Ascobolus immersus artificial gene repeats were shown previously to be subject premeiotically to both cytosine methylation and inactivation. We studied sexual progenies of strains harbouring two wild type copies of the endogenous met2 gene lying either in tandem array or at ectopic unlinked positions, by (i) investigating the methylation status, (ii) searching for mutations and (iii) analysing the inheritance of inactivation both in mitotic and sexual offspring. 100% of the ‘tandem’ progeny and 64% of the ‘ectopic’ progeny had methylated repeats and displayed gene inactivation. Similar methylation patterns involving all or most of the cytosine residues within the repeats were observed in both arrangements. The inactivated met2 copies were totally devoid of mutation, as deduced from: (i) extensive restriction site analysis and DNA sequencing; (ii) the finding that all the Met‐ derivatives tested reverted to prototrophy in selective conditions; and (iii) the finding that an inactivated copy of met2 stripped of its methylation through amplification in Escherichia coli regained activity when reintroduced in A.immersus. In the absence of selection, gene silencing and methylation were faithfully maintained through mitotic divisions and through five successive sexual cycles. Altogether, these data show the epimutational nature of this methylation induced premeiotically (MIP) process.

A gene encoding a protein that shows sequence similarity with the histone H1 family only was cloned in Ascobolus immersus. The deduced peptide sequence presents the characteristic three-domain structure of metazoan linker histones, with a central globular region, an N-terminal tail, and a long positively charged C-terminal tail. By constructing an artificial duplication of this gene, named H1, it was possible to methylate and silence it by the MIP (methylation induced premeiotically) process. This resulted in the complete loss of the Ascobolus H1 histone. Mutant strains lacking H1 displayed normal methylation-associated gene silencing, underwent MIP, and showed the same methylation-associated chromatin modifications as did wild-type strains. However, they displayed an increased accessibility of micrococcal nuclease to chromatin, whether DNA was methylated or not, and exhibited a hypermethylation of the methylated genome compartment. These features are taken to imply that Ascobolus H1 histone is a ubiquitous component of chromatin which plays no role in methylation-associated gene silencing. Mutant strains lacking histone H1 reproduced normally through sexual crosses and displayed normal early vegetative growth. However, between 6 and 13 days after germination, they abruptly and consistently stopped growing, indicating that Ascobolus H1 histone is necessary for long life span. This constitutes the first observation of a physiologically important phenotype associated with the loss of H1.

A large proportion of Ascobolus immersus transformants are highly unstable in crosses: the phenotype conferred by the transgene is not transmitted to the progeny, irrespective of the endogenous or foreign origin of the transgene. They all have integrated multiple transgene copies, clustered at a single chromosomal site or at tightly-linked sites. Clustered non-homologous integrations are always rearranged. Yet they never escape the "methylation induced premeiotically" (MIP) process. This always results in gene silencing, even when the transgene is partially repeated, accounting for the high instability of these transformants.

Repeated genes, artificially introduced in Ascobolus immersus by integrative transformation, are frequently inactivated during the sexual phase. Inactivation is observed in about 50% of meioses if duplicated genes are at ectopic chromosomal locations, and in 90% of meioses if genes are tandemly repeated. Inactivation is associated with extensive methylation of the cytosine residues of the duplicated sequences and is induced in the still haploid nuclei of the dikaryotic cell which will undergo karyogamy and subsequent meiosis. Only repeated sequences become methylated. This raises the intriguing question of how the premeiotic inactivation machinery is informed that a nucleus contains multiple copies of a gene. By using in crosses recombinant strains of A. immersus in which either one, two or three genetically independent copies of the exogenous amdS gene from Aspergillus nidulans had been introduced, we could follow the premeiotic inactivation of each one of the ectopic amdS copies. This led us to propose that a prerequisite for inactivation is a premeiotic pairing of repeated sequences and that each copy can undergo successive cycles of pairing. In fact, once methylated, a copy can pair with a still unmethylated copy, so that an uneven number of copies can be subject to inactivation.