Summary In order to identify new factors involved in Em (a class I Late Embryogenesis Abundant protein) gene expression, Arabidopsis mutants with an altered expression of an Em promoter GUS fusion construct and a modified accumulation of Em transcripts and proteins were isolated. Germination tests on ABA showed that the most affected mutant had a weak abi phenotype. Complementation tests further revealed this mutant to be a new abi5 allele, consequently named abi5–5. In addition to reducing the final level of Em transcripts in the dry seed, the abi5–5 mutation causes a delay in the accumulation of AtEm1 during seed development. An additional characteristic of the abi5–5 mutant, is the ability of its seeds to germinate at high concentrations of salt and mannitol. The abi5–5 mutation was characterized at the molecular level and was shown to result from a two base pair deletion in the coding sequence of the ABI 5 gene. The wild type and mutant recombinant proteins were produced in E. coli and were assayed for DNA‐binding activity on their target promoters by electrophoretic mobility shift assay (EMSA). The ABI5 recombinant protein binds the ABRE sequence in the AtEm6 promoter as shown by Dnase footprinting. Among the ABRE‐type sequences selected on both Em promoters, the G‐box type AGACACGTGGCATGT element of the AtEm6 promoter shows the strongest binding by EMSA quantification.

We analyzed the Arabidopsis thaliana genome sequence to detect Late Embryogenesis Abundant (LEA) protein genes, using as reference sequences proteins related to LEAs previously described in cotton or which present similar characteristics. We selected 50 genes representing nine groups. Most of the encoded predicted proteins are small and contain repeated domains that are often specific to a unique LEA group. Comparison of these domains indicates that proteins with classical group 5 motifs are related to group 3 proteins and also gives information on the possible history of these repetitions. Chromosomal gene locations reveal that several LEA genes result from whole genome duplications (WGD) and that 14 are organized in direct tandem repeats. Expression of 45 of these genes was tested in different plant organs, as well as in response to ABA and in mutants (such as abi3, abi5, lec2 and fus3) altered in their response to ABA or in seed maturation. The results demonstrate that several so-called LEA genes are expressed in vegetative tissues in the absence of any abiotic stress, that LEA genes from the same group do not present identical expression profile and, finally, that regulation of LEA genes with apparently similar expression patterns does not systematically involve the same regulatory pathway.