We have identified and characterized the complete complement of genes encoding the core histones of Neurospora crassa. In addition to the previously identified pair of genes that encode histones H3 and H4 (hH3 and hH4-1), we identified a second histone H4 gene (hH4-2), a divergently transcribed pair of genes that encode H2A and H2B (hH2A and hH2B), a homolog of the F/Z family of H2A variants (hH2Az), a homolog of the H3 variant CSE4 from Saccharomyces cerevisiae (hH3v), and a highly diverged H4 variant (hH4v) not described in other species. The hH4-1 and hH4-2 genes, which are 96% identical in their coding regions and encode identical proteins, were inactivated independently. Strains with inactivating mutations in either gene were phenotypically wild type, in terms of growth rates and fertility, but the double mutants were inviable. As expected, we were unable to isolate null alleles of hH2A, hH2B, or hH3. The genomic arrangement of the histone and histone variant genes was determined. hH2Az and the hH3-hH4-1 gene pair are on LG IIR, with hH2Az centromere-proximal to hH3-hH4-1 and hH3 centromere-proximal to hH4-1. hH3v and hH4-2 are on LG IIIR with hH3v centromere-proximal to hH4-2. hH4v is on LG IVR and the hH2A-hH2B pair is located immediately right of the LG VII centromere, with hH2A centromere-proximal to hH2B. Except for the centromere-distal gene in the pairs, all of the histone genes are transcribed toward the centromere. Phylogenetic analysis of the N. crassa histone genes places them in the Euascomycota lineage. In contrast to the general case in eukaryotes, histone genes in euascomycetes are few in number and contain introns. This may be a reflection of the evolution of the RIP (repeat-induced point mutation) and MIP (methylation induced premeiotically) processes that detect sizable duplications and silence associated genes.
EUKARYOTES employ an elaborate system to package and organize their extensive genetic material. The first order of this packaging system involves the incorporation of DNA into nucleosomes, the fundamental units of chromatin. Each nucleosome consists of 146 bp of DNA wrapped around an octameric protein complex composed of two proteins of each of the core histones H2A, H2B, H3, and H4 (McGhee and Felsenfeld 1980). Structural studies of the nucleosome have revealed that histones contain two distinct structural domains, a globular domain and N- and C-terminal “tails” (Arentset al. 1991). The globular domains of the core histones, which are similar to each other, allow the histones to interact with one another and with the surrounding DNA to form the nucleosome (Lugeret al. 1997). The N- and C-terminal tails protrude from the nucleosome core. Gene expression can be regulated through post-translational modifications of the histone tails including acetylation, methylation, phosphorylation, and ADP ribosylation (Imhof and Becker 2001; Wanget al. 2001). The tails are known to play roles in some epigenetic processes, such as repression of the silent mating loci in yeast (Kayneet al. 1988). In Neurospora crassa, treatment with the histone deacetylase (HDAC) inhibitor, trichostatin A, results in selective loss of DNA methylation (Selker 1998). Studies in animal systems have found that methylated sequences can recruit HDACs via proteins that associate with methylated DNA (reviewed by Dobosy and Selker 2001). A recent study implicates methylation of the N-terminal tail of H3 in DNA methylation (Tamaru and Selker 2001). Finally, the suggestion has been made that histone acetylation may direct demethylation of associated DNA (Cervoni and Szyf 2001). To facilitate further studies on the role of histones in gene regulation and in epigenetic processes such as DNA methylation, we carried out a thorough characterization of the histone genes of N. crassa.
Available information suggests that Neurospora nucleosomes and histones are typical for eukaryotes (Noll 1976; Kornberg 1977). The full complement of histones has been purified and their amino acid profiles are similar to those of rabbit and pea histones (Goff 1976), consistent with the high conservation of histones generally (Thatcher and Gorovsky 1994). This conservation was confirmed in the three instances in which the sequence of a Neurospora histone was determined. A partial sequence of the H2B globular domain and C terminus showed near identity to H2B sequences from plants, animals, and fungi (Karpovaet al. 1986). Single genes encoding H3 and H4 were cloned on a contiguous stretch of genomic DNA and sequenced (Woudtet al. 1983). A comparison of the conceptual translation products of these genes with yeast H3 and H4 revealed 95 and 92% identity, respectively. The two genes are transcribed divergently and separated by ~2 kbp. Southern analysis of Neurospora genomic DNA using these histone genes as probes revealed no other putative H3 or H4 genes, leading the authors to conclude that this gene pair is the only source of H3 and H4 in Neurospora. This contrasts with the situation in most eukaryotes examined, which have multiple, sometimes divergent, histone genes (Maxsonet al. 1983; Old and Woodland 1984; Chabouteet al. 1993).
Whereas metazoan and plant genomes often contain tens or hundreds of genes that encode each histone, fungal genomes appear to have at most three genes for each histone (Walliset al. 1980; Choe et al. 1982, 1985; Smith and Andresson 1983; Matsumoto and Yanagida 1985; May and Morris 1987; Ehingeret al. 1990). The genomic sequence of Saccharomyces cerevisiae (Goffeauet al. 1996) includes two genes for each histone, organized into two H2A-H2B and two H3-H4 gene pairs, in agreement with previous studies (Walliset al. 1980; Choeet al. 1982; Smith and Andresson 1983). The two H3 genes and two H4 genes each encode identical proteins, but the two H2A and H2B genes each encode slightly different H2A and H2B proteins. In addition to S. cerevisiae, comprehensive surveys for histone genes have been carried out in two other fungi, Schizosaccharomyces pombe and Emericella (Aspergillus) nidulans. S. pombe has two H2A genes, each encoding slightly different proteins and a single H2B gene paired with one of the H2A genes (Choeet al. 1985). Three H3-H4 gene pairs have been identified in S. pombe (Matsumoto and Yanagida 1985). All three predicted H4 proteins are identical, but only two of the H3 genes encode identical H3 proteins; the third encodes a slightly different protein. E. nidulans has single genes encoding H2A, H2B, and H3 and two genes encoding slightly different H4 proteins (May and Morris 1987; Ehingeret al. 1990). Again, the H2A and H2B genes are physically paired, as are the H3 and H4 genes.
Given the variability in histone gene numbers in fungi, we were pleased to discover that N. crassa has a relatively simple set of histone genes, as described below. Molecular and genetic experiments were used to demonstrate that the genes we identified are responsible for producing all of the core histones of Neurospora.
MATERIALS AND METHODS
Informatics: Routine sequence analyses were carried out using the program MacVector (Oxford Molecular, Palo Alto, CA). Protein and nucleotide sequence database searches were performed using the BLAST program at the National Center for Biotechnology Information (Altschulet al. 1997). BLAST searches for histone gene ESTs were carried out with on-line databases from the Aspergillus nidulans and Neurospora crassa cDNA Sequencing Project (Roeet al. 2001b) and the New Mexico Neurospora Genome Project (http://www.unm.edu/~ngp/; Nelsonet al. 1997). BLAST searches for histone genes in Neurospora used the MIPS Neurospora crassa database (MNCDB; Schulteet al. 2001) and the second assembly version of the Whitehead Institute Neurospora Sequencing Project (NSP; WICGR 2001).
Manipulation of N. crassa: Standard techniques for culturing Neurospora (Davis and De Serres 1970) were followed, except that a modified crossing medium (Russoet al. 1985) was used. Microconidiation was carried out as described by Pandit and Maheshwari (1993). Transformations for gene replacement at his-3 were carried out by electroporation of conidia as previously described (Margolinet al. 1997) after linearization of pSH10, pJS94, and pJS95 with AseI, pSH25 with NdeI, or pSH14 with NotI. Transformation for gene replacement at hH4-2 was carried out similarly with pSH18 linearized with ApaI, selecting for inositol prototrophy. The genotypes and sources of Neurospora strains are listed in Table 1. Strains N1679 (FGSC no. 5888) and N1997 (FGSC no. 7267) were provided by the Fungal Genetics Stock Center (FGSC; University of Kansas Medical Center, Kansas City).
Southern analysis: DNA was purified from cultures grown to stationary phase in 5 ml liquid Vogel's medium with 1.5% sucrose, necessary supplements and hygromycin (200 μg/ml), as appropriate. Mycelial pads were pat dried and lyophilized and DNA was purified as described (Irelanet al. 1993). Restriction digests were carried out under the conditions recommended by the manufacturer (New England Biolabs, Beverly, MA). Typically, 0.6 μg of genomic DNA was digested overnight with at least 3 units of restriction endonuclease to ensure complete digestion. Digests were analyzed as previously described (Irelan and Selker 1997). Blots were exposed either to film (Kodak X-Omat Blue XB-1) or to a phosphorimager screen, which was read by a Storm 860 phosphorimager (Molecular Dynamics, Sunnyvale, CA) and visualized using ImageQuant software (version 1.11 for Macintosh; Molecular Dynamics). Complete digestion was verified by probing with DNA corresponding to a known unmethylated region of the genome (data not shown).
Restriction fragment length polymorphism mapping: Genomic DNA from either the large or the small standard set of progeny generated by Metzenberg et al. (1984) was digested with one of several restriction enzymes and analyzed by Southern hybridization as described above. Segregation data were compared with published data (Nelson and Perkins 2000).
Cosmid libraries: The Orbach-Sachs (O-S; Orbach 1994) and Vollmer-Yanofsky (V-Y; Vollmer and Yanofsky 1986) N. crassa genomic DNA cosmid libraries, provided by the FGSC, were screened and processed as previously described (Kouzminova and Selker 2001).
Plasmids: pBM60 and pBM61 have been described (Margolinet al. 1997). pOKE01 was a gift from J. Grotelueschen and R. Metzenberg. Plasmids were constructed by standard procedures (Sambrooket al. 1989). pJS94 was constructed by subcloning a 2.7-kb BamHI fragment of Neurospora genomic DNA from O-S cosmid G25:G12 into pBM61 digested with BamHI. The fragment includes the entire hH2A coding region, plus 1.4 and 0.7 kb of upstream and downstream sequences, respectively. pJS95 was constructed by subcloning a 2-kb NdeI-PstI fragment of Neurospora genomic DNA from O-S cosmid G25:G12 into pBM61 cut with SmaI. The fragment includes the entire hH2B coding region, plus 0.6 and 0.75 kb of upstream and downstream sequences, respectively.
pSH2 was identified by colony hybridization, using hH4-1 as a probe, in a sublibrary of Sau3AI fragments cloned from O-S cosmid G11:G5 into BamHI-digested pBluescript II SK(−). pSH2 contains the entire hH4-2 coding region, plus 40 and 550 bp of upstream and downstream sequences, respectively. pSH10 was created by subcloning a 2.2-kb BglII-SpeI fragment of Neurospora genomic DNA from a genomic DNA clone into pBM60 digested with BamHI and SpeI. The subcloned fragment includes the entire hH4-1 coding region, plus 1.0 and 0.9 kb of upstream and downstream sequences, respectively. pSH14 was constructed by subcloning a 2-kb EaeI-SpeI fragment from a PCR-generated fragment of genomic DNA into pBM60 digested with ApaI and SpeI. The subcloned fragment contains a 2-kb BglII-HincII fragment of Neurospora genomic DNA that contains the entire hH3 coding region, plus 1.2 kb and 370 bp of upstream and downstream sequences, respectively. pSH16 was constructed by subcloning a 9-kb NarI-SacI fragment of Neurospora genomic DNA containing hH4-2 from a genomic clone into pBluescript II SK(−) digested with HincI and SacI. pSH18 was created by replacing a 2.3-kb XbaI-Bsp106I fragment of pSH16, which contained all of the hH4-2 coding sequence, plus 890 and 720 bp of upstream and downstream sequences, respectively, with a 3.6-kb XbaI-BstBI fragment from pOKE01, which carries the inl gene. pSH18 retains 1.6 and 5.2 kb of upstream and downstream sequences, respectively, to facilitate homologous recombination at the hH4-2 locus. pSH25 was constructed by subcloning a 3.4-kb EaeI-SpeI fragment of Neurospora DNA from pSH16 into pBM60 digested with NotI and SpeI. The subcloned fragment contains the entire hH4-2 coding region, plus 1.3 and 1.45 kb of upstream and downstream flanking sequences, respectively.
Sequencing: The initial hH4-2 sequence was generated by sequencing pSH2 with standard T3 and T7 primers at the Oregon State University Sequencing Center. All other sequencing was carried out at the University of Oregon Sequencing Facility using custom primers. To minimize the risk of sequencing a mutation created during polymerase chain reaction (Ausubelet al. 1998), products from at least three reactions were pooled for sequencing.
Phylogenetic analyses: The source and accession number of sequences obtained from GenBank is available on request. The following expressed sequence tagged (EST) sequences were obtained from the Cryptococcus neoformans cDNA Sequencing Project (Roeet al. 2001a) and conceptually translated to obtain their putative histone products: H2B from a5f05cn; H3 from b4d04cn.r1; H4.1 from b1f06cn, b1f10j2, and a7c02j2; and H4.2 from b5g08cn. The following EST sequences were obtained from the Fusarium sporotrichioides cDNA Sequencing Project (Roeet al. 2001c) and conceptually translated to obtain their putative histone products: H2A from b2d06fs, j4d05fs.f1, and d3h05fs.f1; H2B from l1g06fs and l3f03fs; and H4.1 from o1c08fs.r1 and f1f06fs.r1. The following is a partial list of EST sequences for the core histone genes of N. crassa from the Aspergillus nidulans and Neurospora crassa cDNA Sequencing Project (Roeet al. 2001b). hH2A: d5c07ne, e9d02ne, and g8g08nm.r1; hH2B: a9c10ne, a6c09ne.f1, and arf06nm.r1; hH3: b7c11ne, f7b11nm, and a4a05np; hH4-1: b2g12ne, b7b12ne, and a4b01ne.f1; hH4-2: d8e05ne, h4b06nm, and a1b03ne.
Protein and nucleic acid sequences were imported into Biology Workbench (http://workbench.sdsc.edu/), where they were aligned with CLUSTALW (Felsenstein 1989; Thompsonet al. 1994) and manually edited. Aligned sequences were used to determine the identity of other fungal histones relative to Neurospora using MVIEW Multiple Alignment Display (Brownet al. 1998). Phylogenetic trees were constructed from the aligned sequences with CLUSTALX and drawn by Neighbor-Joining Plot (Felsenstein 1989).
Identification and sequence analysis of genes encoding the core histones and histone variants of N. crassa: To identify all of the hH3 and hH4 homologs in the N. crassa genome, we screened two genomic cosmid libraries with the previously identified hH3-hH4 gene pair (Woudtet al. 1983). Of the cosmids identified, only SV31:G9 from the V-Y cosmid library (Vollmer and Yanofsky 1986) was found to carry the hH3-hH4 gene pair. The other cosmids, G11:G5, G22:D6, and X18:H1 from the O-S cosmid library (Orbach 1994), were found to carry a second gene encoding H4, which we named hH4-2. The previously identified hH4 gene in the hH3-hH4 gene pair has been renamed hH4-1 (Perkinset al. 2001).
hH4-1 and hH4-2 encode identical proteins and are 96% identical (300/312) at the nucleotide level in their coding regions. Both genes contain two introns, at precisely conserved locations, but no similarity was found in the introns or in the 5′ or 3′ untranslated regions (UTRs). The intron lengths are also different. Introns 1 and 2 of hH4-1 are 69 and 68 bp, whereas the introns of hH4-2 are 316 and 65 bp, respectively (Figure 1).
To identify the genes encoding histones H2A and H2B, we first searched the two publicly available Neurospora EST databases (Nelsonet al. 1997; Roeet al. 2001b) by tBLASTn for corresponding cDNAs. We identified one set of ESTs for each and designated their corresponding genes hH2A and hH2B. Genomic copies of the genes were identified in O-S cosmid G25:G10, subcloned, and sequenced. We found that hH2A and hH2B lie 2.6 kb apart, are transcribed divergently, and are composed of three and four exons, respectively (Figure 1).
We failed to detect additional hH2A, hH2B, hH3, or hH4 genes in the N. crassa genome. Southern analysis of genomic DNA digested with a variety of restriction enzymes using hH2A, hH2B, hH3, hH4-1, and hH4-2 as probes detected only the bands expected for the known genes, as shown for hH2A (Figure 2). tBLASTn searches for genes encoding H2A, H2B, H3, and H4 in the EST databases (Nelsonet al. 1997; Roeet al. 2001b) detected numerous ESTs for hH2A, hH2B, hH3, hH4-1, and hH4-2, indicating that each of these genes is expressed (Figure 1). No other histone ESTs were found.
tBLASTn searches were also carried out using the databases of the two N. crassa genome sequencing projects, MNCDB (Schulteet al. 2001) and NSP (WICGR 2001). No new genes encoding core histones were found, but three histone variants were discovered. One was an H2A variant of the H2A F/Z family (Carret al. 1994), which we named hH2Az (Figure 3A). In the absence of ESTs for hH2Az, we used the high conservation within the hH2A F/Z family of variants to determine that the hH2Az coding region consists of three exons (Figure 1). We also found an H3 variant, which we named hH3v. It showed the greatest similarity to SpCENP-A (Figure 3B), a S. pombe homolog of CSE4, which is an essential gene of S. cerevisiae (Stoleret al. 1995). On the basis of similarity to other H3 variants, we suggest that the hH3v open reading frame is composed of two exons (Figure 1). The intron location in hH3v appears identical to the intron location in hH3. Without a cDNA sequence, we cannot be certain that hH3v does not have a second intron in the region of the expected N terminus, where there is low conservation among H3 variants (Figure 3B). We also identified a fragmented hH4 homolog, hH4v, which appears composed of six exons and five introns at positions different from those in hH4-1 and hH4-2 (Figure 1). The canonical methionine codon (ATG) of the potential hH4v translation product was replaced by a leucine codon (TTG) and its predicted sequence is only 62% identical to that of N. crassa H4 (Figure 3C). This is significantly lower than the >91% homology among H4 proteins across the fungal kingdom (Table 3), indicating that hH4v either produces a highly divergent H4 or is a pseudogene.
Genomic organization of the genes encoding the core histones and histone variants: The genomic location of the hH3-hH4-1 gene pair on the right arm of linkage group (LG) II was previously determined by restriction fragment length polymorphism (RFLP) mapping (Metzenberg and Grotelueschen 1987). The presence of this gene pair in the V-Y cosmid SV31:G9 (Vollmer and Yanofsky 1986), which carries the LG IIR gene aro-1 (Catchesideet al. 1985; Perkinset al. 2001), corroborates this assignment (Figure 4). The genomic locations of hH4-2 and the hH2A-hH2B gene pair were also determined by RFLP mapping using the standard sets of RFLP progeny (Metzenberget al. 1984). hH4-2 cosegregated with con-7 and trp-1 on LG IIIR and hH2A-hH2B cosegregated with ars-1 and the centromere of LG VII (Cen-VII; data not shown).
To map the histone and histone variant genes more precisely, we determined their location in the sequenced regions available from the NSP (Table 2) and MNCDB, searched for nearby genetic loci by BLASTn and BLASTx queries at NCBI and correlated matches with the Neurospora genetic map. The precise locations of all the histone genes were determined (Figure 4). The hH3-hH4-1 pair lies 15 kb centromere-proximal to aro-1 with hH4-1 lying centromere-distal to hH3. hH2Az also lies on LG II, 158 kb centromere-proximal to arg-5. Since no centromere-related sequences (Centola and Carbon 1994) were found between arg-5 and hH2Az, we conclude that it lies on the right arm of LG II and is transcribed toward the centromere. hH4-2 and hH3v were both found to lie on the right arm of LG III and are both transcribed toward the centromere. hH4-2 lies 38 kb centromere-distal to trp-1, whereas hH3v lies ≥121 kb centromere-proximal to ad-2. hH4v is located on the right arm of LG IV ≥416 kb centromere-distal to tol and is transcribed toward the centromere.
The hH2A-hH2B gene pair lies on supercontig 31 from the NSP (WICGR 2001), which also contains Cen-VII, in agreement with our RFLP mapping data. When we compared conceptual restriction digests of the contigs from supercontig 31 to published restriction maps of Cen-VII (Centola and Carbon 1994), we found that hH2A-hH2B lies to the right of the identified centromeric DNA with hH2A lying centromere-proximal to hH2B. Centola and Carbon (1994) originally determined the extent of the A + T-rich Cen-VII by constructing a restriction map of the centromere and flanking regions with the restriction enzyme PacI, which cleaves at an 8-bp recognition site composed solely of A:T pairs. The centromere was found to have a large concentration of PacI sites, while the relatively A + T-poor flanking regions were found to be devoid of PacI sites. To determine the distance between hH2A-hH2B and the centromere, we determined the concentration of PacI sites around hH2A-hH2B. We found that PacI sites were numerous both centromere-proximal and centromere-distal to it, suggesting that this gene pair may lie in centromeric heterochromatin (Figure 4).
Genetic identification of the histone gene complement: All of the core histones are essential in S. cerevisiae (Rykowskiet al. 1981; Schusteret al. 1986; Kayneet al. 1988; Mann and Grunstein 1992) and are presumably essential in all eukaryotes. Thus, as another test of whether we had identified the full complement of genes encoding each of the core histones of Neurospora, we attempted to inactivate each of the identified histone genes. Viability of strains bearing mutations in all known genes encoding a given histone would suggest that we had missed a functional histone gene, whereas lethality would suggest that we had not. Alternatively, lethality could result if redundant genes exist, but the expression level of each lies below a threshold for viability.
In the case of H4, the test required us to inactivate both hH4-1 and hH4-2. For hH4-2, this was accomplished by replacing the hH4-2 gene with a functional inl gene (ΔhH4-2::inl+), which encodes an enzyme in the inositol biosynthesis pathway (Perkinset al. 2001). Two strains (N2014 and N2015) were transformed with pSH18, a plasmid carrying ΔhH4-2::inl+, and Inl+ strains were selected. Southern analysis of the transformants indicated that homologous replacement events occurred at a frequency higher than that normally reported for Neurospora (Fincham 1989; Aronsonet al. 1994): 24% (10/42) for N2014 and 14% (6/43) for N2015 (Figure 5). Two ΔhH4-2::inl+ transformants of N2014 (N2016 and N2017) and one of N2015 (N2018) were rendered homokaryotic by microconidiation (Pandit and Maheshwari 1993) and used in further tests. We compared the growth rates of N2018 and its host N2015 in race tubes and found no difference (data not shown). We also found that the transformants were fully fertile either as males or as females (data not shown). Apparently, either hH4-1 fully complements the loss of hH4-2 or hH4-2 is not a major contributor of H4.
To determine the importance of hH4-1 in the production of H4, we inactivated hH4-1 with RIP, a process in which duplicated sequences are peppered with G:C to A:T transition mutations during the premeiotic stage of crosses (Selker 1990). Many sequences altered by RIP are found methylated in vegetative cells (Selkeret al. 1993; Singeret al. 1995). We targeted hH4-1 to the his-3 locus, crossed the resulting hH4-1 duplication strains and analyzed the progeny for mutations and/or methylation at hH4-1. The nonduplication parent carried a nonfunctional arg-12 allele, which lies <1 map unit from aro-1 (Perkinset al. 2001). Because the hH3-hH4-1 gene pair lies between arg-12 and aro-1, crossovers between arg-12 and hH4-1 should be rare. Therefore, Arg+ progeny were selected, enriching for progeny whose native hH4-1 allele came from the duplication parent and therefore had the opportunity to undergo RIP. Two of 12 progeny screened for RFLPs and methylation, N2022 and N2023, exhibited methylation at hH4-1. N2023 also exhibited an altered digestion pattern with DpnII (data not shown). The hH4-1 gene at the native locus of each of the strains was sequenced and both alleles showed extensive mutations by RIP (Figure 3C). The hH4-1RIP1 allele from N2022 exhibited 32 mutations in the 450 bp from the start codon through the stop codon, and allele hH4-1RIP2 from N2023 showed 22 mutations in this span. A conceptual translation of hH4-1RIP1 revealed 7 silent mutations and 13 missense mutations, whereas a conceptual translation of hH4-1RIP2 revealed 6 silent mutations and 6 missense mutations. Both alleles include a nonsense mutation at Q94, which should cause a deletion of 10 amino acids from the C terminus. In S. cerevisiae, deletion of the 4 C-terminal amino acids of H4 is not lethal, whereas deletion of 6 C-terminal amino acids is (Kayneet al. 1988). Therefore, even disregarding the missense mutations, it seems unlikely that our H4 mutant alleles are functional. The growth rates of the hH4-1RIP1 strains (N2022, N2023) and their parents were found to be equivalent, suggesting that hH4-2 supplied adequate H4. N2022 and N2023 were crossed as a male and female and with each other and in all cases were found to be fertile. Apparently, either hH4-1 or hH4-2 is sufficient for normal growth.
To confirm this conclusion and to verify that histone H4 is essential in Neurospora, we crossed strains carrying each hH4 mutant allele to determine whether or not hH4-1RIP1;ΔhH4-2::inl+ progeny could be generated. Two crosses were carried out. In the first, N2024 × N2025, both parents carried defective inl alleles, so that the inl+ gene that replaced hH4-2 was the only functional inl gene in the cross. Furthermore, the ΔhH4-2::inl+ parent (N2025) carried arg-12, which is tightly linked to hH4-1. Ascospores from the cross were activated and plated directly on medium lacking inositol and arginine, thus selecting for ΔhH4-2::inl+ and the arg-12+ allele near hH4-1RIP1. If strains lacking functional hH4-1 and hH4-2 genes could be viable, for example, due to a hypothetical third hH4 gene in the genome, the vast majority of progeny (>99%) should have been hH4-1RIP1;ΔhH4-2:: inl+. Instead, analysis of 10 progeny by Southern hybridization revealed that all carried the wild-type hH4-1 allele, suggesting that the hH4-1RIP1;ΔhH4-2::inl+ double mutant is not viable.
The second cross was equivalent to the first, except that a functional copy of hH4-2 was inserted at the his-3 locus of N2025 by transformation with pSH25, creating N2026. This artificially provided a third hH4 gene as a control. Progeny were selected as before, except that the functional hH4-2 at his-3 was selected by requiring histidine prototrophy. The majority of progeny (7/9) were hH4-1RIP1;ΔhH4-2::inl+. Therefore, hH4-1RIP1 and ΔhH4-2:: inl+ are null alleles and are synthetically lethal. On the basis of these results and our inability to detect other genes encoding H4, we conclude that hH4-1 and hH4-2 are the only functional hH4 genes in N. crassa.
We also attempted to recover an inactive hH3 allele by RIP. To provide a second copy of hH3 to activate RIP, we targeted hH3 to his-3 by transforming N1674 with pSH14, creating N2027, which was subsequently crossed with N2025. Progeny were selected using the same selection regime used in the hH4-1 RIP cross, described above, and 16 progeny were analyzed for hallmarks of RIP at hH3. No methylation or altered digestion patterns were observed (data not shown). Compared to the success rate of recovering null hH4-1RIP alleles (2/12), the inability to recover hH3RIP alleles (0/16) suggests that the loss of hH3 is lethal. On the basis of these results and our inability to identify other genes encoding H3, we conclude that hH3 is the only source of H3 in N. crassa.
To confirm genetically that hH2A and hH2B are unique in the genome, we employed the sheltered RIP strategy of Metzenberg and Grotelueschen (1992). This strategy relies on the inclusion of a mei-2 mutation in each parent to cause chromosome nondisjunction and thus production of aneuploid progeny. Disomic progeny break down rapidly to give heterokaryons with euploid nuclei that should be genetically identical, except for the chromosome that was subject to nondisjunction. One nuclear type should inherit this chromosome from one parent, while the other nuclear type should inherit this chromosome from the other parent. In this way a strain carrying an essential gene that has been disrupted by RIP can survive due to complementation by an extra wild-type allele.
In the sheltered RIP cross that we carried out, the two parents carried different mutations on the chromosome carrying hH2A-hH2B (LG VII). hH2A and hH2B were targeted to the his-3 locus of N1825 by transformation with pJS94 and pJS95 to create duplication strains N1822 and N1821, respectively. These transformants were crossed with N1679 and the resulting ascospores were germinated on medium lacking nicotinamide and arginine. This selected for the nic-3+ allele from N1821 and N1822 and the arg-10+ allele from N1679, resulting solely in the recovery of progeny disomic for LG VII. Eight progeny from the N1822 (hH2A-duplication) cross and 13 progeny from the N1821 (hH2B-duplication) cross were analyzed by Southern hybridization for hallmarks of RIP at hH2A and hH2B. One strain from the hH2A RIP cross, N2028, exhibited methylation at hH2A. Two of the progeny from the hH2B RIP cross exhibited methylation at hH2B and one of these (N2029) also exhibited an altered DpnII digestion pattern and was selected for further analysis (Figure 6).
To test whether the hH2ARIP and hH2BRIP alleles were null and essential, we crossed N2028 and N2029 with N1997, a mei-2 strain with a genetically distinct LG VII (Figure 7). Ascospores from this second cross were germinated on medium containing hygromycin and arginine, which allows only two classes of progeny to survive: homokaryotic euploid progeny carrying the mutant histone allele and progeny disomic for LG VII, arising from a cross between N1997 and the nucleus bearing the mutant histone allele. Since the homokaryotic progeny would be Arg− and the heterokaryotic progeny would be Arg+, we distinguished between these two expected classes by spot-testing on plates lacking arginine. In the cross involving the hH2BRIP1 allele, all eight progeny were Arg+. The apparent requirement for a wild-type allele of hH2B in any strain carrying hH2BRIP1 strongly argues that hH2BRIP1 is null and that hH2B is essential, indicating that it is the only source of H2B in Neurospora.
In the case of the cross involving the hH2ARIP allele, three of the seven progeny tested were Arg−, indicating either that the hH2ARIP allele is not fully defective or that it is a null and there is a second hH2A gene in the genome that makes up for this deficiency. To distinguish between these two possibilities, we amplified the hH2ARIP allele from the Arg+ progeny by PCR and sequenced it. Mutations from RIP were found, but it was not obvious that they would result in a null allele. In the 594-bp coding region and introns, only five mutations were found, three of which are silent mutations (Figure 3A). Since we were unable to generate a definite null allele of hH2A, we could not determine genetically the number of genes encoding H2A in the N. crassa. Nevertheless, since only hH2A was detected when genomic DNA blots or genomic cosmid libraries were screened with hH2A and since all of the H2A ESTs and genomic sequences that we found belong to hH2A, we conclude that hH2A is unique in the genome.
Evolutionary relationship of the core histones of N. crassa to those of other fungi: Previous phylogenetic analyses of histones included few fungal representatives (Thatcher and Gorovsky 1994). To place the Neurospora histone genes in a fungal phylogeny, we searched the nonredundant databases at NCBI and several publicly available fungal EST databases to identify as many fungal core histone genes as possible. The amino acid sequences of the identified histone genes were then aligned and phylogenetic trees were constructed. All of the histones of Neurospora were found to group consistently with those of other Euascomycetes, as expected from previous phylogenetic studies based on morphological and molecular data (Spatafora 1995; Liuet al. 1999; Berbeeet al. 2000; Figure 8). As known for eukaryotes as a whole (Thatcher and Gorovsky 1994), H2A and H2B are the least highly conserved histones in fungi (Table 3) and most of their divergence is found in the N- and C-terminal tails. In contrast, H3 and H4 are more highly conserved (Table 3) and most of their divergence occurred in the structured globular domains.
Histone H4 of N. crassa is identical to one of the two H4 proteins of Botrytis fuckeliana and to the one H4 protein from F. sporotrichioides for which the entire amino acid sequence is available (Table 3). To estimate the time of divergence between hH4-1 and hH4-2 relative to their divergence from these counterparts, we constructed a phylogenetic tree of the coding sequences of the H4-coding genes from the euascomycetes. Although most of the tree is at low resolution, hH4-1 and hH4-2 cluster together with high confidence (data not shown). Thus, it appears that the two hH4 genes of Neurospora arose from a relatively recent gene duplication event. This is supported by an examination of intron locations between the two hH4 genes of Neurospora and the two H4-coding genes from E. nidulans, the only other euascomycete for which the genomic sequence of the H4 genes are known (Ehingeret al. 1990). Although the location of the second intron in hH4-1 and hH4-2 is not shared by the two H4 genes from E. nidulans, the first intron is located at the same position in all four genes.
We set out to identify all of the genes encoding the core histones in N. crassa. Previously, a single gene pair (hH3 and hH4-1), coding for H3 and H4, had been identified (Woudtet al. 1983). We identified another gene encoding H4 (hH4-2) and single closely linked genes encoding H2A and H2B (hH2A and hH2B). To be certain that we had identified the entire complement of histone genes, we assessed the dispensability of each histone gene genetically. In the cases of hH2B and hH3, we found that we could not generate viable strains carrying null alleles at these loci. Strains carrying null alleles of hH4-1 or hH4-2 were viable and phenotypically wild type, however, suggesting that neither of these genes is solely responsible for producing all the histone H4 in Neurospora. We were unable to generate a double mutant without providing an extra functional hH4 gene. We were unable to generate a null allele of hH2A, but our inability to detect homologs, as discussed above, leads us to conclude that hH2A is unique in the genome. We identified variant forms of histones H2A (hH2Az), H3 (hH3v), and H4 (hH4v), but do not know whether they are involved in the same processes as their homologs in other species.
RFLP mapping and database mining were used to map hH2Az and the hH3-hH4-1 gene pair to LG IIR, hH3v and hH4-2 to LG IIIR, hH4v to LG IVR, and the hH2A-hH2B gene pair to LG VII near the right edge of the centromere. The number and arrangement of the core histone genes in N. crassa is identical to that in the closest relative for which a comprehensive search of histone genes has been undertaken, E. nidulans (May and Morris 1987; Ehingeret al. 1990). Furthermore, an analysis of numerous cDNA sequences available for the euascomycetes F. sporotrichioides (Roeet al. 2001c) and B. fuckeliana suggests that they also have single genes for H2A, H2B, and H3 and two genes coding for H4. Why do euascomycetes appear to maintain two H4 genes, when single genes suffice for the other core histones? We found that under laboratory conditions either gene is sufficient for wild-type viability, fertility, and rate of growth. Interestingly, in E. nidulans the abundance of hhfB mRNA, which codes for H4.2, is regulated differentially from that of hhfA, whose product is 98% identical to H4.2, suggesting that these H4 genes play different roles (Ehingeret al. 1990). Perhaps different euascomycetes maintain two H4 genes for different reasons.
Although the single-copy histone genes from each euascomycete are presumably orthologous, the hH4 genes appear exceptional. For instance, the H4 phylogenetic trees show that H4-1 and H4-2 group together, while H4.1 of E. nidulans forms a close sister group and H4.2 of E. nidulans forms the outgroup (Figure 8B). This indicates that hH4-1 and hH4-2 are not orthologs of hhfA (H4.1) and hhfB (H4.2), respectively, but instead are paralogs of hhfA, having arisen from a duplication of an ancestral ortholog of hhfA.
The progenitor of Neurospora that duplicated the ancestral hH4 gene may have been competent at RIP or an equivalent duplication-sensing mechanism, since the fungus Ascobolus immersus, which is more distantly related to Neurospora than E. nidulans, has an apparently related mechanism, MIP (Goyon and Faugeron 1989). The duplication of an hH4 gene may have escaped detection due to its small size. The hH4 coding region requires only 312 bp, below the apparent minimum (~400 bp) for a duplication to be subject to RIP or MIP (Goyonet al. 1996; Watterset al. 1999). This suggests that RIP/MIP may not be an impediment to the evolutionary development of new genes through gene duplication, so long as the genes are small.
RIP/MIP may be a factor in the unique genomic organization of euascomycete histone genes, including the minimal histone gene set and the existence of introns in the histone genes. Although introns are extremely uncommon in histone genes, generally (Walliset al. 1980; Choe et al. 1982, 1985; Maxsonet al. 1983; Smith and Andresson 1983; Old and Woodland 1984; Matsumoto and Yanagida 1985; Horowitzet al. 1987; Stark and Milner 1989; Wefes and Lipps 1990; Chabouteet al. 1993; Puertaet al. 1994; Mackenzieet al. 2000; Mollapour and Piper 2001), each report of a histone gene from a euascomyte has noted the existence of introns (Woudtet al. 1983; May and Morris 1987; Ehingeret al. 1990). In fact, introns are found in all 14 euascomycete histone genes, for which the genomic sequence is known, which includes all core histone genes from N. crassa, all core histone genes from E. nidulans, H2A from B. fuckeliana, H2A and H2B from Podospora anserina, and H3 from Ajellomycetes capsulatus. Thus introns in histone genes are the rule, not the exception in euascomycetes, setting them apart from yeasts (Walliset al. 1980; Choe et al. 1982, 1985; Smith and Andresson 1983; Matsumoto and Yanagida 1985; May and Morris 1987; Ehingeret al. 1990) and most other eukaryotes (Maxsonet al. 1983; Old and Woodland 1984; Chabouteet al. 1993). The common ancestor of euascomycetes likely carried multiple histone genes. If so, the evolution of RIP/MIP would have made this situation untenable, since the coding regions for histones H2A, H2B, and H3 are >400 bp in length, making uninterrupted duplicate copies of these genes substrates for these genome defense systems. Consequently, only histone genes that were altered, by the insertion of nonhomologous introns to break up the contiguous stretches of homology and/or by a reduction in the number of histone gene sets to one, would be able to survive repeated sexual crosses intact.
We thank Michael Freitag and Lenna Kuzminova for materials used in some experiments, Robert Metzenberg and Greg Kothe for discussions, and Michael Freitag for comments on the manuscript. S.M.H. was supported in part by U.S. Public Health Services training grant GM-07759. This work was supported by U.S. Public Health Services grant GM-35690 to E.U.S.
Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AY062171, AY062172, and AY062173.
Communicating editor: R. H. Davis
- Received November 21, 2001.
- Accepted December 26, 2001.
- Copyright © 2002 by the Genetics Society of America