Donor eye tissue preparation
This study was carried out at the Queens Medical Centre, University Hospital Nottingham, England with approval of Nottingham Research Ethics Committee (REC NO: OY030202). Protocol was consistent with Tenets of Declaration of Helsinski. Informed, written consent was obtained from relatives of all the donors. The eyes were harvested within 48 hours of death under aseptic conditions using conventional techniques in order to maintain the RNA viability. Four pairs of human donor eyes were collected for microarray study. The inclusion criteria were: i) donors aged between 20 to 70 years; ii) donors of either sex; iv) Eyes with intact and undamaged OS epithelium, confirmed with dissecting microscope and patient history as ascertained from the case notes. The corneoscleral button was dissected from the cadaver eye and processed for sectioning using established techniques within our department . Briefly, a 15 mm corneal button was trephined maintaining a 3 mm frill of conjunctiva around the limbus and divided into eight triangular radial segments. Each segment was positioned in the optimum temperature compound (OCT, Emitech Ltd, East Sussex, England) with the long dissected edge parallel to the OCT surface and then gradually frozen using isopentane precooled in liquid Nitrogen. The frozen tissue blocks were stored at -80°c for future cryosectioning.
Processing of Standard probe samples
Depending on the number of biological samples and their replicates sufficient amount of control samples are required for any microarray study. As it was not possible to generate required amount of control sample with LMD, we followed the reference probe hybrid approach for sample processing as described by Neal and Westwood . Briefly, the reference samples were prepared by pooling the corneal and conjunctival epithelial RNA. The tissue for RNA extraction was obtained by scrapping the OS epithelial regions from the cadaver eyes. This approach generated sufficient amount of reference samples for the microarrays, facilitated better comparison between the two regional arrays and also highlighted the variations in gene expression between the biological replicates but not the reference samples. For real time PCR, RNA extraction for each OS region was performed in triplicate from three different cadaver eyes using RNeasy Microkit (Qiagen, Crawley, West Sussex, UK) according to manufacturer's protocol.
Laser Microdissection (LMD)
Under RNase and DNase free conditions, 6-7 μm serial sections of frozen tissue blocks of corneoscleral buttons were prepared with Jung CM 1900 cryostat (Leica, UK) and examined under light microscope for presence of LEC and also for good epithelial histology of other OS regions. Prior to LMD, sections were placed on poly-L-lysine coated PALM® membrane slides, fixed in precooled 70% v/v ethanol for 5 minutes and air dried. Following which, the sections were briefly stained with 0.1% w/v Toludine Blue for 30 seconds, washed in DEPC treated water and air dried. LMD was performed with the PALM® Microbeam systems (Zeiss Instruments, Bernreid, Germany), using the Robo LPC laser function, according to manufacturer's recommended guidelines. The area of interest was cut and catapulted in the caps of collection tubes, coated with special adhesive. Thereafter RLT RNA lysis buffer (QIAGEN) was added to the collection tubes and stored at -80°c until further use. For each of the 4 eyes, multiple LMD samples were collected from five regions creating 5 independent experimental groups; 1) LEC; 2) limbus; 3) cornea; 4) LEC stroma, and 5) conjunctiva.
Total RNA Extraction
For microarrays total RNA extraction from LMD sections for samples and reference sample was performed with RNeasy kit, including DNase treatment, according to manufacturer's protocols (QIAGEN House, West Sussex, UK). RNA quantity and quality was measured with Picoassay 2100 Bioanalyzer (Agilent Technologies, USA). Samples with concentration ranging between 20-90 pg/μl and an average RIN value (RNA Integrity Number) of 5.1  were used for further analysis.
Preparation of reference sample for Standard Probe
1 μg of total RNA from each corneal and conjunctival reference sample was mixed and ethanol precipitated by adding 0.1 volumes Sodium Acetate and 2.5 volumes 100% ETOH followed by incubation at -20°c for 30 minutes. Samples were then centrifuged at 21000 × g at 4°c for 15 minutes and the pellet washed in 250 μl of 70% v/v ETOH before recentrifugation at 21000 × g for 5 min. The RNA pellet was dried and resuspended in 10 μl ultra pure, RNase free water followed by quantitation using NanoDrop ND-1000 UV-Vis Spectrophotometer (Labtech International Ltd-UK).
RNA Amplification, labelling of the samples and reference sample
Each RNA sample and reference sample was further processed for microarray analysis. To 0.2 ng/10 μl of starting RNA, 0.5 μl of 1:5000 diluted spike control (GE Healthcare life Sciences Universal Score card oligonucleotide control), which are sequences from E. coli genes, was added for validation of microarray data. Complimentary RNA (cRNA) amplification was performed with Amino Allyl Message Amp™ II aRNA Amplification kit and labelled with Cy3 and Cy5 reactive dye [Ambion, (Europe) LTD, UK], the Frequency of Incorporation (FOI) of the Cy3 and Cy5 dye in the labelled samples was measured, according to manufacturer's protocols. Quality control of the purified cRNA samples was performed at the end of 1st and 2nd rounds by NanoDrop spectrophotometer. Unsatisfactorily amplified and labelled samples were discarded and new samples were processed.
500 ng of a labelled sample and reference sample with matching FOI were separately blocked with 2 μl Poly A and 2 μl human Cot 1 DNA and then combined. This was followed by ethanol precipitation to generate 2 μl of the hybrid probe in nuclease free water. Prior to hybridisation, the 30K Human spotted oligonucleotide glass arrays manufactured in house, (Post-Genomics technologies facility, University of Nottingham, UK), were blocked with appropriate buffers (5× SSC, 0.2% w/v SDS, 1% w/v BSA), (SSC buffer: Saline, Sodium Citrate buffer; SDS buffer: Sodium Dodecyl. Sulphate; BSA: Bovine Serum Albumin). The slides were then washed thrice with ultra pure water, 100% ETOH and spun dried. For hybridisation, 100 μl of pre warmed Schott 1× Hybridisation Buffer was gently added to the hybrid probe, heated to 95°C, 2 minutes followed by hybridisation on automated hybridisation station TECAN HS 4800 (Tecan UK Ltd) using manufacturers protocols. The conditioned hybridisation station was primed with ultra pure water followed by a 1 min and a 15 min wash in 5× SSC, 0.2% w/v SDS, at 50°C. Next, 100 μl of probe was injected onto the slide and hybridised with agitation at 50°C for 16 hours, followed by sequential washing in four cycles of 2× SSC, 0.1% SDS, 2 cycles of 1× SSC, 0.1% SDS, and 0.1× SSC, 0.2% SDS at 40°C, 2 cycles of 0.1× SSC at 23°C, followed by a final cleaning cycle of ultrapure water before drying. On completion of the programme, the slides were covered to protect against light and scanned immediately.
Scanning and Data analysis
Hybridised microarray slides were scanned to obtain two coloured digital images on an Agilent BA scanner. The images were further analysed with Gene Pix Pro 6.0 software. Poor spots and spots overlying areas of high background intensities were lassoed and removed from further analysis. However, spots of varying sizes but with good intensities were included in analysis. The raw data was uploaded on BASE (Bio Array Software Environment; (Lund University, Lund, Sweden) which is a MIAME (Minimum Information about a Microarray Experiment) compliant system . The raw data sets were expressed as log ratio of channel 1/channel 2 intensities or log ratio of Cy5/Cy3 (sample/standard probe). Bioassay sets created from the raw data sets were further filtered to refine the data by removing 'noise'. Following filtrations, intra-array Lowess normalisation was performed on BASE, and the bioassay sets were then imported to J-Express Pro software (http://www.molmine.com; MolMine AS, Norway). In J-Express inter array scale normalisation was performed on the imported data sets. Further statistical analysis of the data was performed in J-Expresspro using a feature subset selection algorithm (FSS) for two unpaired groups comparison with following parameters: P value was selected for score method, individual ranking of the genes was performed and fold change values were log(2) transformed.
Significance Analysis of Microarrays (SAM)
Differentially expressed gene list was generated from the FSS derived gene list (p value ≤ 0.05) by performing SAM on the data to determine the fold change and the False Discovery Rate (FDR). Cutoff limit of FDR was set to ≤ 5% and genes with FDR above this limit were excluded from the analysis. Following SAM the five individual gene lists generated for each of the OS regions were merged to form an overlapping gene list for each region.
Principal Component Analysis (PCA)
PCA evaluates variation between the samples. The 2D plots representing the samples (Figure 2 left) and the genes (Figure 2 right) were generated following FSS analysis in J-Expresspro. The variance of axes was displayed in percentage. The data with greatest variation is clustered at the first principal component axis.
Gene Ontology (GO)
The enriched GO terms for each region was determined by uploading the significant gene list for each region individually on the Database for Annotation, Visualization, and Integrated Discovery (DAVID) v6.7, 2008, (http://david.abcc.ncifcrf.gov) and the data was analysed according to published methodology [88, 89]. GO terms associated with the biological processes (GO~BP) were mostly considered for analysis. Statistical significance of the GO terms was established by Fishers Exact T test. GO terms with p values < 0.05 were considered statistically significant.
Canonical pathways on Ingenuity Pathway Analysis
The microarray data was also analysed with Ingenuity Pathways Analysis (IPA) version 7.6 (Ingenuity® Systems, http://www.ingenuity.com). IPA identified the canonical signalling and metabolic pathways from the IPA library that were most significant to the data set. The significance of the association between the data set and the canonical pathway was measured as a ratio and the p value calculated with Fischer Exact test. The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE19035 http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE19035.
Quantitative Gene Expression Analysis (Real-time PCR)
The relative quantification of mRNA for genes of interest on OS epithelium (LEC, limbus, cornea, and conjunctiva) was performed with real-time PCR. Approximately 1 ng/μl concentration of RNA was used for cDNA synthesis using QuantiTect Reverse Transcription Kit (QIAGEN) according to manufacturer's protocol. Inventoried Taqman assays (Applied Biosystems, Foster City, CA) were used for selected genes of interest. Each reaction was performed in triplicate with final reaction volume of 20 μl. The reaction components comprised of, 10 μl 2× Taqman Gene Expression Master Mix (Applied Biosystems), 1 μl of 20× Taqman Assay probes (Applied Biosystems), 1 μl cDNA (1:2 dilutions), 8 μl nuclease free water (Promega UK, Southampton, UK). Non template, reverse transcriptase negative and positive cDNA from Universal Human Reference RNA (Stratagene, La Jolla, CA) were run as controls. Amplification was performed on the Mx3005P multicolour 96 well PCR system (Stratagene Europe, Amsterdam, Netherlands) with the following parameters, 50°C for 2 min and then 95°C for 10 min followed by 45 cycles of 95°C for 15 seconds and 60°C for 1 minute. Data analysis was performed on Mxpro ver 4.2 software to measure the threshold cycle (Ct) for each reaction. Set of triplicate Ct values for each sample was averaged. Normalisation of the average sample gene expression was performed with the average Ct value of 18S rRNA endogenous gene control for that sample.
Statistical Analysis of Real time PCR Data
The real time PCR data was statistically analysed on SPSS ver 16. The average normalised Ct values of gene of interest for each of the OS regions were subjected to Levene's test, to measure the equality of variance. The normally distributed data was then analysed with one way analysis of variance (ANOVA) with Bonferonni's posthoc test. Non parametric distributed samples were analysed by Kruskal-Wallis test followed by Mann-Whitney test, p value < 0.05 was considered statistically significant.
Six micron frozen tissue sections mounted on 2% 3-aminopropyltriethoxysilane (APES), (SIGMA) coated slides were fixed in acetone followed by three washes with wash solution consisting of 1%PBS, 1% BSA and 0.1% v/v Triton-X for 5 minutes each. The sections were then blocked using 10% v/v goat serum (Invitrogen, UK), followed by over night incubation with primary antibody prepared in the above mentioned wash solution at 4°C. Primary antibodies were rabbit anti human polyclonal FRZB (H-170), dilution, 1:100 (Santa Cruz Biotechnology, INC Europe); and HES1, rabbit anti human, dilution 1:100, (US Biological, USA). Slides were washed 3 times for 5 minutes in wash solution followed by detection of the primary with secondary antibody Goat anti-rabbit conjugated Alexa Fluor 555 (Dilution 1:300; Invitrogen Ltd, Paisley, Scotland) for 30 minutes. This was followed by washes with above mentioned wash solution for 5 minutes and then counterstained with nuclear stain DAPI (4', 6-diamidino-2-phenylindole) (Dilution: 25 μl in 5 ml PBS of stock solution 1 mg in 4 ml PBS, Santacruz Biotechnology, UK) for 4 minutes followed by further washing. Slides mounted with fluorescent mounting medium (Dako, Ely, UK) were visualised using the B51X fluorescent microscope with software CELL^F (Olympus UK Ltd, UK).