INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In mammals, methylation of CpG islands correlates with loss of gene expression (2). In plants, hypermethylation accompanies gene silencing (52), while silenced genes which recover expression in mutants unable to maintain silencing lose methylation (32). In the fungus Ascobolus immersus, repeated genes are methylated and silenced by a process named MIP (methylation induced premeiotically) that takes place during sexual reproduction (43). In Neurospora crassa, a related process, named RIP (repeat-induced point mutation), leads to a concomitant hypermethylation and hypermutation of the DNA repeats (47). In the latter situation, methylation may spread to an unmutated neighboring gene which becomes silenced as well, suggesting that methylation without mutation may be sufficient to initiate gene silencing (20). It has been shown that methyl-binding proteins recognizing methylated CpG's play an important role in the methylation-associated silencing in vertebrates (33). The methyl-CpG-binding protein MeCP2 can nucleate a complex containing deacetylases which remove acetyl moieties from lysine residues in the core histones H3 and H4 (41), resulting in a repressive nucleosomal array. In Neurospora, a connection has also been established between methylation, deacetylation, and gene silencing (48).

Methylated groups may directly prevent the binding of transcription factors to DNA (23). Methylation could also induce the formation of a silenced higher-order chromatin structure. Methylated DNA was found to be preferentially assembled in nuclease-resistant chromatin after transfection of mouse L cells (24). A ubiquitous component of chromatin might interact preferentially with methylated DNA, thereby stabilizing the higher order of chromatin structure and preventing gene expression.

Linker histones, which bind to linker DNA extending between nucleosomes, have been proposed as potential candidates for playing that role. Linker histones, such as H1, are known to seal nucleosomes, therefore stabilizing a higher order of chromatin structure (54). Histone H1 is abundant in nuclease-resistant, inactive chromatin (57), and it inhibits in vitro transcription (7, 59). Metazoan linker histones have a three-domain structure, with a central globular domain flanked by N- and C-terminal tails rich in basic residues (18). The amino acid sequence of the globular domain is the most conserved region. The basic C-terminal tail is rich in lysine, serine, proline, and alanine and is likely to be involved in the interaction with linker DNA, neutralizing its charge and facilitating chromatin condensation (1). In animals, linker histones show extensive diversification. Various subtypes display different DNA and chromatin-condensing properties in vitro (25). In addition, they exhibit highly regulated patterns of expression during development and differentiation (26).

Several studies aimed at investigating the possibility of a preferential binding of linker histones to methylated DNA have been performed, but the overall results remain inconclusive. In the mouse, 5-methylcytosine was reported to be preferentially located in nucleosomes that contain histone H1 (3). A chicken H1-like protein, MDBP-2, was reported to selectively bind methylated DNA both in vivo and in vitro (21). Further in vitro studies led to contradictory results. While H1 was reported by McArthur and Thomas (31) to bind preferentially to methylated DNA, Campoy et al. (5) and Nightingale and Wolffe (35) concluded that binding of H1 was indifferent to methylation in chromatin reconstitution experiments. It is difficult to make decisive conclusions from these in vitro studies, since factors playing an important role in the assembly of cellular chromatin may be missing in these assays.

A major contribution to understanding the structural and functional roles of linker histones in vivo came from experiments with Xenopus laevis. Extracts from Xenopus eggs depleted of histone B4, the only linker histone present in these eggs, retained the capacity to assemble chromatin from sperm nuclei, to initiate replication, and to condense their chromosomes (36). This finding indicates that linker histones facilitate the in vitro folding of nucleosomal arrays but are not required for chromatin and chromosome assembly. In somatic cells of Xenopus, histone H1 was shown to function as a developmentally regulated gene repressor acting specifically on the set of embryonic 5S RNA genes (4, 22, 46) and mesoderm-specific genes (53). In the case of 5S RNA genes, molecular studies indicated that the repressive effect of histone H1 is related to differential nucleosome positioning (38, 49).

In the unicellular eukaryotes Saccharomyces cerevisiae and Tetrahymena thermophila, putative linker histone genes encoding unusual products have been characterized. The S. cerevisiae candidate H1 histone contains two globular domains (27, 56), while that from Tetrahymena lacks the globular domain (61). In contrast, Ramón et al. recently characterized an H1 gene encoding a canonic linker histone in the filamentous fungus Aspergillus nidulans (40). S. cerevisiae, Tetrahymena, and A. nidulans cells lacking linker histones are viable and display normal growth (37, 40, 61). In A. nidulans, the nucleosomal organization of a number of promoters was shown to be identical in a wild-type strain and in a strain harboring a complete deletion of the H1 gene (40). Knocking out the S. cerevisiae linker histone gene had little effect on gene expression (37, 56). In particular, genes silenced as a consequence of their telomeric location were not activated in mutants devoid of histone H1. Deletion of the H1 gene expressed in the macronucleus of Tetrahymena did not affect transcription, except for a small subset of genes that were either activated or repressed (50). This again suggests that linker histones do not play a general role in gene repression and gene silencing but can occasionally interact with some specific gene targets to modulate their expression. However, these data provide no information on a possible role of linker histone in methylation-associated gene silencing, since S. cerevisiae, Tetrahymena, and A. nidulans do not display cytosine methylation.

The filamentous fungus Ascobolus immersus represents an attractive, well-characterized experimental system with which to test in vivo by a genetic approach the possible interaction between linker histones, methylated DNA, and gene silencing. This organism displays DNA methylation, and MIP provides a convenient tool to methylate and silence at will endogenous genes (10, 13, 43). The cloning and characterization of the H1-like gene from Ascobolus, henceforth named H1, allowed us to inactivate the expression of this gene and to construct strains lacking Ascobolus histone H1. We showed that this histone is not required for methylation-associated gene silencing and protects methylated and unmethylated chromatin equally well against micrococcal nuclease (MNase) digestion. Its loss results in three clear-cut phenotypes: hypermethylation, increased accessibility of MNase to chromatin, and reduced life span.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Transformation procedures, genetic analyses, and media. Standard genetic techniques, transformation procedures, and media were as described elsewhere (42).

Manipulation of DNA and methylation analysis. Most experimental procedures were as described previously (13). Other standard techniques were as in reference 45. PCR amplifications were performed under standard conditions (43). Cytosine methylation was analyzed by Southern hybridization, using the isoschizomers Sau3AI and NdeII, which are sensitive and insensitive, respectively, to C methylation. Methylation status was deduced from the replacement of the expected hybridizing Sau3AI fragments by larger fragments.

Cloning and characterization of the Ascobolus H1 gene. Ascobolus genomic DNA digests were probed in Southern hybridization with the complete open reading frame (ORF) of the A. nidulans H1 gene (40), which was kindly provided by C. Scazzocchio and colleagues. Hybridization and washings were performed at 53°C. To clone the hybridizing 1.7-kb HindIII-PstI fragment, a size-fractionated (1.5- to 1.9-kb) HindIII/PstI digest of the DNA from strain RN42 was subcloned into the HindIII-PstI-digested pBluescript KS vector (Stratagene). Clones were screened by colony hybridization. For reverse transcription-PCR (RT-PCR) experiments, total RNA was purified from mycelium by using the TRIzol reagent (GIBCO/BRL) and reverse transcribed, and PCR amplification of the cDNA was performed with primers H1cDNA1 and H1cDNA2, corresponding to positions 24 to 43 and 908 to 889 (Fig. 1), respectively. The PCR product obtained was sequenced with the same primers.

View larger version (51K): [in this window] [in a new window]   FIG. 1.   Primary structure of histone H1 from Ascobolus. (A) Nucleotide and derived amino acid sequences of the H1 gene. The entire 1.7-kb HindIII-PstI fragment is shown. The two introns are underlined. (B) Comparison of the amino acid sequence of the globular domain of Ascobolus H1 (Asc) with the globular domains of H1 from A. nidulans (Asp) and S. cerevisae (Sac; the first globular domain is shown) and of the H1 consensus sequence resulting from the comparison of 30 animal H1 sequences (58). Boxes indicate the regions identified as  helix and  sheet in the H5 crystal structure (39). Asterisks indicate the conserved amino acids. (C) Tripartite organization of the Ascobolus H1 protein. The box indicates the globular domain flanked by the N- and C-terminal tails. Positions of basic (+) and acidic () residues are indicated.

Methylation by MIP of the endogenous H1 gene. An ectopic duplication of the H1 gene was created in the wild-type strain FB14 by cointegrative transformation using plasmids pH1 and pMP6. Plasmid pH1 carried the 1.7-kb HindIII-PstI fragment encompassing the Ascobolus H1 gene. Plasmid pMP6 carried the hph (hygromycin resistance [Hyg^r]) gene, which was used as a selectable marker (29). Two Hyg^r transformants, T21 and T32, that had integrated a single full-length copy of the HindIII-PstI H1 fragment (identified by Southern hybridization) were selected and crossed with an appropriate tester strain in order to trigger MIP and methylation of the H1 duplication. In the progeny, strains that had segregated away the transgenic element of the duplication through meiotic segregation (Hyg^s strains) and had thus inherited only the resident H1 gene were isolated. Methylation of H1 was checked by Southern hybridization.

Construction of Ascobolus strains. To analyze the effect of H1 silencing on preexisting methylation, strains harboring the methylated and silenced allele of the H1 gene (H1^m) were crossed with strain FC72 in which the 5.6-kb HindIII fragment carrying the resident met2 gene had been previously methylated and silenced by MIP (creating the met2^m gene) (43). Methylation of H1 in the Met⁺ and Met strains obtained was determined by Southern hybridization, allowing identification of the four possible genotypes, H1⁺,met2⁺; H1⁺,met2^m; H1^m,met2⁺; and H1^m,met2^m.

Mycelial growth rate analysis. Ascospores from a cross between the H1-duplicated strain (H1⁺, [H1⁺-hph⁺]) and wild type (H1⁺) were germinated and screened for resistance or sensitivity to hygromycin. Only Hyg^s strains (that had segregated away the transgene and had thus inherited the resident H1 gene only) were further analyzed. H1^m strains were distinguished from H1⁺ strains by Southern hybridization analysis. For that purpose, a small fraction of the mycelium growing on minimal medium obtained 1 day after germination was used to inoculate a liquid culture in order to obtain mycelium for DNA extraction. In the meantime, cultures on minimal medium plates were allowed to grow, and the growth rate was measured every day. On the third day following germination, a small piece of agar cut in front of the growing mycelium was taken and transferred onto a plate containing fresh minimal medium. This procedure was repeated every 3 days.

Nucleus isolation and histone purification. The mycelium obtained from a 3-day culture in liquid medium was harvested by filtration, pressed dry, frozen in liquid nitrogen, and ground to a powder. Nucleus isolation was carried out at 4°C. Powdered mycelium (5 g) was introduced into a Potter homogenizer containing 20 ml of buffer A (1 M sorbitol, 7% Ficoll, 20% glycerol, 5 mM EGTA, 5 mM EDTA, 50 mM Tris-HCl [pH 7.5]). Homogenization was repeated seven times. The material obtained was transferred into a beaker, and 40 ml of buffer B (10% glycerol, 5 mM EGTA, 25 mM Tris-HCl [pH 7.5]) was slowly added and mixed. The mixed material was distributed into four 30-ml tubes, each containing 10 ml of a mix of buffers A and B (1/1.7 vol/vol) by pouring without disturbing the interface. After centrifugation at 4,300 rpm in a Sorvall HB-4 rotor for 7 min, the upper 15 ml of each tube was transferred to a new 30-ml tube containing 3.5 ml of buffer C (1 M sucrose, 10% glycerol, 25 mM Tris-HCl [pH 7.5]) without disturbing the interface. After centrifugation at 9,000 rpm in a Sorvall HB-4 rotor for 20 min, the supernatant composed of two aqueous phases was discarded, and the pelleted nuclei were resuspended in 1 ml of buffer D (0.25 M sucrose, 60 mM KCl, 15 mM NaCl, 15 mM Tris-HCl [pH 7.5]) and immediately used or frozen at 80°C.

Nucleosomal repetition and chromatin analysis. Protoplasts from the different strains were prepared as described elsewhere (9). For the nucleosomal repetition analysis, 2 × 10⁷ protoplasts were resuspended in 250 µl of permeabilization buffer (300 mM sucrose, 0.2% NP-40, 60 mM KCl, 15 mM NaCl, 5 mM MgCl₂, 5 mM CaCl₂, 15 mM Tris-HCl [pH 7.5]), and increasing amounts (0.75, 1.5, 3.0, 6.0, and 12 U for the H1^m strain; 2.25, 4.5, 9.0, 18, and 36 U for the wild-type strain) of freshly added MNase (Boehringer Mannheim) were added. Samples were incubated 3 min at 25°C, and reactions were stopped by the addition of 250 µl of stop buffer (50 mM Tris-HCl pH 7.5, 20 mM EDTA, 1% SDS). After extraction, the DNA was size separated on a 1.5% agarose gel further stained with ethidium bromide.

Nucleotide sequence accession number. The GenBank accession number for the sequence reported in this paper is AF190622.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of the histone H1 gene from Ascobolus. Several restriction enzyme digests of Ascobolus DNA were probed in low-stringency conditions with the coding sequence of the H1 gene from A. nidulans (40). A faint hybridization band was found with almost every DNA digest (data not shown). The hybridizing 1.7-kb HindIII-PstI fragment was cloned. Its sequencing revealed the presence of a discontinuous ORF, split in three putative exons separated by two introns of 72 bp (positions 227 to 298) and 62 bp (positions 487 to 548), respectively (Fig. 1A). The presence and location of the two introns was confirmed by RT-PCR amplification followed by sequencing of the PCR product. The putative protein is 213 amino acids long, with a calculated molecular mass of 21.88 kDa.

Histone H1 is dispensable for gene silencing. If the Ascobolus H1 protein were required for methylation-associated gene silencing, it would be impossible to silence the H1 gene via MIP. To test this prediction, an ectopic duplication of the HindIII-PstI fragment containing the H1 gene was created via integrative transformation of wild-type strain FB14. Two independent transformants harboring the duplication were used in crosses in order to trigger MIP and target methylation at the native H1 gene. Progeny strains that had segregated away the transgene through meiotic segregation were used for further analyses. The methylation status of the native H1 gene was analyzed by Southern hybridization using restriction enzymes sensitive to cytosine methylation. Methylation was found in 20% of the strains (Fig. 2A). We then determined whether the methylated H1 gene was silenced, as expected for genes that have undergone MIP. Total histone proteins were extracted from nuclei of two strains either with or without the native H1 gene methylated. The complete and specific disappearance of histone H1 in H1-methylated strains indicated that the H1 gene was completely silenced (Fig. 2B). Accordingly, when the protocol for extracting only histone H1 from the H1-methylated strains was used, no protein was detected by SDS-PAGE (data not shown).

View larger version (47K): [in this window] [in a new window]   FIG. 2.   Methylation and silencing of the H1 gene. (A) Southern analysis of methylation in four strains derived from the H1-duplicated strain and harboring only the resident H1 copy. Sau3AI DNA digests were probed with the cloned 1.7-kb HindIII-PstI fragment carrying the H1 gene (this work). Two strains (lanes 1 and 4) harbored the unmethylated allele, as revealed by the presence of the 1,184- and 728-bp expected fragments. The two other strains (lanes 2 and 3) harbored the methylated allele, as revealed by the presence of larger fragments. (B) SDS-PAGE of H2SO4-soluble proteins extracted from nuclei of the same strains as in panel A. The bands corresponding to H1 and to the core histones are indicated.

The lack of histone H1 results in hypermethylation. DNA from H1-silenced strains was more resistant to digestion with methylation-sensitive enzymes than the wild type, as revealed by the presence of a large amount of uncut DNA in ethidium bromide-stained agarose gels (Fig. 3A). We checked that this was not the result of an incomplete digestion by probing the digests with the unmethylated met2 gene from Ascobolus (Fig. 3D, lanes 1 to 4).

View larger version (80K): [in this window] [in a new window]   FIG. 3.   Hypermethylation in strains lacking histone H1. (A) Ethidium bromide staining of the gel used for Southern analyses of the Sau3AI DNA digests shown in panels B to D. As determined from panel B, strains analyzed in lanes 1, 3, 5, and 7 harbored the native unmethylated H1 allele, and strains analyzed in lanes 2, 4, 6, and 8 harbored the methylated allele. The gel was probed with H1 (B), Mars3 (C), and met2 (D). The H1 probe was as in Fig. 2. The Mars3 probe corresponded to the EcoRI fragment carrying Mars3 in plasmid pCG57 (16). The met2 probe was a PCR product corresponding to the 2.9-kb HincII-BglII fragment carrying the met2 gene (14). Strains 5 to 8 had inherited a silenced allele of met2, while strains 1 to 4 expressed this gene. The largest met2 Sau3AI fragments from strains 5 and 7 are hardly visible on the autoradiograph presented in panel D.

The methylation-associated chromatin modification does not require histone H1. The above results show that H1 is dispensable for both gene silencing and methylation and that its loss results in DNA hypermethylation. This led us to ask whether this loss would affect the chromatin changes associated with methylation. In another study (J. L. Barra, G. Grégoire, G. Almouzni, J.-L. Rossignol, and G. Faugeron, unpublished data), we had shown, by analyzing MNase-controlled digests of chromatin, that methylation in the Ascobolus met2 gene was associated with a change in chromatin that was confined to the methylated portions of the gene. Here we compared, in the same way, the chromatin configuration of the unmethylated and methylated met2 gene in H1-silenced and wild-type strains (Fig. 4). The loss of H1 did not lead to any change in the MNase banding patterns. Notably, the chromatin changes associated with methylation were similar in the two types of strains, which indicates that the chromatin changes that accompany methylation do not depend on H1. As expected, an analysis of the overall genomic nucleosomal pattern showed that the chromatin was more accessible to MNase in strains lacking H1. Indeed, three to four times less enzyme was sufficient to produce the same nucleosomal patterns as in the wild type (Fig. 5). Although the nucleosomal patterns were similar, oligonucleosome bands were much less diffuse in the absence of H1 than in its presence (Fig. 5). The size heterogeneity exhibited by wild-type oligonucleosomes may result in part from the fact that some of the nucleosomes contain H1 and others do not. In the met2 region also, the lack of H1 made the chromatin more accessible to MNase. Indeed, equal amounts of MNase gave higher levels of digestion in strains lacking H1 than in the wild-type control (Fig. 4A). Interestingly, this happened independently of the methylation status of met2, which indicates that in Ascobolus histone H1 interacts similarly with methylated and unmethylated DNA. These results indicate that there is no preferential binding of histone H1 to methylated DNA in Ascobolus. Furthermore, the hypermethylation phenotype and the increased accessibility of MNase to chromatin in strains devoid of histone H1 indicate that this protein is a chromatin constituent, as expected for linker histones.

View larger version (67K): [in this window] [in a new window]   FIG. 4.   Comparison of the methylation-associated chromatin modifications in H1-silenced and wild-type strains. (A) MNase analysis of the chromatin of the met2 gene was performed in two strains harboring the unmethylated allele of met2 (met2) and two strains harboring the methylated allele of this gene (met2m). One strain of each type harbored the silenced allele of H1 (H1m); the other harbored the wild-type allele (H1). Protoplasts were incubated with increasing amounts of MNase and subjected to indirect end-labeling analysis. Samples were loaded on the same gel. The long vertical box at the left represents the met2 gene, with the transcription start site (arrow), ORF (dark gray box), position of the EcoRV site (EV), and size markers (in kilobases) indicated. The black vertical box indicates the probe used for hybridization. A to H indicate the eight major bands obtained when met2 was unmethylated (met2,H1; met2,H1m). Band A corresponds to the EcoRV fragment carrying the met2 gene. The white vertical box indicates the methylated region of met2. Black dots indicate the positions of bands that changed when met2 was methylated (metm,H1; met2m,H1m). (B) Ethidium bromide staining of the gel used for hybridization shown in panel A.

View larger version (94K): [in this window] [in a new window]   FIG. 5.   Nucleosome repeat ladder obtained after chromatin digestion with MNase of one strain harboring the silenced allele of H1 (H1m) and one strain harboring the wild-type allele (H1). In the central lane is the size marker ladder (100-bp ladder; GIBCO BRL). The relative amounts of MNase used are indicated above the lanes.

Vegetative and sexual phenotypes of strains lacking histone H1: essential role of H1 for normal life span. The loss of H1 did not confer any noticeable phenotype to the mutant strains either during early vegetative life or during sexual reproduction. Spore germination was not affected, and mycelial growth occurred at the normal rate. In contrast to Tetrahymena, where histone H1 knockout cells display enlarged 4',6-diamidino-2-phenylindole (DAPI)-stained nuclei (51), we did not detect any difference in the sizes of the nuclei from either protoplasts (Fig. 6) or mycelium in strains devoid of H1 or wild-type strains after DAPI staining.

View larger version (63K): [in this window] [in a new window]   FIG. 6.   Comparison of DAPI-stained nuclei in protoplasts from the wild-type strain (H1) and strains lacking H1 (H1m).

View larger version (25K): [in this window] [in a new window]   FIG. 7.   Comparison of growth rates of six strains harboring the silenced copy of the H1 gene (H1m) and six wild-type strains (H1).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

By using MIP as a tool, we constructed Ascobolus strains in which the native H1 gene was methylated and totally silenced, as shown by the complete absence of histone H1. This indicates that in Ascobolus mycelium, the presence of H1 depends on the expression of a single copy of the H1 gene. In this respect, Ascobolus behaves like other lower eukaryotes. We cannot exclude, however, the existence of one or more H1 histone variants in the sexual reproduction cycle of Ascobolus, as exists in animals in developmentally regulated systems.

The loss of H1 results in a general increase in MNase accessibility to chromatin. This effect indicates that the Ascobolus H1 protein is associated with a large fraction of the nuclear chromatin and behaves in that respect as expected for a linker histone. This conclusion is reinforced by the fact that a considerable fraction of the methylated genome is protected by H1 against hypermethylation.

Our study provides definitive evidence that at least in Ascobolus, histone H1 does not play a role in vivo in mediating the effects of DNA methylation on chromatin structure and gene expression. The main conclusion of this study is that histone H1 does not mediate methylation-associated gene silencing in Ascobolus, since strains lacking H1 can still undergo MIP and gene silencing. This first point is in line with in vitro observations made by Nan et al. (33) suggesting that histone H1 may not be involved in gene repression mediated by methylation. These authors showed that when accessing its binding sites, rat MeCP2, which acts as a transcriptional repressor, can displace a large fraction of histone H1 from methylated chromatin. Our study of the met2 region suggests three other conclusions. (i) H1 does not play a role in nucleosomal positioning, as indicated by the fact that the loss of H1 did not lead to any detectable change in the nucleosomal pattern over the met2 region. (ii) The chromatin changes associated with methylation are not dependent on histone H1. We showed that Ascobolus strains lacking histone H1 retained the chromatin modifications associated with the methylated met2 region. (iii) H1 is present in both methylated and nonmethylated chromatin, as indicated by the observation that similar increases in chromatin accessibility were observed in the different strains lacking H1, independently of the methylation state. This finding is consistent with observations made in vitro showing that H1 does not bind preferentially to methylated DNA (5, 35).

Histone H1 is required for the normal vegetative life span of Ascobolus. Strains devoid of H1 suddenly stopped growing between 6 and 13 days after germination, i.e., after more than 50 division cycles. The short-life-span phenotype was not observed in Tetrahymena and S. cerevisiae (37, 61), which, unlike Ascobolus, do not display a metazoan-like histone H1. Such a phenotype also was not found in work performed with A. nidulans, although in this fungus H1 exhibits a metazoan-like tripartite structure (40). The effect found in Ascobolus may be accounted for by specific chromatin changes resulting from the loss of H1 that could repress genes essential to growth. Indeed, in Tetrahymena, although the loss of histone H1 has no detectable effect on viability and growth, it can result in the repression of some genes (50). The delay observed in Ascobolus before the arrest of growth suggests that the loss of H1 triggers the progressive accumulation of events that would lead eventually to the inability of the nuclei to divide further. The nature of these events is unknown. In the first hypothesis, one or several control genes ensuring the fidelity of the information flow, such as genes encoding chaperones, could be repressed concomitantly with the loss of H1. This would create a cascade of events that could indirectly lead to the arrest of growth. In a second hypothesis, a gene(s) essential to growth could be progressively repressed. Since strains lacking H1 display hypermethylation, methylation per se might be directly involved by spreading progressively toward essential genes, thus repressing them. This seems unlikely, however, since strains lacking H1 do not experience de novo methylation of nonmethylated genes, and neither methylation nor chromatin changes spread from hypermethylated regions to flanking sequences. Independently of methylation, the loss of H1 could be responsible for the progressive occurrence of chromatin changes in some essential genes. An attractive hypothesis comes from the data of Wolffe and colleagues showing that in vitro, the nonhistone linker protein HMG1 can replace histone H1 in chromatin (34, 55). Indeed, HMG1 was shown, also in vitro, to reversibly inhibit transcription by RNA polymerase II by interacting with the TATA-binding protein, suggesting that HMG1 is likely to affect the basal transcriptional machinery (11). Thus, it may be that the chromatin of strains lacking H1 becomes progressively enriched in HMG1, during early mycelial growth, reaching a stage at which the genome, or some critical genes, become silenced. Interestingly, strains lacking H1 undergo normal sexual reproduction if they are intercrossed before the arrest of growth, and the progeny displays the same 6- to 13-day delay before the arrest of growth. To account for this resetting of the life length expectancy through sexual reproduction, we must assume that expression of the critical gene(s) during this phase of the life cycle is not affected by the postulated changes in the chromatin control of gene expression.

Interestingly, the loss of histone H1 results in hypermethylation of chromosomal regions previously methylated by MIP. Since close to 100% of the C's belonging to CpG dinucleotides are methylated by MIP while other C's are less densely methylated (15), hypermethylation must mainly concern C's not belonging to CpG sites. Previous results suggest that two distinct mechanisms acting at CpG and non-CpG sites underlie maintenance of methylation in Ascobolus (12, 15). The first mechanism would consist in the methylation of CpG sites according to the classical maintenance model (19, 44). The second mechanism would result in the methylation of neighboring C's. It is known that in mammalian cells, in which methylation occurs almost exclusively at CpG sites the methyltransferase localizes to the chromosomal replication complex (28) and maintenance methylation takes place less than 1 min after replication (17). By contrast, chromatin assembly takes 10 to 20 min (8), histone deposition occurs in stages, and it is not until a complete histone octamer is assembled with DNA that histone H1 is stably sequestered (60). If this scenario also holds for Ascobolus, the following model could explain our observations. Methylation at CpG sites would proceed in a first short step associated with DNA replication. The second mechanism directing methylation at non-CpG sites would occur secondarily, during chromatin assembly, and would be hindered by H1 interacting with DNA. In absence of H1, secondary methylation could not be blocked, resulting in hypermethylation. In keeping with this hypothesis is the observation that the variant human histone H1e can inhibit in vitro enzymatic DNA methylation (62).