Molecular Vision 2003; 9:262-276 <http://www.molvis.org/molvis/v9/a38/> Received 22 April 2003 | Accepted 16 June 2003 | Published 20 June 2003

Download Reprint

Gene discovery in the embryonic chick retina

Abigail S. Hackam,¹ Rebecca L. Bradford,² Rita N. Bakhru,² Raza M. Shah,² Ronald Farkas,¹ Donald J. Zack,¹ Ruben Adler²
 
(The first two authors contributed equally to this publication)
 
 

¹Guerrieri Center for Genetic Engineering and Molecular Ophthalmology and ²Retina Degenerations Center, Wilmer Eye Institute, Johns Hopkins University School of Medicine, Baltimore, MD

Correspondence to: Ruben Adler, M.D., The Johns Hopkins University School of Medicine, 519 Maumenee, 600 North Wolfe Street, Baltimore, MD, 21287-9257; Phone: (410) 955-7589, FAX: (410) 955-0749; email: radler@jhmi.edu

Abstract

Purpose: The chick embryo is a powerful model system for the study of retinal development. However, analysis of gene expression in the chick retina has lagged behind biological studies. The purpose of this study was to identity and characterize genes expressed in the chick embryo retina as candidate molecules involved in the development and function of photoreceptors and other retinal cell types.

Methods: RNA from embryonic day (ED) 18 White Leghorn chick embryo retinae was used to generate an oligo dT-primed cDNA library. Bacterial colonies representing five thousand individual clones were arrayed onto nylon membranes using a microarray robot. Replicate membranes were hybridized with cDNA probes synthesized from ED 18 retina, brain and liver. Clones that appeared preferentially expressed in retina were identified by homology searches, and their spatial and temporal expression patterns were analyzed by in situ hybridization.

Results: Two hundred and seventy-two clones were identified. Approximately forty percent of the clones represented potential novel genes, including ESTs, hypothetical proteins and clones with no assigned identities. Furthermore, many genes were identified that are the putative chick orthologues of genes cloned from other species. We determined the expression pattern of several clones for which sequence homologies suggested possible roles in transcriptional regulation, apoptosis or intercellular signaling. Their corresponding mRNAs were expressed in the embryonic retina in topographically specific, developmentally regulated patterns.

Conclusions: We identified and characterized genes in the chick embryo retina using a combination of microarray analysis and in situ hybridization. Analysis of the expression patterns suggests involvement of several of these genes in key events during embryogenesis.

Introduction

The chick embryo has a number of advantages for the experimental analysis of retinal development. The embryonic eye, for example, is easily accessible from very early stages of development, allowing surgical manipulation [1-5], infection of retinal progenitor cells with replication incompetent or replication competent viruses [1,6-10], and pharmacological intervention [11,12]. This has made possible significant progress in a number of areas of retinal development, including cell lineage, the role of various transcription factors in retinal patterning and cell differentiation, and the mechanisms controlling guidance of ganglion cell axons towards specific regions of the CNS. In addition, the chick embryo retina is well suited for in vitro studies, and has been used for decades for explant, reaggregation and dissociated cultures [13-16].

Despite its substantial strengths, however, the chick retina does have certain limitations as a model system, many of which arise from the lack of well-developed genetic and genomic resources. Only a limited number of genes expressed in the developing and adult chick retina have been identified, and in many cases their patterns of expression have not yet been analyzed. This has limited substantially the scope of studies that could take advantage of powerful techniques available for delivering reagents for gain and loss of function experiments in the chick embryo retina, including electroporation and avian-specific retroviral vectors [6-8,17,18]. As an initial approach to this problem, in the present study we have isolated and characterized numerous genes that may be relevant for the development of photoreceptor and other retinal cell types. The strategy involved using an embryonic chick retinal nylon membrane-based microarray for high throughput screening of mRNA expression, followed by sequence analysis of abundantly expressed genes and in situ hybridization to determine their temporal and spatial patterns of expression at several stages of embryonic development. We identified 272 clones that include a large number of genes that had previously been cloned but not described in the chick retina, as well as others that represent altogether novel genes. Topographically-specific, developmentally-regulated patterns of expression were observed for many of these genes, suggesting that they could be involved in key events during embryogenesis.

Methods

Embryonic day 18 (ED 18) chick retina library

All procedures involving chickens were carried out in accordance with the statement by the Association for Research in Vision and Ophthalmology for the Use of Animals in Ophthalmic and Vision Research and was approved by the Animal Care and Use Committee at Johns Hopkins University. Neural retina tissue was dissected from ED 18 White Leghorn chick embryos and total RNA was extracted using TRIZOL reagent (Invitrogen, Carlsbad, CA), followed by polyA(+) RNA purification with the Message Maker kit (Gibco BRL, Rockville, MD). An oligo dT-primed cDNA library was constructed with the polyA(+) RNA using Superscript cDNA Library reagents (Gibco BRL), according to the manufacturer's protocols. The library contained 1.1x10⁷ primary clones, with an insert size range of 0.5-5 kb. The library was subsequently excised and converted to a PZL1 plasmid library (Gibco BRL), using endogenous CRE protein. Ninety percent of the viable colonies contained retina library inserts.

Generation of membrane-based microarrays

The ED 18 chick retina plasmid library was re-plated at low density, and 5,000 bacterial clones were randomly picked into 384-well plates, and cultured overnight in LB-ampicillin/8% glycerol. The bacterial cultures were arrayed in duplicate onto nylon membranes (Pall Biodyne B, Nunc, Rochester, NY) with a Microgrid II Robot (Biorobotics, Cambridge, England) using a 0.4 mm 384-well pin tool. The membranes were placed on an LB-ampicillin agar support, and the printed colonies were grown at 37 °C for 16-24 h. When the majority of the colonies on the membrane were visible, the membranes were treated with 0.5 M NaOH/1.5 M NaCl for 7 min, neutralized twice in 1 M Tris-HCl, pH 7.4 for 5 min, incubated in 0.5 M Tris-HCl (pH 7.4)/1.5 M NaCl for 4 min, and cross-linked with ultraviolet light (120 J/cm² for 30 s).

Probe generation for array analysis

Hybridization probes were synthesized from stage ED 18 brain, liver and retina tissues. Retinas were dissected under a microscope to exclude pigmented epithelium, lens and other extraretinal tissue. Tissue samples were processed immediately, or frozen in liquid nitrogen and stored at -80 °C. Total RNA was extracted using TRIZOL reagent (Invitrogen), the integrity and purity were assessed by gel electrophoresis and A₂₆₀/A₂₈₀ absorbance ratios, and mRNA was then purified using the Message Maker kit (Gibco BRL). First-strand cDNA was synthesized from 2 μg of mRNA using SuperScript II reverse transcriptase (Invitrogen), with random hexamers and ³³P-labeled dCTP. The membranes were prehybridized with 5X Denhardts reagent in 6X SSC/1% SDS for 2 h at 68 °C in a rotating hybridization oven. 5x10⁶ dpm/ml of labeled probe was then added with 100 μg/ml herring sperm DNA and hybridized overnight at 65 °C. The membranes were washed twice with 2X SSC/1% SDS at room temperature for 5 min and then with 0.1X SSC/1% SDS at 65 °C for 20 min.

Array data analysis

Hybridized membranes were exposed to phosphoimager screens for 4 h to 3 days and acquisition of radioactive images was performed using the Cyclone phosphoimager (Packard, Boston, MA) and OptiQuant software. Spot-finding and image analysis on the scanned images was performed using Imagene software (Biodiscovery, Marina Del Rey, CA). Automatic flagging of spots that were empty (indistinguishable from background) or poor (due to non-specific background) allowed elimination of faulty data. Clones preferentially expressed in the retina were identified by the Genesight software analysis program (Biodiscovery) and were confirmed manually by visual inspection of the spots.

Database searches

The plasmids corresponding to differentially expressed genes were purified from the bacteria and sequenced using the M13 reverse primer, yielding on average 600 bp of useful sequence information. Many of the clones were sequenced in both directions, permitting full sequencing of the insert. Clones with no significant matches in the database or those with poor quality sequence were re-sequenced in the opposite direction using a degenerate primer 5'-(T)17(A+C+G)(A+C+G+T)-3'.

Following editing of the sequences to remove segments corresponding to the vector, database searches were performed using the available public databases, with the non-redundant (nr) Blastn or Blastx algorithms at the National Center of Biotechnology Information (NCBI) [19]. Due to the relatively small number of chicken genes deposited in the databases, we expected that most of the matches for our clone set would represent cross-species homologies. Consequently, low stringency cut-off values (expectation values [E] of E=10^-10) were used as the criteria for determining homology. It should be noted that various other reports have used less stringent criteria (e.g., 10^-5) [20]. We used a slightly more stringent E-value to reduce false matches while enabling us to identify potential homologies in our cross-species comparisons; typically, E-values of 10^-15 or lower have been used in mammalian gene discovery efforts. A low complexity filter was used in the analysis. Domain and motif searches were performed using public online search tools. Additionally, all the matched sequence alignments were examined visually to exclude spuriously high probability values arising from repetitive sequences. Functional classification of the genes was based on the primary reported function obtained from the PubMed literature database (NCBI).

Probe synthesis for in situ hybridization

An approximately 400 bp insert PCR product was amplified from individual clone plasmids using a gene-specific forward primer containing a T3 RNA polymerase overhang (18-20 bp of perfect homology; T3 sequence: AATTAACCCTCACTAAAGGGAGA) and a reverse primer corresponding to plasmid sequence (M13/pUC reverse primer: AGCGGATAACAATTTCACACAGG). This amplification strategy resulted in the inclusion of a T7 RNA polymerase sequence from the plasmid. The resulting PCR product was gel-purified using β-agarase digestion. In vitro transcription was performed in the presence of digoxigenin-11-uridine-triphosphate (DIG-UTP) to produce DIG-UTP-labeled single-stranded antisense RNA (using T7 polymerase) or sense RNA (using T3 polymerase) probes using the DIG RNA Labeling Kit (Roche Molecular Biochemicals, Indianapolis, IN), according to the manufacturer's instructions. Transcript concentration was estimated by comparison against a DIG-labeled RNA control (Roche).

In situ hybridization

Tissues were dissected in RNase-free Hanks Balanced Salt Solution, fixed at 4 °C in 4% paraformaldehyde/PBS containing 5% sucrose, incubated in an increasing concentration series of sucrose in 0.1 M phosphate-buffered saline (PBS), and embedded in a 2:1 mixture of 0.1 M PBS with 20% sucrose and OCT in PBS [21]. Seven and 10 μm thick cryosections were collected on Superfrost Plus slides, rinsed in PBS, treated with proteinase K (10 μg/mL) for 1 min, rinsed in PBS, and pre-hybridized for 2 h at 65 °C in 50% formamide, 5X SSC, 100 μg/ml heparin, 0.1% Tween 20, 1 mg/ml tRNA, 1X Denhardts, 0.1% CHAPS and 5 mM EDTA. Sections were then covered with 100 μl of 400 ng/ml probe in the prehybridization solution, and incubated overnight at 60 °C in a humidified chamber. Post-hybridization washes were as follows: (1) twice in a 50% formamide/2X SSC solution at 60 °C for 30 min each; (2) three times in 2X SSC at 37 °C for 5 min; (3) RNAse A and T1 RNAse in 2X SSC at 37 °C for 15 min; (4) 50% formamide, 0.1% CHAPS, in 2X SSC at 60 °C for 15 min; (5) 50% formamide, 0.1% Tween 20, in 0.2X SSC for 15 min. Hybridization was detected with the DIG Nucleic Acid Detection kit according to the manufacturer's instructions (Roche Molecular Biochemicals).

Results

Classification of genes expressed in chick embryo neural retina

The microarray generated for this study included 5,000 randomly selected clones from a non-normalized embryonic day (ED) 18 chick retina cDNA library (see Materials and Methods for details). ED 18 retina represents a fairly advanced stage of development at which time all retina cell types have acquired a mature differentiated phenotype, and their synaptic connections have been established at the plexiform layers. To identify genes that are highly expressed and/or preferentially expressed in the retina, comparative hybridizations were performed with probes generated from ED 18 retina, brain, and liver. Although microarray studies can be used with appropriate statistical analysis to provide quantitative differential expression data, we chose to use them as a qualitative "first-pass" filter to rapidly identify potentially interesting genes, focusing further studies on their sequence and in situ hybridization analysis. This proved to be a rewarding strategy, yielding over 100 genes that had not previously been described in the chick retina.

Approximately 40% of the 5000 arrayed clones hybridized to the retina probe under the stringent conditions used. Of these, 272 hybridized more intensely to the retina probe than to the brain probe (or, in an initial smaller experiment, than to the liver probe), and were therefore selected for sequencing. Readable sequence of at least 300 bp was obtained for 236 clones. The sequences were classified into three main categories based on bioinformatic analysis (see Methods for a description of the criteria used): (1) previously cloned genes with known function from chick and other species, which were further classified into the ten functional groups in Table 1; (2) genes with unknown function including hypothetical proteins, predicted genes and ESTs with no homology to known genes ("ESTs/Unknown function" in Table 1); and (3) clones that were unclassifiable ("Unassigned identity/no homologs" in Table 1), including clones that had no matches in the database with the stringency criteria used, as well as clones that only had homologies to genomic regions. Categories 2 and 3 include clones that potentially represent novel genes.

Sequence analyses demonstrated that 188 (80%) of the clones were in categories 1 and 2, in that they matched genes or ESTs present in public databases (functional classification in Table 1 and Table 2). For the clones in category 1, 71 matched to previously cloned chick sequences whereas 65 were homologous to transcripts not previously cloned from the chick, suggesting that we have identified their chick orthologue (or a gene closely related to their orthologue). Their orthologues were found mostly in mammals, except the homologue of CRG (chick retina gene) 178 that was only present in another bird, the Japanese quail. As described below, there was little information in the literature about ocular expression and/or function of most of these homologues. As expected, we identified known retina-enriched genes in our clone database (for example, melanopsin, transducin). However, the total number of clones representing vision genes was low (4%). The reason why visual function genes were not abundant in the library is unclear, although it is apparent from in situ hybridization results (see below) that there were several examples of clones representing genes with clear photoreceptor expression.

For category 2 clones, we identified 25 clones that matched only to ESTs found in the chick, 8 clones that matched to ESTs found in mammals, and 19 were identified only as novel genes of unknown function (including predicted genes and hypothetical proteins). Some of the ESTs had weak similarity to known genes but the majority was "anonymous" in that they lacked recognizable predicted protein motifs, and their identity and function are currently unknown. Of interest, we have identified a large number of potentially novel genes in this study (clones in categories 2 and 3; clone and sequence information in Table 2). Approximately 40% of the clones with readable sequence belonged to category 3, having no significant matches in the databases (48 clones), or matched ESTs and/or genes of unknown function (category 2, 52 clones). Genes in this group may represent bona fide novel genes, but could conceivably have diverged so significantly from genes in other species that a significant match could not be detected with the available sequences and databases. It is also possible that the lack of sequence homology of clones in this category may in some cases be due to 3' UTR sequences present in the clones, which typically contain larger regions of lower sequence homology than the coding regions of the same mRNA. Additional sequence information could facilitate identification of the clones in those hypothetical cases.

Developmental expression analysis

The temporal and spatial expression of a selection of these newly cloned genes was studied by in situ hybridization, using a developmental series of chick embryo retinas that included the following stages: ED 5, when neuroepithelial cells are actively proliferating, and some cells are beginning terminal mitosis and differentiation; ED 8, when cell proliferation is largely complete, particularly near the fundus of the retina, and most postmitotic retinal progenitors have relocated to their future laminar positions; ED 12 and ED 15, when the various cell types of the retina are undergoing cell differentiation and forming specific neuronal interconnections; and ED 18, when cell differentiation is essentially complete, and the retina has a mature organization [22,23]. For clarity of presentation, the description of the various genes that were analyzed by in situ hybridization will be presented following the classification described above. The frequency of each category is listed in Table 1. Control sense strand hybridization was negative in all cases (Figure 1F,J, Figure 2I,N, Figure 3D,K,O, Figure 4D,H,L,P, and Figure 5G,J,P,V).

ESTs

As described above, many of the clones had sequence matches to ESTs that did not contain motifs or domains that could indicate their identity (see Table 1 and Table 2). As one example, clone CRG73 was homologous only to an EST from a subtracted chick eye library. In our analysis, CRG73 expression appeared relatively low at early developmental stages, but became much higher (and more localized) as the retina matured. At ED 5 there were practically no detectable hybridization signals in the retinal fundus, although some positive cells could be seen at higher magnification (not shown). Signals were more evident at the retinal periphery, near the future ciliary epithelium (Figure 1A). This pattern remained largely unchanged by ED 8 (not shown), but at ED 12 ganglion cells were positive, and the inner nuclear layer INL (INL) showed scattered, but abundant positive cells (Figure 1B). The putative photoreceptor layer remained negative at this stage, as it did at ED 15 (Figure 1C). At the later stage, signals were very strong in the ganglion cell layer (GCL) and in the inner part of the INL, and even stronger in the central region of the INL, in marked contrast with the light staining in the INL adjacent to the outer plexiform layer (OPL; Figure 1C). By ED 18, positively hybridizing cells were found both in the ganglion layer and in the inner part of the INL at the periphery of the retina (Figure 1E), but only in ganglion cells at the midperiphery (not shown) and fundus (Figure 1D).

Clone CRG123 was homologous to an EST isolated from a normalized chicken pituitary/hypothalamus/pineal cDNA library. Neither the clone nor the EST has recognizable motifs or known homologues. The expression of CRG123 changed developmentally from widespread to cell-type specific. Strong signals were seen throughout the neural retina at ED 5, except in the region immediately adjacent to the RPE (Figure 1G). The overall pattern persisted at ED 8, although the periphery of the retina had much lighter hybridization signals than the fundus (not shown). Widespread expression in all retinal layers was noted at more advanced stages (illustrated for ED 15 in Figure 1H), with the outer half of the INL appearing darker than its inner half. Some cells in the INL showed a circular outline and very strong hybridization signals (Figure 1I); their identity remains unknown, but they were reminiscent of apoptotic cells [24].

Genes with unknown function

The first group in Table 1 also contains clones with significant homology to predicted genes or hypothetical proteins in the databases that have no established annotated function. We have performed expression analyses on those genes that contain sequence motifs that could suggest possible roles in retina development. An example is clone CRG9, which is homologous to the predicted human protein KIAA0728. The presence of a dystrophin-like domain in KIAA0728 suggests that it, and by inference, CRG9, may belong to a group of dystrophin-like molecules that have been implicated in development [25]. In ED 5 embryos, hybridization signals were present throughout the fundal region of the neural retina (Figure 2A-B), but were more restricted at the periphery, where they predominated near the RPE and the vitreal chamber (Figure 2A,C). As the retina acquired some degree of lamination by ED 8, the putative ONL appeared negative in the fundal region of the eye adjacent to the optic nerve, while the INL remained positive throughout, with occasional cells appearing darker than their neighbors (Figure 2D,E). At the periphery (Figure 2E) there was some heterogeneity in expression throughout the retinal epithelium. In more mature retinas (illustrated for ED 18 in Figure 2F-H), hybridization signals were strong in ganglion and amacrine cells, and much weaker in the rest of the INL (Figure 2F,G). Hybridization was undetectable in photoreceptors at the periphery (Figure 2F), but was very strong in their inner segments at the fundus (Figure 2G,H).

CRG31 is homologous to the human hypothetical protein FLJ20113, a gene of unknown function that lacks any recognizable motif. Cognate ESTs are broadly distributed in human, mouse and bovine tissues. The most interesting aspects of its expression were seen in ED 15-18 retinas, when very strong signals were observed in ganglion cells, and in the inner segments (but not the cell bodies) of photoreceptors (Figure 2K-M). Within the INL, expression was strong in many cells of the amacrine layer, in scattered cells in the region corresponding to glial cells of Müller, and in the region abutting the OPL (Figure 2K,L). In contrast to this selective pattern of expression, younger retinas (ED 5-8) showed a much more widespread pattern of expression, although the innermost and outermost aspects of the neuroepithelium appeared less positive than its center (Figure 2J).

Unassigned identity/no homologs

The third category of potentially novel genes is the 46 clones that had no significant matches in the databases to genes or ESTs. This category also included genes that had regions of similarity to mammalian genomic clones, but for which no other sequence similarities were found, making them currently unclassifiable. These clones (19% of total) could represent novel genes, or novel splice forms of previously cloned genes, or could correspond to expressed genes that were not previously identified due to low abundance in cDNA libraries (see Discussion). A representative clone is CRG92, which yielded practically no hybridization signals in the retina at ED 5 (not shown), and was only lightly and diffusely positive at ED 8 (Figure 3A); at this stage the fundus was stronger than the periphery (not shown). In a fully differentiated retina (ED 18), the outer part of the INL showed strongest signals, while ganglion cells and photoreceptor cells were lightly stained or negative (Figure 3C). A similar pattern was detectable at ED 15 (Figure 3B).

Newly cloned chick genes

Clones with significant matches to genes previously cloned in other species could be either the chick orthologue of such genes, or a chick gene closely related to the true orthologues. The categories of the various functions of these genes are listed in Table 1. We chose for further analysis genes whose involvement in the development of non-ocular tissues has been reported, but whose expression in the chick retina was unknown.

CRG111 is homologous to human LATS (large tumor suppressor, Drosophila) homolog 1, a serine/threonine kinase [26] that has been reported to suppress cell proliferation and induce apoptosis in mammalian and fly tissues [27]. As shown in Figure 3E, neural retinal signals appeared light (or absent) in the region adjacent to the RPE, but were stronger more vitreally at ED 5. The RPE was negative at the periphery, near the lens (Figure 3E), but appeared positive in the midperiphery (Figure 3F) and fundus (not shown). The periocular mesenchyme appeared positive as well (Figure 3F), as did the neural tube (not shown). By ED 8 the neural retina appeared fairly homogeneously positive (Figure 3G), although near the ora serrata, hybridization signals appeared strong towards the vitreal surface but negative adjacent to the RPE (not shown). The lens showed a negative anterior epithelium, but elongated fibers had perinuclear signals (Figure 3H). By ED 15 (not shown), and particularly by ED 18, maximum expression was seen in ganglion cells, with intermediate levels in the INL and weak staining in photoreceptors (Figure 3I). Analysis at high magnification showed heterogeneity both within the INL and the ONL (Figure 3J), with positive cells (*, arrowhead) alternating with negative ones.

CRG137 is homologous to human and mouse seizure-related gene SEZ-6, a brain-specific gene that is upregulated in response to convulsant drugs [28]. SEZ-6 expression has not been studied in the retina. The overall hybridization patterns with this probe changed from diffuse at early stages of development, to fairly localized in the differentiated retina. On ED 5 there were strong signals throughout the circumference of the retina, although they were stronger in its vitreal than its scleral side (Figure 3L). The extraretinal mesenchyme (Figure 3L) and the neural tube (not shown) were also positive. The retina (except for the extraretinal mesenchyme) maintained this pattern of expression at ED 8 (not shown). The ganglion cells, the INL and some cells in the photoreceptor layer appeared clearly positive at ED 12 (Figure 3M). Further restrictions in signal distribution were noted by ED 15 (not shown) and ED 18 (Figure 3N). In the latter case, intense signals were observed in many (but not all) cells in the amacrine region of the INL. Photoreceptors appeared negative, and the ganglion cells showed very light signals. Conspicuous signals were also detected in (or adjacent to) the OPL (Figure 3N).

CRG150 is homologous to human CGI-130 protein. This protein is predicted to contain a metal-dependent phosphohydrolase domain, found in enzymes involved in nucleic acid metabolism and signal transduction, including the visual transduction protein 3',5'-cGMP phosphodiesterase. Cognate ESTs have been isolated from human neural retina and numerous other tissues. The most striking observation in the chick retina was the exquisite localization of this gene to ganglion cells at ED 18, when there were only very faint hints of expression in some cells in the INL and essentially none in the ONL (Figure 4C). Expression also predominated in the GCL at intermediate stages of development (e.g., ED 12, Figure 4B), although there was also detectable expression in the INL, particularly in the region of amacrine cells. The ONL was negative. Expression was light and fairly diffuse at ED 5 (Figure 4A) and ED 8 (not shown).

The sequence of CRG177 has overall low homology to known genes, but does contain a 130 nucleotide region with 85% identity to human γ-transducin activity polypeptide 2 (GNGT2, cone transducin) and a second region of 180 bp that has 100% identity to an EST from a chick eye library. Therefore, part or all of this clone could be a chick orthologue of human cone transducin, or it may represent a novel transducin family member. A probe from the region homologous to the eye library EST demonstrated that the gene has late, markedly cell type-specific expression: strong signals were seen in photoreceptor inner segments on ED 18, but not in other photoreceptor regions or cell types or at any of the stages studied before ED 18 (Figure 4E-G).

CRG220 is homologous to the human gene sacsin. Mutations in sacsin lead to the neurological disease spastic ataxia of Charlevois-Saguenay [29], which features prominent myelination of retinal nerve fibers. As illustrated in Figure 4I-K, its overall pattern of expression in chick retina was fairly generalized throughout the developmental stages studied. At ED 8, for example, the only detectable heterogeneity was a somewhat darker signal in the future INL (Figure 4I). All cell layers showed clear signals in more differentiated retinas (Figure 4J,K). Some noteworthy details were the presence of patches with more intense signals in the INL, and the differential distribution of signals in the photoreceptors, which predominated in their cell bodies at ED 15 (Figure 4J) but were stronger in their inner segments on ED 18 (Figure 4K).

CRG231 is homologous to human protein kinase C-binding protein Zeta 1, also known as fasciculation and elongation protein (FEZ1). FEZ1 is believed to play a regulatory role in axonal guidance during C. elegans development [30] and is possibly involved in neuronal differentiation of PC12 cells [31]. In our study, a generally diffuse pattern of expression was observed in ED 5 (not shown) and ED 8 retinas (Figure 4M). In the latter case, however, the periphery appeared less stained than the fundus (not shown), and the ganglion cell region and putative ONL appear lighter than the rest of the neuroepithelium. Signals were predominantly localized to the INL and GCL at ED 12 (not shown) and ED 15 (Figure 4N), with little expression in the ONL. By ED 18 the ganglion cells were the most conspicuous positive cell type (Figure 4O); in the INL some amacrine cells appeared darker than the rest of its cells. Photoreceptor cell bodies were negative, but some inner segment had detectable signals (Figure 4O).

Known chick genes

The clones described in this section are identical to chick genes that have been previously characterized in non-retinal tissue (Table 1). The first such gene, CRG110, corresponds to the chicken gene for macrophage migration inhibitory factor (MIF), primarily known for its role in the immune response [32] but is also involved in chick lens differentiation [33]. The spatial expression pattern of CRG110 became substantially restricted during development. Expression in the ED 5 neural retina was widespread, but predominated near the RPE and the vitreal surface of the retina (Figure 5A). The RPE was largely negative (Figure 5A), but some signal could be seen in the RPE near the origin of the optic nerve (Figure 5B). As previously reported, the anterior epithelium of the embryonic lens was negative or lightly positive [33]; we observed that signal intensity increased markedly at the lens equator, and very strong signals were seen in the area where cells are known to become postmitotic and start to differentiate and elongate (Figure 5C, arrows), whereas mature lens fibers were weakly positive or negative (Figure 5C). Some expression was also observed in the extraretinal mesenchyme (Figure 5A) and in the neural tube (not shown). At ED 8 there was conspicuous absence of hybridization signals from the prospective ONL region; the rest of the retina showed widespread but somewhat uneven expression (Figure 5D). The pattern of expression in the lens remained as on ED 5 (not shown). As the retina matured, hybridization signals appeared predominantly localized to the GCL and the inner part of the INL (prospective amacrine cells), as illustrated for ED 12 and 18 (Figure 5E,F). The photoreceptor layer was negative.

CRG100 has significant sequence similarity to the chick Defender Against Death gene (DAD1), a protein that inhibits programmed cell death and enhances cellular proliferation in certain cell types [34,35]. DAD1 is part of the oligosaccharyltransferase enzyme complex that initiates N-linked glycosylation [36]. CRG100 showed little, if any, cell type-specificity of expression throughout development. Fairly widespread signals were seen in the retinal neuroepithelium at ED 5 (Figure 5H) and ED 8 (not shown). Similarly, all retinal layers showed hybridization in more matured retinas (illustrated for ED 18 in I) although ganglion cells appeared lighter than the remaining retinal layers. It is noteworthy that photoreceptor cell bodies were intensely stained, but their inner segments appeared devoid of signal.

Clone CRG196 was identified as stathmin-like 2 (SCG10-like), a member of a family of proteins that regulate microtubules and may play a role in axonal and dendritic outgrowth during neuronal development [37,38]. This probe yielded a very distinct hybridization pattern. On ED 5, when most other genes were either not expressed or expressed in fairly diffuse patterns, CRG196 signals were localized to a small population of cells adjacent to the vitreal surface of the retina, and only in its fundal region (Figure 5K,L). This localization and the large size of the cells suggest that they were newly generated ganglion cells [39,40]. It must be noted, however, that this apparent ganglion cell-specific pattern of expression was only observed in thinner tissue sections, whereas thicker sections also showed some signal in the fundal (but not the peripheral) retinal neuroepithelium (Figure 5L), as well as in the ciliary epithelium (not shown). Expression continued to be restricted to the GCL by ED 8, extending further into the periphery as compared to ED 5 (Figure 5M). As the retina matured, ganglion cells appeared strongly positive throughout the retina, but were accompanied by cells scattered in the region corresponding to putative amacrine cells that were also conspicuously positive (illustrated for ED 15 (Figure 5N) and ED 18 (Figure 5O)). The photoreceptor layer was negative at all time-points.

The CRG233 clone was identified as a fragment of the chicken HT7 antigen gene, which encodes a member of the immunoglobulin super gene family. HT7, also known as 5A11 and basigin, has been implicated in the formation and maintenance of the blood-retina barrier [41] as well as in cell-cell recognition during retinal development [42]. On ED 5, expression appeared stronger in the fundus than the periphery of the retina (Figure 5Q). The fundus continued to be very positive on ED 8 (Figure 5R), with a clear demarcation between the positive retina and the negative optic nerve (Figure 5S). The INL appeared to be much lighter at the periphery than in the fundus, but very darkly stained ganglion cells could be observed in this area (Figure 5T). The ciliary epithelium showed strong, polarized hybridization signals (not shown). As the retina differentiated (illustrated for ED15 in Figure 5U), the INL appeared strongly positive in its outer half, and somewhat lighter more internally; photoreceptors showed weak staining and ganglion cells appeared very lightly stained.

Discussion

The goal of this study was to discover genes expressed in the embryonic chick retina, which could represent candidates for future experimental analysis in retinal development and function. Retinal gene expression was surveyed by differential library screening on robotically printed membranes. Genes abundantly and/or preferentially expressed in the retina were sequenced and analyzed by in situ hybridization to determine their developmental and spatial expression patterns. This gene discovery approach was successful because the majority of the selected clones had no assigned function and/or represented genes not previously identified in the chick. This successful outcome was achieved despite the relatively small number of clones screened (5000), and without a thorough quantitative comparison of expression ratios of genes in different tissues. Moreover, most of the genes that had been previously identified in the chick or other species had not been studied in the retina. Due to the highly specialized nature of the retina, genes involved in retinal homeostasis might be expected to be conserved across species, permitting the identification of chick genes by sequence similarities. In contrast, the anatomical differences between the chick and mammalian eye and the increased number of visual pigments in the chick retina may also result in the identification of genes specific to the avian retina. Even with moderately stringent statistical probability cut-off values to identify putative homologues in the databases, 40% of the clones in this study appeared to be novel, having no significant matches in the databases, or only matching ESTs or genes of unknown function. A caveat for the interpretation of these results is that the library was generated by oligo dT priming, which could have resulted in some messages being predominantly represented by their less well conserved 3' UTR. In some cases, the unknown genes did match weakly to genomic regions, but we considered these genes as "unassigned" since they did not have similarity to definitively predicted or cloned genes. Finally, because this was a first-pass sequencing project we were primarily concerned with obtaining sequence from the clones in order to facilitate homology searches, as has been done in the majority of published sequencing efforts. Sequencing of the entire coding region of the identified genes will be necessary before further investigations of their functional significance are undertaken.

It has been reported that the proportion of novel or "anonymous" clones found in a gene-hunting study varies amongst various species and tissues [43]. The frequency of anonymous clones in our clone set (40%) falls within the range of anonymous ESTs found by other groups [20]. It is of course conceivable that we may have failed in some cases to find significant matches for our genes because their homologues may have low expression, leading to their absence from the database, as well as due to the relatively low number of chick genes in the public databases. Additional investigations of the novel genes isolated in this study, including identification of the full-length sequences and manipulation of gene expression, will be important for determining whether any play critical roles in the development and function of the chick retina.

Developmental patterns of gene expression

From a developmental point of view, the group that contained the largest number of genes was characterized by detectable, although diffuse, expression in the neuroepithelium of ED 5 embryos. Some of these genes (e.g., CRGs 100, 123 and 220) retained a very generalized distribution even in mature retinas, suggesting that they may be associated with general metabolic or other "common" activities. In other cases, however, the diffuse pattern changed over development into a layer-specific distribution (e.g., CRGs 31, 69, and 110). Additional genes that were detectable on ED 5 already had a somewhat restricted expression at this stage; in some cases (e.g., CRG 233) expression was more intense in the fundal than in the peripheral region of the retina, and in others (e.g., CRGs 111 and 137) signals were stronger in the inner (vitreal) than in the outer regions of the retina. Four additional genes were very low or undetectable on ED 5, and appeared subsequently in layer-specific patterns or in a single cell type (e.g., CRG177, γ-transducin, which was undetectable through ED 15 and, when first observed in ED 18, was specifically restricted to photoreceptor cells). A unique expression pattern was seen with CRG196, which even at the earliest developmental stages studied appeared very intensely positive in cells that could be identified as newly generated ganglion cells; this identification [39,40] is based on their large size, position (adjacent to the inner limiting membrane) and regional distribution (restricted to the fundal region of the eye on ED 5). When the above-mentioned temporal variations are evaluated from a cellular perspective, it appears noteworthy that the morphologically undifferentiated neuroepithelium (at ED 5) already expresses many genes that are later found in layer-specific patterns in the mature retina. While the meaning of these molecular similarities between neuropithelial cells and subsets of differentiated cells remain unclear, it is interesting that in several cases these genes are unevenly distributed within the retinal neuropithelium. Such results suggest that the "undifferentiated" neuroepithelium, although morphologically homogeneous and developmentally uncommitted [13], is in fact heterogeneous at the molecular level [6].

Another cellular feature of potential interest is that, in differentiated retinas, cells in the ganglion and amacrine layers are by far the ones that were strongly positive for the largest number of gene products; within the INL, the region adjacent to the IPL frequently had stronger signals than the region adjacent to the OPL (e.g., CRGs 9, 100, 110, 111, 137, 196, 233), while the opposite pattern was less frequent (CRGs 92, 123). Photoreceptor cells were only infrequently positive and, when they were, signals appeared in some cases in the inner segment (e.g., CRGs 177, 233) and in other cases in the perinuclear region of the cell bodies (CRG 100, and in some cells in CRG 123). The explanation for this relative lack of genes whose expression is enriched in photoreceptors is unclear. It remains to be determined whether these differences reflect biologically meaningful cell properties; ongoing studies using microarrays to analyze cDNAs from individual cells may provide insight to these questions (Bradford and Adler, in preparation).

Without detailed biological studies it is not possible to assess how expression of the identified genes contributes to retina development. It is noteworthy, however, that very powerful methods are now available to study the function of specific genes in the chick embryo retina, since reagents for loss of function or gain of function experiments can be readily delivered to retinal cells using lipid-mediated transfection [44], avian retroviral vectors [6-8] or electroporation [17,18]. These methods should allow direct testing of the working hypotheses that can be generated based on the developmental expression patterns and/or sequence characteristics of the genes that we have discovered.

Acknowledgements

This work was supported by NIH grants EYO4859 (RA), EY 09769 (DJZ), EYO0416 (RF), and Core Grant EY1765, a center grant from the Foundation Fighting Blindness (Baltimore, MD), a grant from the Macula Vision Foundation, an unrestricted departmental grant from Research to Prevent Blindness, Inc. (New York, NY), and by funds from Mrs. Harry J. Duffey, Mr. and Mrs. Marshall and Stevie Wishnack, and from Mr. and Mrs. Robert and Clarice Smith. ASH is supported by a Canadian Institutes of Health Research Senior Research Fellowship. RA is the Arnall Patz Distinguished Professor of Ophthalmology; DJZ is the Guerrieri Professor of Genetic Engineering and Molecular Ophthalmology; and both are recipients of Senior Investigator Awards from Research to Prevent Blindness, Inc.

References

1. Fekete DM, Cepko CL. Retroviral infection coupled with tissue transplantation limits gene transfer in the chicken embryo. Proc Natl Acad Sci U S A 1993; 90:2350-4.

2. Dutting D, Meyer SU. Transplantations of the chick eye anlage reveal an early determination of nasotemporal polarity. Int J Dev Biol 1995; 39:921-31.

3. Uemonsa T, Sakagami K, Yasuda K, Araki M. Development of dorsal-ventral polarity in the optic vesicle and its presumptive role in eye morphogenesis as shown by embryonic transplantation and in ovo explant culturing. Dev Biol 2002; 248:319-30.

4. Mueller BK, Dutting D, Haase A, Feucht A, Macchi P. Partial respecification of nasotemporal polarity in double-temporal chick and chimeric chick-quail eyes. Mech Dev 1998; 74:15-28.

5. Trisler D, Rutin J, Pessac B. Retinal engineering: engrafted neural cell lines locate in appropriate layers. Proc Natl Acad Sci U S A 1996; 93:6269-74.

6. Adler R, Belecky-Adams TL. The role of bone morphogenetic proteins in the differentiation of the ventral optic cup. Development 2002; 129:3161-71.

7. Fekete DM, Perez-Miguelsanz J, Ryder EF, Cepko CL. Clonal analysis in the chicken retina reveals tangential dispersion of clonally related cells. Dev Biol 1994; 166:666-82.

8. Yan RT, Wang SZ. neuroD induces photoreceptor cell overproduction in vivo and de novo generation in vitro. J Neurobiol 1998; 36:485-96.

9. Das I, Sparrow JR, Lin MI, Shih E, Mikawa T, Hempstead BL. Trk C signaling is required for retinal progenitor cell proliferation. J Neurosci 2000; 20:2887-95.

10. Skeith A, Dunlop L, Galileo DS, Linser PJ. Inhibition of beta1-integrin expression reduces clone size during early retinogenesis. Brain Res Dev Brain Res 1999; 116:123-6.

11. Desire L, Head MW, Fayein NA, Courtois Y, Jeanny JC. Suppression of fibroblast growth factor 2 expression by antisense oligonucleotides inhibits embryonic chick neural retina cell differentiation and survival in vivo. Dev Dyn 1998; 212:63-74.

12. Bovolenta P, Frade JM, Marti E, Rodriguez-Pena MA, Barde YA, Rodriguez-Tebar A. Neurotrophin-3 antibodies disrupt the normal development of the chick retina. J Neurosci 1996; 16:4402-10.

13. Adler R. A model of retinal cell differentiation in the chick embryo. Prog Retin Eye Res 2000; 20:529-57.

14. Layer PG, Robitzki A, Rothermel A, Willbold E. Of layers and spheres: the reaggregate approach in tissue engineering. Trends Neurosci 2002; 25:131-4.

15. Seigel GM. The golden age of retinal cell culture. Mol Vis 1999; 5:4 <http://www.molvis.org/molvis/v5/a4/>.

16. de la Rosa EJ, Diaz B, de Pablo F. Organoculture of the chick embryonic neuroretina. Curr Top Dev Biol 1998; 36:133-44.

17. Koshiba-Takeuchi K, Takeuchi JK, Matsumoto K, Momose T, Uno K, Hoepker V, Ogura K, Takahashi N, Nakamura H, Yasuda K, Ogura T. Tbx5 and the retinotectum projection. Science 2000; 287:134-7.

18. Sakuta H, Suzuki R, Takahashi H, Kato A, Shintani T, Iemura Si, Yamamoto TS, Ueno N, Noda M. Ventroptin: a BMP-4 antagonist expressed in a double-gradient pattern in the retina. Science 2001; 293:111-5.

19. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol 1990; 215:403-10.

20. Lee Y, Sultana R, Pertea G, Cho J, Karamycheva S, Tsai J, Parvizi B, Cheung F, Antonescu V, White J, Holt I, Liang F, Quackenbush J. Cross-referencing eukaryotic genomes: TIGR Orthologous Gene Alignments (TOGA). Genome Res 2002; 12:493-502.

21. Barthel LK, Raymond PA. Improved method for obtaining 3-microns cryosections for immunocytochemistry. J Histochem Cytochem 1990; 38:1383-8.

22. Belecky-Adams T, Cook B, Adler R. Correlations between terminal mitosis and differentiated fate of retinal precursor cells in vivo and in vitro: analysis with the "window-labeling" technique. Dev Biol 1996; 178:304-15.

23. Belecky-Adams T, Tomarev S, Li HS, Ploder L, McInnes RR, Sundin O, Adler R. Pax-6, Prox 1, and Chx10 homeobox gene expression correlates with phenotypic fate of retinal precursor cells. Invest Ophthalmol Vis Sci 1997; 38:1293-303.

24. Cook B, Portera-Cailliau C, Adler R. Developmental neuronal death is not a universal phenomenon among cell types in the chick embryo retina. J Comp Neurol 1998; 396:12-9.

25. Blank M, Blake DJ, Kroger S. Molecular diversity of the dystrophin-like protein complex in the developing and adult avian retina. Neuroscience 2002; 111:259-73.

26. Xu T, Wang W, Zhang S, Stewart RA, Yu W. Identifying tumor suppressors in genetic mosaics: the Drosophila lats gene encodes a putative protein kinase. Development 1995; 121:1053-63.

27. Xia H, Qi H, Li Y, Pei J, Barton J, Blackstad M, Xu T, Tao W. LATS1 tumor suppressor regulates G2/M transition and apoptosis. Oncogene 2002; 21:1233-41.

28. Shimizu-Nishikawa K, Kajiwara K, Sugaya E. Cloning and characterization of seizure-related gene, SEZ-6. Biochem Biophys Res Commun 1995; 216:382-9.

29. Engert JC, Berube P, Mercier J, Dore C, Lepage P, Ge B, Bouchard JP, Mathieu J, Melancon SB, Schalling M, Lander ES, Morgan K, Hudson TJ, Richter A. ARSACS, a spastic ataxia common in northeastern Quebec, is caused by mutations in a new gene encoding an 11.5-kb ORF. Nat Genet 2000; 24:120-5.

30. Bloom L, Horvitz HR. The Caenorhabditis elegans gene unc-76 and its human homologs define a new gene family involved in axonal outgrowth and fasciculation. Proc Natl Acad Sci U S A 1997; 94:3414-9.

31. Kuroda S, Nakagawa N, Tokunaga C, Tatematsu K, Tanizawa K. Mammalian homologue of the Caenorhabditis elegans UNC-76 protein involved in axonal outgrowth is a protein kinase C zeta-interacting protein. J Cell Biol 1999; 144:403-11.

32. Nishihira J. Macrophage migration inhibitory factor (MIF): its essential role in the immune system and cell growth. J Interferon Cytokine Res 2000; 20:751-62.

33. Wistow GJ, Shaughnessy MP, Lee DC, Hodin J, Zelenka PS. A macrophage migration inhibitory factor is expressed in the differentiating cells of the eye lens. Proc Natl Acad Sci U S A 1993; 90:1272-5.

34. Apte SS, Mattei MG, Seldin MF, Olsen BR. The highly conserved defender against the death 1 (DAD1) gene maps to human chromosome 14q11-q12 and mouse chromosome 14 and has plant and nematode homologs. FEBS Lett 1995; 363:304-6.

35. Hong NA, Flannery M, Hsieh SN, Cado D, Pedersen R, Winoto A. Mice lacking Dad1, the defender against apoptotic death-1, express abnormal N-linked glycoproteins and undergo increased embryonic apoptosis. Dev Biol 2000; 220:76-84.

36. Kelleher DJ, Gilmore R. DAD1, the defender against apoptotic cell death, is a subunit of the mammalian oligosaccharyltransferase. Proc Natl Acad Sci U S A 1997; 94:4994-9.

37. Ozon S, Maucuer A, Sobel A. The stathmin family -- molecular and biological characterization of novel mammalian proteins expressed in the nervous system. Eur J Biochem 1997; 248:794-806.

38. Ozon S, Byk T, Sobel A. SCLIP: a novel SCG10-like protein of the stathmin family expressed in the nervous system. J Neurochem 1998; 70:2386-96.

39. Snow RL, Robson JA. Ganglion cell neurogenesis, migration and early differentiation in the chick retina. Neuroscience 1994; 58:399-409.

40. McLoon SC, Barnes RB. Early differentiation of retinal ganglion cells: an axonal protein expressed by premigratory and migrating retinal ganglion cells. J Neurosci 1989; 9:1424-32.

41. Liebner S, Gerhardt H, Wolburg H. Maturation of the blood-retina barrier in the developing pecten oculi of the chicken. Brain Res Dev Brain Res 1997; 100:205-19.

42. Fadool JM, Linser PJ. 5A11 antigen is a cell recognition molecule which is involved in neuronal-glial interactions in avian neural retina. Dev Dyn 1993; 196:252-62.

43. Cirera S, Wintero AK, Fredholm M. Why do we still find anonymous ESTs? Mamm Genome 2000; 11:689-93.

44. Toy J, Bradford RL, Adler R. Lipid-mediated gene transfection into chick embryo retinal cells in ovo and in vitro. J Neurosci Methods 2000; 104:1-8.