25.1 Introduction

Hyperoxia has been shown to be specifically toxic to photoreceptors in the retina of the rabbit and rodents (Noell 1955; Yamada et al. 1999; Yamada et al. 2001; Walsh et al. 2004; Geller et al. 2006; Natoli et al. 2008a, Natoli et al. 2008b), and may be an important factor in the later stages of photoreceptor degenerations, as the photoreceptor population is depleted (Stone et al. 1999). This study investigates a regional variation in the susceptibility of the mouse to hyperoxia, in which photoreceptor degeneration reliably begins in retina close to (0.5 mm from) and inferior to the optic disc (Smit-McBride et al. 2007; Natoli et al. 2008a). The source of this regional variation is not known, although factors such as rhodopsin content and oxygen delivery have been discussed (Rapp and Williams 1977; Chung et al. 1999). In a recent study, we have examined global changes in gene expression induced in the C57BL/6 J mouse retina by hyperoxic exposure for up to 14d (Natoli et al. 2008b). Here we have examined differences in gene expression between inferior and superior retina of C57BL/6 J mouse, before and after hyperoxic exposure.

25.2 Methods

C57BL/6 J mice, a hyperoxia-vulnerable strain, were raised in dim cyclic illumination (12 h 5 lux, 12 h dark) and maintained in this level of lighting during the experiment. At the age of P (postnatal day) 83–90, some mice were exposed to constant hyperoxia (75% O2) for 14 days. The mice were housed, with free access to food and water, in plexiglass chambers in which the oxygen level was controlled by a feedback device (OXYCYCLER, Reming Bioinstruments, NY).

Retinas from control (normoxic) and hyperoxia-challenged mice were removed and divided into superior and inferior halves, and RNA was extracted. The RNA from 2 animals (1 male and 1 female) at each condition was pooled, purified and hybridized to an Affymetrix GeneChip R Mouse Gene 1.0 ST Array. This array is comprised of more than 750,000 unique oligonucleotide features constituting more than 28,000 genes. Labelling, hybridization and scanning of the microarray were performed at the ACRF Biomolecular Resource Facility at the John Curtin School of Medical Research, Australian National University.

Experiments were run in duplicate and analysis of the expression patterns was performed using GeneSpring (Agilent Technologies, Santa Clara, CA) and Partek Genomics Suite (Partek Inc., St. Louis, MO) software. Analyses of gene ontology and molecular pathways were performed using Ingenuity Pathway Analysis (Ingenuity R Systems, www.ingenuity.com) for the most highly variable genes in inferior and superior retina after hyperoxia.

Quantitative PCR (qPCR) was used to validate the expression changes of nine genes identified in the microarray experiment. Three biologic groups were used and RNA extraction was carried out as for microarray experiments. Taqman R probes were obtained (Applied Biosystems, Foster City, CA) and qPCR performed on

a Rotor-Gene 3,000 (Corbett Robotics, Mortlake, NSW, Australia). GAPDH was included as a reference gene and cycle threshold means were used to calculate fold change using the Pfaffl Equation (Pfaffl, 2001).

25.3 Result

The array quality was assessed by pseudocolour imaging and a sample graph of probe intensities (Fig. 25.1). All eight samples showed consistent pattern of labelling and distribution, indicating that there were no major hybridization defects.

Fig. 25.1 Pseudocolour images (left) and a sample graph (right) of log2 probe intensities for each of the 8 samples. Each pseudocolour image (shown here in black and white) was created from the logged CEL file data to display signal intensities for the whole of the microarray. The graph provides another comparison of the distributions of gene expression as a function of fold change. The pseudocolour images were examined at high resolution (not illustrated), for evidence of discontinuities in hybridisation. These diagrams indicate that there were no major technical flaws in any of the 8 hybridizations involved

Hierarchical clustering was performed using the heat map function in Partek Genomics Suite (Fig. 25.2). The replicates of one group should cluster more closely to each other, than to other samples. The analysis showed strong replicate clustering for three groups (1 with 5; 3 with 7; 4 with 8) but not sample 2 with sample 6. The distance between these two samples suggest some biological variability of

Fig. 25.2 Hierarchical clustering diagram of genes (rows) and samples (columns). The genes are organised vertically in each sample by the level of their expression, with lowest expression genes at the top of the column and highest expression genes at the bottom. The columns are clustered according to the similarity of expression of individual genes. Samples 1 and 5 are replicates of superior regions of control retinas; samples 2 and 6 are replicates of inferior regions of control retinas; samples 3 and 7 are replicates of superior regions of hyperoxia-exposed retinas; samples 4 and 8 are replicates of inferior regions of hyperoxia-exposed retinas

gene expression in inferior control retina. Overall, superior retina samples prior to and after hyperoxia clustered more closely to each other than to other groups. Gene expression in the inferior retina diverged from superior retina, with hyperoxia increasing the separation.

Data were then analysed by two-way ANOVA (analysis of variance) using a 2 fold change as the criterion for significant differences (increases and decreases) in expression (Table 25.1). There are significantly more genes differentially expressed in the inferior than superior retina (e.g. 392 genes changed significantly in the superior retina while 849 genes changed in the inferior at p ≤ 0.01). When the criteria

Table 25.1 Numbers of genes changing expression with hyperoxia, in superior and inferior retina, assessed with different levels of probability

↑, up-regulation, ↓ down-regulation

combined the p-value ≤ 0.01 with fold change > 2, the number of genes differentially expressed was 5.5 times greater in the inferior than in the superior retina (78 vs 14). The genes showing the largest fold difference between inferior and superior retina after hyperoxia include Edn2 (endothelin 2), C1qb (complement component 1, q subcomponent, beta polypeptide), Gfap (glial fibrillary acidic protein) and Bcl-3 (B-cell leukemia/lymphoma 3).

Expression changes of 9 genes chosen using the combined criteria of p-value ≤ 0.01 with fold change > 2 were tested by qPCR, to verify changes found in the gene array, including Edn2, Gfap, Bcl-3, nuclear receptor subfamily 2, group E, member 3 (Nr2e3), wingless-related MMTV integration site 8A (Wnt8a), cyclic nucleotide gated channel alpha 2 (Cnga2), exportin 6 (Xpo6), medium-wave- sensitive cone opsin 1 (Opn1mw) and hypoxia inducible factor 1, alpha subunit (Hif1a). With the exception of Xpo6, the regulation changes of all these genes were confirmed by qPCR (Table 25.2).

Table 25.2 Validation of microarray data using quantitative PCR

The results of quantitative PCR were from three independent experiments; Sup, superior; Inf, inferior