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


     

(Investigative Ophthalmology and Visual Science. 2007;48:5229-5242.)
© 2007 by The Association for Research in Vision and Ophthalmology, Inc.
DOI:  10.1167/iovs.07-0122
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplementary Tables
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
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 ISI Web of Science
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 Google Scholar
Google Scholar
Right arrow Articles by Lupien, C. B.
Right arrow Articles by Salesse, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lupien, C. B.
Right arrow Articles by Salesse, C.

Comparison between the Gene Expression Profile of Human Müller Cells and Two Spontaneous Müller Cell Lines

Caroline B. Lupien,1,2 Carl Bolduc,3 Solange Landreville,1,2 and Christian Salesse1,2

1From Unité de Recherche en Ophtalmologie, 2Département d’ORLO, Faculté de Médecine, and 3Laboratoire de Recherche en Endocrinologie Moléculaire et Oncologie, Centre de Recherche du CHUQ and Université Laval, Ste-Foy, Québec, Canada.


   Abstract
Top
 Abstract
Methods
 Results
 Discussion
 References
 
PURPOSE. Müller cells are the principal glial cells in the retina. They play important functions in this tissue. Alterations in Müller cell behavior were observed in the retinal tissue of patients with proliferative diabetic retinopathy. The purpose of this study was to determine the gene expression profile of normal human Müller cells (NHMCs) and to compare it with that of two spontaneously generated human Müller cell lines (HMCLs) from type 1 and type 2 diabetic donors by using serial analysis of gene expression (SAGE).

METHODS. Approximatively 50,000 tags were sequenced for each SAGE library. Identification of the transcripts was obtained by matching the 15-bp tags with the UniGene and GenBank databases. Classification of the genes was based on the updated information of the genome directory found at the National Center for Biotechnology Information (NCBI) Web site.

RESULTS. SAGE was used to characterize the entire transcriptome of human Müller cells and to compare these data with those from Müller cell lines generated from type 1 and type 2 diabetic donors. The transcriptome of NHMCs and HMCLs demonstrated that "metabolism" and "protein synthesis" are the two main categories of genes expressed by human Müller cells. Only 106 genes are differentially expressed between NHMCs and HMCLs.

CONCLUSIONS. The SAGE libraries reported in this article provide the gene expression profile of NHMCs and HMLCs. It thus represents a valuable source of information regarding the function of Müller cells as well as their role in the development of diabetic retinopathy.

Among the four types of glial cells in the retina, Müller cells are the most important ones. They extend throughout the whole thickness of the neural retina, with their nucleus in the inner nuclear layer, and many fine processes which surround neuronal cell bodies, axons, and blood vessels.1 2 The association between Müller cell processes and retinal neurons implies a close functional relationship between these two cell types. Müller cells subserve many of the metabolic, ionic, and extracellular buffering requirements of neurons.1 Although much is yet to be learned about the functions of Müller cells in the retina, it is clear that they play important roles in its development, in the preservation of its integrity, and in maintaining neuronal survival.3 4 5 By virtue of their transretinal orientation, Müller cells are positioned to respond to several initiating events in the retina.6 7 Virtually all retinal diseases are associated with reactive Müller cell gliosis, which may be protective and support the survival of retinal neurons or may exacerbate the progress of neuronal degeneration.8 Diabetic retinopathy is a complication of diabetes and the leading cause of blindness in the active population of developed countries. There is an emerging body of evidence suggesting that neuronal changes are an early phenomenon in the diabetic retina and that several cell types are affected, including the neuronal and glial cells.9 10

Biological events are associated with changes in the expression of key genes. During the onset and progression of diseases, extensive changes take place in gene expression. By comparing gene expression profiles under different conditions, individual genes or group of genes that play important roles in a particular signaling cascade or process or in disease etiology can be identified. Serial analysis of gene expression (SAGE) is a highly efficient method that can provide the global gene expression profile of a particular type of cell or tissue.11 12 13 Two major principles underlie SAGE: short expressed sequence tags (ESTs) are sufficient to identify individual gene products, and these unique sequence tags (15 bp) can be concatenated into long DNA molecules and identified by sequence analysis.11 14 With the ever-expanding sequence information available in public databases, identification of gene transcripts with SAGE tags has greatly facilitated transcriptome comparison and gene identification.15 The SAGE method has also been used in a wide variety of applications such as analysis of the effect of drugs on tissues or disease-related genes and gaining insights into disease pathways.16

A few SAGE analyses of retinal tissue have been conducted during the past few years. Blackshaw et al.17 were the first ones to study the mouse retina by SAGE, followed with a SAGE analysis of each cell type involved in mouse retinal development including Müller cells.18 Sharon et al.19 were the first to use SAGE to compare different regions of the retina. Recently, Kobashi-Hashida et al.20 analyzed the expression profile of the human choroid and the retinal pigment epithelium. More recently, Bowes Rickman et al.21 presented EyeSAGE, a transcriptome database of human retina and RPE/choroid. To our knowledge, no one has prepared a SAGE library of human Müller cells. In the present study, we used the SAGE method to determine the gene expression profile of normal human Müller cells (NHMCs) and two previously characterized human Müller cell lines (HMCLs)22 generated from donors with type 1 or type 2 diabetes, as well as to compare genes differentially expressed between NHMC and HMCLs.


   Methods
Top
 Abstract
 Methods
Results
 Discussion
 References
 
This study was conducted in accordance with the Declaration of Helsinki and our institution’s guidelines.

Preparation of Samples
The two spontaneous human Müller cell lines obtained from donors with type 1 (HMCL-I) or type 2 (HMCL-II) diabetes have been recently characterized.22 Isolation and culture of HMCLs and NHMC were performed as previously described.22 23 Total RNA from HMCLs and NHMCs has been isolated by using an RNA extraction kit (Qiagen, Mississauga, ON, Canada).24

Transcriptome Analysis
The SAGE method was performed as previously described.11 25 26 27 In brief, polyadenylated RNA was extracted with an mRNA purification system (Oligotex; Qiagen), annealed with the biotin-5-T18-3 primer and converted to cDNA with the cDNA synthesis kit (Invitrogen, Carlsbad, CA). The resultant cDNA library was digested with NlaIII (anchoring enzyme), and the 3' restriction fragments were isolated with streptavidin-coated magnetic beads (Dynal Biotech, Carlsbad, CA) and separated in two populations. Each population was ligated to one of the two annealed linker pairs and extensively washed to remove unligated linkers. The tag beside the most 3' NlaIII restriction site (CATG) of each transcript was released by digestion with BsmFI (tagging enzyme). The blunting kit from Takara Co. (Otsu, Japan) was used for the blunting and ligation of the two tag populations. The resulting ligation products containing the ditags were amplified by PCR with an initial denaturation step of 1 minute at 95°C, followed by 22 cycles of 20 seconds at 94°C, 20 seconds at 60°C, and 2 seconds at 72°C with 27-bp primers.27 The PCR product was digested with NlaIII, and the band containing the ditags was extracted from the acrylamide gel. The purified ditags were self-ligated to form concatemers. The concatemers of 500 to 1800 bp were isolated by agarose gel electrophoresis. The resulting DNA fragments were cloned into the SphI site of pUC19, and bacterial transformation was performed (UltraMAX DH5{alpha}FT; Invitrogen). White colonies were screened by PCR to select long inserts for automated sequencing by the Plateforme de Génomique du Centre de Recherche du CHUL (Centre Hospitalier de l’Université Laval; Service de Séquençage et de Génotypage).

Bioinformatic Analyses
Sequence files were analyzed by the SAGEana program, a modification of SAGEparser (ftp://ftp.pbrc.edu/public/eesnyder/SAGE/). Tags corresponding to linker sequences were discarded, and duplicate concatemers were counted only once. Identification of the transcripts was obtained by matching the 15 bp (CATG + 11 bp tags) with the UniGene (www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=unigene; provided by the National Center for Biotechnology Information [NCBI], Bethesda, MD) and GenBank (http://www.ncbi.nlm.nih.gov/GenBank/; NCBI) databases. The matching procedure used was very restrictive because, to avoid the possibility of sequencing errors in the expressed sequence tag (EST) database, we did not consider the matches that were identified only once among the numerous sequences of a UniGene cluster. Indeed, the chance of matches with ESTs containing sequencing errors decreases dramatically when at least two ESTs are identified in a UniGene cluster for a given tag sequence. In addition, a minimum of one EST with a known polyA tail had to be in the UniGene cluster to identify the last NlaIII site on the corresponding cDNA. Classification of the genes was based on the updated information of the genome directory28 found at the Institute for Genomic Research (TIGR; J. Craig Venter Institute, La Jolla, CA).

Several normal libraries (library ID: 8, 9, 94, 99, 135, and 136; cortex, heart, kidney, liver, lung, and astrocytes) as well as normal retinal libraries (library ID: 162, 163, 164, 1990, 1991, and 1994) from SAGE Genie were normalized to 50,000 tags before being loaded into our database together with our NHMC library. These SAGE Genie libraries were obtained through the SAGE Genie Web site (ftp://ftp1.nci.nih.gov/pub/sage/human/). The Hs.libraries.gz and Hs_short.frequencies.gz files of interest were processed using Perl, a programming language (http://www.perl.org), and information regarding the tags was loaded into a SQLite database (http://www.sqlite.org). Only the tags that were sequenced two times or more were included in the database, to avoid possible sequencing errors and the tremendous background that the single tags introduced in the comparison. SQL queries were then designed to determine the analogy between the NHMC 11-bp tags with the 10-bp tags present in the other libraries. This process was used to compare the NHMC library against a subset of the SAGE Genie libraries so that percentages of similarity between the various libraries could be obtained.

Statistical Analyses
We used the comparative count display (CCD) test to identify the transcripts that were significantly differentially expressed (P < 0.05) between the groups (NHMC and HMCLs libraries) with more than a twofold change. The CCD test makes a key-by-key comparison of two key-count distributions by generating a probability that the frequency of any key in the distribution differs by more than a given factor from the other distribution. This statistical test has been described elsewhere.29 The data are normalized to 50,000 tags to facilitate visual comparison in the tables.

RT-PCR Analyses
Total RNA from HMCLs and NHMCs were isolated and first-strand cDNAs were synthesized and used for semiquantitative determination of mRNA levels of different genes by PCR, as previously described.24 Gene-specific primers were designed to amplify cDNA fragments of fibronectin, cytochrome c oxidase subunit II, NADH dehydrogenase subunit IV, keratin 18, insulin growth factor binding protein 7 (IGFBP7), cathepsin D, and transgelin (Table 1) . The oligonucleotide primers used for the amplification of the 18S ribosomal RNA for the semiquantitative RT-PCR analyses were provided with the kit (Quantum RNA 18S Internal Standards; Ambion Inc., Austin, TX) and the measurement was performed according to the manufacturer’s instructions. The primer-competimer ratio used, the annealing temperature, and the total number of cycles for each gene, as well as the cDNA size of each PCR product are presented in Table 1 . The cycle parameters were the same for both primer sets used (denaturation 94°C, 1 minute; annealing, 1 minute; extension 72°C, 1 minute). Band density was evaluated by the Service d’Analyze d’Image du CHUL with an image analysis software (Scion Image, Fredericksburg, MD).


View this table:
[in this window]
[in a new window]

  		TABLE 1. Sequence of the Primers and Details of the Experimental Conditions of the Semiquantitative RT-PCR Analyses

 

   Results
Top
 Abstract
 Methods
 Results
Discussion
 References
 
Preparation of SAGE Libraries
The characterization of the two novel human Müller cell lines (HMCL-I and -II)22 has shown only a few differences between HMCLs and NHMCs by RT-PCR analyses, but large differences in growth rate and proliferation.22 SAGE libraries were thus prepared with mRNA from NHMC, HMCL-I, and HMCL-II, to determine the gene expression profile of each of these cell types and genes differentially expressed between NHMCs and HMCLs.

In total, 47,933 (NHMC), 49,877 (HMCL-I), and 48,955 (HMCL-II) tags were sequenced for each library (Table 2) . Then, 42,682 (NHMC), 47,155 (HMCL-I), and 43,611 (HMCL-II) tags were selected for further analysis (Table 2) . All SAGE tag sequences from these libraries have been submitted to the GEO database at the NCBI (GSE 3118). The analysis of the libraries determined that 25,743 (NHMC), 22,845 (HMCL-I), and 24,178 (HMCL-II) tags matched a unique sequence (Table 2) that corresponds to a UniGene entry. A significant number of tags corresponding to no match were thus classified as possible novel genes (Table 2) .


View this table:
[in this window]
[in a new window]

  		TABLE 2. Summary of SAGE Analyses of the NHMC, HMCL-I, and HMCL-II Libraries

 
Tag-to-Tag Assignments
Among the tags corresponding to a unique sequence, 54% (NHMC), 52% (HMCL-I), and 62% (HMLC-II; Table 2 ) were associated with known genes. The other tags were associated with ESTs or hypothetical proteins (Table 2) . Based on the expression profile of known genes, Table 3 presents the percentages of the various categories of protein function for each library. The detailed list of tags corresponding to known genes for each library is presented as supplementary data (Tables S1–S3; all Supplementary Tables are online at http://www.iovs.org/cgi/content/full/48/11/5229/DC1). The largest category, which accounted for at least 20% of these genes, corresponded to functionally unknown genes (uncharacterized; Table 3 ). The second largest category comprised proteins involved in metabolism and accounted for 16% to 20% of the genes (Table 3) . The metabolism category was subdivided into six distinct categories, with metabolism energy and metabolism enzyme being the largest ones (Table 3) . The third largest category comprised proteins involved in protein synthesis which accounted for 15% to 18% of the genes (Table 3) . The proteins from the channels/signaling/transport (8%–10%) and cytoskeleton (6%–8%) categories were also expressed at high frequencies (Table 3) .


View this table:
[in this window]
[in a new window]

  		TABLE 3. Main Categories of Functional Proteins Represented in NHMC and HMCLs

 
Validity Assessment of the SAGE Libraries
The three Müller cell libraries analyzed in this study shared between 51% and 59% identical tags. For critical evaluation of the quality of our libraries, we compared the NHMC library with those of various tissues, using libraries from the SAGE Genie database. Although the SAGE Genie libraries contain 10-bp tags, we managed to determine how many 11-bp tags from our NHMC library were analogous to the 10-bp tags of the libraries from various tissues. Furthermore, the total number of sequenced tags from the different SAGE Genie libraries varied between 22,000 and 127,000 tags compared with 50,000 tags for our library which influenced the percentage of similarity between libraries. Therefore, a comparison was performed with selected normalized SAGE Genie libraries to 50,000 tags. When tissues other than the retina were compared with our Müller cell library, 30% to 45% similarity was obtained between these libraries. The highest similarity of 45% was obtained when our NHMC library was compared with the astrocytes library. Astrocytes and Müller cells are both glial cells, and they share similar functions.1 The high similarity between these two libraries is thus not surprising. However, only 20% to 30% similarity was obtained when retinal libraries were compared with NHMCs. This rather low similarity could be explained by the fact that Müller cells account for only 6% of total retinal cells.30 Moreover, given that our NHMC library differed from the retinal libraries in the total number of sequenced tags and in the length of the tags (11 bp vs. 10 bp), we undertook an additional comparison to test the effect of these parameters. We thus compared our NHMC library with a retinal library prepared at our institute (Raymond V, private communication, July 15, 2005) in identical experimental conditions. In this case, 33% similarity was obtained, which is very close to the highest similarity (30%) we obtained with the retinal libraries from the SAGE Genie database. Moreover, this result provides confidence in the comparative analyses of various tissues described earlier. Finally, it is important to stress that none of the libraries tested showed a similarity as high as when our NHMC library was compared with our HMCL libraries (51%–59%).

A comparative analysis was also made with retinal SAGE Genie libraries to determine tags that were specific or enriched in retinal Müller cells. Therefore, specific tags were obtained by seeking sequences that were present only in the NHMC library. Since all retinal SAGE Genie libraries contain more than 50,000 tags, the data were normalized to 50,000 tags to make a valid comparison with our NHMC library of 50,000 tags. Moreover, enriched tags in NHMCs were selected by calculating the ratio of transcript expression between the NHMC library and the retinal libraries. Consequently, 5271 tags were found only in NHMCs and were thus considered specific to Müller cells (Supplementary Table S4). Among those tags, 30 were sequenced 10 times or more in the NHMC library and can be thus considered specific and highly expressed by Müller cells. Furthermore, 239 tags were present in the NHMC library and in at least one of the retinal libraries used for comparison (Supplementary Table S5). These tags were sequenced 10 times more often in the Müller cells than in the retina and can thus be considered enriched. A large number of these specific enriched genes from the NHMC library were included in the extracellular matrix and cytoskeleton categories.

We further assessed the validity of the NHMC SAGE library by evaluating tags from genes previously known to be expressed in Müller cells (Table 4) . The tag corresponding to the cytoskeletal protein vimentin,31 a specific marker of Müller cells, was expressed 168 times in the NHMC library. Moreover, only one tag corresponding to the glial fibrillary acidic protein (GFAP) was found in the NHMC library, which is in good agreement with studies indicating that GFAP is normally expressed at low levels in Müller cells.32 33 34 Compared to NHMCs, a lower expression of VEGF was found in HMCLs, which is consistent with previous RT-PCR analyses with these cells22 as well as with Müller cells from diabetic rats.35 Additional proteins known to be expressed by Müller cells such as CD 4436 37 ; cadherin 2 type 138 ; bFGF39 ; LIF40 ; TGFß241 ; IGFBP242 ; PDGF-{alpha}-R43 ; EGF-R44 45 ; VEGF46 ; agrin47 ; collagen types I, II, and IV48 49 50 ; fibronectin51 ; laminin51 ; tenascin52 53 ; thrombospondin; vitronectin54 ; chondroitin sulfate55 ; heparan sulfate56 ; and decorin57 were found in the NHMC library, as can be seen in Table 4 .


View this table:
[in this window]
[in a new window]

  		TABLE 4. Genes Known to Be Found in Müller Cells

 
Characterization of Each Library
The 50 most highly expressed tags from NHMCs, HMCL-I, and HMCL-II are presented in Table 5 . The most highly expressed tag in the NHMC library is fibronectin, which is part of the extracellular matrix–cell adhesion category. A large number of additional extracellular matrix components are expressed at high level in the NHMC library such as collagen types I, II, and IV48 49 50 ; laminin51 ; tenascin52 53 ; thrombospondin54 ; and vitronectin54 (Table 5) . A significant number of tags corresponding to the protein synthesis and cytoskeleton categories can also be found in the NHMC library. The gene encoding ferritin which is involved in iron metabolism is represented among the 10 most highly expressed tags in NHMCs (Table 5) . A high level of expression of iron metabolism proteins has also been observed in a SAGE library of the retina.19 A large number of tags in the channel/transport protein category was also found in the NHMC library, which is consistent with one of the important functions of Müller cells. Indeed, these cells regulate the extracellular environment of the retina, and thus specific ion channels and transport systems are expressed by Müller cells.58


View this table:
[in this window]
[in a new window]

  		TABLE 5. Fifty Most Highly Expressed Genes in the Different Libraries

 
The mitochondrial protein cytochrome c oxidase subunit II from the metabolism energy category was the most highly expressed tag in the HMCL-I library (Table 5) . Protein synthesis and metabolism were the main categories represented in this library. Consistent with these data, the most highly expressed tag found in the HMCL-II library (Table 5) also corresponded to a mitochondrial protein from the metabolism energy category, NADH dehydrogenase subunit IV. One can thus postulate that the mitochondrial respiratory chain is very active in HMCLs. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a key enzyme in glycolysis, was also highly expressed in HMCL-II (Table 5) . Energy production in Müller cells relies entirely on glycolysis, with a low dependency on oxygen.59 It has been demonstrated that GAPDH accumulates in the nucleus of Müller cells of diabetic rats,60 which is thus in good agreement with our data.

RT-PCR analyses were performed with the most expressed tag–associated gene from each SAGE library (fibronectin, cytochrome c oxidase, and NADH dehydrogenase; Table 5 ), by using cDNA of the corresponding libraries (Fig. 1) . An 18S ribosomal cDNA fragment (489 bp) was coamplified as a control for both cDNA synthesis and PCR efficiency. The results show that fibronectin was more strongly expressed (more than threefold) by NHMCs than by the HMCLs (Fig. 1A) . In contrast, cytochrome c oxidase expression was much higher (more than sixfold) in HMCL-I than in NHMCs (Fig. 1B) , whereas the level of expression of NADH dehydrogenase was stronger (more than twofold) in HMCLs than in NHMCs (Fig. 1C) . These data are consistent with their expression level in each SAGE library (Table 5) .


Figure 1
View larger version (53K):
[in this window]
[in a new window]

  		FIGURE 1. Semiquantitative RT-PCR analyses of the most highly expressed tag in each SAGE library. RT-PCR analyses were performed with RNA from NHMCs (second lane), HMCL-I (third lane), and HMCL-II (fourth lane) using the RNA expression level of the most expressed gene from each SAGE library: (A) fibronectin (830 bp), (B) cytochrome c oxidase subunit II (581 bp), and (C) NADH dehydrogenase subunit IV (955 bp). An 18S ribosomal cDNA fragment was coamplified as a control for both cDNA synthesis and PCR efficiency. The position of both cDNAs and 18S fragments (489 bp) are indicated along with those of relevant markers. Bottom: band density was calculated with an image analysis software, and the intensity of each PCR band was divided by the length of the corresponding PCR product and normalized with the intensity of the corresponding 18S RNA band. The ratio was calculated by using the values obtained for NHMC (A), HMCL-I (B), and HMCL-II (C) as the basal level.

 
Genes Differentially Expressed between NHMCs and HMCLs
The program SAGEana61 was used to analyze genes differentially expressed among the three libraries. This program finds transcripts that are differentially expressed with a ratio of 2 or greater among different samples, according to the statistical CCD test.29 The Venn diagram presented in Figure 2 is a schematic representation of the tag relationship among the three different SAGE libraries. Normal human Müller cells were thus compared to Müller cells generated from diabetic donors. It can be seen that approximately one third (2721) of the tags were common to the three libraries. A detailed description of the genes that are differentially expressed in this diagram is presented in Table 6 . The NHMC library was chosen as the reference. The genes differentially expressed are thus represented as an increase or a decrease in their expression (indicated by an arrow in Table 6 ) with respect to NHMCs. Two mitochondrial proteins, NADH dehydrogenase subunit IV and ATP synthase F0 subunit 6, were the most highly increased genes that were differentially expressed by HMCLs. This increase in the genes involved in cellular respiration in the HMCL libraries is consistent with previous observations with human skeletal muscles from diabetic patients.62 The extracellular matrix/cell adhesion proteins fibronectin and collagen types 1, alpha 1, and 2 accounted for the mostly decreased genes in both HMCLs when compared with NHMCs (Table 6) . We also observed a decrease in the insulin growth factor binding protein (IGFBP) and an increase in basic fibroblast growth factor (bFGF) in both HMCLs compared with the NHMCs. Comparison of the genes differentially expressed among the three SAGE libraries revealed that the expression of seven no-match tags increased in HMCLs, whereas the expression of five no match tags decreased. Those no-match tags could be novel genes.


Figure 2
View larger version (17K):
[in this window]
[in a new window]

  		FIGURE 2. Venn diagram depicting the distribution of SAGE tags differentially expressed among the three libraries. The tags included are only those that are associated with known genes and do not contain unmatched tags or tags corresponding to ESTs. Only differentially expressed tags between the libraries are represented. The tags used were expressed with a twofold difference between libraries.

 

View this table:
[in this window]
[in a new window]

  		TABLE 6. Genes Differentially Expressed between NHMC and HMCLs

 
To validate further the differential expression between NHMCs and HMCLs, we performed additional RT-PCR analyses with tag-associated genes that were differentially expressed among the three libraries (Fig. 3) . Two enriched genes from NHMCs showing decreased expression in HMCLs were selected, IGFBP-7 (Fig. 3A) and transgelin (Fig. 3B) , as well as two genes showing increased expression in HMCLs, cathepsin D (Fig. 3C) and keratin 18 (Fig. 3D) . Almost all IGFBP gene family members were expressed in NHMCs (Table 6) . IGFBP-7 was the most highly expressed member of this family in the NHMC library (339 times)—that is, five- and twofold higher than in HMCL-I and HMCL-II, respectively (Fig. 3A) . Transgelin expression was more than sixfold higher in NHMCs than in HMCLs. Transgelin is an actin gelling protein found in smooth muscle.63 Currently, little is known about its function. As Gunnersen et al.64 showed by SAGE, transgelin is expressed by human astrocytes but its expression decreases in glioblastoma. Cathepsin D has been localized in rabbit and rat Müller cells.65 66 67 68 This enzyme has also been suggested to be involved in protein catabolism after phagocytosis of debris from degenerating retinal neurons by Müller cells.69 70 A fourfold increase in cathepsin D expression was observed in HMCL-I (insulin-dependent) compared with HMCL-II (insulin-independent) and NHMCs (Fig. 3C) , consistent with the data of Nerurkar et al.71 who have shown that insulin treatment of 1-month-old streptozotocin (STZ)-induced diabetic animals restores the decreased activity of cathepsin D to a level greater than normal. Finally, keratin 18 was four times more highly expressed by HMCLs than by NHMCs (Fig. 3D) , which is consistent with our previous RT-PCR analyses and with the epithelioid morphology of the HMCLs.22


Figure 3
View larger version (70K):
[in this window]
[in a new window]

  		FIGURE 3. Semiquantitative RT-PCR analyses of tags differentially expressed between the SAGE libraries. RT-PCRs were performed with RNA from NHMCs (second lane), HMCL-I (third lane), and HMCL-II (fourth lane). Four differentially expressed genes were tested: (A) IGFBP7 (311 bp), (B) transgelin (615 bp), (C) cathepsin D (415 bp), and (D) keratin 18 (240 bp) were analyzed, along with a 18S ribosomal cDNA fragment that was coamplified as a control for both cDNA synthesis and PCR efficiency. The position of the cDNAs and 18S fragments (489 bp) are indicated along with those of relevant markers. Bottom: band density was calculated by an image analysis software. The intensity of each PCR band was divided by the length of the corresponding PCR product and normalized with the intensity of the corresponding 18S RNA band. Ratio levels were calculated according to the value obtained for NHMC (A, B), HMCL-I (C), and HMCL-II (D) as the basal level.

 
Tags Related to Retinal Diseases
Forty tags were found to correspond to genes related to retinal diseases (Table 7) . For example, genes such as the ones associated with Refsum disease (phytanoyl-CoA hydroxylase), Wolfram syndrome (wolframin), macular dystrophy (vitelliform macular dystrophy 2), glaucoma (optineurin), optic atrophy (OPA1 and -3), Bardet-Biedl syndrome (Bardet-Biedl syndrome-2, -4, and -5), retinitis pigmentosa (EST retinitis pigmentosa-9), and Sorsby fundus dystrophy (tissue inhibitor of metalloproteinase-3) can be found in Table 7 . Moreover, vascular endothelial growth factor (VEGF), with expression that is known to be increased during diabetic retinopathy,72 was not increased in HMCLs, which is consistent with our previous RT-PCR analyses.22


View this table:
[in this window]
[in a new window]

  		TABLE 7. Genes Expressed in Retinal Diseases

 

   Discussion
Top
 Abstract
 Methods
 Results
 Discussion
References
 
SAGE is a powerful tool that works by isolating short fragments of genetic information from the expressed genes that are present in the cell under study.16 Its main advantage over other methods is that it does not require prior knowledge of the genes of interest and provides qualitative and quantitative data of potentially every transcribed sequence in a particular cell or tissue type.16 SAGE has the capability of detecting and quantifying the expression of large numbers of known and unknown transcripts11 and has been demonstrated to be reproducible.61 The computer program SAGEana was used to analyze the NHMC and HMCL libraries to eliminate artifacts and to adequately treat the replicated ditags. Indeed, this program eliminates the replicated concatemers and ditags of improper length, detects the presence of any vector sequence in the concatemer sequences, and does not consider them any further.61 In this study, we have generated the first human retinal Müller cell (NHMCs) SAGE library. To create this library and reduce individual variability, we pooled cells from human donors between 2 and 40 years of age. Moreover, we also determined the expression profile of two human Müller cell lines from diabetic donors (HMCL-I and -II).

A small number of human retinal SAGE libraries have been prepared before now.19 21 The mouse retinal developmental study from Blackshaw et al.18 represents the only available data containing partial information on Müller cells. Therefore, the data related to Müller cells in Supplementary Tables S5 and S7 (mouse adult Müller cells) from Blackshaw et al.18 have been loaded into our database and compared to our NHMC library. The 72 genes common to both libraries are presented in Supplementary Table S6. Among these genes, the most important categories are metabolism, transcription and protein synthesis (many tags are also associated with ribosomal genes), and transport. Müller cells play an active role in several metabolic processes that are vital to normal retinal function. They also exhibit many features of other glial cells, such as the expression of high-affinity carrier-mediated transport systems.73 74 75 It is now firmly established that glia are major players in the active removal of neuroactive substances76 77 and are probably engaged in similar activities in the retina. It is thus not surprising to find a high number of transport systems genes expressed in the Müller cells library such as the Solute carrier family 25 member 12 (mitochondrial carrier) and the ATPase H+ transporting (Atp6ap2). Of the 72 genes in Table S6, only 13 correspond to tags specific to Müller cells (Supplementary Table S4). These genes include the four and a half LIM domains 1 (FHL1) and cadherin 11 (CDH11). The exact function of FHL1 remains unclear. In recent studies, the expression patterns of FHL1 appeared to be coordinated with important factors such as myogenin78 and involved in the Notch signaling pathway to repress transcription.79 CDH11 is highly expressed in a subset of cells of the inner nuclear layer of the retina80 and more specifically in Müller cells.81 Cadherins have been shown to be involved in many cell adhesion, morphogenesis, and cell growth and differentiation processes.82 In their study, Marchong et al.80 found that CDH11 mRNA levels decreased in the maturing murine retina, suggesting an important role for this protein during retinal development. Other genes associated with development can be found in both human and mouse Müller libraries, including cyclin D2, stem cell growth factor, chromatin factor assembly 1, dickkopf homolog 3 (Dkk3), and clusterin. Dkk3 is a member of the dickkopf family of Wnt-signaling regulatory genes83 and has been implicated in the control of cell proliferation.84 85 Wnt signaling pathways are important signal-transduction mechanisms that mediate essential processes in embryonic and adult tissues, ranging from cell proliferation and differentiation to body axis determination and synaptic plasticity.86 Expression of the Dkk3 gene has not been reported in the degenerating retina, whereas the expression patterns of several dickkopf family members and Wnt-related genes have been described in the developing and adult normal eye.87 88 Dkk3 transcripts were localized to the inner nuclear layer, possibly in the glia.89 It has been demonstrated that clusterin, a heterodimeric, acidic, and sulfated glycoprotein,90 is a protective molecule that is upregulated during physiological stress such as neuronal development.91 Recently, Kim et al.92 observed the clusterin expression in different phases of retinal development in the inner nuclear layer of the retina until maturity. Furthermore, Blackshaw et al.18 demonstrated that the expression of Dkk3 and clusterin in the adult mouse retina is restricted to Müller cells, which is consistent with the data found in the NHMC library.

The most highly expressed tag in the NHMC library, fibronectin, is an important component of the extracellular matrix and was found to be expressed in the inner limiting membrane of the retina.51 However, it has not yet been shown to be synthesized by Müller cells. The strong fibronectin expression found in NHMCs could be explained by the necessity for Müller cells in vitro to produce their own extracellular matrix when cultured in monolayer. Many Müller cell–enriched genes (Table S5) are part of the extracellular matrix and cytoskeleton categories. It is well recognized that neuronal development is governed by a diverse group of macromolecules that include cell adhesion and receptors as well as growth neurotrophic factors and extracellular matrix constituents, which act in a concerted fashion. These molecules are expressed at specific stages in the developing retina and are intimately involved in processes such as cell determination, differentiation, and migration of retinal neurons.93 In situ hybridization and cell culture studies have shown that Müller cells synthesize many of these developmentally important molecules, but very little is known about the specific roles of Müller cell–derived molecules in retinal development. Ultrastructural studies have shown that Müller cells contain many cytoskeletal components such as microtubules, microfilaments, and intermediate filaments, such as vimentin, which has also been shown by immunohistochemical analysis to be expressed by retinal Müller cells of all vertebrate species during development.94

Two specific markers of Müller cells, glutamine synthetase and cellular retinaldehyde binding protein (CRALBP), were not found in the NHMC library. The absence of glutamine synthetase could be explained by the growth culture protocol of Müller cells. It has been demonstrated that when Müller cells are grown in monolayer cultures, in absence of contact interaction with neurons, they fail to induce glutamine synthetase expression.95 In contrast, CRALBP expression is not affected by the absence of neurons.96 However, in a proteomic study in porcine Müller cells, Hauck et al.97 demonstrated that glutamine synthetase and CRALBP are downregulated among with other proteins when grown as primary cultures. Glutamine synthetase is expressed in HMCL libraries, but no CRALBP expression can be found. Moreover, a low expression level was detected in the most recent retinal library from Bowes Rickman et al.21 The lack of expression of those two known markers of Müller cells in our libraries could be due to the number of tags sequenced. Indeed, approximately 50,000 tags have been sequenced per library. While it is generally accepted that using 50,000 tags per sample studied is enough to obtain a complete cell expression profile,61 98 it has been demonstrated that sequencing more tags (100,000 or 150,000) reduces significantly the number of no-match tags and thus increases the probability of sequencing a gene present at a low expression level.99 100

Traditionally, diabetic retinopathy has been viewed as a disorder of the retinal vasculature. Recent evidence indicates that it also affects the glial and neural cells of the retina.101 102 A growing body of evidence has demonstrated a link between various disturbances in mitochondrial functioning and diabetes.103 A previous diabetic mouse cDNA microarray study104 has shown the upregulation of all mitochondrial DNA-encoded proteins and most of the nuclear DNA-encoded genes for oxidative phosphorylation in the diabetic retina, suggesting that this pathway is potentially activated in the diabetic retina. This notion is consistent with our data, in that we observed an upregulation of NADH dehydrogenase, cytochrome c oxidase, and ATP synthase in the HMCL libraries. Moreover, a subtractive library62 identified an increased expression of these proteins in diabetes and also a higher expression in type II than in type I diabetes, which is consistent with our observations (Table 6) .

It has been well demonstrated in the literature that the expression of extracellular matrix/cell adhesion proteins is modified in diabetes. A study performed to measure the synthesis of basement membrane components in skin samples of diabetic and nondiabetic patients105 demonstrated a significant reduction of mRNA expression of collagen and fibronectin isolated from the skin of diabetic patients, as observed in the present study for the HMCLs. Moreover, Feldmann et al.106 measured the IGFBP serum levels of these patients, with and without diabetic retinopathy. They demonstrated that the level of IGFBP was decreased in patients with type I or II diabetes, which is in agreement with our SAGE data. In addition, we observed an increase of bFGF in the HMCLs which is known to be elevated in diabetic retinopathy.107 108 109

Recently, Gerhardinger et al.35 determined the expression profile of diabetic rat Müller cells by using gene microarray (GeneChips; Affymetrix, Santa Clara, CA). However, this oligonucleotide array method is restricted to known sequences and limitations with quantification have been identified.110 When our SAGE library is compared with the results obtained with this oligonucleotide array method, significant discrepancies have been observed. However, their results are quite consistent with our SAGE library of HMCL-II. In particular, they found a downregulation of VEGF in diabetic rats compared with normal rats, which is consistent with our observation when comparing NHMC and HMCL SAGE libraries.

The SAGE library of human Müller cells reported in this article provides a gene expression profile of these cells that will be very useful in understanding the function of these main glial cells of the retina. Moreover, the HMCL SAGE libraries represent a valuable source of knowledge for understanding the role of Müller cells in the development of diabetic retinopathy.


   Acknowledgements
 
The authors thank Jean-François Cadrin Girard, Pascal Belleau, and Jonny St-Amand (Laboratoire de Recherche en Endocrinologie Moléculaire et Oncologie au Centre de Recherche du CHUL) and Jennifer Gagné (Unité de Recherche en Ophthalmologie) for skillful technical assistance and the Banque d’Yeux Nationale, Inc. for providing retinal tissues.


   Footnotes
 
Supported by the Natural Sciences and Engineering Research Council of Canada. CBL was the recipient of a studentship from the Canadian Institutes of Health Research (CIHR) and the Vision Research Network from the Fonds de la Recherche en santé du Québec (FRSQ). CB and SL hold a studentship from the FRSQ and CIHR, respectively. CS is a Chercheur boursier national from the FRSQ.

Submitted for publication January 31, 2007; revised July 13, 2007; accepted August 27, 2007.

Disclosure: C.B. Lupien, None; C. Bolduc, None; S. Landreville, None; C. Salesse, None

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.

Corresponding author: Christian Salesse, Unité de Recherche en Ophtalmologie, Salle S-5, Centre de Recherche du CHUQ, Pavillon CHUL, 2705 Boul. Laurier, Ste-Foy, Québec, G1V 4G2, Canada; christian.salesse{at}crchul.ulaval.ca.


   References
Top
 Abstract
 Methods
 Results
 Discussion
 References
 

  1. Newman E, Reichenbach A. The Müller cell: a functional element of the retina. Trends Neurosci. 1996;19:307–312.[CrossRef][ISI][Medline][Order article via Infotrieve]
  2. Hollander H, Makarov F, Dreher Z, et al. Structure of the macroglia of the retina: sharing and division of labour between astrocytes and Müller cells. J Comp Neurol. 1991;313:587–603.[CrossRef][ISI][Medline][Order article via Infotrieve]
  3. Pfeiffer B, Grosche J, Reichenbach A, Hamprecht B. Immunocytochemical demonstration of glycogen phosphorylase in Müller (glial) cells of the mammalian retina. Glia. 1994;12:62–67.[CrossRef][Medline][Order article via Infotrieve]
  4. Poitry-Yamate C, Tsacopoulos M. Glial (Müller) cells take up and phosphorylate [3H]2-deoxy-D-glucose in mammalian retina. Neurosci Lett. 1991;122:241–244.[CrossRef][ISI][Medline][Order article via Infotrieve]
  5. Poitry-Yamate CL, Poitry S, Tsacopoulos M. Lactate released by Müller glial cells is metabolized by photoreceptors from mammalian retina. J Neurosci. 1995;15:5179–5191.[Abstract]
  6. Lewis GP, Erickson PA, Guerin CJ, Anderson DH, Fisher SK. Changes in the expression of specific Müller cell proteins during long-term retinal detachment. Exp Eye Res. 1989;49:93–111.[CrossRef][ISI][Medline][Order article via Infotrieve]
  7. Lewis GP, Guerin CJ, Anderson DH, Matsumoto B, Fisher SK. Rapid changes in the expression of glial cell proteins caused by experimental retinal detachment. Am J Ophthalmol. 1994;118:368–376.[ISI][Medline][Order article via Infotrieve]
  8. Bringmann A, Pannicke T, Grosche J, et al. Müller cells in the healthy and diseased retina. Prog Retin Eye Res. 2006;25:397–424.[CrossRef][ISI][Medline][Order article via Infotrieve]
  9. Gardner TW, Antonetti DA, Barber AJ, LaNoue KF, Levison SW. Diabetic retinopathy: more than meets the eye. Surv Ophthalmol. 2002;47(suppl 2)S253–S262.[CrossRef][ISI][Medline][Order article via Infotrieve]
  10. Smith SB. Diabetic retinopathy and the NMDA receptor. Drug News Perspect. 2002;15:226–232.[CrossRef][ISI][Medline][Order article via Infotrieve]
  11. Velculescu VE, Zhang L, Vogelstein B, Kinzler KW. Serial analysis of gene expression. Science. 1995;270:484–487.[Abstract/Free Full Text]
  12. Bartlett J. Technology evaluation: SAGE, genzyme molecular oncology. Curr Opin Mol Ther. 2001;3:85–96.[Medline][Order article via Infotrieve]
  13. Yamamoto M, Wakatsuki T, Hada A, Ryo A. Use of serial analysis of gene expression (SAGE) technology. J Immunol Methods. 2001;250:45–66.[CrossRef][ISI][Medline][Order article via Infotrieve]
  14. Velculescu VE, Madden SL, Zhang L, et al. Analysis of human transcriptomes. Nat Genet. 1999;23:387–388.[ISI][Medline][Order article via Infotrieve]
  15. Boheler KR, Stern MD. The new role of SAGE in gene discovery. Trends Biotechnol. 2003;21:55–57.[CrossRef][ISI][Medline][Order article via Infotrieve]
  16. Tuteja R, Tuteja N. Serial analysis of gene expression: applications in human studies. J Biomed Biotech. 2004;2:113–120.
  17. Blackshaw S, Fraioli RE, Furukawa T, Cepko CL. Comprehensive analysis of photoreceptor gene expression and the identification of candidate retinal disease genes. Cell. 2001;107:579–589.[CrossRef][ISI][Medline][Order article via Infotrieve]
  18. Blackshaw S, Harpavat S, Trimarchi J, et al. Genomic analysis of mouse retinal development. PLoS Biol. 2004;2:E247.[CrossRef][Medline][Order article via Infotrieve]
  19. Sharon D, Blackshaw S, Cepko CL, Dryja TP. Profile of the genes expressed in the human peripheral retina, macula, and retinal pigment epithelium determined through serial analysis of gene expression (SAGE). Proc Natl Acad Sci USA. 2002;99:315–320.[Abstract/Free Full Text]
  20. Kobashi-Hashida M, Ohguro N, Tsujikawa M, et al. Micro serial analysis of gene expression in normal human choroid and retinal pigment epithelial transcriptomes. Jpn J Ophthalmol. 2005;49:15–22.[CrossRef][Medline][Order article via Infotrieve]
  21. Bowes Rickman C, Ebright JN, Zavodni ZJ, et al. Defining the human macula transcriptome and candidate retinal disease genes using EyeSAGE. Invest Ophthalmol Vis Sci. 2006;47:2305–2316.[Abstract/Free Full Text]
  22. Lupien CB, Salesse C. Characterization of two spontaneously generated human Müller cell lines from donors with type 1 and type 2 diabetes. Invest Ophthalmol Vis Sci. 2007;48:874–880.[Abstract/Free Full Text]
  23. Lupien C, Brenner M, Guerin SL, Salesse C. Expression of glial fibrillary acidic protein in primary cultures of human Müller cells. Exp Eye Res. 2004;79:423–429.[CrossRef][ISI][Medline][Order article via Infotrieve]
  24. Proulx S, Guerin SL, Salesse C. Effect of quiescence on integrin alpha5beta1 expression in human retinal pigment epithelium. Mol Vis. 2003;9:473–481.[ISI][Medline][Order article via Infotrieve]
  25. Velculescu VE, Zhang L, Zhou W, et al. Characterization of the yeast transcriptome. Cell. 1997;88:243–251.[CrossRef][ISI][Medline][Order article via Infotrieve]
  26. Kenzelmann M, Muhlemann K. Substantially enhanced cloning efficiency of SAGE (serial analysis of gene expression) by adding a heating step to the original protocol. Nucleic Acids Res. 1999;27:917–918.[Abstract/Free Full Text]
  27. St-Amand J, Okamura K, Matsumoto K, Shimizu S, Sogawa Y. Characterization of control and immobilized skeletal muscle: an overview from genetic engineering. FASEB J. 2001;15:684–692.[Abstract/Free Full Text]
  28. Adams MD, Kerlavage AR, Fleischmann RD, et al. Initial assessment of human gene diversity and expression patterns based upon 83 million nucleotides of cDNA sequence. Nature. 1995;377:3–174.[Medline][Order article via Infotrieve]
  29. Lash AE, Tolstoshev CM, Wagner L, et al. SAGEmap: a public gene expression resource. Genome Res. 2000;10:1051–1060.[Abstract/Free Full Text]
  30. Distler C, Dreher Z. Glia cells of the monkey retina–II. Müller cells. Vision Res. 1996;36:2381–2394.[CrossRef][ISI][Medline][Order article via Infotrieve]
  31. Schnitzer J. Immunocytochemical studies on the development of astrocytes, Müller (glial) cells, and oligodendrocytes in the rabbit retina. Brain Res Dev Brain Res. 1988;44:59–72.[CrossRef][Medline][Order article via Infotrieve]
  32. Guerin CJ, Wolfshagen RW, Eifrig DE, Anderson DH. Immunocytochemical identification of Müller’s glia as a component of human epiretinal membranes. Invest Ophthalmol Vis Sci. 1990;31:1483–1491.[Abstract]
  33. Lewis GP, Erickson PA, Guerin CJ, Anderson DH, Fisher SK. Basic fibroblast growth factor: a potential regulator of proliferation and intermediate filament expression in the retina. J Neurosci. 1992;12:3968–3978.[Abstract]
  34. Vaughan DK, Erickson PA, Fisher SK. Glial fibrillary acidic protein (GFAP) immunoreactivity in rabbit retina: effect of fixation. Exp Eye Res. 1990;50:385–392.[CrossRef][ISI][Medline][Order article via Infotrieve]
  35. Gerhardinger C, Costa MB, Coulombe MC, et al. Expression of acute-phase response proteins in retinal Müller cells in diabetes. Invest Ophthalmol Vis Sci. 2005;46:349–357.[Abstract/Free Full Text]
  36. Chaitin MH, Wortham HS, Brun-Zinkernagel AM. Immunocytochemical localization of CD44 in the mouse retina. Exp Eye Res. 1994;58:359–365.[CrossRef][ISI][Medline][Order article via Infotrieve]
  37. Chaitin MH, Brun-Zinkernagel AM. Immunolocalization of CD44 in the dystrophic rat retina. Exp Eye Res. 1998;67:283–292.[CrossRef][ISI][Medline][Order article via Infotrieve]
  38. Matsunaga M, Hatta K, Takeichi M. Role of N-cadherin cell adhesion molecules in the histogenesis of neural retina. Neuron. 1988;1:289–295.[CrossRef][ISI][Medline][Order article via Infotrieve]
  39. Morimoto A, Matsuda S, Uryu K, et al. Light- and electron-microscopic localization of basic fibroblast growth factor in adult rat retina (in Japanese). Okajimas Folia Anat Jpn. 1993;70:7–12.[Medline][Order article via Infotrieve]
  40. Neophytou C, Vernallis AB, Smith A, Raff MC. Müller-cell-derived leukaemia inhibitory factor arrests rod photoreceptor differentiation at a postmitotic pre-rod stage of development. Development. 1997;124:2345–2354.[Abstract]
  41. Anderson DH, Guerin CJ, Hageman GS, Pfeffer BA, Flanders KC. Distribution of transforming growth factor-beta isoforms in the mammalian retina. J Neurosci Res. 1995;42:63–79.[CrossRef][ISI][Medline][Order article via Infotrieve]
  42. Lee W-H, Javedan S, Bondy CA. Coordinate expression of insulin-like growth factor system components by neuron and neuroglia during retinal and cerebellar development. J Neurosci. 1992;12:4737–4744.[Abstract]
  43. Mudhar HS, Pollock RA, Wang C, Stiles CD, Richardson WD. PDGF and its receptors in the developing rodent retina and optic nerve. Development. 1993;118:539–552.[Abstract]
  44. Roque RS, Caldwell RB, Behzadian MA. Cultured Müller cells have high levels of epidermal growth factor receptors. Invest Ophthalmol Vis Sci. 1992;33:2587–2595.[Abstract/Free Full Text]
  45. Lillien L. Changes in retinal cell fate induced by overexpression of EGF receptor. Nature. 1995;377:158–162.[CrossRef][Medline][Order article via Infotrieve]
  46. Gerhardinger C, Brown LF, Roy S, et al. Expression of vascular endothelial growth factor in the human retina and in nonproliferative diabetic retinopathy. Am J Pathol. 1998;152:1453–1462.[Abstract]
  47. Kroger S, Horton SE, Honig LS. The developing avian retina expresses agrin isoforms during synaptogenesis. J Neurobiol. 1996;29:165–182.[CrossRef][ISI][Medline][Order article via Infotrieve]
  48. Burke JM, Kower HS. Collagen synthesis by rabbit neural retina in vitro and in vivo. Exp Eye Res. 1980;31:213–226.[CrossRef][Medline][Order article via Infotrieve]
  49. von der Mark K, von der Mark H, Timpl R, Trelstad RL. Immunofluorescent localization of collagen types I, II, and III in the embryonic chick eye. Dev Biol. 1977;59:75–85.[CrossRef][Medline][Order article via Infotrieve]
  50. Sarthy V. Collagen IV mRNA expression during development of the mouse retina: an in situ hybridization study. Invest Ophthalmol Vis Sci. 1993;34:145–152.[Abstract/Free Full Text]
  51. Kohno T, Sorgente N, Ishibashi T, Goodnight R, Ryan SJ. Immunofluorescent studies of fibronectin and laminin in the human eye. Invest Ophthalmol Vis Sci. 1987;28:506–514.[Abstract/Free Full Text]
  52. Perez RG, Halfter W. Tenascin in the developing chick visual system: distribution and potential role as a modulator of retinal axon growth. Dev Biol. 1993;156:278–292.[CrossRef][ISI][Medline][Order article via Infotrieve]
  53. Bartsch S, Bartsch U, Dorries U, et al. Expression of tenascin in the developing and adult cerebellar cortex. J Neurosci. 1992;12:736–749.[Abstract]
  54. Neugebauer KM, Emmett CJ, Venstrom KA, Reichardt LF. Vitronectin and thrombospondin promote retinal neurite outgrowth: developmental regulation and role of integrins. Neuron. 1991;6:345–358.[CrossRef][ISI][Medline][Order article via Infotrieve]
  55. McAdams BD, McLoon SC. Expression of chondroitin sulfate and keratan sulfate proteoglycans in the path of growing retinal axons in the developing chick. J Comp Neurol. 1995;352:594–606.[CrossRef][ISI][Medline][Order article via Infotrieve]
  56. Chai L, Morris JE. Distribution of heparan sulfate proteoglycans in embryonic chicken neural retina and isolated inner limiting membrane. Curr Eye Res. 1994;13:669–677.[ISI][Medline][Order article via Infotrieve]
  57. Inatani M, Tanihara H, Honjo M, et al. Expression of proteoglycan decorin in neural retina. Invest Ophthalmol Vis Sci. 1999;40:1783–1791.[Abstract/Free Full Text]
  58. Sarthy V, Ripps H. The Retinal Müller Cell Structure and Function. 2001;278. Kluwer Academic New York.
  59. Winkler BS, Arnold MJ, Brassell MA, Puro DG. Energy metabolism in human retinal Müller cells. Invest Ophthalmol Vis Sci. 2000;41:3183–3190.[Abstract/Free Full Text]
  60. Kusner LL, Sarthy VP, Mohr S. Nuclear translocation of glyceraldehyde-3-phosphate dehydrogenase: a role in high glucose-induced apoptosis in retinal Müller cells. Invest Ophthalmol Vis Sci. 2004;45:1553–1561.[Abstract/Free Full Text]
  61. Dinel S, Bolduc C, Belleau P, et al. Reproducibility, bioinformatic analysis and power of the SAGE method to evaluate changes in transcriptome. Nucleic Acids Res. 2005;33:e26.[Abstract/Free Full Text]
  62. Antonetti DA, Reynet C, Kahn CR. Increased expression of mitochondrial-encoded genes in skeletal muscle of humans with diabetes mellitus. J Clin Invest. 1995;95:1383–1388.[ISI][Medline][Order article via Infotrieve]
  63. Lees-Miller JP, Heeley DH, Smillie LB. An abundant and novel protein of 22 kDa (SM22) is widely distributed in smooth muscles: purification from bovine aorta. Biochem J. 1987;244:705–709.[ISI][Medline][Order article via Infotrieve]
  64. Gunnersen JM, Spirkoska V, Smith PE, Danks RA, Tan SS. Growth and migration markers of rat C6 glioma cells identified by serial analysis of gene expression. Glia. 2000;32:146–154.[CrossRef][ISI][Medline][Order article via Infotrieve]
  65. Bernstein HG, Reichenbach A, Kirschke H, Wiederanders B. Cell type-specific distribution of cathepsin B and D immunoreactivity within the rabbit retina. Neurosci Lett. 1989;98:135–138.[CrossRef][ISI][Medline][Order article via Infotrieve]
  66. Yamada T, Hara S, Tamai M. Immunohistochemical localization of cathepsin D in ocular tissues. Invest Ophthalmol Vis Sci. 1990;31:1217–1223.[Abstract/Free Full Text]
  67. el-Hifnawi E. Localization of cathepsin D in rat ocular tissues. An immunohistochemical study. Ann Anat. 1995;177:11–17.[ISI][Medline][Order article via Infotrieve]
  68. Hartig W, Grosche J, Distler C, et al. Alterations of Müller (glial) cells in dystrophic retinae of RCS rats. J Neurocytol. 1995;24:507–517.[CrossRef][ISI][Medline][Order article via Infotrieve]
  69. Caley DW, Johnson C, Liebelt RA. The postnatal development of the retina in the normal and rodless CBA mouse: a light and electron microscopic study. Am J Anat. 1972;133:179–212.[CrossRef][ISI][Medline][Order article via Infotrieve]
  70. Stolzenburg JU, Haas J, Hartig W, et al. Phagocytosis of latex beads by rabbit retinal Müller (glial) cells in vitro. J Hirnforsch. 1992;33:557–564.[Medline][Order article via Infotrieve]
  71. Nerurkar MA, Satav JG, Katyare SS. Insulin-dependent changes in lysosomal cathepsin D activity in rat liver, kidney, brain and heart. Diabetologia. 1988;31:119–122.[Medline][Order article via Infotrieve]
  72. Caldwell RB, Bartoli M, Behzadian MA, et al. Vascular endothelial g