Global Gene Expression Analysis in a Mouse Model for Norrie Disease: Late Involvement of Photoreceptor Cells

Abstract

purpose. Mutations in the NDP gene give rise to a variety of eye diseases, including classic Norrie disease (ND), X-linked exudative vitreoretinopathy (EVRX), retinal telangiectasis (Coats disease), and advanced retinopathy of prematurity (ROP). The gene product is a cystine-knot–containing extracellular signaling molecule of unknown function. In the current study, gene expression was determined in a mouse model of ND, to unravel disease-associated mechanisms at the molecular level.

methods. Gene transcription in the eyes of 2-year-old Ndp knockout mice was compared with that in the eyes of age-matched wild-type control animals, by means of cDNA subtraction and microarrays. Clones (n = 3072) from the cDNA subtraction libraries were spotted onto glass slides and hybridized with fluorescently labeled RNA-derived targets. More than 230 differentially expressed clones were sequenced, and their expression patterns were verified by virtual Northern blot analysis.

results. Numerous gene transcripts that are absent or downregulated in the eye of Ndp knockout mice are photoreceptor cell specific. In younger Ndp knockout mice (up to 1 year old), however, all these transcripts were found to be expressed at normal levels.

conclusions. The identification of numerous photoreceptor cell–specific transcripts with a reduced expression in 2-year-old, but not in young, Ndp knockout mice indicates that normal gene expression in these light-sensitive cells of mutant mice is established and maintained over a long period and that rods and cones are affected relatively late in the mouse model of ND. Obviously, the absence of the Ndp gene product is not compatible with long-term survival of photoreceptor cells in the mouse.

Ocular symptoms of Norrie disease, a rare X-linked recessive form of congenital blindness, include bilateral leukokoria soon after birth, vascularized retrolental membranes, retinal detachment, and progressive shrinking of the eye. At least one third of affected males have mental retardation and experience progressive hearing loss in later life. ^{1 2} The NDP gene was isolated by positional cloning. Mutation analyses demonstrated that mental handicaps and hearing deficiencies are pleiotropic effects of a single gene defect. ^{3 4 5 6} Mutations in the NDP gene also account for a variety of other familial and sporadic eye diseases, including exudative vitreoretinopathy, retinopathy of prematurity, and Coats disease. ^{7 8 9} Disease symptoms of these traits are confined to the inner eye or retina; neither mental retardation nor deafness is observed. The predicted protein consists of 133 amino acid residues and contains an N-terminal signal sequence and a cystine-knot motif at its carboxyl-terminal end. Three-dimensional computer modeling of the amino acid sequence reveals striking similarity with transforming growth factor-β (TGFβ), ¹⁰ suggesting that this protein is involved in developmental and differentiation processes. However, the precise function of the gene has been hitherto unknown. 

Prior to the current study, we generated a mouse model for ND by homologous recombination in embryonic stem cells. ¹¹ In the mouse, the Ndp gene is expressed in the inner layers of the retina, but also in brain and ear, as shown by RNA in situ hybridization. ^{11 12} Thus, the expression pattern matches the tissues affected in human disease. Morphologic characteristics in the eyes of mutant mice include retrolental structures, reduced number of retinal ganglion cells, regional disorganization of the inner nuclear and photoreceptor cell layers, and malformation of the retinal vasculature. ^{11 13} Electrophysiologic recordings reveal a negatively shaped b-wave of the electroretinogram, suggesting a predominant loss of inner retinal components. ¹⁴ Hearing deficits in mutant mice have also been reported. ¹²  

To shed more light on the cellular and molecular processes underlying ND in the eye, we compared gene expression patterns in mutant and wild-type mice by using cDNA subtraction and microarrays. These techniques enabled us to characterize hundreds of transcripts in parallel. For cDNA subtraction, we used whole-eye RNA from 2-year-old mice. Several thousand clones from the subtractions were printed on glass slides and used as probes to target cDNA from wild-type and Ndp knockout mice for differential expression. Approximately 1600 differentially expressed clones were identified and characterized in more detail. 

Methods

Animals

Ndp knockout mice were obtained by gene-targeting, as described previously. ¹¹ Animal studies were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the regulation for animal experiments of the federal government of Germany. 

Light Microscopy

The enucleated eyes were immersion fixed for 5 hours in 2.5% glutaraldehyde in 0.05 M sodium cacodylate buffer. The cornea and lens were then removed. After postfixation in 2.5% glutaraldehyde overnight, the tissue was fixed in cacodylate-buffered osmium tetroxide (1%) for 1 hour and embedded in Spurr resin. For light microscopy analysis, 0.5-μm-thick sections were cut on a microtome (Ultracut E; Reichert-Jung, Arnsburg, Germany). Sections were stained with toluidine blue. 

RNA Extraction

Total RNA was isolated from mouse eyes with an extraction kit (RNeasy; Qiagen, Hilden, Germany). Briefly, tissues were disrupted and simultaneously homogenized (Polytron PT 3100; VWR International, Darmstadt, Germany) in lysis buffer containing guanidinium isothiocyanate. The sample was then applied to an RNA extraction column, in which the total RNA bound to the silica-gel matrix, and contaminants were efficiently washed away. High-quality RNA (longer than 200 bases) was eluted in water. 

Suppression Subtractive Hybridization

cDNAs were synthesized from 0.75 μg total RNA with a kit (Smart PCR cDNA synthesis kit; BD Biosciences-Clontech, Heidelberg, Germany). Single-stranded cDNA was amplified by long-distance PCR (Advantage cDNA PCR Kit; BD Biosciences-Clontech). Before subtraction, cDNA was size fractionated with spin columns (Chroma Spin-1000; BD Biosciences-Clontech), digested by RsaI, and purified (Nucleo Spin Extract; Macherey and Nagel, Düren, Germany). 

Subtraction was performed (PCR-Select cDNA Subtraction Kit; BD Biosciences-Clontech), according to the method described by Diatchenko et al. ¹⁵ and the manufacturer’s instructions. Two separate subtractive hybridization experiments were performed for the isolation of genes suppressed (wild-type cDNA was the tester for forward subtraction) or overexpressed (knockout cDNA was the tester for reverse subtraction) in the knockout mice, respectively. We applied 27 cycles of primary PCR and 12 cycles of secondary PCR with a cDNA polymerase mix (Advantage cDNA Polymerase Mix; BD Biosciences-Clontech). 

Subtraction Efficiency Test

A comparison of the abundance of known cDNAs before and after subtraction by hybridization analysis can be used to estimate the efficiency of subtraction. Secondary PCR products (10 μL from a 25-μL reaction) of forward- and reverse-subtracted and unsubtracted samples were separated on 1.6% agarose gels and transferred onto membranes (GeneScreenPlus; NEN Life Science Products, Köln, Germany) by alkaline transfer. Hybridizations were performed overnight with ³²P-labeled cDNA probes at 65°C in 0.5 M phosphate buffer, 7% SDS, and 1 mM EDTA. 

Generation of Complete cDNAs

Complete cDNAs of five selected genes were amplified by RT-PCR using whole-eye RNA and sequence-specific primers (Pde6b: 5′-CAGTGAGGAACAGGTACGCA and 5′-AGGCAGAGTCCGTATGCAGT, 2623 bp; Bcp: 5′-GAATATCTCTTCGGTGGGGC and 5′-AGTGAGGGCCAACTTTGCTA, 1007 bp; Sag: 5′-CTGATAGGATTGCACCAGGTC and 5′-ATTTCTGGGAGGAATGCTCA, 1482 bp; Rpr1: 5′-TCAGTGGACAGCGATTCTCA and 5′-GCCCAAATAAAAATGCCAAG, 1158 bp; Xlrs1: 5′-GACCAAGGACAAGGAGAAAATGC, and 5′- TGATCCAAAGGTAGCCAGAAT, 5565 bp). 

Cloning of the Subtraction Libraries

Subtracted secondary PCR products were ligated into a plasmid vector from a T/A cloning kit (pGEM-T Easy plasmid vector; Promega, Mannheim, Germany). The resultant constructs were transformed by electroporation into XL1 Blue–competent cells (Stratagene, Heidelberg, Germany) and selected by blue–white screening. Colonies were chosen randomly, placed in 96-well plates, and grown for 16 hours in 100 μL LB medium containing ampicillin (50 μg/mL). For long-term storage (−80°C), an equal volume of glycerol was added. 

Microarray Construction

Clone inserts were PCR amplified with vector-specific standard M13 primers (M13for-TG: 5′-GTAAAACGACGGCCAGTG and M13rev-C: 5′- caggaaacagctatgacc). Subsequently, a nested PCR-reaction with 5′ amino-modified, adaptor-specific primers NPp1 (5′-amino- TCGAGCGGCCGCCCGGGCAGGT) and NPp2R (5′-amino- AGCGTGGTCGCGGCCGAGGT) was performed. Control genes (62 retina-specific genes [201 expressed sequence tags (ESTs)], 31 housekeeping genes [111 ESTs, used for normalization of fluorescence intensities], and 17 different plant cDNAs [used as the negative control and for background estimation]) were amplified with 5′ amino-modified M13 primers (M13for-TG and M13rev-C). By agarose gel electrophoresis, evaluated PCR products were ethanol precipitated, washed in 70% ethanol, air dried, and resuspended in 8 μL 100 mM sodium carbonate (pH 9). Microarrays were printed on pretreated amino-silane–coated slides (PE-Applied Biosystems, Weiterstadt, Germany) using a robotic spotting device (Beecher Instruments, Silver Spring, MD). Spotting volume was approximately 5 nL per spot, resulting in spots approximately 200 μm in diameter. 

Labeling of Targets and Microarray Hybridization

In general, targets (complete cDNAs, pools of PCR-amplified inserts, and subtraction products) were labeled by random priming in 60 μL labeling buffer containing 6.3 μg decanucleotides; 15 μM each of dGTP, dCTP, and dATP (Decalabel; MBI Fermentas, Vilnius, Lithuania); and 15 μM Cy3- or Cy5-dUTP (Amersham, Freiburg, Germany). Cy3- and Cy5-labeled targets were cohybridized on a single microarray. Each hybridization was replicated at least once, with a color swap. 

Total RNA (25 μg) was reverse transcribed by direct incorporation of fluorescent dyes in a reaction containing 1 μg oligo(dT), 100 μM Cy3- or Cy5-dUTP (Amersham), 200 μM dTTP, and 500 μM each of dGTP, dCTP, and dATP (Roche Diagnostics, Mannheim, Germany). 

Labeled products were purified with a kit (Qiaquick PCR Purification; Qiagen) and combined, and 30 μg mouse DNA enriched for repetitive sequences (mouse-COT-1 DNA; Gibco), 25 μg yeast t-RNA (Gibco), and 20 μg poly(dA-dT) were added. To block unspecific signals caused by cross-hybridization of flanking primer sequences of the subtraction products, an oligonucleotide mix containing the nested primers NPp1 and NPp2R and their complementary sequences was added. Products were dried in a speedvac, dissolved in a formamide-based hybridization buffer (50% formamide, 6× SSC, 0.5% SDS, and 5× Denhardt solution), and hybridized under a coverslip in a humid chamber for 16 hours at 42°C. Unbound targets were washed away by a 5-minute incubation in 0.01% SDS and 0.2× SSC, followed by two 5-minute incubations in 0.2× SSC at room temperature. 

Image Acquisition and Data Analysis

Hybridization signals were detected with a confocal laser scanner (model 418; Affymetrix, Santa Clara, CA). Sixteen-bit TIF-tagged images, with intensities ranging from 1 to 65,536 arbitrary units, were obtained for each dye and virtually merged by using an extension of image analysis software (IPLab Spectrum; Scanalytics, Billerica, MA). Background-corrected mean spot intensities were obtained from each processed picture using the algorithms of Chen et al. ¹⁶ Data sets were transferred to a spreadsheet computer program (Excel; Microsoft, Redmond, WA). Plant genes, clones without an insert, and empty spots were used to estimate nonspecific hybridization signals and background. Signals from clones containing mitochondrial sequences were used to normalize arrays with regard to labeling bias, differences in quantum yield, and photobleaching of the dyes. 

In the comparative analysis, transcript levels were considered significantly different when the ratio of the experimental to the control average fluorescence intensities was more than threefold and the average fluorescence intensity was greater than 1000 arbitrary units. The latter criteria minimized selection of genes that showed multifold changes over a control involving weak absolute responses (low average fluorescence intensities). 

DNA Sequencing

Putative differentially expressed clones were chosen from glycerol stocks. Plasmid DNA was prepared using a robot and a kit (model 9600 robot and Turbo kit; Qiagen). Inserts of recombinant clones were sequenced (Big Dye terminator chemistry; PE-Applied Biosystems). The sequencing gels were run (model ABI377 sequencer; PE-Applied Biosystems), and data were assembled and edited using the GAP4 Contig Editor (provided in the public domain by the Laboratory of Molecular Biology, Medical Research Council, Cambridge, UK, and available at http://www.mrc-lmb.cam.ac.uk/pubseq/manual/gap4_unix_1.html). ¹⁷ Sequences were compared with entries in the GenBank and EMBL databases by using the BLAST homology search program. ¹⁸ (GenBank is provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD, and is available at http://www.ncbi.nlm.nih.gov/Genbank/; EMBL is provided in the public domain by the European Molecular Biology Laboratory, Heidelberg, Germany, and is available at http://www.embl-heidelberg.de/.) 

Virtual Northern Blot Analysis

Plasmids of individual clones were digested with RsaI (New England Biolabs, Frankfurt am Main, Germany), and inserts were recovered for hybridization experiments. PCR-amplified cDNAs (15 μL from a 100-μL reaction; Smart PCR cDNA synthesis kit; BD Biosciences-Clontech) from wild-type and Ndp knockout mice were electrophoresed, blotted, and hybridized, as described for the subtraction efficiency test. 

Results

Suppression Subtractive Hybridization, Characterization of Subtraction Libraries, and Construction of a cDNA Microarray

To identify changes in gene expression in the eyes of Ndp knockout mice, we used suppression subtractive hybridization (SSH), a subtractive hybridization technique that enriches the amount of cDNA fragments from transcripts present in one population of mRNAs (tester) but absent from another (driver). The technique simultaneously reduces fragments of equally expressed genes. Subtraction was performed in both orientations, forward (tester: wild-type whole-eye RNA) and reverse (tester: knockout whole-eye RNA). The forward subtraction was performed to identify genes downregulated or not expressed in 2-year-old knockout mice, and the reverse library was generated to recognize upregulated genes. 

From the forward-subtraction library, 96 randomly chosen clones were analyzed in an initial differential screening test by hybridizing the inserts to blots of wild-type and knockout cDNAs (virtual Northern blot analysis). A large percentage of clones (49/96; 51%) detected normal signal intensities in wild-type but no or weak signals in knockout cDNA, indicating lower levels or complete loss of expression in the eyes of knockout mice. Hybridization of 38 randomly selected clones from the reverse library revealed no expression differences between wild-type and mutant mice. Because of this result, we determined the efficiency of subtractive suppression. In addition to the subtractive enrichment, SSH excludes sequences that the tester and driver have in common (subtractive suppression). The hybridization of probes from two housekeeping genes (Gapdh and mt-Nd1) to wild-type and knockout cDNAs before and after SSH showed their reduction in both subtracted samples (Fig. 1) . To determine the enrichment rate of tester-specific fragments, we hybridized two clones from the forward-subtraction library (clones 90 and 975) to blotted cDNAs and subtraction products. The abundance of both fragments was significantly increased in the forward subtraction (Fig. 1) . In addition, we performed a PCR-based subtraction efficiency test (data not shown) and obtained consistent results. Altogether, these data demonstrate a high fidelity of cDNA subtraction in both orientations (forward and reverse subtraction). 

Because 49 of 96 clones from the forward-subtraction library were identified as differentially expressed genes (i.e., RNA transcripts present in wild-type but absent or reduced in mutant mice), we analyzed this library in more detail. The inserts of further 3072 clones were PCR-amplified and spotted together with controls onto glass slides. Subsequently, the cDNA microarray was hybridized with fluorescently labeled subtracted cDNA targets (e.g., Cy3-labeled forward-subtraction product and Cy5-labeled reverse-subtraction product, Fig. 2A ). Based on the 96 clones assessed in the previous differential screening test, criteria were defined to identify additional differentially expressed cDNA fragments (see the Methods section). On the microarray, 1565 of 3072 clones (i.e., 51%) fulfilled these criteria. Before individual inserts were tested for differential expression, redundant clones were eliminated. The full-length cDNAs of five genes (Pde6b, Opn1sw, Sag, Rpr1, and Rsh1) representing 13 of the precharacterized set of 96 clones were hybridized to the microarray. In this way, 339 of 1565 clones were excluded from further analysis. Of the remaining 1226 clones, 295 were then hybridized on virtual Northern blots. Differential expression was confirmed for 187 (63.4%) of them. Some of the corresponding transcripts were completely absent in knockout mice, whereas others showed a decreased expression level in the affected tissue (Fig. 2B) . Consistent results were obtained for more than 100 of these clones on a second pair of (wild-type and knockout) animals. 

Identity of Differentially Expressed Clones and Calculation of Number of Genes in the Microarray

All 236 confirmed differentially expressed clones (49/96 and 187/3072) were sequenced. Of those, 156 represented 83 known genes (mouse genes or human orthologues). Further, 67 clones were part of 51 different ESTs, and 13 clones did not show homology to any known sequences (sequence information provided on a Web site hosted by the Max Planck Institute for Molecular Genetics, Berlin, Germany, at http://www.molgen.mpg.de/∼norrie/iovs/welcome.html). Among known genes, structural proteins of photoreceptor cells (peripherin and rod outer membrane protein) and numerous components of the phototransduction cascade (rhodopsin, transducin, two phosphodiesterase subunits, arrestin, guanylate cyclase–activating protein, and cyclic guanosine monophosphate [cGMP] gated cation channel) were identified. The human orthologues of 13 of these transcripts have been shown to be mutated in familial forms of photoreceptor degeneration. We also found genes involved in calcium binding, signal transduction, and exocytosis, RNA-binding, apoptosis, and cytoskeleton organization. The results of sequence-based gene identification are summarized in Table 1 1 1 . 

To estimate the number of individual transcripts in the subtraction library we ascertained the frequency of identical fragments by screening the microarray with nine pools each containing 6 to 14 individual clone inserts. On average, each fragment occurred with a frequency of two to three. Subsequently, we calculated the number of amplifiable RsaI fragments per transcript. Sequence data for 12 complete cDNAs were extracted from the database or generated by sequencing of full-length RT-PCR products. On average, each transcript contained five amplifiable RsaI fragments. Correspondingly, clones from the forward-subtraction library (n = 3072) should represent approximately 250 different transcripts. On virtual Northern blot analysis, 49 (51%) of 96 randomly selected clone inserts from the forward-subtraction library detected differentially expressed transcripts. Hence, this library contains at least 126 transcripts that are underrepresented in the eyes of knockout mice. 

Differential Gene Expression during Disease Progression

Initially, the temporal transcription pattern of the Ndp gene itself was analyzed to confirm whether the gene is active throughout life. Indeed, the transcript was detected on virtual Northern blot analysis from postnatal day 14 up to 2 years of age in the eyes of wild-type mice (Fig. 3) . As expected, no transcript was found in all Ndp knockout mice. ¹¹ To monitor differential gene expression early in disease, the cDNA microarray was cohybridized with Cy5- and Cy3-labeled PCR-amplified cDNA targets (Smart PCR cDNA synthesis kit; BD Biosciences-Clontech) from 4-week-old Ndp knockout mice and wild-type littermates, respectively (Fig. 2C) . None of the clones on the microarray produced differential signals according to the defined criteria, whereas marked signal differences were found with PCR-amplified cDNA from 2-year-old animals (Fig. 2D) . 

In addition to PCR-amplified cDNAs, total RNAs from eyes of wild-type and Ndp knockout mice at 4 weeks and 2 years of age, respectively, were used as targets in hybridization experiments (data not shown). To obtain reliable sets of expression data, total RNAs from three independent pairs of animals were hybridized to separate microarrays. Again, the expression of photoreceptor-specific transcripts was drastically reduced only in 2-year-old but not in 4-week-old knockout mice. 

Subsequently, selected differentially expressed clones were hybridized to virtual Northern blots containing total eye cDNA from wild-type and mutant mice at 0.5, 1, 3, 9, 12, and 24 months of age (Fig. 3) . These genes showed normal expression levels up to 12 months. Significant reduction of transcript levels was observed only after 2 years. 

To correlate gene expression with histologic data, we examined retinal morphology in Ndp knockout mice at 24 and 29 months of age by light microscopy (Fig. 4) . All retinal layers were significantly reduced; however, remnants of photoreceptor cells were still present. 

Discussion

To search for alterations in the gene expression pattern in the eyes of Ndp knockout mice, we used RNA from 2-year-old wild-type and mutant animals for SSH, an approach that enriches the amount of differentially expressed genes. The total number of different transcripts in the forward-subtraction library and the quantity of confirmed differentially expressed genes was calculated to be approximately 250 and 126, respectively. Most likely, the absolute numbers are much higher, because numerous clones occur only once in the subtraction library. Although SSH has been reported to detect expression differences efficiently, ¹⁵ the forward-subtraction library may contain only a portion of all differentially expressed genes. Because subtractive enrichment relies on RsaI fragments of cDNAs, it apparently fails to identify transcripts that do not contain amplifiable RsaI fragments. 

Because of the advanced age of the mice (2 years), the subtraction products reflect primarily consequences of disease progression. In the forward subtraction, numerous differentially expressed genes were identified that were present in wild-type, but not in mutant, mice. Absence of specific mRNAs in the eyes of mutant mice is mainly due to disease-associated loss of particular cell types, but is also due to altered gene expression. Histologic examination of the eyes of 2-year-old Ndp knockout mice revealed that all retinal cell layers were drastically reduced but still present, including photoreceptors. Possibly, the almost complete absence of several photoreceptor-specific transcripts does not result from complete loss of photoreceptors but rather from their failure to express the characteristic set of genes. 

Although we used RNA from whole eyes for cDNA subtraction, many of the differentially expressed genes are known to be retina specific, consistent with the fact that disease processes in the Ndp knockout mouse are most conspicuous in retinal cells ¹¹ and that the Ndp gene is highly expressed in mouse retina. Macroscopically, the eyes of Ndp knockout mice are normal, but light and electron microscopy have revealed characteristic morphologic changes in the retina and vitreous. ^{11 13}  

Studies of the electroretinogram in young mutant mice have pointed to a defect of inner retinal components, whereas the pathologic features of photoreceptors are less pronounced. ¹⁴ By analyzing the gene expression of several photoreceptor-specific transcripts over time we found alterations only late in the disease. 

In addition to photoreceptor-specific transcripts, differentially expressed genes encode proteins involved in Fas-mediated apoptotic cell death. Bcl-2 binding protein (Bis) and Fas apoptosis inhibitory molecule (Faim) are expressed in wild-type mice at 2 years, but not in age-matched Ndp knockout mice. Both genes code for negative regulators of apoptotic pathways. Bis itself has only weak antiapoptotic activity, but enhances Bcl-2-mediated prevention of apoptosis. ¹⁹ Ectopic expression of Bcl-2 in photoreceptor cells of mice with a mutation in the peripherin (retinal degeneration slow, rds) gene protects against photoreceptor cell death. ²⁰ It also has been shown that overexpression of Bcl-2 protects photoreceptor cells in vitro against apoptosis induced by exposure to visible light. ²¹ Faim was originally isolated by analyzing Fas-resistant B lymphocytes but was also shown to be expressed in other tissues and cell types. It functions as inducible mediator of Fas resistance, thereby affecting apoptotic processes. ²² Loss of the antiapoptotic functions of Bis and Faim in Ndp knockout mice may contribute to the degeneration of photoreceptor cells during disease progression. 

In summary, we identified numerous alterations in gene expression that reflect pathologic changes at the molecular level in an advanced disease stage of Ndp knockout mice. Because the Ndp gene is predominantly expressed in the inner cell layers of the adult mouse retina, and photoreceptor degeneration is seen relatively late in Ndp knockout mice, the effect of the NDP protein on rods and cones may be indirect. However, our results demonstrate that proximal retinal cell layers and continuous expression of the Ndp gene have an important function for maintenance and integrity of all retinal cell layers, in particular photoreceptor cells. 

Differentially expressed genes are involved in phototransduction, protein trafficking, RNA processing, cytoskeleton and extracellular matrix turnover, apoptosis, and exo- and endocytosis. Also, numerous novel genes were shown to be differentially expressed in Ndp knockout mice, many of which may be candidate genes for retinal diseases, and their characterization promises to provide new insights into physiological processes in the retina. Clues to the initial pathogenetic processes in ND can be expected from the analysis of subtraction libraries prepared from early disease stages. Characterization of differentially expressed cDNAs from these libraries is in progress. 

 Supported in part by Grant Be 1559/3-1 from the German Research Foundation.

 Submitted for publication February 22, 2002; revised April 22, 2002; accepted May 20, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Subtraction efficiency test. Virtual Northern blot analysis of four RsaI fragments from different mRNAs on wild-type and knockout mice cDNAs before and after subtraction (forward: tester is wild-type; reverse: tester is knockout). The concentrations of the housekeeping transcripts (Gapdh and mt-Nd1) were considerably lower after forward and reverse subtraction. The clone inserts 90 and 975 are RsaI fragments from different mRNAs that are absent or significantly reduced in knockout tissue. The abundance of both fragments was highly enriched during forward subtraction.

Figure 2.

 

(A) cDNA microarray hybridization with subtracted targets. The microarray containing 3072 PCR-amplified cDNA clones from the forward-subtraction library and controls was hybridized with fluorescently labeled forward-subtracted (Cy3) and reverse-subtracted (Cy5) targets. Green signals: genes with a higher expression in the forward-subtracted target; red signals: false positives or artifacts of the subtraction process; and yellow signals: transcripts with an equal expression in both targets. Of the 3072 clones in the microarray, 1565 fulfilled the criteria for differential expression. White boxes: cDNA clones from the forward-subtraction library representing RsaI fragments of the indicated transcripts. (B) Confirmation of differential expression on virtual Northern blot analysis. Selected cDNA clones were hybridized on Southern blots containing equal amounts of PCR-amplified cDNA from whole eyes of 2-year-old wild-type and Ndp knockout mice, respectively. A clone for alcohol dehydrogenase family-3 (Ahd3), which exhibits identical expression levels in wild-type and Ndp knockout mice served as a control. (C, D) Hybridization of cDNA microarrays with PCR-amplified whole-eye cDNA targets derived from animals at 4 weeks (C) and 2 years (D) of age. The cDNAs from wild-type and Ndp knockout mice were fluorescently labeled with Cy3 and Cy5, respectively, and cohybridized to the microarray. Numerous differentially expressed clones were detected late in disease (D), but not in young Ndp knockout mice (C).

Table 1.

 

Differentially Expressed Genes

Gene Name	Gene Symbol (Alternate Symbols)	Accession Number	Human Orthologue	Function	Human Chromosome	Disease
Cyclic nucleotide gated channel, cGMP-gated	Cncg	U19717	CNGA1	Phototransduction	4p12-cen	Retinitis pigmentosa
Rod photoreceptor 1 (phosducin)	Rpr1	L08075	PDC	Phototransduction	lq25-q31.1
	(Rpr-1, Pdc)
Phosphodiesterase, gamma-subunit	Pde6g	Y00746	PDE6G	Phototransduction	17q25
	(Pdeg)
Phosphodiesterase, beta-subunit	Pde6b	X60133	PDE6B	Phototransduction	4p16.3	Retinitis pigmentosa, night blindness
	(rd, rd1, Pdeb)
Guanylate kinase 1	Guk1	U53514	GUK1	Phototransduction	lq32-q41
	(GMK)
Rhodopsin (rod opsin, L opsin, long-wavelength sensitive opsin)	Rho	M55171	RHO	Phototransduction	3q21-q24	Retinitis pigmentosa, night blindness
		(Ops, RP4, Opn2)
Opsin 1 (cone pigment), blue cone pigment	Opn1sw	U49720	OPN1SW	Phototransduction	7q31.3-q32	Colorblindness (tritanopia)
		(Bcp)
Guanine nucleotide binding protein, alpha transducing 1 (rod transducin)	Gnat1	U38504	GNAT1	Phototransduction	3p21	Night blindness
		(transducin, Tralpha)
Guanine nucleotide binding protein, alpha transducing 2 (cone transducin)	Gnat2	L10666	GNAT2	Phototransduction	1p13
Recoverin	Rcvrn	X66196	RCV1	Phototransduction	17p13.1
	(CAR, S-modulin)
Guanylate cyclase activator la	Guca1a	L36860	GUCA1A	Phototransduction	6p21.1	Cone dystrophy
	(Guca1, Gcap1)
Arrestin, retinal S-antigen	Sag	M24086	SAG	Phototransduction	2q37.1	Night blindness (Oguchi disease)
	(Irbp)
Interphotoreceptor matrix proteoglycan 1	Impg1	AF229929	IMPG1	Retinal adhesion, photoreceptor survival	6q14.2-q15
		(SPACR, IMP150)
Peripherin 2 (retinal degeneration slow)	Prph2	X14770	RDS	Structural photoreceptor protein	6p21.2-cen	Retinitis pigmentosa, macular dystrophy, retinitis punctata albescens, pattern dystrophy
	(Rd2, Rds)
Rod outer segment membrane protein 1	Rom1	M96760	ROM1	Structural photoreceptor protein	11q13	Retinitis pigmentosa, digenic
Calcium binding protein 5	Cabp5	AF169161	CABP5	Calcium binding, calcium homeostasis	19q13.33
Calumenin	Calu	U81829	CALU	Calcium binding, protein folding, and sorting	7q32
Prominin	Prom	AF039663	PROML1		4pter-p15.31	Early-onset severe retinal degeneration
Guanine nucleotide binding protein (G protein), gamma 1 subunit	Gng1	U38495	Unknown	Signal transduction, G-protein–coupled receptor signaling
		(G(y)1)
ADP-ribosylation-like 6 (ADP-ribosylation-like factor homologue)	Arl6	AF031903	Unknown	Signal transduction, Arf-like small GTPase (vesicle transport, endocytosis)
GTP binding protein 3	Gtpbp3-pending	AF325354	CRFG		10p15-p14
	(Crfg)
LGN protein	(HSU54999)	U54999	HSU54999	Guanine nucleotide exchange factor	1p36.13-q23.3

Table 1A.

 

Differentially Expressed Genes (continued)

Zinc finger protein 289 (RIKEN cDNA 2310032E02 gene)	Zfp289	AF229439	ZFP289	GTPase activating protein for Arf	11p11.2-p11.12
		(2310032E02Rik)
Zinc finger protein 101	Zfp101	U07861	Unknown	Regulation of gene transcription
Synaptosomal-associated protein, 25kDa	Snap25	M22012	SNAP25	Exocytosis	20p12-p11.2
	(sp)
N-myc downstream regulated 3	Ndr3	AB033922	FLJ13556
Karyopherin (importin) alpha 2	Kpna2	U34229	KPNA2	Intracellular protein trafficking	17q23.1-q23.3
	(Rch1, importin pendulin)
FKBP-associated protein	(Fap48)	U73704	FAP48	Protein complex assembly, protein folding	1pter-p21.3
Guanine nucleotide binding protein, betal subunit	Gnb1	U88324	Unknown
Muscleblind-like (Drosophila)	Mbnl	AF231110	MBNL	RNA-binding protein	3q25
			(EXP42, EXP40, EXP35)
CUG triplet repeat, RNA-binding protein 2	Cugbp2	AF090697	CUGBP2	RNA-binding protein	10p13
		(Napor, ETR-3, Napor2)
RNA binding motif protein 7	(Rbm7)	AF156098	RBM7	RNA-binding protein	11q23.1-q23.2
Fas apoptotic inhibitory molecule	Faim	AF130367	Unknown	Apoptosis inhibitor
Bcl2-associated athanogene 3 (Bcl-2-interacting death supressor)	Bag3	AF130471	BAG3	Apoptosis inhibitor	10pter-q21.3
		(Bis)
BCL2/adenovirus E1B 19-kDa-interacting protein 1	Bnip2	AF035207	BNIP2	Apoptosis inhibitor	15q21.2
Retinoschisin	Rs1h	AF084562	XLRS1	Cell adhesion(?)	Xp22.2	X-linked juvenile retinoschisis
	(Xlrs1)
Translocase of outer mitochondrial membrane	KIAA0016	D13641	TOM20	Mitochondrial protein import	1q42
		(TOM20)
RFamide-related peptide	Rfrp	AB040289	(RFRP)	Neuropeptide	7p21-p15
Elongation of very long chain fatty acids-like 4	(ElovL4)	AF277094	ELOVL4	Fatty acid biosynthesis	6q14	Macular dystrophy (Stargardt disease 3)
				(STGD3, STGD2, ADMD)
HSPC016: hypothetical protein	(Hspc016)	AF077202	HSPC016		3p21.33-3p21.1
Erythrocyte protein band 4.1-like 2	Epb4.112	AF044312	EPB41L2	Cytoskeleton, structural protein	6q23
	(4.1G, NBL)
Capping protein alpha 2	Cappa2	U16741	CAPZA2	Cytoskeleton, actin-binding, actin filament organization	7q31.2-q31.3
Keratin complex 1, acidic, gene 13	Krt1–13	U13921	KRTI3	Cytoskeleton, organization, and biogenesis	17q21-q23

Table 1B.

 

Differentially Expressed Genes (continued)

CD151 antigen	Cd151	D89290	CD151	Transmembrane signaling and cell adhesion	11p15.5
	(SFA-1, PETA-3)
Dynactin 6	Dctn6	AF124788	(WS-3)	Mitochondrial biogenesis	8p12-p11
	(p27, WS-3)
N-ethylmalemide sensitive fusion protein	Nsf	BC006627	NSF		17q21
		(SKD2)
Membrane protein, palmitoylated 4 (MAGUK p55 subfamily member4)	Mpp4	AB059357	MPP4		2q33.2
		(Dlg6a)
1-acylglycerol-3-phosphate O-acyltransferase 3	Agpat3	AK015906	AGPAT3	Metabolism of plasma lipoproteins	21q22.3
		(D10Jhu12e)
ATPase, H+ transporting, lysosomal 1	Atp6c1	U13839	ATP6C	Transporter	8p22-q21.3
Budding uninhibited by benzimidazoles 1 homologue, beta (S. cerevisiae)	Bub1b	AF107296	BUB1B	Cell cycle, normal mitotic progression	15q15
FBJ osteosarcoma oncogene B	Fosb	X14897	FOSB	Regulator of cell proliferation, differentiation, and transformation	19q13.32
Alpha glucosidase 2, alpha neutral subunit	G2an	U92793	(KIAA0088)		11q12.2
Heterogeneous nuclear ribonucleoprotein U	Hnrpu	AF073992	HNRPU	RNA processing	1q44
Potassium large-conductance calcium-activated channel, subfamily M	Kcnma1	U09383	KCNMA1		10q22
		(mSlo1)
Lactate dehydrogenase 1, A chain	Ldh1	U13687	LDHA	Anaerobic glycolysis	11p15.4
	(Ldha)
Myocyte enhancer factor 2C	Mef2c	L13171	MEF2C	Transcription factor	5q14
Meningioma-expressed antigen 5 (hyaluronidase)	Mgea5	AK014781	MGEA5	Glycoprotein degradation	10q24.1-q24.3
NMDA receptor-regulated gene 1	Narg1-pending	AK017653	Unknown		4q28.3
	(Tbdn-1)
Ornithine decarboxylase, structural	Odc	M10624	ODC1	Polyamine biosynthesis pathway	2p25
Origin recognition complex, subunit 6 homologue (Saccharomyces cervisiae)	Orc6-pending	AF139659	ORC6L		16q12
Pyruvate kinase 3	Pk3	X97047	PKM2	Glycolysis	15q22
Protease, serine, 18	Prss18	AB015206	Unknown
	(Bssp)
Secretory carrier membrane protein 1	Scamp1	AK015706	SCAMP	Exocytosis, vesicular transport	5q13.3-q14.1
Phosphoserine/threonine/tyrosine interaction protein	Styx	AK002822	STYX
		(hStyxb)

Figure 3.

 

Gene expression analysis during disease progression. The Ndp cDNA and four other confirmed differential clones (representing Bcp, Pde, Rpr1, and Sag) were hybridized to virtual Northern blots containing whole-eye cDNA from wild-type and Ndp knockout mice at 0.5, 1, 3, 9, 12, and 24 months of age. The Ndp transcript is only present in wild-type mice. All others genes show normal expression up to 12 months. Only in the oldest (2-year-old) Ndp knockout mice were the expression levels drastically reduced. Gapdh served as the control.

Figure 4.

 

Light micrographs of mouse retinas from a 16-month-old male wild-type (A), a 24-month-old male Ndp knockout (B), and a 29-month-old female homozygote Ndp knockout (C) mouse. The retinas of Ndp knockout animals (B, C) showed a drastic degeneration of all retinal cell layers, particularly of the photoreceptor inner and outer segments. Sections were stained with toluidine blue. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; IS, inner segment; OS, outer segment; RPE, retinal pigment epithelium. Scale bar, 50 μm.

The authors thank Wolfgang Mann and Roland Kirchner for advice on microarray technology and data analysis; Renate Kirschner, Ulrich Luhmann, and Christina Zeitz for helpful discussions and critical reading of the manuscript; and Ralph Schulz, Jens Fassen, Gerhild Lüder, and Rudi Lurz for technical assistance.