Abstract
Purpose: :
Gene expression patterns in the eye are highly variable among diverse but non–mutant strains of mice. Genetic covariance among transcripts defines networks of genes that are functionally coupled––a new and efficient strategy for annotating multifaceted gene function and even more complex causal linkage between allelic variants and differences in disease susceptibility––glaucoma, macular degeneration, retinoblastoma, and many others.
Methods: :
Estimates of steady–state mRNA levels (Affymetrix M430 2.0) were acquired from sets of eyes taken from healthy young adults belonging to ∼70 strains and both sexes, including over 55 BXD recombinant inbred lines and 15 highly diverse isogenic strains. Excellent sequence and SNP data are available publicly for all of these strains. DBA/2J and ∼15 BXD strains are at high risk for developing glaucoma (Tyrp1 and Gpnmb glaucoma alleles); two strains have severe retinitis pigmentosa; and five are Tyr–negative albinos. Expression data were entered in GeneNetwork (www.genenetwork.org). This new resource is called the Hamilton Eye Institute Mouse Eye Database (HEIMED).
Results: :
Cell–specific gene expression networks can be defined even when whole eye is used for analysis. This is particularly true for photoreceptor–dependent mRNAs that define a robust coexpression network (see abstract by Geisert and colleagues). However, even the expression signals of less numerous cell populations can be statistically deconvolved using genetic covariance. For example, signatures of off–center cone bipolar are well linked (Tacr3 and Vsx1, p = E–7). One of the most interesting uses of HEIMED is to define candidate genes for known variants in retinal structure. The retinal ganglion cell population is tightly coupled to the cancer susceptibility candidate 3 gene (Casc3) on Chr 11 at 98.6 Mb, making this gene an excellent candidate for the ganglion cell number controller locus, Nnc1.
Conclusions: :
We have devised a novel genetic and statistical approach that can simultaneously provide insight into cell–specific and system–wide gene expression networks in the mammalian eye. The power and sensitivity of this global approach is a function of the number of strains, their range of genetic variation, the number of replicates within strain, and the adequacy of control for technical confounds.
Keywords: gene microarray • genetics • retina: proximal (bipolar, amacrine, and ganglion cells)