Main

Chaudhury suggests that the information required to restore correct genetic sequences in hth mutant plants could be stored in short stretches of nucleotide sequence within the genome1. Although the sequences required for restoration are indeed present in the genome, the length of similarity seen in the ‘reverting sequences’ identified by Chaudhury is barely greater than would be expected from random chance. An appropriate control for his in silico experiment would be to establish how many similar sequences (13–18 nucleotides in length, with a single nucleotide mismatch relative to the sequence in the parent plant) are present in the genome that could likewise introduce silent nucleotide substitutions into the hth gene under the same conditions.

Should there be a significant number of these sequences (and given that no such silent mutations occur in the corrected alleles3), then an explanation is needed for why some of them are used for correction whereas the majority are not. Even if silent mutation events occur independently of the reversion of the hth mutation, we should still detect them in our small sample of sequenced reverted alleles owing to the relatively large number of possible silent mutations.

Chaudhury also suggests that the increased permeability of mutant hth female gametophytes could allow DNA or RNA from the degenerating non-functional megaspores to enter the functional hth megaspore and then be archived and carried forward to allow gene conversion in the next generation. This may be a possibility, but our previous work examined only changes in the permeability of the extracellular cuticle covering the outside of the epidermal cell layer4. We have no data concerning increased cellular permeability in hth mutants. The fact that we see no obvious drop in the rate of reversion over several generations3 is inconsistent with Chaudhury's suggestion that this could be a second mechanism to bolster the rate of reversion for only a single generation.

Ray2 claims that our results could be explained by the stable inheritance of supernumerary chromosomal fragments that are archived in a way that makes them inaccessible to DNA hybridization and the polymerase chain reaction. These fragments might also be restricted to meristematic cells and therefore be present in such low concentrations that they are undetectable in a conventional experiment. This is an interesting possibility that is consistent with our observations, but it postulates a novel system of segregation to restrict the chromosome fragments to what would constitute a hitherto undetected germ line in plants. Considering also Ray's explanation for the doubled rate of conversion in embryos, we note that it would be necessary for all conversion events to take place in the generative cell and to fail to be corrected by mismatch repair.

In summary, we agree with Ray that there is little direct evidence to support any given molecular identity for the cryptic templates that allow genetic restoration in hth mutant plants. We proposed that the templates might be a replicating form of RNA, but the data are also consistent with a form of DNA that is segregated into a limited number of cells in the plant or that is not readily detectable by conventional molecular techniques. This sequence archive (whether DNA or RNA) would therefore require the same basic properties as those we proposed3: it would need to be replicated, transmitted with high fidelity over several generations, and retain the ability to restore nuclear DNA sequences.