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


     

(Investigative Ophthalmology and Visual Science. 2003;44:2466-2476.)
© 2003 by The Association for Research in Vision and Ophthalmology, Inc.
DOI:  10.1167/iovs.02-0671
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
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 PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Nielsen, K.
Right arrow Articles by Orntoft, T. F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Nielsen, K.
Right arrow Articles by Orntoft, T. F.

Identification of Differentially Expressed Genes in Keratoconus Epithelium Analyzed on Microarrays

Kim Nielsen,1,2 Karin Birkenkamp-Demtröder,2 Niels Ehlers,1 and Torben Falck Orntoft2

1From the Department of Ophthalmology and the 2Molecular Diagnostic Laboratory, Aarhus University Hospital, Aarhus, Denmark.


   Abstract
Top
 Abstract
Materials and Methods
 Results
 Discussion
 References
 
PURPOSE. The present study was conducted to investigate differential gene expression in individual samples derived from fresh-frozen human keratoconus and normal corneal epithelium, using gene microarrays.

METHODS. Total RNA was extracted (11 keratoconus and 8 normal samples), labeled, and hybridized to microarrays (GeneChip; Affymetrix, Inc., Santa Clara, CA). GeneChip data were validated by verifying the expression profiles of 10 genes by real-time PCR and by recalculation using dChip software (Wong Laboratory, Department of Biostatistics, Harvard School of Public Health, Boston, MA). Furthermore, 3 of the 10 encoded proteins were stained by immunohistochemistry.

RESULTS. In comparison with normal cornea, the expression of 471 of the 5600 genes on the microarrays was changed in the keratoconus samples. This number was reduced to 47 with increased expression and 9 with decreased expression when more stringent selection parameters were applied. These genes are believed to be involved in keratoconus. Two of the candidate genes, lysyl oxidase and tissue inhibitor of metalloproteinase 3, are known to be involved in other eye diseases. Expression profiles were reproduced with the software dChip (Wong Laboratory) and real-time PCR. Increases in keratin 6 and 13 were also detected at the protein level.

CONCLUSIONS. Keratoconus epithelium appears to be characterized by massive changes of the cytoskeleton, reduced extracellular matrix remodeling, altered transmembrane signaling, and modified cell-to-cell and cell-to-matrix interactions. Validation of gene expression with dChip analysis and real-time PCR indicates GeneChip to be a valid technique for investigation of epithelium from single dissected corneal samples. Association between alterations at RNA and protein levels was observed for some of the tested candidates.

Keratoconus (KC; On-line Mendelian Inheritance in Man [OMIM] 148300) is a noninflammatory bradytrophic human corneal disease. The bilateral distortion induces myopia and astigmatism. At first, glasses and contact lenses are used to compensate for the optical defects, but later a corneal transplantation may become necessary. KC is one of the most common causes of penetrating keratoplasty in the Western world. No medical treatment is known today; however, if single factors could be identified, they might serve as drug targets in a medical treatment—for example, in the form of eye drops that would be cost-effective compared with transplantation.

KC is believed to be autosomal inherited, because both eyes are affected, it occurs more often in certain families,1 2 3 4 and it develops in homozygous twins.5 6 7 8 However, contradictory information exists in reports of unilateral KC9 10 11 and cases of recurrent KC after transplantation,12 13 14 but asymmetrical progression and transplantation with KC-affected donor corneas cannot be excluded. Diagnosis is usually based on Placido disk analysis, slit lamp biomicroscopy, and computer-assisted topography, which are often difficult to interpret, especially in young patients with mild symptoms. If genomic mutations can be identified,15 16 they may serve as valuable markers in clinical diagnostics.

KC is characterized by a reduction in corneal thickness.17 18 The epithelium degenerates, loosens its smoothness, and becomes irregular.19 Stromal erosion and breaking of the Bowman’s membrane are often observed. The cause of KC is unknown; it may originate in any of the three corneal cell layers. In many investigations, corneas have been treated as a homogeneous tissue, whereas we dissected the cornea and focused on the epithelium. If one or more genes are differentially expressed between KC and normal corneal epithelium, early detection of KC could be possible by simple gene expression analysis of scraped-off epithelium.

High-density oligonucleotide microarrays (GeneChip; Affymetrix, Inc., Santa, Clara, CA), which provide the ability to measure gene expression of approximately 5600 known genes simultaneously, were used to study gene expression in individual patients. By comparing the expression in KC samples to normal samples derived from patients with myopia, we determined those genes that were differentially expressed and thus isolated 56 candidate genes that may be related to KC.

Microarray analysis has a great potential but should be independently validated. In this case, expression of 10 candidate genes was verified with real-time PCR and, furthermore, data were recalculated with the alternative software dChip (Wong Laboratory, Department of Biostatistics, Harvard School of Public Health, Boston, MA). Both experiments demonstrated comparable expression patterns as determined on the GeneChips.

The process from RNA to protein is regulated in numerous ways. We wanted to determine whether the observed changes at RNA level could also be observed at protein level, using immunohistochemistry. Three of the candidate proteins were antibody-stained and for two of these, an association was observed.


   Materials and Methods
Top
 Abstract
 Materials and Methods
Results
 Discussion
 References
 
Tissue
Fresh human corneal epithelial cells were obtained from 10 patients with KC having a corneal transplantation and from 25 myopes who underwent photorefractive keratectomy. One patient with KC had surgery on both eyes and contributed two samples: Ks3 and Ks122. Patients were asked not to wear contact lenses for at least 2 days before surgery. Tissue from the myopic eyes was considered to be normal. Ten minutes before surgery, the eye was anesthetized with 0.4% oxybuprocaine (SAD, Copenhagen, Denmark). The central 8 mm of the cornea was marked, and the epithelium was scraped off with a scalpel after a brief period of coagulation with absolute ethanol followed by flushing with saline. The procedure took place immediately before the primary surgery was initiated. Cells were quickly dissolved in 4 M guanidine thiocyanate (25 mM citrate, 0.5% lauroylsarcosine, and 100 mM mercaptoethanol) and immediately frozen in liquid N2. Samples were stored at -80°C. Permission to collect human tissue was given from The Scientific Ethics Committee of Aarhus County, and informed consent was obtained from all patients in accordance with the Declaration of Helsinki.

Isolation of Total RNA
Epithelial cells were lysed in a 100-µL glass homogenizer (Jencons, Leighton, UK). Total RNA was isolated with extraction reagent (RNAzol B; Wak Chemie Medical, Bad Soden/Ts, Germany) as described by the manufacturer. The amount of RNA was quantified photospectrometrically and checked for degradation by agarose electrophoresis. Samples labeled Ks and Ns are single-patient samples whereas Np18 is a pool of equal amounts of total RNA from 18 patients with myopia.

cRNA Synthesis
Reverse transcription was performed on 9 µg total RNA for 1 hour at 42°C using a T7-oligo(dT24)-primer and reverse transcriptase (SuperScript II; Life Technologies-Invitrogen, Carlsbad, CA). Second-strand cDNA synthesis was performed for 2 hours at 16°C using Escherichia coli DNA polymerase I, DNA ligase, and RNase H (Life Technologies) followed by incubation in 50 mM NaOH and 0.1 mM EDTA for 10 minutes at 65°C leading to degradation of the RNA. After phenol-chloroform extraction and ethanol precipitation, in vitro transcription was performed for 6 hours at 37°C using biotin-16-UTP and biotin-11-CTP with an RNA transcript labeling kit (BioArray; Enzo Diagnostics, Farmingdale, NY). cRNA was purified on spin columns (RNeasy; Qiagen, Valencia, CA), followed by fragmentation for 30 minutes at 95°C. Finally, spikes necessary for image adjustment after scanning were added.

GeneChip Performance
Before analysis of 19 samples on Hu6800FL GeneChips (Affymetrix, Inc.), the quality of each sample was checked on Test3 arrays (Affymetrix). GeneChips were hybridized with 15 µg biotin-labeled cRNA at 45°C in a hybridization oven (model 640; Affymetrix) overnight, followed by a first staining with streptavidin-phycoerythrin in a fluidics station (model 400; Affymetrix) and a second staining with biotinylated anti-streptavidin and streptavidin-phycoerythrin. Finally, the GeneChips were scanned (GeneArray Scanner; Hewlett Packard, Palo Alto, CA).

Data Mining
Data were extracted from antibody-stained microarray images and analyzed using the Microarray Suite 5.0 (MAS), Data Mining Tool 3.0, and MicroDB software (Affymetrix). GeneChips were globally scaled to the arbitrary value of 150.20 Probe sets for spikes were removed before further analysis of the remaining 7070 probe sets.

Two different approaches were used for data analysis. One strategy was statistical and based on the nonparametric Mann-Whitney test of the arbitrary gene expression values (termed Signal by Affymetrix) in KC (n = 11) and normal (n = 8) epithelium. The other strategy was a comparison analysis of KC samples versus normal samples, based on MAS software interpretations of Detection (transcript scored as Present or Absent), Change (scored as Increased, Decreased, or Not Changed) and Signal Ratio. We emphasize that the Signal Ratio is not the simple ratio between two gene expression levels but is a complex analysis of individual probe cells.

Two-way (gene and sample) hierarchical cluster analysis (GeneCluster 2.11, Michael Eisen, Lawrence Berkeley National Laboratory, University of California at Berkeley)21 was performed on log-transformed, median-centered, and normalized (sum of squares similar to 1.0 in both directions) Signals on the 19 samples. Results were visualized with TreeView 1.5 (http://taxonomy.zoology.gla.ac.uk; developed by Roderic D. M. Page and provided in the public domain by the Institute of Biomedical and Life Sciences, University of Glasgow, Glasgow, Scotland, UK).22

Gene expression in the 19 samples was recalculated using the software dChip 1.2 (Wong Laboratory, Dept. of Biostatistics, Harvard School of Public Health),23 24 which is based on alternative algorithms to MAS.

Real-Time PCR
One microgram total RNA was reverse transcribed into 20 µL cDNA for those samples with excessive RNA, using 50 pmol oligo dT21-primer, 1 mM dNTP mix and 200 U reverse transcriptase (SuperScript II; Life Technologies-Invitrogen). Samples tested negative for DNA contamination, by amplification of microsatellite marker D2S119.

RNA expression was determined with a sequence detection system (Prism 7000; Applied Biosystems, Foster City, CA). Gene-specific primers (Table 1) for 11 genes were designed to be located in the same region as the Affymetrix probe sets, and, if possible, to be intron spanning. Real-time PCR was performed in a volume of 25 µL containing 2 µL cDNA (x20), 7.5 pmol of each primer, and 12.5 µL master mix (SYBR Green PCR Master Mix; Applied Biosystems). The program was 50°C for 2 minutes and 95°C for 10 minutes, with amplification in 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. Fluorescence was monitored after each elongation period. The program terminated with a melting curve analysis by elevating the temperature from 60°C to 95°C while monitoring the fluorescence. A standard dilution (x1, x10, x100, x1,000, and x10,000) composed of a large reference pool of cDNA, was amplified for each run. Samples were quantified compared with the standard curve created from those dilutions.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Primer Sets Used for Real-Time PCR, Designed According to the Manufacturer’s Description

 
The housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was confirmed to be constantly expressed in corneal epithelium from the GeneChip analysis. Samples were amplified with GAPDH primers for determination of the relative starting amount of cDNA in each sample, and all genes were normalized to that amount. Negative controls without template were produced for each run. Tissue samples were amplified in triplicate and from these replicates, averages were calculated.

Immunohistochemistry
Eight archival corneas were fixed in 4% formalin and embedded in paraffin. Four samples were KC corneas, independent of the samples used for GeneChips and real-time PCR, and four were from eyes with ocular tumors (three with retinoblastoma and one with orbital tumor). In this study, the tumor samples were considered to be normal and served as a reference. Tissue sections (5 µm) were transferred to electrostatic slides (Menzel-Gläser, Braunschweig, Germany). Sections were deparaffinized in mineral oil for 10 minutes at 60°C and washed in ethanol. Antigens were retrieved by boiling in a 900-W microwave for 12 minutes in buffer (10 mM tris(hydroxymethyl)aminomethane, 0.5 mM EGTA [pH 9.0]). Endogenous peroxidase activity was quenched in 1.4% hydrogen peroxide for 20 minutes. Sections were blocked in 1:25 horse serum (Sigma, St. Louis, MO) for 20 minutes and incubated with primary antibody at 4°C overnight. Antibodies were cytokeratin 6 (K6, 1:2; Research Diagnostics, Flanders, NJ), cytokeratin 10/13 (K13, 1:100; Dako, Glostrup, Denmark), and vimentin (1:120; Santa Cruz Biotechnology, Santa Cruz, CA). They were rinsed in phosphate-buffered saline (PBS) buffer before the corresponding secondary antibody (biotinylated anti-mouse and anti-rabbit [both from Oncogene Research Products, San Diego, CA], or anti-goat [Santa Cruz Biotechnology]) was applied for 30 minutes. The sections were then rinsed in PBS, incubated in an avidin DH-biotinylated horseradish peroxidase H complex solution (Oncogene Research Products) for 30 minutes, and stained in 3-amino-9-ethyl-carbazole (Sigma) for 10 minutes. Finally, the slides were counterstained in hematoxylin and mounted. Control staining was produced by omitting the primary antibody. All antibodies were diluted in PBS containing 1% bovine serum albumin (Calbiochem, San Diego, CA).


   Results
Top
 Abstract
 Materials and Methods
 Results
Discussion
 References
 
Selection of Candidate Genes
Total RNA was extracted from human corneal epithelium. To ensure comparability of the arrays, we used pair-wise scatterplots. Figure 1 shows three scatterplots representative of the 171 possible plots. Most of the expression pairs were within a narrow range. The correlation coefficients were typically between 0.88 and 0.96 and for the two corneas from the same patient, it was 0.9485. The uniform array performance illustrated by the scaling factors and Present probe sets (Table 2) and by the scatterplots allowed us to proceed with comparative analyses, based on two different methodological approaches.



View larger version (22K):
[in this window]
[in a new window]
  		FIGURE 1. Three representative scatterplots from six randomly chosen microarrays. The correlation coefficient (R2) is shown in each plot.

 

View this table:
[in this window]
[in a new window]
  		TABLE 2. Information on Participants

 
The first selection method was based on MAS software (Affymetrix, Inc.) interpretations. KC and normal samples were compared each to each, resulting in 88 comparisons. When the criteria of Increase in expression in at least 50% of the comparisons was used, 78 increased probe sets were selected and, analogous to this, 23 decreased probe sets. When the parameter Signal Ratio was used instead, 41 probe sets showed an more than twofold increase and 18 probe sets a less than twofold decrease. The second method was based on selection of altering genes, using the nonparametric Mann-Whitney test on the Signal Ratios, and 414 probe sets were found to change significantly (P < 0.05).

A total of 471 altered probe sets were selected. The MAS software analysis was the most conservative, selecting only 101 and 59 probe sets, respectively, compared with 414 selected by the statistical method. Probe sets that were selected by one of the two software-mediated tests and that were also significantly changed according to the nonparametric test were extracted. Those noninformative probe sets that were either increasing to Absent or decreasing from Absent were eliminated, which reduced the number of probe sets from 88 to 62. Some of the genes were detected with more than one probe set, indicating a good reproducibility of the method, and because of this, the final set of candidate genes consisted of 56 different genes.

When we looked at all genes, the two KC samples from the same individual differed as much as two KC samples from different individuals. The two samples from the same individual were further studied. Figure 2 shows the average Signal Ratio between Ks122 and the eight normal samples plotted against similar values for Ks3. Despite the global dissimilarity between the two samples, a strong relation was observable when focusing on the candidate genes. Comparing random samples from different individuals showed much less relation (data not shown).



View larger version (6K):
[in this window]
[in a new window]
  		FIGURE 2. Scatterplot from one individual showing Signal Ratios from sample Ks122 (left eye) plotted against Ks3 (right eye) for the 62 candidate probe sets. The Signal Ratios of both Ks3 and Ks122 were calculated using the average of the eight normal samples as a reference.

 
Candidate Genes
Various functional groups were identified among the candidate genes (Table 3) of which the cytoskeleton, the extracellular matrix (ECM), transmembrane signaling, and cell-to-cell–cell-to-matrix interaction–related genes are some of the most well represented. Intermediate filaments such as keratins, are cytoskeletal building blocks. K6 is highly represented among the candidate genes, with four probe sets increasing 5.2-fold, on average. This gene shows the largest increase in gene expression. K13 and type III vimentin, which aggregate into homopolymers, were also increased, and as were the two small proline-rich proteins SPR2-1 and SPRR2B. They are small cross-linked molecules attached to the cytoplasmic side of the plasma membrane in keratinocytes.25 The cytoskeleton is, in addition to intermediate filaments, formed of microfilaments and microtubules as well. Gene expression of two major components of the microfilaments, {alpha}1-actinin and {alpha}1-tropomyocin increased as did the tropomyosin-binding protein troponin T1. Microtubules are involved in positioning chromosomes to the equatorial plan before cytokinesis is initiated. One element of the microtubules, {alpha}3-tubulin, showed increased expression.


View this table:
[in this window]
[in a new window]
  		TABLE 3. Summary of the 56 Candidate Genes

 
We observed upregulation of different signal proteins in the data set (Table 3) . The protein inhibitor of neuronal nitric oxide synthase, also known as the light chain of the 1.2 MDa dynein complex, is thought to control multiple events through the control of the neurotransmitter nitric oxide.27 Other selected genes involved in signaling are the sortilin-related receptor, and two G protein-coupled receptors GPR37 and RDC1. RDC1 acts as a coreceptor for immunodeficiency viruses.28

Candidate genes encoding proteins related to the ECM included several protease inhibitors—secretory leukocyte protease inhibitor, two serine (or cysteine) proteinase inhibitors, tissue inhibitor of metalloproteinase 3 (TIMP3), and cystatin E/M—all of which were upregulated. No changes in expression of matrix proteins, such as collagen, elastin, and proteoglycan were observed. Only the lysyl oxidase, involved in cross-linking of collagens was increased. Four of the five inhibitors are characterized as being secreted from the cell,29 30 31 with TIMP3 being bound to the ECM.32 Cystatin M/E, an inhibitor of cysteine proteinases, is found both intracellularly and extracellularly.33 The 200-kDa cathepsin C was the only proteinase among the 56 candidate genes. It is an intracellular enzyme involved in the activation of lysosomal serine proteinases34 and thus is not in direct contact with the ECM in living cells.

Genes encoding proteins involved in cell–cell interactions and cell–matrix adhesion also account for many of the probe sets showing increasing expression. These include the bullous pemphigoid antigen 1, carcinoembryonic antigen-related cell adhesion molecule 6, desmoglein 3, {gamma}2-laminin, and lectin galactoside-binding soluble 7 protein. Desmogleins are components of the desmosomes that help the cells to adhere to each other. Intracellularly, they are linked to keratins. Apart from these sets showing increased expression, mesothelin was the only gene with decreased expression. Some of the candidate genes have multiple functions, and others are less characterized.

Cluster Analysis
The 471 candidate genes were analyzed with a two-way hierarchical cluster analysis (Fig. 3A) . The 19 samples formed two groups on the horizontal axis, clearly separating all 11 KC samples from the eight normal samples (Fig. 3B) . Sample Ks3 and Ks122, which originated from the same patient, cluster close together but not on the same branch. In the gene cluster (vertical axis), only the keratins form a distinct biological group (Fig. 3C) . Figure 3D shows an example of a downregulated cluster involving several lipid-associated genes.



View larger version (74K):
[in this window]
[in a new window]
  		FIGURE 3. Hierarchical cluster analysis based on 19 samples and 471 candidate probe sets. (A) Gene and microarray cluster, (B) dendrogram of samples, (C) keratin-cluster, and (D) downregulated cluster.

 
Validation of Data with Real-Time PCR
Gene expression was validated with real-time PCR. Often the low yield of epithelial RNA is adequate only for analysis on a single GeneChip, but excessive material (>9 µg) was available from 13 samples. Expression of 10 candidate genes, showing some variation (low/high and increased/decreased expression), was tested on these samples. Figure 4 shows the expression profiles determined by the two techniques. The curve for real-time data is very similar to the GeneChip data, indicating a good reproducibility of the GeneChips, even for genes with low expression, such as cyclin D2 and 5'-nucleotidase. A striking finding was the rather heterogeneous expression level of some of the genes among the KC samples. When single genes were the focus, they appeared to be heterogeneously expressed, but when the total gene expression intra- and intergroup was examined with scatterplots, no difference in expression were observable between the group of KC and normal (data not shown).



View larger version (45K):
[in this window]
[in a new window]
  		FIGURE 4. Gene expression profiles from MAS and dChip compared with real-time PCR in arbitrary values. dChip values were reduced 10-fold to fit the axis. Real-time PCR values are averages of triplicate analyses. Left axis ({square}), real-time PCR; right axis ({circ}), dChip; ({triangleup}) MAS.

 
Validation of Data with dChip
Various software programs provide different algorithms for calculation of gene expression levels. Expression data were validated for some of the candidates by comparing them with data calculated with dChip. Profiles for the 10 candidate genes are included in Figure 4 . The expression profiles appear similar, although the dChip expression levels are much higher than those calculated by MAS, but, because all values are in arbitrary units, the profiles are not influenced by the different levels.

Immunohistochemistry
A possible relation between alterations at the RNA and protein level was examined. Three of the verified candidate genes were antibody stained (Fig. 5) . The anti-keratin 10/13 antibody recognized only K13 on formalin-fixed, paraffin-embedded tissue sections (information from supplier) which was also confirmed by the absence of staining in the negative control.



View larger version (94K):
[in this window]
[in a new window]
  		FIGURE 5. Immunostaining of corneas with keratin 6 (A, D), keratin 13 (B, E), and vimentin (C, F). Top: KC samples; bottom: reference samples (tumors). Bar, 50 µm.

 
Four KC and four reference corneas were immunohistochemically stained. Anti-K6 antibody stained the epithelium in three fourths of the KC samples and anti-K13 antibody halves of the samples in a heterogeneous way, along the longitudinal axis. One sample (Fig 5B) showed staining of only the superficial epithelium with K13. The two keratin antibodies did not stain the corneal epithelium of the four eyes with ocular tumor. No systematic difference was detectable for vimentin, which weakly stained both KC and reference.


   Discussion
Top
 Abstract
 Materials and Methods
 Results
 Discussion
References
 
Gene expression in KC epithelium was examined with microarrays. Epithelium from myopes was considered to be normal and used as the reference. Alternatively, corneas can be obtained from a tissue bank, but the RNA may be degraded by postmortem RNase activity, and gene expression may change when tissue is stored in culture medium. Consequently, we developed a protocol to snap freeze fresh corneal epithelium and thereby also prevent induction of new transcripts.

Corneal epithelium is separated from the stroma by the Bowman membrane, a rough layer of collagen that prevents contamination with keratocytes during isolation. The cornea is a unique tissue because it has no vascular system and therefore has no immune cells. Our samples can be regarded as very homogeneous, only infiltrated by nerve cells. The purity of the tissue samples is also implied by the absence of immunoglobulin gene expression findings in the data.

The samples included in this study were epithelium from 9 KC corneas from single patients, a left and right cornea from one patient with KC, 7 single normal reference corneas, and a pool of 18 reference corneas. The left and right samples were investigated in a comparison analysis that showed 12% changed probe sets. The proportion was 10% to 20% when the two samples were compared with any of the normal samples (data not shown). Based on this information, the samples were analyzed independently of each other.

Selection of candidate genes was performed in multiple steps. Low-stringency parameters were used to select 471 potential candidate probe sets. When stringency was increased, only 56 genes fulfilled the stated requirements. We assume that these candidates are likely to be involved in the KC disease; however, it cannot be ruled out that some of the excluded genes may be involved in the pathogenesis. The close relationship, illustrated in Figure 2 , between the left and right eye from the same individual supports the consistency of the candidates. Most of the differentially expressed genes are changing in the same direction and with similar magnitude for these two samples.

Cytokeratins are mostly expressed in epithelium. They are strong markers of differentiation.35 Keratins combine in pairs of a type I and II subunits. K5 and K14 are expressed in nonkeratinized stratified epithelium but when cells detach from the basement membrane and differentiate in the suprabasal layers, K4 and K13 expression is switched on.36 37 Superficial K13 staining was observed in one of the samples. K6 and K16 are markers for hyperproliferation and are expressed during wound healing and in skin with psoriasis.38 K6 (type II) and K13 (type I) were upregulated in KC but none of their partners was coexpressed. Perhaps, and this is strictly speculation, the keratins dimerize in KC into an atypical K6/K13 pair, a combination that has been shown in vitro.39 At present it is unknown whether such dimerization could have an impact on epithelial cell function.

Cytokeratins are involved in several diseases (reviewed by Corden and McLean40 ) but have so far only been related to a single corneal disease, Meesmann corneal dystrophy (OMIM 122100) where mutations in the K3 or K12 gene prevent heterodimerization of keratin.41 42 Two of the other selected candidate genes lysyl oxidase and TIMP3 are involved in the eye diseases Ehlers-Danlos syndrome type V (OMIM 305200) and Sorsby fundus dystrophy (OMIM 136900), respectively.

We observed changes, not only in the keratins but in all parts of the cytoskeleton: microfilaments, intermediate filaments, and microtubules. They all showed an increase in gene expression. This suggests the internal cell structure to be heavily reinforced in KC epithelium.

The finding of the two PDZ/LIM domain-containing proteins is most interesting, because they are able to construct a linkage between the cytoskeleton and intracellular kinases as part of signal transduction through the cell membrane. The extracellular part of a signal transduction complex is the ligand-recognizing receptor. Different receptors were selected: the LDL-binding sortilin-related receptor 1 suspected to be involved in Golgi-endosome sorting,43 44 and the uncharacterized G protein-coupled receptors GPR37 and RDC1. Information about their functions is very limited.

Homeostasis of the ECM is regulated by the processes of synthesis and degradation, the latter being tightly controlled by balancing the activity of proteinases and proteinase inhibitors. The observed increase of secreted proteinase inhibitors does not support the traditional idea of corneal protease degeneration (reviewed by Collier45 ), but it should be kept in mind that the data presented in this article originate from the epithelium and cannot be extrapolated to the entire cornea. If corneal degradation is caused by increased proteinase activity, it does not appear to be directed from the epithelium. On the contrary, our data suggest a reduced ECM remodeling. At present, a protocol is being developed for analysis of the stroma.

Cell-to-matrix and cell-to-cell interacting proteins were also well represented. Two of them, desmoglein 3 and bullous pemphigoid antigen (BP230), are involved in severe skin diseases.46 47 Another hemidesmosomal component besides BP230, the bullous pemphigoid antigen BP180, is suspected to be involved in KC, but Cheng et al.48 found no difference at the protein level. Mutations in {gamma}2-laminin cause the skin disease junctional epidermolysis bullosa.49 Galectin 7 and {gamma}2-laminin have also been found to increase in whole corneas from mice during wound healing.50

One of the differentially expressed transcripts between normal and KC tissues has not previously been described in the human cornea. Nagase et al.51 cloned KIAA0095 from a human myeloid cell line in 1994. We recently blasted the protein sequence to GenBank (http://www.ncbi.nlm.nih.gov/Genbank; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD) and found similarity to nucleoporin-interacting component NIC96 in yeast, An4 in frog, and dye (dead eye) in zebrafish; but now, 8 years after its cloning, it remains uncharacterized. Mutations in the dye gene were found to cause necrosis in eyes of zebrafish embryos.52

Most of the 56 altered candidate genes were in the direction of an increase. KC appeared to be characterized by relative small changes in gene expression with changes ranging from -2.7 to +6.6-fold (averages of 88 comparisons). Similar low magnitudes of change have been observed in other quantitative studies of KC,53 54 55 which may be related to the bradytrophic character of the disease.

The incidence of KC in patients with Down syndrome is higher than in the normal population56 57 and thus has been speculated to be related to chromosome 21 in some way. We examined the chromosomal localization of the 56 candidate genes, but none of them was positioned on chromosome 21.

This year, Fullerton et al.26 published the results of a genome scan study involving eight patients with KC, based on the founder population in Tasmania, Australia. We examined the loci of the candidate genes with the inherited regions from the study and found six of the candidates to be located within these conserved regions. Although the 10-cM genome region comprises several genes of interest, these six candidates are obvious subjects for further studies. Most recently, a Finnish study comprising 20 KC-affected families linked the disease to locus 16q22-q23,58 but none of our candidate genes located to this position.

The candidate genes represent multiple functional groups, whereas our focus was on the larger changes. Summarizing the results from the screening of 5600 genes, we observed increases in all three parts of the cytoskeleton and in proteins closely related thereto, and we saw signs of reduced ECM remodeling in the form of secreted proteinase inhibitors. KC causes changes in transmembrane signaling, together with altered cell-to-matrix cellular properties of the corneal epithelium. No indication of apoptosis was registered, as described by Kim et al.59 and recently by Kaldawy et al.60 in a TUNEL assay.

Analyses of microarray data are dependent on the algorithms and integrated cutoff values of the software. The MAS-calculated data were recalculated with dChip, which provides an alternative normalization and calculation of gene expression. First, dChip normalizes to a group of microarrays, in contrast to MAS, which normalizes microarrays separately; and second, the conservative MAS normalization uses the scaling factor across the entire expression set, whereas dChip uses a running median. Finally, the calculation of expression levels is diverse.23 24 Despite the differences in software, the two analyses yielded similar results for the 10 candidate genes. We also verified the expression with real-time PCR. The expression profiles were well reproduced, confirming that it is possible to determine gene expression in dissected corneas of individual patients with GeneChip microarrays. It also showed that when candidate genes are identified with the use of GeneChip microarrays, high throughput and low-cost real-time PCR is appropriate for subsequent and more extensive investigations.

High myopic refraction, corneal thinning, and opacification suggest an advanced stage of KC, but these cannot explain the heterogeneous gene expression among the KC samples observed with real-time PCR. The differences may be an indication of subgroups, but more samples must be investigated. Unsupervised cluster analysis was performed on the candidate probe sets, and the 19 samples were correctly separated into KC and normal groups, but no new clinical subgroups of KC were detected. That the two samples derived from the same patient did not cluster to the same node suggests that the samples are too few, compared to the variation in gene expression, to be divided into subgroups.

Changes at the mRNA level are not necessarily reflected at the protein level. We examined whether the alterations found on the GeneChip array would also be observed at protein level. Cytoskeletal changes were indicated by an increase of K6, K13, and vimentin at the RNA level. These three genes had been verified by real-time PCR, and commercial antibodies were available. From antibody staining of a small series of samples 4+4, we registered an overall relation between RNA and protein level for the two keratins. The heterogeneous staining of KC samples possibly demonstrates the natural variation that was also observed on the GeneChip array. Unfortunately, we did not have sets of cornea for both microarray and immunohistochemistry.

Microarray technology is rapidly entering the field of vision, as recent publication of studies involving cell cultures, whole corneal buttons, and knockout animals demonstrate.15 50 61 62 The current study is, to our knowledge, the first differential microarray study on human corneal epithelium that identifies genes differentially expressed between KC and normal tissue. We have proposed 56 candidate genes that changed expression level between 11 KC and 8 normal samples, based on different selection strategies. Each gene deserves further study in larger sample collections, with characterization of the respective corneal proteins. Instead of considering the cornea as a homogeneous tissue, we dissected the tissue and examined the epithelium. In the future we will use microarrays to investigate the stroma.


   Acknowledgements
 
The authors thank Jesper Hjortdal, MD, for his contribution to the collection of cornea samples, and Casper Moller Frederiksen, MSc, for the dChip analysis.


   Footnotes
 
Supported by the Danish Eye Research Foundation (Copenhagen, Denmark), the Danish Eye Health Society (Copenhagen, Denmark), Aarhus University Hospital Research Initiative (Aarhus County, Denmark), and the Institute of Experimental Clinical Research (Skejby University Hospital, Denmark).

Submitted for publication July 5, 2002; revised November 6 and November 27, 2002; accepted January 30, 2003.

Disclosure: K. Nielsen, None; K. Birkenkamp-Demtröder, None; N. Ehlers, None; T.F. Orntoft, 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: Torben Falck Orntoft, Molecular Diagnostic Laboratory, Skejby Sygehus Brendstrupgaaardsvej, DK-8200 Aarhus N, Denmark; orntoft{at}kba.sks.au.dk.


   References
Top
 Abstract
 Materials and Methods
 Results
 Discussion
 References
 

  1. Rabinowitz, YS. (1998) Keratoconus Surv Ophthalmol 42,297-319[CrossRef][Medline][Order article via Infotrieve]
  2. Rabinowitz, YS, Garbus, J, McDonnell, PJ. (1990) Computer-assisted corneal topography in family members of patients with keratoconus Arch Ophthalmol 108,365-371[Abstract]
  3. Forstot, SL, Goldstein, JH, Damiano, RE, Dukes, DK. (1988) Familial keratoconus Am J Ophthalmol 105,92-93[Medline][Order article via Infotrieve]
  4. Hammerstein, W. (1974) Zur Genetik des Keratoconus [Genetics of conical cornea] Albrecht Von Graefes Arch Klin Exp Ophthalmol 190,293-308[CrossRef][Medline][Order article via Infotrieve]
  5. Etzine, S. (1954) Conical cornea in identical twins SA Med J ,154-155
  6. Franceschetti, A, Lisch, K, Klein, D. (1958) Zwei eineiige Zwillingspaare mit konkordantem Keratokonus Monatsbl Augenheilk 133,15-30
  7. Hammerstein, W. (1972) Konkordanter Keratokonus bei eineiigen Zwillingen [Keratoconus concurrent in identical twins] Ophthalmologica 165,449-452[Medline][Order article via Infotrieve]
  8. Parker, J, Ko, WW, Pavlopoulos, G, et al (1996) Videokeratography of keratoconus in monozygotic twins J Refract Surg 12,180-183[Medline][Order article via Infotrieve]
  9. Lee, LR, Hirst, LW, Readshaw, G. (1995) Clinical detection of unilateral keratoconus Aust NZ J Ophthalmol 23,129-133[Medline][Order article via Infotrieve]
  10. Kennedy, RH, Bourne, WM, Dyer, JA. (1986) A 48-year clinical and epidemiologic study of keratoconus Am J Ophthalmol 101,267-273[Medline][Order article via Infotrieve]
  11. Rabinowitz, YS, Nesburn, AB, McDonnell, PJ. (1993) Videokeratography of the fellow eye in unilateral keratoconus Ophthalmology 100,181-186[Medline][Order article via Infotrieve]
  12. Abelson, MB, Collin, HB, Gillette, TE, Dohlman, CH. (1980) Recurrent keratoconus after keratoplasty Am J Ophthalmol 90,672-676[Medline][Order article via Infotrieve]
  13. Nirankari, VS, Karesh, J, Bastion, F, Lakhanpal, V, Billings, E. (1983) Recurrence of keratoconus in donor cornea 22 years after successful keratoplasty Br J Ophthalmol 67,23-28[Free Full Text]
  14. Kremer, I, Eagle, RC, Rapuano, CJ, Laibson, PR. (1995) Histologic evidence of recurrent keratoconus seven years after keratoplasty Am J Ophthalmol 119,511-512[Medline][Order article via Infotrieve]
  15. Kennan, A, Aherne, A, Palfi, A, et al (2002) Identification of an IMPDH1 mutation in autosomal dominant retinitis pigmentosa (RP10) revealed following comparative microarray analysis of transcripts derived from retinas of wild-type and Rho(-/-) mice Hum Mol Genet 11,547-557[Abstract/Free Full Text]
  16. Weber, BH, Vogt, G, Pruett, RC, Stohr, H, Felbor, U. (1994) Mutations in the tissue inhibitor of metalloproteinases-3 (TIMP3) in patients with Sorsby’s fundus dystrophy Nat Genet 8,352-356[CrossRef][Medline][Order article via Infotrieve]
  17. Patel, S, McLaughlin, JM. (1999) Effects of central corneal thickness on measurement of intra-ocular pressure in keratoconus and post-keratoplasty Ophthalmic Physiol Opt 19,236-241[CrossRef][Medline][Order article via Infotrieve]
  18. Avitabile, T, Marano, F, Uva, MG, Reibaldi, A. (1997) Evaluation of central and peripheral corneal thickness with ultrasound biomicroscopy in normal and keratoconic eyes Cornea 16,639-644[Medline][Order article via Infotrieve]
  19. Jongebloed, WL, Worst, JF. (1987) The keratoconus epithelium studied by SEM Doc Ophthalmol 67,171-181[CrossRef][Medline][Order article via Infotrieve]
  20. Thykjaer, T, Workman, C, Kruhoffer, M, et al (2001) Identification of gene expression patterns in superficial and invasive human bladder cancer Cancer Res 61,2492-2499[Abstract/Free Full Text]
  21. Eisen, MB, Spellman, PT, Brown, PO, Botstein, D. (1998) Cluster analysis and display of genome-wide expression patterns Proc Natl Acad Sci USA 95,14863-14868[Abstract/Free Full Text]
  22. Page, RD. (1996) TreeView: an application to display phylogenetic trees on personal computers Comput Appl Biosci 12,357-358[Free Full Text]
  23. Li, C, Wong, WH. (2001) Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection Proc Natl Acad Sci USA 98,31-36[Abstract/Free Full Text]
  24. Li, C, Hung, WW. (2001) Model-based analysis of oligonucleotide arrays: model validation, design issues and standard error application Genome Biol 2,research0032.1-research0032.11
  25. Cabral, A, Voskamp, P, Cleton-Jansen, AM, et al (2001) Structural organization and regulation of the small proline-rich family of cornified envelope precursors suggest a role in adaptive barrier function J Biol Chem 276,19231-19237[Abstract/Free Full Text]
  26. Fullerton, J, Paprocki, P, Foote, S, et al (2002) Identity-by-descent approach to gene localisation in eight individuals affected by keratoconus from north-west Tasmania, Australia Hum Genet 110,462-470[CrossRef][Medline][Order article via Infotrieve]
  27. Rodriguez-Crespo, I, Straub, W, Gavilanes, F, Ortiz de Montellano, PR. (1998) Binding of dynein light chain (PIN) to neuronal nitric oxide synthase in the absence of inhibition Arch Biochem Biophys 359,297-304[CrossRef][Medline][Order article via Infotrieve]
  28. Shimizu, N, Soda, Y, Kanbe, K, et al (2000) A putative G protein-coupled receptor, RDC1, is a novel coreceptor for human and simian immunodeficiency viruses J Virol 74,619-626[Abstract/Free Full Text]
  29. Cho, SH, Tam, SW, Demissie-Sanders, S, Filler, SA, Oh, CK. (2000) Production of plasminogen activator inhibitor-1 by human mast cells and its possible role in asthma J Immunol 165,3154-3161[Abstract/Free Full Text]
  30. Packard, BZ, Lee, SS, Remold-O’Donnell, E, Komoriya, A. (1995) A serpin from human tumor cells with direct lymphoid immunomodulatory activity: mitogenic stimulation of human tumor-infiltrating lymphocytes Biochim Biophys Acta 1269,41-50[Medline][Order article via Infotrieve]
  31. Si-Tahar, M, Merlin, D, Sitaraman, S, Madara, JL. (2000) Constitutive and regulated secretion of secretory leukocyte proteinase inhibitor by human intestinal epithelial cells Gastroenterology 118,1061-1071[CrossRef][Medline][Order article via Infotrieve]
  32. Leco, KJ, Khokha, R, Pavloff, N, Hawkes, SP, Edwards, DR. (1994) Tissue inhibitor of metalloproteinases-3 (TIMP-3) is an extracellular matrix-associated protein with a distinctive pattern of expression in mouse cells and tissues J Biol Chem 269,9352-9360[Abstract/Free Full Text]
  33. Zeeuwen, PL, Vlijmen-Willems, IM, Jansen, BJ, et al (2001) Cystatin M/E expression is restricted to differentiated epidermal keratinocytes and sweat glands: a new skin-specific proteinase inhibitor that is a target for cross-linking by transglutaminase J Invest Dermatol 116,693-701[CrossRef][Medline][Order article via Infotrieve]
  34. Rao, NV, Rao, GV, Hoidal, JR. (1997) Human dipeptidyl-peptidase. I. Gene characterization, localization, and expression J Biol Chem 272,10260-10265[Abstract/Free Full Text]
  35. Quinlan, R, Hutchison, C, Lane, B. (1995) Intermediate filament proteins Protein Profile 2,795-952[Medline][Order article via Infotrieve]
  36. Sawaf, MH, Ouhayoun, JP, Forest, N. (1991) Cytokeratin profiles in oral epithelial: a review and a new classification J Biol Buccale 19,187-198[Medline][Order article via Infotrieve]
  37. Nelson, WG, Sun, TT. (1983) The 50- and 58-kdalton keratin classes as molecular markers for stratified squamous epithelia: cell culture studies J Cell Biol 97,244-251[Abstract/Free Full Text]
  38. Machesney, M, Tidman, N, Waseem, A, Kirby, L, Leigh, I. (1998) Activated keratinocytes in the epidermis of hypertrophic scars Am J Pathol 152,1133-1141[Abstract]
  39. Hatzfeld, M, Franke, WW. (1985) Pair formation and promiscuity of cytokeratins: formation in vitro of heterotypic complexes and intermediate-sized filaments by homologous and heterologous recombinations of purified polypeptides J Cell Biol 101,1826-1841[Abstract/Free Full Text]
  40. Corden, LD, McLean, WH. (1996) Human keratin diseases: hereditary fragility of specific epithelial tissues Exp Dermatol 5,297-307[CrossRef][Medline][Order article via Infotrieve]
  41. Irvine, AD, Corden, LD, Swensson, O, et al (1997) Mutations in cornea-specific keratin K3 or K12 genes cause Meesmann’s corneal dystrophy Nat Genet 16,184-187[CrossRef][Medline][Order article via Infotrieve]
  42. Nishida, K, Honma, Y, Dota, A, et al (1997) Isolation and chromosomal localization of a cornea-specific human keratin 12 gene and detection of four mutations in Meesmann corneal epithelial dystrophy Am J Hum Genet 61,1268-1275[CrossRef][Medline][Order article via Infotrieve]
  43. Herz, J. (2001) The LDL receptor gene family: (un)expected signal transducers in the brain Neuron 29,571-581[CrossRef][Medline][Order article via Infotrieve]
  44. Jacobsen, L, Madsen, P, Nielsen, MS, et al (2002) The sorLA cytoplasmic domain interacts with GGA1 and -2 and defines minimum requirements for GGA binding FEBS Lett 511,155-158[CrossRef][Medline][Order article via Infotrieve]
  45. Collier, SA. (2001) Is the corneal degradation in keratoconus caused by matrix-metalloproteinases? Clin Exp Ophthalmol 29,340-344[Medline][Order article via Infotrieve]
  46. Adams, MJ, Reichel, MB, King, IA, et al (1998) Characterization of the regulatory regions in the human desmoglein genes encoding the pemphigus foliaceous and pemphigus vulgaris antigens Biochem J 329,165-174
  47. Mueller, S, Klaus-Kovtun, V, Stanley, JR. (1989) A 230-kD basic protein is the major bullous pemphigoid antigen J Invest Dermatol 92,33-38[CrossRef][Medline][Order article via Infotrieve]
  48. Cheng, EL, Maruyama, I, SundarRaj, N, et al (2001) Expression of type XII collagen and hemidesmosome-associated proteins in keratoconus corneas Curr Eye Res 22,333-340[CrossRef][Medline][Order article via Infotrieve]
  49. Aberdam, D, Galliano, MF, Vailly, J, et al (1994) Herlitz’s junctional epidermolysis bullosa is linked to mutations in the gene (LAMC2) for the gamma 2 subunit of nicein/kalinin (LAMININ-5) Nat Genet 6,299-304[CrossRef][Medline][Order article via Infotrieve]
  50. Cao, Z, Wu, HK, Bruce, A, Wollenberg, K, Panjwani, N. (2002) Detection of differentially expressed genes in healing mouse corneas, using cDNA microarrays Invest Ophthalmol Vis Sci 43,2897-2904[Abstract/Free Full Text]
  51. Nagase, T, Miyajima, N, Tanaka, A, et al (1995) Prediction of the coding sequences of unidentified human genes. III. The coding sequences of 40 new genes (KIAA0081-KIAA0120) deduced by analysis of cDNA clones from human cell line KG-1 DNA Res 2,37-43[Abstract]
  52. Allende, ML, Amsterdam, A, Becker, T, et al (1996) Insertional mutagenesis in zebrafish identifies two novel genes, pescadillo and dead eye, essential for embryonic development Genes Dev 10,3141-3155[Abstract/Free Full Text]
  53. Sawaguchi, S, Yue, BY, Sugar, J, Gilboy, JE. (1989) Lysosomal enzyme abnormalities in keratoconus Arch Ophthalmol 107,1507-1510[Abstract]
  54. Sawaguchi, S, Twining, SS, Yue, BY, et al (1990) Alpha-1 proteinase inhibitor levels in keratoconus Exp Eye Res 50,549-554[CrossRef][Medline][Order article via Infotrieve]
  55. Whitelock, RB, Fukuchi, T, Zhou, L, et al (1997) Cathepsin G, acid phosphatase, and alpha 1-proteinase inhibitor messenger RNA levels in keratoconus corneas Invest Ophthalmol Vis Sci 38,529-534[Abstract/Free Full Text]
  56. Cullen, JF, Butler, HG. (1963) Mongolism (Down’s syndrome) and keratoconus Br J Ophthalmol 47,321-330
  57. Shapiro, MB, France, TD. (1985) The ocular features of Down’s syndrome Am J Ophthalmol 99,659-663[Medline][Order article via Infotrieve]
  58. Tyynismaa, H, Sistonen, P, Tuupanen, S, et al (2002) A locus for autosomal dominant keratoconus: linkage to 16q22.3-q23.1 in Finnish families Invest Ophthalmol Vis Sci 43,3160-3164[Abstract/Free Full Text]
  59. Kim, WJ, Rabinowitz, YS, Meisler, DM, Wilson, SE. (1999) Keratocyte apoptosis associated with keratoconus Exp Eye Res 69,475-481[CrossRef][Medline][Order article via Infotrieve]
  60. Kaldawy, RM, Wagner, J, Ching, S, Seigel, GM. (2002) Evidence of apoptotic cell death in keratoconus Cornea 21,206-209[CrossRef][Medline][Order article via Infotrieve]
  61. Mahajan, VB, Wei, C, McDonnell, PJ, III (2002) Microarray analysis of corneal fibroblast gene expression after interleukin-1 treatment Invest Ophthalmol Vis Sci 43,2143-2151[Abstract/Free Full Text]
  62. Yoshida, S, Yashar, BM, Hiriyanna, S, Swaroop, A. (2002) Microarray analysis of gene expression in the aging human retina Invest Ophthalmol Vis Sci 43,2554-2560[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 IOVSHome page
V. Barbaro, E. Di Iorio, S. Ferrari, L. Bisceglia, A. Ruzza, M. De Luca, and G. Pellegrini
Expression of VSX1 in Human Corneal Keratocytes during Differentiation into Myofibroblasts in Response to Wound Healing
Invest. Ophthalmol. Vis. Sci., December 1, 2006; 47(12): 5243 - 5250.
[Abstract] [Full Text] [PDF]

Home page
 IOVSHome page
X. Li, Y. S. Rabinowitz, Y. G. Tang, Y. Picornell, K. D. Taylor, M. Hu, and H. Yang
Two-stage genome-wide linkage scan in keratoconus sib pair families.
Invest. Ophthalmol. Vis. Sci., September 1, 2006; 47(9): 3791 - 3795.
[Abstract] [Full Text] [PDF]

Home page
 J. Biol. Chem.Home page
M. Zhou, X.-m. Li, and R. M. Lavker
Transcriptional Profiling of Enriched Populations of Stem Cells Versus Transient Amplifying Cells: A COMPARISON OF LIMBAL AND CORNEAL EPITHELIAL BASAL CELLS
J. Biol. Chem., July 14, 2006; 281(28): 19600 - 19609.
[Abstract] [Full Text] [PDF]

Home page
 IOVSHome page
M. Saghizadeh, A. A. Kramerov, J. Tajbakhsh, A. M. Aoki, C. Wang, N.-N. Chai, J. Y. Ljubimova, T. Sasaki, G. Sosne, M. R. J. Carlson, et al.
Proteinase and Growth Factor Alterations Revealed by Gene Microarray Analysis of Human Diabetic Corneas
Invest. Ophthalmol. Vis. Sci., October 1, 2005; 46(10): 3604 - 3615.
[Abstract] [Full Text] [PDF]

Home page
 IOVSHome page
Y. S. Rabinowitz, L. Dong, and G. Wistow
Gene Expression Profile Studies of Human Keratoconus Cornea for NEIBank: A Novel Cornea-Expressed Gene and the Absence of Transcripts for Aquaporin 5
Invest. Ophthalmol. Vis. Sci., April 1, 2005; 46(4): 1239 - 1246.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
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 PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Nielsen, K.
Right arrow Articles by Orntoft, T. F.
Right arrow Search for Related Content