina; however, Guca1a is shown to be up-regulated 5.5-fold by Q-PCR (same as microarray) but Kibra 26-fold (5-fold underestimate by microarray). Similar examples are evident amongst the down-regulated genes. Q-PCR analysis using additional retinal samples for six of the genes revealed similar AFCs (data not shown).


View this table:
[in this window]
[in a new window]
  		Table 2. Real-time Q-PCR validation
 
Hierarchical clustering and functional annotation
Relative expression profiling across multiple developmental time-points can provide information on the potential role of a given gene in the context of known biological events occurring within that time frame. Comparison of relative profiles can allow clustering of genes into groups that show similar patterns of behavior. To compare expression patterns between all 161 differentially expressed genes, the average signals from the four replicate GeneChips were first normalized to z-scores, and then run through a hierarchical clustering algorithm. Ten major clusters were identified by visual inspection, and Gene Ontology was used to assign functional annotation of 101 genes (62%) (Fig. 1).



View larger version (79K):
[in this window]
[in a new window]
  		Figure 1. Hierarchical clustering and functional annotation. Clustering of differentially expressed genes based on normalized average signals (z-scores). Bright green boxes indicate lowest signal with increasing values indicated by darkening color towards bright red, representing peak signal. The set of 161 non-redundant genes are divided, based on similarity of profiles, into 10 clusters. (A) Down-regulated genes are in groups I–VI and (B) up-regulated in groups VII–X. Functional annotation is based on defined biological processes assigned by the Gene Ontology (GO) consortium. Though listed as having no defined function, Mtap6 (indicated by an asterisk) is associated with microtubules and has a presumed role in synaptic plasticity and function. The far right column indicates whether the gene is known to be associated with (black square) or is a candidate for (empty square) retinal disease (see Table 3 also).

 
Cluster I contains genes that display a bimodal (peaks at P2 and 2 month) or constant pattern of expression in wild-type, but show significantly decreased expression at P10 or 2 months of age in the Nrl–/– retina. Cluster II contains three {gamma}-crystallin genes (E, D and F); for these, the peak expression is in the wild-type adult retina, but in the Nrl–/– retina there is increased expression at P2 and P10 but a significant decrease at 2 months. Although these genes show AFCs >2-fold at P10, none of these is considered statistically significant (Table 1). This may be due to the signal noise associated with the high degree of sequence identity between different crystallins. Q-PCR confirmed the decreased expression of Crygd and Crygf at 2 months (Table 2). Crystallins are expressed in neural retina and may play a role in stress response (33). For the genes of Cluster V, their expression peaks at P10 but then decreases (though still detectable) in the wild-type adult retina. In the Nrl–/– retina, the peak expression may still be at P10, but is reduced for all these genes, suggesting a potential role in differentiation, as indicated for Ndr1, Ndrl and Lmo1. Cluster VI contains only two ESTs that are expressed across all three time-points but are down-regulated in the Nrl–/– retina.

Almost 80% of genes showing decreased expression in the Nrl–/– retina belong to clusters III and IV. These genes demonstrate an increasing (relative) level of expression, reaching peak expression by P10 (cluster III) or 2 months (cluster IV) in wild-type, suggesting a role in the mature retina/photoreceptors. In the Nrl–/– retina, these genes are down-regulated showing, typically, only a moderate (or no) increase in expression at later time-points. Genes of these clusters are strong candidates for direct positive regulation by Nrl and include Rho, Pde6b and Pde6a (known targets of Nrl) as well as Gnat1 and Gnb1.

The genes showing higher expression in the Nrl–/– retina can be organized into four major clusters. The genes of the largest cluster VIII show an increase in expression at P10 or 2 months in the Nrl–/– retina relative to wild-type. As anticipated, this includes genes encoding proteins with a role in cone-mediated visual function (e.g. Opn1sw and Gnat2). Expression of cluster VII genes peaks at P2 in wild-type, suggesting a primary role in early development, but in the Nrl–/– retina they show elevated expression peaking at P10 or 2 months. Their sustained high expression in the adult retina may be indicative of an aberrant reactivation of gene expression, possibly related to stress, cell death or reactive gliosis. Cluster IX genes show an elevated differential expression (and peak) in the Nrl–/– retina, primarily at P10, and may play a role in cone differentiation. Of the 14 genes in this cluster, six are associated with signaling, development or cell cycle/growth. Cluster X includes genes showing peak expression at P2 in wild-type but the expression declines (often rapidly) by P10 or 2 months, suggesting a primary role in early development. In the Nrl–/– retina, the expression profile is similar but the expression is elevated and maintained for a longer period.

Direct targets of Nrl identified by ChIP
We hypothesized that targets of Nrl will be enriched among the genes exhibiting reduced expression in the Nrl–/– retina. Hence, we examined the enrichment of the promoter regions that include a potential AP-1 like or Nrl-response element (NRE) of ‘candidate Nrl targets’ by ChIP with a polyclonal anti-Nrl antibody (8) using the wild-type mouse retina. Twenty different gene promoters were assayed by PCR amplification; of these, 18 (90%) showed enrichment in the antibody fractions (Nrl-ChIP) over the no antibody control (Fig. 2), demonstrating in vivo promoter occupancy by Nrl. The positive target promoters included three genes (Rho, Pde6b and Pde6a) that are modulated by Nrl. The promoters of other photoreceptor genes (such as Cnga1, Gnat1, Gnb1, Rom1 and Pdc) were also enriched. In addition, a few widely expressed genes, such as Aqp1 (water channel) and adiponectin receptor 1 (AdipoR1), appear to be the target of Nrl regulation in mature rods. It should be noted that although only down-regulated genes were analyzed by ChIP, Nrl might negatively regulate (i.e. repress) the expression of cone-specific genes, much akin to the predicted role of Nr2e3 (21).



View larger version (69K):
[in this window]
[in a new window]
  		Figure 2. ChIP analysis. PCR products are shown for 20 different gene promoters in which the most proximal putative AP-1/NRE-like site was assayed. Each set includes the input genomic DNA as positive control (lane 1), chromatin DNA fraction immunoprecipitated with anti-NRL antibody (lane 2), chromatin DNA obtained without the antibody (background control; lane 3) and water (negative control; lane 4). Enrichment (greater amount of product) with antibody IP over background control (no Ab) indicates the in vivo occupancy of a sequence element within the amplified region by Nrl. All assayed gene promoters showed enrichment, except Wisp1 (inconclusive) and Csda (negative). Direct regulation of Rho and Pde6b by Nrl (at these sites) has been demonstrated previously, and as such these are positive controls for the ChIP protocol.

 
Identification of retinal disease candidate genes
Many genes showing photoreceptor-enriched expression are associated with retinal disease; these encode diverse functions, including phototransduction (e.g. rhodopsin), transcriptional regulation (e.g. Crx and Nrl), outer segment structure (e.g. Rom1 and Prom1) or maintenance of the extracellular matrix (e.g. Rs1h). Expression profiling of a mouse model with retinal degeneration (Rho–/–) was utilized previously to identify a retinitis pigmentosa disease gene (RP10), inosine monophosphate dehydrogenase type 1 (IMPDH1) (34), which was not an obvious candidate due to its ubiquitous expression and role in guanosine nucleotide biosynthesis. We, therefore, determined the chromosomal location of genes that are expressed differentially in the Nrl–/– retina using in silico methods. On the basis of the map position of the human homolog, 41 of the differentially expressed genes (25%) have been associated previously with or are candidates for retinal diseases (Table 3). A few of these (e.g. Mef2c, Nt5e and Cdr2) were also identified in the Rho–/– gene profiling study (34), providing further evidence of their rod-preferred expression. Up-regulated genes that are candidates for macular or cone associated diseases include S100A6, RXRG, ADCY2, NP and SOCS3, whereas down-regulated genes that map to the region of rod associated disease loci (such as RP) include NT5E and CDR2.


View this table:
[in this window]
[in a new window]
  		Table 3. Disease association or candidacy for differentially expressed genes
 
Analysis of differentially expressed genes
Light response and vision.
The genes displaying restricted expression to rods or cones show the most dramatic changes in expression. For the rods, these include genes encoding rod-specific phototransduction proteins such as rhodopsin (Rho), cGMP phosphodiesterase subunits (Pde6a and b), rod transducin subunits (Gnat1 and Gnb1) and the cyclic nucleotide gated channel subunit (Cnga1). By Q-PCR, transcripts of these genes are virtually undetectable in the Nrl–/– retina with expression typically <1% of wild-type. Modest expression of Pde6a (~7%) and Pde6b (~2%) in the adult Nrl–/– retina can be attributed to their expression in non-photoreceptor neurons, as observed for Pde6a (35). Genes encoding cone phototransduction proteins, such as the photopigment S-opsin (Opn1sw), cone transducin subunits (Gnat2, Gnb3 and Gngt2) and the cyclic nucleotide gated channel subunit (Cnga3), show dramatically higher expression in the Nrl–/– retina. A number of genes that are expressed in both photoreceptor subtypes show varying degrees of expression change, which may reflect a moderate quantitative bias towards one class (or expression in multiple cell types). These include guanylate cyclase activator 1a (Guca1a or Gcap1), recoverin (Rcvrn), prominin 1 (Prom1), phosducin (Pdc), retinal S-antigen (Sag), retinal outer segment membrane protein (Rom1) and an ATP-binding cassette (ABC) transporter (Abca4). Guca1a displays a 5.5-fold increase in expression in the Nrl–/– retina suggesting preferential expression in cones. Notably, although expressed in both rods and cones, mutations in this gene are primarily associated with cone or cone–rod dystrophies (36,37). Other down-regulated genes may indicate their preferential expression in rods.

Gene regulation, differentiation and development.
Transcription factors and signaling molecules that are expressed differentially in the Nrl–/– retina may provide insights into the regulatory networks associated with photoreceptor development and/or function. Q-PCR analysis of E14–P21 retina for the cone photopigment Opn1sw (S-opsin) showed that the increase in its expression occurred at P6.5 in the Nrl–/– retina (Fig. 3). This second-wave of cone differentiation likely corresponds to the post-mitotic photoreceptors that are normally destined to become rods. Therefore, it is predicted that genes associated with rod or cone differentiation would be down- or up-regulated, respectively, at this time-point. The expression of MADS-box containing myocyte enhancer factor 2c (Mef2c) (38,39) is reduced in the matured Nrl–/– retina to 20% of the wild-type levels. Zfp36l2 (a C3H-type zinc finger protein) is down-regulated ~8-fold in the Nrl–/– retina. A significant decrease in expression of LIM domain only 1 (Lmo1), a developmentally associated transcription factor, is observed in the adult retina (10-fold by Q-PCR) suggesting its role in mature rods. Similar profiles are also observed for N-myc downstream regulated 1 (Ndr1) and Ndr-like (Ndrl).



View larger version (24K):
[in this window]
[in a new window]
  		Figure 3. Temporal expression of S-opsin in the wild-type and Nrl–/– retina. The profile of relative expression of S-opsin determined by Q-PCR on the developing mouse retinas is shown after normalization to Hprt. Error bars indicate s.e.m.

 
A number of genes encoding transcription regulatory proteins are up-regulated in the Nrl–/– retina. Retinoid X receptor gamma (Rxrg), localized to cones in the adult retina (40) and shown to be induced by retinoic acid (RA) (41), shows 9-fold higher expression in the Nrl–/– retina. Rxrg maps to the region of cone-dystrophy locus CORD8 (Table 3) and is an excellent candidate for this disease. Sal-like 3 (Sall3), a C2H2 zinc finger transcription factor, is required for terminal differentiation of photoreceptors in Drosophila (42); its augmented expression is therefore of considerable interest. Validation by Q-PCR, which detects two of the six alternative transcripts, reveals that Sall3 is highly differentially expressed at P10 (20-fold) but is only moderately increased at 2 months (2-fold), suggesting a potential role in cone differentiation. Engrailed-2 (En2), a homeobox transcription factor, shows sustained expression in the mature wild-type retina but in Nrl–/– retina it is highly elevated (30-fold increase). The positive regulatory domain zinc finger protein, Prdm1, shows elevated expression (8-fold) in the matured Nrl–/– retina. It is expressed earlier in the wild-type retina and is undetectable in the adult.

Apoptosis and stress response
Several genes encoding proteins associated with stress response or apoptosis exhibit decreased expression in the Nrl–/– retina; these include the chaperone heat shock proteins Hsp70.3 (Hspa1a) and Hsp70.1 (Hspa1b). Serum/glucocorticoid regulated kinase (Sgk), which shows peak expression in the adult retina and is down-regulated in the Nrl–/– retina, is shown to be anti-apoptotic and induced in response to multiple forms of stress in epithelial cells (43). Tumor necrosis factor alpha induced protein 3 (Tnfaip3), which inhibits NF-kappa B (Nfkb1) (44), has been associated with light-induced photoreceptor degeneration (45). Tnfaip3 is first detected at P10, and its expression peaks at 2 months. In contrast, Nfkb1 expression is relatively constant in the wild-type retina but exhibits a moderate peak at P2. In the Nrl–/– retina, Tnfaip3 is down-regulated 8-fold, whereas its inhibitory target, Nfkb1, is up-regulated. This observation, may at least in part, provide clues to the mechanism through which stress response and cell death may be mediated in the Nrl–/– retina during late stages (unpublished data). Caspase-7, which is detected in the wild-type retina during development, is the only caspase showing elevated (10-fold) expression in the adult Nrl–/– retina.

Calcium homeostasis and retinal function
During the recovery of light response in photoreceptors, cGMP is regulated by cytoplasmic Ca2+ via Guca1a (or Gcap1). Both Guca1a and rod arrestin (Sag) are associated with retinal diseases and are expressed differentially in the Nrl–/– retina. Calcium/calmodulin-dependent kinase II beta (Camk2b) is up-regulated (15-fold) in the Nrl–/– retina. Calcyclin (S100a6) is expressed highly in neurons (46) and shows elevated levels in the Nrl–/– retina. The human homolog of this gene maps to a cone–rod dystrophy locus (CORD8). S100a6 is regulated by NF-kappaB (47), which is also augmented in the Nrl–/– retina. Two calcium channels genes Trpc1 and Cacnb2 are down-regulated in the Nrl–/– retina. Syntrophin acidic 1 (Snta1) is a component of the dystrophin glycoprotein complex (DGC) which may play a significant structural and signaling role (neurotransmission) in the retina (48). Mutations of dystrophin or disruption of the DGC may account for scotopic (rod response) defects in patients with Duchenne muscular dystrophy (49), consistent with rod-enrichment of Snta1 and its down-regulation in the Nrl–/– retina.

Melatonin signaling.
Retinal melatonin, acts as a local neuromodulator through the melatonin receptors, which then may control the release of dopamine (50). Three genes of the melatonin pathway, tryptophan hydroxylase (Tph1), dopamine receptor 4 (Drd4) and melatonin receptor 1a (Mtnr1a), are expressed differentially in the Nrl–/– retina. Tph1 is the first enzyme in the biosynthetic pathways of melatonin in the photoreceptors and is believed to be synthesized primarily in the cones (51), consistent with its up-regulation in the Nrl–/– retina. The melatonin receptor 1a, which normally shows peak expression around P2–P4, is highly elevated in the Nrl–/– retina, and peaks at P8 before rapidly decreasing in expression. The dopamine receptor Drd4, which plays a role in regulating cAMP metabolism, is not highly expressed until P10–P12 in the wild-type retina (52), but is down-regulated to <10% of the wild-type levels in the P10 Nrl–/– retina, indicating a role in rods.

Novel functions and novel genes
Although a majority of the differentially expressed genes have a defined function, in many cases their specific role in the retina or their possible bias towards rods or cones is not understood. Deleted in polyposis 1-like 1 (Dp1l1) is the top FDRCI ranked down-regulated gene and is expressed at <3% of the wild-type levels. It shows peak expression in the adult retina and is detected in the outer nuclear layer (data not shown) but its function is unknown. A function can be inferred but is not known for calcium activated chloride channel 3 (Clca3), which is up-regulated 44-fold in the Nrl–/– retina. Kibra is a novel WW-domain containing protein expressed primarily in brain and kidney (53) and is up-regulated 26-fold in the Nrl–/– retina. In addition, 18 of the differentially expressed genes identified by microarray analysis match only ESTs. These novel genes could provide new leads for elucidating retinal development and function.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
Expression profiling and data mining
Appropriate microarray design and data analysis are essential for extracting meaningful results in genome-wide expression profiling studies (54). We utilized RMA for normalization (29,32) and chose an AFC cut-off of 1.5. A new two-stage gene filtering procedure (55) was applied that controls both FDR and minfc levels. This procedure is based on construction of a set of simultaneous FDRCI on the temporal fold-changes of each gene. Genes having at least one confidence interval that covers a range of fold-changes larger than the specified AFC cut-off, which we call minfc, are declared significant at the specified FDRCI level. As FDRCI is more stringent than FDR, the associated significance levels are generally not as high as those of the FDR procedure. For each minfc level studied, the two-stage procedure was used to generate a list of genes ranked according to decreasing FDRCI significance or, equivalently, increasing FDRCI P-value. For an AFC cut-off of 1.5, the complete ranked list, excluding probesets having FDRCI P-values >0.99, consisted of 173 probesets. Of the 54 data points tested by Q-PCR, 51 (94%) were verified. If the minfc is reduced to 1.25, the probeset list is expanded to over 300 probesets (see Supplementary Material, Table A). These additional genes may display a reduced validation rate by Q-PCR but add to cluster analysis and pathway construction based on the microarray data. Replicate experiments and statistical analysis are critical for extracting such probesets.

Temporal profiling and clustering analysis add a new dimension for predicting the functional role, possible interactions and regulatory relationships that may exist amongst the genes that are being analyzed. Our studies should identify the genes that are presumably associated with photoreceptor development (P2), terminal differentiation (P10) and function (2 month). Although our data are based on a mixed cell population (whole retina), the generated profiles are dominated by photoreceptors (about 70% of total cells) and can direct future studies to prioritize candidate genes of interest for positional cloning or functional analysis. Of particular interest are the differentially expressed genes encoding proteins associated with visual process, transcriptional regulation, signal transduction and development, as they may provide insights into the regulatory networks and signaling pathways underlying the differences between rods and cones.

Genes encoding metabolism-related proteins represented the single largest class of differentially expressed genes (24%) in the Nrl–/– retina. In addition, one-third of the genes are associated with light response/vision (11%), signaling (18%) and transcription (6%). There was no significant difference between up- and down-regulated genes in terms of the specific biological processes affected; however, more genes associated with vision or cell adhesion are down-regulated in the Nrl–/– retina (Fig. 4). This can be attributed to greater representation of rod- rather than cone-specific transcripts on the MGU74Av2 Chips. A decreased expression of genes encoding structural proteins may reflect the abnormalities of the retinal organization in the Nrl–/– mouse. It should be noted that cones contain more mitochondria when compared with rods (56,57); expression changes in mitochondria associated genes (Aqp1, Mscs, Skd3 and Clic4) may therefore reflect numerical and physiological differences between the populations of mitochondria in the two classes of photoreceptors.



View larger version (37K):
[in this window]
[in a new window]
  		Figure 4. Biological processes associated with differentially expressed genes. (A) Overall distribution of all differentially expressed genes, (B) a comparison between up- and down-regulated genes.

 
Expanding the data set: MOE430 GeneChips and custom cDNA arrays
The MGU74Av2 GeneChip contains over 12 000 known genes and ESTs but the retina-specific transcripts are represented poorly. For example, neither Nr2e3 or cone arrestin are on these arrays. Affymetrix has since significantly improved the mouse arrays and the new MOE430 GeneChips now comprise over 36 000 genes and ESTs. These arrays are superior in design showing greater sensitivity and improved specificity of probesets. One problem with GeneChips is that the probesets are based on public databases and if transcripts are exclusively or predominantly expressed in the retina, they may not have been identified. Custom retinal cDNA arrays (28,58–60) should therefore complement GeneChip-based analysis of the Nrl–/– retina (J. Yu and A. Swaroop, unpublished data).

Differential expression and reactive gliosis in the Nrl–/– retina
The ready-extraction of rod- or cone-specific genes from the microarray analysis is complicated by the fact that the Nrl–/– retina undergoes a slow form of retinal degeneration (after 4–6 months, unpublished data). A marker of retinal stress, glial fibrillary acidic protein (Gfap), is up-regulated in the Nrl–/– retina (18). Reactive gliosis or glial hypertrophy is observed as part of the complex neuronal remodeling that occurs during retinal degeneration (61,62). Discrimination between photoreceptor-based differential expression and changes due to retinal remodeling must be evaluated carefully, especially when dealing with genes that encode proteins with a poorly defined function. One experimental strategy would be to compare gene profiles, reported here, to those of mouse models of retinal degeneration.

Cones or ‘cods’
In the original characterization of the Nrl–/– mouse, the photoreceptor population was referred to as ‘cods’ as there was uncertainty as to whether the later developing but functional cones were in fact a type of hybrid photoreceptor. Subsequent analysis with cone-specific markers (such as PNA), suction electrode recordings of isolated photoreceptors (S.S. Nikonov, L. Daniele, A.J. Mears, A. Swaroop and E.N. Pugh Jr, unpublished data) and ERG of whole retina, nuclear morphology of the ONL (punctate staining typical of cones) and extensive molecular studies are all consistent with these photoreceptors being cones. Histologically, the retina is abnormal with rosettes and whorls disrupting the ONL, and short, sparse and disorganized OS. These changes, however, may be a consequence of inappropriate nuclear and OS packing within ONL and the sub-retinal region, and may be secondary to the actual identity and differentiation of the photoreceptors. The gene profiling data, presented here, provide strong evidence in favor of the photoreceptors of the Nrl–/– retina being cones and not cods.

Photoreceptor plasticity and identity
In the absence of Nrl, the failure of the retinal photoreceptors to adopt their appropriate rod identity results in their transformation into cones primarily expressing S-opsin (S-cones). Nrl therefore appears to act as a molecular switch during photoreceptor differentiation by promoting the rod differentiation program while simultaneously repressing the cone identity. The suppression of the cone fate is achieved, at least in part, through direct or indirect regulation of the transcription factor Nr2e3 (20,21), whose expression is undetectable in the Nrl–/– retina (18).

How does Nrl orchestrate the coordinated expression of a broad array of genes that are required for making a mature and functional rod? Delineation of direct downstream targets is the essential first step towards assembling the Nrl-mediated transcriptional regulatory network(s) underlying rod differentiation. Our study has identified several potential direct targets of Nrl by a combined approach of microarray profiling and ChIP. Several of these are known or putative transcription factors or signaling proteins that may play a role in rod or cone differentiation. Comparative retinal gene profiling studies of mouse loss-of-function mutants of other photoreceptor transcription factors (e.g. Crx, Trß2, Nr2e3) should provide considerable insights into the gene regulatory networks that govern differentiation and homeostasis.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
SUPPLEMENTARY MATERIAL
 REFERENCES
 
Animal use and tissue collection
University Committee on Use and Care of Animals of the University of Michigan approved all procedures involving mice. Both the Nrl–/– mice and the wild-type controls were of a matched mixed genetic background (R1 and C57BL/6 strains) (18). Mice were sacrificed by cervical dislocation, and the retinas were excised rapidly, frozen on dry ice and stored at –80°C. No signs of pathology were detected in any of the animals used. To isolate sufficient total RNA for labeling protocols, retinas from two mice were pooled into a single sample. To minimize false positives due to biological variation, different samples were utilized for four replicate experiments per genotype/time-point (biological replicates). For the developmental Q-PCR studies, retinas were dissected from the embryos of timed-pregnant Nrl–/– or wild-type females and pooled. Retinas from post-natal time-points were also pooled (entire litter) after dissection.

RNA preparation
Tissues were placed into TRIzol (Invitrogen, Carlsbad, CA, USA) (added to the frozen tissues at ~1.3 ml per four retinas) and homogenized (Polytron, Kinematica, Lucerne, Switzerland) at maximum speed for 120 s. Subsequent steps were done according to the manufacturer's instructions.

Gene expression analysis
The GeneChips (Affymetrix, Santa Clara, CA, USA) used in the study contained ~12 000 probe sets, corresponding to over 6000 genes and 6000 ESTs (Murine Genome U74A Array v2).

Total retinal RNA was used to generate double-stranded cDNA (ds-cDNA) with SuperScript Choice System (Invitrogen) and oligo-dT primer containing a T7 RNA polymerase promoter. After second-strand synthesis, the reaction mixture was extracted with phenol–chloroform–isoamyl alcohol, and ds-cDNA was recovered by ethanol precipitation. In vitro transcription was performed by using a RNA transcription labeling kit (Enzo) with 10 µl of ds-cDNA template in the presence of a mixture of unlabeled ATP, CTP, GTP and UTP and biotin-labeled CTP and UTP [bio-11-CTP and bio-16-UTP (Enzo Life Sciences, Farmingdale, NY, USA)]. Biotin-labeled cRNA was purified by using an RNeasy affinity column (Qiagen, Valencia, CA, USA), and fragmented randomly to sizes ranging from 35 to 200 bases by incubating at 94°C for 35 min. The hybridization solutions contained 100 mM MES, 1 M NaCl, 20 mM EDTA and 0.01% Tween-20. The final concentration of fragmented cRNA was 0.05 µg/µl in the hybridization solution. After hybridization, the solutions were removed and GeneChips were washed and stained with streptavidin–phycoerythrin. GeneChips were read at a resolution of 6 µm with a Hewlett-Packard GeneArray Scanner. Initial data preparation (i.e. generation of CHP files) were performed by Affymetrix MICROARRAY SUITE v5.0. Normalization (quantile method) and calculation of signal intensities were performed with the software package RMA from the R project (http://www.r-project.org/). Data were based on four Affymetrix MGU74Av2 GeneChips (biological replicates) for each time-point per genotype (i.e. total of eight GeneChips per timepoint). Of the total 24 GeneChips, only one had to be repeated due to a negative quality report based on raw image and MAS5 analysis. Ratios of average signal intensity (log2) were then calculated for the probesets (Nrl–/– relative to wild-type) and then converted to an AFC. Statistical validation was performed on probesets showing a minimum AFC of 1.5. If due to low signal, any of these probesets were reported as having an absent signal (based on MAS5) in all GeneChips (i.e. for both genotypes) for a given time-point then it was reported as absent and reported signal values and relative expressions were ignored.

FDR and P-values
The statistical method used to assign P-values to the fold-changes of gene responses is a two-step procedure based on the Benjamini and Yekutieli construction of FDRCI (63–65) on the fold-changes between the Nrl–/– and the wild-type response profiles (55). FDRCIs are (1–q)% confidence intervals where the level ‘q’ is corrected for error amplification inherent to performing multiple comparisons on many genes and many time-points. For specified minimum fold-change (fcmin) and a given level of significance q, a gene response is declared as ‘positive’ if the range of the FDRCI is either greater than fcmin (positive fold-change) or less than –fcmin (negative fold-change). The FDRCI P-value for a given gene is defined as the minimum level q for which the gene's FDRCI does not intersect the interval [–fcmin, fcmin]. For this data, we formed a ranked list of genes according to increasing FDRCI significance level having minfc of 1.5 (0.58 log2). All probesets with a P-value <1 were reported.

Q-PCR
RNA was treated with RQ1 DNAse (Promega, Madison, WI, USA) following manufacturer's guidelines. Oligo-dT-primed reverse transcription was performed using 2.5 µg of DNAse-treated total retinal RNA with Superscript II (Invitrogen). Primers for the validated genes were designed typically from the 3' UTR region using Primer 3 (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3). The PCR reactions on the cDNA template were then performed in triplicate in an I-cycler thermocycler with optical module (BioRad, Hercules, CA, USA). Amplified products were quantified based on the level of fluorescence of SybrGreen I (Molecular Probes, Eugene, OR, USA) in each reaction. Specificity of reactions was confirmed by melt curve analysis and gel electrophoresis. AFCs were then calculated based on the difference in the threshold cycles (Ct) between the Nrl–/– and the wild-type samples after normalization to Hprt.

Clustering analysis
Clustering based on similarity of temporal expression profiles and visualization was performed using the software program Spotfire DecisionSite 7.2 (www.spotfire.com). The signal data of statistically significant differentially expressed genes were standardized to z-scores (66), and hierarchical clustering performed using the ‘Euclidean distance’ method.

Annotation
Functional annotation of proteins was assigned through Gene Ontology (http://www.geneontology.org) or Locuslink (http://www.ncbi.nlm.nih.gov/LocusLink) classifications obtained through appropriate public databases such as NetAffx (http://www.netaffx.com/indexp2.jsp) (67) and DAVID (http://apps1.niaid.nih.gov/david/upload.asp) (68).

ChIP analysis
Retinas were obtained from the C57BL/6 wild-type mice and snap frozen on dry ice. ChIP was performed using a commercial assay kit (Upstate Biotechnologies, Charlottesville, VA, USA). Briefly, four retinas were crosslinked in PBS containing proteinase inhibitors and a final concentration of 1% formaldehyde for 15 min at 37°C. The retinas were washed four times in ice-cold PBS with proteinase inhibitors and then incubated on ice for 15 min. The tissue was then sonicated on ice with 10 pulses of 20 sec. The remaining steps were performed as described by the manufacturer, using an anti-NRL polyclonal antibody (8).

The putative promoter region for each of the genes analyzed was determined using in silico methods (http://www.ncbi.nlm.nih.gov/mapview). Each promoter DNA sequence was analyzed using Matinspector (http://www.genomatix.de/index.html) and PCR primers were designed to flank putative AP1-like sites either predicted by Matinspector or predicted manually. If there was more than one AP-1 like site, the sequence element closest to the 5' untranslated sequence was used. Equal amounts of input DNA, with and without antibody, were used in each PCR reaction.


   SUPPLEMENTARY MATERIAL
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
REFERENCES
 
Supplementary Material is available at HMG Online.


   ACKNOWLEDGEMENTS
 
The authors thank S. Zareparsi, M.I. Othman and S.P. MacNee for discussions, and Sharyn Ferrara for administrative assistance. This research was supported by grants from the National Institutes of Health (EY11115 including administrative supplements, EY07003), The Foundation Fighting Blindness (Owings Mills, MD, USA) and Research to Prevent Blindness (RPB; New York, NY, USA). A.J.M is a recipient of a Tier 2 Canada Research Chair. J.S.F. is a recipient of an FFB-Canada post-doctoral fellowship. A.S. is Harold F. Falls Collegiate Professor and RPB Senior Scientific Investigator.


   FOOTNOTES
 
* To whom correspondence should be addressed at: Department of Ophthalmology and Visual Sciences and Department of Human Genetics, W.K. Kellogg Eye Center, University of Michigan, 1000 Wall Street, Ann Arbor, MI 48105-0714, USA. Tel: +1 7347633731; Fax: +1 7346470228; Email: swaroop{at}umich.edu

{dagger} The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors. Back

{ddagger} Present address: Department of Ophthalmology, Kyushu University Graduate School of Medical Sciences, Fukuoka, Japan. Back

Present address: University of Ottawa Eye Institute and Ottawa Health Research Institute, Ottawa, Ontario, Canada. Back

§ Present address: Department of Cell Biology, University of Oklahoma Health Sciences Center, Oklahoma City, OK, USA. Back

|| Present address: Merck Research Laboratories, San Diego, CA, USA. Back


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 

  1. Masland, R.H. (2001) The fundamental plan of the retina. Nat. Neurosci., 4, 877–886.[CrossRef][ISI][Medline]

  2. Curcio, C.A., Sloan, K.R., Kalina, R.E. and Hendrickson, A.E. (1990) Human photoreceptor topography. J. Comp. Neurol., 292, 497–523.[CrossRef][ISI][Medline]

  3. Williams, D.S. (2002) Transport to the photoreceptor outer segment by myosin VIIa and kinesin II. Vision Res., 42, 455–462.[CrossRef][ISI][Medline]

  4. Tan, E., Wang, Q., Quiambao, A.B., Xu, X., Qtaishat, N.M., Peachey, N.S., Lem, J., Fliesler, S.J., Pepperberg, D.R., Naash, M.I. et al. (2001) The relationship between opsin overexpression and photoreceptor degeneration. Invest. Ophthalmol. Vis. Sci., 42, 589–600.[Abstract/Free Full Text]

  5. Pacione, L.R., Szego, M.J., Ikeda, S., Nishina, P.M. and McInnes, R.R. (2003) Progress toward understanding the genetic and biochemical mechanisms of inherited photoreceptor degenerations. Annu. Rev. Neurosci., 26, 657–700.[CrossRef][ISI]

  6. Rattner, A., Sun, H. and Nathans, J. (1999) Molecular genetics of human retinal disease. Annu. Rev. Genet., 33, 89–131.[CrossRef][ISI][Medline]

  7. Swaroop, A., Xu, J.Z., Pawar, H., Jackson, A., Skolnick, C. and Agarwal, N. (1992) A conserved retina-specific gene encodes a basic motif/leucine zipper domain. Proc. Natl Acad. Sci. USA, 89, 266–270.[Abstract/Free Full Text]

  8. Swain, P.K., Hicks, D., Mears, A.J., Apel, I.J., Smith, J.E., John, S.K., Hendrickson, A., Milam, A.H. and Swaroop, A. (2001) Multiple phosphorylated isoforms of NRL are expressed in rod photoreceptors. J. Biol. Chem., 276, 36824–36830.[Abstract/Free Full Text]

  9. Mitton, K.P., Swain, P.K., Chen, S., Xu, S., Zack, D.J. and Swaroop, A. (2000) The leucine zipper of NRL interacts with the CRX homeodomain. A possible mechanism of transcriptional synergy in rhodopsin regulation. J. Biol. Chem., 275, 29794–29799.[Abstract/Free Full Text]

  10. Rehemtulla, A., Warwar, R., Kumar, R., Ji, X., Zack, D.J. and Swaroop, A. (1996) The basic motif-leucine zipper transcription factor Nrl can positively regulate rhodopsin gene expression. Proc. Natl Acad. Sci. USA, 93, 191–195.[Abstract/Free Full Text]

  11. Pittler, S.J., Zhang, Y., Chen, S., Mears, A.J., Zack, D.J., Ren, Z., Swain, P.K., Yao, S., Swaroop, A. and White, J.B. (2004) Functional analysis of the rod photoreceptor cGMP phosphodiesterase alpha subunit gene promoter: Nrl and Crx are required for full transcriptional activity. J. Biol. Chem., 279, 19800–19807.[Abstract/Free Full Text]

  12. Lerner, L.E., Gribanova, Y.E., Ji, M., Knox, B.E. and Farber, D.B. (2001) Nrl and Sp nuclear proteins mediate transcription of rod-specific cGMP-phosphodiesterase beta-subunit gene: involvement of multiple response elements. J. Biol. Chem., 276, 34999–35007.[Abstract/Free Full Text]

  13. Bessant, D.A., Payne, A.M., Mitton, K.P., Wang, Q.L., Swain, P.K., Plant, C., Bird, A.C., Zack, D.J., Swaroop, A. and Bhattacharya, S.S. (1999) A mutation in NRL is associated with autosomal dominant retinitis pigmentosa. Nat. Genet., 21, 355–356.[CrossRef][ISI][Medline]

  14. Bessant, D.A., Payne, A.M., Plant, C., Bird, A.C., Swaroop, A. and Bhattacharya, S.S. (2000) NRL S50T mutation and the importance of ‘founder effects’ in inherited retinal dystrophies. Eur. J. Hum. Genet., 8, 783–787.[CrossRef][ISI][Medline]

  15. DeAngelis, M.M., Grimsby, J.L., Sandberg, M.A., Berson, E.L. and Dryja, T.P. (2002) Novel mutations in the NRL gene and associated clinical findings in patients with dominant retinitis pigmentosa. Arch. Ophthalmol., 120, 369–375.[Abstract/Free Full Text]

  16. Martinez-Gimeno, M., Maseras, M., Baiget, M., Beneito, M., Antinolo, G., Ayuso, C. and Carballo, M. (2001) Mutations P51U and G122E in retinal transcription factor NRL associated with autosomal dominant and sporadic retinitis pigmentosa. Hum. Mutat., 17, 520.[Medline]

  17. Bessant, D.A., Holder, G.E., Fitzke, F.W., Payne, A.M., Bhattacharya, S.S. and Bird, A.C. (2003) Phenotype of retinitis pigmentosa associated with the Ser50Thr mutation in the NRL gene. Arch. Ophthalmol., 121, 793–802.[Abstract/Free Full Text]

  18. Mears, A.J., Kondo, M., Swain, P.K., Takada, Y., Bush, R.A., Saunders, T.L., Sieving, P.A. and Swaroop, A. (2001) Nrl is required for rod photoreceptor development. Nat. Genet., 29, 447–452.[CrossRef][ISI][Medline]

  19. Akhmedov, N.B., Piriev, N.I., Chang, B., Rapoport, A.L., Hawes, N.L., Nishina, P.M., Nusinowitz, S., Heckenlively, J.R., Roderick, T.H., Kozak, C.A. et al. (2000) A deletion in a photoreceptor-specific nuclear receptor mRNA causes retinal degeneration in the rd7 mouse. Proc. Natl Acad. Sci. USA, 97, 5551–5556.[Abstract/Free Full Text]

  20. Haider, N.B., Naggert, J.K. and Nishina, P.M. (2001) Excess cone cell proliferation due to lack of a functional NR2E3 causes retinal dysplasia and degeneration in rd7/rd7 mice. Hum. Mol. Genet., 10, 1619–1626.[Abstract/Free Full Text]

  21. Haider, N.B., Jacobson, S.G., Cideciyan, A.V., Swiderski, R., Streb, L.M., Searby, C., Beck, G., Hockey, R., Hanna, D.B., Gorman, S. et al. (2000) Mutation of a nuclear receptor gene, NR2E3, causes enhanced S cone syndrome, a disorder of retinal cell fate. Nat. Genet., 24, 127–131.[CrossRef][ISI][Medline]

  22. Livesey, F.J. and Cepko, C.L. (2001) Vertebrate neural cell-fate determination: lessons from the retina. Nat. Rev. Neurosci., 2, 109–118.[CrossRef][ISI][Medline]

  23. Nishida, A., Furukawa, A., Koike, C., Tano, Y., Aizawa, S., Matsuo, I. and Furukawa, T. (2003) Otx2 homeobox gene controls retinal photoreceptor cell fate and pineal gland development. Nat. Neurosci., 6, 1255–1263.[CrossRef][ISI][Medline]

  24. Ng, L., Hurley, J.B., Dierks, B., Srinivas, M., Salto, C., Vennstrom, B., Reh, T.A. and Forrest, D. (2001) A thyroid hormone receptor that is required for the development of green cone photoreceptors. Nat. Genet., 27, 94–98.[ISI][Medline]

  25. Zhang, J., Gray, J., Wu, L., Leone, G., Rowan, S., Cepko, C.L., Zhu, X., Craft, C.M. and Dyer, M.A. (2004) Rb regulates proliferation and rod photoreceptor development in the mouse retina. Nat. Genet., 36, 351–360.[CrossRef][ISI][Medline]

  26. DeRyckere, D. and DeGregori, J. (2002) Identification and characterization of transcription factor target genes using gene-targeted mice. Methods, 26, 57–75.[CrossRef][ISI][Medline]

  27. Livesey, F.J., Furukawa, T., Steffen, M.A., Church, G.M. and Cepko, C.L. (2000) Microarray analysis of the transcriptional network controlled by the photoreceptor homeobox gene Crx. Curr. Biol., 10, 301–310.[CrossRef][ISI][Medline]

  28. Mu, X., Zhao, S., Pershad, R., Hsieh, T.F., Scarpa, A., Wang, S.W., White, R.A., Beremand, P.D., Thomas, T.L., Gan, L. et al. (2001) Gene expression in the developing mouse retina by EST sequencing and microarray analysis. Nucl. Acids Res., 29, 4983–4993.[Abstract/Free Full Text]

  29. Irizarry, R.A., Bolstad, B.M., Collin, F., Cope, L.M., Hobbs, B. and Speed, T.P. (2003) Summaries of Affymetrix GeneChip probe level data. Nucl. Acids Res., 31, e15.[Abstract/Free Full Text]

  30. Young, R.W. (1985) Cell differentiation in the retina of the mouse. Anat. Rec., 212, 199–205.[CrossRef][Medline]

  31. Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D., Antonellis, K.J., Scherf, U. and Speed, T.P. (2003) Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics, 4, 249–264.[Abstract]

  32. Barash, Y., Dehan, E., Krupsky, M., Franklin, W., Geraci, M., Friedman, N. and Kaminski, N. (2004) Comparative analysis of algorithms for signal quantitation from oligonucleotide microarrays. Bioinformatics, 20, 839–846.[Abstract/Free Full Text]

  33. Xi, J., Farjo, R., Yoshida, S., Kern, T.S., Swaroop, A. and Andley, U.P. (2003) A comprehensive analysis of the expression of crystallins in mouse retina. Mol. Vis., 9, 410–419.[ISI][Medline]

  34. Kennan, A., Aherne, A., Palfi, A., Humphries, M., McKee, A., Stitt, A., Simpson, D.A., Demtroder, K., Orntoft, T., Ayuso, C. et al. (2002) Identification of an IMPDH1 mutation in autosomal dominant retinitis pigmentosa (RP10) revealed following comparative microarray analysis of transcripts derived from retinas of wild-type and Rho(–/–) mice. Hum. Mol. Genet., 11, 547–557.[Abstract/Free Full Text]

  35. Taylor, R.E., Shows, K.H., Zhao, Y. and Pittler, S.J. (2001) A PDE6A promoter fragment directs transcription predominantly in the photoreceptor. Biochem. Biophys. Res. Commun., 282, 543–547.[CrossRef][ISI][Medline]

  36. Payne, A.M., Downes, S.M., Bessant, D.A., Taylor, R., Holder, G.E., Warren, M.J., Bird, A.C. and Bhattacharya, S.S. (1998) A mutation in guanylate cyclase activator 1A (GUCA1A) in an autosomal dominant cone dystrophy pedigree mapping to a new locus on chromosome 6p21.1. Hum. Mol. Genet., 7, 273–277.[Abstract/Free Full Text]

  37. Downes, S.M., Holder, G.E., Fitzke, F.W., Payne, A.M., Warren, M.J., Bhattacharya, S.S. and Bird, A.C. (2001) Autosomal dominant cone and cone–rod dystrophy with mutations in the guanylate cyclase activator 1A gene-encoding guanylate cyclase activating protein-1. Arch. Ophthalmol., 119, 96–105.[Abstract/Free Full Text]

  38. Naya, F.S. and Olson, E. (1999) MEF2: a transcriptional target for signaling pathways controlling skeletal muscle growth and differentiation. Curr. Opin. Cell Biol., 11, 683–688.[CrossRef][ISI][Medline]

  39. Parker, M.H., Seale, P. and Rudnicki, M.A. (2003) Looking back to the embryo: defining transcriptional networks in adult myogenesis. Nat. Rev. Genet., 4, 497–507.[ISI][Medline]

  40. Mori, M., Ghyselinck, N.B., Chambon, P. and Mark, M. (2001) Systematic immunolocalization of retinoid receptors in developing and adult mouse eyes. Invest. Ophthalmol. Vis. Sci., 42, 1312–1318.[Abstract/Free Full Text]

  41. Li, A., Zhu, X., Brown, B. and Craft, C.M. (2003) Gene expression networks underlying retinoic acid-induced differentiation of human retinoblastoma cells. Invest. Ophthalmol. Vis. Sci., 44, 996–1007.[Abstract/Free Full Text]

  42. Mollereau, B., Dominguez, M., Webel, R., Colley, N.J., Keung, B., de Celis, J.F. and Desplan, C. (2001) Two-step process for photoreceptor formation in Drosophila. Nature, 412, 911–913.[CrossRef][Medline]

  43. Leong, M.L., Maiyar, A.C., Kim, B., O'Keeffe, B.A. and Firestone, G.L. (2003) Expression of the serum- and glucocorticoid-inducible protein kinase, Sgk, is a cell survival response to multiple types of environmental stress stimuli in mammary epithelial cells. J. Biol. Chem., 278, 5871–5882.[Abstract/Free Full Text]

  44. Heyninck, K., De Valck, D., Vanden Berghe, W., Van Criekinge, W., Contreras, R., Fiers, W., Haegeman, G. and Beyaert, R. (1999) The zinc finger protein A20 inhibits TNF-induced NF-kappaB-dependent gene expression by interfering with an RIP- or TRAF2-mediated transactivation signal and directly binds to a novel NF-kappaB-inhibiting protein ABIN. J. Cell Biol., 145, 1471–1482.[Abstract/Free Full Text]

  45. Wu, T., Chiang, S.K., Chau, F.Y. and Tso, M.O. (2003) Light-induced photoreceptor degeneration may involve the NF kappa B/caspase-1 pathway in vivo. Brain Res., 967, 19–26.[CrossRef][ISI][Medline]

  46. Jastrzebska, B., Filipek, A., Nowicka, D., Kaczmarek, L. and Kuznicki, J. (2000) Calcyclin (S100A6) binding protein (CacyBP) is highly expressed in brain neurons. J. Histochem. Cytochem., 48, 1195–1202.[Abstract/Free Full Text]

  47. Joo, J.H., Kim, J.W.,