Genetic regulation of gene expression during shoot development in Arabidopsis

Abstract

The genetic control of gene expression during shoot development in Arabidopsis thaliana was analyzed by combining quantitative trait loci (QTL) and microarray analysis. Using oligonucleotide array data from 30 recombinant inbred lines derived from a cross of Columbia and Landsberg erecta ecotypes, the Arabidopsis genome was scanned for marker-by-gene linkages or so-called expression QTL (eQTL). Single-feature polymorphisms (SFPs) associated with sequence disparities between ecotypes were purged from the data. SFPs may alter the hybridization efficiency between cDNAs from one ecotype with probes of another ecotype. In genome scans, five eQTL hot spots were found with significant marker-by-gene linkages. Two of the hot spots coincided with classical QTL conditioning shoot regeneration, suggesting that some of the heritable gene expression changes observed in this study are related to differences in shoot regeneration efficiency between ecotypes. Some of the most significant eQTL, particularly those at the shoot regeneration QTL sites, tended to show cis-chromosomal linkages in that the target genes were located at or near markers to which their expression was linked. However, many linkages of lesser significance showed expected "trans-effects," whereby a marker affects the expression of a target gene located elsewhere on the genome. Some of these eQTL were significantly linked to numerous genes throughout the genome, suggesting the occurrence of large groups of coregulated genes controlled by single markers.

Figure 1.

Plot of gene expression vs. shoot regeneration phenotype (shoots per explant on square-root scale) for At5g48330 in the 30 RI lines used in this study. At5g48330, encoding a regulator of chromosome condensation family protein, shows the strongest relationship between gene expression and phenotype. Linear regression line and parental allele at the shoot regeneration QTL are indicated by C (Col) or L (Ler). Expression values represent MAS 5.0 signals that were logged and mean centered for each gene chip. A negative expression reflects a gene with a logged MAS 5.0 value below the average logged expression of genes on a gene chip.

Figure 2.

Density map of significant linkages in a genomewide scan. The Arabidopsis genome was scanned for marker-by-gene associations for 288 evenly spaced markers. Scans were conducted at different thresholds as described in the text. Results from two thresholds and the corresponding number of significant linkages are shown here. Plot is corrected by elimination of data for SFP-affected probe pairs.

Figure 3.

Examples of the effect of the genotype at marker 270 on the expression of various genes to which the marker is significantly linked. Expression levels in the 30 RI lines are grouped according to the presence of the Landsberg erecta or Columbia allele at marker 270. Horizontal lines represent QTL genotype group means.

Figure 4.

Allele frequency distribution in a genomewide scan. (A) Out-of-balance group size or the proportion of Col or Ler alleles in the larger group at each of the 288 markers used in genome scans. (B) The comparison-wise power of detecting a 1.5 standard deviation in the average gene expression at each marker based on a type I error rate of 0.05. Power is greatest when genotypes are known and equal numbers of the two different parental alleles are present at a given marker. The two spikes associated with lower power are due to missing genotypes at the given markers. (C) Distribution of a number of significant linkages at a threshold associated with a FDR of 2.3%.

Figure 5.

Location of genes linked to markers at the eQTL hot spots in the Arabidopsis genome at a threshold associated with a FDR of 2.3%. Markers 190 and 270 are centered on the shoot regeneration QTLs. Ticks pointing upward show the location of upregulated genes and downward pointing ticks are downregulated genes. Markers are located about every 2 cM and position of markers are indicated by an oval.

Figure 6.

Significant marker-by-gene linkages plotted as (x, y) coordinates with the x-axis representing the genome location of the marker and the y-axis representing the genome location of the linked gene. Each dot represents a single gene plotted against its best controlling marker. Significant linkages at two thresholds (see text) are shown. Solid dots represent significant linkages at a threshold associated with a FDR of 2.3%, and shaded dots represent significant linkages at a less stringent threshold associated with a FDR of 10.2%. Chromosome endpoints are indicated by dashed lines.