Nuclear gene-induced variegation mutants provide a powerful system to dissect interactions between the genetic systems of the nucleus-cytoplasm, the chloroplast, and the mitochondrion. The immutans (im) variegation mutation of Arabidopsis is nuclear and recessive and results in the production of green- and white-sectored leaves. The green sectors contain cells with normal chloroplasts, whereas the white sectors are heteroplastidic and contain cells with abnormal, pigment-deficient plastids as well as some normal chloroplasts. White sector formation can be promoted by enhanced light intensities, but sectoring becomes irreversible early in leaf development. The white sectors accumulate the carotenoid precursor phytoene. We have positionally cloned IM and found that the gene encodes a 40.5-kD protein with sequence motifs characteristic of alternative oxidase, a mitochondrial protein that functions as a terminal oxidase in the respiratory chains of all plants. However, phylogenetic analyses revealed that the IM protein is only distantly related to these other alternative oxidases, suggesting that IM is a novel member of this protein class. We sequenced three alleles of im, and all are predicted to be null. Our data suggest a model of variegation in which the IM protein functions early in chloroplast biogenesis as a component of a redox chain responsible for phytoene desaturation but that a redundant electron transfer function is capable of compensating for IM activity in some plastids and cells.

No abstract

Results from Unification Theory and type inference with coercions are combined to produce a new method for performing strictness analysis of functional programs. A formal deduction system is developed in which extended types are derivable for terms of the A-calculus. These extended types contain Boolean rings describing the reduction behaviour of terms. Algorithms implementing the method are described, as well as proofs of their correctness. The method is extended to deal with recursion, polymorphism, and constants.

A combinator-based parser is a parser constructed directly from a BNF grammar, using higher-order functions (combinators) to model the alternative and sequencing operations of BNF. This paper describes a method for constructing parser combinators that can be used to build efficient predictive parsers which accurately report the cause of parsing errors. The method uses parsers that return values (parse trees or error indications) decorated with one of four tags.

No abstract