Differential expression of genes within the cochlea as defined by a custom mouse inner ear microarray

Abstract

Microarray analyses have contributed greatly to the rapid understanding of functional genomics through the identification of gene networks as well as gene discovery. To facilitate functional genomics of the inner ear, we have developed a mouse inner-ear-pertinent custom microarray chip (CMA-IE1). Nonredundant cDNA clones were obtained from two cDNA library resources: the RIKEN subtracted inner ear set and the NIH organ of Corti library. At least 2000 cDNAs unique to the inner ear were present on the chip. Comparisons were performed to examine the relative expression levels of these unique cDNAs within the organ of Corti, lateral wall, and spiral ganglion. Total RNA samples were obtained from the three cochlear-dissected fractions from adult CF-1 mice. The total RNA was linearly amplified, and a dendrimer-based system was utilized to enhance the hybridization signal. Differentially expressed genes were verified by comparison to known gene expression patterns in the cochlea or by correlation with genes and gene families deduced to be present in the three tissue types. Approximately 22-25% of the genes on the array had significant levels of expression. A number of differentially expressed genes were detected in each tissue fraction. These included genes with known functional roles, hypothetical genes, and various unknown or uncharacterized genes. Four of the differentially expressed genes found in the organ of Corti are linked to deafness loci. None of these are hypothetical or unknown genes.

Fig. 1

Dissection of the mouse cochlea. A histological representation of the dissected fractions of the mouse inner ear is depicted. Three functionally distinct regions of the cochlea were chosen for isolation and represent the lateral wall (LW) comprising the stria vascularis, spiral ligament and spiral prominence, the organ of Corti (OC), and the spiral ganglion (SG) neurons. All samples are comprised of tissue from all three turns including the basal hook region.

Fig. 2

Assessment of total RNA and amplified RNA samples. Quality assessment of total and amplified RNA. Bioanalyzer gel image (A) of total RNAs is represented by duplicated samples from each of the three dissected cochlear regions. In each sample, the 28S, 18S, and 5S ribosomal bands are visible, and this is depicted by the bioanalyzer electropherogram (B) of total RNA isolated from the OC. (C) Bioanalyzer electropherogram of 1× amplified RNA (aRNA) from the OC overlaid on the corresponding electropherogram of a 6-kb RNA ladder (Ambion, Inc., Austin, TX). For all aRNA samples, the size range was about 100–6000 bases, with a midpoint of about 1000 bp.

Fig. 3

Utilization of genes specifying ribosomal proteins in normalization assessment. Intensity levels of a large set of genes specifying ribosomal proteins present on the array were examined. Data are for one of the organ of Corti versus spiral ganglion comparisons, but similar results were observed across all experiments. The ribosomal protein genes were categorized into high (>1000 ESTs), moderate (>500 ESTs), and low (<250 ESTs) expression levels based on the number of ESTs in the entire mouse GenBank database. Genes with moderate and low expression levels were cytoplasmic ribosomal proteins and are designated. Those with background signal levels were mitochondrial ribosomal protein genes and are Mrpl10, Mrpl51, Mrps9, Mrps16, and Mrps19a.

Fig. 4

Locally weighted (lowess) log ratio normalization of microarray data. A microarray comparison between SG (red signal) and LW (green signal) is shown. (A) A typical M versus A scatter plot of unnormalized microarray data. The x-axis is 0.5log2 (red intensity × green intensity) or A. The y-axis is log intensity ratio or M. The log ratio values are skewed to the negative, especially at low intensities. (B) M versus A scatter plot of normalized microarray data. The distribution is now symmetrical about zero at all intensity values. Data are typical for all comparisons performed.

Fig. 5

SAM plots for cochlear dissected fraction expression comparisons. Comparisons were made between LW vs. OC (A), SG vs. OC (B), and SG vs. LW (C). For each comparison, the total number significant, median number false significant, and delta value are shown. Significant differentially expressed genes are represented by either green (tissue 1) or red (tissue 2) data points shown outside of the delta cutoff lines. The delta value is set to maximize the significant data points and minimize the false significant.