A cytosine methyltransferase homologue is essential for repeat-induced point mutation in Neurospora crassa

Abstract

During sexual development, Neurospora crassa inactivates genes in duplicated DNA segments by a hypermutation process, repeat-induced point mutation (RIP). RIP introduces C:G to T:A transition mutations and creates targets for subsequent DNA methylation in vegetative tissue. The mechanism of RIP and its relationship to DNA methylation are not fully understood. Mutations in DIM-2, a DNA methyltransferase (DMT) responsible for all known cytosine methylation in Neurospora, does not prevent RIP. We used RIP to disrupt a second putative DMT gene in the Neurospora genome and tested mutants for defects in DNA methylation and RIP. No effect on DNA methylation was detected in the tissues that could be assayed, but the mutants showed recessive defects in RIP. Duplications of the am and mtr genes were completely stable in crosses homozygous for the mutated potential DMT gene, which we call rid (RIP defective). The same duplications were inactivated normally in heterozygous crosses. Disruption of the rid gene did not noticeably affect fertility, growth, or development. In contrast, crosses homozygous for a mutation in a related gene in Ascobolus immersus, masc1, reportedly fail to develop and heterozygous crosses reduce methylation induced premeiotically [Malagnac, F., Wendel, B., Goyon, C., Faugeron, G., Zickler, D., et al. (1997) Cell 91, 281-290]. We isolated homologues of rid from Neurospora tetrasperma and Neurospora intermedia to identify conserved regions. Homologues possess all motifs characteristic of eukaryotic DMTs and have large distinctive C- and N-terminal domains.

Figure 1

Structure of rid and the DMT domain of its predicted protein. (A) Segment map of N. crassa linkage group I (top line) and structure of rid (bottom line). The rid gene was mapped near the mating type (mat) locus. Genetic and physical maps were aligned by identifying leu-4, cys-5, and ser-3 (based on orthologs from yeast) and tef-1 (30). The positions of the centromere (Cen-I) and his-3 are shown. The rid coding region spans nt 259-2797 of the rid transcript (filled boxes). One intron (nt 17–177; open box) is present in the untranslated leader sequence. The triangle marks the position of an intron in Ascobolus masc1 and Aspergillus dmtA that is absent from Neurospora rid. A consensus TATA promoter element is indicated by the asterisk, and a putative polyadenylation signal is indicated by a diamond. The 286-aa DMT domain is indicated by a line below the transcript. Arrowheads represent primers used to amplify rid. Restriction sites (E = EcoRV and S = SpeI) used for targeting to his-3 are shown. The nonsense mutation identified in mutant rid alleles is indicated by the open square. (B) Alignment of the catalytic DMT domain of RID homologues from N. crassa (RID), A. fumigatus (DmtA), and A. immersus (Masc 1), DMT motifs are indicated by Roman numerals. Residues identical in all eukaryotic DMTs are marked by asterisks; positions that have conservative substitutions are indicated by dots.

Figure 2

Generation of rid mutants by RIP. Genomic DNA from host strain N623 (H), the rid his-3∷rid duplication strain N1963 (D) and strains bearing rid^RIP1 (N2248, 1) and rid^RIP4 (N2257, 2) alleles was digested with RsaI, transferred to nylon membranes, and probed with a 2.5-kb rid fragment to detect novel fragments resulting from RIP. Molecular weight size standards (kb) are shown on the right.

Figure 3

Disruption of rid does not affect DNA methylation in vegetative cells. Genomic DNA from rid⁺ (WT; N150) and rid^RIP1 (RIP1; N1977) strains was digested with Sau3AI (S) or DpnII (D), fractionated in 1% agarose, stained with ethidium bromide (EtBr), transferred to a nylon membrane, and probed sequentially for the indicated methylated chromosomal regions. Molecular weight size standards (kb) are shown on the left.

Figure 4

Structure of predicted RID proteins from N. crassa, N. intermedia, and N. tetrasperma. Conserved motifs in the DMT domain are shown as black boxes (Roman numerals). Conservative and nonconservative substitutions in the protein sequences of RID from N. intermedia and N. tetrasperma, relative to that of N. crassa, are indicated by short and long vertical ticks, respectively. The number of amino acids inserted or deleted is indicated within the inverted and upright triangles, respectively; small triangles indicate single insertions or deletions. The 8- and 9-aa deletions in the C-terminal domain of N. intermedia and N. tetrasperma RID remove one copy of a 24-bp identical direct repeat in the N. crassa DNA sequence.

Figure 5

Relationship between bona fide and putative DNA methyltransferases. Eukaryotic DMTs fall into four major classes: (i) de novo DMTs (DIM-2 and DNMT3) from N. crassa (Nc) and animals (Dr, D. rerio; Hs, H. sapiens); (ii) chromomethylases (CMT) from plants (At, A. thaliana; As, A. suecia; Aa, A. arenosa); (iii) maintenance DMTs (MET- and DNMT1-like) from plants (Nto, N. tabacum; Dc, D. carota; Ps, P. sativum, Zm, Z. mays), fungi (Ai, A. immersus Masc2), and animals (Xl, X. laevis; Gg, G. gallus; Mm, M. musculus; P/, P. lividus); and (iv) Masc1 homologues (Masc1, DmtA, RID) from fungi (En, E. nidulans; Af, A. fumigatus; Ng, N. galapagosensis; Nte, N. terricola; Nt, N. tetrasperma). Alignment of the catalytic domain of bona fide and putative DMTs was performed by clustalw analyses with default settings (39). The sequence of the bacterial DNA methyltransferase most closely related to RID, Neisseria gonorrhoeae VII methylase (M.NgoVII), was used as an outgroup. Proteins with known in vivo and/or in vitro DMT activity or phenotypes potentially associated with DNA methylation are shown in bold, whereas putative DMTs identified only by sequence homology are shown in plain type. For simplicity, we have not included the Dnmt2 group of putative DMTs (e.g., see ref. 33), because to date there are no indications that these proteins are involved in methylation.