JBC Biosymposia, Inc.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published In Press as doi:10.1074/jbc.M408223200 on July 28, 2004

J. Biol. Chem., Vol. 279, Issue 40, 42211-42220, October 1, 2004
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
279/40/42211    most recent
M408223200v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Yu, J.
Right arrow Articles by Swaroop, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Yu, J.
Right arrow Articles by Swaroop, A.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Altered Expression of Genes of the Bmp/Smad and Wnt/Calcium Signaling Pathways in the Cone-only Nrl-/- Mouse Retina, Revealed by Gene Profiling Using Custom cDNA Microarrays*

Jindan Yu,ab Shirley He,a James S. Friedman,ac Masayuki Akimoto,d Debashis Ghosh,e Alan J. Mears,af David Hicks,g and Anand Swaroop, Harold F. Falls Collegiate Professor and RPB Senior Scientific Investigatorahi

From the Departments of aOphthalmology and Visual Sciences, bBiomedical Engineering, eBiostatistics, and hHuman Genetics, University of Michigan, Ann Arbor, Michigan 48015, the dTranslational Research Center, Kyoto University Hospital, Sakyo-ku, Kyoto 606-8507, Japan, and the gLaboratoire de Physiopathologie Cellulaire et Moléculaire de la Rétine, INSERM U.592, 67091 Strasbourg Cedex, France

Received for publication, July 20, 2004


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
Many mammalian retinas are rod-dominant, and hence our knowledge of cone photoreceptor biology is relatively limited. To gain insights into the molecular differences between rods and cones, we compared the gene expression profile of the rod-dominated retina of wild type mouse with that of the cone-only retina of Nrl-/- (Neural retina leucine zipper knockout) mouse. Our analysis, using custom microarrays of eye-expressed genes, provided equivalent data using either direct or reference-based experimental designs, confirmed differential expression of rod- and cone-specific genes in the Nrl-/- retina and identified novel genes that could serve as candidates for retinopathies or for functional studies. In addition, we detected altered expression of several genes that encode cell signaling or structural proteins. Prompted by these findings, additional real-time PCR analysis revealed that genes belonging to the Bmp/Smad and Wnt/Ca2+ signaling pathways are expressed in the mature wild type retina and that their expression is significantly altered in the Nrl-/- retina. Chromatin immunoprecipitation analysis of adult retina identified Bmp4 and Smad4, which are down-regulated in the Nrl-/- retina, as possible direct transcriptional targets of Nrl. Consistent with these studies, Bmp4 and Smad4 are expressed in the mature rod photoreceptors of mouse retina. Modulation of Bmp4 and/or Smad4 by Nrl may provide a mechanism for integrating diverse cell signaling networks in rods. We hypothesize that Bmp/Smad and Wnt/Ca2+ pathways participate in cell-cell communication in the mature retina, and expression changes observed in the Nrl-/- retina reflect their biased utilization in rod versus cone homeostasis.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
In mammals, vision is initiated in the retina, which is a highly structured part of the brain consisting of over 50 types of neurons that are organized in three distinct layers (1, 2). The outer nuclear layer consists exclusively of two types of photoreceptors, rods and cones, responsible for detection and transduction of light energy. Rods function under low ambient light and form the major photoreceptor population of many mammals, including humans (95%) and mice (97%). Cones are responsible for phototransduction in bright light, providing high acuity and color vision (1). Cones are needed for maintaining central vision, and their survival is vital for preserving visual function in retinal and macular diseases (3, 4). Rods and cones possess distinct subsets of proteins involved in the phototransduction cascade (5, 6), but despite significant neuroanatomical and physiological advances (2, 7, 8), little progress has been made toward delineating the molecular mechanisms that underlie functional distinctions between the two photoreceptor types, their communication with other neurons, and their maintenance, survival, or remodeling in response to extrinsic or intrinsic insults.

One approach to systematically dissect the regulatory networks and molecules associated with rod or cone photoreceptor function is to take advantage of animal models that exhibit preferential utilization of one or the other photoreceptor sub-type. Unfortunately, many species with cone-rich retinas (e.g. ground squirrel and chick) present difficulties with respect to experimentation and are less amenable to genetic manipulations. Because of a large number of naturally occurring and experimentally generated mutants (available at jaxmice.jax.org/info/index.html), the mouse offers a unique opportunity to examine this complex question. The Nrl1-knockout (Nrl-/-) mouse, recently generated in our laboratory, exhibits a unique rod-less and cone-only retinal phenotype (9). Nrl was originally identified from a subtracted human retina library and shown to be expressed, by Northern analysis, specifically in the retina and retinoblastoma cell lines (10). In situ hybridization analysis identified Nrl transcripts in developing mouse brain and lens, although the expression became restricted to the retina after birth (11). Later studies, however, demonstrated that Nrl is specifically and highly expressed in the rod (and not cone) photoreceptors (9, 12) and pineal gland2 and that the transcripts in developing brain and lens probably represented cross-hybridization with p45, L-Maf, or another homologous sequence in the mouse genome (12, 13). Nrl is shown to interact with other transcription factors, such as the homeodomain protein Crx, zinc finger protein Fiz1, and orphan nuclear receptor Nr2e3, and regulate (either alone or synergistically) the expression of several rod-specific genes (14–20). Missense mutations in the human NRL gene are associated with autosomal dominant retinitis pigmentosa (21, 22). Consistent with these findings, the targeted deletion of Nrl (Nrl-/-) in mouse resulted in a retina with no rod photoreceptors; instead, a concomitant increase in functional S-opsin expressing cones was observed (9). This apparent functional switching of photoreceptor sub-types (from rods to cones) has been validated by histology, electrophysiology, and biochemical and molecular analysis (9, 23, 24).3

Microarray-based global profiling of gene expression, in combination with bioinformatic tools, can yield valuable insights into cell- or tissue-specific functions. Expression profiling of tissues from mice deficient in a transcription factor gene can point to downstream regulatory targets, provide candidates for functional studies, and facilitate positional cloning of human disease loci (25–28). Analysis of the retinal transcriptome during development and aging and in mouse models of retinal dysfunction has been the subject of intense investigation (28–31). We have used the Nrl-/- mouse model to identify molecular differences between rods and cones (24). However, commercially available microarrays do not have adequate representation of genes transcribed in developing and mature eye/retina; hence, several groups have produced custom slide microarrays of eye/retina-expressed genes (28, 30, 32). For gene profiling, we have isolated and sequenced cDNAs from mouse eye/retina libraries, annotated over 10,000 expressed sequence tags (ESTs), and produced cDNA microarrays (called I-gene microarrays) (33, 34).

Here, we report the expression profile of the mature retina from the rod-less (and cone-only) Nrl-/- mice using I-gene microarrays and compare it to the gene profile of the rod-dominated wild type mouse retina. We demonstrate differential expression of several genes, encoding phototransduction, structural, and signaling proteins, in the Nrl-/- retina. Our data reveal novel differentially expressed genes for future functional studies. Of particular importance are the findings that genes encoding components of Bmp/Smad and Wnt/Ca2+ signaling pathways are expressed in the mature retina and their expression is altered in the Nrl-/- retina. Expression analysis by real-time PCR and chromatin immunoprecipitation studies suggest that the activity of the Smad-mediated Bmp signaling pathway is modulated by Nrl in the mature retina. In support of this, we show that Bmp4 and Smad4 are expressed in the rod photoreceptors of the mature mouse retina. We also propose that rods and cones exhibit selective bias in the utilization of different signaling pathways for cell-cell communication and controlling intracellular functions.


   EXPERIMENTAL PROCEDURES
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
RESULTS
 DISCUSSION
 REFERENCES
 
Tissue Preparation, RNA Isolation, and Northern Analysis—All procedures involving mice were approved by the University Committee on Use and Care of Animals of the University of Michigan. Retinas were dissected from the wild type and Nrl-/- mice at postnatal day (P) 21 and snap-frozen on dry ice. Total RNA was isolated using TRIzol reagent (Invitrogen) and purified by using an RNeasy kit (Qiagen). RNA integrity was verified by denaturing formaldehyde-agarose gels. Total RNA samples with a 260- to 280-nm absorbance ratio of greater than 1.9 were used for studies. Northern analysis was performed as described previously (10).

Reference RNA for Retinal Gene Profiling—To generate the reference RNA for microarray hybridizations, total RNA was pooled from the following tissues and cell lines: mouse eye or retina at embryonic day (E) 14–16, P2–3, P10–12, and adult (5.5 mg); mouse adult brain (1.5 mg); P19 embryonic carcinoma cells (35) (3 mg); retinoic acid-induced P19 cells that differentiated into neuronal and glial-like cells (3 mg); and neuroblastoma cell line N1E-115 (36) (4 mg). The pooled RNA was divided into aliquots (at a concentration of 1.28 µg/µl) and stored at -80 °C until use.

Target Labeling and Microarray Hybridization—Mouse I-gene cDNA microarray slides (34) contained PCR-amplified products from over 6,500 cDNAs, printed in duplicate; cDNAs were isolated from libraries constructed from E15.5 eyes, P2 eyes, and adult retinas, sequenced, and annotated (33). Target RNA (10 µg of total RNA) was labeled using a 3DNA Submicro Expression Array Detection kit (Genisphere, Hatfield, PA) and hybridized to microarray slides, as described (37). The slides were scanned using an Affymetrix 428 scanner (Affymetrix, Santa Clara, CA) to obtain the highest intensity of signal, without reaching saturation for a maximum of 10 out of the 13,440 spots.

Image Processing and Data Analysis—Scanned array images without major defects (such as scratches or blobs) were analyzed using AnalyzerDG (MolecularWare Inc., Cambridge, MA) in a batch mode. A "contour shape" was utilized to detect spots for intensity calculations, whereas a "cell method" was set to calculate background on an individual basis in a local square region centered on the spot. A data file containing spots' intensities and annotations was exported for each array in the tab-delimited text format and then imported into the statistical package, R (available at www.r-project.org/). Intensities of Cy3 and Cy5 channels for each spot were calculated separately by subtracting corresponding median background from the mean signal intensity. Genes with negative background-corrected intensities in either channel were filtered out.

To reduce systematic variation caused by experimental procedures (such as dye effects), a data-driven normalization was applied to individual datasets using a cluster of least-altered genes on the array identified by a rank-based algorithm. Briefly, let (Rj, Gj) denote the measurements for the jth gene in the red (Cy5) and green (Cy3) channels, respectively, j = 1,..., m. We calculate the ranks in each channel separately, take the difference in ranks between the two channels, and fit a three-component normal mixture model to the difference in ranks for the m genes. There will be three classes of genes to consider: those for which the rank in the red channel is significantly higher than that in the green channel, those for which the rank in the green channel is significantly higher than that in the red channel, and those for which the ranks do not substantially change between the two channels.

The genes whose ranks did not change between the two channels were used to perform a slide-dependent normalization based on a locally weighted linear squares procedure (38). This allowed us to normalize genes and redefine a new horizontal zero axis that was used to compute log ratios of intensities in two channels. For indirect comparisons, log ratios of the wild type slide were subtracted from the Nrl-/- slide to obtain reference-corrected values for each gene. An Empirical Bayes method was then applied to the replicated arrays to obtain a B statistic for each gene (39); the B statistic estimates the posterior log odds of differential expression.

Quantitative Reverse Transcription-PCR—To validate gene expression changes, qRT-PCR analysis was performed as described (33). Briefly, total RNA (2.5 µg) treated with RQ1 RNase-free DNase (Promega, Madison, WI) was subjected to reverse transcription using oligod(T) primers and with (+RT sample) or without (-RT) SuperScript II (Invitrogen). For each gene, real-time PCR reactions from wild type and Nrl-/- samples were performed in triplicate with SybrGreen I (Molecular Probes, Eugene, OR) and analyzed with the iCycler IQ real-time PCR detection system (Bio-Rad). The average threshold cycle (Ct) differences between the two samples were normalized against Hprt (control) in the corresponding cDNA preparation.

Immunohistochemistry—Anti-Gnb3 polyclonal antibodies were generated in rabbit against the peptide (ADITLAELVSGLEVV) and affinity-purified (Invitrogen). Anti-Smad4 rabbit polyclonal antibody (H-552) was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Due to the low number of cones in wild type mouse retina (3% of photoreceptors), we also performed immunohistochemical analyses on frozen sections of adult pig retina, which contains ~15–20% cones (40). Cryostat sections of mouse and pig retina were permeabilized with 0.1% Triton X-100 (5 min), and then preincubated in blocking buffer (PBS supplemented with 0.2% bovine serum albumin, 0.1% Tween 20, 5% normal rabbit serum, and 0.1% NaN3) for 30 min. The sections were then incubated overnight at 4 °C in anti-Gnb3 and anti-Smad4 polyclonal antibodies (diluted 1:200 in blocking buffer), combined either with rho-4D2 anti-rod opsin monoclonal antibody (1 µg/ml) (41) or with biotinylated peanut agglutinin (PNA, Vector Laboratories, 10 µg/ml). After extensive washing in PBS, sections were incubated in a mixture of goat anti-rabbit IgG-Alexa488 or -Alexa594 (Molecular Probes; 2 µg/ml), and either rabbit anti-mouse IgG- or streptavidin-Alexa594 or -Alexa488 (each 2 µg/ml) in blocking buffer for 2 h. Slides were washed extensively, mounted, and viewed by fluorescence microscopy (Nikon Optiphot 2) or by laser scanning confocal microscopy (Zeiss LSM 510 version 2.5 scanning device with Zeiss Axiovert 100 inverted microscope). Control experiments were performed by omitting the primary antibody.

Chromatin Immunoprecipitation—A commercially available assay kit (Upstate Biotechnologies, Charlottesville, VA) was used for ChIP studies. Briefly, four snap-frozen retinas from wild type mice were cross-linked for 15 min at 37 °C with 1% formaldehyde in PBS containing proteinase inhibitors. 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 eight times using 20-s pulses. The remaining steps were essentially performed as described by the manufacturer, using anti-NRL polyclonal antibody (12).

Putative promoter regions (5' upstream of the transcription start site) for Rho, Bmp4, Smad4, and Bmpr1a were determined in silico (www.ncbi.nlm.nih.gov/mapview). Each DNA sequence was analyzed using MatInspector (www.genomatix.de/index.html). PCR primers were designed to flank the putative Nrl binding sites (Nrl response element, NRE) (18) either predicted manually or by MatInspector (Table I). The sequence closest to the transcription start site was chosen. ChIP DNAs from two independent experiments were used for PCR using equal amounts for input, with antibody, and no antibody reactions.


View this table:
[in this window]
[in a new window]
  		TABLE I
Putative NRE and the PCR primers used for ChIP enrichment assays

 
In Situ Hybridization—The 35S-labeled antisense and sense cRNA transcripts were synthesized from the mouse Bmp4 and Smad4 cDNAs cloned in pSPORT1 vector, according to the manufacturer's instructions (Ambion). The labeled transcripts were subjected to alkali hydrolysis (to give an average size of 70 nucleotides) and then hybridized overnight to 5- to 7-µm serial sections of adult mouse retina using standard protocols (Phylogeny Inc., Columbus, OH). After stringent washing, the slides were dipped in Kodak NTB-2 nuclear track emulsion and exposed for 1–2 weeks. Slides were developed in Kodak D-19, counterstained with toluidine blue, and analyzed by light and dark-field optics.

Expression Analysis Using Purified Rod Photoreceptors—To obtain a highly enriched population of rod photoreceptors, we used Nrl-GFP transgenic mice that express enhanced green fluorescent protein (EGFP) under the control of Nrl promoter specifically in rods.4 Retinas of Nrl-GFP mice at P28 were dissected and dissociated using the Papain Dissociation System (Worthington). Cells that were positive and negative for EGFP were flow-sorted by FACSAria (BD Biosciences). Total RNA was used for RT-PCR using gene-specific primers, as follows. Bmpr1a: forward, AAGGAATGGGTGGGATTAGC; reverse, TGGCGATTGCCAACTAGATA; Grm6: forward, CAAGTAGCAAGGTTGAGTGT; reverse, GGAAGAATGCTGGAAGCAAG; Hprt: forward, CAAACTTTGCTTTCCCTGGT; reverse, CAAGGGCATATCCAACAACA; Nrl: forward, GCTGCATTTTCACCGAATCT; reverse, GGTGGTTTGGGTTGTGGTAG; Rho: forward, CTTCCTGATCTGCTGGCTTC; reverse, ACAGTCTCTGGCCAGGCTTA; Smad4: forward, ACCCGCGTATGCCGCCCCATCC; reverse, ACAGCGTCGCCAGGTGCTCGGC; Thy1: forward, AACTCTTGGCACCATGAACC; reverse, AGGCTGAACTCATGCTGGAT.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
DISCUSSION
 REFERENCES
 
We chose to generate gene expression profiles of P21 retinas from rod-dominated wild type and rodless (cone-only) Nrl-/- mice. At this stage, the differentiation and laminar organization of retinal neurons are complete, and phototransduction pathways in the retina (from photoreceptors to ganglion cells) are fully functional.

High Concordance between Direct and Indirect Microarray Comparisons—We examined two different experimental designs, direct and indirect comparison (42), for their ability to identify differentially expressed genes. Two direct comparison experiments (two slides), in which Cy3-labeled wild type and Cy5-labeled Nrl-/- retinal RNA targets were hybridized simultaneously to the same slide, were performed. Five indirect comparisons were carried out with the reference RNA labeled by Cy3 and hybridized in conjunction with Cy5-labeled either wild type or Nrl-/- retinal RNA (total of 10 slides). Of the 13,440 spots on the I-gene microarray slides, 97.4% (13,092 spots) showed higher spot intensity than the background (i.e. were considered detected) in both direct comparison slides, whereas 91.6% (12,307 spots) spots were detected in all 10 slides with indirect comparisons. Scatter-plot analysis of signal intensities of the wild type and Nrl-/- retinas revealed high similarity between direct and indirect comparisons (Fig. 1). A majority of genes have log2 ratios centered at 0, indicating no change between the two tested samples. Only a few spots displayed over 4-fold change in both methods, although direct comparison showed a tighter scatter within the 4-fold lines. Both methods were able to successfully identify the duplicate spots of S-opsin and rhodopsin as the most up- or down-regulated genes in the Nrl-/- retina, respectively.



View larger version (22K):
[in this window]
[in a new window]
  		FIG. 1.
Representative scatter-plots showing differential expression of two copies each of S-opsin (circled spots above 0 axis) and rhodopsin (circled spots below 0 axis). A, direct comparisons. MA-plot of wild type (G) and Nrl-/- (R) mouse retinal gene expression in one direct comparison experiment, where M = log2(R/G) and A = [log2(RxG)]/2. B, indirect comparisons. Samples from the wild type (R1) retina and the common reference (G1) were hybridized to one slide, whereas samples from the Nrl-/- retina (R2) and the reference (G2) were hybridized to another slide. Expression levels of the wild type and Nrl-/- retinas were compared by removing the signal from common reference after data normalization. X-axis, A = [log2(R1 x G1) + log2(R2 x G2)]/2. Y-axis, M = log2(R1/G1) - log2(R2/G2).

 
To compare the power of direct versus indirect methods in identifying differential gene expression, similar statistical analysis protocols were applied to both datasets to rank the genes based on their odds of being altered. The top 40 genes identified by direct comparison included 31 genes that were also ranked in top 40 by indirect comparison. Furthermore, 37 of the top 40 genes revealed by indirect experiments were present in the top 75 genes with direct comparison. This indicates a possible false-positive rate of 3/40 in the indirect comparison, assuming that only genes ranked in the top 75 by both methods are truly positive.

Differential Expression of Genes in the Nrl-/- Retina—Based on the greatest odds of difference identified by both direct and indirect comparisons, a non-redundant set of 74 highly ranked genes was selected; of these, 50 genes are known (Table II), whereas 24 cDNAs only show homology to sequences in the EST or genomic databases (Table III). Of the 50 known differentially expressed genes, only 8 demonstrate higher expression in the Nrl-/- retina as compared with the wild type. The majority of these genes can be divided into the following functional groups: phototransduction, transcriptional regulation, signaling pathways, and structural or membrane-associated proteins. One-third of the altered genes (24/74) represent proteins of unknown function (Table III); such genes can be used as candidates for mutation screening of patients with rod and cone dystrophies or for biological studies.


View this table:
[in this window]
[in a new window]
  		TABLE II
The top 50 differentially expressed genes Microarray and qRT-PCR columns show -fold change, where (-) and (+) indicate lower and higher expression in the Nrl-/- retina, respectively.

 


View this table:
[in this window]
[in a new window]
  		TABLE III
The top 24 differentially expressed unknown genes/ESTs. Microarray and qRT-PCR columns show fold-change, where (-) and (+) indicate lower and higher expression in the Nrl-/- retina, respectively

 
We first wanted to determine whether our microarray profiling could validate changes in the expression of 16 phototransduction genes that had been analyzed previously by Northern analysis of P10 Nrl-/- retina (see Fig. 5A of Ref. 9). Our microarray data corroborated the expression changes of all genes (data not shown), except cone arrestin 3 (Arr3) and Pde6b, which did not exhibit any alteration. To further establish the fidelity of the microarray data, we examined the overlaid microarray images of spots corresponding to three genes: Gnb3 (cone transducin {beta} subunit) and its counterpart in rods, Gnb1, and one random EST (AC021049 [GenBank] ) (corresponding to the clone MRA-1648, with partial sequence homology to cold-shock domain protein A). The overlaid images of these three genes were in agreement with their microarray data (Fig. 2A). Northern blot and qRT-PCR analysis further confirmed the expression differences detected by microarray experiments (Fig. 2, B and C). Consistent with these observations, Gnb3 immunoreactivity was specific to cones: intense within outer segments, and moderate in cone cell bodies and synaptic pedicles (Fig. 2D, panels a and b). The staining within the outer segments was distinct from that observed for rod opsin (Fig. 2D, panel c), but co-localized precisely with PNA (Fig. 2D, panels d and e), a commonly used marker of cone outer matrix sheaths, demonstrating the cone-specific distribution of Gnb3 immunolabeling.



View larger version (73K):
[in this window]
[in a new window]
  		FIG. 2.
Differential expression of three genes: Gnb3,Gnb1[b], and an EST (AC021049 [GenBank] , corresponding to the clone MRA-1648). A, the overlaid images of red (Nrl-/-) and green (wild type) channels showing microarray spots corresponding to these three genes (white arrows). The spots appear as green if expression is higher in wild type, red if higher in knockout, and yellow if equivalent expression is observed. B, Northern analysis comparing total RNA levels in wild type (+/+) and Nrl-/- (-/-) retinas. Gapd was utilized to normalize the amount of loaded samples. C, qRT-PCR analysis of transcript levels in wild type (wt) and Nrl-/- (ko) retinas. D, immunohistochemical localization of Gnb3 in adult porcine retina: confocal images of retinal sections viewed by normal illumination to show outer layers (a), and with anti-Gnb3 alone (b), Gnb3 merged with rho-4D2 rod opsin staining (c), PNA alone (d), or anti-Gnb3 merged with PNA staining (e). Intense Gnb3 immunoreactivity was observed in cone outer segments (arrow, a–c), as well as in the cytoplasm of cone cell bodies (*, b) and within cone synaptic pedicles (#, b). Labeling with rod opsin antibody (red) and Gnb3 antibody (green) shows the separation of the two patterns (c). Cone outer segments were outlined by specific staining with PNA (d), and merging of signals for PNA (red) and Gnb3 (green) shows overlap of the two stains (wide arrows in both d and e). CB, cone cell bodies; IS, photoreceptor inner segments; ONL, outer nuclear layer; OPL, outer plexiform layer; OS, photoreceptor outer segments. Scale bar in panel e = 20 µm.

 
QRT-PCR Validation of Differentially Expressed Genes—Due to the variability inherent in microarray technology, qRT-PCR analysis was applied to validate expression data (Tables II and III). For 33 out of 35 genes examined (95%), qRT-PCR data were in good agreement with the microarray results with respect to the direction of observed changes (i.e. qualitative). For 21 genes, microarray data underestimated the expression change. Two genes, Rxrg and Mtap6 (also called Stop), showed a discrepancy between the two methods. Microarray data indicated slight reduction, whereas qRT-PCR demonstrated clear augmentation of their expression in the mutant retina. This discrepancy may be the result of differential expression of various isoforms (generated by alternative splicing) encoded by the two genes. Although microarray method is hybridization-based and may reflect cumulative effects, qRT-PCR using gene-specific primers generates expression data for specific isoform(s).

Repression of Bmp/Smad Signaling Pathway in the Nrl-/- Retina—Microarray analysis showed that the expression of four genes belonging to the Bmp/Smad pathway (Bmp2, Bmp4, Bmpr1a, and Tob1) was reduced in the Nrl-/- retina. Because the expression and role of Bmp/Smad proteins have not been documented in mature retina, we decided to explore this further by qRT-PCR analysis of other pathway components that were not represented on I-gene microarray. All genes of the Smad-mediated Bmp signaling pathway exhibited lower expression in the cone-only Nrl-/- retina (Fig. 3A). The expression levels of Bmp4, Bmp2, and Bmpr1a were reduced by 12-, 4-, and 32-fold, respectively, in the Nrl-/- retina, whereas R-Smads (Smad 1 and Smad 5) and co-Smad (Smad 4) transcripts were marginally decreased. The expression of anti-Smads (Smad 6 and Smad 7) and Tob1 was also reduced in the Nrl-/- retina.



View larger version (17K):
[in this window]
[in a new window]
  		FIG. 3.
Expression analysis by qRT-PCR. The number associated with each bar indicates -fold change in the Nrl-/- retina relative to the wild type. A, reduced expression of genes associated with the Bmp/Smad pathway in the Nrl-/- retina. B, altered expression of genes associated with Wnt/Ca2+ signaling and Ca2+ homeostasis in the Nrl-/- retina. Expression changes in genes encoding Wnt5a ligand, Fzd2 receptor, and ten Ca2+ regulated proteins are shown.

 
Altered Expression of the Components of Wnt/Ca2+ Signaling Pathway in the Nrl-/- Retina—Altered expression of several Ca2+/calmodulin-regulated genes, including Myo5a, Myo7a, and Mtap6, prompted us to examine the transcript levels of calmodulin (Calm2) and Camk2b by qRT-PCR. Both genes demonstrated altered expression in the Nrl-/- retinal RNA as compared with the wild type (Fig. 3B). We also detected higher expression of the Wnt5a ligand by qRT-PCR, indicating the overall increased activity and utilization of this pathway in the Nrl-/- retina.

ChIP Analysis—To investigate if the reduced expression of Bmp/Smad genes in the Nrl-/- retina is a direct result of the lack of Nrl, we performed ChIP assays that permit in vivo analysis of the binding of a transcription factor to its cognate cis-regulatory sequence elements (43, 44). DNA fragments bound to Nrl were isolated from mouse retina using an anti-Nrl antibody (12). The ChIP DNA was examined for the enrichment of putative promoter regions of the three Bmp/Smad pathway genes (Bmp4, Smad4, and Bmpr1a) (Fig. 4). PCR amplification of the NRE region in the rhodopsin promoter (14, 18) was used as positive control for ChIP assay. We were able to amplify the genomic DNA fragments, including the putative Nrl binding sites upstream of transcription start sites of Bmp4 and Smad4; however, no enrichment of the promoter region of Bmpr1a was observed.



View larger version (40K):
[in this window]
[in a new window]
  		FIG. 4.
Chromatin immunoprecipitation. A, a flowchart of the ChIP protocol. B, an example of the sonicated input DNA used for ChIP PCR. C, enrichment of Bmp4 and Smad4, but not Bmpr1a, promoter region sequences in ChIP DNA obtained with anti-Nrl antibody. Amplification of rhodopsin promoter sequence in Nrl-ChIP DNA establishes the validity of the procedure. Sonicated input DNA and H2O lanes serve as positive and negative controls, respectively.

 
Expression of Bmp4 and Smad4 in Mature Rod Photoreceptors—To further validate the possible role of the Bmp/Smad pathway and its regulation by Nrl, we decided to determine the expression of Bmp4 and Smad4 in mature rod photoreceptors. In situ hybridization to RNA in adult mouse retina sections revealed a positive signal for Bmp4 and Smad4 in all three nuclear layers, including the outer nuclear layer (ONL) of photoreceptors that primarily contains rods (Fig. 5A). RT-PCR analysis of the FACS-purified rod photoreceptors (GFP-positive and rhodopsin-positive) from the Nrl-GFP retina also demonstrated high enrichment of Smad4 transcripts in GFP(+) versus GFP(-) cells (Fig. 5B). GFP-positive cells also expressed Bmpr1a; however, the data for Bmp4 was inconsistent with the primers used for RT-PCR.



View larger version (101K):
[in this window]
[in a new window]
  		FIG. 5.
Expression of Bmp4 and Smad4 in rod photoreceptors. A, in situ hybridization of 35S-labeled Bmp4 and Smad4 cRNAs to adult mouse retina sections. Phase contrast image is shown to identify the retinal layers. Sense transcripts served as control and showed background hybridization. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer (photoreceptor nuclei); OS, outer segments of photoreceptors. Magnification, x10. B, expression analysis using purified rod photoreceptors. a, almost 95% of FACS-purified cells of dissociated retina from adult Nrl-GFP mice express GFP, and all GFP-expressing cells are immunolabeled with rhodopsin RET-P1 antibody. b, RT-PCR analysis using RNA from FACS-purified GFP(+) and GFP(-) cells. Smad4 and Bmpr1a transcripts are detected primarily in GFP(+) rod photoreceptors. Nrl and rhodopsin expression are highly enriched in GFP(+) cells, whereas Thy1 (a marker of retinal ganglion cells) and Grm6 (a bipolar cell marker) are expressed in GFP(-) cells. Hprt expression serves as control for RNA. C, immunohistochemistry of Smad4 within adult mouse retina. Frozen sections of PN20 wild type mouse retinas were incubated with either anti-Smad4 polyclonal antibody (b and e) and, respectively, either rhodopsin (rho4D2 monoclonal: c) or PNA (f). Cellular layers were revealed by 4',6-diamidino-2-phenylindole staining (a and d). Smad4 immunoreaction product was seen in the inner segments (IS) and as discrete strands running down through the ONL till the level of the OPL (arrow, b and e). Scale bar in f = 30 µm for a–c and 50 µm for d–f.

 
We then examined whether Bmp4 or Smad4 protein is detectable in mature photoreceptors. Various commercially available Bmp4 antibodies gave either high background or did not work well on retinal sections; hence, we were unable to observe specific Bmp4 immunostaining. Nevertheless, one Smad4 polyclonal antibody displayed a distinct pattern in accordance with its diminution in the Nrl-/- mouse retina (Fig. 5C). Within the adult wild type mouse retina, Smad4 immunostaining was observed throughout the inner segments (IS) and reaction product percolated down through the ONL as discrete fiber-like profiles (Fig. 5C, panels b and e). Immunostaining was more intense toward the scleral surface of the ONL but was far more widespread than the few cone cell bodies located in this position. Smad4 immunolabeling partially overlapped with that of rhodopsin (panel c) and did not correspond to PNA staining (panel f). Taken together, these results suggest Smad4 is within a subset of rod cells.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
REFERENCES
 
Our knowledge of signaling molecules and cell-cell communication pathways that specify physiological divergence between rods and cones and their homeostasis is currently limited. The availability of the Nrl-/- mouse, which exhibits a complete lack of rods and significantly enhanced cone function, permits dissection of molecular differences between the two types of photoreceptors and associated pathways. We took advantage of the advances in microarray technology and utilized custom I-gene microarrays to compare the gene expression profiles of mature P21 retinas of the rod-dominant wild type mouse with that of the Nrl-/- mouse. In addition to a large number of differentially expressed genes encoding known proteins, we identified 24 novel ESTs that might serve as candidate genes for human retinal diseases but would require additional molecular and functional analyses. The major findings from expression profiling presented in this report are that: (i) components of Bmp/Smad and Wnt/Ca2+ signaling pathways are expressed in the mature retina, suggesting their physiological relevance in regulating cell-cell communication and/or homeostasis; (ii) Bmp/Smad and Wnt/Ca2+ pathways are altered in the cone-only Nrl-/- retina, indicating their differential utilization in rod- and cone-associated inter- and intra-cellular signal transduction; (iii) at least two Bmp/Smad pathway genes (Bmp4 and Smad4) may be direct targets of regulation by Nrl; and (iv) Bmp4 and Smad4 are expressed in the rod photoreceptors of mature retina.

Global gene expression profiling using microarrays is a powerful methodology to generate molecular signatures of specific cell types, yielding valuable data for investigating complex biological processes. However, a careful experimental design is critical; for cDNA microarrays, issues related to direct versus indirect comparisons have been repeatedly emphasized (42, 45). Indirect comparison is more advantageous, because it allows researchers to cross-compare studies performed under different conditions. Recent advances have greatly minimized systematic sources of variations by improved technique (37), appropriate normalization (46–48), and sufficient replication (49, 50). In this study, we used I-gene microarrays and evaluated the performance of indirect method using a reference RNA target relative to the direct comparison of wild type and Nrl-/- samples. Using the reference, we were able to obtain signals for >97% of spots on a single slide and about 91% of spots on all 12 slides studied. This high representation of retinal genes in the reference sample provides a basis for future indirect comparisons using I-gene microarrays and meta-analysis of the data. The percentage of positive spots on I-gene microarrays is substantially higher than that obtained with commercial microarray slides (52%) (31) or Affymetrix GeneChips (~60%) using the retinal tissue (24). Although the precise -fold change was not identical, similar patterns of gene expression were revealed by direct and indirect comparisons. As one would expect, increasing the number of independent replicates decreased the number of false positives. Despite systematic variation introduced by the use of a reference sample, we found that the indirect method provided comparable results to direct comparison.

Gene profiling of whole retina from wild type and Nrl-/- mice would yield expression differences that reflect the morphological and physiological changes associated with the loss of Nrl function. As predicted, the lack of rods and their replacement by cones resulted in higher expression of cone phototransduction genes with a concomitant reduction in rod gene transcripts. A majority of the differentially expressed genes showed lower expression in the Nrl-/- retina, reflecting the bias toward rod genes in the I-gene microarrays (containing cDNAs derived from rod-dominant wild type retinal tissue). Gene profiling suggested a remodeling of the Nrl-/- retina, as evident by altered expression of many genes encoding structural proteins (e.g. Prph2, Myo5a, Myo7a, Krt1–18, Cct4, and Mtap6) (see Table II). This is consistent with the collapse of the sub-retinal space, shortening of photoreceptor outer segments, and rewiring of neurons in the outer plexiform layer (9, 51). It should be noted that many of these expression changes probably reflect secondary rather than primary effects of the loss of Nrl.

Proteins of Wnt (wingless) and transforming growth factor-{beta} families initiate a diverse array of signal transduction pathways that control cell-cell communication and play prominent roles during development (52–56). Our studies demonstrated altered expression of components of Wnt/Ca2+ and Bmp-Smad pathways in the mature Nrl-/- retina (Fig. 6). In the Wnt/Ca2+ pathway, Wnt5a binds to its receptor Frizzled 2 (Fzd2), stimulating intracellular Ca2+ release and activating Ca2+/calmodulin-dependent kinase II beta (Camk2b) in a G-protein-dependent manner (52). The gene profiling data suggested alterations in Ca2+ homeostasis, as indicated by increased expression of three (Gcap1, Gc1, and Rcvrn) and reduced expression of two (Gcap2 and Gc2) genes encoding Ca2+-regulated proteins. Furthermore, we observed altered expression of several Ca2+-dependent genes, such as Calm2, Camk2b, Mtap6, Myo5a, and Myo7a. Because these genes are expressed in both rods and cones (57, 58), we hypothesize that the expression changes may reflect differences in Ca2+ homeostasis in rods versus cones (59, 60). We suggest that the Wnt/Ca2+ pathway may modulate intracellular Ca2+ concentration and may have more direct control on cone function or homeostasis. A regulatory role of the Wnt/Ca2+ pathway in photoreceptor activity has been indicated in previous studies. Wnt5a and Fzd2 are known to signal through G{alpha}t, which is highly and specifically expressed in photoreceptors, and are altered in mouse retinal mutants (61, 62). In addition, two Fzd2 homologous genes, sFRP2 and Mfrp, known to inhibit the Wnt/Ca2+ signaling pathway, were shown to have altered transcript levels in retinitis pigmentosa patients and in the rd6 mouse (63–65). Regulated expression of synaptic proteins, such as Myo5a, Myo7a, and Hrg4 (Unc119h), is in good agreement with the known roles of the Wnt/Ca2+ signaling in the synthesis and release of neurotransmitters (66).



View larger version (44K):
[in this window]
[in a new window]
  		FIG. 6.
A schematic representation of signaling molecules and corresponding pathways altered in the Nrl-/- mouse retina. Expression changes were identified in several genes that could be classified as components of the Bmp/Smad pathway (54) and/or the Wnt/Ca2+ signaling pathway (52). Proposed cross-talk (blue arrows) between these two signaling pathways and their regulation of cytoskeletal (light-blue rectangle box) or Ca2+-dependent proteins (purple ellipse) are indicated. Genes, upregulated in the Nrl-/- retina, are represented in red, whereas down-regulated genes are shown in green. The color bar indicates the -fold change in expression.

 
Bone morphogenetic proteins (Bmps) are members of transforming growth factor-{beta} superfamily of secreted proteins that primarily transduce signals through Smad proteins (53–56). Bmp signaling is essential for lens development and is involved in retinal differentiation (67–70). Although a role for Bmps has been suggested in adult retinal pigment epithelium (71), little is known about Bmp and Smad pathways in the mature retinal photoreceptors. We postulate that reduced expression of genes of the Bmp/Smad pathway has relevance to the lack of rod function in the Nrl-/- retina. ChIP and expression data strongly suggest that Bmp4 and Smad4 are expressed in rod photoreceptors of adult mouse retina and that their expression is directly regulated by Nrl (see Figs. 4 and 5). The absence of enrichment of Bmpr1a (although it is expressed in rods) by ChIP suggests that Nrl may not directly act on its promoter (at least in the region tested). Modulation of Bmp/Smad signaling by Nrl argues in favor of its role in rod-mediated cell-cell communication. Activated Smad complexes can positively or negatively regulate transcription of numerous target genes and modify cellular response. We hypothesize that in addition to directly controlling the expression of specific phototransduction genes in mature rods (15–18), Nrl influences rod function and homeostasis by integrating discrete signaling events.


   FOOTNOTES
 
* This research was supported in part by National Institutes of Health Grants EY11115 (including administrative supplements), EY07003, and GM72007, by The Foundation Fighting Blindness (Owings Mills, MD), by Research to Prevent Blindness (RPB, New York, NY), by the Macula Vision Research Foundation (West Conshohocken, PA), and by the British Retinitis Pigmentosa Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Back

c A recipient of Foundation Fighting Blindness-Canada Post-Doctoral Fellowship. Back

f Current address: University of Ottawa Eye Institute, Ottawa Health Research Institute, Ottawa, Ontario K1H 8L6, Canada. Back

i To whom correspondence should be addressed: W. K. Kellogg Eye Center, University of Michigan, 1000 Wall St., Ann Arbor, MI 48105. Tel.: 734-615-2246; Fax: 734-647-0228; E-mail: swaroop{at}umich.edu.

1 The abbreviations used are: Nrl, neural retina leucine zipper; ChIP, chromatin immunoprecipitation; NRE, Nrl response element; Bmp, bone morphogenetic protein; EST, expressed sequence tag; P, postnatal day; E, embryonic day; qRT-PCR, quantitative reverse transcription-PCR; Hprt, hypoxanthine guanine phosphoribosyl transferase; PBS, phosphate-buffered saline; PNA, peanut agglutinin; GFP, green fluorescence protein; EGFP, enhanced GFP; ONL, outer nuclear layer; FACS, fluorescence-activated cell sorting; IS, inner segment. Back

2 A. J. Mears and A. Swaroop, unpublished data. Back

3 S. S. Nikonov, L. Daniele, C. Lillo, A. J. Mears, A. Swaroop, D. Williams, and E. N. Pugh, Jr., submitted for publication. Back

4 M. Akimoto and A. Swaroop, unpublished data. Back


   ACKNOWLEDGMENTS
 
We thank R. Farjo, M. I. Othman, and S. P. MacNee for generation of microarrays, Alfred Hero for discussions on data analysis, P. Hitchcock, T. Glaser, and A. J. Hunt for constructive suggestions, Jérôme Mutterer for confocal microscopy, Jennifer Masterson for technical, and Sharyn Ferrara for administrative assistance.



   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Dowling, J. E. (1987) The Retina: an Approachable Part of the Brain, Harvard University Press, Cambridge, MA
  2. Masland, R. H. (2001) Curr. Opin. Neurobiol. 11, 431-436[CrossRef][Medline] [Order article via Infotrieve]
  3. Hicks, D., and Sahel, J. (1999) Invest. Ophthalmol. Vis. Sci. 40, 3071-3074[Free Full Text]
  4. Curcio, C. A., Owsley, C., and Jackson, G. R. (2000) Invest. Ophthalmol. Vis. Sci. 41, 2015-2018[Free Full Text]
  5. Nathans, J. (1999) Neuron 24, 299-312[CrossRef][Medline] [Order article via Infotrieve]
  6. Molday, R. S. (1998) Invest. Ophthalmol. Vis. Sci. 39, 2491-2513[Medline] [Order article via Infotrieve]
  7. Leskov, I. B., Klenchin, V. A., Handy, J. W., Whitlock, G. G., Govardovskii, V. I., Bownds, M. D., Lamb, T. D., Pugh, E. N., Jr., and Arshavsky, V. Y. (2000) Neuron 27, 525-537[CrossRef][Medline] [Order article via Infotrieve]
  8. Marc, R. E., and Jones, B. W. (2003) Mol. Neurobiol. 28, 139-147[CrossRef][Medline] [Order article via Infotrieve]
  9. Mears, A. J., Kondo, M., Swain, P. K., Takada, Y., Bush, R. A., Saunders, T. L., Sieving, P. A., and Swaroop, A. (2001) Nat. Genet. 29, 447-452[CrossRef][Medline] [Order article via Infotrieve]
  10. Swaroop, A., Xu, J. Z., Pawar, H., Jackson, A., Skolnick, C., and Agarwal, N. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 266-270[Abstract/Free Full Text]
  11. Liu, Q., Ji, X., Breitman, M. L., Hitchcock, P. F., and Swaroop, A. (1996)Oncogene 12, 207-211[Medline] [Order article via Infotrieve]
  12. 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) J. Biol. Chem. 276, 36824-36830[Abstract/Free Full Text]
  13. Ogino, H., and Yasuda, K. (1998) Science 280, 115-118[Abstract/Free Full Text]
  14. Chen, S., Wang, Q. L., Nie, Z., Sun, H., Lennon, G., Copeland, N. G., Gilbert, D. J., Jenkins, N. A., and Zack, D. J. (1997) Neuron 19, 1017-1030[CrossRef][Medline] [Order article via Infotrieve]
  15. Lerner, L. E., Gribanova, Y. E., Ji, M., Knox, B. E., and Farber, D. B. (2001) J. Biol. Chem. 276, 34999-35007[Abstract/Free Full Text]
  16. Mitton, K. P., Swain, P. K., Chen, S., Xu, S., Zack, D. J., and Swaroop, A. (2000) J. Biol. Chem. 275, 29794-29799[Abstract/Free Full Text]
  17. Mitton, K. P., Swain, P. K., Khanna, H., Dowd, M., Apel, I. J., and Swaroop, A. (2003) Hum. Mol. Genet. 12, 365-373[Abstract/Free Full Text]
  18. Rehemtulla, A., Warwar, R., Kumar, R., Ji, X., Zack, D. J., and Swaroop, A. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 191-195[Abstract/Free Full Text]
  19. 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) J. Biol. Chem. 279, 19800-19807[Abstract/Free Full Text]
  20. Cheng, H., Khanna, H., Oh, E. C., Hicks, D., Mitton, K. P., and Swaroop, A. (2004) Hum. Mol. Genet. 13, 1563-1575[Abstract/Free Full Text]
  21. 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) Nat. Genet. 21, 355-356[CrossRef][Medline] [Order article via Infotrieve]
  22. DeAngelis, M. M., Grimsby, J. L., Sandberg, M. A., Berson, E. L., and Dryja, T. P. (2002) Arch. Ophthalmol. 120, 369-375[Abstract/Free Full Text]
  23. Zhu, X., Brown, B., Li, A., Mears, A. J., Swaroop, A., and Craft, C. M. (2003) J. Neurosci. 23, 6152-6160[Abstract/Free Full Text]
  24. Yoshida, S., Mears, A. J., Friedman, J. S., Carter, T., He, S., Oh, E., Jing, Y., Farjo, R., Fleury, G., Barlow, C., Hero, A. O., and Swaroop, A. (2004) Hum. Mol. Genet. 13, 1487-1503[Abstract/Free Full Text]
  25. Livesey, F. J., Furukawa, T., Steffen, M. A., Church, G. M., and Cepko, C. L. (2000) Curr. Biol. 10, 301-310[CrossRef][Medline] [Order article via Infotrieve]
  26. DeRyckere, D., and DeGregori, J. (2002) Methods 26, 57-75[CrossRef][Medline] [Order article via Infotrieve]
  27. Horton, J. D., Shah, N. A., Warrington, J. A., Anderson, N. N., Park, S. W., Brown, M. S., and Goldstein, J. L. (2003) Proc. Natl. Acad. Sci. U. S. A. 100, 12027-12032[Abstract/Free Full Text]
  28. Mu, X., Beremand, P. D., Zhao, S., Pershad, R., Sun, H., Scarpa, A., Liang, S., Thomas, T. L., and Klein, W. H. (2004) Development 131, 1197-1210[Abstract/Free Full Text]
  29. Blackshaw, S., Fraioli, R. E., Furukawa, T., and Cepko, C. L. (2001) Cell 107, 579-589[CrossRef][Medline] [Order article via Infotrieve]
  30. Swaroop, A., and Zack, D. J. (2002) Genome Biol. http://genomebiology.com/2002/3/8/reviews/1022
  31. Yoshida, S., Yashar, B. M., Hiriyanna, S., and Swaroop, A. (2002) Invest. Ophthalmol. Vis. Sci. 43, 2554-2560[Abstract/Free Full Text]
  32. Chowers, I., Gunatilaka, T. L., Farkas, R. H., Qian, J., Hackam, A. S., Duh, E., Kageyama, M., Wang, C., Vora, A., Campochiaro, P. A., and Zack, D. J. (2003) Invest. Ophthalmol. Vis. Sci. 44, 3732-3741[Abstract/Free Full Text]
  33. Yu, J., Farjo, R., MacNee, S. P., Baehr, W., Stambolian, D. E., and Swaroop, A. (2003) Genome Biol. http://genomebiology.com/2003/4/10/R65
  34. Farjo, R., Yu, J., Othman, M. I., Yoshida, S., Sheth, S., Glaser, T., Baehr, W., and Swaroop, A. (2002) Vision Res. 42, 463-470[CrossRef][Medline] [Order article via Infotrieve]
  35. McBurney, M. W., Jones-Villeneuve, E. M., Edwards, M. K., and Anderson, P. J. (1982) Nature 299, 165-167[CrossRef][Medline] [Order article via Infotrieve]
  36. Amano, T., Richelson, E., and Nirenberg, M. (1972) Proc. Natl. Acad. Sci. U. S. A. 69, 258-263[Abstract/Free Full Text]
  37. Yu, J., Othman, M. I., Farjo, R., Zareparsi, S., MacNee, S. P., Yoshida, S., and Swaroop, A. (2002) Mol. Vis. 8, 130-137[Medline] [Order article via Infotrieve]
  38. Cleveland, W. S. (1979) J. Am. Stat. Assoc. 74, 829-836[CrossRef]
  39. Lonnestedt, I., and Speed, T. (2002) Statistica Sinica 12, 31-46
  40. Hendrickson, A., and Hicks, D. (2002) Exp. Eye Res. 74, 435-444[CrossRef][Medline] [Order article via Infotrieve]
  41. Hicks, D., and Molday, R. S. (1986) Exp. Eye Res. 42, 55-71[CrossRef][Medline] [Order article via Infotrieve]
  42. Yang, Y. H., and Speed, T. (2002) Nat. Rev. Genet. 3, 579-588[CrossRef][Medline] [Order article via Infotrieve]
  43. Boyd, K. E., and Farnham, P. J. (1999) Mol. Cell. Biol. 19, 8393-8399[Abstract/Free Full Text]
  44. Wells, J., and Farnham, P. J. (2002) Methods 26, 48-56[CrossRef][Medline] [Order article via Infotrieve]
  45. Churchill, G. A. (2002) Nat. Genet. 32, (suppl.) 490-495
  46. Quackenbush, J. (2002) Nat. Genet. 32, (suppl.) 496-501
  47. Tseng, G. C., Oh, M. K., Rohlin, L., Liao, J. C., and Wong, W. H. (2001) Nucleic Acids Res. 29, 2549-2557[Abstract/Free Full Text]
  48. Yang, Y. H., Dudoit, S., Luu, P., Lin, D. M., Peng, V., Ngai, J., and Speed, T. P. (2002) Nucleic Acids Res. 30, e15[Abstract/Free Full Text]
  49. Lee, M. L., Kuo, F. C., Whitmore, G. A., and Sklar, J. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 9834-9839[Abstract/Free Full Text]
  50. Pan, W., Lin, J., and Le, C. T. (2002) Genome Biol. http://genomebiology.com/2003/5/research/0022
  51. Strettoi, E., Mears, A. J., and Swaroop, A. (2004) J. Neurosci., 24, 7576-7582[Abstract/Free Full Text]
  52. Kuhl, M., Sheldahl, L. C., Park, M., Miller, J. R., and Moon, R. T. (2000) Trends Genet. 16, 279-283[CrossRef][Medline] [Order article via Infotrieve]
  53. Derynck, R., and Zhang, Y. E. (2003) Nature 425, 577-584[CrossRef][Medline] [Order article via Infotrieve]
  54. Miyazawa, K., Shinozaki, M., Hara, T., Furuya, T., and Miyazono, K. (2002) Genes Cells 7, 1191-1204[Abstract]
  55. Wrana, J. L. (2000) Cell 100, 189-192[CrossRef][Medline] [Order article via Infotrieve]
  56. Shi, Y., and Massague, J. (2003) Cell 113, 685-700[CrossRef][Medline] [Order article via Infotrieve]
  57. Yang, R. B., and Garbers, D. L. (1997) J. Biol. Chem. 272, 13738-13742[Abstract/Free Full Text]
  58. Ebrey, T., and Koutalos, Y. (2001) Prog. Retin. Eye Res. 20, 49-94[CrossRef][Medline] [Order article via Infotrieve]
  59. Rebrik, T. I., and Korenbrot, J. I. (1998) J. Gen. Physiol. 112, 537-548[Abstract/Free Full Text]
  60. Krizaj, D., and Copenhagen, D. R. (2002) Front. Biosci. 7, d2023-d2044[Medline] [Order article via Infotrieve]
  61. Yang-Snyder, J., Miller, J. R., Brown, J. D., Lai, C. J., and Moon, R. T. (1996) Curr. Biol. 6, 1302-1306[CrossRef][Medline] [Order article via Infotrieve]
  62. Liu, X., Liu, T., Slusarski, D. C., Yang-Snyder, J., Malbon, C. C., Moon, R. T., and Wang, H. (1999) Proc. Natl. Acad. Sci. U. S. A. 96, 14383-14388[Abstract/Free Full Text]
  63. Jones, S. E., Jomary, C., Grist, J., Stewart, H. J., and Neal, M. J. (2000) Neuroreport 11, 3963-3967[Medline] [Order article via Infotrieve]
  64. Jones, S. E., Jomary, C., Gris